DRUGS AGAINST CANCER and the Quest for a Cure From discovery to clinical application by Kurt W. Kohn, MD, PhD Developmental Therapeutics Branch, NCI, NIH DNA backbones - DNA base-pair Normal DNA with bound DNA lntercalator (red) Leonard Lerman's DNA intercalaion hypothesis (1961) Doxorubicin (Adriamycin) Sidney Farber and methotrexate https://discovery.nci.nih.gov /kohn-book-d rugs-against-cancer.jsp Titlitpo!Jll22JIJ09d Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@nih.gov � This work wos produced by the Notional Concer Institute, NIH, ond the text therein is freely available in the public domain. Many of the figures however hove been copied from previously published articles and their further reproduction may require permission from the publishers of the articles in which those figures appeared. Dedication To my son Philip David Kohn And my granddaughter Melissa Sara Naugle. And to the memory of My Father Dr. Siegfried Kohn, My Mother Sara (Margulies) Kohn, And my daughter Julia Alison (Kohn) Naugle. And to the memory of My childhood friend and cousin, Helga Charlotte Kohn, and her father Albert Kohn. And the many other close relatives who perished in HaShoah. And to the memory of My father's older brother, Wilhelm Kohn, Who died in 1914, shortly before Yorn Kippur, at the age of 17, fighting in the Austrian army during World War I, and after whom I was given my middle name. Kurt W. Kohn, MD, PhD Potomac, Maryland August 26, 2022. Contents Preface and acknowledgements...•.•..•..•..•..•..•..•..•..•..•..•..•..•..•..•..•..•..•..•..•..•..•..•.•..•..•..•..•..•..•..••..•..•.•..•..• 6 Introduction..•..•..•..•..•..•..•..•..•..•..•..•..•..•..•..•..•..•..•..•..•..•..•.•..•..•..•..•..•..•..•..••..•.•..•..•..•..•..•..•..•..•..•..•..•..•..• 8 1. Anti-cancer drugs that crosslink DNA ............................................................................................... 22 2. The temozolomide story: DNA-GO6 alkylation and repair ....................................................... 58 3. The Platinum Story: From Imagination to a New Anticancer Drug ......................................................................................................................................... 84 4. The DNA Intercalation Story: Drug-DNA sandwiches ..............................................................101 5. The methotrexate story: folic acid analogs ...................................................................................111 6. The 5-Fluorouracil Story: from a simple idea to a major anti-cancer drug .......................................................................................................................................128 7. The 6-mercaptopurine (6MP) story .................................................................................................148 8. The Doxorubicin Story: a star with a nearly fatal tlaw ............................................................. 158 9. The DNA filter elution story: a new way to measure DNA Damage ..............................................................................................................................................169 10. The Topoisomerase II Story: from methodology to a new anticancer drug target ...........................................................................................................................183 11. The Topoisomerase I Story: camptothecin, from a Happy Chinese Tree ..............................................................................................................................................212 12. The Mitotic Inhibitor Story: taxol and vinca ................................................................................. 232 13. The Bleomycin Story: an anticancer drug with a unique mode of action...........................................................................................................................................262 14. The Philadelphia Chromosome Story and a new era of targeted cancer therapy ..........................................................................................................................................276 15. The Oncogene Discovery Story ..........................................................................................................299 16. The Oncogene Addiction Story: a conceptual basis for cancer Therapy ........................................................................................................................................................ 307 17. The EGFR Oncogene story .................................................................................................................... 312 18. The RAS oncogene story ....................................................................................................................... 354 19. The BRAF-melanoma story .................................................................................................................. 376 20. Anticancer drug discovery and development at the National Cancer Institute (NCI) ............................................................................................................................ 396 21. The DNA Repair Story: early discoveries .......................................................................................432 22. Genetic diseases reveal DNA nucleotide excision repair ........................................................453 23. The DNA Nucleotide Excision Repair Story: cutting out the Damage ........................................................................................................................................................ 471 24 . The DNA Base Excision Repair Story: removing bad bases ................................................... 483 25. The DNA Mismatch Repair Story: fixing base-pairs that don't Match ............................................................................................................................................................ 494 26. The BRCA and homologous recombination story ...................................................................... 512 27 A. DNA double-strand break repair by homologous recombination....................................... 526 27B. DNA double-strand break repair by nonhomologous end joining ...................................... 546 28. The gamma-H2AX story: DNA double-strand breaks revealed in cell nuclei ................. 559 29. The ataxia telangiectasia story and the ATM gene .................................................................... 565 30. The PARP story and a new strategy for cancer therapy .......................................................... 578 31. The Fanconi anemia story and the repair of DNA crosslinks ................................................ 614 32. The p53 story and Li-Fraumeni Syndrome ................................................................................... 637 33. The retinoblastoma story - control of cell division ................................................................... 660 K. W. Kohn Drugs Against Cancer Preface PREFAC£110llflrt3 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@oib gov Preface and acknowledgements First and foremost, I must express my gratitude to the United States Public Health Service and the National Cancer Institute for giving me the opportunity for a lifetime of wonderful association with biomedical researchers and to contribute a little to efforts to understand and hopefully to eventually control cancer. In a sense, this work is a tribute to those efforts, especially on the historical efforts to develop new drugs for cancer chemotherapy and the investigation of their mechanisms of action. I owe special gratitude to Gordon Zubrod, David P. Rall, Emil (Tom) Frei, Emil J Freireich, Ti Li Loo, and Jack D. Davison for acquainting me with the clinical, pharmacological, and chemical aspects of cancer research and helping me find my way to the research area to which I seemed best adapted. Several others to whom I owe debts of gratitude will appear in the narrative of the following chapters. Any huge research and development endeavor is unlikely to be free of all shortcomings from which lessons could be learned. Although I have tried to give an accurate historical account, I have no doubt committed errors, which might be corrected by future authors examining the available records and interviewing surviving protagonists. Starting on this work, I felt a responsibility to make a record of the part of the cancer drug development effort with which I was associated, going back nearly 60 years, and to relate it to the global anti-cancer drug discovery and development efforts. However, I don't pretend to have written a scholarly history, which remains for others to accomplish. This work may be viewed as a combination of science, history, medicine, and memoir, and hopefully could mostly be understood without a great deal of prerequisite knowledge. However, some technical material is included for experts and students and for those who may have the fortitude to read some perhaps 6 K. W. Kohn Drugs Against Cancer Preface challenging parts of the text However, I have endeavored to use non-technical language even in explaining those more complicated aspects. To that end, I have minimized the use of abbreviations. For the most part, the chapters are self- contained and can be read in any order. I have tried to give an account of the earliest published work leading to the anticancer drug discovery stories that I will relate. My aim then was to explain how the knowledge and application of cancer chemotherapy drugs developed. However, I could not possibly have thoroughly covered the entire territory of cancer drug discovery in this writing. There are no doubt areas of omission or misunderstanding. and I apologize to those whose work I failed to mention or give adequate credit Preliminary versions of some chapters are available on the website of the Laboratory of Molecular Pharmacology in the Developmental Therapeutics Branch (DTB) of the National Cancer Institute. They were uploaded and managed under the direction of William C. Reinhold, Head of the Genomics and Bioinformatics unit in the DTB, and Yves Pommier, Director of the DTB. It is a pleasure also to acknowledge many colleagues and friends who have commented and helped to correct and improve some of these chapters, particularly Albert Fornace, Silvio Parodi, Mirit Aladjem, Yves Pommier, Leonard Zwelling. Kenneth Kraemer, Vilhelm Bohr, William Reinhold, as well as my old literary friend Barbara Herrnstein Smith. 7 K. W. Kohn Drugs Against Cancer INTRODUCTION I NTRODIJCTtOilf 1207J&ou3 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@oib gov INTRODUCTION I should perhaps begin by relating how it happened that I was to devote nearly 60 years of my life to anti-cancer drug research at the National Cancer Institute (NCI) in Bethesda, Maryland. In his recent book, "The Death of Cancer, 2015" (DeVita Jr., 2015), Vincent DeVita writes of his early life, which in some ways parallels my own. We both grew up in New York in the 1940's, although he was a few years younger, and we did not know each other. We attended different medical schools, and independently joined the United States Public Health Service as an alternative to becoming inducted as young doctors in the armed forces of the United States. We both became Clinical Associates (physicians in training) at the National Cancer Institute, which, for both ofus was a second choice. DeVita's first choice would have been the Heart Institute, while I had at first aimed for a research position in a neurophysiology group. He relates his early formative experience at the age of 6 when a beloved aunt died of disseminated cancer. Mine would have been our family's narrow escape from Nazi Vienna in August 1938, when I was 7 years old. De Vita then spent his career developing cancer cures in clinical research, whereas my bent was for laboratory research to try to find out how anti-cancer drugs work. Upon arriving at NCI in 195 7, I was assigned to a Clinical Pharmacology unit and thought it a promising direction to take in view of my interest in chemistry, in which I had majored at Harvard College (chemistry and physics combined major). My first assignment however was to help care for children on the acute leukemia service, after which I was assigned to the adult cancer ward. That occupied much of my first two years at NCI. In his book, DeVita tells of a pretty 10 year old girl with acute leukemia to whose room he was called one night, while he was an intern at the University of Michigan in 1962 (DeVita Jr., 2015). She was in the last stage of her disease and he was called upon to restart an I.V. that had stopped working. Her overused veins were difficult to access, but he was lucky and slipped the needle into a vein on the first try. She 8 K. W. Kohn Drugs Against Cancer INTRODUCTION smiled, because it was long since a needle had not hurt ... and she gave him a SO-cent coin in appreciation. He says that it was this image that he took with him when he came to work in the childhood leukemia ward at NCI. When I was a third-year medical student I saw a very intelligent young man of about 20 who was in the last stage (blast crisis) of acute leukemia, and for whom nothing more could be done at the time. I saw him vomiting blood continuously . .. but he kept apologizing to the nurses for making such a mess! That poignant image was with me when I arrived to work in the same childhood leukemia ward in 1957 -- 2 East on the second floor of the Clinical Center-- a few years earlier than DeVita. I was recruited to the Clinical Associate program of NCl's new clinical research program by NCI Clinical Director C. Gordon Zubrod (Figure 0.5), who recognized my bent and background in physical sciences and wisely assigned me to work with David P. Rall (Figure 0.6) in the new Clinical Pharmacology Service in NCl's Medicine Branch to begin after my first 2 years in the childhood leukemia and adult cancer clinical programs. The childhood leukemia ward in 1957 had some of the same feel to it that DeVita describes when he was a Clinical Associate there about 5 years later. Though depressing to those who knew the score, the nurses managed to keep everyone who was not in immediate pain cheerful. Although some of the children would go into briefremission, all them soon died. Nevertheless, we all suppressed the dreadful facts as we watched cheerful television cartoons with the children on sunny weekend mornings. The designers of the Clinical Center were wise to place the patient wards on the South side of the building. In order to avoid being drafted, I did not go on to residency training and came directly to NCI right after internship. Therefore, I was really under-qualified for my role to help care for very sick children. But fortunately, I had excellent guidance and supervision from Emil J Freireich ("Jay"), who was chief of the unit. In our regular meetings with Emil Frei ("Tom"), who headed the entire NCI clinical program, we reviewed the drug treatments and clinical problems and continued our discussions at breakfast in the cafeteria. Tom always had a stack of big yellow punched cards (the best data management system at the time) sticking out of a pocket of his long white coat. I asked many questions, and was impressed by his scholarly answers, which he always backed up with specific data references. J Freireich was full of wild ideas that he expressed with great confidence, in contrast to Tom Frei's scholarly precision. But both were intent on getting an effective drug treatment for leukemia. Their different ways of thinking complemented each other, and their collaboration was extraordinarily productive. Any idea of a cure however was far away in the fog. The personalities and extraordinary successes of the Freireich-Frei duo is well described by John Laszlo in his book "The Cure of Childhood Leukemia" (Laszlo, 1995). Freireich's delight with any new idea that 9 K. W. Kohn Drugs Against Cancer INTRODUCTION might come to mind and Frei's scholarly demeaner, are well shown in old photos I found on the internet (Figure 0.4). Frei's scholarly demeaner, by the way, hid a brilliant sense of humor and a talent for vaudeville accentuated by his height and long legs. In 1957, we were giving only one drug at a time, methotrexate or 6-mercaptopurine. Since two or more antibiotics given together were known to help patients with tuberculosis, I naively suggested in our meetings with Tom Frei that we try giving both drugs together. That idea was promptly shot down, because we didn't yet know enough about the actions of each of those drugs by itself. DeVita describes the triumvirate: Freireich, Frei and Rall, and their single-minded efforts to cure cancer, particularly leukemia. He well described Freireich and Frei, and how their eccentric personalities helped to eventually get a cure. My wife, Elaine, and I attended many parties at the Rail's and Frei's and were entertained by Tom's Vaudeville talents, although we missed the wild party described by DeVita (DeVita Jr., 2015). After my stint in the childhood leukemia and adult cancer wards, I joined David Rall (Figure 0.6) in his studies of how drugs could be made to pass the blood-brain barrier. Like De Vita, I had been recruited by Rall for the Clinical Associate position. The blood-brain barrier was one of the road-blocks to effective treatment, because it kept the drugs out of the brain where residual leukemia cells were often lurking. In those early days, I did not have much hope for the trial-and-error clinical drug testing. and felt it was better to take the basic science tack. Therefore, I stuck to the lab and gradually reduced my attendance at ward rounds. The first 4-drug combination, VAMP, seemed bizarre ... but it produced a few cures (DeVita Jr., 2015). 1couldn't have been more surprised! It had seemed to me that the drug combination was designed empirically without sufficient basic knowledge. I was mistaken, however, because the combination design was based on careful dosage and toxicity considerations, as well as some general notions of drug mechanisms. The combination worked, even though our knowledge of how it worked was mostly lacking. I was greatly encouraged and inspired by close association with members of the Clinical Pharmacology Service to which I had been assigned: Dave Rall, Ti Li Loo, Jack Davidson and we would often continue our discussions at lunch in the cafeteria in the Bl level of the Clinical Center. The labs of the Clinical Pharmacology Service, by the way, were located in the East wing on the 6 th floor of the Clinical Center. Dave Rall (Figure 0.6) had wonderful insights in physiology, particularly about the anatomy and physiology of the blood-brain barrier and the chemical factors (ionization states) of drugs that determine their ability to penetrate the barrier and enter the brain. 10 K. W. Kohn Drugs Against Cancer INTRODUCTION Ti Li Loo (Figure 07) was an organic and medicinal chemist with whom I had many fruitful discussions about chemical mechanisms and how the different ionization states of compounds can be determined by spectral analysis. Jack D. Davidson had keen physical insight and liked to design and modify experimental apparatus. He was a leader in the field of scintillation counting of radioactivity in biological samples and had one the first automatic multi-sample scintillation counters manufactured. He solved the problem of how to dissolve an aqueous biological sample in the toluene-based solution required to produce the scintillations. His reading and commenting on my first manuscripts before submission for publication were extremely helpful. He went on to head the Nuclear Medicine department in the NIH Clinical Center. My first research in the Clinical Pharmacology Service of the Medicine Branch (1957-1959) followed up on a clever (albeit unsuccessful) attempt to discover a new anti-cancer vitamin analog. One of the senior scientists in our group, Montague ("Monty") Lane, was engaged in a project that caught my attention. Since methotrexate, an analog of folic acid, was successful as an anti-cancer drug. Monty decided to test an analog of another vitamin, riboflavin. In animal tests, Monty had assured himself that his riboflavin analog was not toxic. He then tried the riboflavin analog on one of his patients whose cancer was very advanced and terminal, and for whom no further therapy was known. (In 1956, constraints on clinical testing were not yet formalized.) When the patient soon died of her disease, Monty wanted to look for the strong fluorescence of riboflavin to see whether it had gotten into her tumors. He indeed saw bright fluorescence in the cancer tissues. But he found that the fluorescence in the tumor was not due to his riboflavin analog. Checking the clinical record, he found that the patient had been given tetracycline to treat an infection. He found that the bright fluoresce indeed was due to tetracycline that was concentrated in the tumor tissue. That piqued my interest, and my first research projects at NIH focused on tetracycline as an agent that selectively bound to tumor tissue (Kohn, 1961a, b). It later turned out, however, that tetracycline was binding only in the necrotic (i.e., dead) parts of the cancer tissue. So, that idea led to a dead end as far as therapy was concerned. Dave Rall, however, was interested in my work for another reason: it seemed a good way to investigate how a drug passed through the blood-brain barrier. I had found that, when tetracycline binds calcium, it can be induced to migrate into a lipid solvent (analogous to the lipid layer that constitutes the blood-brain barrier), where its fluorescence in the lipid phase was greatly enhanced, which made it easy to detect and measure by means of an assay I had developed (Kohn, 1961a). (That first paper of mine, by the way, was published in Analytical Chemistry, which was ironic, because I had managed to avoid courses on that seemingly uninteresting area of chemistry at college.) 11 K. W.Kohn Drugs Against Cancer INTRODUCTION In order to induce the tetracycline-calcium complex to become highly lipid-soluble, however, a third component was necessary - the most effective turned out to be a barbiturate. In order to test various barbiturates, I was able to order them from chemical companies, because access to many drugs that were to become widespread abuse problems was not yet restricted. Years later, I discovered in the back of a high shelf a dusty old brown bottle that contained nearly half a pound of pen tobarbital that I had recrystallized and later forgotten, and I had to explain why I had in my possession a large quantity of a restricted drug without proper approval and documentation. Anyway, I applied those findings to devise a new sensitive assay for tetracycline, which we used to investigate the permeation of tetracycline from the blood into the brain. I also found that calcium mediates the binding tetracycline to DNA (Kohn, 1961b). Those were my first studies at NIH, before I moved to work in Paul Doty's laboratory at Harvard for two years. When DeVita was Director of the Division of Cancer Treatment at NCI, I was Chief of the Laboratory of Molecular Pharmacology that I had founded in the Division. In our discussions and conferences, we sometimes disagreed, but he listened carefully to alternative views and was always cordial. He had the difficult task of encouraging creative research while keeping on track to curing cancer. It was a difficult task, where one could never be sure that his/her decision was the best one. His viewpoint was well expressed at a Division meeting when a major new effort was being debated. When I brought up why I thought that the planned project was unlikely to succeed, his response was a quote he attributed to Winston Churchill to the effect that "the demand for perfection spells paralysis." In 1981, my Laboratory was joined by Yves Pommier who had received an MD degree from the University of Paris and carried out chemotherapy studies there. He became active in our studies of the effects of anti-cancer drugs on the DNA- associated enzyme topoisomerase II. We carried out much of the early work together and remained close colleagues and friends. His investigations progressed and expanded, leading to his appointment in 1995 to succeed me as Chief of the Laboratory of Molecular Pharmacology upon my retirement from that position. He later became Director of a new Developmental Therapeutics Branch that included the Laboratory as well a clinical research component I tell about many of those investigations in the chapters that follow, including his role in the more recent drug discovery accomplishments at NCI. Going back, my interest in science began around age 12 with a passion for astronomy, which led to my becoming active in the Junior Astronomy Club at the American Museum of Natural History in New York, among other teenagers who were passionate about mathematics and science; several of whom went on to become well known in those fields. Four of us came to Washington, DC as winners in the Westinghouse Science Talent Search of 1947, where we met other like-minded 12 K. W.Kohn Drugs Against Cancer INTRODUCTION young people; several of us then attended Harvard College and continued our friendship there. I wanted to become a physicist, but soon found that several of my friends had much more math talent than I did. Therefore, I thought to settle for physical chemistry. But I came to feel that the future was in biology, in which I also had an intense interest (my Science Talent Search project was on how ants recognize members of their nest mates). I guess I was driven by a desire to know how the physical and biological world worked as revealed by the sciences. Finally, I settled on the new field of biophysics, but was advised that in order to get into that field I should go to medical school, because that was where the best research was being done. I became enthusiastic about that idea, because I wanted to find out how the human body and medical treatments work. And so it happened that my father's urging me to go to medical school came to fruition after all, whereas I had always denied any intention of doing that because I had thought it a diversion from pure science. I attended Columbia's medical school, the College of Physicians and Surgeons (P&S) in New York. I had been set to attend Harvard Medical School but could not give up a New York State Medical Scholarship. However, P&S turned out to become a key to my future research, as I will go on to explain. My professional life and experiences at P&S were in several ways entangled with some of the stories I tell in this book. That entanglement began in 1953 in Alfred Gilman's pharmacology lectures during my first year as a medical student at P&S. Gilman's lectures, although masterly in clarity and scope, were a challenge in note taking, because his crystal-clear delivery was rapid and unrelenting, and there were no practical recording devices, no internet, and the only informative pharmacology textbook, the first edition of Goodman and Gilman's classic, "The Pharmacological Basis of Therapeutics" (1941), was hopelessly out of date. As already mentioned, I had entered medical school after majoring in chemistry and physics, with the idea of preparing for research in some area of biophysics. For me, Gilman's key lecture turned out to be his description of the chemistry of nitrogen mustard, which he had a major role in elucidating during World War II (Gilman and Philips, 1946) (see Chapter 1). My ears perked up even more than usual in his always insightful lectures when he described how the nitrogen mustard molecule had to have two reactive sites in order to be effective as an anti-cancer drug. Moreover, the chemistry Gilman had helped to unravel showed that the two sites could each bind tightly (covalently) to something. Evidently nitrogen mustard worked by forming tight cross-links in or between some important biomolecules; but what were they? That question intrigued me but remained latent in my mind until aroused unexpectedly 7 years later. I may have already associated the number two from Gilman's lecture with the number of strands in Watson and Crick's 1953 model of DNA Those two stands have to separate in order to form new copies of the genetic material before the cell divides. If some tight crosslink held them together, the DNA could not be duplicated, 13 K. W. Koh n Drugs Against Cancer INTRODUCTION and the cell could not divide normally. I'm not sure whether that notion was already hazily in my mind at the time, but it became loud and clear when I started working in Paul Doty's laboratory at Harvard as a post-doc in 1959 (see Chapter 1). While at medical school, there was another connection with nitrogen mustard. My closest classmate friend, with whom I did my first research projects, was Edgar Haber (1932-1997) - we worked together on projects ranging from electronics to physiology and pharmacology - went on to an illustrious clinical and research career in cardiology. Ed was a relative of Fritz Haber, the inventor of mustard gas. Chapter 1 relates the story of how that World War I poison gas was a prelude to the development of nitrogen mustard as the first anti-cancer drug. I will also touch upon the ethical issues of Fritz Haber's poison gas research (see Chapter 1). 1 will also note that Fritz Haber's development of the process to produce ammonia from atmospheric nitrogen, so as to provide a desperately needed new source for agricultural fertilizer saved more lives from starvation than were lost during both World Wars combined. Benefit-harm dualities, however, are often complicated: the Haber-Bosch process for converting atmospheric nitrogen to ammonia and hence nitrates, also fueled Germany's production of explosives for weaponry and may have prolonged the war. In the meantime, I had graduated from medical school, interned at Mount Sinai Hospital in New York, joined the U.S, Public Health Service as a medical officer, and gotten a job as a Clinical Associate at the National Cancer Institute in Bethesda, Maryland. Also in the meantime, I had married Elaine Kay Mogels (1931-2013) and we soon had two children, Philip and Julia. During our first year in Bethesda, Elaine and I both had jobs in the new NIH Clinical Center. Living and working in Bethesda in 1957 was wonderful, especially for us New York city dwellers. It was a very pleasant rustic rural setting in which the NIH was mostly manicured lawns, trees, shrubs and flowers. The new Clinical Center stood out as a huge all-brick monument (Figure 0.1). But my entanglement with the events to be related also had a clinical and biological side, which has to do with how it happened that I came to do research at the then little-known National Institutes of Health (NIH). During my 3rd medical school year (1955), I had a 3-month elective at Goldwater Memorial Hospital on Welfare Island (now known as Roosevelt Island) in New York's East River (Figure 0.2). It was a chronic disease hospital where Columbia had two clinical research units; I was assigned to a unit of all-male patients that was studying hypertension; the other was an all-female unit focused on rheumatoid arthritis. During my time at Goldwater, we had a guest lecturer from the NIH, Sidney Udenfriend (1918-1999), who had worked on the anti-malarial drug discovery program at Goldwater during the World War II. He was one of the many researchers from Goldwater who were recruited to staff the expanding NIH. Indeed, I heard the NIH referred to as "the Goldwater on the Potomac"! For me, however, it spelled an opportunity for a research career. Moreover, Udenfriend's lecture influenced my initial research direction at NIH. He described his use of the new spectrophotofluorometer he had developed together 14 K. W.Kohn Drugs Against Cancer INTRODUCTION with Robert Bowman and that was manufactured by AMIN CO (Udenfriend, 1995) (Figure 0.3). It became the major research instrument in my tetracycline studies (Kohn, 1961b). It was a strange feeling recently to see that instrument in the museum exhibits in the Clinical Center. It was also a strange feeling to see the analytical ultracentrifuge, which was the major research instrument in my DNA crosslinking studies (Kohn et al., 1966), also consigned to "ancient history" museum displays (see Chapter 1). In both cases, I witnessed the origin and finite lifespan of a new research technology. The present work summarizes some of the main stories behind the discoveries, successes and failures in the early efforts to develop drug treatments for cancer. Tremendous advances have been made in recent decades and the outlook for the future is bright. However, the early history that contributed to those recent advances is not so well known. I focus on past discoveries with the view that the era of cytotoxic cancer chemotherapy may gradually wane as molecularly targeted therapies come to the fore. It may be appropriate therefore for some of us who have experienced that early period, to assemble the highlights of how those discoveries were made and how they progressed to the current era of research and clinical application. Chapters 1 through13 relate the history of discovery and mechanisms of action of each of the major cancer chemotherapy drugs. The subsequent chapters relate the advent of targeted cancer therapeutic agents, such as inhibitors and monoclonal antibodies directed against tyrosine kinase enzymes and molecules required to repair damaged DNA. They also focus on the interaction networks that determine life or death of a cell in response to anticancer treatment. 15 K. W. Kohn Drugs Against Cancer INTRODUCTION Figure 0.1. The National Institutes of Health (NIH) in 1955, shortly after completion of the Clinical Center (NIH Building 10, upper right), about 2 years before I started working there as a Clinical Associate in the National Cancer Institute. The Clinical Center was by far the largest building on campus (it was said to be one of the largest all-brick buildings in the world) and was to become even much larger with modern additions to right side (the North side) of the building. From the beginning, the Clinical Center included a hospital and research laboratories for several of the NIH Institutes. The meadow in the upper left was later replaced by new buildings. The building with pillars, near the center of the picture, is Building 1, which was the administrative center of NIH. All of the buildings in the foreground that are similar in architecture still existed in 2016, albeit with remodeled interiors. The rectangular building at the left edge of the picture, building 13, was the first major addition after the Clinical Center; it included mechanical shops for construction of new apparatus and an area for surplus equipment. The building visible through the trees at the top is the old Suburban Hospital across Old Georgetown Road from NIH. 16 K. W.Kohn Drugs Against Cancer INTRODUCTION Figure 0.2. Goldwater Memorial Hospital in 1938, as seen from the Queensborough Bridge. This vast chronic disease hospital was located on Welfare Island (later called Roosevelt Island), a two-mile sliver of land in the East River nestled between the Upper East Side and Astoria. In addition to caring for a large number of chronic disease patients, Goldwater included clinical research departments associated with the Columbia, Cornell, and NYU medical schools. The hospital, opened in 1939, was an immense facility designed to be a new model of medical care for patients with chronic illnesses. Researchers in the Columbia unit solved the anti-malaria drug problem during World War II. Many of those researchers were recruited to lead the clinical and research programs of the newly expanded NIB. The hospital closed in December 2013, but before its destruction, a detailed photographic record was made (http; //nrhanomnjhns net/2 014 /04 {antopsy-of-a-hospita)-a-photographic-record- of-coler-goidwater-on-roosevelt-island/). 17 K. W. Kohn Drugs Against Cancer INTRODUCTION Figure 0.3. Sidney Udenfriend (1918-2000) in the late 1950's. Resting on the tabletop in the background is an AMINCO-Bowman spectrophotofluorometer that he helped to develop, and that I used in my first studies at NIH, (Udenfriend, 1995). Figure 0.4. Emil J Freireich (1927-2021) (left) and Emil "Tom" Frei Ill (1924-2013) (right) as I remember them in 1958; they taught me the essentials of clinical cancer chemotherapy research. After enduring family hardships in Chicago during the Great Depression, Freireich attended the University of Illinois and received an M.D. degree in 1949. After post-doctoral work in hematology, he joined the U.S. Public Health Service and came to the NIH in 1955. Emil ("Tom") Frei received his medical degree from Yale University in 1948, served in the Korean War and came to NCI in 1955. Frei and Freireich worked together to help develop the first combination chemotherapy for the cure of acute lymphoblastic leukemia in children. 1965 they 18 K. W.Kohn Drugs Against Cancer INTRODUCTION both moved to the M. D. Anderson Cancer Center in Huston, Texas, and continued their work there. In 1972, Frei moved to Boston to become physician-in-chief at the Dana-Farber Cancer Institute of Harvard University. Figure 0.5. C. Gordon Zubrod (1914-1999) was clinical and research director of the National Cancer Institute (NCI) from 1956 until 1974, when he moved to direct the oncology program at the University of Miami Medical School and the Florida Comprehensive Cancer Center. He received an MD degree at Columbia College of Physicians and Surgeons in 1940. During World War II, he worked at Goldwater Memorial Hospital to find a replacement for quinine for the treatment of malaria. The sources of quinine had been cut off during the war. That work resulted in the development of chloroquine as a substitute for quinine. Figure 0.6. David P. Rall (1926-1999) as I remember him in the 1960's. After an MS degree in Pharmacology, he went on to receive MD and PhD degrees from Northwestern University Medical School in 1951. He interned at Bellevue Hospital 19 K. W. Kohn Drugs Against Cancer INTRODUCTION in New York and then joined the National Cancer Institute in 1954 as a Medical Officer in the U. S. Public Health Service. At NCI, he was a Senior Scientist and later Chief of the Clinical Pharmacology Service, which later became the Laboratory of Chemical Pharmacology. His early studies of the blood-brain barrier were carried out together with Gordon Zubrod and aimed to eliminate leukemia cells lurking in the brain which most anticancer drugs could not penetrate into. Their experimental animal was the dogfish shark whose brain anatomy suited those studies, which they carried out in Maine and Bimini. He had prominent roles in investigations of the toxicity of anticancer drugs and on the screening of large numbers of compounds for anticancer activity. In 1971, he left the NCI and became Director of a new National Institute of Environmental Health Sciences (NIEHS) in North Carolina. He had prominent roles in governmental and international policy on environmental carcinogens until 1999, when he died as the result of a car accident in France. In his memory, the main NIEHS building was named after him. Figure 0.7. Ti Li Loo (1918-2007) completed his undergraduate studies in China and then in 1943 began his studies of chemistry and pharmacology at Oxford University. After post-doctoral work at the University of Maryland and acquiring United States citizenship, he was recruited by Gordon Zubrod to the NCI in 1955 and was assigned to the Clinical Pharmacology Service, where I learned much from him. In 1965, he moved to the M. D. Anderson Cancer Center and continued his research activities there ( Loo Ti Li Oral History 1998). References DeVita Jr., V.T. (2015). The Death of Cancer (New York: Farrar, Straus and Giroux). Gilman, A., and Philips, F.S. (1946). The biological actions and therapeutic applications of the B-chloroethyl amines and sulfides. Science 103, 409-415. Kohn, K.W. (1961a). Determination of tetracyclines by extraction of fluorescent complexes. Anal Chem 33, 862-866. Kohn, K.W. (1961b). Mediation of divalent metal ions in the binding of tetracycline to macromolecules. Nature 191, 1156-1158. 20 K. W. Kohn Drugs Against Cancer INTRODUCTION Kohn, K.W., Spears, C.L., and Doty, P. (1966). Inter-strand crosslinking of DNA by nitrogen mustard. Journal of molecular biology 19, 266-288. Laszlo, J. (1995). The Cure of Childhood Leukemia: into the Age of Miracles. (New Brunswick, New Jersey, USA: Rutgers University Press). Udenfriend, S. (1995). Development of the spectrophotofluorometer and its commercialization. Protein Sci 4, 542-551. 21 K.W. Kohn Drugs Against Cancer CHAPTER 1 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@nih.gov CHAPTER 1 Anti-cancer drugs that crosslink DNA. The Story of Nitrogen Mustard: From poison gas to anti- cancer drug. ".. . and they shall beat their swords into plowshares, and their spears into pruning hooks: nation shall not lift up sword against nation, neither shall they learn war anymore." -- lsaia 2:3-4 lsaia's words may not have been the inspiration for how it happened that a poison war gas was "beaten" into an anti-cancer drug, but that nevertheless is what happened. It came about by way of a wartime disaster that was part design and part accident or coincidence. It came about, as it were, "from out of the blue." Nor did the admonishment in lsaia's words come to pass, as humans went on to use science to devise ever mightier swords, and one terrible war led to another. But few would have imagined that a poison war gas would lead to some of the most useful drugs for cancer therapy. On December 2, 1943, in the evening, in the Adriatic harbor of Bari on the Eastern coast of Southern Italy, a still little-known military disaster took place. Some in the United States Navy and Merchant Marine called this World War II event "The Little Pearl Harbor" (Figures 1.1-1.2) (Infield 1971: Remjnjck 2001). The harbor at Bari was filled that morning with ships waiting to unload their military cargo to supply the Allied push up the Italian boot. No one in the base was 22 K.W. Kohn Drugs Against Cancer CHAPTER 1 aware that Nazi Luftwaffe bombers were at that moment approaching from the East. The Nazi high command under the direction of General Albert Kesselring had decided that their best chance to slow the Allied advance was to put the Bari harbor out of commission by sinking as many ships as possible while the harbor was crowded with them. The Luftwaffe was by that time pretty much decimated, but General Kesselring was able to assemble enough bombers for a surprise attack. Figure 1.1. The Italian port of Bari, where the Nazi German attack, known as "the little Pearl Harbor," took place on December 3, 1943. Figure 1.2. German Junkers JUBB bombers (top). Bombed ships exploding in the Bari harbor (bottom). (Source: HistoricWings.com) 23 K.W.Kohn Drugs Against Cancer CHAPTER 1 And surprise indeed it was. The Allies had discounted the remaining capacity of the Luftwaffe to the extent that they kept the harbor lights on all night to speed the unloading of the ships, didn't make sure their radar was working, and ignored occasional German reconnaissance planes. First to note that something was happening that fateful evening were sailors of the Merchant Marine who saw strange strips of metal foil landing on the decks of their ships (Remjnjck 2001). High flying German planes, in advance of the low flying bombers, were dropping metal foil strips to evade and confuse the unbeknownst to them non-functioning Allied radar. What followed was terrible. Some of the ships blown up were carrying munitions and oil, and the harbor became covered with burning oil in which many sailors were immersed and desperately looking for a way to survive. About 1000 military and merchant marine personnel and about 1000 civilians are estimated to have perished, and 28 of the 30 ships that were in the harbor at the time were sunk or destroyed. The sailors who made it to shore were covered with oil and didn't think it urgent to change clothes or bathe. During that part of the war, Washington was worried that Nazi Germany might in desperation resort to poison gas. To meet that threat, they made it known that they would retaliate in kind. To back it up, they secretly dispatched ships to deliver mustard gas bombs to key places - one of which, as you will have guessed, was Bari. One of those ships, the John Harvey, a Liberty Ship, was anchored in the port waiting its turn to unload its terrible cargo. Some who experienced it called the disaster "the little Pearl Harbor". General Dwight Eisenhower said it was the worst setback on his watch, and cancer researchers felt that it launched the first treatment of cancer with a chemical agent What happened was recently clarified and filled out in a well-documented book by Jennet Conant (Conant 2020). I first heard mention of this World War II disaster in a lecture given during the 1960's by Dr. Joseph Burchenal, who participated in the secret mustard gas research during the war and later became a leader in the new field of cancer chemotherapy. I learned nothing more about it during the years that I was studying nitrogen mustard at Harvard and at the National Cancer Institute -- until Glen Infield's little- known 1971 book Disaster at Bari was brought to my attention by an Israeli Physician during a course of lectures I was giving in the NIH evening program (Infield. 1971). Conant's book clarifies and corrects inaccuracies in Infield's book that were due to difficulties he had acquiring reliable information about the event, which remained hidden in a cloud of silence even years after the war. Sailors with burns of various degrees were arriving at overcrowded military hospitals, but many of the burns failed to heal as they should. That, together with rumors of smell of garlic, fed suspicions about poison gas, and led to a call for a chemical warfare specialist to come and investigate. The specialist, Lieutenant 24 K.W. Kohn Drugs Against Cancer CHAPTER 1 Colonel Dr. Stewart F. Alexander (Figure 1.3), soon arrived from North Africa, and was credited with making the connection between mustard gas and a potential anti- cancer drug - although that accolade was ironically to be snatched from him. Infield's book says that the John Harvey's captain asked the British port commander for priority for unloading his ship, but that secrecy prevented him from revealing why, and his urgent request was denied. Conant's investigations, however, reveal the opposite: port commanders knew about the mustard gas bombs in the John Harvey, and it was they who couldn't reveal the secret Had the medical staffs known about the mustard gas dissolved in the surface layer of oil in the harbor, many lives could have been saved by removing the sailors' contaminated clothing. Alexander's investigation was hampered by closely held secrecy by those who knew about the delivery of mustard gas bombs. He was nevertheless able to collect undeniable evidence and even pinpointed the John Harvey as the source of the mustard gas. He did that by making a crude map of where the ships were located in the harbor and plotting on the map how seriously the casualties from each ship were affected (Figure 1.5). Faced with all that evidence, the British authorities had to admit that poison gas was released. After the disaster, an investigating committee advised that saving lives should be more important than secrecy. Examining the patients and their medical records, Alexander was surprised that, after the spike in number of deaths that occurred during the first 4 days due to acute injuries, there was a second wave of deaths a few days later (Figure 1.4). But what really "made the hair at the back of [his] neck stand on end" was that many of the patients who survived the first 3 days then had rapidly falling white blood cell counts and died. He had seen that pattern before: in rabbits, in research he had done at Edgewood Arsenal in 1942. After exposure to nitrogen mustard, the rabbits' white blood cell counts plummeted and their lymph nodes "melted away." All of the Bari casualties, like the rabbits, whose white blood cell counts fell to extremely low levels died. Conant's investigations reveal that German scientists devised nitrogen mustard in their search for a better war gas. It was an improvement over mustard gas in being odorless and devoid of the tell-tale garlic odor, and it was more quickly absorbed through the skin to produce internal injury. Ironically, nitrogen mustard also had an essential advantage as an anticancer drug. In a weak hydrochloric acid solution, it becomes inactive and can safely be injected intravenously. After reaching the blood, it rapidly converts back to its reactive form able to form crosslinks between DNA strands (Kohn et a) 1966). Thus, nitrogen mustard ironically had advantages over mustard gas, both as a war gas and as a therapeutic drug. Conant's book tells how a sample of the new nitrogen mustard was smuggled out of Nazi Germany early in 1942 and immediately analyzed at Edgewood Arsenal, Maryland and studied for its effects on animals. Through great risk, samples of two compounds had been smuggled out of Germany. They were immediately studied by 25 K.W.Kohn Drugs Against Cancer CHAPTER 1 Howard Skipper (Figure 2.10), who found them to be potent blistering agents. Chemical analysis revealed one of them to be a chemical relative of mustard gas, the sulfur atom being replaced by a nitrogen (Figure 1.6). The new compound was named nitrogen mustard. The second compound was like nitrogen mustard, except that it had three chloroethyl groups attached to the nitrogen, instead of just two. Alexander was assigned to study the effects of nitrogen mustard on rabbits. In the course of that work, he made an astounding discovery: nitrogen mustard (but not mustard gas) depleted the rabbits' white blood cells and shrank their lymph nodes. He immediately imagined a possible nitrogen mustard therapy for lymphomas, which are malignant tumors made up of cells that are like white blood cells gone wild. Alexander began the study on April 13, 1942 and reported his findings on June 30, 1942. Several copies of his report were distributed to leading military doctors and academic scientists who were carrying out classified wartime research on the effects of the mustards. Alexander wanted to pursue his concept of nitrogen mustard as possible treatment for lymphomas, but his research proposal was turned down as "not beneficial" to the war effort Researchers at Yale, however, received a copy of Alexander's 1942 report and began investigating the effect of nitrogen mustard on the white blood cell count and lymph nodes in rabbits and other animals. Within a few months, they felt ready to try the drug on a nearly moribund lymphoma patient. They were astonished by a patient's response. His tumors disappeared and he seemed entirely well. But two months later, the tumors reappeared and resisted further treatment. Trial of the drug on other patients, however, were disappointing and further studies languished. When the Yale scientists later received Alexander's report early in 1944 of the effects of mustard gas on the casualties at Bari, as well as samples of their affected tissues, it spurred new attempts to develop nitrogen mustard as an anticancer drug. An apparent discrepancy remained: why did mustard gas suppress the white blood counts of the Bari casualties but not of the experimental animals? The difference was likely due to its very slow penetration through the skin. The mustard gas exposures in World War I and in the experimental animals were short-lived exposures of the skin, causing severe burns and blisters, but the mustard gas exposure was not long enough for much of it to penetrate through the skin. The Bari victims, however, were exposed for many hours in their contaminated clothing - which gave time for the toxic stuff to get into the blood stream. After the war, when Alexander was at last permitted to publish his research, his paper was rejected, because similar results had already been reported by the academic scientists. Eventually however, Alexander was offered the position of assistant director of the new Sloan-Kettering Institute for Cancer Research, but he 26 K.W. Kohn Drugs Against Cancer CHAPTER 1 declined that enticing opportunity because he had promised to join his father's practice of medicine and cardiology in New Jersey. Conant lays out the moral complexities of General Motors president Alfred Sloan who sought to allay public criticism of his industrial ties with Nazi Germany by founding what became known as the Sloan-Kettering Institute for Cancer Research. From 1937 to 1941, GM's Opal subsidiary in Germany was manufacturing war machinery for the Nazis, including the engines for the JU88 bombers of the type that were to decimate Bari harbor. GM's continuing profits from its German subsidiary were severely criticized by the American public and press. Sloan sought to restore his reputation by founding the Cancer Research Institute that bears his name. American scientists invented the name "nitrogen mustard" and dubbed the compound HN2, because it had 2 chloroethyl groups on the nitrogen. The compound with 3 chloroethyl groups on the nitrogen was called HN3. HN2 and HN3 had similar biological activities; wisely, HN2 became the drug preferred for chemotherapy. HNl was a similar compound, but with just one chloroethyl group on the nitrogen; it was therapeutically worthless for a simple reason: with only one chloroethyl group, it could not form crosslinks. In fact, this turned out to be the first evidence that HN2 and HN3 worked by forming crosslinks: two reactive chloroethyl groups on the nitrogen were need for biological and therapeutic activity. Two reactive groups could bind firmly to two biological molecules, thereby forming a crosslink between them. Who deserves the creditfor triggering the chemotherapy on cancer? On April 13, 1942, Alexander began his 2-month-long study in which he led a research group to study the effects of HN2 and HN3 on animals, mainly rabbits. He was amazed by the white blood cell depletion the lymph node shrinkage, leaving "shrunken little shells." This had never before been reported in the scientific literature. Moreover, mustard gas did not have these effects (Conant 2020) - presumably, because in those experiments the mustard gas exposure was not long enough for much of it to penetrate into the internal tissues. Alexander reported those findings on June 30, 1942 in a secret memorandum: Medical Division Edgewood Arsenal MD Memorandum Report 59, Preliminary Report on Hematological Changes in the Rabbit Following Exposure to Lethal Doses of 1130 [codename for HN2] (cited by (Conant. 2020), and the classified report was distributed to leading scientists of the National Research Council. Presumably, the academic clinicians who conducted the first pre-clinical studies of nitrogen mustard on lymphomas would have received that report, which was prepared within 4 months after the compounds were smuggled from Germany. Alexander had wanted to go on to study the effects of nitrogen mustard on lymphomas already in 1942, to see whether the compound would cause those 27 K.W. Kohn Drugs Against Cancer CHAPTER 1 tumors to shrink as he had seen lymph nodes to shrink His proposed study however was not approved, because it did not help the war effort (Conant. 2020). It seems therefore that Alexander was the first to propose the nitrogen mustard treatment of lymphomas. Figure 1.3. Lt. Col. Dr. Stewart Alexander at age 29 was the chemical warfare expert dispatched to Bari, Italy to investigate the suspected poison gas incident consequent to the German bombing of the port on 2 December 1943. His investigation led him to propose nitrogen mustard as treatment for lymphosarcoma. (Source: Jennet Conant, Smithsonian Magazine, September 2020.) 1st dcy 2nd do.y -- 4 9 doaths donths 15th dny 16th doy - - 1 dooths l donths 3rd day - ll do•ths 17th dsy - 0 donths - --- 4th day 8 dccths 18th dn,y 0 docths 5th day 6th day -- 4 4 dc•ths dccths 19th dny 20th day 2 deoths l dooths 7th doy 8th day -- 5 dodhs 9 dorths 21st d:1y 22nd day - -- 0 donths O dosths 9th dny 10th de;r -- 9 dcstha 2 do•ths 23rd dny 24th dcy - 0 dooths l donths 11th dn;r - 2 dc<lths 25th day -- 0 deaths 12th da;r - 4 dontho 26th dr,¥ - l donths 13th da;r 14th dny -- l d •ths l d the 27th dny 28th dny - l deaths 0 duoths Aft,or tho 28th - 2 dsstJla Figure 1.4. Number of deaths on each day after the bombing, listed by Lt Col. Dr Stewart Alexander. A second w ave of deaths occurred on days 8 and 9. (From: Stewart F. Alexander, 1943. "Final Report of Bari Mustard Casualties." Records of the Office of the Surgeon General. National Archives and Records Administration.) 28 K. W. Kohn Drugs Against Cancer CHAPTER 1 7.~ 0a-- == (/=0 -·- r'O ~::' a'4' >~ l cJ,~f: P.. (\) -:r: Qci r::,. [[\0 Cl) r~= = == j C C-· t <.I - \b:~ -J ~ ~ = = i I~ _, ~ ~ ~ ~~ ' I /( S ■ ll ■li Figure 1.5. Lt. Col. Stewart Alexander's drawing of the positions of ships in Bari Harbor on 2 December 1943. (From: Stewart F. Alexander, "Final Report of Bari Mustard Casualties." Records of the Office of the Surgeon General. National Archives and Records Administration.) Cl ~ s~ cI Mustard gas Nitrogen mustard Figure 1.6. Nitrogen mustard was found to be like mustard gas in its chemical mechanism of action, with the sulfur atom replaced by a nitrogen. 29 K.W. Kohn Drugs Against Cancer CHAPTER 1 Mustard gas and the controversy about Fritz Haber Mustard gas was deployed for the first time in 1917 before the third battle at Ypres. !twas developed during World War I by German chemist Fritz Haber (1868-1934) (Figure 1.8) with the idea that it would shorten the war and thus reduce overall casualties. However, it did not shorten the war and only created more misery. Moreover, he was wrong in imagining that poison gas would help Germany win, because the Allies soon countered with their own poison gases and gas masks. Chlorine and phosgene had been used previously by both sides as poison gas, but these could be protected against with gas masks. The idea behind mustard gas was that it might break the stalemate by sinking into the trenches and be absorbed through the skin even if a gas mask covered the face; mustard gas differed from the previously used poison gases in being able to dissolve in the oily substance of skin and produce disabling burns. Most mustard gas casualties survived, but their burns were often terrible, and many became blind (Figure 1.7). There is no record of what may have happened to them later in life; mustard gas, like radiation, causes mutations and cancer (Paoabi et al 2015). Figure 1.7. Left, British soldiers blinded by mustard gas, April 1918 (Wikipedia; trcs.wikispaces.com). Right, blisters caused by mustard gas (reference: medscape.com) Fritz Haber was awarded the Nobel Prize in Chemistry in 1918 for developing the chemical process whereby nitrogen in the atmosphere is used to make ammonia for agricultural fertilizer, which made up for the limited supply from natural sources and saved millions from starvation. However, it also made up for the limited supply of nitrates for the manufacture of explosives and thereby may have prolonged the duration of World War I. The Haber-Bosch process for converting atmospheric 30 K.W.Kohn Drugs Against Cancer CHAPTER 1 nitrogen to ammonia was produced on an industrial scale from 1910 to the present day. Haber was demonized as "the father of gas warfare." He personally supervised the first use of poison gas, chlorine in April 1915. In addition to his belief that it would shorten the war, Haber's enthusiasm for poison gas may have come from wanting to show that a scientist of Jewish descent was loyal to the German war effort (Dnnjkowska and Tnrko, 2011). Haber's Nobel Prize, awarded for developing the Haber-Bosch process, was credited for saving millions from starvation. In the 1920 ceremony presenting the Prize to Haber, however, there was no mention of poison gas, either in the presentation speech or in Haber's acceptance speech. Perhaps the Nobel committee at that time felt it inconclusive whether poison gas was prohibited in warfare. Haber's view was that "in times of peace, a scientist belongs to the world; in times of war, he belongs to his country" and that "death is death, no matter how it is inflicted." A German military point of view at the time was "War is self-defense that knows no rules" (Deimling,.1.2.3.Q.cited by (Dnnjkowska and Turko, 2011)) (Frjedrjch and Hoffmann .2.!Un). Those arguments however may not have persuaded Haber's first wife, Clara lmmerwahr (Figure 1.8). On May 1, 1915, there was a party to celebrate Haber's promotion to captain in recognition of the success of first deployment of chlorine gas, which took place in the battle of Ypres on April 22, 1915. During the party, Clara reportedly had an argument with her husband; some say it was because of her conviction that her husband was misusing science for war. She then left the party and went out into the garden and shot herself in the heart with his revolver. Haber's views also did not satisfy a public whose outcry about the use of poison gas during the war led to the Geneva Protocol of 1925, which banned the use of chemical or biological weapons. Clara lmmerwahr was an outspoken critic of her husband's poison gas work, even to the point of being threatened about disloyalty. However, she had been unhappy in her marriage for several years, possibly depressed, and frustrated at being unable to pursue her scientific career (Friedrich and Hoffmann. 2016). Opinions have become polarized about the immediate reason that she shot herself; the truth may lie in a combination of factors. A German view of the first and second world wars, as well as the period between them, 1914-1945, equated its impact on the country to a "second 30-years' war" (Stern 2012). The Haber-Bosch process has been likened to Janus, the 2-faced figure of Roman legend that presided over war and peace (Figure 1.9): credited with saving millions from starvation, but prolonging the war by providing critically needed nitrate for explosives (Stern 2012). As inventor of the Haber-Bosch process, Fritz Haber also acquired a dual reputation as both benefactor and detractor of human welfare. This incongruity caused Haber's hometown city of 31 K.W. Kohn Drugs Against Cancer CHAPTER 1 Breslau (Wroclaw) to display his portrait upside down among notable figures who stemmed from that city (Figure 1.9) (Dunikowska and Turko 2011). The German military also were two-sided on the use of poison gas: many high- ranking German officers at first detested the use of poison gas (Dnnjkowska and Turko ZQJJ ). Figure 1.8. Left:, Fritz Haber (1868-1934) in his laboratory in 1905 (Bundesarchiv; Wikimedia Commons.) Right, his first wife, Clara lmmerwahr (1870-1915), the first woman to be awarded a doctorate in chemistry in Germany. She committed suicide in May 1915 at the age of 44, some say in dismay of her husband's work on poison gas. She was perhaps true to the meaning of her name: lmmerwahr = always true. 32 K.W. Kohn Drugs Against Cancer CHAPTER 1 Figure 1.9. Left: Roman deity Janus presided over the beginning and ending of conflict, and hence war and peace. Head of Janus, Vatican museum, Rome, from Wikimedia Commons. Right: Haber's portrait, upside down in the Salon Slaski in Breslau (Wroclaw) (Dunjkowska and Turko 2011) expressed the ambivalent opinions about him . (CONTACT WILEY'S PEAMISSK>NS DEPARTMENT ON PERMISSIONSOWILEY£OM 0A USE fl.IE RIGKTSUNK SEAw;E fJY CLICKING ON TIE 'REOUEST P£AMISSK)NS' UNK AOCOMPAtmNG fl.II$ AATICI.E. 'Ml.EV OR AIJTl«)R OWNED £MAGES Mo\Y BE USED FOR NON-00.....al'"L PURPOSES. SUBJECT TO PAOPER CtTATK)NQF THE ARTIQ.E, NJTI«)A..ANO PUBUSHEA.) Nitrogen mustard becomes the first anti-cancer drug. For many years after World War I, it was thought that mustard gas caused burns by reacting with water in the cells to produce hydrochloric acid. It was only during World War II that organic chemists revealed its more sinister chemistry. The work was carried out by scientists in the United States and was top secret. The remarkable results were not made known until 1946, after the war (Gilman and Philips l 946; Goodman et a) 1946) (Figure 1.10). Several of the scientists who worked on mustard gas and nitrogen mustard during World War II, including several of the authors of those 2 landmark papers, became leaders in the war against cancer, in accord with swords into plowshares. 33 K.W. Kohn Drugs Against Cancer CHAPTER 1 SCIEN C E Nitrogen Mustard Therapy n,c Biologb.J Action.t :tnd Thcnpcutk Applic:ittOos of U,e of Methyl-Bis(Beta-Chl0<oethyl)amine Hydrochloride and A che 8,.0doroethyl Aminet 2nd SuHidcs . AlfM C:..U-11\, ~ ~~ .flll( OOlli:t,t'tl!OCII •11,o(,q••~-s,..llbl- .. -(), or V.'6111.0 WU t ~~ ia.•--lo ... - 1,1_......., .....a, .......... - -.• •• ..d .,,_,Jdl &. 1'1lolli.,... b l tift.,...111, S.,C, AU$ ....... .. ~ -.'lo.,""' ..,.it. ...... ., Tri<(Bcta-Chlo,oethyl)amine Hydrochl0<ide f0< Hodgkin's o,,..,.,Lrmphosarcoma, Leukemia and Certain Allied and Miscellaneous Disorder, I.mus.~nun. /11.D.. s.11 t..1.oOir ..u1,.-10o11,. ~t, •· i.. 'l\.tr...... ._....._...._lW"'_ - ................. - . " . . I , ~ - - .. ~~- .. _._... -'""_...,._.,._ ~~- ..,_,.i...--i:..i...~,...t- . ..,,_~ ..... ._.,._ _ ,,~~. la.,......l,l,<,f_do<t ... -4.1. . . . ., . Ml•,.tll M, Wi((l(ce_ M,0 .. Sd t..i..~· ,_.-..._.,,_ __.... ___ 1-"'""• ,_.,_ "' ..._i..wi.. .. ..... " ' ......,;it,.1a\UAI .,...'n.______ (NI, . .J uol..-i,t. , _ _....,• ...,.,_, ~,..,.- Ui<rol- _ _.,.., _,11-., •-.j_,_.., ... ..._..u,w Wilt.a lbmeihrl M.D.• . _ "1"1~ , CxJmr.. M.O.• r-i..,t <>'-. ""--••fl'-(1.-ltt>IM,_,..., _ ___,_., ____ ~•""'••- i.:,p,.,_...... .....,.., ....t-.._....., ~-(fl,..;..,,J _ ""''""~ ,..•.._..,_.,,,.,..w n_.,..,,,, __ . . . . ."U'-•flWr-'"'•(11o,1"'1,l'l)li.. - · · -... .._ . . _ . , _ _ _ _ ~ _ ; , ,_....,,...., _ ..• .,..,.• ., - ;._,...... ,.. ..,.. .., .. _, ~ .. ~bj,rA!fm! Ciln.- Nt6:-ol~.,_,,(II_.UolbJ .... Mlirp~ T, M('Wlfh. ~I.D., Sllti.....o, Figure 1.10. Two landmark papers that ushered in the era of cancer chemotherapy. Both published inl 946, they reported the essential findings of the secret studies of mustard gas and nitrogen mustard that were conducted during the war years. The chemical and pharmacologic findings were reported in the paper in Science by Alfred Gilman and Fred Philips (left); the remarkable clinical results were reported in the journal of the American Medical Association by Louis Goodman and his colleagues (right). Several authors of these papers became leaders in cancer chemotherapy, hematology and pharmacology. The application of the new chemistry of alkylation to the new cancer chemotherapy began in 1942, during the war, under a cloak of secrecy; even the identity of the medications was encoded. But it was not until 1946 that the results of the clinical investigations of mustard gas and its chemical relative, nitrogen mustard, were published (Gilman and Philips 1946: Goodman et a). 1946) (Figure 1.10); the story was further clarified by Alfred Gilman in 1963 (Gilman. 1963). Nitrogen mustard, rather than mustard gas, was used in the biological and clinical studies, because as a crystalline hydrochloride salt it could be freshly dissolved and safely injected. Mustard gas would be very difficult or impossible to use as a drug. but the two substances engage in similar chemical reactions. Goodman, Gilman and their colleagues during the war found that injections of nitrogen mustard in tumor-bearing mice dramatically reduced the size of the tumors and prolonged the survival of the mice. They did extensive tests in animals to determine the nature of the drug's toxicity and to estimate the dose that would be safe in patients. Only then did they try the drug in cancer patients. Most of their patients had large lymphoma tumors that had become resistant to x-ray treatments and who were not expected to survive much longer (Gilman 1963: Gilman and Philips. 1946; Goodman et al., 1946). It is here that Stewart Alexander's information from Bari may have had an impact in putting the focus on lymphomas. And that was a fortunate choice, because those cancers were particularly sensitive to drugs like nitrogen mustard, and the clinical responses and promise of the drug were plain to see. 34 K.W. Kohn Drugs Against Cancer CHAPTER 1 The response of those large tumors to nitrogen mustard must have astounded both physician and patient and given them hope (Figure 1.11) (Goodman et a) 1946. 1984). It was the first time that a chemical agent obliterated a large internal tumor in humans. It was in fact the beginning of the era of cancer chemotherapy. Remarkably. nitrogen mustard was sometimes effective after radiation had failed. The tumor however soon grew back and lost its responsiveness to further treatment. The tumor had become resistant to the nitrogen mustard, as well as to radiation. Nonetheless. the treated patients sometimes lived several months longer than would otherwise have been expected. But the bugaboo of drug resistance was to plague cancer chemotherapy from then on; the reason for the acquired resistance to nitrogen mustard remained enigmatic. Figure 1.11. This 48-year-old lymphoma patient was one of the first whose tumors shrank after treatment with a drug. nitrogen mustard. This famous case was reported inl 946 by Louis Goodman and his colleagues (Goodman et a) 1946). Left, large tumors in armpits. neck and chest as they looked before treatment. Right, after treatment the tumors have disappeared. The full story of this patient and his treatment is told by Vincent DeVita in his book "The Death of Cancer" (DeVita )r., .2.!ll.5.) In the initial clinical trial. 67 patients in the last stages of their disease, most of whom had received radiation treatment that was no longer effective. were treated at New Haven Hospital (L. S. Goodman and A. Gilman). Salt Lake County General Hospital (M. M. Wintrobe and M. T. McLennan). and Tufts College Medical School, Boston (W. Dameshek). All of these authors were to become leaders in the new oncology and hematology. Search for better nitrogen mustards 35 K.W.Kohn Drugs Against Cancer CHAPTER 1 When the response of lymphoma patients became widely known after the war, it was hoped that better results could be achieved, and resistance perhaps avoided with chemically modified nitrogen mustards. The alkylation chemistry of nitrogen mustards had been worked out during the war and was well understood, and the chemical structure of the nitrogen mustard molecule made it relatively easy to synthesize many active modifications. A huge number of modified nitrogen mustards were synthesized and tested in tumor-bearing mice. Despite massive effort, however, none of the modified nitrogen mustards were distinctly superior in animal tests (Shapiro et al.. 1949). Clinical experience, especially the problem of acquired drug resistance, soon revealed the limitations of what could be achieved with nitrogen mustards and related drugs, and clinicians stopped thinking of "cure" in the context of cancer. Unless a malignant tumor could be eliminated before it spread, there was at the time little hope for more than a brief reprieve. Of the large number of nitrogen mustards tested, a few did become part of the chemotherapy armamentarium; these will be considered individually after a review of some to the basic science. Nitrogen mustard may from crosslinks -- but between what? During the war, chemists learned how nitrogen mustards react, which made it possible to understand the chemical behavior of a variety of related compounds, either natural or synthetic products. The essential reaction, called "alkylation," causes the drug molecule to bind firmly (covalently) to biomolecules such as DNA and proteins. Drugs that work by this mechanism are called "alkylating agents." What nearly all effective nitrogen mustard-like alkylating agents have in common is a feature that was noted already during the early nitrogen mustard studies. To be effective, the nitrogen mustard had to have 2 reactive sites; when similar molecules with only 1 reactive site were made and tested, the great majority were found to be inactive. It was as if the anticancer and toxic actions required the formation of a crosslink between 2 other molecules; in other words, the effects required the linking together of 2 target molecules (Figure 1.12) (Golda ere et al 1949; Loveless and Reven 1949). It was not known what those critical target molecules were, and it took more than a decade to find out. 36 K.W.Kohn Drugs Against Cancer CHAPTER 1 Figure 1.12. An early concept of how nitrogen mustard works was that it forms stable (covalent) crosslinks between 2 important biomolecules, A and B. In 1946 it was not known what A and B might be. The only evidence was that a nitrogen mustard needed 2 reactive sites to be effective. I first learned of nitrogen mustard alkylation chemistry and the idea of crosslink production in Alfred Gilman's pharmacology lectures at Columbia's College of Physicians and Surgeons in 1952 (Figure 1.13). That idea lingered in my mind until 1960, when I joined Paul Doty's lab at Harvard. Concepts and methods had by that time been developed to permit me to show that the bifunctional reactions of nitrogen mustard formed crosslinks between the paired strands of DNA and was the main factor that killed cells (Kohn and Green, 1966; Kohn et al.. 1966). Before proceeding to that story, however, it may be helpful to explains some fundamentals about the chemistry of alkylation and crosslink formation. 37 K.W. Kohn Drugs Against Cancer CHAPTER 1 Figure 1.13. Alfred Gilman (1908-1984), Professor of Pharmacology, Columbia College of Physicians and Surgeons, was a key figure in the elucidation of alkylation chemistry as it applies to nitrogen mustards. He taught the pharmacology course while I was a medical student there, which is how I first learned about nitrogen mustard and its reaction mechanism. Alkylation and DNA cross/inking - the chemistry. The essential property of alkylating agents is the ability to form stable bonds with molecules such as DNA and proteins. Since it has such a fundamental role in drug actions, I will explain how alkylation reactions work. I'll try to explain the essentials in a way that those without much chemistry background could understand. It takes quite a few words to do that, but taken one step at a time, it's pretty simple. We'll take nitrogen mustard as a classic example. So, here goes (in the following, refer to Figure 1.14). lf you're familiar with organic chemistry, please just look at Figures 1.14 and 1.15 and skip the rest of this section. 38 K.W. Kohn Drugs Against Cancer CHAPTER 1 R 1.(\ Cl ~ N~ I .. Nu Figure 1.14. Nitrogen mustard (left), showing the unshared electron pair on the nitrogen. The curved arrow shows the unshared electron pair from the N attacking the carbon atom to which the chloride (Cl) is attached. At the same time, the Cl leaves, one electron richer, to become a happy chloride ion. The resulting triangle, consisting of a nitrogen and 2 carbons, is shown on the right The bond angles however like to be about 109°, whereas confined to 60° in the triangle puts them under much stress. The curved arrows show what happens next: an atom with a loose pair of electrons (here designated Nu, which stands for "nucleophile"), such as the nitrogen at the 7-position of guanine in DNA, can come in, and form a bond with a carbon in the triangle. The triangle opens and the stress is relieved. (By organic chemistry convention, a CH2 is assumed to exist at any angle between two lines.) A carbon atom has 4 bonds coming out of it. (If you know about such things, you may object: how about double-bonds, pi-bonds and such? Well, for the present purposes, we don't have to trouble with those cases.) The 4 bonds of a carbon atom like to be directed towards the comers of a regular tetrahedron, the carbon atom being at the center of the tetrahedron. That means that the bonds are most stable when the angles between them is about 109°. If three atoms were connected in an equilateral triangle, the bonds would be forced to be at an angle of 60°, which would put a lot of stress on them. So, what has that to do with forming bonds to DNA? Before we get to that, there is something else to know .. . about nitrogen atoms. Like carbon atoms, they too like to have a tetrahedral configuration of bonds, but they often have only 3 bonds in place; the 4th direction, where a bond could be but isn't, is occupied by "a pair of unshared electrons" (because nitrogen has one more electron out there than carbon does). That unshared pair of electrons has the potential of forming a bond with another atom when there is an opportunity to do so. An opportunity arises when that other atom can create a space for that electron pair to go into. With those ideas in place, let's look at the business end of the nitrogen mustard molecule. The only atoms that we have to be concerned with are -N-C-C-Cl. (To make up the 4 bonds, each carbon atom also has 2 hydrogen atoms bound to it, which are not shown.) 39 K.W. Kohn Drugs Against Cancer CHAPTER 1 Now here is another important idea about alkylation reactions, it's called "the leaving group." In this case, the leaving group is just the chlorine atom (Cl) with an extra electron to make it a happy chloride ion. The main thing about a leaving group, you may not be surprised to learn, is that it tries to leave the molecule. But, and this is the key, when it leaves, it carries with it both electrons in the bond. A leaving group will only work as such if it attracts electrons much more strongly than the atom from which it is leaving. In this case, the Cl has a much stronger affinity for the electrons in the C-CI bond than the C does. However, the Cl can't leave right away, because it would make the C very unhappy: a C cannot tolerate an empty place where there could be a bond. Here is where the N, with its unshared electron pair, comes in; it bends around and takes the place of the Cl; thus, the Cl can make off with the extra electron and becomes a very happy chloride ion. What is left is an N bound to where the Cl was bound before it left ... which creates a triangle of bonds. As already said, a triangle of bonds puts the carbon atoms under stress. That arrangement of 2 C's and lN in a triangle, would very much like to open up; in technical terms, it has a lot of energy in it (Figure 1.14). Now at last we come to how the bond to DNA forms. There is one atom in the base- pairs of DNA that has the greatest possibility of providing an unshared electron pair to relieve the stress in the now activated nitrogen mustard (the 3 atoms in the stressed triangle). That DNA atom is a nitrogen, the so-called N7 atom of the guanines in DNA. What happens is that this "GN7" atom binds to one of the C's in the triangle, releasing the N from the stressed triangle. The stress is relieved, and we are left with a stable bond between the nitrogen mustard atoms and DNA: CI-CH2-CH2-N(CH3)-CH2-CH2-GN7-DNA. Finally, the second CI-CH2-CH2-N(CH3)- part of the nitrogen mustard can engage a GN7 on the opposite DNA strand and carry out the same sequence of reactions described above. The result is a stable crosslink between the 2 DNA stands, linked together through atoms from the nitrogen mustard (Figure 1.15). 1 There are many variations of this theme in the mechanisms of DNA damage and repair. The preceding concepts and explanation can help to understand those different cases. 1I have omitted some details about the charges on the atoms, which however are not needed to understand the essentials. When the chloride ion leaves with the extra electron, it has a negative charge; that leaves behind a positive charge, which resides in the now 4-bonded nitrogen atom in the triangle. When the bond to DNA forms, that positive charge is transferred to the attached guanine. 40 K.W. Kohn Drugs Against Cancer CHAPTER 1 © 0 { N: . (NH N~ NH I z [DNA] Guanine in strand 2 of DNA Guanine in strand 1 of DNA Figure 1.15. A DNA inter-strand crosslink, showing how nitrogen mustard links the 2 DNA strands. The crosslink is between a guanine in one DNA strand and a guanine in the opposite DNA strand. The nitrogen mustard moiety is attached to the nitrogen at position 7 of each guanine. The crosslink prevents the separation of the two strands that must happen when DNA is replicated. The DNA cr oss/inking story To recapitulate, the ability of nitrogen mustard to bind important molecules in the cell had been established in the 1940's, and the alkylation mechanism that brings it about had been elucidated. The observation that 2 reactive groups were needed for potent effects on cells and animals suggested that the drug worked by crosslinking something in the cell (Figure 1.12) (Goldacre et al 1949· Loyeiess and Reven 1949). The question remained: what was the important target that was being crosslinked? In the late 19S0's Paul Doty's laboratory at Harvard had elucidated how the 2 strands of DNA come apart when heated and how they come back together when cooled to let the complementary DNA bases (A-T; G-C) find each other again (Figure 1.16). 41 K.W.Kohn Drugs Against Cancer CHAPTER 1 Figure 1.16. Paul Doty (1920-2011), Mallinckrodt Professor of Biochemistry at Harvard, developed the principles of the DNA helix-coil transition, the process whereby the paired strands of the DNA helix come apart and re-associate. When I joined Doty's lab in 1960 and looked over their most recent data, it was apparent that the DNA strands would remain separated only if heated past a critical temperature where strand separation was complete. If the temperature was a little below that point, the strands would separate partially, but almost instantaneously snap back when cooled; the complementary base pairs would be confined to a small region and could quickly find each other again. Only if the temperature was above that critical point would the strands remain separated. As long as even a small region of the strands remained together, the separated regions of complementary sequences could quickly find each other again. That brought to mind what I had learned from Alfred Gilman in his medical school pharmacology lectures in 1953 about the nitrogen mustard reaction mechanism and that to be effective the molecule had to be able to firmly link 2 sites. It seemed possible that nitrogen mustard linked the DNA strands so that they could not separate completely, in same way as heating to a sub-critical temperature. The new concepts developed in Doty's lab of how complementary DNA strands dissociate and re-associate suggested how we could test the idea that nitrogen mustard forms crosslinks between the 2 strands in the DNA helix, thereby preventing the strands from separating completely. I found that even one crosslink could keep the strands connected and near each other when all the base pairs had dissociated, and the normal base-paired double helix could quickly reassemble when the base-pair-separating conditions were reversed, because all the complementary bases would remain in a small region of space (Kohn et a) 1966). The same year that I joined the Doty lab, in 1960, Brookes and Lawley reported that mustard gas can bind to the nitrogen atom at position 7 of guanine (Brookes and Law)ey J96Q). We thought that nitrogen mustard could do the same. The two 42 K.W.Kohn Drugs Against Cancer CHAPTER 1 alkylating groups on nitrogen mustard could bind to guanines in DNA, and, moreover, if nitrogen mustard's 2 alkylating groups each bound to one of the strands in a DNA double helix, the 2 strands would be crosslinked and unable to separate completely. Lawley and Brookes however reported that the bond between the mustard and the DNA guanine was not very stable. Therefore, instead of using heat to separate the strands, as had been the general practice, I separated them by briefly making the solution alkaline (pH12). After neutralizing the solution, the DNA strands remained separated -- unless they were crosslinked. Thus, normal DNA would end up single- stranded, whereas crosslinked DNA would end up as normal double-stranded helix. We needed a way to measure how much of the DNA remained intact helix after the procedure and how much was separated single strands. The clearest way to make that measurement was by means of the analytical ultracentrifuge (Figure 1.17). Using that remarkable instrument and other physical-chemical methods, I was able to prove that nitrogen mustard indeed crosslinked the DNA and that even a single crosslink would allow the double helix to quickly re-associate (Figure 1.18) (Kohn et a! J966). Figure 1.17. The analytical ultracentrifuge was a mainstay in DNA research from the about 1953 until about 1980. Left, an ultracentrifuge at the University of Connecticut being tended in 1968 by my former college roommate, David Yphantis, who became a leader in the development of the technology (Correia et al.. 2004). Right, this ultacentrifuge may have been the very one I used in the Doty lab in 1960; it has the same sign taped to it. It was sold at auction in 2014 for $105, presumably for its parts (at NIH, in 1968 or so, we bought a new one for about $7000). 43 K.W. Kohn Drugs Against Cancer CHAPTER 1 Normal double• stranded DNA Normal DNA after alkali to separate t he strands Crosslinked DNA DNA t reated wi th a "' crosslinki ng drug and then with alkali. Figure 1.18. Analytical ultracentrifuge tracings showing how I detected and measured crosslinked DNA. The DNA was dissolved in a highly concentrated solution of a cesium (Cs) salt and then centrifuged at a high speed for 48 hours. The heavy Cs atoms tended to move in the direction of centrifugal field, reaching an equilibrium between centrifugal force and back-diffusion and forming a Cs salt concentration density gradient. The critical fact was that double-helical DNA banded at a lower density than single-stranded DNA. The two strands of crosslinked DNA did not separate completely when the pH was raised to 12.0, because the strands were held together at the point where there was even by a single inter-strand crosslink. When the solution was neutralized 2 minutes later, the crosslinked strands quickly found their matching base-pairs, thereby restoring the double-helix. Together with Donald McDonald Green in the Doty lab, I showed that the nitrogen mustard-treated DNA whose base pairs had completely dissociated and then re- associated retained its gene coding ability (genetic transformation activity in bacteria) (Kohn and Green 1966). In addition, we showed that DNA crosslinking by nitrogen mustard determined the sensitivity of bacterial cells to being killed by the drug (Kohn et a) 1965). Later studies in many laboratories established that the ability of human cells to survive treatment with nitrogen mustards and related drugs depends in large part on the cells' ability to repair DNA crosslinks. There is however a different kind of alkylating agents that bind largely to the oxygen at guanine position 6. These drugs differ from the nitrogen mustard-like drugs, (which crosslink between guanines in the 2 strands) in that they produce crosslinks between a guanine and its base-paired cytosine. Importantly, there is a special repair enzyme that prevents those crosslinks from forming. These drugs have unique chemical and biological properties and are the subject of the next chapter. 44 K.W. Kohn Drugs Against Cancer CHAPTER 1 There is yet another class of DNA crosslinking drugs, very important ones in cancer chemotherapy, which have an entirely different chemistry. They are not alkylating agents, but instead use a platinum atom to carry out analogous reactions. The fascinating story of the platinum drugs is the subject of Chapter 3. Alkylating agents in clinical research and practice The first clinical study of nitrogen mustard, already described above, was officially summarized in 1946 by Cornelius P. Rhoades (Rhoades 1946,) . At the time of that report, 160 patients with lymphoma, leukemia, and allied conditions had been treated. Rhoades indicated that the drug was available for experimental purposes only through the National Research Council in cooperation with the Chemical Warfare Service. He summarized information about dosage, side effects, and toxicity, noting that divided doses over several days was safer than injecting a single large dose. The most frequent toxicity was suppression of white cells, anemia, and bleeding tendency due to fall in platelet count, which was to become a well-known toxicity pattern in cancer chemotherapy. It was already suspected that rapidly dividing tissues, whether normal or cancer, were particularly vulnerable to the drug. Cl ~N~ CI ~S~CI Mustard gas HOOC ~ - Cl Chlorambucil " Cl ~N~ H2N-( '=./ '---- COCH Cl Nitrogen mustard Melphalan o r ct C N I P ~ ~ N .......__.,..-. Cl II 0 H Cyclophosphamide Bendamustine Figure 1.19. Chemical structures of nitrogen mustards in clinical use, in comparison with mustard gas (also known as sulfur mustard). 45 K.W.Kohn Drugs Against Cancer CHAPTER 1 The Search for better nitrogen mustards throug h chemical modification The remarkable ability of nitrogen mustard to shrink large lymphoma tumors in some of the patients treated by Goodman and his team in 1942-5 (Goodman et al.. ~ ) inspired medicinal chemists to prepare modified versions of the drug. The easiest modifications to make was to add various chemical groups to the methyl group, a change that would not disturb the ability of the drug to form crosslinks. Of the large number of structures prepared, a few became important in chemotherapy, in particular, chlorambucil, melphalan, and cyclophosphamide (Figure 1.19). Chlorambucil and Melphalan were developed by Alexander Haddow in 1953 at the Chester Beatty Institute in England, and both were for decades in the mainstream of cancer chemotherapy. Chlorambucil found its place in the treatment of chronic lymphatic leukemia (CLL) and chronic myelogenous leukemia (CML), while melphalan for a time became standard treatment for multiple myeloma (Catovsky et a) 2011). Chlorambucil Chlorambucil was one of the first modified nitrogen mustards to become widely used in cancer therapy. Early studies indicated that it was effective, although of course not curative, in the treatment of lymphomas and chronic myelogenous leukemia, and it was thought to have less side-effects than nitrogen mustard (Ge))horn et a) l 956' Krakoff et a) l 958; JJ)tmann et a) l 958). The difference in biological properties caused by the change in chemical structure on going from nitrogen mustard to chlorambucil may largely be due to the negative charge from the carboxyl group on the side chain that replaces the methyl group of nitrogen mustard (Figure 1.19), which may affect the drug's distribution in tissues and ability to enter cells. Chlorambucil became the drug of choice for chronic lymphatic leukemia (CLL). However, the rate of complete response was increased by adding to the treatment regimen an antibody to CD20 (also known as MS4Al). CD20 is a protein that is displayed on the surface of B-type lymphocytes, which is the cell type that is over- produced in CLL; the anti-CD20 antibody helps to kill the CLL cells (Lepretre et al.. .2.Ql.5.). Melphalan Melphalan (L-phenylalanine mustard) was, with chlorambucil, one of the first nitrogen mustard derivative to become a significant part of our chemotherapy armamentarium. The idea behind its synthesis in 1953 was that the L-phenylalanine part of the melphalan molecule would serve as a carrier to steer the mustard 46 K.W.Kohn Drugs Against Cancer CHAPTER 1 warhead into cancer cells. L-phenylalanine is one of the amino acid building blocks that make up proteins. A cell's ability to take up this amino acid from the outside is enhanced by specific transporter channels in the surface membrane; it was hoped that cancer cells would have relatively large numbers of these active transport channels in their surface membranes through which the L-phenylalanine mustard would be taken up.2 The malignant tumor most susceptible to treatment with melphalan was multiple myeloma (Musto and D'Auria, 2007), a disease of antibody-producing white blood cells, plasma cells, that grow wild, invade the bone marrow, dissolve calcium from bone, make bones prone to fractures, and cause bone pain. Before the advent of melphalan, there was no effective therapy. The only available therapy was urethane, which was rarely effective (Hoogstraten et al., 1967). Melphalan, although not by itself curative, prolonged the lives of many patients. First approved for the treatment of multiple myeloma and ovarian cancer, melphalan became part of drug combinations for treatment of a variety of malignancies (Fa)co et a) 2007). Early reports of melphalan as a promising treatment for multiple myeloma appeared in 1964 (Speed et al 1964: Wa)denstroem 1964). In 1968, Raymond Alexanian, Daniel Bergsagel and their colleagues at M. D. Anderson Hospital in Huston, Texas, found that 40% of their patients with multiple myeloma responded to melphalan, which prolonged their lives by more than 2 years (Alexanian et al., .19fill). The addition of prednisone increased responses to 70%, although all of the patients eventually relapsed (Alexanian et al., 1969). Until recently, when additional modalities further improved the therapy, the melphalan-prednisone combination remained standard treatment for multiple myeloma (Falco et al., 2007: Musto and P'Auria 2007). Bendamustine Bendamustine was synthesized in the 1960's in the German Democratic Republic (East Germany) and was commonly used there, although not very much studied. After Germany was reunited, studies eventually showed it to be effective in breast cancer and certain lymphomas, and only partially cross-resistant to other nitrogen mustards. In some chemotherapy combinations it was used in place of cyclophosphamide, sometimes giving less toxicity and longer time before progression of the disease (Herold et a) 2006' Ka)aycjo, 2009; yon Mjnckwjtz et a) 2005). Unlike cyclophosphamide, it does not require activation in the liver. 2 The "L" in L-phenylalanine indicates which one of the two possible mirror image structures the molecule has. All natural amino acids of proteins have the L- configuration, and the active transport channels only take up this form of the amino acid or ofmelphalan. The "ef' in me/phalan emphasizes that the drug molecule has the "L" configuration. 47 K.W. Kohn Drugs Against Cancer CHAPTER 1 Cyclophosphamide Cyclophosphamide is a modified nitrogen mustard that became one of the most commonly used chemotherapy drugs and is on the World Health Organization's List of Essential Medicines. Recently, however, its use has been declining as other drugs have begun to replace it The original concept behind the development of cyclophosphamide was that its phosphate moiety would tend to draw electrons away from the nitrogen of the mustard moiety and thereby prevent that nitrogen from releasing the chloride to form the 3-membered ring of an active nitrogen mustard. The drug would then remain inactive until, it was thought, the bond between the P and N would be cleaved by an enzyme thought to be present at high levels in cancer cells. That idea, it turned out, was only partly correct The drug is inactive and requires activation, as predicted. But this activation does not occur in the cancer cell; it occurs in the liver and does not involve cleavage of the P-N bond. A series of chemical steps (which involves removal of the 3 carbons in the ring containing the phosphorous atom (Figure 1.19)) yield the active form of the drug. called phosphoramide mustard, which then crosslinks DNA in both tumor and normal cells (Dong et al..1995). The activation of the drug depends on liver enzymes whose activity can vary from patient to patient, which might make the drug effect delivered by a given administered dose inconsistent (Madondo et al., 2016). Some of the first careful studies of the effects of cyclophosphamide on leukemia in mice were carried out in 1958 by Montague ("Monty") Lane, with the technical assistance of Sidney Yancey, in the former Clinical Pharmacology Service of the General Medicine Branch of the National Cancer Institute, while I was a member of that group upon coming to NIH in 1957. They found that, at the optimum dose, cyclophosphamide greatly extended the life span of the leukemic mice and was more effective than previous drugs, including nitrogen mustard (Lane 1959). Some of the first clinical observations on the toxicological effects of various doses of cyclophosphamide in cancer patients were reported from the University of Pennsylvania by Peter Coggins and his coworkers. The drug seemed to be less toxic than nitrogen mustard and produced partial regression of tumors in many of the 130 patients with measurable tumors of various kinds in the study. Although it was a preliminary uncontrolled study, the investigators felt that the drug produced better results than what was previously available (Kovacs et a) 1960). Early experience put cyclophosphamide on the road to becoming one of the most commonly used drugs in cancer chemotherapy. A lingering question however was the role of the liver activation that the drug required. In what way was that helpful, or did it produce variability among patients, depending on the activity of their liver enzymes? An important advantage however was that the drug could be given orally; 48 K.W. Kohn Drugs Against Cancer CHAPTER 1 once absorbed from the intestinal tract, it passed directly to the liver, where it was activated. Chemical activation of cyclophosphamide, however, produced a toxic by-product: acrolein (Madondo et a) 2016). This situation of a drug that, when activated, generates two different reactive compounds, one of which only adds to toxicity, was seen also with the nitrosoureas, which will come up in the next chapter. An intriguing result of recent investigations is that cyclophosphamide may potentiate the anti-cancer immune system. The immune system's cytotoxic T-cells are part of a surveillance system that can eliminate small nests of cancer cells before they grow into tumors. They also attack developed tumors but are held in check by so-called Treg cells that normally function to prevent cytotoxic T-cells from attacking normal tissues. Tumors can stimulate the proliferation of Treg cells in their neighborhood, which reduces the ability of the cytotoxic T-cells to attack the tumor. The exciting new findings are that regular treatments with low non-toxic doses of cyclophosphamide can directly or indirectly inhibit Treg cells, which would free the immune system to mount a stronger attack on the cancer (Madon do et a) .2.0.1.n). The Mitomycin C story Mitomycin C is produced by certain microorganisms for the purpose of biochemical warfare in nature. It crosslinks DNA by way of a much more complicated chemistry than nitrogen mustards (Figure 1.20). Although not in the nitrogen mustard class, it alkylates DNA guanines at the 7 position and goes on the form inter-strand crosslinks. I once heard Waclaw Szybalski, who discovered the DNA crosslinking activity of mitomycin C (Iyer and Szyha)skj 1963) (Iyer and Szyha)skj 1964). aptly describe the molecule as "bristling with reactive groups," a phrase that was especially effective when delivered with his sharp Polish accent (you may not think of Polish as being "sharp," but the way he rolled his r's for emphasis in that phrase was striking). He told me the story of the discovery this way: He was using the analytical ultracentrifuge to study the breakage of DNA in bacteria when they are deprived of the essential DNA building block, thymine. V. N. Iyer had just joined the lab, and Szybalski asked him to do a simple control experiment to get some experience with the analytical ultracentrifuge. A control experiment was needed to check whether the DNA breakdown was merely a consequence of the DNA synthesis inhibition caused by thymine deprivation. So, he looked around the lab to see what DNA synthesis inhibitor he happened to have on the shelf and found a vial of the known DNA synthesis inhibitor, mitomycin C. 49 K.W. Kohn Drugs Against Cancer CHAPTER 1 Now, in order check on DNA strand breakage, it was necessary first to separate the DNA strands, because the intact DNA helix would hold the whole structure together and hide the breaks. They separated the DNA strands by heating the solution to near boiling (as described above in the context of our findings in Paul Doty's lab). The strands then normally stay separated after quick cooling, because the complementary strands then cannot find each other again. The result of the first experiment, however, was strange: the strands of the heated DNA did not separate. Szybalski thought, well, Iyer must not have heated the solution to a high enough temperature. But repeated careful experiments always gave the same result: mitomycin prevented the strands from separating. Then the light dawned: mitomycin prevented the DNA strands from separating. because it produced crosslinks between them! 0 11 N Figure 1.20. Mitomycin C, an alkylating agent and DNA crosslinker unrelated to the nitrogen mustards. Its chemistry is complicated. It is activated in the cell by reduction of the quinone moiety (adding a hydrogen atom to each to the double- bonded oxygens on the 6-membered ring). That allows the methoxy group (red encircled) to come off and create an alkylating center. A key to this reaction is the N that connects between the two 5-membered rings; its unshared electron pair forms a double-bond that allows the methoxy group to leave. Reducing the quinone allows enough negative charge to flow to the N, so that its unshared electron pair can form the double-bond. A second alkylating groups is the 3-membered ring consisting of an N and 2 C's in the upper right, which is analogous to the alkylating group in activated nitrogen mustard. Thus 2 alkylating groups are generated, which together form DNA inter-strand crosslinks. The Psoralen story. 50 K.W. Kohn Drugs Against Cancer CHAPTER 1 Psoralen is, like mitomycin, another natural product capable of forming inter-strand crosslinks in DNA It is produced by many plants, but it can react with DNA only when activated by ultraviolet light (UV). The activation is an electronic excitation that has a brief life-time; therefore, the UV exposure has to be while psoralen is at the site where it is to react. The psoralen story traces back to the treatment ofvitiligo (unpigmented patches of skin) using plants that happened to contain psoralen-like compounds. In Egypt about 4000 years ago, the juice of Ammi majus (Figure 1.21) was rubbed on patches of vitiligo, after which, patients were to go out into the sun. The ancient Egyptians apparently had already noted the combined effect of the plant material and sunlight Then, in the 13 th century, ground seeds of the plant were used to treat vitiligo (Si.di and Bourgeois-Gavardin. 1952) (Lerner et al.. 1953). Psoralen derives its name from Psoralea corylifolia (Figure 1.22), whose seeds contain psoralen among several related compounds; it was included in the Chinese system of traditional medicine. Psoralen is also found in figs, limes, celery, and parsnips. Ammi majus was tested in 1951 by dermatologists in Paris and found to have some benefit; they swabbed the vitiligo areas with solutions of compounds derived from the plant and then exposed the areas to ultraviolet light (Sidi and Ronrgeojs- Gayardjn 1952). They had already determined that Ammi majus contains compounds related to psoralen (Fahmy et al.. 1947) . In 1953, Aaron Lerner and colleagues at the University of Michigan Medical School reported a detailed study of the chemical properties of 8-methoxypsoralen and its use for treatment of vitiligo (Lerner et al.. 1953). They gave this psoralen derivative to patients orally and found it to be non-toxic. They then exposed the vitiligo areas of skin to ultraviolet light, in order to cause the white areas to become pigmented. The effectiveness of this treatment was dramatically shown when a laboratory worker accidentally exposed an area of arm to an alcoholic solution of 8- methoxypsoralen, followed by ultraviolet light (Figure 1.23). So, what does treatment of vitiligo have to do with cancer treatment? There are two parts to the answer. First, psoralen is a flat molecule having the size and shape suitable for binding to DNA by intercalation (Figure 1.24). UV-activated psoralen produces DNA inter-strand crosslinks (Figure 1.25) (Cole 197Q) (Gasparro et al, 1985). The double-bond pattern of the psoralen molecule allows the molecule to absorb a quantum of UV light that elevates an electron orbital to an excited state that makes the molecule reactive. Second, psoralen, together with long-wavelength ultraviolet light (UVA), was found useful for treatment of mycosis fungoides, a malignant lymphoma that is localized, in its early stages, to skin (Gilchrest et a) 1976) (Abel et a) 1981). This was obviously a logical treatment because skin can easily be exposed to ultraviolet light. The treatment was called PUVA for psoralen plus UVA light. The long-wavelength UVA was by itself less damaging than shorter wavelengths of ultraviolet light, or of 51 K.W. Kohn Drugs Against Cancer CHAPTER 1 sunlight Psoralen effectively absorbs UVA, thereby becoming reactive and able to produce DNA inter-strand crosslinks that kill the malignant lymphoma cells in the skin. In later years, the malignancy was found to be of T-lymphocytes, and the term "mycosis fungoides" was dropped in favor of "cutaneous T-cell lymphoma." Most studies of PUVA treatment of the disease reported complete disappearance of tumor in over 80% of patients (Gasparro et al l 965). But it was still difficult to eliminate all of the malignant cells, and the disease usually recurred within a few years. Treatment usually failed if malignant cells had grown deeper below the skin or metastasized to lymph nodes or other tissues. Figure 1.21. Ammi majus Linn. Figure 1.22. Psoralea cory/ifolia, whose seeds contain psoralen, among several related compounds; it is included in the Chinese system of traditional medicine. 52 K.W. Kohn Drugs Against Cancer CHAPTER 1 Psoralen is also found in figs, limes, celery, and parsnips. Psoralen's flat shape and double bonds allow the molecule to bind DNA by intercalation (the subject of Chapter 4). When activated by ultraviolet light, intercalated psoralen can react with and bind to thymines in DNA and form inter-strand crosslinks. Figure 1.23. Hyperpigmented area of the arm of a laboratory worker, whose arm was accidentally exposed to an alcoholic solution of 8-mehtoxypsoralen and then to ultraviolet light (Lerner et a) J953). 0 0 0 Figure 1.24 Chemical structure of psoralen. Its flat shape allows it to bind DNA by intercalation (see Chapter 4). Upon activation by ultraviolet light, the intercalated molecule can react with thymines in the DNA and form inter-strand crosslinks. Thymine Thymine 0 ~ CH, HN/ '-...C-+--+- - - -< I I o"'c'-...N/ T-->,c+----c I H H Psoralen 53 K.W. Kohn Drugs Against Cancer CHAPTER 1 Figure 1.25. Psoralen with crosslink between two thymines, as occurs in DNA after treatment with ultraviolet light (UVA, long wavelength UV). UV excites psoralen to activate a double bond to react with the thymine double bond, forming a cyclobutene connection between the two molecules (4-membered rings in the structure). This can occur at double bonds at both ends of the psoralen molecule (shown at the intersection between the blue and red ovals). Inter-strand crosslinks are produced when the psoralen molecule can reach a thymine on each of the two DNA strands (Gasparro et al l 965). Synopsis It was a long road from the mustard gas war tragedies to the current application of DNA crosslinking drugs in cancer therapy. There was hope, disappointment, and some surprises. Along with those developments, detailed knowledge of the chemistry and molecular biology of these drugs emerged and has continued to grow. This chapter has been about anti-cancer drugs that produce DNA inter-strand crosslinks. Except for the natural product, psoralen, they are all alkylating agents that attack DNA at the guanine-N7 positions. The next chapter will be about alkylating agents that attack DNA at the guanine-O6 position. The chapter following that will be about platinum complexes: drugs that crosslink DNA, but that are not alkylating agents; starting with a surprising discovery, they became some of the most useful DNA drugs for cancer therapy. Refe rences Abel, E.A., Deneau, D.G., Farber, E.M., Price, N.M., and Hoppe, R. T. (1981 ). PUVA treatment of erythrodermic and plaque type mycosis fungoides. J Am Acad Dermatol 4, 423-429. Alexanian, R., Bergsagel, D.E., Migliore, P.J., Vaughn, W.K., and Howe, C.D. (1968). Melphalan therapy for plasma cell myeloma. Blood 31, 1-10. Alexanian, R., Haut, A., Khan, A.U., Lane, M., McKelvey, E.M., Migliore, P.J., Stuckey, W.J., Jr., and Wilson, H.E. (1969). Treatment for multiple myeloma. Combination chemotherapy with different melphalan dose regimens. Jama 208, 1680-1685. Brookes, P., and Lawley, P.O. (1960). The reaction of mustard gas with nucleic acids in vitro and in vivo. The Biochemical journal 77, 478-484. Catovsky, D., Else, M., and Richards, S. (2011). Chlorambucil--still not bad: a reappraisal. Clinical lymphoma, myeloma & leukemia 11 Suppl 1, S2-6. Cole, R.S. (1970). Light-induced cross-linking of DNA in the presence of a furocoumarin (psoralen). Studies with phage lambda, Escherichia coli, and mouse leukemia cells. Biochimica et biophysica acta 217, 30-39. Conant, J. (2020). The Great Secret. The classified World War II disaster that launched the war on cancer., First Edition edn (New York and London: W. W, Norton & Company). 54 K.W. Kohn Drugs Against Cancer CHAPTER 1 Correia, J.J., Johnson, M.L, Laue, T., Stafford, W.F., and Williams, R.C. (2004). Spinning with Dave: David Yphantis's contributions to ultracentrifugation. Biophysical chemistry 108, 23-42. DeVita Jr., V.T. (2015). The Death of Cancer (New York: Farrar, Straus and Giroux). Dong, Q., Barsky, D., Colvin, M.E., Melius, C.F., Ludeman, S.M., Moravek, J.F., Colvin, O.M., Bigner, D.D., Modrich, P., and Friedman, H.S. (1995). A structural basis for a phosphoramide mustard-induced DNA interstrand cross-link at 5 '-d(GAC). Proceedings of the National Academy of Sciences of the United States of America 92, 12170-121 74. Dunikowska, M., and Turko, L. (2011). Fritz Haber: the damned scientist Angew Chem Int Ed Engl 50, 10050-10062. Fahmy, LR., Abushady, H., and et al. (194 7). A crystalline principle from Ammi majus L. Nature 160,468. Falco, P., Bringhen, S., Avonto, I., Gay, F., Morabito, F., Boccadoro, M ., and Palumbo, A. (2007). Melphalan and its role in the management of patients with multiple myeloma. Expert review of anticancer therapy 7, 945-957. Friedrich, B., and Hoffmann, D. (2016). Clara Haber, nee lmmerwahr (1870-1915): Life, Work and Legacy. Z Anorg Alig Chem 642, 437-448. Gasparro, F.P., Chan, G., and Edelson, R.L. (1985). Phototherapy and photopharmacology. Yale J Biol Med 58, 519-534. Gellhorn, A., Hyman, G.A., and Ultmann, J.E. (1956). Chlorambucil in treatment of chronic lymphocytic leukemia and certain lymphomas. J Am Med Assoc 162, 178- 183. Gilchrest, B.A., Parrish, J.A., Tanenbaum, L., Haynes, H.A., and Fitzpatrick, T.B. (1976). Oral methoxsalen photochemotherapy ofmycosis fungoides. Cancer 38, 683-689. Gilman, A. (1963). The initial clinical trial of nitrogen mustard. American journal of surgery 105, 574-578. Gilman, A., and Philips, F.S. (1946). The biological actions and therapeutic applications of the B-chloroethyl amines and sulfides. Science 103, 409-415. Goldacre, R.J., Loveless, A., and Ross, W.C. (1949). Mode of production of chromosome abnormalities by the nitrogen mustards; the possible role of cross- linking. Nature 163, 667-669. Goodman, L.S., Wintrobe, M.M., Dameshek, W., Goodman, M.J., Gilman, A., and McLennan, M.T. (1946). Nitrogen mustard therapy. Use ofmethyl-bis(beta- chloroethyl)amine hydrochloride for Hodgkin's disease, lymphosarcoma, leukemia and certain allied and miscellaneous disorders. J Am Med Assoc 132, 126-132. Goodman, L.S., Wintrobe, M.M., Dameshek, W., Goodman, M.J., Gilman, A., and McLennan, M.T. (1984). Landmark article Sept. 21, 1946: Nitrogen mustard therapy. Use of methyl-bis(beta-chloroethyl)amine hydrochloride and tris(beta- chloroethyl)amine hydrochloride for Hodgkin's disease, lymphosarcoma, leukemia and certain allied and miscellaneous disorders. By Louis S. Goodman, Maxwell M. Wintrobe, William Dameshek, Morton J. Goodman, Alfred Gilman and Margaret T. McLennan. Jama 251, 2255-2261. Herold, M., Schulze, A., Niederwieser, D., Franke, A., Fricke, H.J., Richter, P., Freund, M., lsmer, B., Dachselt, K., Boewer, C., et al. (2006). Bendamustine, vincristine and 55 K.W. Kohn Drugs Against Cancer CHAPTER 1 prednisone (BOP) versus cyclophosphamide, vincristine and prednisone (COP) in advanced indolent non-Hodgkin's lymphoma and mantle cell lymphoma: results of a randomised phase III trial (OSHO# 19). Journal of cancer research and clinical oncology 132, 105-112. Hoogstraten, B., Sheehe, P.R., Cuttner, J., Cooper, T., Kyle, R.A., Oberfield, R.A., Townsend, S.R., Harley, J.B., Hayes, D.M., Costa, G., et al. (1967). Melphalan in multiple myeloma. Blood 30, 74-83. Infield, G.B. (1971). Disaster at Bari (New York: Macmillan). Iyer, V.N., and Szybalski, W. (1963). A Molecular Mechanism of Mitomycin Action: Linking of Complementary DNA Strands. Proceedings of the National Academy of Sciences of the United States of America 50, 355-362. Iyer, V.N., and Szybalski, W. (1964). Mitomycins and Porfiromycin: Chemical Mechanism of Activation and Cross-Linking of DNA. Science 145, 55-58. Kalaycio, M. (2009). Bendamustine: a new look at an old drug. Cancer 115, 473-479. Kohn, K.W., and Green, D.M. (1966). Transforming activity of nitrogen mustard- crosslinked DNA. Journal of molecular biology 19, 289-302. Kohn, K.W., Spears, C.L., and Doty, P. (1966). Inter-strand crosslinking of DNA by nitrogen mustard. Journal of molecular biology 19, 266-288. Kohn, K.W., Steigbigel, N.H., and Spears, C.L. (1965). Cross-linking and repair of DNA in sensitive and resistant strains of E. coli treated with nitrogen mustard. Proceedings of the National Academy of Sciences of the United States of America 53, 1154-1161. Kovacs, E.T., Ohno, S., and Kinosita, R. (1960). Studies on cyclophosphamide, antitumor alkylating compound. I. Effect on mouse leukemias. Journal of the National Cancer Institute 24, 759-768. Krakoff, I.H., Karnofsky, DA, and Burchenal, J.H. (1958). Remissions induced by chlorambucil in chronic granulocytic leukemia. J Am Med Assoc 166, 629-631. La ne, M. (1959). Some effects of cyclophosphamide (cytoxan) on normal mice and mice with L1210 leukemia. Journal of the National Cancer Institute 23, 1347- 1359. Lepretre, S., Dartigeas, C., Feugier, P., Marty, M., and Salles, G. (2015). Systematic review of the recent evidence for the efficacy and safety of chlorambucil in the treatment of B-cell malignancies. Leukemia & lymphoma, 1-14. Lerner, A.B., Denton, C.R., and Fitzpatrick, T.B. (1953). Clinical and experimental studies with 8-methoxypsoralen in vitiligo. J Invest Dermatol 20, 299-314. Loveless, A., and Revell, S. (1949). New evidence on the mode of action of mitotic poisons. Nature 164, 938-944. Madondo, M.T., Quinn, M ., and Plebanski, M. (2016). Low dose cyclophosphamide: Mechanisms ofT cell modulation. Cancer treatment reviews 42, 3-9. Musto, P., and D'Auria, F. (200 7). Melphalan: old and new uses of a still master drug for multiple myeloma. Expert opinion on investigational drugs 16, 1467-1487. Panahi, Y., Gholami, N., Ghojazadeh, M., Moslemi, F., Naghavi-Behzad, M ., Azami- Aghdash, S., Ghaffari, A., and Piri, R. (2015). Complica tions and Carcinogenic Effects of Mus tard Gas - a Systematic Review and Meta-Analysis in Iran. Asian Pacific journal of cancer prevention : APJCP 16, 7567-7573. Reminick, G. (2001). Nightmare at Bari (Glencannon Press). 56 K.W. Kohn Drugs Against Cancer CHAPTER 1 Rhoades, C.P. (1946). Nitrogen mustards in the treatment of neoplastic disease. J Am Med Assoc 131, 656-658. Shapiro, D.M., Goldin, A., and et al. (1949). Cancer chemotherapy; analysis of results of different screening techniques with nitrogen-mustard analogues. Cancer 2, 100-112. Sidi, E., and Bourgeois-Gavardin, J. (1952). The treatment of vitiligo with Ammi Majus Linn; a preliminary note. J Invest Dermatol 18, 391-395. Speed, D.E., Galton, D.A., and Swan, A. (1964). Melphalan in the Treatment of Myelomatosis. British medical journal 1, 1664-1669. Stern, F. (2012). Fritz Haber: flawed greatness of person and country. Angew Chem Int Ed Engl 51, 50-56. Ultmann, J.E., Hyman, G.A., and Gellhorn, A. (1958). Chlorambucil and triethylene thiophosphoramide in the treatment of neoplastic disease. Annals of the New York Academy of Sciences 68, 1007-1016. von Minckwitz, G., Chernozemsky, I., Sirakova, L., Chilingirov, P., Souchon, R, Marschner, N., Kleeberg, U., Tsekov, C., Fritze, D., Thomssen, C., et al. (2005). Bendamustine prolongs progression-free survival in metastatic breast cancer (MBC): a phase Ill prospective, randomized, multicenter trial ofbendamustine hydrochloride, methotrexate and 5-fluorouracil (BMF) versus cyclophosphamide, methotrexate and 5-fluorouracil (CMF) as first-line treatment of MBC. Anti- cancer drugs 16, 871-877. Waldenstroem, J. (1964). Melphalan Therapy in Myelomatosis. British medical journal 1, 859-865. 57 K.W. Kohn Drugs Against Cancer CHAPTER2 Chapur-2 TM t!'mmolomfdt stor y: DN.A·C06 :ilkybllon :Ind ttp:iir ZZ071tJe-n3 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@nih.gov CHAPTER2 The temozolomide story: DNA-G06 alkylation and repair.1 Medical researchers have sought to cure cancers, or at least to arrest cancerous activity in a patient's body, by removing or destroying cancerous tissues or cells. One way of doing this, the most ancient, is through surgery, that is, by cutting away a part of the body in which cancerous cells have proliferated. Another important way was by irradiating with x-rays or other ionizing radiation the area of the body where the cancer was located. A third way, the subject of this book, was by introducing into the patient's body chemicals capable of targeting cancer cells and destroying or disabling them without causing excessively harmful effects to the patient. Chapter 1 told the story of the first successful such effort at what has come to be called "chemotherapy." It was the story of nitrogen mustard and its progeny of bi functional alkylating agents capable of crosslinking DNA. In this chapter, I continue the story with another type of alkylating agents that attack DNA at a different site, do not form crosslinks, and yet became useful anticancer drugs. Although the nitrogen at guanine position 7 is the most readily alkylated site on DNA, some alkylating drugs are potent enough to attack also the oxygen at guanine position 6, where the impact on the cell is much greater, leading to mutations and eventual cell death. As a reminder, "alkylation" means that a chemical group, such as methyl, ethyl, or chloroethyl, becomes bound tightly (covalently) to the atom that is "alkylated", such as the nitrogen atom at position 7 or the oxygen atom at position 6 of guanine (Figure 2.1). Alkylating agents chemically attack DNA and other cell constituents, producing mutations and potentially lethal effects on the cell. Fortunately, normal cells often have greater capacity than cancer cells to repair and 1 Reproducti on of some of t he figures in t his chapter may need publisher's permission. 58 K.W.Kohn Drugs Against Cancer CHAPTER2 recover from the toxic actions of these drugs - which is what makes chemotherapy possible. Guanine Figure 2.1. The alkylating agents, such as nitrogen mustard, discussed in Chapter 1, bind DNA predominantly at guanine-N7 positions. This chapter, however, focuses on alkylating agents that are strong enough to bind to 06 as well as N7 of guanine. The drugs of this chapter also differ in that they do not themselves form crosslinks - although later in the chapter we will encounter a special case where crosslinks do form. But the most useful anticancer drugs in this chapter have their therapeutic effect because of a simple alkylation at guanine-06. The MGMT story and its impact on cancer treatment The alkylating agents discussed in the preceding chapter, as well as the platinum drugs to be discussed in the next chapter, all attack DNA primarily at the nitrogen at guanine N7 (GN7). However, there is another class of alkylating agents, which additionally attack DNA at the oxygen at guanine position 06 (G06). What is special about alkylation at G06 is that it drastically affects the pairing of guanine with cytosine in the DNA double helix (Figures 2.2): alkylation at G06 allows the guanine to pair just as well with thymine as with cytosine -- which is apt to cause a mutation in the cell's DNA. G06 alkylations cause a host of troubles for the cell, as we shall see. Since some chemical carcinogens in the environment can alkylate DNA at guanine- 06 and therefore cause mutations or potentially lethal damage to the cell, a special enzyme, called MGMT (for methylguanine-methyltransferase), has evolved to quickly and efficiently remove such alkylations before they can cause trouble. MGMT simply removes the offending alkylation and restores a normal guanine. Hence, repair by MGMT is error-free, in contrast to most other DNA repair processes, which are prone to making mistakes. As we shall see, some anticancer drugs capable of alkylating guanine-06 positions on DNA take advantage of the fact that the cells of some cancers are deficient in the MGMGT enzyme. One of those anticancer drugs, metazolomide, works by adding a methyl group (methylating) to G06 positions in DNA. Those drugs are particularly effective against cancers that have low levels of MGMT (Figure 2.3). 59 K.W. Kohn Drugs Against Cancer CHAPTER2 Inadequate MGMT can cause a mutation that is an early step in the development of cancer. In colon cancer, for example, the production of MGMT is suppressed (by promoter methylation; see below) in about 40% of cases as an early event in the development of these cancers (Fornaro et al., 2016). As already mentioned, low levels of MGMT in cancer make those cancers vulnerable to drugs that alkylate DNA at GO6. If cancer cells lack adequate amounts of MGMT to remove the alkylation, the consequent DNA damage is apt to kill those cells. Some cancers indeed have low MGMT levels and are therefore sensitive to GO6-alkylating drugs. Herein was an opportunity for therapy targeted to tumors that have low levels of a DNA repair enzyme (Hegi and Stupp, 2015). Figure 2.2. A guanine:-cytosine base-pair in DNA If the 06 position of guanine is alkylated (e.g., methylated), the hydrogen-bonds that holds the G:C base pair together are disrupted. (A hydrogen bond is a weak bond between Hand 0, or between Hand N, indicated by dashed lines.) The O6-alkylated guanine then can base-pair with thymine rather than cytosine. The result, after DNA replication, is that the G:C base pair is replaced by an A:T base pair, which may change an amino acid in a protein. Figure 2.3. Temozolomide alkylates guanine by adding a methyl group (CH3) to the GO6 position. The DNA repair enzyme, MGMT, removes methyl groups, as well as other alkylations, from that position and regenerate normal guanine. Discovery of a deficiency of rep air of 06-methy l-g uanines in DNA. 60 K.W.Kohn Drugs Against Cancer CHAPTER2 The MGMT story began in 1980, with a groundbreaking observation by Rufus Day, a former colleague at the National Cancer Institute (Day et al., 1980a; Day et al., 1980b). His investigation was inspired by the work of Paul Kornblith, a neurosurgeon colleague of ours at NCI, who had found that cells derived from brain tumors from different patients varied greatly in their sensitivity to BCNU/carmustine (a GO6-trageting DNA crosslinking drug that will be discussed later in this chapter) (Kornblith and Szypko, 1978). Following up on that finding. Rufus demonstrated that cells from some cancers were abnormally sensitive to GO6-targeting alkylating agents, because they had a defect in a DNA repair process. In that work, Rufus used an assay based on the fact that DNA repair-deficient cells cannot support the growth of a DNA-damaged adenovirus in the cells. He first treated the adenovirus with a GO6-targeted alkylating agent (a compound that added a methyl group to GO6 of DNA), so that the virus could grow only in cells that could repair the GO6 methylations. Using that assay to identify the cells had the repair deficiency, he showed that cells whose DNA repair system was defective, were unusually sensitive to being killed by GO6-targeted methylating agents. In other words, the cells that could not repair the guanine-O6-methylated virus, could not repair their own DNA either; therefore they died upon treatment with relatively low concentrations of those drugs (Day and Ziolkowski, 1979; Day et al., 1980a; Day etal., 1980b). Rufus surmised correctly that there was a phenotype (a cell type that had particular functional characteristics), which he called Mer for "methylation repair minus." The Mer phenotype made some tumors abnormally sensitive to alkylating agents of the GO6-alkylating type (Day et al., 1980a). The reason he called that phenotype methylation repair deficient was because the agents he used added a methyl groups to O6-guanine on DNA, and the presumed repair involved removal of those methyl groups from DNA guanine. The high drug sensitivity was present only for alkylating agents that targeted GO6 and only to cells that were unable to remove the GO6 alkylations efficiently. In 1983, Dan Yarosh, working with Rufus Day, confirmed that Mer human tumor cells were unable to repair O6-methylguanine in DNA by demethylation (Yarosh et al., 1983). In a companion paper that accompanied Rufus Day's in Nature, Leonard Erickson and I, together with our laboratory colleagues, demonstrated that, after treatment with the GO6-targeted DNA crosslinking drug. chloroethylnitrosourea, the repair- deficient (Mer) cells, were not only consistently killed by low concentrations of the drug. but also sustained higher levels of DNA inter-stand crosslinks (Erickson et al., 1980). That result was confirmed by Eric Sariban, Len Erickson and me for human cell strains derived from glioblastoma tumors (Sariban et al., 1987). (How chloroethylating drugs produced DNA crosslinks, while methylating agents did not, will be explained later in this chapter.) The enzyme that specifically removes alkylations from DNA guanine-O6 sites, as well as the gene that codes for it, were soon identified. The gene was called "MGMr' 61 K.W.Kohn Drugs Against Cancer CHAPTER2 for "O6-methylguanine-methyltransferase," but it (that is, its protein product) removes a variety of GO6 alkylations, not only methyl groups. The MGMT gene was found to be turned off ("silenced") in the sensitive (Mer) cells; they were sensitive, because they could not remove the GO6 alkylations from the DNA. The cause of the MGMT silencing was also soon discovered. The gene was silenced, because the part of the DNA sequence that turns on the MGMT gene had methyl groups attached to it; this methylation is not on guanine; rather, it is a normal gene- regulation process in which cytosines in the vicinity of gene start regions ("promoter region") in DNA are methylated. Thus, when the MGMT gene's promoter region was methylated, little or no MGMT enzyme was produced. Two classes of anti-cancer drugs were found to alkylate guanine-O6 on DNA: (1) chloroethylnitrosoureas, which add chloroethyl groups at GO6 and form crosslinks, and (2) temozolomide and dacarbazine, which add methyl groups at GO6 and do not form crosslinks. These drugs' stories follow. The temozolomide story and the treatment of brain cancer. Temozolomide was the most notable advance in the treatment of the highly malignant brain tumor, glioblastoma, up to the time of this writing (Ajaz et al., 2014; Stupp et al., 2005). This "blockbuster drug" came at the pinnacle of a series of compounds investigated at Aston University in Birmingham, UK, beginning in an antitumor pharmacology group organized by John Hickman and Andy Gesher (Stevens and Newlands, 1993). Figure 2.4. Malcolm Stevens, developer oftemozolomide (Sansome, 2009). (Photograph from Chemistry World, 2009.) 62 K.W.Kohn Drugs Against Cancer CHAPTER2 The temozolomide story began in 1978, when Robert Stone, a PhD student, joined Malcolm Stevens' drug discovery laboratory at Aston University. Stevens' instruction to Stone was brief and open-ended: "make some interesting molecules" (Sanso me, 2009) (Figure 2.4); the modus operandi of the laboratory was to synthesize creative and potentially useful organic compounds. Stevens evidently felt that allowing a talented young mind freedom of action could lead to something out of the box, as indeed it did. Stone was interested in ring compounds with several nitrogens and that had a nitrogen atom at the junction of 2 rings (a so-called bridgehead nitrogen). He had read about a new route to the synthesis of some compounds of that sort. With that start and Stevens' chemical insights, they came up with a brand new 2-ring system (called imidazotetrazinone) that had never been seen before (Sansome, 2009). They knew they were heading into the realm of alkylating agents that had some resemblance to dacarbazine (Figure 2.5), which has 3 nitrogens in a row, although not in a ring. and which was in use for the treatment of melanoma. They were worried however that so many alkylating agents had already been tried and their problems were well known, that such drugs had lost much of their luster. In fact, when they finally came up with temozolomide, despite its remarkable effectiveness against almost all mouse tumors tested, Stevens had difficulty convincing clinical researchers to put it in clinical trial. An advantage that may have helped its acceptance for clinical trial was that, as a pro-drug. it could conveniently and safely be taken by mouth. Moreover, since it was lipid soluble and had a chemical structure that could generate a nitrosourea-like moiety, it was reasonable to test it against the highly lethal glioblastoma brain tumors. Stevens and Stone were not happy with the name, temozolomide that the manufacturer assigned to it, because it gave no hint of its chemical nature or origin. They wanted to call it "azolastone," which would combine "azo" for nitrogen, "Aston" for the name of the University where it was made, and "Stone" for the name of the student who made it. That creative name, however, did not prevail, because the manufacturer feared it could be confused with the name of an antihistamine then in use, and also because an unkind person called it "azo-last-one" (Sansome, 2009). Along the way to temozolomide, a drug (mitozolomide) having a chloroethyl in place of the methyl, and therefore a DNA crosslinker, had been in clinical trial, but was dropped because of excessive toxicity (Stevens and Newlands, 1993). Replacing the choloethyl with a methyl in temozolomide did not seem a promising move but was motivated by its effectiveness in mice. Despite the fragile rationale, it was put in clinical trial, which revealed temozolomide's surprising potential as an antitumor drug. 63 K.W.Kohn Drugs Against Cancer CHAPTER2 How does temozolomide work? Temozolomide was found to be a "pro-drug" that is inactive until converted in the liver to form the active drug (Figure 2.5). Moreover, it was one of the few anti- cancer drugs able to penetrate the "blood-brain barrier" to get into the brain and have access to tumors in the brain. Temozolomide proved so effective that, combined with radiation, it became the standard treatment for glioblastoma (after surgery, where possible) (Stupp et al., 2015). Temozolomide (after activation) was found to methylate guanine-O6 positions in DNA and did not form crosslinks. However, although GO6 methylation could kill cells, it was also noted for producing mutations and cancer. But those nasties took many years to show up, whereas glioblastoma patients, even with the best available therapy, rarely survived that long. A propose, the ancient Greek word pharmakon means both remedy and poison, and this dual pharmacological action applies to most chemotherapy drugs. H , 0 e I C=N= N H Methylation at guanine-06 and N7 Figure 2.5. Temozolomide (upper left) is activated by enzymes in the liver that cleaves of the bonds indicated by red arrows. Dacarbazine (lower left) is activated by a liver enzyme that cleaves of a bond (red arrow) to remove a CH3 group. The activations of both drugs yield the structure on the upper right, which decomposes spontaneously to form a highly reactive molecule potent enough to methylate DNA guanines at 06. 64 K.W. Kohn Drugs Against Cancer CHAPTER2 After Rufus Day, Leonard Erickson and 1, together with our colleagues, had reported that deficiency in MGMT enhanced the response of cancer cells to G06-targeted drugs, such as temozolomide, dacarbazine, and chloroethylnitrosoureas (BCNU/ carmustine and CCNU/lomustine), our findings were confirmed in clinical studies, which were made possible by development of suitable assays (Belanich et al., 1996; Esteller et al., 2001; Esteller et al., 2000b; Hegi et al., 2005) (Figure 2.6), as well as later by using a more precise assay method (Barault et al., 2015) (Figure 2.7). 100 Met hylated 90 (MGMT <ti - 80 BCNU (carmustine) Silenced) ·-> .., (/) 70 > ... C Q) ::J •-.., 60 CJ) <ti - 0. 50 <ti'+- ... 0 40 Q) Unmethylated P< 0 .001 30 o>* (MGMT not - 20 Silenced) 10 0 0 6 12 18 24 30 36 M onths 100 ..-c~ ~ 90 80 10 Temozolomid e :, "' 60 ~ ., > 50 ... 0 0 40 promoter Methyla1ed 2:- MGMT 30 .. ~ .0 20 (MGMT not SIienced) promoter e 0.. 10 P<0.001 0 0 6 12 18 24 30 36 42 Months Figure 2.6. Patients treated for malignant brain tumors (glioblastomas) survived longer if their cancer's MGMT gene was not functioning (d ue to DNA-methylation of the MGMT gene's promoter). The patients were treated with BCNU (carmustine, upper panel (Esteller et al., 2000a)) or temozolomide ((lower panel, (Hegi et al., 2005)). (The BCNU and temozolomide studies cannot be compared with each other, because they were carried out at different times in different universities using different protocols.) (From the New England Journal of Medi cine.) 65 K.W. Kohn Drugs Against Cancer CHAPTER2 -<>- Unmethylated MGMT -<r Methylated MGMT P<0.0001 o-l-~~~~~~.==:;: =: :i;l,..-~~~...:.... 0 6 12 18 24 30 36 42 48 54 60 66 Overall survival (months) Figure 2.7. Increased survival of temozolomide-treated glioblastoma patients whose tumors had low levels of MGMT, compared with those whose tumors had high levels. Unmethylated MGMT gene indicated high MGMT production (black curve); methylated MGMT gene indicated low MGMT production (blue curve). High MGMT prevented the beneficial action of temozolomide. This study confirmed the results in Figure 2.6 by using a more precise method of analysis. (Barault et al., 2015). Temozolomide treatment of brain cancer patients. In 2000, Esteller and coworkers showed that the G06-targeted DNA crosslinking drug, BCNU (carmustine), produced more benefit to glioblastoma patients whose tumors' MGMT genes were inactive due to DNA methylation (Esteller et al., 2000a) (Figure 2.6). In 2005, Monika Hegi and Roger Stupp reported similar results for temozolomide; they found that the MGMT gene was silenced (by promoter methylation) in the tumors of 45% of their malignant glioblastoma patients, and that it was only those patients who benefited from treatment with temozolomide: they lived longer, whereas patients with tumors whose MGMT genes were not silenced had little or no benefit from the drug (Hegi et al., 2005). These studies eventually defined the standard of care for newly diagnosed glioblastomas. In 2015, it was again reported that temozolomide was effective only against glioblastomas whose MGMT genes were silenced (Bara ult et al., 2015; Lombardi et al., 2015) (Figure 2.7). Similar conclusions were reported also for patients with colon cancer (Fornaro et al., 2016; Pietrantonio et al., 2015). However, despite the accumulating evidence for the importance of MGMT status, it was some time before MGMT status was routinely considered in deciding whether a patient's glioblastoma brain cancer was likely to respond to temozolomide. Glioblastoma patients continued to be treated with temozolomide, regardless of their tumor's MGMT status. In 2015, Hegi and Stupp published an article in the New England Journal ofMedicine, asking why that was the case (Hegi and Stupp, 2015). Why were more than half of glioblastoma patients continuing to be treated with a 66 K.W.Kohn Drugs Against Cancer CHAPTER2 drug that the MGMT test indicated would not benefit them? The authors pointed out that, by omitting temozolomide in the treatment of patients with MGMT-active tumors, there would have been an opening to test innovative therapies for those patients who were unlikely to be benefited by temozolomide. Worse, research emphasis on glioblastoma brain cancer continued to aim in the direction of the conventional idea that the main barrier to successful chemotherapy was drug-resistance of the tumor, and that the obvious thing to do was to overcome the cause of the resistance. Drugs were therefore developed to inhibit the MGMT enzyme. The clinical results of combining temozolomide with an MGMT inhibitor were disappointing. which was not at all surprising. because the inhibition of MGMT also sensitized critical normal tissues to the drug. This misguided clinical research direction delayed the opportunity to select the treatment that would be most likely to increase survival and minimize toxicity in glioblastoma patients. (Since I was engaged in the pre-clinical research, a disclosure is needed. I had argued strongly for emphasis on MGMT status and against the use of MGMT inhibitors. But to no avail, perhaps because, despite my efforts, I lacked the ability to make the argument convincing enough. Or perhaps because there was vested interest in the MGMT inhibitors.) Recent findings indicated that MGMT status was important for treatment decisions also for the less malignant gliomas brain tumors, as it was for the highly malignant glioblastomas (Figure 2.8) (Bell et al., 2018). Temozolomide worked only when the cancer had little or no MGMT enzyme that would have prevented the anticancer action of the drug. 100 Methylated 80 "";;; > 60 ·;;: ~ ::, V, ;;; 40 .,> ~ 0 20 0 0 1 2 3 4 s Time After Registration, y Figure 2.8. A recent study showing that low-grade gliomas whose MGMT gene is silenced by DNA methylation respond better to temozolomide plus radiation than do gliomas whose MGMT gene is unmethylated and therefore not silenced. Thus, 67 K.W.Kohn Drugs Against Cancer CHAPTER2 MGMT status was important for these less malignant brain cancers, as well as for the highly malignant glioblastomas (Bell et al., 2018). DNA mismatch r epair and a surprise. If DNA replication encounters a GO6-alkylated guanine, which would often happen in MGMT-deficient cancers, a more sinister DNA repair process comes into play that could paradoxically increase the anti-cancer effect of temozolomide. Called "mismatch repair," it detects and tries to repair places in DNA that are not properly base-paired (DNA mismatch repair is the subject of Chapter 25). In the case of GO6- methylated DNA, however, the repair back-fires and kills more cells than it helps. It has in fact turned out to be an important factor in clinical response to temozolomide. Surprisingly, patients whose tumors had high mismatch repair capacity (indicated by high content of the mismatch repair enzyme, MSH2), survived longer than those whose tumors low in MSH2 (Figure 2.9). The repair process, instead of making the tumors resistant to temozolomide, enhanced the killing of the tumor cells by the drug. This effect was prominent in cancers that were low in MGMT, because those cancers were likely to have persistent O6-alkylated guanines (because the MGMT that would have reversed them was lacking). Thus, the O6-alkylated guanines looked like a base-pair mismatch to the mismatch repair system, which however was often unable to repair them and instead produced more DNA damage. Here is what was surmised to happen in MGMT-deficient temozolomide-treated cancer cells, because of the many persistent O6-methylated guanines in their DNA. When such cells replicate their DNA, the replication machinery would soon encounter an O6-methylated guanine in the DNA template strand it is trying to copy. Because O6-methylguanine can pair with thymine as well as cytosine, the replication machinery often mistakenly inserted a thymine instead of a cytosine in the new DNA strand it was making. The resulting methyl-O6-guanine : thymine base-pair would be detected as a DNA defect by the mismatch repair system, which would proceed to remove and replace a section of one of the strands that included the now mis-paired methyl-guanine or thymine. The replaced strand segment, however, would still be apt to have a methyl-guanine : thymine mis-pair. The next DNA repair machine that came along would then repeat the cycle. This futile repair cycle would continue until it came to the attention of another surveillance system that concluded that it was time to give up trying to repair this mess and signaled the cell to commit suicide by apoptosis (McFaline-Figueroa et al., 2015). When that happens in a tumor cell, it's good news. The surprise was that the mismatch repair system, instead of repairing the problem, assisted in killing the MGMT-deficient temozolomide-treated cancer cell. This phenomenon was found to make itself felt in glioblastoma brain cancer patients (Figure 2.9) 68 K.W.Kohn Drugs Against Cancer CHAPTER2 100 ~ MSH2 low (n = 57) ctS _;::,: 80 ~ MSH2 high (n = 52) C: ::::, 60 -CJ) C Q) u ~ 40 P = 0.0004 Q) 0.. 20 0 0 500 1,000 1,500 Survival (days) Figure 2.9. Temozolomide-treated glioblastoma patients whose tumors had high DNA mismatch repair capacity (blue cuive) paradoxically survived longer than those with tumors low in this repair capability (red cuive). Mismatch repair capacity was gauged by the level of the MSH2 protein in the tumor (McFaline- Figueroa et al., 2015) (see Chapter 25). The paradox was that here was a case where a DNA repair process, instead of helping cells to recover from DNA damage, helped to kill them. The chloroethylnitrosourea story: promise and disappointment One of the most promising leads to come out of the early years of the NCl's anti- cancer drug screening program was the chloroethylnitrosoureas; these drugs aroused strong interest because they were found to be highly effective against tumors in mice and, particularly, because of their unusual effectiveness against tumors in the brain. True to the complexity of their name, however, they were fraught with several concurrent chemical reaction paths, which frustrated efforts to attain a consistent balance between therapeutic and toxic effects. Toxicity tended to be delayed, unpredictable and difficult to manage. Although the chloroethylnitrosoureas were more potent than temozolomide, they were disappointingly not any more effective than temozolomide in the treatment of glioblastoma brain tumors. The story of the rise and decline of the chloroethylnitrosoureas is a good example of how chemistry and therapy interact, although the story may not yet be over. In 1972, Joseph Burchenal and Steven Carter, in their review of available anti-cancer drugs, listed two chloroethylnitrosoureas, BCNU and CCNU (also known as carmustine and lomustine, respectively), as "agents of proven clinical value" (Burchenal and Carter, 1972). It was thought that the ability of choroethylnitrosoureas to crosslink DNA, an ability that temozolomide lacked, would make the former more effective in patients. However, the chloroethylnitrosoureas turned to have less clinical benefit and more toxicity problems than temozolomide. 69 K.W. Kohn Drugs Against Cancer CHAPTER2 This story began in 1960 at the Southern Research Institute in Birmingham, Alabama, with the work of three remarkable cancer researchers, who became noted for many contributions to experimental cancer chemotherapy: Howard E. Skipper, Frank M. Schabel, and John A. Montgomery (Figure 2.8). Howard E. Skipper Frank M. Sdlabel, Jr. John A. Montgomery 1915-2006 1918-1983 1924- Figure 2.10. Leaders in experimental chemotherapy research that led to the development of the chloroethylnitrosoureas BCNU (carmustine) and CCNU (lomustine). Howard Skipper was one of the many researchers and clinicians who were engaged in the mustard gas and nitrogen mustard studies during World War II (Chapter 1), who were eager to apply their new knowledge to cancer, and who became leaders in anti-cancer drug development and cancer chemotherapy. A biochemist by training. Skipper served in the U.S. Army Chemical Warfare Service, which was led by Cornelius P. Rhoads, the organizer of the first anti-cancer trials of nitrogen mustard. Rhoads selected Skipper to start a biochemistry department at the new Southern Research Institute in Birmingham, Alabama, where Skipper established a world- famous experimental cancer research program (Simpson-Herren and Wheeler, 2006). He became well-known for his precise models of cancer cell growth in mice, which were fundamental concepts later used by clinician researchers to design drug dosage scheduling and combinations, including those that led to the cure of childhood acute leukemia and Hodgkin's lymphoma (DeVita Jr., 2015). Frank M. Schabel (1918-1983) worked closely with Skipper at Southern Research Institute to develop important principles of cancer chemotherapy; their names were associated together in some of their most notable contributions. John Montgomery in his 1982 Cain Memorial Award Lecture of the American Association for Cancer 70 K.W.Kohn Drugs Against Cancer CHAPTER2 Research described Schabel as "the most able cancer chemotherapist in the world" (Montgomery, 1982). Dr. Schabel's untimely death while at the helm of cancer chemotherapy research was an unfortunate setback. On the morning of August 30, 1983, Dr. Schabel had taken his place in the front row of a conference room in the Hotburg Palace in Vienna, Austria, at the 13th International Congress on Chemotherapy. He was scheduled to give the second talk that morning. A few minutes before the start of the session, he had a sudden cardiac arrest from which the physicians in the room were unable to revive him (Freireich, 1984). His colleagues and friends were devastated and cancelled the session. I was at a different session at the time and was looking forward to discussing the nitrosourea problem with him, when later that morning I was shocked to hear from a stunned J Freireich what had happened. Frank Schabel's crystal clear analyses, and his --1 would say in the spirit of Vienna, "gemuetlich" -- style of conversation that exuded collegial friendship, were always enjoyable and enlightening. and I deeply regretted his untimely passing. John A. Montgomery joined the Southern Research Institute in 1952 and served as Director of Organic Chemistry Research from 1956 to 1986. He spearheaded the development of several new classes of anti-cancer drugs, including the chloroethylnitrosoureas, and was highly regarded for his opinions and judgment that contributed much to the drug development program of the National Cancer Institute. The chloroethylnitrosourea story dates from 1961, when Howard Skipper and Frank Schabel reported a systematic study of drug effects in mouse leukemia L1210, in which they noted that methylnitrosourea, one of the many compounds they studied, extended the life-span of the leukemia-bearing mice even when the leukemia cells were growing in the brain (Skipper et al., 1961). That was unusual, because few, if any, of the previous promising drugs were able to cross the blood- brain barrier. Therefore, they modified the methylnitrosourea molecule to try to increase its potency while hopefully retaining its activity against tumors in the brain. The most promising of these were compounds that had a chloroethyl group (CICH2CH2-) attached to the nitrosourea moiety (Figure 2.11). The first of that series to be further investigated was BCNU (carmustine). BCNU was made with two chloroethyl groups, because it was originally designed to resemble nitrogen mustard. However, only the chloroethyl attached to the nitroso (-N=O) end of the molecule was important; the other turned out to be irrelevant and was replaced without loss of activity by a non-reactive cyclohexyl group in the next of the series to be investigated, CCNU/lomustine (Figure 2.11). Two years later, in 1963, Schabel and Skipper reported that BCNU had marked activity against L1210 leukemia in mice and that it appeared to be a new type of alkylating agent with an anti-tumor profile different from the nitrogen mustards (Schabel et al., 1963). Particularly encouraging was that the drug, as hoped, was 71 K.W. Kohn Drugs Against Cancer CHAPTER2 effective even when the leukemia cells were inoculated into the brain. That was remarkable because other drugs did not get into the brain and were ineffective against those brain tumors. The researchers realized that BCNU was lipid soluble (that is, it dissolves in fat), and therefore could penetrate the fatty substance of the blood-brain barrier. Interest in chloroethyl-nitrosoureas mounted further when, in 1977, John Montgomery reported that those drugs were highly active against advanced Lewis lung cancer in mice, a tumor that was notoriously resistant to treatment with other drugs; and, most remarkably, some of the mice with advanced Lewis lung tumors were even cured (Montgomery et al., 1977). Because of their remarkable effectiveness against malignant tumors in mice and their ability to cross the blood-brain barrier, chloroethylnitrosoureas, particularly BCNU/carmustine and CCNU/lomustine, were used to treat patients with malignant brain tumors, such as glioblastomas. Their effectiveness, however, was limited by their toxic side effects, which were delayed, unpredictable, and difficult to manage. Therefore, the chloroethylnitrosoureas were largely replaced by the less potent, but more effective, temozolomide, which could be given orally, and whose toxicity was easier to manage. The standard treatment for glioblastoma then became surgery, radiation, and temozolomide. Despite intensive therapy, however, patients generally survived for little more than one year. BCNU, given to patients after relapse, had little benefit (Reithmeier et al., 2010). As an alternative to temozolomide in the treatment of glioblastoma, a 3-drug combination was tried, consisting of CCNU / lomustine plus procarbazine (an early variant of dacarbazine) and vincristine (discussed in Chapter 12), but without notable benefit. CCNU by itself increased survival by no more than a few months and then only in a minority of patients. Adding other drugs, such as procarbazine or vincristine, to the regiment yielded no further benefit The outlook was bleak indeed (Ajaz et al., 2014). Why they failed: too many reaction paths? With such remarkable anti-tumor effect in mice, why did the chloroethylnitrosoureas fail in cancer patients? We still don't know. But it might have been due to the multiple and complicated reactions of which these drugs were capable. Because the drugs were so effective against mouse tumors, enormous effort went into unraveling their chemistry and their mechanism of action, in hope of finding out how to separate their toxicity away from their anti-cancer activity (Habraken et al., 1990; Kohn, 1977, 1981; Li et al., 2003; Ludlum, 1997; Matijasevic et al., 1993; Sariban et al., 1984). The studies pointed to DNA crosslinks, mainly of the inter-strand type, as the major cause of the cell killing. Research therefore focused on bringing to light the chemical reaction paths that led to the crosslinking. 72 K.W. Kohn Drugs Against Cancer CHAPTER2 Chloroethy lnitrosoureas, their complicated reactions, and DNA crosslinking. DNA inter-strand crosslinks were the most likely cause of cell killing by chloroethylnitrosoureas, but there were also several chemical pathways that could damage cells in a variety of ways. The chloroethylnitrosourea molecule is inherently unstable and breaks apart spontaneously into 2 chemically reactive pieces (Figure 2.11). The left half of the molecule forms a powerful alkylating agent (chloroethyldiazohydroxide) that was found, first to bind the 06 position of guanines in DNA, and then go on by way of a peculiar dance (elucidated by David Ludlum and diagrammed in Figure 2.12) to form inter-strand crosslinks. (In addition, there is a lesser reaction path that can contribute to the toxicity of chloroethylnitrosoureas: they can alkylate the nitrogen at position 3 of adenine, forming alkylated adenines, which can be removed by a specific DNA repair enzyme, called alkyladenine glycosylase (Li et al., 2003; Matijasevic et al., 1991).) Before explaining how the crosslink forms, however, a few words about the right half of the cleaved molecule (blue arrows in Figure 2.11), which generates reactive isocyanates that can react with and damage many proteins (Cheng et al., 1972; Montgomery et al., 1967), including some involved in DNA repair (Ali-Osman et al., 1985; Kann, 1978; Kann et al., 197 4; Wheeler et al., 1975). The commonly used BCNU/carmustine and CCNU/lomustine produce these extraneous and potentially harmful" carbamoylation" reactions. Although there were chloroethylnitrosoureas that did not produce isocyanates (Dive et al., 1988), they were not developed, because of clinicians' disillusionment with chloroethylnitrosoureas in general (Kohn, 1981). We return now to the chloroethylnitrosourea reaction pathway leading from attack at guanine-O6 to the production of guanine:cytosine crosslinks (Tong et al., 1982) (Ludlum, 1997). This pathway is important, because it can be blocked in cells that have active MGMT, whereas cancer cells deficient in this enzyme were highly vulnerable to being killed by chloroethylnitrosoureas (Sariban et al., 1987). The reactions via the GO6 alkylation pathway leading to the production of DNA crosslinks between guanine and its paired cytosine are explained in Figures 2.11 and 2.12. In brief, the chloroethyl group (CICH2CH2-) alkylates the guanine-O6 positions in DNA. At that point, the MGMT repair enzyme can remove that chloroethyl group to regenerate a perfectly normal guanine. Competing with that repair reaction, the chloroethyl group that the drug added at guanine-O6 can react with a nitrogen in the guanine ring to produce a new 5-membered ring. The new 5- membered ring then opens and leads to a G-C crosslink. Unless the crosslink is repaired by other DNA repair processes, the cell is likely to die. 73 K.W. Kohn Drugs Against Cancer CHAPTER2 The details of how the crosslink forms, as described by David Ludlum, is shown in Figure 2.13. ;••···········••-. R ./ 0 :···················-I', II ! R ; : H !. BCNU/carmuslfne CCNU/lomustfne -ot2CH20 -cyclohexyl !CICHaCH. • ! ··7 • - N·:....c-N I •, N=o·:, ·················· \j CICH2CH1 - N = NOH O=C= N- R Chloroethydiazohydroxlde Isocyanate CH2CH2CI / \ H N A, N N H NA I~ N N J? Proteins 2 H 2 06-chloroethylguanine N7--chloroethylguanine Figure 2.11. Reactions of the chloroethylnitrosoureas (BCNU/carmustine and CCNU/lomustine. The molecule spontaneously breaks into 2 pieces: chloroethyldiazohydroxide (left branch, red) and an isocyanate (right branch, blue). The former (red) chloroethylates DNA, mainly at guanine-06 and guanine-N7. The latter (blue) binds to proteins and inactivate enzymes, which would likely be a source of toxicity. Chloroethyl- nitrosourea Chloroethyl added at G06 74 K.W.Kohn Drugs Against Cancer CHAPTER2 Figure 2.12. This scheme shows how chloroethylnitrosoureas crosslink between the guanine and cytosine in a DNA base pair and how the repair enzyme, MGMT, prevents that from happening. Chloroethylnitrosourea (top) spontaneously breaks (at red arrow) to form a reactive intermediate (Figure 2.11) that adds a chloroethyl group to guanine-06 (upper right), which then undergoes either of two reactions: (1) MGMT removes the chloroethyl group to regenerate a normal guanine, which would repair the DNA perfectly, or (2) the Cl come off as the C to which it was attached binds to an N in the guanine ring, forming a new 5-membered ring (lower left). That ring is unstable and opens by reacting with the cytosine on the opposite strand to form a crosslink (lower right in the Figure) (Tong et al., 1982). The crosslink is between the G:C base pair, and the extent of the crosslink formation depends on the balance between reactions (1) and (2). 75 K.W.Kohn Drugs Against Cancer CHAPTER2 dC-dG Cross-Link dC-dG Cross-link Fonnation Prevention Step I CENU + HS-AT Step2 Ranoval of CIOl2Clli Group by Alkyltnnafe~ (AT) + Step 3 O.oxyoytldlne Figure 2.13. How chloroethylnitrosoureas form DNA interstrand crosslinks via the G06 pathway, as determined and depicted by David Ludlum (Ludlum, 1997); this is a more detailed view of how the crosslinks for, showing Ludlum's concept of the reaction steps. Step 1: the chloroethylnitrosourea (CENU) adds a CH2CH2CI group to the oxygen at position 6 of a guanine in DNA Step 2: the CH2CH2Cl group can be removed by alkyltransferase (AT, which another name for MGMT), thereby preventing crosslink formation; otherwise, the Cl comes off, and a new 5-membered 76 K.W. Kohn Drugs Against Cancer CHAPTER 2 ring forms on the guanine. Step 3: the transient 5-membered opens as a crosslink forms between the guanine and the cytosine of the base-pair. (R = deoxyribose of the DNA; "HS-AT" in Ludlum's diagram is to indicate the sulfhyryl (SH) group on AT that is the enzyme's reaction site.) A nitrosourea targeted to a specific tissue: streptozotocin Malignant tumors of the insulin-producing islets of the pancreas are rare. Something else rare about them, which makes them of special interest, is that there is a drug that targets this specific tissue. The drug is streptozotocin, a methylnitrosourea connected to a glucose moiety (Figure 2.14). Streptozotocin is made by a microorganism (a Streptomyces mold), which perhaps evolved the strange compound as a biological warfare toxin to kill competitor organisms that would take up the toxin as if it was glucose. The competing organism would take up streptozotocin, thinking it was taking up glucose, but, like a Trojan Horse, the streptozotocin would proceed to methylate the competitor's DNA and kill it Aside from mitomycin (see Chapter 1), streptozotocin is the only other alkylating agent I know of that is made by an organism in nature. Notable also about the tissue selectivity of streptozotocin is that it is almost completely devoid of bone marrow toxicity (Moertel et al., 1977). Glucose NH o==< Nitrosourea ;N@ O= N--········· Streptozotocin Figure 2.14. Streptozotocin consists of a glucose part that targets the drug to insulin- producing islet cells in the pancreas and to tumor cells arising from those islets. The islet cells and the cancers derived from them take up glucose avidly. The cells also take up streptozotocin avidly, because they recognize the glucose part of the drug. Once inside the cell, the nitrosourea part of the drug then methylates guanine-06 positions in DNA, thereby killing the cell. (The CH3 in the nitrosourea part is the methyl group that is transferred to the guanine-06 position in DNA) Insulin-producing pancreatic islet cells take up glucose from the blood in order to regulate the rate of insulin production according to the blood glucose concentration. 77 77 K.W. Kohn Drugs Against Cancer CHAPTER2 The glucose moiety of streptozotocin targets the drug to the islet cells and the methylnitrosourea part of the molecule kills them (Evans-Molina et al., 2007) (Figure 2.14). In fact, streptozotocin causes diabetes by destroying normal islet cells. Some pancreatic islet tumors (about 30% of patients) respond to treatment with streptozotocin (Moertel et al., 1994). The possible relationship between streptozotocin responsiveness and MGMT levels however seems not to have been investigated. The concept of malignant pancreatic islet cell tumors that overproduce insulin ("insulinomas") was expanded to "pancreatic endocrine tumors," because some of those rare tumors produce other hormones than insulin. Streptozotocin in combination with other drugs was used to treat those tumors (Fjallskog et al., 2008; Moertel et al., 1992). Chemists made a more potent variant of the naturally occurring streptozotocin, called chlorozotocin, in which the methyl group (circled red in Figure 2.14) was replaced by a chloroethyl, thereby conferring DNA crosslinking ability. On clinical trial, however, chlorozotocin, although more potent, was no better than streptozotocin at the optimum dose of each drug; hence, chlorozotocin was dropped from further study (Moertel et al., 1992). Final word. Evaluation of the MGMT DNA repair protein became a useful predictor of response to DNA guanine-O6 targeted drugs, which enabled clinicians to avoid administering toxic chemotherapy in patients whose cancers would not respond to it The experience with chloroethylnitrosoureas showed the problems that can arise with therapeutic agents that are highly reactive and that engage in many potentially toxic reactions. Their remarkable ability to cure tumors in mice however points to anti-cancer potential that may not have been fully tapped. Further development of drugs of this class, however, was impeded by disappointment due to the difficult toxicities that were encountered in treated patients. Summary Chapter 1 was about anticancer drugs that react (alkylate) at the N7 position of guanine in DNA The current chapter was about more powerful alkylating drugs that attack also the guanine-O6 position. The most important of these was temozolomide, which became useful, especially in the treatment of brain cancer, because it was one of the few drugs able to penetrates the blood-brain barrier and get into the brain. However, the guanine-O6 (GO6)-alkylating drugs were found to be effective only against cancers that lacked an enzyme (methylguanine-methyltransferase, MGMT) 78 K.W.Kohn Drugs Against Cancer CHAPTER2 that would remove the GO6-alkylations before they could exert their cancer killing effects. Patients whose cancers had active MGMT received little or no benefit from those drugs, including temozolomide. Only patients in whose cancers MGMT genes were suppressed (by DNA methylation, an epigenetic mechanism) benefited from these drugs. Another factor that came into play was the DNA mismatch-repair system that detects and repairs base-pairs that do not match, i.e., base-pairs other than G:C or A:T. The mismatch repair enzymes paradoxically increased the cancer cell killing effect of temozolomide and related drugs against the MGMT-deficient cancers. Thus, the patients who received the most benefit from temozolomide were those whose cancers were both MGMT-deficient and mismatch repair active. In other words, if the mismatch repair system was inactive (due to mutation in one of its enzymes), then the drug was less effective, even against cancers that were MGMT deficient. Therefore, measuring MGMT and mismatch repair enzyme activities helped to predict how effective GO6-alkylating drugs such as temozolomide would be against a cancer in a particular patient Another class of GO6-targetted alkylating agents were the chloroethylnitrosoureas (carmustine (BCNU) and lomustine (CCNU)). These drugs were extraordinarily effective against mouse cancers, but disappointing against human cancers, in large part because their toxicities were difficult to manage. Like, temozolomide, they were most effective against cancers with MGMT deficiency. Unlike temozolomide, however, they were able to produce inter-strand crosslinks in DNA. A problem with chloroethylnitrosoureas, however, was that they engaged in a complicated set of chemical reactions that led to toxicity in addition to therapeutic action. The possibility of modifying these drugs in a manner that would reduce their undesired reactions, however, was not fully explored. It seemed that it might be possible to modify GO6-alkylating drugs in a manner that would allow them to enter certain cancer cells but not normal cells. A drug that suggested that possibility was streptozotocin, which consists of a glucose part linked to a methylnitrosourea part. The glucose part carried the drug into the islet cells of the pancreas and the cancers derived from them. Once inside the cell, the methylnitrosourea part killed them. Consequently, the drug was useful in the treatment of the rare islet cell tumors of the pancreas. Thus, it seemed that chemical modifications of nitrosoureas might lead to new drugs for particular cancer types. References Ajaz, M., Jefferies, S., Brazil, L., Watts, C., and Chalmers, A. (2014). Current and investigational drug strategies for glioblastoma. Clin Oncol (R Coll Radio() 26, 419-430. Ali-Osman, F., Giblin, J., Berger, M., Murphy, M.J., Jr., and Rosenblum, M.L. (1985). Chemical structure of carbamoylating groups and their relationship to bone 79 K.W.Kohn Drugs Against Cancer CHAPTER2 marrow toxicity and antiglioma activity ofbifunctionally alkylating and carbamoylating nitrosoureas. Cancer research 45, 4185-4191. Barault, L., Amatu, A., Bleeker, F.E., Moutinho, C., Falcomata, C., Fiano, V., Cassingena, A., Siravegna, G., Milione, M ., Cassoni, P., et al. (2015). Digital PCR quantification of MGMT methylation refines prediction of clinical benefit from alkylating agents in glioblastoma and metastatic colorectal cancer. Annals of oncology: official journal of the European Society for Medical Oncology/ ESMO 26, 1994-1999. Belanich, M., Pastor, M., Randall, T., Guerra, D., Kibitel, J., Alas, L., Li, B., Citron, M., Wasserman, P., White, A., et al. (1996). Retrospective study of the correlation between the DNA repair p rotein alkyltransferase and survival of brain tumor patients treated with carmustine. Cancer research 56, 783-788. Bell, E.H., Zhang, P., Fisher, B.J., Macdonald, D.R., McElroy, J.P., Lesser, G.J., Fleming, J., Chakraborty, A.R., Liu, Z., Becker, A.P., et al. (2018). Association of MGMT Promoter Methylation Status With Survival Outcomes in Patients With High- Risk Glioma Treated With Radiotherapy and Temozolomide: An Analysis From the NRG Oncology/RTOG 0424 Trial. JAMA Oncol. Burchenal, J.H., and Carter, S.K. (1972). New cancer chemotherapeutic agents. Cancer 30, 1639-1646. Cheng, C.J., Fujimura, S., Grunberger, D., and Weinstein, J.B. (1972). lnteraction ofl- (2-chloroethyl)-3-cyclohexyl-1-nitrosourea (NSC 79037) with nucleic acids and proteins in vivo and in vitro. Cancer research 32, 22-27. Day, RS., 3rd, and Ziolkowski, C.H. (1979). Human brain tumour cell strains with deficient host-cell reactivation of N-methyl-N'-nitro-N-nitrosoguanidine- damaged adenovirus 5. Nature 279, 797-799. Day, RS., 3rd, Ziolkowski, C.H., Scudiero, DA, Meyer, S.A., Lubiniecki, A.S., Girardi, A.J., Galloway, S.M., and Bynum, G.D. (1980a). Defective repair of alkylated DNA by human tumour and SV40-transformed human cell strains. Nature 288, 724-727. Day, R.S., 3rd, Ziolkowski, C.H., Scudiero, D.A., Meyer, S.A., and Mattern, M.R. (1980b). Human tumor cell strains defective in the repair of alkylation damage. Carcinogenesis 1, 21-32. DeVita Jr., V.T. (2015). The Death of Cancer (New York: Farrar, Straus and Giroux). Dive, C., Workman, P., and Watson, J.V. (1988). Inhibition of intracellular esterases by antitumour chloroethylnitrosoureas. Measurement by flow cytometry and correlation with molecular carbamoylation activity. Biochemical pharmacology 37, 3987-3993. Erickson, L.C., Laurent, G., Sharkey, N.A., and Kohn, K.W. (1980). DNA cross-linking and monoadduct repair in nitrosourea-treated human tumour cells. Nature 288, 727-729. Esteller, M., Garcia-Foncillas, J., Andion, E., Goodman, S.N., Hidalgo, O.F., Vanaclocha, V., Baylin, S.B., and Herman, J.G. (2000a). Inactivation of the DNA-repair gene MGMT and the clinical response of gliomas to alkylating agents. The New England journal of medicine 343, 1350-1354. Esteller, M., Risques, RA, Toyota, M., Capella, G., Moreno, V., Peinado, M.A., Baylin, S.B., and Herman, J.G. (2001). Promoter hypermethylation of the DNA repair 80 K.W.Kohn Drugs Against Cancer CHAPTER2 gene O(6)-methylguanine-DNA methyltransferase is associated with the presence of G:C to A:T transition mutations in p53 in human colorectal tumorigenesis. Cancer research 61, 4689-4692. Esteller, M., Toyota, M., Sanchez-Cespedes, M., Capella, G., Peinado, M.A., Watkins, D.N., Issa, J.P., Sidransky, D., Baylin, S.B., and Herman, J.G. (2000b). Inactivation of the DNA repair gene O6-methylguanine-DNA methyltransferase by promoter hypermethylation is associated with G to A mutations in K-ras in colorectal tumorigenesis. Cancer research 60, 2368-2371. Evans-Molina, C., Garmey, J.C., Ketchum, R., Brayman, K.L., Deng, S., and Mirmira, R.G. (2007). Glucose regulation of insulin gene transcription and pre-mRNA processing in human islets. Diabetes 56, 827-835. Fjallskog, M.L., Janson, E.T., Falkmer, U.G., Vatn, M.H., Oberg, K.E., and Eriksson, B.K. (2008). Treatment with combined streptozotocin and liposomal doxorubicin in metastatic endocrine pancreatic tumors. Neuroendocrinology 88, 53-58. Fornaro, L., Vivaldi, C., Caparello, C., Musettini, G., Baldini, E., Masi, G., and Falcone, A. (2016). Pharmacoepigenetics in gastrointestinal tumors: MGMT methylation and beyond. Front Biosci (Elite Ed) 8, 170-180. Freireich, E.J. (1984). A Memorial Issue for Dr. Frank M. Schabel, Jr. Introduction. Cancer 54, 1132-1133. Habraken, Y., Carter, C.A., Kirk, M.C., Riordan, J.M., and Ludlum, D.B. (1990). Formation of N2,3-ethanoguanine in DNA after in vitro treatment with the therapeutic agent, N-(2-chloroethyl)-N' -cyclohexyl-N-nitrosourea. Carcinogenesis 11, 223-228. Hegi, M.E., Diserens, A.C., Gorlia, T., Hamou, M.F., de Tribolet, N., Weller, M., Kros, J.M., Hainfellner, J.A., Mason, W., Mariani, L., et al. (2005). MGMT gene silencing and benefit from temozolomide in glioblastoma. The New England journal of medicine 352, 997-1003. Hegi, M.E., and Stupp, R. (2015). Withholding temozolomide in glioblastoma patients with unmethylated MGMT promoter-still a dilemma? Neuro-oncology 17, 1425-1427. Kann, H.E., Jr. (1978). Comparison of biochemical and biological effects of four nitrosoureas with differing carbamoylating activities. Cancer research 38, 2363-2366. Kann, H.E., Jr., Kohn, K.W., and Lyles, J.M. (1974). Inhibition of DNA repair by the 1,3- bis(2-chloroethyl)-1-nitrosourea breakdown product, 2-chloroethyl isocyanate. Cancer research 34, 398-402. Kohn, K.W. (1977). lnterstrand cross-linking of DNA by 1,3-bis(2-chloroethyl)-1- nitrosourea and other 1-(2-haloethyl)-1-nitrosoureas. Cancer research 37, 1450-1454. Kohn, K.W. (1981 ). Mechanistic approaches to new nitrosourea development Recent results in cancer research Fortschritte der Krebsforschung Progres dans Jes recherches sur le cancer 76, 141-152. Kornblith, P.L., and Szypko, P.E. (1978). Variations in response of human brain tumors to BCNU in vitro. Journal of neurosurgery 48, 580-586. 81 K.W.Kohn Drugs Against Cancer CHAPTER2 Li, Q., Wright, S.E., Matijasevic, Z., Chong. W., Ludlum, D.B., and Volkert, M.R. (2003). The role of human alkyladenine glycosylase in cellular resistance to the chloroethylnitrosoureas. Carcinogenesis 24, 589-593. Lombardi, G., Pace, A., Pasqualetti, F., Rizzato, S., Faedi, M., Anghileri, E., Nicolotto, E., Bazzoli, E., Bellu, L, Villani, V., et al. (2015). Predictors of survival and effect of short (40 Gy) or standard-course (60 Gy) irradiation plus concomitant temozolomide in elderly patients with glioblastoma: a multicenter retrospective study of AINO (Italian Association of Neuro-Oncology). Journal ofneuro-oncology 125, 359-367. Ludlum, D.B. (1997). The chloroethylnitrosoureas: sensitivity and resistance to cancer chemotherapy at the molecular level. Cancer investigation 15, 588-598. Matijasevic, Z., Bodell, W.J., and Ludlum, D.B. (1991). 3-Methyladenine DNA glycosylase activity in a glial cell line sensitive to the haloethylnitrosoureas in comparison with a resistant cell line. Cancer research 51, 1568-1570. Matijasevic, Z., Boosalis, M ., Mackay, W., Samson, L., and Ludlum, D.B. (1993). Protection against chloroethylnitrosourea cytotoxicity by eukaryotic 3- methyladenine DNA glycosylase. Proceedings of the National Academy of Sciences of the United States of America 90, 11855-11859. McFaline-Figueroa, J.L., Braun, C.J., Stanciu, M., Nagel, Z.D., Mazzucato, P., Sangaraju, D., Cerniauskas, E., Barford, K., Vargas, A., Chen, Y., et al. (2015). Minor Changes in Expression of the Mismatch Repair Protein MSH2 Exert a Major Impact on Glioblastoma Response to Temozolomide. Cancer research 75, 3127-3138. Moertel, C.G., Douglas, H.O., Jr., Hanley, J., and Carbone, P.P. (1977). Treatment of advanced adenocarcinoma of the pancreas with combinations of streptozotocin plus 5-fluorouracil and streptozotocin plus cyclophosphamide. Cancer 40, 605-608. Moertel, C.G., Johnson, C.M., McKusick, M.A., Martin, J.K., Jr., Nagorney, D.M., Kvols, L.K., Rubin, J., and Kunselman, S. (1994). The management of patients with advanced carcinoid tumors and islet cell carcinomas. Annals of internal medicine 120, 302-309. Moertel, C.G., Letkopoulo, M., Lipsitz, S., Hahn, R.G., and Klaassen, D. (1992). Streptozocin-doxorubicin, streptozocin-fluorouracil or chlorozotocin in the treatment of advanced islet-cell carcinoma. The New England journal of medicine 326, 519-523. Montgomery, J.A. (1982). Has the well gone dry? The First Cain Memorial Award Lecture. Cancer research 42, 3911-3917. Montgomery, J.A., James, R., Mccaleb, G.S., and Johnston, T.P. (196 7). The modes of decomposition of 1,3-bis(2-chloroethyl)-1-nitrosourea and related compounds. Journal of medicinal chemistry 10, 668-674. Montgomery, J.A., Mccaleb, G.S., Johnston, T.P., Mayo, J.G., and Laster, W.R., Jr. (1977). Inhibition of solid tumors by nitrosoureas. 1. Lewis lung carcinoma. Journal of medicinal chemistry 20, 291-295. Pietrantonio, F., de Braud, F., Milione, M., Maggi, C., Iacovelli, R., Dotti, K.F., Perrone, F., Tamborini, E., Caporale, M., Berenato, R., et al. (2015). Dose-Dense 82 K.W.Kohn Drugs Against Cancer CHAPTER2 Temozolomide in Patients with MGMT-Silenced Chemorefractory Colorectal Cancer. Targeted oncology. Reithmeier, T., Graf, E., Piroth, T., Trippel, M., Pinsker, M.O., and Nikkhah, G. (2010). BCNU for recurrent glioblastoma multiforme: efficacy, toxicity and prognostic factors. BMC cancer 10, 30. Sansome, C. (2009). Temozolomide - birth of a blockbuster. In Chemistry World, pp. 49-51. Sariban, E., Erickson, L.C., and Kohn, K.W. (1984). Effects of carbamoylation on cell survival and DNA repair in normal human embryo cells (IMR-90) treated with various 1-(2-chloroethyl)-1-nitrosoureas. Cancer research 44, 1352-1357. Sariban, E., Kohn, K.W., Zlotogorski, C., Laurent, G., D'lncalci, M., Day, R., 3rd, Smith, B.H., Kornblith, P.L., and Erickson, L.C. (1987). DNA cross-linking responses of human malignant glioma cell strains to chloroethylnitrosoureas, cisplatin, and diaziquone. Cancer research 47, 3988-3994. Schabel, F.M., Jr., Johnston, T.P., Mc, C.G., Montgomery, J.A., Laster, W.R., and Skipper, H.E. (1963). Experimental evaluation of potential anticancer agents VIII. Effects of certain nitrosoureas on intracerebral L1210 leukemia. Cancer research 23, 725-733. Simpson-Herren, L., and Wheeler, G.P. (2006). Howard Earle Skipper: in memoriam (1915-2006). Cancer research 66, 12035-12036. Skipper, H.E., Schabel, F.M., Jr., Trader, M .W., and Thomson, J. R. (1961). Experimental evaluation of potential anticancer agents. VI. Anatomical distribution of leukemic cells and failure of chemotherapy. Cancer research 21, 1154-1164. Stevens, M.F., and Newlands, E.S. (1993). From triazines and triazenes to temozolomide. Eur J Cancer 29A, 1045-1047. Stupp, R., Mason, W.P., van den Bent, M.J., Weller, M., Fisher, B., Taphoorn, M.J., Belanger, K., Brandes, A.A., Marosi, C., Bogdahn, U., et al. (2005). Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. The New England journal of medicine 352, 987-996. Stupp, R., Taillibert, S., Kanner, A.A., Kesari, S., Steinberg, D.M., Toms, S.A., Taylor, L.P., Lieberman, F., Silvani, A., Fink, K.L., et al. (2015). Maintenance Therapy With Tumor-Treating Fields Plus Temozolomide vs Temozolomide Alone for Glioblastoma: A Randomized Clinical Trial. Jama 314, 2535-2543. Tong, W.P., Kirk, M.C., and Ludlum, D.B. (1982). Formation of the cross-link 1-(N3- deoxycytidyl),2-(Nl-deoxyguanosinyl]ethane in DNA treated with N,N'-bis(2- chloroethyl)-N-nitrosourea. Cancer research 42, 3102-3105. Wheeler, G.P., Bowdon, B.J., and Struck, R.F. (1975). Carbamoylation of amino acid, peptides, and proteins by nitrosoureas. Cancer research 35, 297 4-2984. Yarosh, D.B., Foote, R.S., Mitra, S., and Day, R.S., 3rd (1983). Repair ofO6- methylguanine in DNA by demethylation is lacking in Mer- human tumor cell strains. Carcinogenesis 4, 199-205. 83 K. W. Kohn Drugs Against Cancer CHAPTER3 a»p:u37lllf f'todn11111 Sroryn100fb&l.doot Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@nih.gov CHAPTER3 The Platinum Story: From Imagination to a New Anticancer Drug. The previous 2 chapters were about alkylating agents: anticancer drugs that damage DNA by binding tightly (covalently) to DNA bases, especially to guanine. Surprisingly, it turned out that certain molecules centered on a platinum atom can bind and damage DNA in a manner that is rather similar to that of alkylating agents, again especially by attacking DNA at guanines. The anticancer activity of platinum complexes was one of the most surprising and far-reaching discoveries in all of anti- cancer drug research. Particularly remarkable is how that landmark discovery was made. The first and structurally simplest of the platinum complexes to be discovered was cisplatin, which, with its modified forms, became a mainstay of cancer chemotherapy. Cisplatin would not have been discovered in the drug screening programs, because it is an inorganic chemical, while all cancer drug research had been in the realm of organic chemistry, which is based on carbon atoms. Cisplatin is made up entirely of an atom of the heavy metal, platinum, 2 chlorine atoms, 2 nitrogen atoms, and a few hydrogens; there is not a single carbon atom in it (Figure 3.1). Nor would it have been discovered by searching natural products made by animals, plants, fungi, or microorganisms, because platinum has not been found in any natural biological system. Even if heavy metal complexes had been screened for anti-cancer activity, cisplatin could easily have been missed, because the atoms and their configuration have to be just right For example, cisplatin and transplatin consist of the same atoms and bonds, differing only in whether the 2 chlorines are next to each other (cis) or across 84 K.W.Kohn Drugs Against Cancer CHAPTER3 from each other (trans), yet only the cis configuration has anti-cancer activity (Figure 3.1). cisplatin transplatin Figure 3.1. Chemical structures of cisplatin and transplatin. The 2 chlorides (Cl·) and 2 ammonias (NH3) are arranged in a plane around the platinum (Pt..) atom. The chlorides are next to each other (cis) in cisplatin and across from each other (trans) in transplatin. The platinum atom has 2 positive charges, while the chlorides have one negative charge each; therefore. these platinum complexes are electrically neutral, which allows them to enter cells easily. Both cisplatin and transplatin bind to DNA at guanine-N7 positions. But only cisplatin has the geometry to form DNA crosslinks, and only cisplatin was effective as an anticancer drug. Discovery by imagination The clue to the discovery of cisplatin came from an accidental and seemingly bizarre observation by an imaginative and persistent investigator. As noted by Pestko (Petsko, 2002): "cisplatin came from outside the box - so far outside that the box wasn't even visible; it came from a place no one would have dreamt of looking in for an anticancer drug". Also relevant to the story is a remark by Albert Einstein: "Imagination can be even more important than knowledge." Barnett Rosenberg-- his friends called him "Barney") (Figure 3.2) -- was a biophysicist, working in a small laboratory at Michigan State University with just one assistant He had graduated from Brooklyn College in 1948 and obtained a PhD in physics from New York University in 1956. As a biophysicist, a natural way to combine biology and physics in those early days was to examine the effects of electric current on the growth of bacteria; the techniques for such a study were straight forward and easily set up in a modest laboratory. There was little reason to expect any interesting findings. 85 K.W.Kohn Drugs Against Cancer CHAPTER3 Figure 3.2. Barnett ("Barney") Rosenberg (1926-2009), discoverer of cisplatin. (Picture from Wikipedia. Source: General Motors Cancer Research Foundation.) Here is what happened ((Rosenberg et al., 1965), and from what he told me). Barney had a culture of E. coli bacteria growing in a growth medium in which he had inserted 2 electrodes connected to pass an electric current through the medium while the bacteria were in there growing and dividing. The electrodes were made of platinum, which was considered to be an inert metal that would not react chemically with anything in the culture. One day, according to Barney, the culture didn't grow. He thought his technician must have forgotten to inoculate the bacteria into the broth. The technician may have known that he did indeed inoculate it, but thought that, well, he'd inoculate it again in the morning. But the broth was perfectly good, he thought, so why bother preparing it again in the morning, so he just put it into the refrigerator. Next day, despite a fresh inoculum of bacteria, still nothing seemed to be growing in that old medium. Now, under similar circumstance, many a researcher would have just dumped the old medium and started fresh. But Rosenberg was curious, so he took a bit of the medium, which was still perfectly clear, as if nothing was growing -- and looked at it under the microscope. What he saw was astonishing. 86 K.W. Kohn Drugs Against Cancer CHAPTER3 ' •• , 'I ', '' I.,, ) I I\, I '\ I I , ,~ • , ,, I ("<1 ,, .,.- ' • I • I ,I •, "" ' , • • I I - \.I I• ..... ' ' , • , ' • Nothing added Platinum complex added Figure 3.3. The platinum complex prevented the bacteria from dividing but allowed them to grow into long filaments. E. coli grown without (left) or with (right) an active platinum complex (X600) (Rosenberg et al., 1967a). The image on the right may have been concentrated by centrifugation, because what Rosenberg saw would have been a much sparser number of filaments in the microscope field. E. coli bacteria are normally short rods. But what Rosenberg saw under the microscope were long filaments (Figure 3.3). It seemed that the E. coli were growing in length but not dividing. It is impossible to know exactly what was going through his mind when he saw that, but it might have been something like this: There is something in this medium that is preventing the bacteria from dividing, even though it is allowing them to grow in substance, thereby producing those long threads. It has recently been reported, he might have reasoned, that x-rays and agents like nitrogen mustard do exactly that: they let the bacteria grow but inhibit their ability to divide: they were reported to grow into long threads just like what is here now in this medium. Furthermore, x rays and nitrogen mustard have anticancer activity. So, maybe an anti-cancer substance might somehow have gotten into the medium. But how? The only thing different from the original medium in which the bacteria were growing normally is that there were thin platinum bars (the electrodes) immersed in it and I had passed an electric current through them. But platinum is chemically inert. Or is it? What about the electricity that was going through those platinum electrodes? It might have caused some platinum atoms to come off and bind to the ammonia or chloride in the medium. So that was it! Following up on those ideas, Rosenberg ordered several platinum complexes that were available from a chemical supply company. He soon discovered the active material that prevented the bacteria from dividing while allowing them to grow into long threads: it was indeed a complex of platinum with ammonia and chloride (Rosenberg et al., 1967b); it was in fact the drug that we now call cisplatin (Figure 3.1). In short order, it was tested at the National Cancer Institute and at the Chester Beatty in England for anti-cancer activity in mice. And the results were spectacular! The anticancer activity of the platinum complex was astounding. 87 K.W.Kohn Drugs Against Cancer CHAPTER3 Moreover, when cisplatin was combined with other drugs, such as cyclophosphamide, the anti-cancer effects were even more impressive (Rosenberg and Vancamp, 1970) (Woodman et al., 1973). Cisplatin and nitrogen mustard are chemical cousins. The chemical structures of cisplatin and nitrogen mustard may look different (Figure 3.4); nevertheless they both work by crosslinking DNA (Zwelling et al., 1981). The favored site of attack on DNA was found to be the same for cisplatin and nitrogen mustard: the nitrogen at position 7 of guanine. The only similarity between the two structures is that each has 2 chlorine atoms -- which indeed is the key to the similarity in their chemical actions. In both drugs, each chlorine can come off, leaving behind a reactive site capable of forming a covalent bond with DNA or proteins. Both drugs have 2 reactive sites whereby DNA can become crosslinked. Although cisplatin, like nitrogen mustard, can produce interstrand crosslinks (Zwelling et al., 1981), cisplatin differs in that it more frequently produces DNA-disabling intra-strand crosslinks (Figure 3.5). Crosslink repair is paramount to the drug-treated cell's survival. The cell has highly effective DNA repair mechanisms, which however are not foolproof. Sometimes it leaves the repaired DNA with missing bases, thus producing mutations. However, there was a puzzle about cisplatin: the powerful anti tumor action of cisplatin was completely abolished if the 2 chlorine atoms were across from each other (trans configuration) rather than next to each other as in cisplatin (cis configuration). (Figure 3.4). Leonard Zwelling, who was then a Clinical Associate in my laboratory, decided to investigate this puzzle using the DNA filter elution methods we had been developing at that time to measure both DNA interstrand and DNA-protein crosslinks (Kohn, 1996) (see Chapter 9). Len's results were quite remarkable: he showed that the trans compound produced almost exclusively DNA- protein crosslinks, and lacked the potent cell-killing and mutation-producing actions of cisplatin (Zwelling et al., 1979a; Zwelling et al., 1979b). Evidently, having the active chlorines across from each other (trans geometry) was unsuited for DNA crosslinking, whereas it easily crosslinked between DNA and proteins. We were then able to measure the rates of formation and repair of both types of crosslinks. The trans-platinum compound was also useful in our developing a method to quantify DNA-protein crosslinks (Kohn and Ewig, 1979), a method that proved key to our discovery of topoisomerase-targeted anti-cancer drugs, as related in Chapters 9 and 10. We found that the interstrand crosslinks produced by cisplatin in cells did not form right away. The number of crosslinks increased for 12 hours before reaching its peak and then declined as they were repaired (left panel in Figure 3.6). DNA-protein 88 K.W. Kohn Drugs Against Cancer CHAPTER3 crosslinks produced by transplatin, on the other hand, formed rapidly and then were repaired (center panel in Figure 3.6). We were astonished by the remarkable difference between the two isomers in that the trans isomer (transplatin) produced many DNA-protein crosslinks, but virtually no interstrand crosslinks, a finding that was substantiated by more precise quantitative measurements (Kohn and Ewig. 1979). That difference went along with cisplatin being tremendously more effective than transplatin in killing cells (right panel of Figure 3.6). We concluded that, at least in the cells we studied, cisplatin-induced DNA inter-strand crosslinks (or the intra- strand crosslinks that correlated with them) were highly toxic, but DNA-protein crosslinks were effectively repaired and were toxic only at very high drug concentrations. Cisplatin's platinum atom (Pt) binds to the same nitrogen atom on guanines (GN7) as nitrogen mustard does, although nitrogen mustard more often forms inter-strand crosslinks than cisplatin does (Figure 3.5). Although it may seem surprising. the 2 DNA crosslinking drugs, cisplatin and cyclophosphamide (a derivative of nitrogen mustard, see chapter 1), often are more effective when used together (synergistic) than either of them used separately (Woodman et al., 1973). The reason may be that the crosslinks produced by the two drugs produce different alterations in DNA structure, such as intra-strand versus inter-strand crosslinks, which are repaired by different molecular systems. Some tumor cells may have a high ability to repair one or the other type of crosslink, while relatively few could repair both types well. Another remarkable finding was in a cell line that was selected for resistance to cisplatin but that remained sensitive to L-phenylalanine mustard (melphalan). The cisplatin-resistant line had completely lost its interstrand crosslinking response to cisplatin, as might be expected for a cell line selected for resistance to this drug. That same cisplatin-resistant cell line however produced just as many interstrand crosslinks to melphalan as the original sensitive line cell line (Figure 3.7). This result supported the idea that the sensitivity or resistance of these cell lines depended on whether interstrand crosslinks were produced. /'-../Cl CH3- N 'v-"---.. Cl nitrogen cisplatin must ard 89 K.W.Kohn Drugs Against Cancer CHAPTER3 Figure 3.4. The chemical structures of cisplatin and nitrogen mustard look different, but their key chemical reactions are similar. Nitrogen mustard has 2 carbons separating the Cl from the N; that arrangement facilitates the loss of the Cl, leaving behind a reactive group on the molecule (see Chapter 1). Cisplatin too is made reactive by the loss of a chloride, which is facilitated by an abundance of water molecules that replace the CJ-. An H20 molecule bound to the Pt'• constitutes the reactive site in cisplatin (as well as in transplatin) . ln both cisplatin and nitrogen mustard, two chlorines are essential, because the departure of each one leaves behind an active center, thus allowing 2 reactions to form a crosslink; the 2 chlorines in cisplatin have to be next to each other for the geometry to allow crosslink formation in DNA. DNA I I Figure 3.5. After its 2 chlorine ions have come off (and replaced by 2 water molecules), cisplatin can bind to 2 guanines on the same DNA strand. The resulting intra-strand crosslink distorts the DNA, which has to be repaired before the DNA can continue to function; but if the repair fails, the crosslink has lethal potential. Cisplatin: interstrand crosslinks ... "- WITIC PflQttllU.SE »>.--~- ~- ~ - ..--, u - Transplatin: ONA-protein crosslinks NO MOfflHASf Cell survival :fJ .!: , " 'Bi ... g eu l trans-platin C ·;;; ' ••• < .,..,.,, PDO < . e0. •• ,.,,... i, ..,, ~ cisplatin • u,l"OO z 0, 0 10011M M TIME Nl(R OIIUG TAUl,T-..Qn" IHOl,lll$1 02 •, • u " ,.,. ,. TIMI Al'T(ll OIIUC TMATMEHT !HOIJIIS) ..... '"' "' ,., ,., "' OIUJ(l ('ONCENTRATIQH 1.,MJ Figure 3.6. Left: formation and repair of cisplatin-induced DNA interstrand crosslinks. Center: formation and repair of transplatin-induced DNA-proteins crosslinks. Right: survival after treatment of cells with cisplatin or transplatin. lnterstrand crosslinks and DNA-protein crosslinks in mouse leukemia L1210 cells were measured by filter elution (Zwell ing et al., 1978) (see Chapter 9). Cell survival was measured by survival of colony-forming ability of 90 K.W.Kohn Drugs Against Cancer CHAPTER3 human V79 cells (Zwelling et al., 1979b). (The production ofinter-strand crosslinks may be an approximately proportional measure of the production of intra-strand crosslinks.) Cisplatin L-phenylalanine mustard Parent Resistant Parent Resistant l 1210 Lt2\0 POO l l110 LlllOIPOO ,.• , . l'l(N'J UN( PA..ifHT UNE ~ .: vi ...u~ ,.• ... ,. iii ~ 0 0 V V ,. 1 ~ C: ,.,,. l 12l0.'l'OO l 1210 PAREN l U N( l lZU)/POO , M t-" 'it I..INE -~ •• • 0 ...u .., 0 V "O • V C: ...'" ... t( ., " Qj £ 0.0 0 • " • " ' ,, ",, ,. . • .. • Hours after treatment of cells"with drug 0 ,. 0 , " .. Figure 3.7. A cell line that had been made resistant to cisplatin did not form DNA crossllnks after being treated with clsplatin (left). However, the cisplatin-resistant cells retained their sensitivity to L-phenylalanine mustard (mel phalan) and melphalan produced crosslinks despite the cell's resistance to cisplatin (right) (Zwelling et al., 1981). From cisplatin treatment to cancer cell death. There is joy when a patient's cancer responds to chemotherapy. The factors that determine whether the cancer will respond however were complicated and not very well understood. Much attention was paid on identifying factors that gave some degree of predictability of response. Some of them were empirical clinical factors and some were physiology- or molecular-based and all were given much attention as chemical and molecular details were elucidated. A general idea of what was involved is shown in Figure 3.8, which is an overview of some of the main factors that were thought to determine whether a cisplatin-treated cell will live or die. As the steps governing cancer cell killing were elucidated, that information was used to help improve the clinical effectiveness of the platinum drugs (Galluzzi et al., 2014; Kelland, 2007; O'Grady et al., 2014). The main reaction steps that were found to affect the death or survival of cisplatin-treated cells are diagrammed in Figure 3.8: 91 K.W. Kohn Drugs Against Cancer CHAPTER3 first, cisplatin must pass through the cell's plasma membrane to get into the cell, which occurs in part by way of specific channels in the membrane that normally allow essential copper compounds to enter. Particularly important however are other channels, which actively pumped cisplatin (normally copper) out of the cell. The intake and export channels affect how much cisplatin is inside the cell (a and b in Figure 3.8). When those particular export pumps were defective, copper was known to accumulate in cells and cause Wilson's disease. One might expect that cisplatin would also accumulate in the cells of Wilson's disease patients, thereby making those cells, whether normal or cancerous more sensitive to the drug. However, Wilson's disease cells express a copper-binding protein that can bind and detoxify cisplatin. The protein binds cisplatin tightly at a pair of cysteines separated by two amino acids (CxxC) and affects the sensitivity of the cells to cisplatin despite the higher concentrations of the sulfur-containing cisplatin binders glutathione and metallothionine (Dolgova et al., 2013). Once inside the cell, cisplatin is activated by its platinum-bound chlorides being replaced by water molecules (c in Figure 3.8). The reason that this replacement activated cisplatin is that the platinum-water bond is weak, and the platinum atom would much rather bind to a nitrogen atom, such as, for example, the one at position 7 of guanine in DNA. That would be the first bond in a prospective DNA crosslink. The main reason that this chloride-water replacement would occur inside the cell, but not outside, is that the concentration of free chloride is much lower inside the cell than outside in the blood. When cisplatin is water-activated (seems odd to put it that way, so scientists instead say "aquated"), what happens next? Sometimes the aquated cisplatin will bind to N7 of DNA guanine, as mentioned above; that toxic reaction however was relatively rare overall, but its impact overrode its rarity. More often, the aquated cisplatin would have been inactivated by binding tightly to one of the many sulfur compounds in the cell (such as glutathione, a common sulfhydryl compound, or to metallothionein, a metal-binding protein that has a large number of exposed sulfhydryl groups) -- platinum loves sulfur (din Figure 3.8). The aquated site on the second arm of cisplatin could then proceed to form a DNA crosslink (e in Figure 3.8). The resulting DNA damage would be detected by surveillance systems, which signal to the cell's molecular response systems that danger is afoot (tin Figure 3.8). The signals activate a remarkable network of logically integrated components that cause the cell to delay DNA replication and cell division to give more time for the cell to adapt and to repair the damage before the replication machinery boggles by trying to replicate through a crosslink, or the chromosomes scramble if the cell tries to mitose while its genome was unrepaired. It also put the cell's DNA repair machinery on high alert (.g and h in Figure 3.8). 92 K.W.Kohn Drugs Against Cancer CHAPTER3 After all that, if DNA crosslinks still remained as the cell tried to move forward in the cell division cycle, a lethal event would happen when the DNA replication machinery encountered a crosslink (i in Figure 3.8). Another process that determined life or death for the cell came as an output from the damage detection and response network If the damage persisted too long, the system took no chances and actively suicided the cell -- aficionados call it "apoptosis" (from Greek "falling off," as in falling off of leaves) Uin Figure 3.8). This helped avoid the production of mutated cells that could lead to cancer. However, apoptosis of cancer cells obviously was desirable and was a positive factor for chemotherapeutic response. Cisplatin outside Outside the cell Plasma membrane Inside the cell b Cisplatin inside C d Activated - - ~ Inactivated cisplatin cisplatin e I DNA i-:--1C>I DNA damage crosslinks i Repaired Death by Cell death DNA apoptosis Figure 3.8. Simplified scheme showing factors that were thought to govern the effects of cisplatin on cells. a Channels allow cisplatin to enter the cell. b Channels that pump cisplatin out of the cell. c Clsplatin becomes activated inside the cell by its chlorides being replaced by water molecules (favored inside the cell, where chloride concentration is low). d Cisplatin can react with sulfur-containing molecules inside the cell to form inactive products. e Clsplatin produces DNA crosslinks. /DNA crosslinks stimulate the cell's DNA damage detection and response systems.g DNA repair machinery of the cell repairs the crossllnks. h DNA damage response system stimulates the cell's DNA repair machinery. I Unrepaired DNA crosslinks lead to death of the cell. JIf the DNA damage repair system fails, it induces the cell to suicide by apoptosis (Galluzzi et al., 2014) (O'Grady et al., 2014). 93 K.W.Kohn Drugs Against Cancer CHAPTER3 A problem: Cisplatin damages the kidney. The main toxicity that limited how much cisplatin could safely be given was damage to the kidneys (Figure 3.9). Cisplatin is actively taken up by cells of the kidney tubules, resulting in deleterious drug concentrations in the cells (Yao et al., 2007). Although the problem was not fully solved, the kidney toxicity was reduced by giving patients lots of fluid and diuretics to increase urine flow that would reduce the concentration of the drug in the urine as it flows through the kidney. Figure 3.9. Damage to cells of the kidney tubules after a course of cisplatin treatment in mice (Kruger et al., 2016). The yellow arrows point t o some of the cell nuclei that have sustained extensive DNA damage (as revealed by staining with antibody to gamma-H2AX; see Chapter 28). After cisplatin treatment (right), the number of DNA damaged cells increased 6-fold compared with untreated cells (left) . Why some patients refused to take cisplatin. Cisplatin-containing therapy made most patients so nauseous that many could not stand it. Even entering the hospital where the drug was given was sometimes sufficient to trigger nausea. Many chemotherapeutic drugs can cause nausea, but cisplatin stood out as an extreme case, and the vomiting it induced was sometimes alarming. The cause might be a direct action on certain centers in the brain, but drugs to interfere with that direct action were not yet available. Ordinary available medications would control nausea shortly after the drug was administered. However, extreme nausea occurred later (perhaps after the drug had entered and affected certain neurons in the vomiting center in the brainstem), at which time it was not easily controlled (De Jonghe et al., 2016; lshido et al., 2016; Shi et al., 2016). 94 K.W.Kohn Drugs Against Cancer CHAPTER3 Treatment of cancer patients with Cisplatin and related drugs. The testing of cisplatin in tumor-bearing mice, which began in 1968, produced results that were so encouraging that only 3 years later the first clinical trial was begun. That was a remarkably short time between finding anti-tumor activity in mice and approval for clinical trial. Not only was there strong evidence of antitumor activity, but the toxicity and dose-levels were well enough understood to try the drug on human patients. Substantial antitumor effect was first reported in 1974 for testicular and ovarian cancer, and several subsequent clinical trials reported increased survival also in lung cancer patients (Lebwohl and Canetta, 1998). Moreover, cisplatin was unusual because it was not toxic to the blood-forming cells in the bone marrow. By 1978, the benefit to cancer patients was promising enough to make cisplatin available for general oncology practice. The chemical structure of cisplatin is modified in search for better drugs. Many variations on the cisplatin structure were tested in search for compounds with increased effectiveness against a greater variety of tumors, with less toxicity, or with lack of cross-resistance to cisplatin (Lebwohl and Canetta, 1998). Two structural relatives became widely used: carboplatin and oxaliplatin (Figure 3.10). In carboplatin, the two chlorine atoms are replaced by a chemical group that has two carboxyl (COO-) groups attached to the platinum atom (Figure 3.10). The carboxyl groups activated the platinum in the same way that the chlorides did, that is, by being replaced by water molecules; however, carboplatin was activated more slowly than cisplatin. Another feature was that the two carboxyl groups were part of a structural unit that might favor both carboxyls coming off at nearly the same time; thus, the two platinum sites would be aquated together, setting the stage for efficient crosslink production. Carboplatin formed the same kinds of DNA crosslinks as cisplatin, but formed them 10-times more slowly, and 30-times higher drug concentrations were needed. The clinical benefit in terms of increased survival time however was not very different from cisplatin; the main benefit of carboplatin was that it was less toxic: it did not damage the kidneys, and spared the gastrointestinal tract and central nervous system; its dose-limiting toxicity, instead, was suppression of platelet and white blood cell production in the bone marrow (Kelland, 2007; Lebwohl and Canetta, 1998). In oxaliplatin, like carboplatin, the chlorides were replaced by carboxyl groups in a structural unit. Unlike cisplatin and carboplatin, however, the two amino groups were linked together via a 6-membered ring. a cyclohexyl group (Figure 3.10). Oxaliplatin-mediated DNA crosslinks therefore retained this cyclohexyl structure, which might impair the binding of some DNA repair proteins to the damage site. A notable finding in the clinical experience with oxaliplatin was that, unlike 95 K.W. Kohn Drugs Against Cancer CHAPTER3 carboplatin, it sometimes was effective in patients whose tumors had stopped responding to cisplatin (Kelland, 2007). 0 CX NH\ O- r ,p, I' / /\o- c..,._ NH2 0 clsplatin oxaliplatin ! I Or NH\ NHi Bulky group blocks binding of )iiuanine /r1 ,IGuanine DNA I DNA repair enzymes Figure 3.10. The upper row shows the chemical structures of cisplatin and its two relatives that were most commonly used: carboplatin and oxaliplatin. The bottom row shows the structures of the DNA crosslinks formed by each of the drugs. The DNA crosslink produced by oxaliplatin differed from the kind produced by cisplatin or carboplatin in that it had an additional 6-membered ring (cyclohexyl group) sticking out from the DNA; this may block the binding of some DNA repair proteins, and may be why oxaliplatin sometimes worked against tumors that were resistant to cisplatin or carboplatin (Chaney et al., 2005; Kelland, 2007). Cisplatin cures advanced testicular cancer. The greatest benefit of cisplatin was for patients with testicular cancer: cisplatin produced lasting remissions in nearly 80% of the cases, and most of the patients who had the common germ cell type of testicular cancer were cured with a drug combination based on cisplatin (Figure 3.11). Germ cell cancer of the testis, although relatively rare compared to some other types of cancer, was one of the very few types of cancers that could be cured by chemotherapy after the cancer has spread (metastasized). What made cure possible was the addition of cisplatin to previously established drug combinations that by themselves were much less effective (Einhorn, 1997; Einhorn, 1981; Hinton et al., 2003). The susceptibility of testicular cancer to chemotherapy was in part due to a relatively rapid cell proliferation rate compared to other cancers, and to the fact that testicular cancers rarely become dormant Tumors with active cell division tended to be susceptible to chemotherapy, as was the case for acute leukemias and choriocarcinoma, and those tumors were often curable. Also contributing to their 96 K.W.Kohn Drugs Against Cancer CHAPTER3 being curable, may be that testicular cancers were nearly unable to repair the DNA damage caused by the drug. (Kelland, 2007). Prognosi.s ,.. b \ '==-tu______....;;g'-o_o_d_ _ _ _ _ _ _ _ _ __ __ ... \..,\, '-'""'---- ---.. ________________________ "'> ' '-...,- --'- - intermediate •~ ll -·--·---- - - ..., - - - -\..-·-l--.-·, ::, V, poor 4- i_·---- ·- 0 .~ ... .c .."' ... .c 0 0.. IJJQ fllrjC Ta pdJXII ... • • , • • • 1 • • n Time after treatment (years) Figure 3.11. A cancer cure. Patients with t e.sticular cancer that had already spread were treated with a cisplatin-based drug combination. The three curves are for patients who w ere judged at the beginning of treatmentto have a good, intermediate, or poor prognosis, based on how advanced their disease was at the time. The results showed that patients whose disease was not highly advanced had greater than 80% chance of remaining alive after 10 years; even highly advanced cases had a 50% chance (Hinton et al., 2003). The curves leveled off with time, showing that patients who survived the first few years were unlikely to die of the disease in the years to come. Summary Cisplatin, which became one of the most useful anticancer drugs, was discovered as the result of an accidental event in a very modest laboratory. Its discovery was due to bold thinking by Barnett Rosenberg. The story is remarkable, because there had been no clue that a heavy metal compound, such as cisplatin, could have anticancer activity, and the major drug discovery programs had never considered testing such compounds. Cisplatin therapy was so promising that great effort was made to overcome the drug's toxicities, and those efforts had significant success. Cisplatin was highly effective and enabled the cure of testicular cancer. The aim then was to modify cisplatin-type therapy so as to find treatments that would be as effective for the common cancers as cisplatin was for testicular cancer. One possibility was that 97 K.W.Kohn Drugs Against Cancer CHAPTER3 higher drug doses could be used if toxicity were controlled. Another possibility was to develop better platinum drugs or better drug combinations. The major toxicity of cisplatin was kidney damage, which however was largely overcome by increased hydration of the patient Many variations of the cisplatin structure were put in clinical trial, but nothing was found better than the old standbys: cisplatin, carboplatin, and oxaliplatin. Unfortunately, many cancer patients were not helped by any platinum regimens or other available chemotherapy. When tumor regression did occur, it was often brief and survival was extended for but a few months. But complete durable remission did sometimes occur, which gave reason for hope, especially if the exact reasons for the effectiveness against testicular cancer could eventually be worked out. References Chaney, S.G., Campbell, S.L., Bassett, E., and Wu, Y. (2005). Recognition and processing of cisplatin- and oxaliplatin-DNA adducts. Critical reviews in oncology/hematology 53, 3-11. De Jonghe, B.C., Holland, R.A., Olivos, D.R., Rupprecht, L.E., Kanoski, S.E., and Hayes, M.R. (2016). Hindbrain GLP-1 receptor mediation of cisplatin-induced anorexia and nausea. Physiology & behavior 153, 109-114. Dolgova, N.V., Nokhrin, S., Yu, C.H., George, G.N., and Dmitriev, O.Y. (2013). Copper chaperone Atoxl interacts with the metal-binding domain of Wilson's disease protein in cisplatin detoxification. The Biochemical journal 454, 147-156. Einhorn, E.H. (1997). Testicular cancer: an oncological success story. Clinical cancer research : an official journal of the American Association for Cancer Research 3, 2630-2632. Einhorn, L.H. (1981). Testicular cancer as a model for a curable neoplasm: The Richard and Hinda Rosenthal Foundation Award Lecture. Cancer research 41, 3275-3280. Galluzzi, L., Vitale, I., Michels, J., Brenner, C., Szabadkai, G., Harel-Bellan, A., Castedo, M., and Kroemer, G. (2014). Systems biology of cisplatin resistance: past, present and future. Cell death & disease 5, e1257. Hinton, S., Catalano, P.J., Einhorn, L.H., Nichols, C.R., David Crawford, E., Vogelzang, N., Trump, D., and Loehrer, P.J., Sr. (2003). Cisplatin, etoposide and either bleomycin or ifosfamide in the treatment of disseminated germ cell tumors: final analysis of an intergroup trial. Cancer 97, 1869-1875. Ishido, K., Higuchi, K., Azuma, M., Sasaki, T., Tanabe, S., Katada, C., Yano, T., Wada, T., and Koizumi, W. (2016). Aprepitant, granisetron, and dexamethasone versus palonosetron and dexamethasone for prophylaxis of cisplatin-induced nausea and vomiting in patients with upper gastrointestinal cancer: a randomized crossover phase II trial (KDOG 1002). Anti-cancer drugs. Kelland, L. (2007). The resurgence of platinum-based cancer chemotherapy. Nature reviews Cancer 7, 573-584. 98 K.W.Kohn Drugs Against Cancer CHAPTER3 Kohn, K.W. (1996). DNA filter elution: a window on DNA damage in mammalian cells. BioEssays : news and reviews in molecular, cellular and developmental biology 18, 505-513. Kohn, K.W., and Ewig, R.A. (1979). DNA-protein crosslinking by trans- platinum(II)diamminedichloride in mammalian cells, a new method of analysis. Biochimica et biophysica acta 562, 32-40. Kruger, K., Ziegler, V., Hartmann, C., Henninger, C., Thomale, J., Schupp, N., and Fritz, G. (2016). Lovastatin prevents cisplatin-induced activation of pro-apoptotic DNA damage response (DOR) of renal tubular epithelial cells. Toxicology and applied pharmacology 292, 103-114. Lebwohl, D., and Canetta, R. (1998). Clinical development of platinum complexes in cancer therapy: an historical perspective and an update. Eur J Cancer 34, 1522- 1534. O'Grady, S., Finn, S.P., Cuffe, S., Richard, D.J., O'Byrne, K.J., and Barr, M.P. (2014). The role of DNA repair pathways in cisplatin resistant lung cancer. Cancer treatment reviews 40, 1161-1170. Petsko, G.A. (2002). A christmas carol. Genome biology 3, COMMENT1001. Rosenberg, B., Renshaw, E., Vancamp, L., Hartwick, J., and Drobnik, J. (1967a). Platinum-induced filamentous growth in Escherichia coli. Journal of bacteriology 93, 716-721. Rosenberg, B., Van Camp, L., Grimley, E.B., and Thomson, A.J. (1967b). The inhibition of growth or cell division in Escherichia coli by different ionic species of platinum(IV) complexes. The Journal of biological chemistry 242, 1347-1352. Rosenberg, B., and Van Camp, L. (1970). The successful regression of large solid sarcoma 180 tumors by platinum compounds. Cancer research 30, 1799-1802. Rosenberg, B., Vancamp, L., and Krigas, T. (1965). Inhibition of Cell Division in Escherichia Coli by Electrolysis Products from a Platinum Electrode. Nature 205, 698-699. Shi, Q., Li, W., Li, H., Le, Q., Liu, S., Zong, S., Zheng, L, and Hou, F. (2016). Prevention of cisplatin-based chemotherapy-induced delayed nausea and vomiting using triple antiemetic regimens: a mixed treatment comparison. Oncotarget. Woodman, R.J., Sirica, A.E., Gang, M., Kline, I., and Venditti, J.M. (1973). The enhanced therapeutic effect of cis-platinum (II) diamminodichloride against L1210 leukemia when combined with cyclophosphamide or 1,2-bis(3,5- dioxopiperazine-1-yl)propane or several other antitumor agents. Chemotherapy 18, 169-183. Yao, X., Panichpisal, K., Kurtzman, N., and Nugent, K. (2007). Cisplatin nephrotoxicity: a review. The American journal of the medical sciences 334, 115-124. Zwelling, L.A., Anderson, T., and Kohn, K.W. (1979a). DNA-protein and DNA interstrand cross-linking by cis- and trans-platinum(II) diamminedichloride in L1210 mouse leukemia cells and relation to cytotoxicity. Cancer research 39, 365-369. Zwelling, L.A., Bradley, M.O., Sharkey, N.A., Anderson, T., and Kohn, K.W. (1979b). Mutagenicity, cytotoxicity and DNA crosslinking in V79 Chinese hamster cells 99 K.W. Kohn Drugs Against Cancer CHAPTER3 treated with cis- and trans-Pt(II) diamminedichloride. Mutation research 67, 271-280. Zwelling, L.A., Kohn, K. W., Ross, W.E., Ewig, R.A., and Anderson, T. (1978). Kinetics of formation and disappearance of a DNA cross-linking effect in mouse leukemia L1210 cells treated with cis- and trans-diamminedichloroplatinum(II). Cancer research 38, 1762-1 768. Zwelling, L.A., Michaels, S., Schwartz, H., Dobson, P.P., and Kohn, K.W. (1981). DNA cross-linking as an indicator of sensitivity and resistance of mouse L1210 leukemia to cis-diamminedichloroplatinum(II) and L-phenylalanine mustard. Cancer research 41 , 640-649. 100 K.W.Kohn Drugs Against Cancer CHAPTER4 Chapur-4. T~ DNA lnu.rrtJltl tlon .story 220719ao3 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@nih.gov CHAPTER4 The DNA Intercalation Story: Drug-DNA sandwiches. In Chapters 1-3, we saw how alkylating agents and platinum complexes bind to DNA tightly and irreversibly (covalently). We come now to other DNA-binding anticancer drugs that bind tightly, but not covalently - which means that the DNA-binding of these drugs was spontaneously reversible, in contrast to the covalent binders of the previous three chapters, whose binding was irreversible. The non -covalent DNA- binding drugs of this chapter have a flat multi-ring system having size and shape resembling a DNA base-pair- which allows them to slip in between base-pairs of DNA in a sandwich-like configuration that is called "DNA intercalation" (Figures 4.1- 4.3). The DNA helix unwinds slightly to open a space between adjacent base-pairs that is just sufficient in size to accommodate the intercalating ring system. The intercalation is stabilized in part by the electron distribution patterns of intercalator and the DNA base-pairs that are snuggly stacked against each other (Figure 4.3). The intercalation story began in 1960. We had been discussing the notion of DNA intercalation in Paul Doty's laboratory, but it was first put on solid footing by Leonard Lerman in studies of the physical consequences of the DNA binding of the dye, proflavine (Figure 4.2). The DNA intercalation concept was to have unexpected applications in DNA studies, particularly to topoisomerase blocking agents (Chapter 10), and the mode of action of doxorubicin (Chapter 8). 101 K.W. Kohn Drugs Against Cancer CHAPTER4 )-' .J-l2N - O:P'.~ (N O·····...• f '\ _ ~H······° Cytosine O , o Guanine 2 e5 ~-.-~" I I 1-1----,,---~ I Sugar-phosphate Bases Sugar-phosphate backbone backbone Figure 4.1. A guanine-cytosine base-pair and connections to the backbones of the 2 DNA strands. The dashed lines indicate the hydrogen bonds that allow the 2 bases (guanine-cytosine or adenine-thymine) to fit together. The base pairs are flat and stack one upon another in the DNA double helix. (From Wikimedia Commons. File:0322 DNA Nucleotides.) "- N I Figure 4.2. Outline structure ofa base pair (above) and of the DNA intercalator, proflavine (below) (modified from (Lerman, 1961)). The double-bonds are not shown. The DNA base-pair and proflavine have similar size and shape, which allows proflavine to stack against the base-pair and to intercalate between base-pairs in a DNA double helix. 102 K.W.Kohn Drugs Against Cancer CHAPTER4 DNA backbones Normal DNA with bound DNA lntercalator(red) Figure 4.3. Simplified picture of a DNA intercalation structure as proposed by Lerman in 1961 (Lerman, 1963). The drug, with its flat multi-ring structure (red) is sandwiched between base-pairs of a slightly unwound DNA double helix. The DNA helix is unwound just enough for the intercalator to slip in between base-pairs Another simple intercalator, ellipticine, derived from the bark of the Australian tree Ochrosia el/iptica, was found to have substantial anticancer activity in mice in NCI screens (Figures 4.4 and 4.5). Although it has four rings, rather than the three in proflavine, ellipticine's size and shape closely approximates that of a base-pair. The drug is an effective DNA intercalator and was found to block topoisomerase II (Chapter 8) (Kohn et al., 1975; Ross et al., 1979; Ross et al., 1978). On the down-side, however, ellipticine's ability to block topoisomerase II was considered mediocre compared with other more potent drugs. Moreover, it was difficult to use clinically because of low solubility, and problematic toxicities were encountered. The chemical structure of ellipticine was therefore modified in hope of producing novel effective drugs. That effort succeeded in producing more potent topoisomerase II blockers with ability to kill cancer cells in culture. However, toxicity still precluded their approval for general use in cancer chemotherapy (Auclair et al., 1987; Vann et al., 2016). 103 K.W.Kohn Drugs Against Cancer CHAPTER4 Figure 4.4. Chemical structure of ellipticine. The nitrogen in the 5-membered ring can pick up a proton (hydrogen ion) from water and become positively charged. The attraction of ellipticine's positive charge to the DNA's negative charge helps stabilize the binding. Figure 4.5. Ellipticine is found in the bark of plants oftheApocynaceae family, such as Ochrosia borbonica (Tmejova et al., 2014). (Picture from Wikipedia.) DNA intercalating drugs, such as doxorubicin, exert their anticancer action by blocking topoisomerase II (Chapter 8). But intercalating ability by itself does not guaranty action against topoisomerase. In point is the case of the anticancer drug, m-AMSA (amsacrine). An isomer ofm-AMSA, called o-AMSA, intercalates equally well (Waring, 1976), but only m-AMSA blocked topoisomerase II and only it had anticancer activity. The structure of o-AMSA differs only in a small chemical group on a ring (not involved in the intercalation). The chemical group (H3CO) is moved over by one carbon atom (Figure 4.6). Despite its ability to intercalate and its close structural similarity to its active isomer, o-AMSA was totally inactive. Evidently, the part of the drug that sticks out from the DNA intercalation structure must interact 104 K.W.Kohn Drugs Against Cancer CHAPTER4 with the topoisomerase protein. The configuration in o-AMSA presumably is incompatible with that essential interaction. Thus, we see that, although o-AMSA intercalates in DNA (Waring, 1976), it lacked the ability to block topoisomerase 11, and had no antitumor activity (Zwelling et al., 1981). It seems that the repositioned group on the external ring prevented the interaction of the intercalated compound with the topoisomerase II enzyme, perhaps due to its effect on how the 6-membered external ring lies in the DNA minor groove (Jangir et al., 2013). A few words about amsacrine (m-AMSA) as an anticancer drug. The DNA binding and anticancer activities of the drug were discovered in 1980 by Bruce Cain and his colleagues at the University of Auckland, New Zealand, as outcome of an intensive investigation of certain positively charged compounds that also had lipid-binding capability (Atwell and Cain, 1967; Baguley et al., 1981a, b). They carefully studied the relationship between chemical structure and ability to prolong the life of leukemic mice. This eventually led to m-AMSA as best of the set of compounds. Interestingly, m-AMSA was one of the first anti-cancer drugs to be designed by chemists, rather than biological organisms. Clinical trials of amsacrine produced responses, especially in leukemias, but the responses relative to toxicity were not good enough to merit its general use (Jelic et al., 1997). Nevertheless, it was occasionally used in combination with other drugs. NH NH m-AMSA o-AMSA Figure 4.6. Chemical structure of amsacrine (m-AMSA) and its isomer o-AMSA, which differs only in the position of the H3CO- group on the upper ring in the diagram. Both compounds intercalate in DNA (Waring, 1976), but only m-AMSA blocked topoisomerase II, and only this isomer was effective in killing cancer cells (Zwelling et al., 1981). The three-ring part of the molecule intercalates, while the external ring lies in the minor groove where it could interact with topoisomerase II. 105 K.W.Kohn Drugs Against Cancer CHAPTER4 An intercalator anticancer drug with a more complicated structure was actinomycin D, an effective inhibitor of RNA synthesis (transcription). Derived from Streptomyces soil bacteria (Figure 4.7), it was the first antibiotic shown to have anticancer activity; it is still used in combination with other anticancer drugs in the treatment of certain cancers (Cortes et al., 2016). Actinomycin has a flat ring system that slips nicely between DNA base pairs. In addition, a part of the molecule fits compactly between the DNA backbone chain in the DNA minor groove (Figures 4.8 and 4.9). The intercalated ring system and the side chains in the minor groove combine to bind actinomycin tightly to DNA (Hollstein, 1974) (So bell, 1973). Clinical use of actinomycin however was limited by excessive toxicity, perhaps due to its unusually strong inhibition of RNA synthesis. Actinomycin's toxicity and strong RNA synthesis inhibition may well be related to its interaction with the amino (NH2) at position 2 of guanine, which it requires for strong DNA binding (Figure 4.10) (Sobell, 1973). This may account for its action differences from other DNA intercalating anticancer drugs. Figure 4.7. Streptomyces soil bacteria, the source of several important anticancer drugs, including actinomycin. (Scanning electron microscope image, from Wikipedia). 106 K.W. Kohn Drugs Against Cancer CHAPTER4 Figure 4.8. Model of actinomycin bound to DNA. The actinomycin ring system (green) is intercalated between DNA base-pairs, while the external part of the molecule (blue) fits nicely into the DNA minor groove. The intercalation and minor- groove binding cooperate to produce strong binding to DNA, even though actinomycin bears no positive charge (from educational portal PDB-101; http://pdb101.rcsb.org/motm/160). / Sar' ,,..Sar, L-Pro L-Meval L-Pro L-Meval I I I I D-Va( ,. . . o D-Va( ,......o L-Thr O L-Thr N 0 0 CHs CHs Figure 4.9. Chemical structure of actinomycin D. The 3-ring system at the bottom intercalates between base-pairs in DNA, and the two rings of amino acids at the top bind in the DNA minor groove (see Figure 4.8). Tight binding of actinomycin occurs preferentially to DNA regions that are rich in guanine-cytosine base-pairs (Lohani et 107 K.W.Kohn Drugs Against Cancer CHAPTER4 al., 2016), which might have something to do with why actinomycin is particularly effective in blocking RNA synthesis. Figure 4.10. Strong binding ofactinomycin to DNA required an amino group (NH2) at position 2 of guanine (encircled in the structure at upper left). Replacement of guanine's oxygen by an NH2, which allowed it to base-pair with thymine (T) instead of cytosine (C) retained strong actinomycin binding (upper right). But base pairs that Jack the NH2 at guanine position-2 did not bind actinomycin (lower two structures) (Sobell, 1973). The discovery of doxorubicin, The most important DNA intercalating drug, however, was doxorubicin. The story began with the discovery of daunomycin, a close chemical relative of doxorubicin, by Di Marco and his colleagues at the Institute Nazionale dei Tumori in Milan, Italy (Di Marco et al., 1964). Members of the same research group demonstrated its binding to DNA, and noted that the drug increased the viscosity of DNA solutions (Calendi et al., 1965), but they did not surmise the cause: intercalation between base-pairs, lengthens the DNA helix and thereby increases its viscosity in solution. The full story is the subject of Chapter 8. 108 K.W.Kohn Drugs Against Cancer CHAPTER4 References Atwell, G.J., and Cain, 8.F. (1967). Potential antitumor agents. V. Bisquaternary salts. Journal of medicinal chemistry 10, 706-713. Auclair, C., Pierre, A., Voisin, E., Pepin, 0., Cros, S., Colas, C., Saucier, J.M., Verschuere, 8., Gros, P., and Paoletti, C. (1987). Physicochemical and pharmacological properties of the antitumor ellipticine derivative 2-(diethylamino-2-ethyl)9- hydroxy ellipticinium-chloride, HCI. Cancer research 47, 6254-6261. Baguley, B.C., Denny, W.A., Atwell, G.J., and Cain, 8 .F. (1981a). Potential antitumor agents. 34. Quantitative relationships between DNA binding and molecular structure for 9 -anilinoacridines substituted in the anilino ring. Journal of medicinal chemistry 24, 170-177. Baguley, B.C., Denny, W.A., Atwell, G.J., and Cain, 8 .F. (1981b). Potential anti tumor agents. 35. Quantitative relationships between antitumor (L1210) potency and DNA binding for 4'-(9-acridinylamino )methanesulfon-m-anisidide analogues. Journal of medicinal chemistry 24, 520-525. Calendi, E., Dimarco, A., Reggiani, M., Scarpinato, 8 ., and Valentini, L. (1965). On Physico-Chemical Interactions between Daunomycin and Nucleic Acids. Biochimica et biophysica acta 103, 25-49. Cortes, C.L., Veiga, S.R., Almacellas, E., Hernandez-Losa, J., Ferreres, J.C., Kozma, S.C., Ambrosio, S., Thomas, G., and Tauler, A. (2016). Effect of low doses of actinomycin Don neuroblastoma cell lines. Molecular cancer 15, 1. Di Marco, A., Gaetani, M., Orezzi, P., Scarpinato, 8 .M., Silvestrini, R., Soldati, M., Dasdia, T., and Valentini, L. (1964). 'Daunomycin', a New Antibiotic of the Rhodomycin Group. Nature 201, 706-707. Hollstein, U. (1974). Actinomycin. Chemistry and Mechanism of Action. Chemical reviews 74, 625-652. Jangir, D.K., Kundu, S., and Mehrotra, R. (2013). Role of minor groove width and hydration pattern on amsacrine interaction with DNA. PloS one 8, e69933. Jelic, S., Nikolic-Tomasevic, Z., Kovcin, V., Milanovic, N., Tomasevic, Z., Jovanovic, V., and Vlajic, M. (1997). A two-step reevaluation of high-dose amsacrine for advanced carcinoma of the upper aerodigestive tract: a pilot phase II study. J Chemother 9, 364-370. Kohn, K.W., Waring, M.J., Glaubiger, D., and Friedman, C.A. (1975). Intercalative binding of ellipticine to DNA. Cancer research 35, 71-76. Lerman, L.S. (1961). Structural considerations in the interaction of DNA and acridines. Journal of molecular biology 3, 18-30. Lerman, L.S. (1963). The structure of the DNA-acridine complex. Proceedings of the National Academy of Sciences of the United States of America 49, 94-102. Lohani, N., Singh, H.N., and Moganty, R.R. (2016). Structural aspects of the interaction of anticancer drug Actinomycin -D to the GC rich region of hmgbl gene. International journal of biological macromolecules 87, 433-442. Ross, W.E., Glaubiger, D., and Kohn, K. W. (1979). Qualitative and quantitative aspects of intercalator-induced DNA strand breaks. Biochimica et biophysica acta 562, 41-50. 109 K.W.Kohn Drugs Against Cancer CHAPTER4 Ross, W.E., Glaubiger, D.L., and Kohn, K.W. (1978). Protein-associated DNA breaks in cells treated with adriamycin or ellipticine. Biochimica et biophysica acta 519, 23-30. So bell, H.M. (1973). The stereochemistry of actinomycin binding to DNA and its implications in molecular biology. Progress in nucleic acid research and molecular biology 13, 153-190. Tmejova, K., Krejcova, L., Hynek, D., Adam, V., Babula, P., Trnkova, L., Stiborova, M., Eckschlager, T., and Kizek, R. (2014). Electrochemical study of ellipticine interaction with single and double stranded oligonucleotides. Anti-cancer agents in medicinal chemistry 14, 331-340. Vann, K.R., Ergun, Y., Zencir, S., Oncuoglu, S., Osheroff, N., and Topcu, Z. (2016). Inhibition of human DNA topoisomerase llalpha by two novel ellipticine derivatives. Bioorganic & medicinal chemistry letters 26, 1809-1812. Waring, M.J. (1976). DNA-binding characteristics of acridinylmethanesulphonanilide drugs: comparison with antitumour properties. Eur J Cancer 12, 995-1001. Zwelling, L.A., Michaels, S., Erickson, L.C., Ungerleider, R.S., Nichols, M., and Kohn, K.W. (1981). Protein-associated deoxyribonucleic acid strand breaks in L1210 cells treated with the deoxyribonucleic acid intercalating agents 4' -(9- acridinylamino) methanesulfon-m-anisidide and adriamycin. Biochemistry 20, 6553-6563. 110 K.W. Kohn Drugs Against Cancer CHAPTERS Chapter 5. The methotrexat.e story 220719dj3 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@oib gov CHAPTERS The methotrexate story: folic acid analogs. Discovery of methotrexate as an anti-leukemia drug Acute leukemia in the 1940's was relentless and invariably fatal, and there was no way of even slowing down the disease. That terrible disease, often of children, was caused by abnormal white blood cells growing unchecked: they overgrew the bone marrow and blocked normal blood cell production there. The result was depletion of red blood cells with consequent anemia, dearth of normal white blood cells that were needed to fight infections, and reduction in platelets needed to prevent bleeding. In June 1948, just 2 years after Goodman, Gilman and their coworkers reported the lymphoma tumor-melting effect of nitrogen mustard (Goodman et al., 1946) (see Chapter 1), Sidney Farber and his coworkers at Harvard Medical School and The Children's Hospital in Boston reported that aminopterin, an analog and antagonist of folic acid, was able to slow the progress of childhood leukemia (Farber and Diamond, 1948) (Figures 5.1). That was the second breakthrough, after nitrogen mustard, that hastened the era of cancer chemotherapy. Although it was not a cure, it did set the stage for a cure. Aminopterin was a chemically modified folic acid that was known to inhibit the actions offolic acid. This inhibition impaired the production of building blocks for the synthesis of DNA and RNA. Consequently, the drug impaired the ability of cells to grow and divide. 111 K.W.Kohn Drugs Against Cancer CHAPTERS Farber had followed up on a report in 1947 from Lederle Laboratories in Pearl River, New York that folic acid antagonists suppressed white blood cell counts in rats. A modest stretch of the imagination suggested that such anti-folate drugs might impair the growth of leukemic blood cells (Franklin et al., 1947). Farber began cautiously treating children in the last stages of the disease with modified forms of folic acid. After some encouraging results, Farber selected aminopterin for the further studies, because it was the most potent folic acid antagonist available. To the investigators' surprise and delight, some of the children had a remarkable response: their symptoms improved and their leukemia cells disappeared (Farber and Diamond, 1948). For a short time, it even seemed as if they might be cured. But within a few months, leukemia cells began to grow again, and those newly growing leukemia cells did not respond to the drug. Similar temporary responses were soon reported also in adult patients with acute leukemia (Dameshek, 1949). In addition to aminopterin, the latter trials used another folate antagonist, amethopterin, which came to be called methotrexate and was to become a mainstay of cancer therapy. Dameshek likened acute leukemia to a wildfire, which, although dampened by aminopterin, continued to smolder and could suddenly light up again (Dameshek, 1949). The temporary responses of childhood acute leukemia to the "antifols" were impressive and beyond previous experience. However, aminopterin or methotrexate, used by itself, was far from a cure. The roots of the anti fol discovery however can be traced further back to the 1930's and early 1940's, when researchers found that a folic acid deficiency often caused anemia (Hotlbrand and Weir, 2001). The bone marrow of some of the anemic patients contained unusual enlarged cells that they thought resembled leukemia cells. The researchers therefore thought that leukemia might result from a folic acid deficiency. This was incorrect, however, because those enlarged cells were abnormal precursors of red blood cells, not leukemic white blood cells. Even though the conjecture was wrong, it led to a major break-through. Following up on that erroneous idea, Henle and Welch treated a leukemia patient with folic acid, thinking that the leukemia was caused by a folic acid deficiency. Instead of slowing the disease, however, folic acid caused it to progress even faster. Well, they thought, if folic acid speeded up the disease, maybe folic acid deficiency would slow it down. Indeed, when they treated another leukemia patient with a crude folic acid antagonist, there was a dramatic reduction in the number of leukemia cells in the blood. Henle and Welch published this observation in a very brief report in 1948 (Heinle and Welch, 1948). It was the first clue that folic acid antagonists could suppress the progress of leukemia. That brief report, spurred chemists at Lederle Laboratories to synthesize new folic acid antagonist. The most potent of these was aminopterin, which was the drug Sidney Farber used in his landmark findings in the treatment of childhood leukemia 112 K.W.Kohn Drugs Against Cancer CHAPTERS -- which was also published in 1948, showing how quickly a preliminary observation led to a substantial clinical result. It was not yet a cure, however, because the patients inevitably relapsed, and their leukemia then no longer responded to the drug. This experience however provided a foundation for the eventual cure of acute leukemia in children. Figure 5.1. Sidney Farber (1903-1973), discoverer of aminopterin and methotrexate as effective drugs for the treatment of acute leukemia in children. Although they did not cure, the drugs did temporarily shut down the disease and prolonged life. Aminopterin's action against childhood leukemia was soon confirmed and extended to leukemia in adults, as well as solid tumors, such as breast cancer ( reviewed by Farber and by Dameshek in 1949 (Farber, 1949) (Dameshek, 1949)). The speed of this progress in discovery and clinical application is notable, especially when compared with the delays and difficulties that new therapies now often encounter (DeVita Jr., 2015). Still, temporary remissions in those early studies were achieved in only a fraction of patients, and at the cost of sometimes severe toxicity. At about the same time, Chester Stock and Abraham Goldin and their colleagues showed that aminopterin inhibited the growth of malignant tumors in mice (Schoenbach et al., 1949; Sugiura et al., 1949) (Figure 5.2). Moreover, the effect of the drug was prevented by folic acid, which supported the idea that aminopterin did in fact inhibit the tumor by competing with folic acid (Goldin et al., 1949). Aminopterin differed from folic acid only in that an oxygen atom was replaced by an amino group (Figure 5.3). It is now known that aminopterin or methotrexate 113 K.W. Kohn Drugs Against Cancer CHAPTERS compete with an active form of folic acid for binding to two critical enzymes, as will be explained later in this chapter. Figure 5.2. Effect of aminopterin on a mouse tumor (sarcoma 180). Left, before treatment; right, after treatment After treatment, the tumor cells were dying and disintegrating (240X) (Sugiura et al., 1949). Abraham Goldin and his coworkers at NCI found that amethopterin (methotrexate) had a better therapy-versus-toxicity ratio than aminopterin. Therefore, in 1956, methotrexate replaced aminopterin in treatment of patients. Aminopterin and methotrexate are chemically and pharmacologically very similar; however it seems that the two drugs may never have been compared head-to-head in human patients (Bertino, 1993). Interestingly, methotrexate had better anti tumor properties (in animals), despite being much less potent (a higher dose was needed) than aminopterin (Ferguson et al., 1950)). Both aminopterin and methotrexate killed most of the leukemia cells, but also depleted the bone marrow of the normal red blood cells, white blood cells, and platelets needed to prevent anemia, fight infection, and prevent bleeding (Thiersch, 1949). Therefore, the normal bone marrow was given time to recover between doses of the drug, which was a major advance in effectiveness of the drugs. Another major step toward the eventual cure of childhood leukemia was the development of platelet transfusion, which prevented bleeding during the time required for the bone marrow to recover. This critical development was spearheaded by Emil J Freireich at the National Cancer Institute. Methotrexate by itself produced remissions that only lasted several months to about a year. Life was prolonged, but the leukemia invariably recurred and no longer responded to the drug. Cancer researchers however were relentless in their quest to 114 K.W.Kohn Drugs Against Cancer CHAPTERS cure the disease; it was a long struggle, but over the next three decades they succeeded in doing so. Folic acid antagonists were an essential part of the story, but eventual success required the careful design of therapy using multi-drug combinations, as well as platelet transfusions and bone marrow implants (DeVita Jr., 2015; Laszlo, 1995). The road to the cure of childhood leukemia was a long and difficult struggle. Some clinicians in the 1940's and 1950's felt that the children should be allowed to die in peace, rather than being subjected to the additional discomforts of drug toxicities and the pain of bone marrow aspirations that were needed to gage the effects of the drugs. Even in 1957, when I arrived at NCI and served on the childhood leukemia ward, some of my fellow Clinical Associates felt that way and at least one of my close friends refused to serve on the cancer wards, because he felt that some of the research was unethical. However, if left untreated, these children were all fated soon to die of their disease, and many parents felt that anything was worth a try. We did succeed in temporarily suppressing the disease with methotrexate, as Farber had described, as well as with other drugs that were being tried. My clinical associate colleagues on the NCI cancer wards in the late 1950's however would have been surprised, as I myself was, that the clinical studies of those early days were the beginning of a path that really did lead to a cure. Figure 5.3. Simple modifications of folic acid yielded the anticancer drugs aminopterin and amethopterin (now called methotrexate). Placing an amino group (NH2) in place of the oxygen on the pteridine ring of folic acid yielded aminopterin; further addition of the methyl (CH3) group (encircled red) yielded methotrexate. The replacement of the pteridine oxygen by an NH2 group, caused the molecule to become an antagonist of folic acid: it inhibited the actions of folic acid and put a monkey wrench (the English might say "spanner") into the mechanisms where folic acid was critical. 115 K.W.Kohn Drugs Against Cancer CHAPTERS Methotrexate cures choriocarcinoma. Only 2 years after Abraham Goldin's discovery of the superior effectiveness of methotrexate, Roy Hertz and his colleagues at NCI reported that methotrexate was remarkably effective against choriocarcinoma, a rapidly fatal cancer arising from embryonic tissues of the placenta in pregnant women (Figure 5.4) (Hertz et al., 1961; Hertz et al., 1956; Li et al., 1958). Methotrexate's dramatic cure of many cases of choriocarcinoma was soon confirmed by James Holland at the Roswell Park Memorial Institute in Buffalo, New York (Holland, 1958). Methotrexate, given in the appropriate dose schedule, cured most of the patients, even if the tumor had already metastasized (Hertz et al., 1964). The reported cure of a metastatic cancer astounded many cancer researchers who at first found it hard to believe. Choriocarcinoma was the first malignant tumor to be cured by chemotherapy, and, most remarkably, it could be cured with the administration of just a single drug, an antifol such as methotrexate. Chemotherapy worked so well against choriocarcinoma, because the cells derive from the embryo, which is a foreign tissue, as far as the patient's immune system is concerned. After methotrexate killed most of the rapidly dividing cancer cells, the remainder were often mopped up by the patient's immune system reacting against the choriocarcinoma cells that are genetically derived from the embryo. The immune system sees this cancer as foreign tissue, because mother and child are not genetically identical: half of the embryo's genes come from the father. Figure 5.4. Choriocarcinoma, a malignant tumor that methotrexate cured. It usually arose in the placenta of pregnant women and was made up of wildly growing cells of various sizes and shapes. Some, such as the large dark one in near the center, have several nuclei. These were the cells that produced chorionic gonadotropin (HCG) in 116 K.W. Kohn Drugs Against Cancer CHAPTERS the placenta, as well as in the tumor. When the HCG hormone in the blood declined to undetectable levels, it was a sign ofresponse and eventual cure. How methotrexate works -- Overview Skipping the details for now, the essential point is that methotrexate inhibits the synthesis of DNA: it prevents the chromosomes from being duplicated for cell division. In other words, it blocks the step in the cell division cycle where DNA has to be duplicated. True, it is good to block the division of leukemia or tumor cells, but normal cells in some important tissues also have to duplicate at a high rate and inhibiting those cells often caused major problems. The normal tissues most sensitive to blockage of cell division by methotrexate were the rapidly dividing blood-forming cells in the bone marrow and in the lining ("mucosa") of the intestines. In some malignancies, particularly leukemias, tumor cells can enter the brain, where methotrexate is kept out by the blood-brain barrier. The drug was therefore also injected into the cerebrospinal fluid by way of a spinal tap, in order to kill tumor cells that may be lurking in the central nervous system (Whiteside et al., 1958). Folic acid was needed to produce the chemical building-blocks required to make DNA To do so, however, the folic acid molecule had to be altered, first by addition of two hydrogen atoms to produce dihydrofolate, and then addition of two more hydrogen atoms to produce tetrahydrofolate. The latter is the reaction step that methotrexate blocks (Figure 5.5). The enzyme that carries out this reaction is dihydrofolate reductase (DHFR), and it is this enzyme that methotrexate bound and blocked. Methotrexate usually had to be combined with other anticancer drugs to have lasting benefit. There was one type of cancer however that was cured by methotrexate alone, and that was choriocarcinoma. This rare cancer, as already mentioned, occurs during pregnancy from cells in the placenta of the embryo. The cells of this form of cancer divide rapidly, which is one reason that this cancer responds so well to a DNA synthesis inhibitor such as methotrexate. Before methotrexate, metastatic choriocarcinoma was fatal in 90% of cases (Yarris and Hunter, 2003). 117 K.W. Kohn Drugs Against Cancer CHAPTERS H I oo'(o d 0 0 . . • •• 7~ ' H N •~ H 0 H,N ) LN I H N I H ~-y ~ · '-' · N I H Dihydrofolate (FH2) ~( NADP' !y' NADPH DHFR H 0 r·· d H 0 0 7~ ' H H N II ' N """N 0 N N . .~.. . .• ·. ·•. ·· N ~ A A A Tetrahydrofolate (FH4) Figure 5.5. An essential reaction that methotrexate inhibits. The reaction is carried out by the enzyme dihydrofolate reductase (DHFR), which methotrexate binds and blocks. What DHFR does is to convert the double-bond enclosed by the dashed oval (upper structure) to a single bond (lower structure). This involves adding 2 hydrogens (not shown). The product, tetrahydrofolate, is required for the manufacture of building-blocks for DNA. How methotrexate kills cancer cells. Methotrexate, like folic acid, enters the cell by way of channels through the cell surface membrane. Cancer cells that have too few of those channels don't let much methotrexate in and therefore do not respond well to the drug (Chen et al., 2013). Once inside the cell, an enzyme adds several more glutamates to the end of the methotrexate molecule, and this polyglutamated form cannot exit from the cell, because the glutamates bear negative charges, and electrical charge impairs the ability of molecules to pass through the cell surface membrane (Figure 5.6). Moreover, molecular pumps that pump many drugs out of the cell do not work with the polyglutamated form ofmethotrexate (Chen et al., 2003). That was important because drug resistance was often caused by increased quantities of those drug efflux pumps, and this resistance mechanism would not work to remove the polyglutamated methotrexate from the cell. Thus, the polyglutamate avoided this common mechanism ofresistance (Szabo et al., 2016). 118 K.W. Kohn Drugs Against Cancer CHAPTERS As usual, however, the real-life situation was more complicated: there were enzymes in the cell that removed the extra glutamates; folic acid competes with methotrexate for the polyglutamating enzyme; methotrexate polyglutamate drugs cannot be gotten directly into the cell, because they will not pass through the cell membrane (Szabo et al., 2016). Also, when methotrexate reduces the amount of thymidylate in the cell, a feedback signal initiates an attempt to compensate by making more dihydrofolate reductase enzyme (Rushworth et al., 2015). Strategies were being developed to overcome these problems. Methotrexate then binds and inhibits the key target enzyme, dihydrofolate reductase (Volpato and Pelletier, 2009). The cell needs these enzymes to produce components required for DNA synthesis, particularly thymine, adenine, and guanine. In a little more detail, here are the steps that were found to be relevant for the action of methotrexate: • In the cell, folic acid (folate) readily picks up 2 hydrogen atoms to become dihydrofolate (FH2). • Dihydrofolate reductase (DHFR) then adds 2 more hydrogens to FH2 to form tetrahydrofolate (FH4) (Figure 5.5). Methotrexate binds to and inhibits DHFR and therefore blocks this reaction. What happens is that the drug binds to the folate binding site on the enzyme and prevents normal folate from coming in and binding there (Volpato and Pelletier, 2009). • Another enzyme in the cell adds a methyl group to FH4 to form methylene- tetrahydrofolate (meFH4), a very important molecule that makes methyl groups available for the syntheses of thymine, adenine, and guanine. • meFH4 provides a methyl group for the enzyme thymidylate synthase to make thymine from uracil (the enzyme converts deoxyuridine phosphate to thymidine phosphate). Since methotrexate inhibits dihydrofolate reductase, the production of FH4 needed to make meFH4 is blocked. Without meFH4, thymidylate synthase function and the production thymine components for DNA are impaired. Moreover, methotrexate also blocks thymidylate synthase directly, which more completely inhibits thymidylate production. Result: by inhibiting dihydrofolate reductase and thymidylate synthase, methotrexate blocks the production of thymine, adenine, and guanine components for DNA synthesis (Fang et al., 2016). 119 K.W.Kohn Drugs Against Cancer CHAPTERS Figure 5.6. Methotrexate polyglutamate. Three glutamates are shown, but there can be as many as 8. Note the negative charges on the -CO2's, which prevent the molecule from passing through the cell's surface membrane. When the glutamates are added inside the cell, the polyglutamate methotrexate cannot exit from the cell. How cells become resistant to methotrexate. Drug resistance, either intrinsic to the tumor, or acquired through selective proliferation of resistant cells, was the major bugaboo of chemotherapy. Resistance to methotrexate was found to be due to any of several factors; some of the best understood were the following (Walling. 2006): • Methotrexate uptake channels that are too few in number or that have a reduced binding affinity for the drug; the drug then cannot enter the cell (Sirotnak et al., 1968). • Reduced addition of glutamates to the methotrexate molecule, thereby reducing the retention of the drug in the cell (Chen et al., 2003). • Increased activity of the enzyme that removes the extra glutamates from methotrexate polyglutamates inside the cell. Without the extra glutamates, the drug can escape from the cell. • Reduced binding affinity by mutation of the dihyrofolate reductase enzyme for methotrexate (Volpato and Pelletier, 2009). Methotrexate would then be unable to inhibit the enzyme. • Overproduction of dihydrofolate reductase (DHFR) by amplification of the gene, i.e., by an increase in the number of copies of the gene in the cell's chromosomes (Flintoff et al., 1982). The methotrexate would then be unable to block all of the increased amount of DHFR inside the cell. This shows how complex the problem of overcoming drug resistance can be. Most of the changes causing drug resistance were due to mutations, which were much more frequent in cancer cells than in normal cells. Only the more resistant tumor cells survived, but these could keep on dividing to form cancers that did not respond to the drug. 120 K.W. Kohn Drugs Against Cancer CHAPTERS Amplification of the DHFR gene in homog eneously s taining regions (HSR) of chromosomes. A striking and unexpected observation was made in 1976, by June Biedler and Barbara Spengler at Memorial Sloan-Kettering Cancer Center in New York. They were examining the chromosomes of cells that had been made resistant to methotrexate or other anti-folate drugs. I imagine that it might have been a surprise, or perhaps even a shock, to see that among the cell's chromosomes there was one that was greatly elongated. The reason for its greater length appeared to be that the chromosome had an insertion of a long region that was devoid of the usual banding pattern, a region that they therefore dubbed "homogeneously staining region" (HSR) (Figure 5.7). They surmised correctly that the HSR contained or was made up of a huge number of DHFR genes - which was the cause of the cell's drug resistance (Biedler and Spengler, 1976). The story was confirmed by Jack Nunberg and coworkers at Columbia University in 1978 (Figure 5.8) (Nunberg et al., 1978). HSR's have since been found in chromosomes of many cancers, generally associated with drug resistance attributable to a gene amplified in the HSR • Figure 5.7. An example of a homogeneously staining region (HSR) (arrow) in a chromosome of a cancer cell observed by Biedler and Spengler in 1976. An HSR was presumed to be an amplification of a gene, resulting in drug resistance (Biedler and Spengler, 1976). 121 K.W. Kohn Drugs Against Cancer CHAPTERS HSRl Figure 5.8. Amplification of the DHFR gene in a homogeneously staining region (HSR) in a chromosome of a methotrexate-resistant Chinese hamster cell reported by Nun berg and coworkers in 1978. The HSR contained a huge number of DHFR genes, which greatly extended the length of the chromosome and caused the resistance to the DH FR-inhibitor drug. The corresponding normal chromosome is on the right (Nun berg et al., 1978). Leucovorin comes to the rescue. Chemotherapy with methotrexate often required high dosage that produced troubling toxicity, especially to the bone marrow and gastrointestinal tract Fortunately, an antidote was available: leucovorin (also known as folinic acid or citrovorum factor) (Flombaum and Meyers, 1999; Schoenbach et al., 1950) (Figure 5.9). Patients could tolerate up to SO-fold higher methotrexate doses if leucovorin was administered within 24-48 hours (Frei et al., 1980). This so called "high-dose methotrexate/leucovorin rescue" regimen given at weekly intervals was found effective, especially in cancers that do not take up methotrexate well; the high dose helps to push the drug into the cells. How much better this regimen was than methotrexate by itself, however, remained uncertain (Frei et al., 1980; Zelcer et al., 2008). Methotrexate inhibited DNA synthesis by blocking both dihydrofolate reductase and thymidylate synthase. These inhibitions could be reversed by administering leucovorin (Schoenbach et al., 1950). Therefore, when high doses of methotrexate were needed for effective anticancer treatment, leucovorin successfully countered methotrexate's major toxic effects on the bone marrow, gastrointestinal tract and kidney, as well as toxicity to the brain and spinal cord, particularly if the drug was administered into the spinal fluid to kill cancer cells lurking in the central nervous system (Whiteside et al., 1958). Leucovorin provided excess dihydrofolate (FH2), which circumvented the methotrexate-blocked reaction steps (Howard et al., 2016). 122 K.W. Kohn Drugs Against Cancer CHAPTERS Leucovorin (Folinic acid) Figure 5.9. Leucovorin (folinic acid, also known as citrovorum factor) is a natural active derivative of folic acid. It differs from folic acid in having 4 hydrogens added to make single-bonds (arrows) from the double-bonds in folic acid. There is also a C=O addition (red oval), which makes a methyl group available for the synthesis of thymine, adenine, and guanine (in a manner similar to the case of meFH4 described in Chapter 6). Platelet transfusion to control bleeding becomes essential in the search for a cure. In the advanced stages of acute leukemia, the normal bone marrow cells become replaced by leukemic cells. A life-threatening consequence was that not enough platelets were made to control bleeding. and the patient was in danger of bleeding to death. That danger limited the amount of drug that could be safely administered. When the problem of how to transfuse fresh platelets was solved, the amount of drug that could be safely administered was increased. Platelet transfusion was essential for patient to survive the dosage of the drug combinations that were needed for cure. Here is how platelet transfusion became possible: Much of the credit goes to Emil J Freireich ("Jay"), whose personality, determination, and thinking outside the box is entwined in the story. Freireich's remarkable career and accomplishments was described in poignant detail by John Laszlo (Laszlo, 1995). Freireich came to the National Cancer Institute shortly after the NIH Clinical Center was opened in 1953. Since he had trained in hematology, Gordon Zubrod asked him to start a Leukemia program. Emil J Freireich immediately met Emil ("Tom") Frei 111, who directed the NCI's clinical program and whose office was next door. The 123 K.W. Kohn Drugs Against Cancer CHAPTERS remarkable coincidence of the similarity of their names caused some confusion. However, Tom was precise and systematic in contrast to Jay's predilection for "wild" ideas, which he pursued relentlessly, and which often worked out Their names were always "Tom" and "Jay"; their common name "Emil" was never used. They complemented each other and their collaboration worked extraordinarily well. Their different personalities and ways of thinking were very evident when I served as a Clinical Associate on the Childhood Leukemia Service in 1957. Tom Frei impressed me in the scope of his knowledge. He always had a stack of punched cards in his long white coat and answered my questions with reference to published evidence. Jay had some extraordinary idea that I had difficulty accepting; some of them however led to important breakthroughs, such as the way concentrated blood platelets could be stored for transfusion. Combination chemotherapy including methotrexate cures childhood leukemia. "Full speed ahead and damn the torpedoes." Freireich was full of'crazy ideas' for new treatments to try. Gordon Zubrod as head of the Medicine Branch was often skeptical, but nevertheless often supported him, because the outlook for the children was so bleak. Zubrod's instincts bore fruit as Emil J Freireich ("Jay") was to deserve much of the credit for the first cures of childhood leukemia. The details of how childhood leukemia was eventually cured is told in the book by John Lazio, which also describes the personalities who made it possible (Laszlo, 1995). Methotrexate was a key part of the drug combination that enabled the cures. The stories of the other anti-cancer drugs that made up the first successful combination are told in their respective chapters: vincristine, Chapter 10; amethopterin (methotrexate), this chapter; 6-mercaptopurine, Chapter 7; and prednisone. The therapy was named VAMP, a combination of the first letters of the aforementioned drug names. By 1962, Jay Freireich and Tom Frei (Emil Frei) had decided that it made sense to combine some of the drugs that individually had shown some activity. They eventually sought to combine 4 drugs with different mechanisms of action, but in addition avoiding multiple drugs having the same toxicity. The idea was to attack the leukemic cells from different directions, while avoiding synergistic toxicity. The idea nevertheless struck some clinicians as far-fetched or even crazy (Laszlo, 1995). Some thought the studies were needlessly poisoning the very sick children. Nevertheless, Jay Freireich's dogged determination, much to the surprise of many, cured some of the children. 124 K.W. Kohn Drugs Against Cancer CHAPTERS References Bertino, J.R. (1993). Karnofsky memorial lecture. Ode to methotrexate. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 11, 5-14. Biedler, J.L., and Spengler, B.A. (1976). Metaphase chromosome anomaly: association with drug resistance and cell-specific products. Science 191, 185-187. Chen, C., Ke, J., Zhou, X.E., Yi, W., Brunzelle, J.S., Li, J., Yong. E.L., Xu, H.E., and Melcher, K. (2013). Structural basis for molecular recognition of folic acid by folate receptors. Nature 500, 486-489. Chen, Z.S., Robey, R.W., Belinsky, M.G., Shchaveleva, I., Ren, X.Q., Sugimoto, Y., Ross, D.D., Bates, S.E., and Kruh, G.D. (2003). Transport of methotrexate, methotrexate polyglutamates, and 17beta-estradiol 17-(beta-D-glucuronide) by ABCG2: effects of acquired mutations at R482 on methotrexate transport Cancer research 63, 4048-4054. Dameshek, W. (1949). The use of folic acid antagonists in the treatment of acute and subacute leukemia. Blood 4, 168-171. DeVita Jr., V.T. (2015). The Death of Cancer (New York: Farrar, Straus and Giroux). Fang. M., Hutchinson, L, Deng. A., and Green, M.R. (2016). Common BRAF(V600E)- directed pathway mediates widespread epigenetic silencing in colorectal cancer and melanoma. Proceedings of the National Academy of Sciences of the United States of America 113, 1250-1255. Farber, S. (1949). Some observations on the effect of folic acid antagonists on acute leukemia and other forms of incurable cancer. Blood 4, 160-167. Farber, S., and Diamond, L.K. (1948). Temporary remissions in acute leukemia in children produced by folic acid antagonist, 4-aminopteroyl-glutamic acid. The New England journal of medicine 238, 787-793. Ferguson, F.C., Jr., Thiersch, J.B., and Philips, F.S. (1950). The action of 4-amino-Nl0- methyl-pteroylglutamic acid in mice, rats, and dogs. The Journal of pharmacology and experimental therapeutics 98, 293-299. Flintoff, W.F., Weber, M.K., Nagainis, C.R., Essani, A.K., Robertson, D., and Salser, W. (1982). Overproduction of dihydrofolate reductase and gene amplifica tion in methotrexate-resistant Chinese hamster ovary cells. Molecular and cellular biology 2, 275-285. Flombaum, C.D., and Meyers, P.A. (1999). High-dose leucovorin as sole therapy for methotrexate toxicity. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 17, 1589-1594. Franklin, A.L., Stokstad, E.L., and et al. (194 7). Biochemical experiments with a synthetic preparation having an action antagonistic to that of pteroylglutamic acid. The Journal of biological chemistry 169, 427-435. Frei, E., 3rd, Blum, R.H., Pitman, S.W., Kirkwood, J.M., Henderson, J.C., Skarin, A.T., Mayer, R.J., Bast, R.C., Garnick, M.B., Parker, L.M., et al. (1980). High dose methotrexate with leucovorin rescue. Rationale and spectrum of antitumor activity. The American journal of medicine 68, 370-376. 125 K.W.Kohn Drugs Against Cancer CHAPTERS Goldin, A., Goldberg, B., and et al. (1949). Reversal of aminopterin-induced inhibition of sarcoma 180 by folic acid. Cancer 2, 857-862. Goodman, L.S., Wintrobe, M.M., Dameshek, W., Goodman, M.J., Gilman, A., and McLennan, M.T. (1946). Nitrogen mustard therapy. Use ofmethyl-bis(beta- chloroethyl)amine hydrochloride for Hodgkin's disease, lymphosarcoma, leukemia and certain allied and miscellaneous disorders. J Am Med Assoc 132, 126-132. Heinle, R.W., and Welch, A.D. (1948). Experiments with pteroylglutamic acid and pteroylglutamic acid deficiency in human leukemia. J Clin Invest 27, 539. Hertz, R., Lewis, J., Jr., and Lipsett, M.B. (1961). Five year's experience with the chemotherapy of metastatic choriocarcinoma and related trophoblastic tumors in women. American journal of obstetrics and gynecology 82, 631-640. Hertz, R., Li, M.C., and Spencer, D.B. (1956). Effect of methotrexate therapy upon choriocarcinoma and chorioadenoma. Proc Soc Exp Biol Med 93, 361-366. Hertz, R., Ross, G.T., and Lipsett, M.B. (1964). Chemotherapy in Women with Trophoblastic Disease: Choriocarcinoma, Chorioadenoma Destruens, and Complicated Hydatidiform Mole. Annals of the New York Academy of Sciences 114, 881-885. Hoftbrand, A.V., and Weir, D.G. (2001). The history offolic acid. British journal of haematology 113, 579-589. Holland, J.F. (1958). Methotrexate therapy of metastatic choriocarcinoma. American journal of obstetrics and gynecology 75, 195-199. Howard, S.C., McCormick, J., Pui, C.H., Buddington, R.K., and Harvey, R.D. (2016). Preventing and Managing Toxicities of High-Dose Methotrexate. Oncologist 21, 1471-1482. Laszlo, J. (1995). The Cure of Childhood Leukemia: into the Age of Miracles. (New Brunswick, New Jersey, USA: Rutgers University Press). Li, M.C., Hertz, R., and Bergenstal, D.M. (1958). Therapy of choriocarcinoma and related trophoblastic tumors with folic acid and purine antagonists. The New England journal of medicine 259, 66-74. Nunberg, J.H., Kaufman, R.J., Schimke, R. T., Urlaub, G., and Chasin, L.A. (1978). Amplified dihydrofolate reductase genes are localized to a homogeneously staining region of a single chromosome in a methotrexate-resistant Chinese hamster ovary cell line. Proceedings of the National Academy of Sciences of the United States of America 75, 5553-5556. Rushworth, D., Mathews, A., Alpert, A., and Cooper, L.J. (2015). Dihydrofolate Reductase and Thymidylate Synthase Transgenes Resistant to Methotrexate Interact to Permit Novel Transgene Regulation. The Journal of biological chemistry 290, 22970-22976. Schoenbach, E.B., Goldin, A., and et al. (1949). The effect of folic-acid derivatives on sarcoma 180. Cancer 2, 57-64. Schoenbach, E.B., Greenspan, E.M., and Colsky, J. (1950). Reversal of aminopterin and amethopterin toxicity by citrovorum factor. J Am Med Assoc 144, 1558-1560. Sirotnak, F.M., Kurita, S., and Hutchison, D.J. (1968). On the nature of a transport alteration determining resistance to amethopterin in the L1210 leukemia. Cancer research 28, 75-80. 126 K.W.Kohn Drugs Against Cancer CHAPTERS Sugiura, K., Moore, A.E., and Stock, C.C. (1949). The effect of aminopterin on the growth of carcinoma, sarcoma, and melanoma in animals. Cancer 2, 491-502. Szabo, I., Orban, E., Schlosser, G., Hudecz, F., and Banoczi, Z. (2016). Cell-penetrating conjugates of pentaglutamylated methotrexate as potential anticancer drugs against resistant tumor cells. Eur J Med Chem 115, 361-368. Thiersch, J.B. (1949). Bone-marrow changes in man after treatment with aminopterin, amethopterin, and aminoanfol; with special reference to megaloblastosis and tumor remission. Cancer 2, 877-883. Volpato, J.P., and Pelletier, J.N. (2009). Mutational 'hot-spots' in mammalian, bacterial and protozoa! dihydrofolate reductases associated with antifolate resistance: sequence and structural comparison. Drug Resist Updat 12, 28-41. Walling, J. (2006). From methotrexate to pemetrexed and beyond. A review of the pharmacodynamic and clinical properties of antifolates. Invest New Drugs 24, 37-77. Whiteside, J.A., Philips, F.S., Dargeon, H.W., and Burchenal, J.H. (1958). lntrathecal amethopterin in neurological manifestations of leukemia. AMA Arch Intern Med 101, 279-285. Yarris, J.P., and Hunter, A.J. (2003). Roy Hertz, M.D. (1909-2002): the cure of choriocarcinoma and its impact on the development of chemotherapy for cancer. Gynecologic oncology 89, 193-198. Zelcer, S., Kellick, M., Wexler, L.H., Gorlick, R., and Meyers, P.A. (2008). The Memorial Sloan Kettering Cancer Center experience with outpatient administration of high dose methotrexate with leucovorin rescue. Pediatric blood & cancer 50, 1176- 1180. 127 K.W. Kohn Drugs Against Cancer CHAPTER6 Chapter 6. The S-fluaraurocil Story 220927dh3 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@oib gov CHAPTER6 The 5-Fluorouracil Story: from a simple idea to a major anti-cancer drug.1 The molecule was simple, as was the idea it was based on, but its impact on cancer therapy was profound. The idea was merely to add a small fluorine atom to uracil, one of the building blocks for the production of RNA and DNA (Figure 6.1). The product, 5-fluorouracil (5FU), despite the minimal change in chemical structure, turned out to disturb DNA and RNA production in a surprisingly complex manner. 5FU was found to be toxic to rapidly dividing cancer cells, and the effects of 5FU on the RNA and DNA synthesis pathways were worked out, but exactly how these actions produced the drug's anti-cancer activity remained obscure. Discovery through knowle dge and intuition 5-Fluorouracil (5FU) was one of the first anti-cancer drugs to be discovered (after nitrogen mustard and methotrexate) and one of the most important. The story began in 1954 at the McArdle Memorial Laboratory in Madison, Wisconsin, with the discovery by Charles (Charlie) Heidelberger that adding a fluorine atom to uracil yielded a compound that had anti-cancer activity in mice and rats (Heidelberger et al., 1957) (Figure 6.2). The fluorine atom was cleverly placed at the 5-position, which, as we shall see, turned out to be critical for its anti-cancer action. The intuition to place a fluorine atom on the 5-position, came from the idea that thymine, a critical constituent of DNA but not RNA, had a methyl group on that 5-position of uracil: the cell made thymine from uracil by adding a methyl group to uracil 1 Reproducti on of some of t he figures in t his chapter may need publisher's permission. 128 K.W. Kohn Drugs Against Cancer CHAPTER6 (actually, by adding a methyl to a nucleotide of uracil to make a nucleotide of thymine). Charlie Heidelberger, son of the famous immunologist Michael Heidelberger, was a major figure in cancer research, noted for his sharp mind and demand for research excellence, until his untimely death at the age of 63, ironically of cancer. He was much loved and respected, but also feared by some younger researchers, because of his challenging questions and comments when in sessions that he chaired. (1 was in fact one of his victims in 1961, when, during the discussion period after my talk, I claimed too much future progress, whereupon he said, "I congratulate Dr. Kohn ahead of time for his future success." It was a lesson I never forgot.) Experience had shown that adding a fluorine atom, even though it is one of the smallest atoms, sometimes changed the properties of a drug. With that in mind, Heidelberger focused on uracil as a target compound, because of an idea that was circulating at the time that cancers might incorporate uracil into RNA more actively than did normal tissues. But, where on the uracil molecule would he add the fluorine? Insightfully, as already mentioned, he chose the 5-position, because that is where thymine had a methyl group (Figure 6.1), and he may have conjectured that the fluorine might then somehow interfere with the production or function of thymine, an essential component of DNA. He was absolutely right! But the way 5FU caused its effects turned out to be much more complicated than anticipated (Longley et al., 2003). When clinical investigators at McArdle in 1958 in the first clinical trial (prior FDA approval was not yet required), administered 5FU to patients who had a variety of cancers, they found that some of the patients' tumors showed signs ofregression, but this occurred only when the drug dose was high enough to produce severe toxicity (Curreri et al., 1958). (Note how quickly - within just a few years - discovery progressed to successful testing in patients.) In 1962, clinical researchers at the University of Wisconsin reviewed the clinical experience with 5FU up to that time. They noted that, although some of the published reports failed to find significant benefit, taking all the results together and focusing on patients whose tumors were large enough to be measured, about 21 % of patients who received the drug to the point of mild toxicity had a reduction in the size of the tumor, and that the remission lasted an average of 9 months (Ansfield et al., 1962). At the time, that was a significant effect against hard-to-treat cancers. Extensive clinical studies were then carried out to determine the best dosage schedule (the timing and amount of successive doses) in various types of cancer, and it was found that the drug was especially effective against colon cancer (Ansfield, 1964) (Ansfield et al., 1977). We will see later in this chapter how new therapies based on 5FU had much better outcomes. 129 K.W.Kohn Drugs Against Cancer CHAPTER6 0 0 0 CNH ''CNH H NA ,C O 'C I N AO N AO NH H H H Uracil 5-Fluorouracil Thymine Figure 6.1. Uracil is a building block for the production of RNA; in the cell, a methyl group (CH3) is added at position 5 of uracil, producing thymine building blocks for DNA 5-Fluorouracil (5FU) is simply uracil with a tluorine atom (F) added at the location (position 5) where thymine has a methyl group. Figure 6.2. Charles ("Charlie") Heidelberger (1920-1983). (Phow from a Biographical Memoir, National Academy of Sciences, Washington DC, 1989).} Improving the effectiveness of5-fluorouracil (SFU) in cancer treatment. Since the therapeutic ability of 5-FU by itself was meager ( response rates of only 10 to 15% in colon cancer (Longley et al., 2003)), much effort was put into making the drug more effective. The first improvement was to use leucovorin (folinic acid) (see Chapter 5), the natural form of folic acid, together with 5FU. That combination enhanced 5FU's cell killing action. The cell converts leucovorin to methylene- tetrahydrofolate, which is required at high concentration to effect the binding of 5FU to thymidylate synthase (Longley et al., 2003). How leucovorin acts together with 5FU to inhibit thymidylate synthase will be explained later in this chapter. The addition ofleucovorin to 5FU doubled the response rate of colon cancer, compared to 5FU alone (from about 11% to about 23%), but it unfortunately had little effect on survival time (Longley et al., 2003). Killing most of the cancer cells 130 K.W.Kohn Drugs Against Cancer CHAPTER6 was able to improve quality of life temporarily, but the remaining malignant cells eventually grew into new tumors that no longer responded to the drugs. In another report, adding Ieucovorin to SFU significantly increased the survival of patients with advanced colon cancer with distant metastases (stage IV). However, the chance of surviving one year with either treatment was not very good, and the chance of surviving 2 years was dismal (Figure 6.3) (Poon et al., 1989). 100 80 na 70 60 *Q) .:1: <( 40 SFU alone 20 0 0 6 12 18 24 Time from randomization, mo Figure 6.3. Survival of colorectal cancer patients who already had metastases (stage IV) or whose tumor could not be removed by surgery. The horizontal axis shows number of months after treatment began. The survival of patients treated with 5- fluosoruracil (SFU) alone is shown by the lower curve. The upper curves show the survival of patients who received leucovorin (2 dosage levels) in addition to SFU (Poon et al., 1989). Very few patients survived for 24 months, regardless of which treatment was given. After surgical removal of a colon cancer that had not metastasized to distant sites, patients often continued to be treated with drugs, usually including SFU or one of its relatives ("adjuvant chemotherapy"). Patients who had extensive regional lymph node involvement, but no distant metastases (advanced stage 111) had a much better outlook than those whose cancer had already metastasized to distant sites. Their long-term survival after surgery followed by SFU plus leucovorin was about 50%, which was increased to about 60% if oxaliplatin also was added. Oxaliplatin (see Chapter 3) seemed to have been effective only on the more advanced parts of the cancer, because it did not benefit patients with earlier stages of the disease (Figure 6.4). 131 K.W. Kohn Drugs Against Cancer CHAPTER6 1.0 0.8 •••• co >- --2: :.= ~ •~~•••~:::::::::S~F~U~+~L~V~+o ............... = x:a:li:Pt.:____,._ ::, . - C/) .D co 0.6 SFU+LV ............................, =.D co 0 0.4 '- '- Q) a. n Events >- - FOLFOX4 229 90 0 0.2 •• LV5FU2 231 117 Log-rank P= .012 HR, 0.705; 95% Cl, 0.535 to 0.928 0 1 2 3 4 5 6 7 8 9 10 Tim e Since Enrollment (years) Figure 6.4. Long-term survival of advanced stage Ill colorectal cancer patients (tumor in many regional lymph nodes, but no distant metastases). Patients in this group who were treated with SFU+leucovorin had about 50% chance of surviving 10 years; if oxaliplatin was added to the adjuvant therapy of SFU+leucovorin, the survival probability rose to about 60%. Oxaliplatin however did not benefit patients with less advanced disease (Andre et al., 2015). How could one tell who needed adjuvant therapy (continued chemotherapy after complete surgical removal to the tumor) and who did not? A blood test was developed that was promising. The test used a highly sensitive DNA analysis method to detect cancer cells or their DNA in the blood. If cancer DNA was detected in the blood of stage II colon cancer patients, it was surmised that they would benefit from adjuvant chemotherapy (Figure 6.5) (Tie et al., 2016). However, a small fraction of the patients who had no detectable cancer DNA in the blood, did have recurrence of the cancer, and they might also have been helped by adjuvant chemotherapy. This is a situation where patients would be called upon to make the decision, based on their consideration of risk and toxicity versus benefit. 132 K.W.Kohn Drugs Against Cancer CHAPTER6 100 Postoperative ctDNA-negative (n = 164) .. 1 80 4 "eC 60 HR, 18 (95% Cl, 7.9- 40) ...." ~ '5 0) 40 ...."' E Ii 20 Postoperative ctDNA-positive (n = 14) a.. 0 0 12 24 36 48 60 Months since su-gery Figure 6.5. Patients with stage II colon cancer who did not receive adjuvant chemotherapy, and who had detectable cancer DNA circulating in the bloodstream, were at risk of soon having a recurrence of the cancer (lower curve, 14 patients). Similar patients who did not have detectable cancer DNA in the blood had a high likelihood of being cured (upper curve, 164 patients) (Tie et al., 2016). (Note that only 8% of patients had detectable cancer DNA in the blood, and these patients were at risk and would probably have benefited from adjuvant chemotherapy; most of those who did not have cancer DNA in the blood did not need adjuvant chemotherapy.) (www.ScienceTranslationalMedicine.org 6 July 2016 Vol 8 Issue 346 346ra92). Cape citabine : a pro-drug for SFU A difficulty for therapy with SFU was that it was rapidly destroyed by enzymes in the blood, and therefore had to be administered around the clock Much of the intravenously administered SFU was destroyed in the blood before the drug entered the cell. Capecitabine was developed as a pro-drug that is converted to SFU by enzymes in the cell. It was not destroyed in the blood, entered cells and only then was converted to SFU. Moreover, unlike SFU, it could be given orally, a major practical advantage during treatment. The new drug was itself inactive (it was a "prod rug") and had to be activated by reactions, first in the liver and then by enzymes that are highly active in some cancers. Thus the active SFU was generated right in the tumor cell (Miwa et al., 1998) (Johnston and Kaye, 2001). Capecitabine was inactive, because of the side chain shown at the top of the structure in Figure 6.6. Enzymes in the liver and cancer cell removed the side chain to yield bare 5-fluorouracil, which formed directly in the cell, thereby evading destruction by enzymes outside of the cell. 133 K.W.Kohn Drugs Against Cancer CHAPTER6 Despite its theoretical advantages, however, capecitabine produced only a modestly higher response rate than SFU in colon cancer with somewhat less toxicity. But it unfortunately had little effect on the survival time of patients (Longley et al., 2003). It nevertheless had the benefit of oral rather than prolonged intravenous administration. Figure 6.6. Structure of capecitabine, an inactive form of SFU (a pro-drug), which is converted by SFU by enzymes, first in the liver, and then in the cell. Active SFU thus forms directly in the cell. The activation occurs by removal of the inactivating chain from the amino group (NH) at the top. Capecitabine had the advantage that it was not destroyed in the blood and oral administration was effective. But the hoped-for benefit in terms of prolonging the lives of cancer patients was disappointing. How colon cancer came to b e treated with 5-Fluorouracil (SFU), We will look back now on the history of this dreadful disease and its treatment Colorectal cancer, at least up to 2015, was the 4th most common cancer, after prostate cancer in men, breast cancer in women, and lung cancer in both sexes. Also, it was the 3rd most frequent cause of cancer-related death in the United States. Figure 6.7 gives an idea of the culprit that had to be deal with; it shows a typical view (histology) of the cells in a moderately differentiated colon cancer, which was the most common type. Cancers that were poorly differentiated (few gland-like structures) were more aggressive and had a worse prognosis than tumors that were highly differentiated (many gland-like structures) (Fleming et al., 2012). 134 K.W.Kohn Drugs Against Cancer CHAPTER6 Figure 6.7. A typical picture of the culprit: a moderately differentiated colon carcinoma. Many of the cancer cells (cells with large nuclei) are arranged in a manner resembling the gland structure of the colon, but in a disorganized fashion: large-scale tissue order is lost. The tumors are surrounded by fibrous tissue cells (stroma), which may contribute to the malignancy of the cancer (Fleming et al., 2012). (Fleming M, Ravula S, Tatishchev SF, Wang HL. Colorectal carcinoma: Pathologic aspects.Journal ofGastrointestinal Oncology. 2012;3(3):153-173). Colon cancers were noted to be of two general types, depending on where in the colon they arose. From the point where the small intestine joins the colon in the lower right side of the abdomen, the colon ascends on the right, crosses over the midline and descends on the left side to the rectum. The first ("proximal") and second ("distal") parts of the colon along that path have differences that trace back to the way they form in the embryo (Bufill, 1990). The two sections of the colon are like different tissues and have different types of cancers with different drug sensitivities. Most colon cancers arise in pre-malignant outgrowths, called polyps; these are found only in the distal (descending on the left side) colon, and could be removed during colonoscopy. That was important, because it is within those polyps that the great majority (about 80%) of colon cancers developed (Figure 6.8). Fortunately, malignant cancers in those polyps take years to develop, which gave time for them to be removed during regular colonoscopy. However, a few people have a rare inherited mutation of the APC (adenopolyposis coli) gene that caused continuous formation of many polyps that had to be removed by frequent colonoscopies. 135 K.W.Kohn Drugs Against Cancer CHAPTER6 The cancers arising in polyps in the distal colon usually responded to 5FU. On the other hand, the less common cancers in the proximal (ascending on the right side) colon did not form polyps, and rarely responded to 5FU (Carethers et al., 2004; Kawakami et al., 2015). Another difference was that cancers in the proximal colon often had a mutation in one of the DNA mismatch repair (MMR) genes (discussed in Chapter 25). Pathways APC RAS Figure 6.8. How cancer develops from a polyp in the distal (descending) colon, as conceptualized by Bert Vogelstein (Vogelstein et al., 2013). Shown here are the stages from normal to cancerous polyps as conceived by Vogelstein. In the normal colonic epithelium (left), a polyp develops with a small adenoma in it (second picture from the left); this happens occasionally when there is a rare mutation in the APC (adenopolyposis coli) gene. After several years, cells in this still benign but pre- cancerous adenoma may acquire a mutation in the RAS gene, which then allows the tumor to grow to become a large adenoma (third picture from the left). Up to this stage, the tumor is still benign and could be removed during colonoscopy. After several more years, the tumor may invade deeper tissues (right) and become malignant and could metastasize. As long as the cancer remained local and without spread to the regional lymph nodes, it was stage II and could be cured by surgery with or without adjuvant drug therapy (continued drug therapy after surgical removal of the tumor). If it had spread to the regional lymph nodes, but not yet metastasized to other organs, it was stage III and could still often be cured by surgery followed by adjuvant drug therapy. For cancers that had metastasized, for example to the liver or brain (stage IV), there was no cure (Vogelstein et al., 2013). Knowledge of how cancers develop in polyps in the distal (descending. left side) colon reached the point where a large majority of malignant cancers that would arise in such polyps could be prevented by surgically removing the polyps. Cancer development in polyps that are not removed constituted the greatest risk of malignant colon cancer. Polyps were surmised to initiate due to a mutation in the APC gene, which normally functions to limit cell division in the colon's epithelium. A mutation in the APC gene inactivates this function and consequently allows cells to divide excessively, thereby producing a polyp (second panel from the left in Figure 6.8). Some of the cell's division controls however remains intact and puts a limit on the size to which the polyp could grow. The next step was found often to be a 136 K.W.Kohn Drugs Against Cancer CHAPTER6 mutation that over-activates a gene of the RAS family, which pushes cells to divide faster. That would cause the polyp to grow larger (third panel from the left in Figure 6.8). Another event that enhanced polyp growth was an inactivating mutation in the TP53 gene (the topic of Chapter 32), a gene that normally stimulated DNA repair and caused division-control-defective cells to commit suicide by apoptosis. This progression from early polyp to malignant cancer usually took many years, because mutations are rare events -- which was fortunate, because it allowed most colon cancers to be preventable by removing any polyps that may be seen during regular colonoscopies. A full-blown malignancy with distant metastases usually required several other mutations or gene function modifications; at that stage (stage IV), chemotherapy could prolong life, but was rarely curative. Colon cancers, as well as most other types of cancer, only became truly malignant after mutations disturbed several genes that together normally kept a cell from growing wild. Figure 6.9 shows one of the methods that were used to display the pattern of gene mutations occurring in particular types of cancer (Vogelstein et al., 2013; Wood et al., 2007). We see that, in colon cancers, the genes for APC, KRAS, and TP53 stood out, although there were also many rare mutations that together could be important (Figure 6.9). Figure 6.9. Genes that were found to be mutated in colon cancers. Three genes stood out: TP53, KRAS, and APC, and to a lesser degree PIK3CA and FBXW7. Less frequent mutations also occurred in many other genes. The gene mutation pattern differed from one type of cancer and another, although a few, in particular TP53, were found to be mutated in many or most cancer types. An APC gene mutation was common specifically in cancers arising in polyps in the descending colon. (The genes in this 2- dimensional landscape were arranged according their location on the chromosomes (Wood et al., 2007)). 137 K.W.Kohn Drugs Against Cancer CHAPTER6 How leucovorin enhances the anti-cancer action of SFU. Although leucovorin reversed the effects ofmethotrexate (see Chapter 5), in the case of SFU leucovorin was found to enhance the drug's effect on its key target: thymidylate synthase, an enzyme required for production of thymidylate that is required for DNA synthesis (Figure 6.10). The way in which SFU and leucovorin together conspire to permanently block the thymidylate synthase enzyme was reported in classic papers by Dan Santi (Santi and McHenry, 1972) (Santi, 1980) (Figure 6.10). First of all, leucovorin easily converts to the active form: 5,10-methylene-tetrahydrofolate (CH2-FH4). Normally, the thymidylate synthase enzyme binds both CH 2-FH4 and a uracil nucleotide and then transfers its CH2 group as a methyl group to the uracil 5-position, thereby converting the uracil nucleotide to a thymine nucleotide that is essential for DNA synthesis. SFU binds to the enzyme-CH2-FH4 combination just as well as uracil does, but the transfer of the CH2 to the uracil 5-position cannot proceed, because the fluorine atom tightly bound to that position. The enzyme becomes permanently trapped in a tight combination with CH2-FH4 and SFU -- and can no longer function of make the thymidine needed for DNA synthesis (Figure 6.10). It is beautiful how the tiny fluorine atom destroys the big enzyme. Leucovorin increased the amount of CH2-FH4 available, and thereby increased the rate or extent to which the thymidylate synthase enzyme binds SFU. Leucovorin thus enhanced the trapping of the enzyme in the complex with SFU. The result was that leucovorin enhanced the potency of SFU in blocking DNA synthesis and thereby killing cancer cells. A limitation, however, was that, by the same process, leucovorin also increased the toxicity SFU for rapidly dividing normal cells. Nevertheless, the net effect of leucovorin was to increase the anticancer effectiveness of SFU (Figure 6.3). 138 K.W. Kohn Drugs Against Cancer CHAPTER6 H H Tetrahydrofolate (FH4) H2N'f~NNx;_ HN I N H H-Oqo ~ /2 N C- Glv II 0 0 11 CH Methylene {CH2)connect ing ' JJ,,:,,~ 2 - tetrahydrofolate with SFU. Fluorine at om of SFU. Figure 6.10. How SFU, together with methylene-tetrahyrofolate (CH2-FH4, an activated form ofleucovorin) blocks the thymidylate synthase enzyme, as described by Dan Santi (the labels and lines in color were added to a figure copied from San ti's 1980 review paper (Santi, 1980)). Thymidylate synthase normally causes the CH2 group of methylene-tetrahydrofolate to link to uridine on the way to converting the uridine to thymidine (as nucleotides) for DNA synthesis. The enzyme reacts similarly with SFU, but the enzyme-SFU-CH2-FH4 complex (shown in the Figure) then remains trapped because of the fluorine on position-5 and can proceed no further. That is how SFU plus CH2FH4 (or leucovorin, which converts to CH2FH4), work together to block DNA synthesis. The thymidylate synthase enzyme binds, by way of a sulfhydryl group ('X' in the Figure), to the 6-position of the uracil, which puts the enzyme in position to carry out its normal work at the adjacent 5-position - - which however it cannot do when there is a fluorine atom there. How 5-jluorouracil (SFU) kills cancer cells. As described above, it is remarkable how much trouble a simple molecule like SFU can cause to a cell. But to understand it better, we have to delve further into the complicated reaction pathways that SFU gets into and messes up. First, in order to have any effects on a cell SFU has to get into it. How SFU enters the cancer cell. Since the cell needs a great deal of uracil for RNA and DNA syntheses, there are special transporter molecules that help move uracil rapidly in from outside the cell. Those same transporters allowed the cell to slurp up SFU, leading to high concentrations of SFU inside the cell. Cancer cells that grow rapidly need more 139 K.W.Kohn Drugs Against Cancer CHAPTER6 uracil and therefore make more transporter molecules, and these allow SFU to enter rapidly. To the transporters, SFU looks so much like normal uracil, that it moves SFU into the cell as easily as it does uracil. SFU stops the production of thymidylate for DNA synthesis. Once inside the cell, 5-fluorouracil (SFU) enters uracil's metabolic pathways, because the enzymes that catalyze those reactions act on SFU almost as well as on uracil (Figure 6.11). Again, the structure of SFU is so much like uracil that the enzymes don't distinguish between them. It is much like a Trojan horse: it looks like a gift: but turns out to be a poison (in German, Gift: means Poison). SFU becomes incorporated into DNA and RNA The most important action of SFU was found to be inhibition of thymidylate synthase and consequent inhibition of DNA synthesis (Figure 6.11). The thymine- containing building blocks for DNA are depleted and uracil-containing units accumulate. Because there then is little thymidylate available, uracil and 5 FU nucleotides, are mis-incorporated into newly synthesized DNA (Longley et al., 2003). Where there should be thymine in newly synthesized DNA, there often then is uracil or SFU. Thus, the scarcity of thymine units forces the DNA-synthesizing enzymes (DNA polymerases) to incorporate uracil or 5-fluorouracil (SFU) in place of thymine in DNA. Then, DNA repair enzymes come into play to remove the uracil and SFU from the DNA, so that they could be replaced by thymine. But there is insufficient thymine nucleotide to replace the mis-incorporated uracil or SFU with thymine efficiently! Hence, the DNA damage remains, and DNA functions are perturbed. The metabolic scheme in Figure 6.11, shows how SFU was thought (probably correctly) to become incorporated into DNA and RNA. SFU first combines with ribose-phosphates to form SFUDP (SFU-ribose-PP) and then SFUTP (SFU-ribose- PPP). The latter (SFUTP) then is incorporated into RNA. For incorporation into DNA, on the other hand, the ribose part first has to be changed to deoxyribose, which entails removing the hydroxyl group from the 3' position ofribose. This is accomplished by ribonucleotide reductase, which converts SFU-ribose-PP to SFU- deoxyribose-PP). Ribonucleotide reductase converts UDP (U-ribose-PP) to dUDP (U- deoxyribose-PP), which is an essential step for DNA synthesis. Because it is required for DNA synthesis, ribonucleotide reductase is itself an important anti-cancer drug target Hence SFU makes its way into both RNA and DNA and thereby messed up functions in both realms (Akpinar et al., 2015). These deleterious effects, especially those 140 K.W.Kohn Drugs Against Cancer CHAPTER6 messing up DNA occur mainly in the phase of the cell cycle where DNA is being replicated, i.e., during S phase. That is why SFU was often administered around the clock -- so as to give time for all the tumor cells to enter S phase, while the drug was still present and able to kill the cancer cells. SFU was most effective against leukemias and lymphomas, most of whose cells were actively progressing around the cell division cycle. Although much became known about what happens to SFU in the cell, as described above, exactly how this complicated network of reactions added up to toxicity for the cell was not completely worked out (Huehls et al., 2016). SFU kills cancer cells mainly by inhibiting thymine production (by inhibition of thymidylate synthase) and by becoming incorporated into DNA. Incorporation of SFU into RNA adds to the toxicity and under some conditions could be the main factor causing cell death (Geng et al., 2011; Longley et al., 2003; Pettersen et al., 2011). A closely related drug, SFU-deoxyribose (5-fluorodeoxyuridine, FdUR) simplified the situation a bit by becoming incorporated mainly into DNA and less into RNA. But FdUR unfortunately did not seem to be much better than SFU in cancer treatment experience. For many cancers, the situation was complicated by the fact that only a fraction of the cancer cells is in the cell-division cycle at any one time. Great effort was therefore made to pin down the details of how a population of cancer cells progress around the cell cycle. Those efforts however did not have much impact on treatment of the major solid tumors, such as lung, breast, and colon cancers. Those cancers are slow growing and only a small fraction of the cancer cells is dividing. For leukemias and lymphomas, however, most of whose cells are in the cycle, the detailed studies of the kinetics of the cell-division cycle did have a major impact A major finding was that the time-scheduling of the treatments was critically important. Much better than continuous treatment with DNA synthesis-inhibiting drugs, was intermittent treatment with rest periods inserted to allow the bone marrow to recover. This anticancer drug scheduling was used successfully by Vincent DeVita and his colleagues in curing patients with Hodgkins lymphoma and is described in DeVita's book (DeVita Jr., 2015). (Although SFU was not part of their 4-drug regimen, the principle was the same.) What happens to DNA that has mis-incorporated SFU and uracil in place ofthymine. Most of the mis-incorporated uracil and SFU is rapidly removed by special DNA repair enzymes: uracil DNA glycosylases, as well other kinds of DNA glycosylases that could remove uracil and SFU from DNA under different circumstances. (Huehls et al., 2016; Pettersen et al., 2011) (see Chapter 24). One of the DNA glycosylases 141 K.W. Kohn Drugs Against Cancer CHAPTER6 (MBD4) that helped to remove SFU from DNA was specialized to remove mismatched thymine that was occasionally produced in certain places in the genome by spontaneous deamination of 5-methyl-cytosine at CpG sequences in gene promoters (Suzuki et al., 2016). Some mis-incorporated nucleotides may however remain and cause trouble. SFU Le ukovorin Thymidyfate / ~ (Folinic acid) synthase SFlJ-ribose-P SFU-deoxyribos~ ! SFU-ribose-PP ---•► !i SFU-deoxyribose-PP SFlJ-rJ ose-PPP ! I SFU-deot yribose-PPP Ribonucleotide reductase ! Incorporation of Incorporation of DNA re plication SFU into RNA SFU into DNA and repair ! RNA damage ! DNA damage Figure 6.11. How 5-fluorouracil (SFU), a drug that simply has a fluorine atom added to uracil, has dramatic effects on the cell. SFU enters the metabolic pathways that normally process uracil. One path leads to DNA, another to RNA. When SFU enters these pathways, the result is DNA damage and disturbance of RNA function. In the DNA pathway, SFU combines with deoxyribose, which is specific for DNA, whereas in the RNA pathway SFU combines with ribose, which is specific for RNA. When SFU combines with deoxyribose-P, it inhibits thymidylate synthase (Figure 6.10), the enzyme that adds a methyl group to a uracil unit to convert it to thymine, an essential for DNA synthesis. That is how SFU exerts its major action: inhibition of DNA synthesis. Also required for DNA synthesis is the step where ribose is converted to deoxyribose, which is carried out by ribonucleotide reductase. which is itself an anticancer drug target. (P stands for phosphate. The technical terms are uridine mono- (di-, or tri-) phosphate; deoxyuridine mono- (di-or tri-) phosphate; similarly for the 5-fluoro compounds.) 142 K. W. Kohn Drugs Against Cancer CHAPTER6 The DNA mismatch repair system paradoxically helps kill cancer cells. What happens to the 5FU that remains mis-incorporated in the DNA and what kind of trouble does it cause? 5FU in DNA was found to be recognized by DNA mismatch repair enzymes (enzymes that repair base-pairs that do not match); it was also found that, in order to give more time for this repair, these enzymes signal to the cell cycle control system that DNA synthesis should be delayed (Li et al., 2009). You would think that DNA repair machinery should help cells to survive DNA damage. Cancers having competent DNA repair therefore ought to be relatively insensitive to DNA-damaging drugs. This is often true. However, in the case of certain kinds of DNA damage the opposite was found to be the case. We saw this paradoxical situation for the case oftemozolomide, which damages DNA by adding a methyl group to the 06 position of guanine (Chapter 2). The DNA mismatch repair system seemed to cooperate with temozolomide in producing anticancer activity. This mismatch repair paradox occurred similarly for the case of 5FU: like O6- methyl-guanine, 5FU in DNA is recognized by the mismatch repair system, which fails in its attempts to repair the defect, leading to a persistent problem that eventually causes cell death (see Chapter 25). In both cases (5FU and temozolomide), the mismatch repair system goes into futile repair cycles and sends distress signals to tell the cell cycle control system to stop the cell from dividing. Eventually, it alerts the last-resort molecular decision makers that consign the cell to suicide by apoptosis (Li et al., 2009). That is how these complicated mechanisms were thought (probably correctly) to work. Thus, cancers that have normal DNA mismatch repair were paradoxically more sensitive to 5FU than cancers that lacked this repair machinery (Iwaizumi et al., 2011; Suzuki et al., 2016). I will try to explain the reason for this strange state of affairs, where 5FU produced mismatched base pairs and the machinery intended to fix this defect instead helped to kill the cells. More details will be in a forthcoming chapter about DNA mismatch repair. But first a quick reminder about the two parts of the colon that differ in whether the DNA mismatch repair system is intact or defective. Cancers in the proximal colon, which do not arise in polyps and do not respond to 5FU, usually had an inactivating mutation in one of the mismatch repair proteins. There are 4 proteins that make up this repair machinery, the most commonly mutated one being MLHl (Figure 6.12) (Fleming et al., 2012) (Chapter 25). 143 K.W.Kohn Drugs Against Cancer CHAPTER6 Figure 6.12. A DNA mismatch repair-deficient cancer in the proximal colon. Mismatch repair requires the function of 4 genes, among which are MLHl, which is frequently mutated in these cancers, and MSH6, which is rarely mutated. The cancer in this figure had an inactivating mutation of the MLHl gene; therefore, the MLHl protein was absent in the cancer (ri9ht) . The MSH6 gene however was normal and its protein product was present in the tumor (left) (Fleming et al., 2012) (Permission needed.) (Fleming M, Ravula S, Tatishchev SF, Wang HL. Colorectal carcinoma: Pathologic aspects Journal ofGastrointestinal Oncolo9y. 2012;3(3):153-173). Why SFU in DNA looks like a mismatch to the repair system. The mismatch occurs, because SFU can become ionized and then pairs better with guanine than its normal partner, adenine (Figure 6.13). The Fluorine atom's strong affinity for electrons attracts negative charge out of the ring, thereby facilitating ionization by loss of a hydrogen ion. As a result, the SFU:adenine pair would be unstable and SFU would preferentially pair with guanine, as further explained in the legend to Figure 6.13. (The ionization of SFU is an equilibrium that depends on pH: the fraction of the time that the SFU is ionized would be greater when the pH is higher.) 144 K.W.Kohn Drugs Against Cancer CHAPTER6 H N N- H-- --- 0 F fW ,N--{ ~:N-----H- ~ 'H) R N=./ \._ N It . 0 R Adenine SFU Guanine Ionized SFU Figure 6.13. When 5-fluorouracil (SFU) has been incorporated into DNA, it can sometimes pair with guanine instead of adenine, which the mismatch repair machinery interprets as a mismatch. The pairing of SFU with guanine occurs when the SFU loses a hydrogen ion and becomes negatively charged, as shown in the lower part of the figure (Iwaizumi et al., 2011). SFU becomes ionized, because the F pulls some of the electron charge out of the ring, allowing loss of a hydrogen from a ring nitrogen, which then can serve as an H-bond receptor, instead of H-bond donor. Summary The simplest anti-cancer drug, 5-fluorouracil (SFU) and one of the earliest to be developed, became the most important drug for the treatment of colon cancer. It was developed based on insightful intuition, which merely entailed addition of a fluorine atom to uracil, a nucleic acid building block. Despite this simple modification, 5-fluorouracil (SFU) disturbs several essential steps in DNA and RNA synthesis that cancer cells need in order to grow and multiply. In combination with surgery and other drugs, SFU was able to cure a large fraction of colon cancer patients who had extensive local disease, but no distant metastases. It is remarkable how a drug simply made up of a normal uracil with a fluorine atom attached has a complicated mix of toxic actions and helps in the therapy of certain cases of colon cancer, as well as other cancers. 145 K.W. Kohn Drugs Against Cancer CHAPTER6 References Akpinar, B., Bracht, E.V., Reijnders, D., Safarikova, B., Jelinkova, I., Grandien, A., Vaculova, A.H., Zhivotovsky, B., and Olsson, M. (2015). 5-Fluorouracil-induced RNA stress engages a TRAIL-DISC-dependent apoptosis axis facilitated by p53. Oncotarget 6, 43679-43697. Andre, T., de Gramont, A., Vernerey, D., Chibaudel, B., Bonnetain, F., Tijeras- Raballand, A., Scriva, A., Hickish, T., Tabernero, J., Van Laethem, J.L., et al. (2015). Adjuvant Fluorouracil, Leucovorin, and Oxaliplatin in Stage II to III Colon Cancer: Updated 10-Year Survival and Outcomes According to BRAF Mutation and Mismatch Repair Status of the MOSAIC Study. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 33, 4176-4187. Ansfield, F., Klotz, J., Nealon, T., Ramirez, G., Minton, J., Hill, G., Wilson, W ., Davis, H., Jr., and Cornell, G. (1977). A phase III study comparing the clinical utility o f four regimens of 5-fluorouracil: a preliminary report. Cancer 39, 34-40. Ansfield, F.J. (1964). A Less Toxic Fluorouracil Dosage Schedule. Jama 190, 686-688. Ansfield, F.J., Schroeder, J.M., and Curreri, AR (1962). Five years clinical experience with 5-fluorouracil. Jama 181, 295-299. Bufill, J.A. (1990). Colorectal cancer: evidence for distinct genetic categories based on proximal or distal tumor location. Annals of internal medicine 113, 779-788. Carethers, J.M., Smith, E.J., Behling, C.A., Nguyen, L., Tajima, A., Doctolero, R.T., Cabrera, B.L., Goel, A., Arnold, C.A., Miyai, K., et al. (2004). Use of 5-fluorouracil and survival in patients with microsatellite-unstable colorectal cancer. Gastroenterology 126, 394-401. Curreri, A.R., Ansfield, F.J., Mc, I.F., Waisman, H.A., and Heidelberger, C. (1958). Clinical studies with 5-fluorouracil. Cancer research 18, 478-484. DeVita Jr., V.T. (2015). The Death of Cancer (New York: Farrar, Straus and Giroux). Fleming, M., Ravula, S., Tatishchev, S.F., and Wang, H.L. (2012). Colorectal carcinoma: Pathologic aspects. Journal of gastrointestinal oncology 3, 153-173. Geng, L., Huehls, A.M., Wagner, J.M., Huntoon, C.J., and Karnitz, L.M. (201 1). Checkpoint signaling, base excision repair, and PARP promote survival of colon cancer cells treated with 5-fluorodeoxyuridine but not 5-fluorouracil. PloS one 6, e28862. Heidelberger, C., Chaudhuri, N.K., Danneberg, P., Mooren, D., Griesbach, L., Duschinsky, R., Schnitzer, R.J., Pleven, E., and Scheiner, J. (1957). Fluorinated pyrimidines, a new class of tumour-inhibitory compounds. Nature 179, 663-666. Huehls, A.M., Huntoon, C.J., Joshi, P.M., Baehr, C.A., Wagner, J.M., Wang, X., Lee, M.Y., and Karnitz, L.M. (2016). Genomically Incorporated 5-Fluorouracil that Escapes UNG-Initiated Base Excision Repair Blocks DNA Replication and Activates Homologous Recombination. Molecular pharmacology 89, 53-62. Iwaizumi, M., Tseng-Rogenski, S., and Carethers, J.M. (2011). DNA mismatch repair proficiency executing 5-fluorouracil cytotoxicity in colorectal cancer cells. Cancer biology & therapy 12, 756-764. Johnston, P.G., and Kaye, S. (2001). Capecitabine: a novel agent for the treatment o f solid tumors. Anti-cancer drugs 12, 639-646. 146 K.W. Kohn Drugs Against Cancer CHAPTER6 Kawakami, H., Zaanan, A., and Sinicrope, F.A. (2015). Microsatellite instability testing and its role in the management o f colorectal cancer. Current treatment options in oncology 16, 30. Li, L.S., Morales, J.C., Veigl, M., Sedwick, D., Greer, S., Meyers, M., Wagner, M., Fishel, R., and Boothman, D.A. (2009). DNA mismatch repair (MMR) -dependent 5- fluorouracil cytotoxicity and the potential for new therapeutic targets. British journal of pharmacology 158, 679-692. Longley, D.B., Harkin, D.P., and Johnston, P.G. (2003). 5-fluorouracil: mechanisms of action and clinical strategies. Nature reviews Cancer 3, 330-338. Miwa, M., Ura, M., Nishida, M., Sawada, N., Ishikawa, T., Mori, K., Shimma, N., Umeda, I., and lshitsuka, H. (1998). Design of a novel oral fluoropyrimidine carbamate, capecitabine, which generates 5-fluorouracil selectively in tumours by enzymes concentrated in human liver and cancer tissue. Eur J Cancer 34, 1274-1281. Pettersen, H.S., Visnes, T., Vagbo, C.B., Svaasand, E.K., Doseth, B., Slupphaug. G., Kavli, B., and Krokan, H.E. (201 1). UNG-initiated base excision repair is the major repair route for 5-fluorouracil in DNA, but 5-fluorouracil cytotoxicity depends mainly on RNA incorporation. Nucleic acids research 39, 8430-8444. Poon, M.A., O'Connell, M.J., Moertel, C.G., Wieand, H.S., Cullinan, SA, Everson, L.K., Krook, J.E., Mailliard, J.A., Laurie, J.A., Tschetter, L.K., et al. (1989). Biochemical modulation of fluorouracil: evidence o f significant imp rovement o f survival and quality of life in patients with advanced colorectal ca rcinoma. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 7, 1407- 1418. Santi, D.V. (1980). Perspective on the design and biochemical pharmacology of inhibitors ofthymidylate synthetase. Journal of medicinal chemistry 23, 103-111. Santi, D.V., and McHenry, C.S. (1972). 5-Fluoro-2'-deoxyuridylate: covalent complex with thymidylate synthetase. Proceedings o f the National Aca demy of Sciences of the United States of America 69, 1855-1857. Suzuki, S., lwaizumi, M., Tseng-Rogenski, S., Hamaya, Y., Miyajima, H., Kanaoka, S., Sugimoto, K., and Carethers, J.M. (2016). Production o f truncated MBD4 protein by frameshift mutation in DNA mismatch repair-deficient cells enhances 5- fluorouracil sensitivity that is independent ofhMLHl status. Cancer biology & therapy, 1-9. Tie, J., Wang, Y., Tomasetti, C., Li, L, Springer, S., Kinde, I., Silliman, N., Tacey, M., Wong. H.L., Christie, M., et al. (2016). Circulating tumor DNA analysis detects minimal residual disease and predicts recurrence in patients with stage II colon cancer. Science translational medicine 8, 346ra392. Vogelstein, B., Papadopoulos, N., Velculescu, V.E., Zhou, S., Diaz, L.A., Jr., and Kinzler, K.W. (2013). Cancer genome landsca pes. Science 339, 1546-1558. Wood, L.D., Parsons, D.W., Jones, S., Lin, J., Sjoblom, T., Leary, R.J., Shen, D., Boca, S.M., Barber, T., Ptak, J., et al. (2007). The genomic landscapes of human breast and colorectal cancers. Science 318, 1108-1113. 147 K. W. Kohn Drugs Against cancer CHAPTER 7 Chapt,er-7. T~ 6NP$1JJ,Y ZZOllOOo.3 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@nih.gov CHAPTER 7 The 6-mercaptopurine (6MP) story. When I arrived at NIH in 1957 and was assigned to NCl's childhood leukemia ward -- for historical record, it was on the 2nd floor (2East) on the South side of the Clinical Center -- I helped to care for very sick children with acute leukemia. Using methotrexate or 6- mercaptopurine (6MP), we sometimes could reverse blast crisis, a late event in the disease that otherwise would be rapidly fatal. Although neither drug extended survival very much, both had a role in developing drug combinations that eventually led to lasting cures of the disease. The methotrexate story was related in Chapter 5. This chapter focusses on 6MP and its relative 6-thioguanine, how they were discovered, their mechanisms of action, and clinical application. The 6MP story began in 1949, perhaps a few years earlier than the 5-fluororuracil story related in Chapter 6. It began with the work of Gertrude Elion and George Hitchings (Figure 7.1), who shared a Nobel Prize in 1988. In 1951, they published their investigation of 100 compounds related to purines (adenine and guanine) (Figure 7.2). They studied the effects of the compounds in inhibiting or stimulating the growth Lactobacillus casei bacteria. An example of one of their early experiments is shown in Figure 7.3. Their major and most enduring finding was that their most potent inhibitory compound was a purine with a sulfur atom added at the 6-position (6-mercaptopurine, 6MP) (Figure 7.2); the inhibition was reversed by purines such as adenine (Elion and Hitchings, 1950; Elion et al., 1951). (Similar effects were produced by the earlier known, but less potent, 2,6-diaminopurine (Figure 7.3).) They went on to test the effect of 6MP on cancers in mice and were impressed that the drug inhibited the S180 sarcoma tumor and even cured a few of the animals (Clarke et al., 1953). Some of their other purine compounds had inhibited the growth of more sensitive tumors but had little effect on S180. 148 K. W. Kohn Drugs Against cancer CHAPTER 7 In those early days (prior FDA approval had not yet been established), the progress from a new finding in the laboratory to a clinical test was remarkably rapid. Thus within two years after 6MP's inhibition of bacteria was noted in 1951 and the same year as the first report of inhibition of tumors in animals in 1953, a test in humans was also published-- by investigators led by Joseph H. Burchenal in coordination with Hitchings and Elion (Burchenal et al., 1953). They began clinical tests after finding that the drug inhibited the growth of many different transplanted tumors in animals and after carrying out toxicity tests in mice, rats, cats, and dogs to determine a safe starting dose in patients. It was a concerted effort by at least 10 well-established clinicians and researchers from the Sloan- Kettering Institute and Cornell University Medical College in New York. Those early clinical tests were considered justified by the desperate need of the patients, including children who were dying of acute leukemia. The clinical trials began in April 1952 and included 107 patients with various advanced cancers who had been recruited to the study by March 1953 (Burchenal et al., 1953). The study included 45 children and 18 adults with acute leukemia. The staff carefully monitored the patients' blood counts, bone marrow cellularity, as well as kidney and liver functions. The children tolerated the treatment relatively well with only rare serious toxicity. Of the 45 children with acute leukemia, 15 had good remissions lasting a few weeks to a few months. As a successful clinical trial carried out so soon after the discovery of a new drug. it was a remarkable achievement and launched 6MP as a promising new anticancer drug candidate. Further studies soon confirmed that the drug increased the survival time of some of the acute leukemia children (Burchenal et al., 1954). Those early results were the foundation for the studies in which I participated in 195 7 on the childhood leukemia ward at the National Cancer Institute. The 6-thioguanine story. In 1958, Donald Clarke and Chester Stock together with Elion and Hitchings went on to test 102 close structural relatives of 6MP for their ability to inhibit the growth of the S180 tumor in mice. All of the compounds were less effective than 6MP, except for 6-thioguanine which was about as effective as 6MP (Clarke et al., 1958). The only difference in their chemical structures was that that 6MP lacked the amino (NH2) of the guanine part of the molecule (Figure 7.3). Moreover, the two drugs were cross-resistant: a tumor that was resistant to one was also resistant to the other. Jack D. Davidson, then at Columbia University, had shown that both drugs inhibited the synthesis of DNA in tumors (Davidson and Freeman, 1955). From those lines of evidence, Clarke and coworkers suspected (correctly) that enzymes in the tissues converted one of the drugs to the other. 149 K. W. Kohn Drugs Against cancer CHAPTER 7 Figure 7.1. Gertrude B. Elion (1918-1999) and George H. Hitchings (1905-1998) before and after their Nobel Prize in Physiology or Medicine in 1988. Adenine Guan ine H:CS N N ~N I NJJ H 2, 6-diam inopurine 6-Mercaptopurine 6-Thioguanine (6MP) Figure 7.2. Among the first variants of adenine or guanine found by Gertrude Elion and George Hitchings to inhibit nucleic acid function was 2,6-diaminopurine. The most potent were 6-mercaptopurine (6MP) and 6-thioguanine. In cells, 6MP is converted to 6- thioguanine, which was responsible for most of the therapeutic actions of 6MP. 150 K. W. Kohn Drugs Against cancer CHAPTER 7 1~ - - - - -- - - - -- ---, 2,6· DIAMI N0PURINE T~YM INE I Y PER ML. 6 067 0.2 .67 2 6.7 20 67 ADEN IN E Y PER M L . Figure 7.3. An early experiment by Elion and Hitching in 1950 showing the inhibition by 2,6-diaminopurine of the growth of Lactobacil/us casei and its reversal by adenine (Elion and Hitchings, 1950). The bacteria were unable to grow in the absence of adenine. When adenine concentration was low, 2,6-diaminopurine inhibited growth (lower 3 curves), while higher adenine concentrations restored the ability to grow. The vertical axis was a measure of growth. The horizontal axis showed the concentration of adenine. The inference was that 2,6-diaminopurine competed with adenine as a requirement for growth. The crucial role of6-mercaptopurine (6MP) for continuation (maintenance) therapy needed for cure of acute ly mphocytic leukemia (ALL). The cure of acute lymphatic leukemia in children had two parts. First, the malignant cells had to be decimated using drugs, such as vincristine and dexamethasone, to induce an initial remission of the clinical disease. To get a lasting cure, however, this had to be followed by continuation therapy for 2 or 3 years or longer with 6MP plus methotrexate; no better drugs for this purpose had been found (Schmiegelow et al., 2014). That was a bit of a surprise, because 6MP was not very good for the first (induction) phase of the therapy, and the addition of the drugs that were best for the induction phase reduced the effectiveness or increase the toxicity of the continuation therapy. In an editorial reflecting on this surprising circumstance, Barton A. Kamen in 2009 invoked a definition of"serendipity" by Horace Walpole in 1754 as "sagacity of being able to link together apparently innocuous facts to come to a valuable conclusion" (Kamen, 2009). Kamen's commentary addressed a report of a clinical trial by Kjeld Schmiegelow and coworkers that appeared in the same issue of the journal. His comments converged on 151 K. W. Kohn Drugs Against cancer CHAPTER 7 three points: (a) the success of clinical trials for children who had acute lymphocytic leukemia, (b) some characteristics of the leukemic stem cells, and (c) thattumor stem cells may be primarily on the de novo pathway for purine (guanine and adenine) synthesis, the de novo and salvage pathways being critically regulated and tied to cell differentiation and proliferation. Therefore, tumor stem cells, which must be suppressed during continuation therapy, would be particularly sensitive to inhibition of the de novo pathway by 6MP and methotrexate (Figure 7.4). Inhibition of the de novo pathway by methotrexate would enhance the utilization of 6MP via the purine salvage pathway and increase its cytotoxic action. 6MP and 6-thioguanine enter the purine salvage pathway by way of an enzyme that adds to them a phosphoribosyl group (Karran and Attard, 2008). Kamen argued that it was by Walpole's definition of serendipity that consideration of those three points led to the counter-intuitive use of 6MP plus methotrexate for the continuation therapy that was needed to cure acute lymphocytic leukemia. In 2011, William E. Evans and coworkers at St Jude Children's Research Hospital in Memphis, Tennessee reported that some children with acute lymphocytic leukemia (ALL) lacked MSH2, one of the proteins required for DNA mismatch repair (the subject of Chapter 25) (Diouf et al., 2011). The lack of MSH2 was not due to inadequate expression of its gene but rather was caused by overactivity of an enzyme that destroyed the MSH2 protein in the leukemia cells. The ALL children whose leukemic cells lacked MSH2 and therefore were deficient in DNA mismatch repair had a reduced cure rate: their 10-year survival was 78.7% compared to 97.5% (P = 0.009) for those who had higher MSH2 levels. A high mismatch repair activity therefore helped long-term survival. That may seem counter-intuitive. The ability of the leukemia cells to repair damage to their DNA one might think should help the cells to survive, and that is usually the case. For mismatch repair, however, the opposite was the case (as we have seen for temozolomide in Chapter 2 and fluorouracil in Chapter 5). When 6-thioguanine was incorporated in DNA, it paired with C like a normal G would do. But there was an enzyme that added a methyl group to the sulfur atom of 6- thioguanine in DNA (step 18 in Figure 7.4). The resulting 6-methylthioguanine was recognized as a mismatch by the repair system. But the repair was futile : it usually left a new mismakh after each repair attempt, as will be explained in connection with step 19 in Figure 7.4. How 6-mercaptopurine (6MP) and 6-thioguanine kill cancer cells. Figure 7.4 shows the chemical steps and biological actions of 6-mercaptopurine (6MP) and 6-thioguanine, as well as methotrexate. In what follows, I explain the essential steps and processes as currently understood, using the number assigned to each step in the diagram. These steps and actions determine whether the drug-treated cell survives or dies. To begin with, there are inactivation steps. The more drug that is inactivated in a given patient, obviously the less effect the drug could have. After 6MP is absorbed from the intestines, it is subject to inactivation in the liver, which converts 6MP to an inactive product, 6-thiouric acid [1 ]. Another enzymatic inactivation process adds a methyl group to the sulfur atom of 6MP or 6-thioguanine [Z, Za] . Variants of that enzyme (thiopurine- 152 K. W. Kohn Drugs Against cancer CHAPTER 7 methyltransferase, TPMT, explained later in this Chapter) affect its activity and therefore the sensitivity of children with acute lymphatic leukemias to toxicity from 6MP treatments (Chan et al., 2019). Both 6MP and 6-thioguanine can become incorporated as a thioguanine component in place of guanine in nucleic acids. The first step along the pathway for incorporation into DNA is the addition of a deoxyribose-phosphate group to 6MP (3, 7) or to 6-thioguanine (6). 6MP and 6-thioguanine converge to the same DNA synthesis precursor, 6-thioguanine- deoxyribose-phosphate, to which two more phosphates are added to form the triphosphate [BJ. The latter becomes incorporated as a thioguanine in place of guanine in DNA (10). Once thioguanine is in the DNA structure, it inhibits DNA synthesis (14] and causes additional trouble, as we shall see. Thioguanine incorporation into DNA however was recently found to be limited by an enzyme called NUDT15 that removes two phosphates from thioguanine-deoxyribose- triphosphate and sends it back to the monophosphate level (9) (Chan et al., 2019; Singh et al., 2017). The enzyme normally protects DNA from incorporation of 8-oxoguanine, a troublesome product of reactive oxygen species - it selectively removes phosphates from 8-oxoguanosine triphosphates. The NUDT15 gene was found to have single-nucleotide polymorphisms (SNIPs). The particular NUDTlS SNIP in a patient's genome affected how sensitive the patient was to bone marrow suppression by 6MP. If a patient's NUDTlS SNIP was such that activity of the enzyme was unusually low or absent, the 6MP dose would be reduced for safety - because low activity of reaction (9) would allow more 6-thioguanine (as well as more 8-oxoguanine) to enter the DNA. After thioguanine was incorporated into DNA (10), an S-methyltransferase enzyme added methyl groups to the sulfur atom of 6MP or 6-thioguanine (2, 2a] and methyl groups to the DNA-incorporated thioguanines (18). As already mentioned above, the resulting S-methyl- thioguanines are recognized by the DNA mismatch repair system (the topic of Chapter 25). The repair system removed a section of either the strand containing the S-methyl- thioguanine or a section containing the base mispaired with it In either case however the repair process would be apt to produce a new mismatch. This cycling of futile repair attempts eventually causes the cell to give up and die (19, 20). While all of that is going on, normal guanine is also being incorporated into DNA (11, 12). However, normal guanine and 6-thioguanine compete for incorporation into DNA [15). Thus, when guanosine-triphosphate is high, the DNA-incorporation of thioguanine would be relatively low. The amount of guanine nucleotides would depend on the rate that guanine is made available by de novo synthesis (16) and by recovery from degraded nucleic acids ("salvage pathway") [17). The de novo pathway is inhibited by methyl- mercaptopurine-ribose-phosphate [5), which is produces by the S-methyltransferase that adds a methyl group to the sulfur (4). De novo purine synthesis is also inhibited by methotrexate (22). Both inhibitions reduce the production guanine nucleotides, which reduces competition with thioguanine and consequently allows increased incorporation of thioguanine into DNA [15). Methotrexate enhances the action of 6MP both by inhibiting de novo purine synthesis and by inhibiting the destruction of 6MP by blocking its oxidation to 153 K. W. Kohn Drugs Against cancer CHAPTER 7 6-thiouric acid [1] (Larsen et al., 2021; Schmiegelow et al., 2014). Although treatment of children who have acute lymphocytic leukemia, including long-term maintenance (or "continuation") therapy with 6MP plus methotrexate, cured most of them, many eventually relapsed (Figure 7.5). Relapse was thought perhaps related to low activity of a necessary enzyme in the metabolic scheme shown in Figure 7.4. This possibility was investigated by testing whether relapsing children sometimes were heterozygous in a critical gene, in other words having one normal gene and an inactivating mutation in the other one. The net effect would be reduced activity of that gene. The investigation pointed particularly to the gene for thiopurine-methyltransferase (TPMT), which is responsible for reactions 2, 2a, 4, and 18 in Figure 7.4 (Schmiegelow et al., 2014). It would seem that deficiency in reactions 2 and 2a would increase 6MP and 6- thioguanine actions, whereas deficiency in reactions 4 and 18 would have the opposite effect (assuming that all those inhibitions involve exactly the same gene). Heterozygous mutation that inactivated only one copy of the gene was found in about 10% of patients, although 1 in 300 had inactivating mutations in both copies of the gene that put them at risk of life-threatening toxicity if treated with customary 6MP doses. Patients with a heterozygous TPMT mutation had higher 6-thioguanine levels in their cells, more toxicity but higher cure rates -- but they were thought to have higher risk of developing new cancers in coming years. In 2018, it was recommended that the genotypes of both TPMT and NUDTlS be evaluated in deciding on the dose of 6MP to be used. The TPMT gene, which is located on chromosome 6, had about 40 variant forms resulting from single-nucleotide polymorphisms (SNPs), of which four accounted for 80-95% of the TPMT variant cases found in patients; each of these four variants had an amino acid change from the normal gene and encoded a protein with relatively low enzymatic activity (Franca et al., 2021). The normal function ofTMPT remained unknown; the absence of the gene seemed not to cause disease. However, the TMPT protein was reported to methylate the selenium atom that replaces sulfur in selenocysteine, a rare encoded form of the normal amino acid cysteine (selenium is below sulfur in the periodic table and is highly nucleophilic). 154 K. W. Kohn Drugs Against cancer CHAPTER 7 ,._ .,_ Thlog,,anlne ~ CH,S (Xj - 12 M:lrm;il lncotl)Oralon of 1-u,,nlnc Into ONA. I Cel death ~ oe nova purine H - synthesis. sl --.1._ _,._ ___. r ,, - <.... - - - - l"--"'I ----1 - p urlncs from .tfi!'l oclb. Zl 22 Methotrexate 1 - - ~ - - - ~ -- - - - - -- - - - - -- - - - - ~ Figure 7.4. The chemical changes and biological effect of 6-mercaptopurine (SMP), 6- thioguanine, and methotrexate in the cancer cell. See the preceding text for explanation of the steps and processes shown in the diagram. 0,4 91 a. .. ..!l! ~ 0,2 High whit e blood cell count -·--·--·-·------ - - ... 'o •• •• .>I a: _ • .•.~·_.,..------------ Low white blood cell count 0,0 .00 5.00 10.00 15,00 Years from diagnosis Figure 7.5. Relapse rates of children with acute lymphocytic leukemia after long-term maintenance therapy with 6-mercaptopurine (6MP) and methotrexate. If the effect of the therapy was strong enough to suppress the white blood cell count to a relatively low level (lower curve), then relapse rate was relative low (10.9%), compared with patients whose treatment had a lower effectiveness as indicated by a higher white blood cell count, who had a higher relapse rate (upper curve, 23.5%; P<0.001) (Schmiegelow et al., 2014). 155 K. W. Kohn Drugs Against cancer CHAPTER 7 Final word We saw (again, as in the case o f 5-fluorouracil) how drugs with simple chemical structures, discovered on the basis of simple principles, entered metabolic pathways and caused multiple complex perturbations o f those pathways, leading to sometimes surprising therapeutic and toxic actions. Also remarkable was how quickly those early discoveries, crucial to the eventual cures of acute leukemias and lymphomas, moved from discovery to treatment of p atients. References Burche nal, J.H., Ellison, R.R., Murphy, M. L., Karnofsky, D.A., Sykes, M.P., Tan, T.C., Mermann, A.C., Yuceoglu, M., Myers, W.P., Krakoff, I., et al. {1954). Clinical studies on 6- mercaptopurine. Anna ls of the New York Academy of Sciences 60, 359-368. Burche nal, J.H., Murphy, M. L., Ellison, R.R., Sykes, M.P., Tan, T.C., Leone, LA., Ka rnofsky, 0.A., Craver, L.F., Dargeon, H.W., and Rhoads, C.P. {1953). Clinical evaluation of a new antimetabolite, 6-mercaptopurine, in the treatment of leukemia and allied diseases. Blood 8, 965-999. Chan, H.T., Chin, Y.M., and Low, S.K. {2019). The Roles of Common Variation and Somatic Mutation in Cancer Pharmacogenomics. Oncol Ther 7, 1-32. Clarke, D.A., Elion, G.B., Hitchings, G.H., and Stock, C.C. {1958). Structure-activity relationships among purines related to 6-mercaptopurine. Cancer resea rch 18, 445-456. Clarke, D.A., Philips, F.S., Sternberg, S.S., Stock, C.C., Elion, G.B., and Hitchings, G.H. {1953). 6- Mercaptopurine: effects in mouse sa rcoma 180 and in normal animals. Cancer research 13, 593-604 . Davidson, J.O., and Freeman, B.B. {1955). The effects of antitumor drugs upon P32 incorporation into nucleic acids of mouse tumors. Cancer research 15, 31-37. Diouf, B., Che ng, Q ., Krynetska ia, N.F., Yang, W., Cheok, M., Pei, 0., Fan, Y., Cheng, C., Krynetskiy, E.Y., Geng, H., et al. {2011). Somatic deletions of genes regulating MSH2 protein stability cause DNA mismatch repair deficiency a nd drug resista nce in human leukemia cells. Nature medicine 17, 1298-1303. Elion, G.B., and Hitchings, G.H. {1950). Antagonists of nucleic acid derivat ives. IV. Reversal studies with 2-aminopurine a nd 2,6-diaminopurine. The Jou rnal of biological chemist ry 187, 511-522. Elion, G.B., Hitchings, G.H., and Va nderwerff, H. {1951). Antagonists of nucleic acid derivatives. VI. Purines . The Journal of biological chemistry 192, 505-518. Fra nca, R., Braidotti, S., Stocco, G., and Decorti, G. {2021). Understanding t hiopurine methyltransferase polymorphisms for t he targete d treat ment of hematologic malignancies. Expert opinion on drug metabolism & toxicology 17, 1187-1198. Kamen, B.A. {2009). Sere ndipity-methotrexate and 6-mercaptopurine for continuation the rapy for patients wit h acute lymphoblastic leukemia: the leukemic stem cell and beyond? J Pediatr Hematol Oncol 31, 383-384. 156 K. W. Kohn Drugs Against cancer CHAPTER 7 Ka rran, P., and Attard, N. (2008). Thiopurines in current medical practice: molecular mechanisms and contributions t o t herapy-related cancer. Nature reviews Ca ncer 8, 24-36. Larsen, R.H., Utke Rank, C., Grell, K., Norgaard Moller, L , M althe Overgaard, U., Kampmann, P., Nersting, J., Oegn, M., Nygaard Nielsen, S., Holst , H., et al. (2021). Increments in ONA- t hioguanine level during thiopurine-enhanced maint enance t herapy of acut e lymphoblast ic leukemia. Haematologica 106, 2824-2833. Schmiegelow, K., Nielsen, S.N., Frandsen, T.L, and Nerst ing, J. (2014). M ercapt opurine/ M ethot rexate maint enance t herapy of childhood acute lymphoblastic leukemia: clinical fact s and fict ion. J Pediatr Hematol Oncol 36, 503-517. Singh, M ., Bhatia, P., Khera, S., and Trehan, A. (2017). Emerging role of NUOT15 polymorphisms in 6-mercapt opurine metabolism and dose related t oxicit y in acut e lymphoblast ic leukaemia. Leuk Res 62, 17-22. 157 K.W. Kohn Drugs Against Cancer CHAPTERS Chapur-8 TkdoxDrubkin 'flDry ZZ07Zfkn3 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@nih.gov CHAPTER S The Doxorubicin Story: a star with a nearly fatal flaw. Introduction Doxorubicin (also known as Adriamycin) is a DNA intercalator (Chapter 4) and a topoisomerase II blocker (Chapter 10). It became one of the most useful anticancer drugs; it was found effective, although not curative, for many cancers. Its usefulness however was bedeviled by toxic effects on the heart, brain, liver, and kidney. Of those toxicities, the most serious was damage to the heart; patients often died of congestive heart failure if the cumulative amount of the drug administered was above a certain limit, (Von Hoff et al., 1979) (Figure 8.1). The drug sometimes caused the cancer to disappear, but the remissions lasted only a few months before the tumor reappeared and was then resistant to the drug (Benjamin et al., 1974). Some breast cancer patients who were successfully treated by surgery followed by a period of doxorubicin as "adjuvant treatment" still had some degree of heart damage even 10 years later (Murtagh et al., 2016). Thus, the heart damage was irreversible, and it could be so severe that the only remedy was transplantation of a new heart. The potentially lethal damage to the heart prevented administration of higher doses that might have cured the cancer. Enormous effort was made therefore to find out exactly how the drug damaged the heart Although the mechanism of the heart damage was clarified, no preventative was found other than to limit the amount of drug administered. Moreover, it was not fully determined exactly how doxorubicin suppressed cancers, although the general opinion was that its action on topoisomerase II was the main therapeutic mechanism. 158 K.W. Kohn Drugs Against Cancer CHAPTERS Figure 8.1. Heart tissue damaged by doxorubicin after treatment for breast cancer in a 63-year old woman (40X). The normal fibrils in a heart muscle cell (lower right corner) are disrupted in many of the muscle cells. A disintegrating heart muscle cell with many blebs is seen in the center. It is mostly white, because it lacks essential proteins that would stain dark This was from a small heart biopsy from an ailing patient (Singal and Iliskovic, 1998). An anticancer drug is discovered by tracking down a red substance. The doxorubicin story began with a search for antibiotics made by microorganisms in the soil. Researchers found organisms that were associated with a red color. As chemists often like to do, the researchers wanted to isolate the substance that had the color. In 1963, the researchers, led by Aurelio Di Marco at Farmitalia Research Institute in Milan, Italy, isolated a red antibiotic from Streptomyces peucetius, which they called "daunomycin"; later they renamed it "daunorubicin" in view of its red color. Di Marco's group was struck by finding that their new antibiotic not only killed bacteria but inhibited cancer growth in mice (Di Marco et al., 1964; Di Marco et al., 1965). Lengthy investigations eventually disclosed that daunorubicin inhibited DNA synthesis and bound to DNA by intercalation between the DNA base pairs (Pigram et al., 1972) (see Chapter4). As an aside, coming to mind in the context of intensely colored anticancer drugs is a rather amusing. but instructive, talk at a Cancer Chemotherapy Gordon Conference many years ago given by Daniel von Hoff, in which he showed a series of slides that 159 K.W.Kohn Drugs Against Cancer CHAPTERS suggested that color was as good way as any available at the time to predict whether a compound might have anticancer activity! Indeed, many anticancer drugs were intensely colored, because their chemical structures often had multiple rings with alternating double bonds (aromatic structure). This is particularly true for DNA intercalating agents, because to fit between the base pairs, the molecule (or at least a part of it) had to be planar (i.e., flat), which requires a multi-ring structure with alternating double bonds (aromatic rings). Moreover, the charge distribution over the ring system, which depends on the nitrogen atoms that most intercalators have in their ring system, helped the molecule to stack firmly against particular base pairs. Then, in 1967, Frederico Arcamone and his coworkers at Farmitalia isolated another red antibiotic from a mutant strain of the organism that produced daunorubicin (Arcamone et al., 1969). The new antibiotic was very similar to daunorubicin in chemical structure, as well as in chemical and biological properties, so they named it "doxorubicin"; they found that the only structure difference from daunorubicin was that doxorubicin had a hydroxyl group added at position 14 (Figure 8.2). It is notable that useful variations of a drug can sometimes be obtained from mutant variants of a particular organism. Doxorubicin was superior to daunorubicin in its pattern of antitumor activity relative to toxicity (Bonadonna et al., 1970). Great effort was then made to find the cause of the toxicity and how to combat it. Although the former objective was achieved, the latter was recalcitrant. 0 OH 0 OH ,., ,... COCH~ □ OCHo 0 OH H OH H H H " " DAUNORUBICIN DOXORUBICIN Figure 8.2. Chemical structures of daunorubicin (left) and doxorubicin (right). The only difference is that doxorubicin has an OH group added at position 14 (red box). The four rings with their double-bonds have the size, shape, and electronic structure to bind stacked against a DNA base-pair, as occurs in intercalation and in DNA- topoisomerase II trapped complexes (Figures 8.3). The amino group in both drugs (circled red) confers a positive charge to the six-membered ring that lies near the negatively charged DNA backbone (Figure 8.4). 160 K.W. Kohn Drugs Against Cancer CHAPTERS I ~ ~\. --·... ~ ·• .~.\:.""..:~, ·'",;~W,[ Figure 8.3. Molecular model showing how doxorubicin (DOX) intercalates between base-pairs in DNA (Pigram et al., 1972). Figure 8.4. A computer-generated model of how Doxorubicin (DOX) intercalates between base-pairs of a DNA double helix (Agudelo et al., 2014). Doxorubicin is show in green. Note that doxorubicin's flat ring system intercalates, while its side- chain with its positive charge is located outside of the stack of base-pairs and close to the negatively charged DNA backbones. 161 K.W.Kohn Drugs Against Cancer CHAPTERS How doxorubicin damages the heart Since heart damage was so prominent in doxorubicin's action, I will discuss first what was found out about it and then consider how doxorubicin exerts its anticancer action. After decades of inconclusive speculation, the chemical culprit that damages the heart was finally identified (Mukhopadhyay et al., 2009) (Pacher et al., 2007). In accord with one of the leading hypotheses, the culprits were found to be "free radicals" (to be explained in a moment) (Sinha et al., 1987) (Rajagopalan et al., 1988). As long suspected, the toxic free radicals are produced by mitochondria as they use oxygen to make ATP, the energy currency of the cell. Doxorubicin was found to increase the production of free radicals during the energy production process in mitochondria. It turned out, however, that most of the damage was due to a particular free radial product As electrons flow through the chain of proteins (cytochromes) in the membranes of mitochondria, they generate ATP, and also some free radicals as an unavoidable side effect. The process is rather complicated and is described in standard biochemistry textbooks. Here, the main thing to know is that some of those electrons flowing through this cytochrome chain occasionally go astray and produce the aforementioned free radicals; doxorubicin, as we shall see, facilitates free radical production. Also important is that heart muscle cells are particularly rich in mitochondria, which supply the high energy needs of the heart. Therefore, the heart's high energy needs entail high electron flow through the mitochondrial cytochrome chains with likely production of damage-inducing free radicals, especially in the presence doxorubicin. Here are a few details about how that happens: Doxorubicin's aromatic ring system easily picks up an electron as it flows through the cytochrome chain, and the doxorubicin molecule then easily transfers the electron to an oxygen molecule. The extra electron on the oxygen molecule makes it highly reactive, because the odd electron desperately wants to pair up with another electron. One way it does that is to grab an electron off of some important protein in the heart, which, in the end, damages the heart muscle. More explicit detail in a moment. Another factor making doxorubicin particularly dangerous to the heart was that it tended to bind to lipid membranes, including membranes where the cytochrome chain is located in the mitochondria: the drug then is in position to pick up an electron flowing through the cytochromes. As already said, heart cells are particularly vulnerable, because of their unusually large numbers of mitochondria - which they have in order to provide for the high energy requirement of the heart. Tissue cells, including those of the heart, do have enzymes (superoxide dismutase and catalase) that rapidly destroy free radicals. In the presence of doxorubicin, 162 K.W.Kohn Drugs Against Cancer CHAPTERS however, the increased free radical production can overwhelm the capacity of those protective enzymes. A little more about free radicals: A free radical is a molecule or atom with an odd number of electrons. Molecules are stable only if they have an even number of electrons (except for some molecules that have a heavy metal atom, such as iron). The chemical structure of doxorubicin, with its multiple aromatic rings (Figure 8.2), easily picks up an extra electron, forming a 'semiquinone' type of free radical (Keizer et al., 1990). The doxorubicin semiquinone gets some stability from its odd electron being distributed over the alternating single and double bonds in the ring system. The odd electron eventually transfers to oxygen (02), forming superoxide (02-), which engages in further reactions producing highly reactive oxidants that can damage many essential molecules in the cell. Intensive investigation revealed more detail about how doxorubicin damages the heart by way of free radicals (Mukhopadhyay et al., 2009) (Pacher et al., 2007). As already said, doxorubicin binds to the inner membrane of mitochondria in heart muscle cells, where it can pick up an electron from the electrons flowing through the energy-generating electron transport chain. The resulting doxorubicin semiquinone free radical then transfers its extra electron to a molecule of oxygen to produce superoxide (02-), which is an 02 molecule with an extra electron added (which gives the molecule a negative charge). At this point, another reactive biologically important molecule with an odd number of electrons comes into play: nitric oxide (NO). Nitric oxide consists of an atom of nitrogen bound to an atom of oxygen. The two together have an odd number of electrons; hence NO is a free radical; however, it is relatively stable, as free radicals go, having a half-life in tissues of a few seconds. That is a long time for a free radical in tissues: long enough for the free radical to move around among the cells and find a vulnerable target. Its high reactivity and relatively short life span, however, allow it to be useful as an important signaling molecule in cells and tissues. It was recently found that the entity that actually causes most of the heart damage is a chemical combination of nitric oxide (NO) and superoxide (02-); the combination forms the highly reactive peroxynitrite (O=N-O-O·) molecule (Mukhopadhyay et al., 2009) (Pacher et al., 2007). Peroxynitrite prod uction is favored when superoxide is generated in increased amounts through the ability of doxorubicin to shuttle unpaired electrons, and when NO levels are also high. The production of peroxynitrite from nitric oxide and superoxide is highly efficient: the two molecules combine whenever they come in contact. Peroxynitrite is a strong oxidant with a short half-life, but long enough to reach targets where it can cause trouble. Its short half-life is countered by a high and sustained production rate (Szabo et al., 2007) (Pacher et al., 2007). Peroxynitrite can react with many different constituents in the cell; however its major toxic action in the heart is thought to be at mitochondria, where it can enhance the production of superoxide, thus producing a positive 163 K.W.Kohn Drugs Against Cancer CHAPTERS feedback loop of problems (Szabo et al., 2007) (Pacher et al., 2007) (Mukhopadhyay et al., 2009). Another aspect of the path to heart cell destruction is that peroxynitrite stimulates mitochondria to produce molecules that cause damaged cells to die in a cell suicide process (apoptosis). Thus, mitochondria are involved both in the doxorubicin- facilitated production of superoxide and in the cell-killing effect (apoptosis) of peroxynitrite in the heart. Over extended periods of treatment with doxorubicin, increasing numbers of heart muscle cells die. Since adult heart muscle cells are not replaced, the damage is irreversible and progressive. That is why the extent of damage to the heart depends on the cumulative amount of doxorubicin a patient receives over time, and the damage persists for many years (Singal and Iliskovic, 1998). However, some investigators thought that a free radical may also contribute to doxorubicin's extraordinary anti-cancer action (Keizer et al., 1990). How do the molecular actions of doxorubicin produce anti-cancer activity? After being submerged in discussion of heart-damaging free radicals, we come at last to the question of how doxorubicin produces its anticancer action. We have noted that doxorubicin binds to DNA by intercalation (Figures 8.3 and 8.4) and blocks DNA and RNA syntheses. Several other DNA-intercalating drugs share these actions, but lack doxorubicin's remarkable anti-cancer activity. Hence DNA intercalation and inhibition of DNA synthesis by themselves were thought insufficient for the anticancer action. Our laboratory had found that DNA intercalation is often associated with a blocking action on topoisomerase II (Pommier et al., 1985) (Zwelling et al., 1981), which became top-of-the-list of likely causes of doxorubicin's anticancer activity. Drug actions on topoisomerase II is the subject of Chapter 10. Nevertheless, there was doubt as to whether the actions of doxorubicin on topoisomerase was the full answer, because there are other drugs that have similar actions, but lack the broad anti-cancer activity of doxorubicin (Burden and Osheroff, 1998). lt was proposed that doxorubicin's unusually strong anticancer action results from a combination of (1) topoisomerase II blocking and (2) free radical damage in mitochondria with consequent apoptosis (programmed cell death) of cancer cells (Mizutani et al., 2005). 164 K.W.Kohn Drugs Against Cancer CHAPTERS Topoisomerase II transiently cleaves both DNA strands in the double helix, so as to allow one double helix to pass through another, thereby disentangling the DNA in order to allow the chromosomes to separate during mitosis (discussed in Chapter 10). Doxorubicin was found to prevent that action by binding to the topoisomerase II-DNA complex while the DNA strands are cleaved, which would prevent the strands from rejoining. This cleaved DNA-drug complex apparently was lethal to the cell unless the cleavage complex reversed spontaneously or was resolved by a molecular repair process. The DNA-topoisomerase II complexes trapped by doxorubicin thus would kill the cell if the trapped complexes were not repaired before an encounter with a replication or transcription process occurred. Also contributing to the anticancer action of doxorubicin may be a relatively high stability of its complex with topoisomerase II that might be difficult to repair. Liposomal preparations of doxorubicin for better delivery to tumors. Attempts were made to improve the delivery of the poorly soluble doxorubicin into cancer cells by putting the drug molecules into microscopic lipid vesicles called liposomes {Uziely, 1995 #10353}, and such preparations were approved for clinical use. A further improvement was to coat the liposomes with polyethylene glycol (Papahadjopoulos et al., 1991); such forms ofliposomes were said to be "pegylated" (Figure 8.5). Pegylation prevented phagocytic cells from taking up and inactivating the drug. Doxorubicin in pegylated liposomes somehow reduced the toxicity to the heart (Markman, 2006) (Thigpen et al., 2005) (Gabizon, 2001). Delivering drug-liposome combinations into cancer cells was also enhanced by the high permeability of blood vessels in tumors, which allowed liposome-sized bodies to exit from the blood vessels and reach cancer cells. Blood vessels in cancer tissue tend to have increased permeability that may allow liposome-coated doxorubicin to enter. Once inside the cancer tissue, the drug tended to be retained for a relatively long period of time, because of poor drainage through lymphatic vessels in tumors (Torchilin, 2011). Thus, various forms ofliposomes were developed and used clinically to improve drug delivery into cancers. A further improvement was to link folic acid molecules to the poly(ethylene glycol) layer on the surface of the drug-containing liposomes. The idea was to take advantage of the relatively large number of folic acid receptors on the surface of many cancer cells. The folic acid on the liposome's surface would bind the receptors on the tumor cell surface and increase the amount drug-containing liposome that gets into the cell (Sriraman et al., 2016). 165 K.W. Kohn Drugs Against Cancer CHAPTER S Internal Aqueous Compartment Contains Doxorubicin Liposome Surface 85nm Coated with PEG Polymer Layer Lipid Bilayer Membrane Composed of HSPC:Chofesterol Figure 8.5. The structure of a doxorubicin-containing pegylated liposome. Doxorubicin was encapsulate in the center, surrounded by a double-layered lipid membrane with poly(ethylene glycol) (PEG) chains on the surface (Gabizon, 2001). Sy nopsis Doxorubicin is one of the most useful drugs in cancer chemotherapy, but is plagued by toxicity, particularly to the heart, which limits the cumulative amount of drug that can safely be administered. The anti-cancer activity of doxorubicin was thought to be due mainly to its ability to trap DNA-topoisomerase II complexes, perhaps combined with its tendency to generate free radicals in mitochondria, leading to programmed cell death "apoptosis." The free radical action was found to be the cause of the toxicity to the heart, but preventing this toxicity was not fully achieved. The anti-cancer usefulness of doxorubicin was enhanced by incorporating the drug in liposomes, which selectively delivered the drug to cancer cells. That action was enhanced further by coating the liposomes with poly-ethylene glycol (PEG). Selectivity for delivery to some tumors was also enhanced by attaching folic acid to the surface of the PEG-coated liposomes molecules, in order to favor selective uptake by cancer cells that have large amounts of folic acid receptors on their surface. References 166 K.W. Kohn Drugs Against Cancer CHAPTERS Agudelo, 0 ., Bourassa, P., Berube, G., and Tajmir-Riahi, H.A. (2014). Intercalation of antitumor drug doxorubicin and its analogue by DNA duplex: structural features and biological implications. International journal of biological macromolecules 66, 144-150. Arcamone, F., Franceschi, G., Penco, S., and Selva, A. (1969). Adriamycin (14- hydroxydaunomycin), a novel antitumor antibiotic. Tetrahedron letters, 1007- 1010. Benjamin, R.S., Wiernik, P.H., and Bachur, N.R. (197 4 ). Adriamycin chemotherapy-- efficacy, safety, and pharmacologic basis of an intermittent single high-dosage schedule. Cancer 33, 19-27. Bonadonna, G., Monfardini, S., De Lena, M., Fossati-Bellani, F., and Beretta, G. (1970). Phase I and preliminary phase II evaluation of adriamycin (NSC 123127). Cancer research 30, 2572-2582. Burden, D.A., and Osheroff, N. (1998). Mechanism of action of eukaryotic topoisomerase II and drugs targeted to the enzyme. Biochimica et biophysica acta 1400, 139-154. Di Marco, A., Gaetani, M., Orezzi, P., Scarpinato, B.M., Silvestrini, R., Soldati, M., Dasdia, T., and Valentini, L. (1964). 'Daunomycin', a New Antibiotic of the Rhodomycin Group. Nature 201, 706-707. Di Marco, A., Silvestrini, R., Di Marco, S., and Dasdia, T. (1965). Inhibiting effect of the new cytotoxic antibiotic daunomycin on nucleic acids and mitotic activity of HeLa cells. The Journal of cell biology 27, 545-550. Gabizon, A.A. (2001). Pegylated liposomal doxorubicin: metamorphosis of an old drug into a new form of chemotherapy. Cancer investigation 19, 424-436. Keizer, H.G., Pinedo, H.M., Schuurhuis, G.J., and Joenje, H. (1990). Doxorubicin (adriamycin) : a critical review of free radical-dependent mechanisms of cytotoxicity. Pharmacology & therapeutics 47, 219-231. Markman, M. (2006). Pegylated liposomal doxorubicin in the treatment of cancers of the breast and ovary. Expert opinion on pharmacotherapy 7, 1469-1474. Mizutani, H., Tada-Oikawa, S., Hiraku, Y., Kojima, M., and Kawanishi, S. (2005). Mechanism of apoptosis induced by doxorubicin through the generation of hydrogen peroxide. Life sciences 76, 1439-1453. Mukhopadhyay, P., Rajesh, M., Batkai, S., Kashiwaya, Y., Hasko, G., Liaudet, L., Szabo, C., and Pacher, P. (2009). Role of superoxide, nitric oxide, and peroxynitrite in doxorubicin-induced cell death in vivo and in vitro. American journal of physiology Heart and circulatory physiology 296, H1466-1483. Murtagh, G., Lyons, T., O'Connell, E., Ballot, J., Geraghty, L., Fennelly, 0 ., Gullo, G., Ledwidge, M., Crown, J., Gallagher, J., et al. (2016). Late cardiac effects of chemotherapy in breast cancer survivors treated with adjuvant doxorubicin: 10- year follow-up. Breast cancer research and treatment Pacher, P., Beckman, J.S., and Liaudet, L. (2007). Nitric oxide and peroxynitrite in health and disease. Physiological reviews 87, 315-424. Papahadjopoulos, 0 ., Allen, T.M., Gabizon, A., Mayhew, E., Matthay, K., Huang, S.K., Lee, K.D., Woodle, M.C., Lasic, 0 .0., Redemann, C., et al. (1991). Sterically stabilized liposomes: improvements in pharmacokinetics and antitumor 167 K.W. Kohn Drugs Against Cancer CHAPTERS therapeutic efficacy. Proceedings of the National Academy of Sciences of the United States of America 88, 11460-11464. Pigram, W.J., Fuller, W., and Hamilton, L.D. (1972). Stereochemistry of intercalation: interaction of daunomycin with DNA. Nature: New biology 235, 1 7-19. Pommier, Y., Minford, J.K., Schwartz, R.E., Zwelling, L.A., and Kohn, K.W. (1985). Effects of the DNA intercalators 4'-(9-acridinylamino)methanesulfon-m- anisidide and 2-methyl-9-hydroxyellipticinium on topoisomerase II mediated DNA strand cleavage and strand passage. Biochemistry 24, 6410-6416. Rajagopalan, S., Politi, P.M., Sinha, B.K., and Myers, C.E. (1988). Adriamycin-induced free radical formation in the perfused rat heart: implications for cardiotoxicity. Cancer research 48, 4766-4769. Singal, P.K., and Iliskovic, N. (1998). Doxorubicin-induced cardiomyopathy. The New England journal of medicine 339, 900-905. Sinha, B.K., Katki, A.G., Batist, G., Cowan, K.H., and Myers, C.E. (1987). Adriamycin- stimulated hydroxyl radical formation in human breast tumor cells. Biochemical pharmacology 36, 793-796. Sriraman, S.K., Salzano, G., Sarisozen, C., and Torchilin, V. (2016). Anti-Cancer Activity of Doxorubicin-Loaded Liposomes Co-Modified with Transferrin and Folic Acid. European journal of pharmaceutics and biopharmaceutics : o fficial journal of Arbeitsgemeinschaft: fur Pharmazeutische Verfahrenstechnik eV. Szabo, C., lschiropoulos, H., and Radi, R. (2007). Peroxynitrite: biochemistry, pathophysiology and development of therapeutics. Nature reviews Drug discovery 6, 662-680. Thigpen, J.T., Aghajanian, C.A., Alberts, D.S., Campos, S.M., Gordon, A.N., Markman, M., McMeekin, D.S., Monk, B.J., and Rose, P.G. (2005). Role ofpegylated liposomal doxorubicin in ovarian ca ncer. Gynecologic oncology 96, 10-18. Torchilin, V. (2011). Tumor delivery of macromolecular drugs based on the EPR effect. Advanced drug delivery reviews 63, 131-135. Von Hoff, D.D., Layard, M .W., Basa, P., Davis, H.L, Jr., Von Hoff, A.L., Rozencweig, M ., and Muggia, F.M. (1979). Risk factors for doxorubicin-induced congestive heart failure. Annals of internal medicine 91, 710-717. Zwelling, L.A., Michaels, S., Erickson, L.C., Ungerleider, R.S., Nichols, M., and Kohn, K.W. (1981). Protein-associated deoxyribonucleic acid strand breaks in L1210 cells treated with the deoxyribonucleic acid intercalating agents 4'-(9- acridinylamino) methanesulfon-m-anisidide and adriamycin. Biochemistry 20, 6553-6563. 168 K. W. Kohn Drugs Against cancer CHAPTER9 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@nih.gov CHAPTER9 The DNA filter elution story: a new way to measure DNA damage. I had long wondered how the remarkable physical properties of long DNA chains might be used to develop improved methods of measuring DNA damage, in particular DNA strand breaks. I had made several attempts in that direction, but without finding anything very useful - until an answer came from a totally unexpected direction. I had imagined that DNA replication sites might be large enough to be retained on filters after passing gently dissolved cell nuclei through the filters. In trying different ways of dissolving the cell nuclei, one particular solution gave a totally unexpected result. A research breakthrough in a long-desired direction sometimes happens unexpectedly during studies aimed in a different direction. In 1973, I was following up on a then current idea that DNA replication occurred in a part of the chromatin that was attached to the surface membrane of the cell's nucleus. I thought we could filter out the nuclear membrane-bound chromatin and test whether it mostly contained newly replicated DNA. Working with my assistant, Regina Ewig, we deposited cell nuclei on filters that had very small pores and measured how much newly replicated DNA was retained on the filter after passing various solutions of detergents through the filter to dissolve the nuclei and wash away any DNA not attached to nuclear material large enough to be retained on the filter. We saw some selectivity for newly replicated DNA in the nuclear material retained on the filters as we had hoped and therefore tried to optimize this selectivity by using various solutions to wash away the unbound DNA. When we tested various detergent solutions to dissolve the nuclei and letting the unbound DNA drip out, we found that our hypothesis was at least partly correct: an increased amount of newly replicated DNA was retained on the filter. 169 K. W. Kohn Drugs Against cancer CHAPTER9 However, when we tested a solution made alkaline with sodium hydroxide to dissolve the cells, the opposite happened: the newly replicated DNA quickly ran out through the filter, instead of being retained on it, whereas the bulk of the DNA was retained (Kohn and Grimek-Ewig, 1973) (Figure 9.1). I soon realized that the alkaline solution would have caused the paired DNA strands to separate and would have release the short newly replicated DNA segments, allowing them to pass freely through the filter. It seemed that this might be a way to measure DNA damage that produced strand breaks yielding relatively short DNA single-strands that would pass relatively easily though the filter. I had been trying for a long time to think of how to measure DNA breaks in a more sensitive way than the problematic centrifugation methods then in use, and here was a new possibility to explore. The first thing we did to check whether short DNA strands would in fact selectively pass through the filter was to expose the cells to a hefty dose ofx-rays (1 think it was 1000 rad in the first trial), which was known to cause DNA breaks. We trekked over to the Clinical Center with our samples in an ice bucket, to a basement facility where a radiation research group had installed a pair of x-ray tubes, one above and one below the platform for cell dishes or mice so as to subject them to a uniform dose of radiation. We were extraordinarily pleased to see that the DNA from the x-ray'd cells ran through the filter quickly as we had hoped and expected. That was all good and well, but to make it into a useful biological measure of DNA strand breaks, we would have to increase the sensitivity of this alkaline elution phenomenon, as we called it, by a factor of at least 20. Was it possible to do that? A dictum attributed to Thomas Edison came to mind, to the effect that one could often improve a new phenomenon by a factor of 10 by systematically varying the conditions used. In accord with Edison's dictum, we systematically examined the effects of varying the conditions -- such as pH, flow rate, filter type, and compositions of both the detergent solution used to disrupt the cells and the alkaline solution used to elute the DNA through the filter. First, we controlled the outflow from the filter by means of a peristaltic pump, thereby allowing us to control the flow rate of the alkaline solution through the filter. That turned out to be important: slowing the flow to several hours' duration increased the sensitivity of the assay several fold. Our hopes became a reality when optimizing the conditions gave us a method that was sensitive enough to measure DNA damage in mammalian cells at therapeutically relevant dosage of x-rays or other DNA-damaging agents, which no other method of the time had achieved (Kohn and Grimek-Ewig, 1973) (Kohn et al., 1976). During several years of experience and further development, the method acquired a sound quantitative basis for measuring several types of DNA damage, including single strand breaks, double-strand breaks, inter-strand crosslinks, and DNA-protein crosslinks (Kohn, 1996). 170 K. W. Kohn Drugs Against cancer CHAPTER9 A condition that had worried me was the number of cells that was to be loaded onto the filter. Our procedure was to lyse (dissolve) the cells on the filter with a detergent solution in order to free the DNA from other cell constituents before starting the controlled flow with the alkaline solution. I was afraid that the measurements would be sensitive to the number of cells loaded, because that would affect the viscosity of the lysed cell material. I was almost sure that the high viscosity of the DNA would slow the elution rate. It was a great surprise and a great relief, as well as somewhat of a puzzle, to find that the DNA elution rates were completely unaffected by number of cells loaded, as long as the number of cells was not so large as to clog the filter and stop the flow. Evidently, the viscosity was somehow cancelled out by some other factors, or perhaps the viscosity was irrelevant to what was happening on the filter. But, regardless of its cause, the fact that it didn't matter how many cells were loaded onto the filter was important in making filter elution a useful assay for DNA damage. In addition to the method for measuring single-strand breaks, we worked out variations, whereby we could measure other types of DNA damage, such as double-strand breaks, inter-strand crosslinks, and DNA-protein crosslinks (Kohn, 1996). For nearly two decades, the filter elution methods were the methods used in most laboratories for measuring DNA damage and repair in cells. Our early experiments were fraught by an annoying variability in the background DNA elution from undamaged cells, which should have shown very little DNA elution. I came to suspect that the room fluorescent lighting might cause some DNA strand breakage and raise the background measurements. What happened next may be of some interest or at least amusing. I asked Reggie Ewig. my assistant, who at that time carried out most of the assays, to arrange covers for the funnel units that had the filters with the lysed cells, in order to block the fluorescent light coming from the ceiling. Reggie was skeptical and proceeded to obey my direction by preparing dunce-like conical caps to put over the funnels, and she pasted labels on them marked "Kohn's folly." That was an example of Reggie's delightfully independent way of thinking that I treasured. But, when she saw the results, which indeed reduced the background to almost nil, she insisted that there be only enough light to allow the work to be done. She taped the room's light switch in the off position and used only a single desk lamp turned away from the counter on which the work was done. Later, we had yellow ceiling lights installed, which solved the problem. We were now able to measure as little as one DNA strand break per average mouse chromatid (Gantt et al., 1978). A few years later, Matt Bradley, a post-doctoral fellow in our laboratory, did quantitative experiments to determine the extent and conditions for fluorescent light-induced DNA single-strand breaks in cells (Bradley et al., 1978). He also determined the rate that high intensity fluorescent light kills cells and how it relates to production of mutations in the cells (Bradley and Sharkey, 1977). Our new filter methods for measuring various types of DNA damage gave us a new way to study the DNA damage and its repair in cells treated with various anticancer drugs, 171 K. W. Kohn Drugs Against cancer CHAPTER9 carcinogens, and radiation. The filter methods were more sensitive and precise than other methods of the time and allowed us for the first time to actually quantify several types of DNA damage at pharmacologically relevant dosage. The most notable result of our experiments using the filter methods was the unexpected discovery that some anti-cancer drugs blocked topoisomerases (although the term 'topoisomerase' was not yet in vogue). That story, however, will be told in the next chapter (Chapter 10), which is about drug actions on DNA topoisomerases. In the paper by Ray Gantt et al. 1978, by the way, we collaborated with Katherine Sanford's laboratory, which she had continued to lead after Wilton R Earle, who had founded the laboratory, died unexpectedly in 1964 at the age of 62. I am now going to take the liberty to digress with a bit of history about Katherine Sanford and Wilton R Earle (Figure 9.3) who were the first to clone a culture from a single mammalian cell. They did so by sucking a single cell into a micropipette and letting it divide a few times before allowing the new cells to grow out of the end of the capillary (Sanford et al., 1948) (Figure 9.4). I was greatly impressed by that work, when I read their paper in 1953 while at medical school. The laboratory Earle had established at NIH and continued by Sanford was at first meticulously constrained with gloves, masks and gowns like a surgical suit in order to avoid contaminating the cell cultures, which were grown without antibiotics. Katherine had a photograph above her desk of the first mammalian cell to grow as a clone in culture. Around 1970, I used their method to clone a single mouse leukemia cell in a capillary, but without needing their elaborate apparatus, because we used antibiotics and a laminar flow hood to avoid contaminating the cultures. 172 K. W. Kohn Drugs Against cancer CHAPTER9 0.1 w ,- : 0 .01 z 0 ;:: ., :, I ..' ••• w • ••• • I h 0.001 100 ELUTION TIME lmin} Figure 9.1. One of our first DNA alkaline elution experiments. Newly replicated DNA (after 3 or 20 minutes of replication) eluted more rapidly than the cell's full-length mature DNA. Our procedure was to deposit cells on a membrane filter and lyse them with a detergent-containing solution that would disrupt the cells and loosen proteins that were bound to the chromatin. We then pumped a solution containing 0.1 M sodium hydroxide (pH 12.0) slowly through the filter. The solid black symbols show the rate at which full-length DNA strands from eluted from the filter as a function of time. The lower two curves show the higher rates of release of newly replicated DNA from the filter (Kohn and Grimek-Ewig. 1973). The DNA that eluted during the first 10 minutes was mostly newly replicated DNA. ~ :, C Unirrat iated cells 0 r,: ~ b e I I ~ I ,··· \; unirradiated 0.01 1000 rad , ~" 1 • '~ \ • ••• / ·•.. 30 m in repair : "' !!l z~ 0 ., ,·, ··/ ·.. '. ;:: <t :, z ... 0 "'\ "\. ..-: ....· -- ••·· . 'I w- 0 C }".....~-...-. .g 0.001 ..:...,. ~ 50 100 50 100 ELUTION TIME (min) ➔ Figure 9.2. Proof that DNA single-strand segments elute from filters at rates inversely related to the length of the strands (Kohn and Grimek-Ewig. 1973). Cells were x-rayed to produce random breaks 173 K. W. Kohn Drugs Against cancer CHAPTER9 in their DNA. The cells were lysed on a filter and the DNA was eluted with 0.1 M sodium hydroxide as described in Figure 9.1. a) Unirradiated cells. b) Cells irradiated with a 1000 rad dose ofx-rays: their DNA eluted more rapidly than DNA from unirradiated cells. After 30 minutes, the cells repaired most of the damage, and the elution rate of their DNA reverted close to that of unirradiated cells. Figure 9.3. Katherine Sanford (1915-2005) (left) worked with Wilton R Earle (1902-1964) (right) who had come to NIH in 1937 and founded a Laboratory to study the development of mutations and cancer by cells in culture. Having joining Earle's laboratory in 1947, Sanford was appointed to assume the role of Chiefof the Laboratory when Earle died unexpectedly in 1964. Figure 9.4. Katherine Sanford and Wilton R. Earle were first to clone colonies from single mammalian cells (Sanford et al., 1948). On the left is a diagram of the apparatus they used. A bent glass capillary (E) was inserted into a Carrell chamber (F) where cells were growing. Under a microscope, a cell was selected and sucked into the capillary. A cotton-filled syringe (D) prevented bacteria from entering into the capillary. On the right, we see pictures of the cell dividing in a glass capillary. From top to bottom, we see a single live cell in the capillary; cells having divided in the capillary; cells growing out of the end of a capillary and into a growth plate. 174 K. W. Kohn Drugs Against cancer CHAPTER9 How the DNA filter elution phenomenon works. The separated DNA single-strands are long enough to enter multiple pores in the filter as an alkaline solution flows slowly through the filter. The longer the DNA strand, the longer it would take to pass through the filter pores (Figure 9.SA). SINGLE-STRANDED ONA FILTER Figure 9.SA. How a filter may impair a long DNA strand from passing through. The DNA strands were long enough to enter several pores at once, longer strands would take longer to get through (Kohn, 1996). The filter elution procedure was essentially as follows. Cells were deposited on the filter and lysed by adding a detergent-containing solution. The strands of the cell's DNA were then separated by means of an alkaline solution (usually a little above pH 12.0), which caused the paired strands of DNA double helix to separate. As the alkaline solution was then slowly pumped through the filter, the shorter DNA stands came through the filter more rapidly than longer strands. The reason that the longer strands took longer to pass through may be that each strand entered a larger number of pores simultaneously, which made it more difficult for the strand to pass through. The rate at which the DNA strands eluted from the filter told us the average length of the strands, which told us the frequency of single-strand breaks in the cell's DNA. Next, we had to find out how the DNA elution rate would tell us how many strand breaks there were. In other words, we had to calibrate the elution rate relative to strand break frequency. We did that by irradiating the cells with various doses ofx-rays that produced known frequencies of DNA strand breaks and then measured the elution rate for each x-ray dose. The elution rate was beautifully proportional to x-ray dose (Figure 9.5B)! That allowed us to quantify precisely the DNA strand break frequency produced by given treatments with anticancer drugs, carcinogens, or other DNA damaging agents. By comparing the elution rate of DNA from the x-rayed cells with the elution rate of DNA from drug-treated cells, we calculated the DNA break frequency in the drug-treated cells (Figure 9.5B). Moreover, the elution rate measurements were highly sensitive: they could detect DNA strand breaks produced by x-ray doses as low as 30 rad, which is less than 1/100 of the mean lethal dose. This was a major advance over other methods available at the time, both in sensitivity and precision. 175 K. W. Kohn Drugs Against cancer CHAPTER9 A B 0.3 lSOR 0 0 • • 0 0 '£ 0.2 • 0 iI'! 0 '-' • • 300R Ig 0.1 0 • ·= 0 i5 • 0 0 0.2' -- ~- - > -- • ~-- 10 20 0 200 400 600 HOURS OF ElUTION X➔ay Dose (R) Figure 9.5B. The rate at which DNA s ingle-strands passed through the filter depended on the average length of the strands. The DNA was cut at random places by subjecting the cells to various doses of x-rays that produced known frequencies of strand breaks. Panel Ashows the rates of DNA elution from cells irradiated with 0, 150, or 300 rad of x-rays. The vertical axis indicates the fraction of the DNA remaining on the filter as a function time as an alkaline solution (pH1 2.l) was slowly pumped through the filter. The open and closed symbols were from experiments using alkaline eluting solutions of different pH, which showed that beyond a critical alkaline pH, raising the pH further had no effect, which was in accord with the theory of DNA strand separation developed by Paul Doty. Panel B showed that the DNA elution rate increased linearly with x-ray dose, thus linearly with the frequency of DNA strand breaks (Kohn et al., 1976). (The vertical axis in panel A is a logarithmic scale; thus, the elution rates were linear in this plot, which indicated that elution of DNA strands cut randomly by a given dose of x-ray was random in time.) Filter methods adapted to measure several types ofDNA damage. We discovered that another property of the filters we used w as that proteins tended to adhere to them. Therefore, a single-stranded piece of DNA that w as linked to a protein did not pass through the filter, regardless of the length of the strand. To avoid this complication, we applied an enzyme to digest away most of the protein. Also, we used filters that had less tendency to bind proteins. Those two measures together eliminated protein binding to the filters. 176 K. W. Kohn Drugs Against cancer CHAPTER9 However, we took advantage of the ability of certain types of filter bind proteins to create a quantitative assay for DNA-protein crosslinks. The procedure was to subject the cells to a relatively large dose of x-ray (3000 rad) to produce relatively short DNA strand segments and then carry out an alkaline elution. The strand segments that had a DNA-protein crosslink in it stuck to the filter and did not pass through, whereas the other strand segments were able to pass through. The fraction of the DNA that could not pass through the filter allowed us to calculate (with a little bit of algebra) the frequency of DNA-protein crosslinks. We validated the procedure by using the trans-isomer of cisplatin, which we had determined produced exclusively DNA-protein crosslinks (Chapter 3), as shown in Figure 9.6. At that point, we were able to quantify both DNA single-strand breaks and DNA-protein crosslinks in drug-treated cells, which was to lead us to the conclusion that some drugs cause certain enzymes (later identified as topoisomerases) to produce DNA strand breaks and to become linked to the strand ends (Chapter 10). ,.or-- - - - -- ----, p" ■■ ■ ■ ■ • ■■ 2001-1M 20,,M 5 10 15 HOUAS OF ELUTION Figure 9.6. How we measured DNA-protein crosslinks. We treated cells with various concentrations of the trans isomer of cisplatin, which we had found produced exclusively DNA-protein crosslinks (see Chapter 3). We then subjected the cells to a relatively high dose ofx-ray (3000 rad). The bottom curve (marked 0µM) showed that, for cells that received no trans-platinum treatment, about 90% of the DNA eluted rapidly. With increasing doses of trans-platinum, progressively less DNA eluted rapidly - which was the fraction of the DNA that was protein-linked. In order to get a true measure of the fraction of the DNA that was protein-linked, the curves (which were nearly linear) were extrapolated back to zero time (Kohn and Ewig. 1979). I was pleased that my propensity for quantitative studies, engendered by my undergraduate background in physics and experience in Paul Doty's laboratory (see Introduction), yielded these quantitative measures of DNA single-strand breaks and DNA- 177 K. W. Kohn Drugs Against cancer CHAPTER9 protein crosslinks. However, I had no idea that those quantitative measurements would lead us to a major discovery that was pertinent to cancer treatment. My background in physics and medicine seemed to come together, as if out of the blue. We discovered that certain anticancer drugs produced protein-linked DNA strand breaks that were caused by enzymes that came to be known as DNA topoisomerases. But that story will be told in the next chapter (Chapter 10). It was an exciting time for us as additional results poured in from our DNA filter experiments. Some of our most memorable findings (aside from the topoisomerase story that is told in the next chapter): In 1974, we measured the rate of DNA chain growth during DNA replication (Kohn et al., 1974). Then, in 1979, Len Erickson, who was a post-doctoral fellow in our laboratory, carefully investigated intermediate elution pH values and succeeded in fractionating and measuring the growth rates of newly replicated DNA chains of different lengths (Erickson et al., 1979). In 1975, Al Fomace, who was at the time a post-doctoral fellow in our laboratory, used the filter methods to investigate the DNA repair in ultraviolet light (UV)-treated cells. He was first to detect the transient DNA strand breaks expected during nucleotide excision repair (NER) of the UV-induced thymine dimers. He showed the cells from xeroderma pigmentosum (XP) patients failed to produce those strand breaks - as expected, because XP was known to be deficient in NER (see Chapter 23). He also found that XP cells could not repair UV induced DNA-protein crosslinks. When he tested cells from the different subtypes ("complementation groups") of the disease, he noted different degrees of DNA repair deficiencies in the various subtypes (Chapter 23) (Fornace and Kohn, 1976; Fomace et al., 1976). In 1977, we worked out how to measure DNA inter-strand crosslinks by means of the filter methods. We investigated the production and repair of those DNA lesions in cells treated with nitrogen mustard (HN2), BCNU and related drugs (Ewig and Kohn, 1977, 1978) (Figure 9.7). HN2 produced inter-strand crosslinks rapidly and then repaired them over a period of several hours (left panel of Figure 9.7). BCNU, on the other hand, produced crosslinks slowly, as monoadducts were slowly converted to crosslinks (ri9ht panel of Figure 9.7), and there was no evidence of repair, perhaps because those crosslinks, which are between paired guanine and cytosine are more difficult to repair (see Chapter 2). We developed a quantitative model to estimate the frequencies of strand breaks, inter- stand crosslinks and DNA-protein crosslinks, even when these DNA lesions are all present at the same time. Figure 9.8, for example, shows that the calculated lesion frequencies were proportional to the BCNU concentrations and therefore were suitable for quantitative studies of drug actions on DNA (Ewig and Kohn, 1978). 178 K. W. Kohn Drugs Against cancer CHAPTER9 .. 1.0 ... Nitrogen mustard (HN2) BCNU !I .. ;;: s "0 .. cu ... C .iii ~ < .. 25 'o C 0 -e ...l! .2 G.2J11MHNZUIY MJIM IICM.I tlw 10 :ID 10 -· °"" Hours of elution Figure 9.7. DNA inter-strand crosslinking by nitrogen mustard (HN2) (left) and BCNU (right) (Ewig and Kohn, 1977). left panel: after treating the cells with HN2 for 0.5 hours (solid symbols), there was a marked reduction of DNA elution in 300-rad-irradiated cells, indicating that there were many DNA inter-strand crosslinks (curve marked •o hr"). There was little change after 3 hours, but by 25 hours the crosslinks were almost all repaired as shown by the elution curve being nearly back to 300 rad without drug. Right panel: after treating the cells with BCNU for 1 hour (solid symbols) there was only a modest reduction of DNA elution in 300-rad-irradiated cells (curve marked •o hr"). After further incubation, progressively more crosslinks appeared (curves marked •3 hr" and "25 hr"). Thus, BCNU was slow to produce inter-strand crosslinks, whereas HN2 produced them rapidly. These experiments were repeated under conditions where DNA-protein crosslinks had no effect (by digesting the proteins with proteinase-K and using filters having low protein binding ability) and gave essentially the same results. 179 K. W. Kohn Drugs Against cancer CHAPTER9 Otv 181) ... 120 0 IO 100 IO 100 BCNU COHCENTAATION i.,MI Figure 9.8. A quantitative model for inter-stand crosslink and DNA-protein crosslink production gave estimates that were proportional to drug concentrations shown on the horizontal axis of the graphs. Cells were treated with BCNU for 1 hour and were then, either assayed immediately (left) or incubated for 4 hours to allow time for more crosslinks to form (right). Circles (upper line}, assay combination of inter-strand and DNA-protein crosslinks (without proteinase). Triangles (lower line}, inter-strand crosslinks, which developed slowly. At zero time (immediately after 1-hour treatment with BCNU) almost no inter-strand crosslinks had yet formed; hence the results showed DNA-protein crosslinks only (left). After 4 hours (right), inter-strand crosslinks appeared (triangle) (Ewig and Kohn, 1978). In 1978, Matt Bradley, as already mentioned, discovered that fluorescent lights caused DNA single-strand breaks in cells illuminated in culture medium (Bradley et al., 1978). Tissues, however, normally have enzymes that destroy the free radicals that are the likely cause of the observed DNA damage -- the cells in the experiments were illuminated in solutions that lacked those enzymes. In 1979, Bradley and I worked out how to measure DNA double-strand breaks by using solutions of neutral pH for elution from filters (Bradley and Kohn, 1979). Using that method, he together with Warren Ross later observed double-strand break production in cells treated with doxorubicin or other DNA intercaltors, and inferred that the double- strand DNA breaks were produced by a trapped topoisomerase (Ross and Bradley, 1981). In 1985, Neil Gibson and Len Erickson found that a new drug related to the chloroethylnitrosoureas (see Chapter 2) produced inter-strand crosslinks and that the ability of cells to survive the treatment was related to their ability to repair or prevent the formation of those crosslinks (Gibson et al., 1985). Then, in 1987, we found that brain 180 K. W. Kohn Drugs Against cancer CHAPTER9 cancer cell lines grouped according to whether or not they were able to prevent DNA crosslink formation by chloroethylnitrosoureas and that this was related to the ability of the cells to survive (Sariban et al., 1987). As explained in Chapter 2, the ability of the cells to survive these treatments was dependent on whether or not the cells produced an enzyme (MGMT) that quickly removed the chloroethyl groups from the DNA before they could go on to form inter-strand crosslinks. The next chapter will relate what was for us the most exciting finding: the discovery that a DNA topoisomerase enzyme was the target of action of several anti-cancer drugs. References Bradley, M.O., Erickson, L.C., and Kohn, K.W. (1978). Non-enzymatic DNA strand breaks induced in mammalian cells by fluorescent light Biochimica et biophysica acta 520, 11-20. Bradley, M.O., and Kohn, K.W. (1979). X-ray induced DNA double strand break production and repair in mammalian cells as measured by neutral filter elution. Nucleic acids research 7, 793-804. Bradley, M.O., and Sharkey, N.A. (1977). Mutagenicity and toxicity of visible fluorescent light to cultured mammalian cells. Nature 266, 724-726. Erickson, L.C., Ross, W.E., and Kohn, K.W. (1979). Isolation and purification of large quantities of DNA replication intermediates by pH step alkaline elution. Chromosoma 74, 125-139. Ewig, R.A., and Kohn, K.W. (1977). DNA damage and repair in mouse leukemia L1210 cells treated with nitrogen mustard, 1,3-bis(2-chloroethyl)-1-nitrosourea, and other nitrosoureas. Cancer research 37, 2114-2122. Ewig, RA., and Kohn, K. W. (1978). DNA-protein cross-linking and DNA interstrand cross- linking by haloethylnitrosoureas in L1210 cells. Cancer research 38, 3197-3203. Fornace, A.J., Jr., and Kohn, K.W. (1976). DNA-protein cross-linking by ultraviolet radiation in normal human and xeroderma pigmentosum fibroblasts. Biochimica et biophysica acta 435, 95-103. Fornace, A.J., Jr., Kohn, K.W., and Kann, H.E., Jr. (1976). DNA single-strand breaks during repair of UV damage in human fibroblasts and abnormalities of repair in xeroderma pigmentosum. Proceedings of the National Academy of Sciences of the United States of America 73, 39-43. Gantt, R., Parshad, R., Ewig, RA., Sanford, K.K., Jones, G.M., Tarone, RE., and Kohn, K.W. (1978). Fluorescent light-induced DNA crosslinkage and chromatid breaks in mouse cells in culture. Proceedings of the National Academy of Sciences of the United States of America 75, 3809-3812. Gibson, N.W., Erickson, L.C., and Kohn, K.W. (1985). DNA damage and differential cytotoxicity produced in human cells by 2-chloroethyl (methylsulfonyl)methanesulfonate (NSC 338947), a new DNA-chloroethylating agent. Cancer research 45, 1674-1679. 181 K. W. Kohn Drugs Against cancer CHAPTER9 Kohn, K.W. (1996). DNA filter elution: a window on DNA damage in mammalian cells. BioEssays : news and reviews in molecular, cellular and developmental biology 18, 505-513. Kohn, K.W., Erickson, L.C., Ewig, R.A., and Friedman, C.A. (1976). Fractionation of DNA from mammalian cells by alkaline elution. Biochemistry 15, 4629-4637. Kohn, K.W., and Ewig, RA. (1979). DNA-protein crosslinking by trans- platinum(II)diamminedichloride in mammalian cells, a new method of analysis. Biochimica et biophysica acta 562, 32-40. Kohn, K.W., Friedman, C.A., Ewig, R.A., and Iqbal, Z.M. (1974). DNA chain growth during replication of asynchronous L1210 cells. Alkaline elution of large DNA segments from cells lysed on filters. Biochemistry 13, 4134-4139. Kohn, K.W., and Grimek-Ewig, RA. (1973). Alkaline elution analysis, a new approach to the study of DNA single-strand interruptions in cells. Cancer research 33, 1849-1853. Ross, W.E., and Bradley, M.O. (1981). DNA double-stranded breaks in mammalian cells after exposure to intercalating agents. Biochimica et biophysica acta 654, 129-134. Sanford, K.K., Earle, W.R., and Likely, G.D. (1948). The growth in vitro of single isolated tissue cells. Journal of the National Cancer Institute 9, 229-246. Sariban, E., Kohn, K.W., Zlotogorski, C., Laurent, G., D'lncalci, M., Day, R., 3rd, Smith, B.H., Komblith, P.L., and Erickson, L.C. (1987). DNA cross-linking responses of human malignant glioma cell strains to chloroethylnitrosoureas, cisplatin, and diaziquone. Cancer research 47, 3988-3994. 182 K.W.Kohn Drugs Against Cancer CHAPTER 10 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@oib gov CHAPTER 10 The Topoisomerase II Story: from methodology to a new anticancer drug target. Introduction The DNA double helix is naturally twisted. Normally, it has one full twist for every 10.5 base-pairs. But what happens to the twists when the DNA strands are pulled apart during replication or transcription? It is like trying to pull apart a long 2- strand twisted rope. The twists could bunch up becoming harder and harder to pull the two strands of the rope apart. Soon after Watson and Crick solved the structure of double-stranded DNA, Max Delbruck pointed out that the number of twists in the parental DNA helix would still be present after the DNA duplicated (Delbruck, 1954). Those twists would somehow have to be eliminated to allow the pair of newly replicated DNA helices to separate. Delbruck concluded that, in order to accomplish that trick, the cell must have a way to transiently cut DNA strands and allow strands to pass through the gap. Many years later, enzymes that accomplish that feat were discovered and came to be called topoisomerases. Two general types oftopoisomerases were discovered. Type I topoisomerases transiently cut one strand of double-stranded DNA helix and allow the other strand to pass through the gap. Those topoisomerases of type I are the topic of the next chapter. In the current chapter, I focus on type II topoisomerases that transiently cut both strands of a double-stranded DNA helix and allow another double-stranded DNA helix to pass through the gap. That was the type of topoisomerase activity we 183 K.W.Kohn Drugs Against Cancer CHAPTER 10 discovered to be targeted by several anticancer drugs. I will tell how we came to that finding in the first part of this chapter. It was hard to imagine how enzymes could possibly break the DNA while keeping hold of both ends of the broken DNA to prevent them from drifting away from each other, because the ends had to be rapidly resealed after the strand passage. Nature, however, as usual, found a straight-forward solution. When the DNA has replicated and the chromosomes begin to condense on their way to mitosis, the mother and daughter strands remain entangled in a manner that pulling the long DNA stands apart becomes like trying to pull two interlocked rings apart. It can't be done without cutting one of the rings. Type II topoisomerases manage to cut both strands of a DNA double-helix, allow another DNA double-helix to pass through the gap, and then quickly reseal the broken ends. An extreme case where a type II topoisomerase is required is the remarkable DNA structure in the mitochondrium oftrypanosomes, one species of which causes sleeping sickness, transmitted through the bite of the tsetse fly. The structure, called a kinetoplast, consists of a large number of interlocked circles. When the organism divides, its kinetoplast also duplicates. But for that to happen, the grossly interlocked DNA circles have to be disentangled by transiently cutting DNA helices, which is what a type II topoisomerase does (Figure 10.1). Figure 10.1. An extreme case where a type II topoisomerase is urgently needed. The kinetoplast of a trypanosome is its mitochondrial DNA, which consists of a large number of interlocked (catenated) DNA circles. When the kinetoplast duplicates during the organism's division, a type II topoisomerase transiently cuts those DNA circles to allow them to disentangle. Upper panels: left, a trypanosome among red blood cells (which are 7 µmin diameter); center, what a kinetoplast looks like when spread out in an electron microscope; right, the 184 K.W. Kohn Drugs Against Cancer CHAPTER 10 ring symbol of the Olympics Oeux olympiques), illustrating how the kinetoplast DNA circles are interlocked. Lower panels: left, edge ofan untreated kinetoplast; center, brief treatment with a type II topoisomerase; right, full treatment (From (Marini et al., 1980)) Interest in topoisomerases blossomed when we discovered that some important anticancer drugs work by blocking one or another of those two types of topoisomerases. This chapter is about drugs that block topoisomerase II, which was the first topoisomerase found to be blocked by some anticancer drugs. The following chapter (Chapter 11) will be about drugs that bock type I topoisomerases. Discovery The topoisomerase enzymes themselves were discovered before there was any indication that they might be targets of drug action. Topoisomerase type I enzymes were first to be discovered. They were discovered in bacteria and viruses and were initially called "DNA nicking-closing enzymes" or "DNA swivelases" (Champoux, 1978b; Champoux and Dulbecco, 1972; Radding, 1978). As explained by Champoux, the enzymes "introduce a transient single-strand break in duplex DNA and thereby provide a swivel for helix unwinding (DNA swivelase)" (Champoux, 1978b). Those names were later replaced by "topoisomerase" to indicate that the enzyme changes the topology of the DNA (a change in topology of an object occurs when the object has to be cut to make the change). Type II topoisomerases, discovered later, cleave both strands of the DNA so as to form a double-strand break through which another double-stranded DNA can pass before the enzyme reseals the break (Liu et al., 1980; Miller et al., 1981). This amazing ability is important during and after DNA replication, because the new chromosomal DNA would otherwise remain entangled in loops analogous to the interlocking circles in the symbol of the Olympics (Figure 10.1), the interlocking would hinder the proper separation of chromosomes during mitosis. Although the nuclear DNA molecules of animal cells are not circular, they are so very Jong that, when duplicated, they can only be disentangled by transient cutting and strand passage. How topoisomerase II accomplishes that trick will be explained later in this chapter. First clues of anticancer drugs acting on topoisomerases. The previous chapter explained how we discovered that DNA strand breaks increase the rate at which the DNA from lysed cells can pass through filter pores, and how we used this phenomenon to work out methods to quantify the frequencies of both DNA single-strand breaks and DNA-protein crosslinks produced by various DNA- damaging drugs and agents. Quite unexpectedly, those studies led us to discover 185 K.W.Kohn Drugs Against Cancer CHAPTER 10 drug actions that we attributed to an effect on an enzyme that had the properties of a nicking-closing enzyme (later called topoisomerase); such enzymes were known in bacteria but had not yet been found in animal cells. Here is how that discovery came about: In 1978, a young physician, Warren E. Ross, having completed his first year as a Clinical Associate in the National Cancer Institute, joined my laboratory to gain some research experience. At that time, we were studying DNA damage and repair produced by various anticancer drugs in cells. We had developed a new technique using filters that allowed us to measure DNA stand breaks and DNA crosslinks, both between the paired strands, and between DNA and proteins (Kohn and Ewig, 1979). The story of that technique was told in the previous chapter (Chapter 9). Warren wanted to apply that methodology to doxorubicin, a promising drug that interested him in his Clinical Associate year. Doxorubicin had been reported to break DNA strands in studies that used a previous less precise and less sensitive ultracentrifugation method. We fully expected that using our new filter-based technique, we would easily confirm the production of DNA breaks by doxorubicin in mammalian cells, as had invariably been the case with several other DNA-breaking agents that we had tested (Erickson et al., 1977; Fomace et al., 1976). However, Warren's repeated attempts to confirm doxorubicin-induced DNA breaks using our filter method failed to show any sign of DNA breakage whatsoever (arrow in the left panel of Figure 10.2). His experiment however suggested that doxorubicin produced DNA-protein crosslinks: the lower two curves in the left panel of Figure 10.2, showed that using x- rays to produce strand breaks yielded less than the expected rate of elution (see legend to Figure 10.2) . We thought that doxorubicin failed to show any DNA strand breaks, because the drug might have produced an excess of DNA-protein crosslinks, which could have hidden the strand breaks -- because the DNA-bound proteins could have caused all of the DNA fragments to stick to the filter. That idea seemed to be confirmed, because digesting the lysed cells with a proteinase before alkaline elution, produced an increased elution rate that confirmed the production of DNA strand breaks by doxorubicin (ri9ht panel of Figure 10.2). Moreover, when Warren applied our protocol for protein-digestion (see Chapter 9), the results were astounding: doxorubicin then produced a beautiful pattern of dose-dependent strand breakage (Figure 10.3). But protein digestion was needed to reveal those breaks - because the DNA fragments were completely hidden by being linked to proteins that stuck to the filter. In order to hide the strand breaks so completely, however, we thought a large excess of DNA-protein crosslinks relative to strand breaks would be needed. We were able to check on that, because we had recently worked out how to quantify both strand breaks and DNA-protein crosslinks (Chapter 9) (Kohn and Ewig, 1979). 186 K.W.Kohn Drugs Against Cancer CHAPTER 10 Our ability to quantify the strand breaks and DNA-protein crosslinks became essential to attributing the drug action to a nicking-closing enzyme (topoisomerase). The results of those quantifications at first presented a big surprise and a puzzle. They showed that there was NO excess of DNA-protein crosslinks over strand breaks. In fact, repeated measurements with doxorubicin, as well as some other DNA intercalators (such as ellipticine) consistently showed that the number of the two types of DNA lesions were equal, within experimental error! I thought that amazing and suspected that there must be some connection between the strand breaks and the DNA-protein crosslinks. They must be causally connected in someway. Without proteinase With proteinase 300 A~,. Ad1•amvc1ri, 2.8..-M A B 08 06 04 08 0 .6 04 Elution Time ➔ Figure 10.2. This experiment by Warren E. Ross in 1977 in my laboratory showed, surprisingly, that doxorubicin (Adriamycin) at first showed no increase in DNA alkaline elution rate, thus no indication ofany strand breaks (arrow in left panel). However, the elution rate of the DNA after subjecting the cells to 300 rad of x-rays just before lysis of the cells was reduced in the doxorubicin-treated cells, which suggested the presence of DNA- protein crosslinks (lower two curved in the left panel). When the assays included digestion of the lysed cells with proteinase, however, doxorubicin showed the increased DNA elution rate expected for the presence of DNA stand breaks (arrow in the right panel). All together, these results suggested that doxorubicin produced both strand breaks and DNA-protein crossllnks (Ross et al., 1978). 187 K.W.Kohn Drugs Against Cancer CHAPTER 10 1.0~~===~ 0.9 0.8 Control 0.7 fil ~ 06 I;; 0: <C 0.6 z 0 z~ 0.4 0 ['; <( 0: IL 0.3 ADRIAMYCIN 1.4 02.__....,,._,_.,...,__ .,...,__....,,._,_ _.,...,__ _....,,...~ _, 0.9 0.8 0.7 0.6 0.6 0.4 FRACTION 'H-ONA RETAINED Figure 10.3. Doxorubicin-induced DNA strand breaks were beautifully revealed after the DNA-linked protein was digested away. In this experiment, doxorubicin-treated cells that were then lysed on the filter were subjected to a protein-digesting enzyme (proteinase K) before pumping an alkaline solution through the filter (as described in Chapter 9). DNA strand breaks are seen to be in proportion to the doxorubicin dose. When the protein- digestion step was omitted, no DNA strand breaks could be seen (elution curve similar to that labeled "control") (Ross et al., 1979; Ross et al., 1978). (Adriamycin is another name for doxorubicin.) The next notion that dawned was that maybe the DNA-linked protein was actually an enzyme that produced the strand break and that the drug caused the enzyme to remain linked to one end of the break it produced. Then every DNA strand segment would have a protein linked to it and the number breaks and DNA-protein crosslinks would be equal, as observed in our experiments (Ross et al., 1979). It is not often that one experiences the delight of imagining something important that perhaps no one had thought of before and having it come to fruition. But to verify that idea required some calculation. Three models could be considered for the distribution of the strand breaks and the DNA-protein crosslinks (Ross et al., 1979). The models are described in Figure 10.4. Model I assumed a random distribution of both strand breaks and DNA-protein crosslinks; this model failed, 188 K.W.Kohn Drugs Against Cancer CHAPTER 10 because an equal frequency of the two DNA lesions would have left some DNA strands without protein links, contrary to our evidence. Model Ill assumed one protein bound to every strand segment anywhere along the segment; this seemed an unlikely circumstance, because it was difficult see what could have brought about such an arrangement. Model II was plausible if the linked protein was in fact an enzyme that produced the breaks and if doxorubicin caused the enzyme to remain attached to one end of the break it produced. To recapitulate, the equal frequency we observed of doxorubicin-induced strand breaks and DNA-protein crosslinks was at first puzzling. because if the two types of DNA lesions were randomly distributed along the DNA, some broken strands would by chance have been free of protein and therefore should have passed though the filter, contrary to our observations. I then reasoned that the breaks could have been completely hidden, as observed, if there were just one protein linked to each broken strand. That, at first seemed unlikely, but I soon realized that it could be the case if each protein molecule were bound consistently to one end of each break site (Figure 10.4, Model II). Algebraic analysis of our data was consistent with that possibility (Ross et al., 1979; Ross et al., 1978). Support for that idea came from measurements of several other drugs that too produced DNA strand breaks that were hidden by DNA-linked protein with equal frequencies of DNA strand breaks and DNA-protein crosslinks. As already said, if a protein were linked consistently to one end (5' or 3') of each strand break, then perhaps the protein was an enzyme that produced the break. An enzyme with that property had already been described time in bacteria: the already mentioned "DNA nicking-closing enzyme" or "DNA swivelase" (Champoux and Dulbecco, 1972). We therefore proposed that a type of nicking-closing enzyme existed in mammalian cells and that doxorubicin (as well as other DNA intercalating agents that we observed to produce similar results (equal numbers of DNA strand breaks and DNA- protein crosslinks) caused the enzyme to become blocked in an intermediate state where the break had been produced but had not yet resealed. Therefore in 1979 we "proposed that intercalation-induced distortion of the DNA helix leads to strand scission by a nuclease which becomes bound to one terminus of the break so as to form a DNA-protein crosslink" (Ross et al., 1979). Nicking-closing enzymes (also called "swivelases" or "DNA unwinding enzymes"), were soon found in mammalian cells (Champoux, 1978a) and were later dubbed "topoisomerases." Those studies gave the first clue that drugs, such as doxorubicin, trap a topoisomerase in a state where the DNA strands are cleaved while the enzyme remains bound to the ends of the broken strands. (The discovery oftopoisomerases would not by itself have suggested that those enzymes might be therapeutic targets of certain anticancer drugs.) 189 K.W.Kohn Drugs Against Cancer CHAPTER 10 ,, I. Random , Il. Bound to One Terminus _ _ ____., _j ' - - ~' - ~' m. One Crosslink per Segment ' _t_ ' ' , Figure 10.4. 1considered three models to account for our observation that doxorubicin (as well as some other DNA intercalators, such as ellipticine) produced equal numbers of strand b reaks and DNA-protein crosslinks. The lines in the diagrams represent DNA strands with interruptions at break sites. The black circles represent protein molecules bound to the DNA strands. Model I was for proteins bound at random places on the DNA; note that by chance some broken DNA pieces have no protein attached. Model Ill for one and only one protein randomly placed on each DNA segment was unlikely, because how would the linked protein know where the breaks were located? Model II, the bound-to-one-terminus model, was for a protein bound consistently to one end of each break. Quantitative examination of the data together with some algebra was consistent only with Model II. The conclusion that there was a protein bound consistently to one end of each break (Model II) suggested that the DNA-bound protein molecules were in fact enzyme molecules that produced the breaks and remained bound consistently to one end (S' or 3') (Ross et al., 1979), as was later found to be the case for topoisomerases. The next step was to demonstrate the effect of the drugs on purified topoisomerase enzyme or in solutions extracted from cells containing the enzyme. Janek Filipski, a Polish visiting scientist in our lab experienced considerable frustration trying to accomplish this. He succeeded in showing that cell extracts contained an enzyme that produced the expected drug effects - DNA stand breaks with associated DNA- protein crosslinks. However, when he tested the then-known topoisomerase enzyme, the drugs had no effect (Filipski et al., 1983a, b ). Soon after he published that work, however, the difficulty was revealed: there were in fact two kinds of topoisomerases, and the one he tested - the only one that was known at the time -- was the wrong one. Only topoisomerase I was known at the time of his experiments. But the enzyme the drugs acted on was topoisomerase II exclusively. That new enzyme was being discovered, unbeknownst to us, during the latter part of his studies. In 1980, Leroy Liu, working with Bruce Alberts at the University of California in San Francisco, had isolated the enzyme that came to be known as topoisomerase II (Liu et al., 1980). About 2 years later, after Leroy Liu had moved to Johns Hopkins University in Baltimore, I visited his laboratory and we discussed the possibility that the drug effects that we could not attribute to actions on topoisomerase I were 190 K.W. Kohn Drugs Against Cancer CHAPTER 10 actually caused by actions on his topoisomerase II. After preliminary experiments to get the drug treatment conditions right, Leroy Liu and his colleagues, as well as John Minford, Yves Pommier and Leonard Zwelling in my laboratory, soon confirmed that indeed doxorubicin trapped topoisomerase II bound to one end of a DNA break, an intermediate state in the enzyme's breakage/resealing cycle (Minford et al., 1986; Nelson et al., 1984; Tewey et al., 1984a; Tewey et al., 1984b). In addition to doxorubicin, we found that some other DNA intercalating drugs, such as amsacrine (m-AMSA) and ellipticine, also trapped topoisomerase II DNA-cleavage complexes in a fashion similar to doxorubicin (Pommier et al., 1985). The nature of the protein-linked DNA strand breaks that we attributed to trapping to topoisomerase II was further revealed by studies of the action of amsacrine ( m- AMSA) by Leonard Zwelling in my laboratory (Zwelling et al., 1981). Len added m- AMSA to cultures of mouse leukemia cells and measured the production ofprotein- linked DNA strand breaks using our filter methods (Kohn, 1996) (Figure 10.5). Ifm- AMSA produced DNA breaks like an ordinary DNA damaging agent, the breaks would continue to accumulate while active agent was present. He found, however, that the breaks produced by m-AMSA soon leveled off, and then remained at a constant level as long as the drug was present. When the drug was removed, the breaks rapidly vanished, showing that the enzyme continued to be active and was necessary to maintain the high plateau levels of protein-linked DNA strand breaks. We concluded that, in the presence ofm-AMSA, there was a rapid equilibrium between the formation and the reversal of the DNA breaks. The simplest explanation was that the drug bound to and trapped an intermediate state of an enzyme that continually opened and closed DNA breaks. In other words, the drug stabilized a state of the enzyme between cleavage and resealing of the DNA Ordinarily, the enzyme resealed the breaks so quickly that one did not see the cleaved state. With drug bound to the enzyme, the duration of the cleaved state was prolonged, producing the steady-states seen in Figure 10.5. 191 K.W.Kohn Drugs Against Cancer CHAPTER 10 I m-AMSA 60 90 TIM! FOLLOWING OAUG ADDITION CminutH) Figure 10.5 Treatment of cells with m-AMSA (amsacrine) caused DNA strand breaks to appear and reseal rapidly, consistent with an effect on a topoisomerase. In the continued presence of drugs, the number of strand breaks increased and soon reached a plateau that was higher when the drug concentration was higher. After 60 minutes, when the drug was removed (arrow), the strand breaks soon vanished. This result showed that there was a rapid equilibrium between formation and reversal of the strand breaks, and the number of strand breaks at equilibrium increased with drug concentration (Zwelling et al., 1981). An ordinary DNA damaging agent would have continued to increase the number of strand breaks, in contrast to the flat equilibria seen here. Later it turned out that another drug, camptothecin, trapped topoisomerase I in a reversible reaction where only one of the strands of the DNA double helix was cleaved. Topoisomerase I, like topoisomerase II, undid excessive DNA twists, but did so by producing DNA single-strand breaks, as opposed to the double-strand breaks produced topoisomerase II. The camptothecin story is related in the next chapter. How doxorubicin and other intercalator-type drugs trap DNA- topoisomerase II complexes. In 1989, when purified topoisomerase II and DNA sequencing gels had become available, we wondered whether the drugs had preferences for the DNA sequences where they incited the enzyme to cleave the DNA. We found that DNA cleavage sites did indeed occur at preferential sites (Figure 10.5). To determine whether the enzyme preferred to cleave at particular DNA sequences, we examined a large number of topoisomerase II DNA cleavage sites trapped by 192 K.W.Kohn Drugs Against Cancer CHAPTER 10 various intercalator-type drugs (Capranico et al., 1990a; Capranico et al., 1990b) (Pommier et al., 1991). Figure 10.6 shows one of our first DNA sequencing gels that indicated exactly where in the DNA sequence the drug-induced cleavage sites were located. When we began that investigation, however, we did not suspect that it was to give a clue to the structure of the trapped DNA-topoisomerase complex itself. Our first notable observation was that doxorubicin breaks occurred preferentially where there was an A (adenine) adjacent to the cleavage site on the side toward the 5' end of the DNA strand. For amsacrine (m-AMSA) there was also a preference for a particular base at the cleavage site, but in that case the preference was for an A on the side towards the 3' end of the broken DNA strand. For etoposide and teniposide (VP-16 and VM-26), again there was a preference for a particular base at the cleavage site, but the preference in that case was for a Con the 5' side (Figure 10.7). Those findings were exciting. because they had the feel of a mathematical quality, like a code of some kind. --- AB>.CDEF - -- -- Figure 10.6. One of our first electrophoretic DNA sequencing gels showing cleavage of DNA at specific sites induced by mammalian topoisomerase II in the presence of doxorubicin (Capranico et al., 1990b). A, DNA alone. 8, DNA plus topoisomerase II; these 2 lanes show that neither DNA alone nor topoisomerase alone nor DNA with only drug. caused breaks. C- F, DNA plus topoisomerase II plus increasing concentrations of doxorubicin; the bands show where in the DNA sequence cleavage occurred in the presence of topoisomerase II plus doxorubicin. As the concentration of doxorubicin was increased, the bands became darker, indicating increased frequency of breaks at those sites. (The lane labeled A. shows marker bands for determination of the exact positions of the cleavage sites in the DNA sequence.) 193 K.W. Kohn Drugs Against Cancer CHAPTER 10 The preference for a base on one side or the other of the break site, and its dependence on the identity of the drug, suggested that the drug molecule stacks against one side or the other of the break site the way DNA intercalators stack against the base-pairs (Pommier et al., 1991; Pommier et al., 2000). We guessed (correctly) that the drug stacked against a particular base-pair at the cleavage site (Figure 10.7), as later shown in a crystallographic structure . The drug plastered against a base-pair at the end of the break prevented the topoisomerase II from closing the break. The drug thus trapped the DNA- topoisomerase complex in a state where the DNA was cleaved and could not reseal. Since the bindings were reversible, the drug eventually dissociated and allowed the break to reseal, as proved in Figure 10.5. The cell however would not rely on the spontaneous dissociation of the drug, because it took some time, during which an encounter with a transcription or replication fork could have lethal consequences, as will be explained later in this chapter. The cell therefore has repair machinery to clean up (albeit slowly) the trapped complexes. DOX mAMSA s .........,........ 3 ..1.1..i..i.u.i..:~LU Elllpt lclne VM26 194 K.W. Kohn Drugs Against Cancer CHAPTER 10 Figure 10.7. Preferred positions of the drugs (solid rectangles) atthe drug-topoisomerase II cleavage sites. We inferred these configurations from our observed site preference observations (such as shown in Figure 10.5 (Pommier et al., 1991)). This model was later confirmed by x-ray crystallography (Wu et al., 2013). The DNA base preferences for the immediate neighbors at the break site were, as indicated in the figure: for doxorubicin (DOX), A on the S' side of the break; for amsacrine (m-AMSA), A on the 3' side; for ellipticine, Ton the S' side; for teniposide (VM26) and etoposide (VP16), Con the 3' side. Topoisomerase II consists of two identical molecules bound together but here shown separated for clarity (Figure 10.7): one cleaves one DNA strand, and the other cleaves the other strand. The two cleavage sites were always separated by 4 base-pairs, and the base preferences were similar at the two sites. How type-2 topoisomerases undo entangled DNA helices. The problem of separating interlocked newly replicated DNA loops (Figure 10.1) at first seemed almost insurmountable, but topoisomerase II manages to do it! It is like a conjuring trick that passes one rope through the middle of another. How one DNA double strand could be made to pass through another, while keeping hold of the strand ends, was at first hard to imagine. But, as so often is the case, evolution discovered a solution, which turned out to be quite simple. It was discovered that the magic happens through the cooperation of two identical topoisomerase II molecules (Figure 10.8): the topoisomerase molecules first cut one DNA double-helix (green), then allow the other (red) to pass through the gap and out the other side; then the molecules quickly and perfectly make the green DNA whole again. It happens quickly and perfectly. The key is that two topoisomerase molecules cooperate so that the cut DNA ends are always bound to the topoisomerases and never free to drift apart, and that the topoisomerase II pair of molecules have two places where they can bind each other alternately to let the passing double helix come in from one side and out the other (Figure 10.8). 195 K.W. Kohn Drugs Against Cancer CHAPTER 10 A B C ATPase Cleavage domai ns Figure 10.8. How topoisomerase II passes one DNA double helix (red) through another (green) (Vos et al., 2011). Two identical topoisomerase molecules cooperate to accomplish this magic. The ATP/ADP units provide the energy that drives the machine. (From Nature Reviews Molec Cell Biol 2011.) Doxorubicin and other DNA intercalation-type drugs bind to an intermediate state (such as B, in Figure 10.8), where the DNA is broken; the bound drug prevents the break from being resealed. Figure 10.8. shows the structure of this intermediate state as revealed by x-ray crystallography; we were happy to see our inferred model (Figure 10.7) confirmed by x-ray crystallography (Figure 10.9). 196 K.W.Kohn Drugs Against Cancer CHAPTER 10 Figure 10.9. Structure of DNA-topoisomerase II (Top2) trapped by amsacrine (m -AMSA) in a state where both DNA strands are cleaved (Wu et al., 2013). The structure was based on x- ray crystallography. The upper part of the figure shows the topoisomerase II homodimer (yellow and pink) and the bound DNA (red) . Below is a detailed view of the cleaved DNA with amacrine intercalated as we had surmised in Figure 10.7. The DNA (maroon) is shown with the base-pairs edge-on connected to the DNA backbone via the pentagonal deoxyribose units. Two amsacrine molecules (blue) are DNA-bound at the two break sites, which are separated by 4 base-pairs as we saw in Figure 10.7. In the absence of drug, those 4 base- pairs would come apart and the DNA double-strand break would open and allow another DNA helix to pass through. The complementarity of those 4 base-pairs then helps the two parts of the broken strand to fit together and restore the original unbroken DNA. Two alpha-helical parts of the topoisomerase II protein that interact with the DNA and/or amsacrine at the break sites are shown in yellow and pink (Wu eta!., 2013). How the topoisomerases, their structure and functions were discovered. The story goes back to 1969, when James C. Wang, then at the University of California at Berkeley discovered an enzyme activity in E. coli extracts that relaxed over-twisted (supercoiled) DNA. Two years later, he had purified the enzyme and called it omega protein (Wang, 1969, 1971). He knew that to relax this circular 197 K. W.Kohn Drugs Against Cancer CHAPTER 10 supercoiled double-stranded DNA, a strand had to be cut to allow the DNA to relieve its excessive twists and relax to its normal degree of twisting (about 10.5 base-pairs per twist). He thought at first that two enzymes were needed: one to cut a strand and another to re-ligate the broken strand after the DNA spontaneously relieved its excessive twists, but soon found that it was all done by his single purified enzyme. Moreover, the DNA-relaxing activity of the enzyme did not require energy - which a ligase enzyme would have required. His new purified enzyme needed a name. His name for the enzyme - omega protein - had a non-committal quality to it, and so, that name was soon replaced by names reflecting what the enzyme did: DNA-unwinding protein or DNA swivelase. After more investigation by several researchers, its name settled down to the modern: topoisomerase I. Topoisomerase 1 relaxed supercoiled double-helical DNA by passing one of the strands through a gap created by cleaving the other strand of the same double-helix. In 1980, as already mentioned, Leroy Liu and Bruce Alberts, discovered another type of topoisomerase, which functioned by passing one double-helix through a break created in both strands of another double-helix. They called their new enzyme topoisomerase II. In 1983, Leroy Liu, having moved to Johns Hopkins University in Baltimore, Maryland, had led his research group to carry out the purification of topoisomerase II from several types of mammalian cells (Liu et al., 1983). The new topoisomerase served to disentangle DNA during entry into mitosis (Champoux, 2001) and, as described above, was the enzyme that his laboratory and mine discovered to be targets for anticancer drugs, such as doxorubicin. How drugs that poison topoisomerases kill cancer cells. You might think that the toxic effects of a drug that poisons an enzyme would be overcome if the cell increased the amount of the enzyme, so that some enzyme activity would still be retained even in the presence to the drug. According to that viewpoint, cells would become resistant to the drug if the amount of the drug's target enzyme were increased, which is often the case of other enzymes. However, for topoisomerases the opposite was found to be true. Cells became drug-resistant if they reduced the amount of topoisomerase they made, because it was the drug- topoisomerase combination that was toxic to the cell (Nitiss, 2009) (Pommier, 2013). But why would a drug-topoisomerase complex, sitting quietly on the DNA cause trouble? The trouble arose when a DNA replication or transcription machine came along and encountered one of those complexes. The encounter created an abnormal DNA structure, such as a double-strand end, which was hard to repair, and such lesions in the DNA could ultimately kill the cell (Hsiang et al., 1989). 198 K.W.Kohn Drugs Against Cancer CHAPTER 10 How cells defend against drugs that poison topoisomerases. Three processes were discovered that helped prevent poisoned topoisomerase from leading to a lethal outcome. First, the cell had enzymes that removed the trapped topoisomerase from the DNA before anything bad happened. Second, if a DNA replication or transcription machine had already collided with a drug-trapped topoisomerase, a DNA repair mechanism -- DNA nucleotide excision repair - was found to come into play to restore the integrity of the DNA. Third, a defense against a lethal outcome was initiated by signals to the cell cycle control systems to delay replication and mitosis, so as to give more time for repair to take place before disastrous consequences occur. If there were too many trapped complexes to handle, however, the cell could give up and undergo programmed suicide (apoptosis). The first countermeasure mentioned above -- removal of the trapped topoisomerase - became fairly well understood. It was found to be accomplished by enzymes called tyrosine-DNA-phosphodiesterases (TD Pl and TDP2). Phosphodiester bonds normally link between nucleotide units in the DNA sequence When a DNA- topoisomerase complex has cleaved a DNA strand, a phosphodiester bond links one end of the cleaved DNA to a tyrosine amino acid of the topoisomerase. TDPl and TDP2 juggled the phosphodiester bond to make the topoisomerase protein come off (at which point the drug also came oft), which then allowed the DNA break to reseal. When the DNA strand break could not close because of an intercalated drug, TD Pl or TDP2 would break the bond between the DNA end and the topoisomerase's tyrosine. The importance of this action was shown in a report that TDP2 helps cells survive topoisomerase II trapping by the Top2 blocker, etoposide (Kont et al., 2016) (Figure 10.10). Actually, the process was a bit more complicated. Before the TD Pl or TDP2 could have access to cleave the tyrosine bond to the DNA, a large part of the topoisomerase protein had to be digested away. This was done by an important (and amazing) machine in the cell, called a proteasome. (Other types of DNA damage, such as produced by alkylating agents can also trap topoisomerases (Schellenberg et al., 2016), but that is generally a minor action relative to other effects of those agents.) The second defense: nucleotide excision repair, as well as the third defense: signaling to the cell cycle control system to delay replication and the initiation of apoptosis will be the subjects of later chapters. Much effort aimed to unravel the complexities of how the DNA lesions caused by topoisomerase-trapping drugs signaled to the DNA repair and cell cycle control systems to initiate further survival actions in the cell (Cristini et al., 2016) (Sakasai and lwabuchi, 2016). 199 K.W.Kohn Drugs Against Cancer CHAPTER 10 DNA Before repair by 5' ~ Base O P -~· TOP, proteasome digests away most of the T0P0 protein O Intercalated l TDPl/2 I drug O=P - o- Tyrosine-OH bp"". 0 3' s DNA Figure 10.10. How a trapped topoisomerase-DNA complex is repaired by the TDPl and TDP2 enzymes. The first two steps listed below are part of the normal function of a topoisomerase. The subsequent steps describe what happens if a DNA-intercalating drug traps the topoisomerase on the DNA. (1) The topoisomerase's tyrosine oxygen atom attacks the phosphorus atom (P) that joins two nucleotide units in a DNA strand (red arrow). (2) At the same time that the tyrosine oxygen binds to the P, an oxygen atom dissociates from the P, producing a break in the DNA strand. The oxygen atom that dissociates from the Pis either the one connected to the S' part of the DNA or the one connected to the 3' part of the DNA, depending on the type oftopoisomerase, but that is a minor point here. (3) An intercalator-type drug (red box) binds by being plastered against a base-pair adjacent to the strand break and prevents the resealing of the break, thereby trapping the topoisomerase in this never-never state. Some drugs bind to the base towards the S' part of the DNA strand, and some towards the 3' part of the DNA strand, as shown in Figure 10.6. (4) A "proteasome" digests away most of the topoisomerase protein. (5) Finally, TD Pl or TDP2 (depending on the type of topoisomerase) breaks the bond between the tyrosine oxygen atom and the DNA's P atom, while reforming the bond between the P and the previously dissociated DNA oxygen atom. In the end, normal DNA structure has been perfectly restored. The Etoposide Story So far, all the Top2-blocking drugs mentioned had the ability to intercalate in DNA, which aided their discovery. But there was a different group ofTop2 blockers. Here is the story. It starts with Hartmann Stahelin and coworkers at Sandoz in Basie, Switzerland, who were manipulating the chemistry of podophyllotoxin, which was known to prevent cells from passing through metaphase of mitosis (Keller-Juslen et al., 1971). 200 K.W.Kohn Drugs Against Cancer CHAPTER 10 The drug was obtained by extracting it from the roots of a poisonous plant: the American mandrake or Mayapple (Figure 10.11). Podophyllotoxin had anticancer activity in mice but was found to be too toxic for use in patients. Therefore, the chemists at Sandoz made chemical modifications of the compound in search of a less toxic drug. They made almost SO variations of the chemical structure of podophyllotoxin, several of which increased the survival of mice bearing leukemia L1210. There was a big surprise, however -- a modest structural change in the podophylotoxin structure completely changed what the drug did in the cell: the toxicity to cell was retained, but the mechanism responsible was entirely different: blocking mitosis was not what caused the cell toxicity. Moreover, the structurally altered drugs were much more effective against cancer. The chemical change was merely to remove a methyl group and to switch the steric configuration of one of the bonds (Figure 10.12). This modest change eliminated (or greatly reduced) the ability of the drug to inhibit cells in metaphase of mitosis. Instead, the cells were prevented from even starting the process toward mitosis. This was reported by Stahelin in 1970, who surmised correctly that the demethylepipodophyllotoxins (the chemical name of the new compounds) killed cells by an entirely new mechanism (Stahelin, 1970). The new compounds were later discovered to block topoisomerase 11. We became accustomed to that unwieldy chemical name and were pleased to let it fade in memory when it was superseded by new names for the drugs: etoposide and teniposide. (You might suppose that the name "etoposide" referred to its action on topoisomerase, but it seems that name was applied before its action on topoisomerase was known!) Thinking back on this story, the remarkable switch in biological target of action produced by simple changes in chemical structure was remarkable and instructive. It challenged the presumption that the drugs with similar chemical structure would necessarily act on the same target 201 K.W.Kohn Drugs Against Cancer CHAPTER 10 Figure 10.11. The American mandrake or mayapple, a poisonous plant, whose roots were the source podophyllotoxin, an inhibitor of mitosis. Chemical modifications of the compound yielded the topoisomerase II blockers and anticancer drugs, etoposide and teniposide. (Photograph from Wikipedia.) 0 ~ podophyllotoxin Etoposide (VP16) Figure 10.12. Chemical structures of podophyllotoxin and etoposide (VP16). The chemical changes that switched the mode ofaction were (1) removal of the methyl (CH3) group (red square); and (2) change of the direction ofone of the bonds (red arrow), from the bond that points up to the one pointing down relative to the plane of the page. Teniposide (VM26) was a minor chemical modification of etoposide. (The chain in the upper part of the structure on the right was not essential to the change in the manner of the drug's action.) 202 K.W. Kohn Drugs Against Cancer CHAPTER 10 Thus, the demethylepipodophyllotoxins surprised the researchers, because, although this modest chemical modification of podophyllotoxin increased the ability to extend the survival of mice with cancer, the new compounds did so by an entirely new action. Instead of blocking cells in the middle of mitosis, they instead blocked the ability of cells to begin condensing chromosomes as prelude to entry into mitosis (Grieder et al., 1974). Because of this drastic change in how the new compounds worked, they were given the tentative drug names, VP16 (later, etoposide) and VM26 (later, teniposide). It was natural to suppose that cells were stopped from starting mitosis by inhibiting DNA synthesis. But the problem with that supposition was that the inhibition of entry into mitosis occurred sooner and at lower drug dose than the inhibition of DNA synthesis (Grieder et al., 1974). Therefore, something other than DNA synthesis inhibition had to be what caused the inhibited cell division. It was a puzzle. Then, in 1976, Susan Horwitz at Albert Einstein College of Medicine in New York reported that etoposide produced DNA strand breaks that gradually disappeared, presumably by being repaired. But the cause and significance of that finding remained a mystery. In 1984, Leroy Liu's research group focused their attention on the demethylepipodophyllotoxins - etoposide (VP16) and teniposide (VM26) - because of the chromosome anomalies produced by those drugs. They thought, it seems, that the drugs might be preventing the DNA from untangling at mitosis by inhibiting their newly discovered topoisomerase II (Top2), based in part on our finding that doxorubicin and several other anticancer drugs acted by way of a topoisomerase, in particular, topoisomerase II (Minford et al., 1986; Ross et al., 1979). Sure enough, when they tested the effect of the drug on purified topoisomerase, it was clear that the demethylepipodophyllotoxins inhibited the enzyme (Ross et al., 1984). That was particularly interesting. because, unlike the previously found Top2 inhibitors that were all DNA intercalators (Pommier et al., 1985), the epipodophyllotoxins were thought to lack DNA intercalating activity. The molecules, however, do have polycyclic aromatic groups that may intercalate in the DNA-topoisomerase complex by stacking against a base-pair (Figure 10.12). In our studies ofTop2 inhibitors, we had looked to see what base-pair preference the drug may have for where it traps Top2 in a state where the DNA strands are cleaved. When we tested etoposide and teniposide, we found they had unique preferences for where they trapped Top2 (Figure 10.7) (Pommier et al., 1991). The mechanism of the reaction seemed to involve an initial interaction between drug and enzyme, rather than between drug and DNA (Burden et al., 1996). Therefore, these Top2-blocking drugs were inferred to act in a manner distinct from the direct DNA-damaging drugs that have Top2 as their target . 203 K.W.Kohn Drugs Against Cancer CHAPTER 10 Etoposide became one of the most important anticancer drug and was often used in combination with cisplatin or cyclophosphamide; it was found to be particularly effective against small cell lung cancer and testicular cancer (Belani et al., 1994) (Meresse et al., 2004). The TDP story revisited: cutting off the fuzz a t topoisomerase-DNA break sites. As already explained, topoisomerase-blocking drugs impede the resealing of the normally transient DNA strand breaks that form during normal topoisomerase function. The trouble is that the topoisomerase protein remains persistently bound to the DNA, where its presence blocks repair machinery from coming to the rescue. The topoisomerase cannot dislodge from the DNA in the normal fashion, because the drug. bound to the same site, prevents it from doing so. The blocked topoisomerase becomes troublesome protein material stuck to the DNA Protein-digesting machinery was found to come into play to cut away much of the bound topoisomerase molecule but leave behind a DNA-bound protein fragment that it cannot access. The remaining fragment of topoisomerase protein is finally cut away by TD Pl or TDP2 (tyrosyl-DNA-phosphodiesterase 1 and 2), as explained in the legend of Figure 10.10. The enzymes that came to be known as TD Pl and TDP2 were first discovered in 1996 in yeast by Howard Nash and his coworkers at NIH (Yang et al., 1996) (Pouliot et al., 1999). The process, as conceived by Howard Nash and his colleagues is diagrammed in Figure 10.13 and explained in the Figure's legend as understood at the time. 204 K.W.Kohn Drugs Against Cancer CHAPTER 10 5' - -- - -- - - - - - 5' l c5Unfolding/ Protoolysis 5'-------- 5' ~ l Tyrosyl-DNA Phosphodiesterase 5' - - - - - ~ -- 5' l p Phosphatase + Kinase 5' =====-=-=-=-=-=-=..::;-;=== 5' l p DH DNA Ligase 5'- ------- 5' Figure 10.13. The essentials of the process by which a DNA-trapped topoisomerase is removed and the DNA repaired, as surmised by Howard Nash in 1996. The repair is accomplished by tyrosyl-DNA-phosphodiesterase enzymes (TD Pl or TDP2) that Nash discovered. These enzymes removed the topoisomerase fragment from the DNA, so as to allow the DNA strand break to reseal or become repaired (Yang et al., 1996). Shown at the top is a topoisomerase firmly bound to an end of a DNA strand break. A protein-digestion process then removed much of the topoisomerase protein but left behind a DNA-bound fragment that the protease could not reach. TD Pl or TDP2 then would come in in to finish the job of topoisomerase removal. So, what relevance would the TOP enzymes have for cancer therapy? On further study of the enzyme in yeast, Nash and his coworkers already in 1999 suspected that inhibition ofTDP might increase the effectiveness of topoisomerase-inhibiting anti-cancer drugs, because TDP would then not be available to cut away from the DNA break the potentially lethal protein fragment; persistence of the protein link to the DNA could kill the cell •· which would be good if it were a cancer cell that was killed (Pouliot et al., 1999). Therefore, much work was begun to discover TOP- inhibiting drugs that could be tried in cancer therapy together with topoisomerase inhibitors (Pommier et al., 2014). TD Pl was found to process trapped topoisomerase I, and TDP2 was found to process trapped topoisomerase II (Pommier et al., 2014). The cell, therefore, was normally able to repair both types of topoisomerases trapped by drugs targeted to each of them. Hence, there were therapeutic possibilities for combining a TD Pl or TDP2 inhibitory drug with a drug targeted against the respective topoisomerase. 205 K.W.Kohn Drugs Against Cancer CHAPTER 10 However, as usual, there were complication. TDPl could remove trapped topoisomerase I (topic of next chapter) in a camptothecin-treated cell, only if the trapped complex had not yet been encountered by a moving DNA replication machine. If a collision had already occurred, TD Pl was powerless to repair the mess, and a different, more complicated and more imperfect repair process was needed to fix the problem. It turned out that, in addition to cleaning off trapped topoisomerase complexes from the DNA, the TDP enzymes were able to clean off a variety of other anticancer drugs and toxin molecules that bound and became trapped at the end of a DNA strand break (Pommier et al., 2014). Summary and further comments. The discovery that topoisomerases were important targets of anticancer drug action came at a time when those DNA processing enzymes were scarcely known to exist outside of microorganisms. It was one of the few important discoveries to which I could lay claim in nearly 60 years of research on anticancer drug mechanisms. Looking back, I think it a rather remarkable story of a series of unanticipated experiment results that were in no way aimed to the final result It all began with my unanticipated discovery - related in the previous chapter -- that large DNA strands from cells dissolved on a filter eluted in a strand size-dependent manner when an alkaline solution was pumped slowly through the filter. By means of quantitative experiments, I was able to use that phenomenon to devise methods to measure the frequencies of DNA strand breaks and DNA-protein crosslinks in drug-treated cells. The next unanticipated result was that a drug, doxorubicin, that was known to produce DNA strand breaks failed to show any sign of strand breaks in our standard alkaline elution procedure. I thought the failure might be due to doxorubicin producing an excess frequency of DNA-protein crosslinks and that the protein would stick to the filter and prevent the broken DNA strands from eluting. That possibility was confirmed, because digesting away the proteins in the cell lysate beautifully revealed the expected elution of DNA strands. A puzzle remained however: measurements showed that the drug did not produce enough DNA-protein crosslinks to hide all of the DNA strand breaks produced. Moreover, incredibly, the frequency of DNA-protein crosslinks was equal (within experimental error) to the frequency of the DNA-strand breaks. That seemingly incredible equivalence was also true for two other DNA-intercalating drugs. Further quantitative considerations led me to conclude that the DNA-protein crosslinks were probably located at the sites of the breaks and, furthermore, that 206 K.W.Kohn Drugs Against Cancer CHAPTER 10 the DNA-linked protein might in fact be an enzyme that caused the break. That was the first evidence that certain anticancer drugs trap a topoisomerase on the DNA in a state where a DNA strand break exists. Consequent to that published finding, tremendous interest arose in finding out how various anticancer drugs trap topoisomerases on the DNA and in studying the consequent biological actions, eventually showing that those drug actions on topoisomerases were responsible for the toxic effect on cancer cells for the effectiveness of the drugs in cancer treatment Those studies were at first of topoisomerase II. The topoisomerase I story is related in the next chapter. I then wanted to see whether topoisomerase-targeted drugs had individual preference as to where in a DNA nucleotide sequence they most frequently become trapped and cause strand breaks. We found that each drug had its own preference for topoisomerase-trapping, depending on the base-pair at one end or the other of the break. I surmised that the drugs staked against a base-pair at the end of the DNA break, which was consistent with the drugs' capabilities of intercalation in the DNA helix. Each drug had its own preference for the type of base pair on one side or the other to which it preferred to stack against It was gratifying that our proposed model of drug-trapped topoisomerase II was eventually confirmed by crystallography. Repair of a persistent DNA strand break that has a drug-trapped topoisomerase bound to the break must first remove the topoisomerase protein from the DNA. Most of the protein was found to be digested away by a proteasome. But a remaining undigested protein fragment remained impervious to removal by proteasome. The enzymes TDPl and TDP2 then come in to play to complete the removal. The scope of their DNA cleaning abilities was later shown to be much broader in terms of the kinds of strand-break-linked chemical entities they could cut away. Therapeutic applications were contemplated where TOP inhibitors might enhance the potency of drugs that trap DNA at strand breaks created by those drugs. Much of my work on topoisomerases was carried out together with my colleague and friend, Yves Pommier (Figure 10.14), who, after my retirement as Chief of the Laboratory, went on to carry the studies further as Chief of a newly established Developmental Therapeutics Program at NCI. 207 K.W.Kohn Drugs Against Cancer CHAPTER 10 Figure 10.14. My colleague and friend, Yves Pommier, had a major role in our topoisomerase studies from a few years after he joined my laboratory as a Research Fellow in 1981 up to the present time as he continues, as Chief of the Developmental Therapeutics Branch of the National Cancer Institute, to investigate many aspects of those remarkable enzymes. Before coming to NCI, Yves had received MD and PhD degrees from the University of Paris. References Belani, C.P., Doyle, L.A., and Aisner, J. (1994). Etoposide: current status and future perspectives in the management of malignant neoplasms. Cancer chemotherapy and pharmacology 34 Suppl, Sl 18-126. Burden, D.A., Kingma, P.S., Froelich-Ammon, S.J., Bjornsti, M.A., Patchan, M.W., Thompson, R.B., and Osheroff, N. (1996). Topoisomerase 11.etoposide interactions direct the formation of drug-induced enzyme-DNA cleavage complexes. The Journal of biological chemistry 271, 29238-29244. Capranico, G., Kohn, K.W., and Pommier, Y. (1990a). Local sequence requirements for DNA cleavage by mammalian topoisomerase II in the presence of doxorubicin. Nucleic acids research 18, 6611-6619. Capranico, G., Zunino, F., Kohn, K. W., and Pommier, Y. (1990b). Sequence-selective topoisomerase II inhibition by anthracycline derivatives in SV40 DNA: relationship with DNA binding affinity and cytotoxicity. Biochemistry 29, 562- 569. Champoux, J.J. (1978a). Mechanism of the reaction catalyzed by the DNA untwisting enzyme: attachment of the enzyme to 3'-terminus of the nicked DNA. Journal of molecular biology 118, 441-446. Champoux, J.J. (1978b). Proteins that affect DNA conformation. Annual review of biochemistry 47, 449-479. Champoux, J.J. (2001). DNA topoisomerases: structure, function, and mechanism. Annual review of biochemistry 70, 369-413. Champoux, J.J., and Dulbecco, R. (1972). An activity from mammalian cells that untwists superhelical DNA--a possible swivel for DNA replication (polyoma- ethidium bromide-mouse-embryo cells-dye binding assay). Proceedings of the National Academy of Sciences of the United States of America 69, 143-146. 208 K.W.Kohn Drugs Against Cancer CHAPTER 10 Cristini, A., Park, J.H., Capranico, G., Legube, G., Favre, G., and Sordet, 0 . (2016). DNA- PK triggers histone ubiquitination and signaling in response to DNA double- strand breaks produced during the repair of transcription-blocking topoisomerase I lesions. Nucleic acids research 44, 1161-1178. Delbruck, M. (1954). On the Replication of Desoxyribonucleic Acid (DNA). Proceedings of the National Academy of Sciences of the United States of America 40, 783-788. Erickson, L.C., Bradley, M.O., and Kohn, K.W. (1977). Strand breaks in DNA from normal and transformed human cells treated with 1,3-bis(2-chloroethyl)-1- nitrosourea. Cancer research 37, 3744-3750. Filipski, J., Yin, J., and Kohn, K.W. (1983a). Reconstitution of intercalator-induced DNA scission by an active component from nuclear extracts. Biochimica et biophysica acta 741, 116-122. Filipski, J., Yin, J., and Kohn, K.W. (1983b). Topoisomerase I activities in L-1210 cell nuclei. Enzyme 30, 89-94. Fornace, A.J., Jr., Kohn, K.W., and Kann, H.E., Jr. (1976). DNA single-strand breaks during repair of UV damage in human fibroblasts and abnormalities of repair in xeroderma pigmentosum. Proceedings of the National Academy of Sciences of the United States of America 73, 39-43. Grieder, A., Maurer, R., and Stahelin, H. (197 4). Effect of an epipodophyllotoxin derivative (VP 16-213) on macromolecular synthesis and mitosis in mastocytoma cells in vitro. Cancer research 34, 1788-1793. Hsiang. Y.H., Lihou, M.G., and Liu, L.F. (1989). Arrest of replication forks by drug- stabilized topoisomerase I-DNA cleavable complexes as a mechanism of cell killing by camptothecin. Cancer research 49, 5077-5082. Keller-Juslen, C., Kuhn, M., Stahelin, H., and von Wartburg. A. (1971). Synthesis and antimitotic activity of glycosidic lignan derivatives related to podophyllotoxin. Journal of medicinal chemistry 14, 936-940. Kohn, K.W. (1996). DNA filter elution: a window on DNA damage in mammalian cells. BioEssays : news and reviews in molecular, cellular and developmental biology 18, 505-513. Kohn, K.W., and Ewig, R.A. (1979). DNA-protein crosslinking by trans- platinum(ll)diamminedichloride in mammalian cells, a new method of analysis. Biochimica et biophysica acta 562, 32-40. Kont, Y.S., Dutta, A., Mallisetty, A., Mathew, J., Minas, T., Kraus, C., Dhopeshwarkar, P., Kallakury, B., Mitra, S., Uren, A., et al. (2016). Depletion of tyrosyl DNA phosphodiesterase 2 activity enhances etoposide-mediated double-strand break formation and cell killing. DNA repair 43, 38-47. Liu, L.F., Liu, C.C., and Alberts, B.M. (1980). Type II DNA topoisomerases: enzymes that can unknot a topologically knotted DNA molecule via a reversible double- strand break Cell 19, 697-707. Liu, L.F., Rowe, T.C., Yang, L., Tewey, K.M., and Chen, G.L. (1983). Cleavage of DNA by mammalian DNA topoisomerase II. The Journal of biological chemistry 258, 15365-15370. Meresse, P., Dechaux, E., Monneret, C., and Bertounesque, E. (2004). Etoposide: discovery and medicinal chemistry. Curr Med Chem 11, 2443-2466. 209 K.W.Kohn Drugs Against Cancer CHAPTER 10 Miller, K.G., Liu, L.F., and Englund, P.T. (1981). A homogeneous type II DNA topoisomerase from HeLa cell nuclei. The Journal of biological chemistry 256, 9334-9339. Minford, J., Pommier, Y., Filipski, J., Kohn, K.W., Kerrigan, D., Mattern, M., Michaels, S., Schwartz, R., and Zwelling. L.A. (1986). Isolation of intercalator-dependent protein-linked DNA strand cleavage activity from cell nuclei and identification as topoisomerase II. Biochemistry 25, 9-16. Nelson, E.M., Tewey, K.M., and Liu, L.F. (1984). Mechanism of antitumor drug action: poisoning of mammalian DNA topoisomerase II on DNA by 4' -(9- acridinylamino)-methanesulfon-m-anisidide. Proceedings of the National Academy of Sciences of the United States of America 81, 1361-1365. Nitiss, J.L. (2009). Targeting DNA topoisomerase II in cancer chemotherapy. Nature reviews Cancer 9, 338-350. Pommier, Y. (2013). Drugging topoisomerases: lessons and challenges. ACS chemical biology 8, 82-95. Pommier, Y., Capranico, G., Orr, A., and Kohn, K. W. (1991). Local base sequence preferences for DNA cleavage by mammalian topoisomerase II in the presence of amsacrine or teniposide. Nucleic acids research 19, 5973-5980. Pommier, Y., Huang. S.Y., Gao, R., Das, B.B., Murai, J., and Marchand, C. (2014). Tyrosyl-DNA-phosphodiesterases (TDPl and TDP2). DNA repair 19, 114-129. Pommier, Y., Laco, G.S., Kohlhagen, G., Sayer, J.M., Kroth, H., and Jerina, D.M. (2000). Position-specific trapping oftopoisomerase I-DNA cleavage complexes by intercalated benzo[a]- pyrene diol epoxide adducts at the 6-amino group of adenine. Proceedings of the National Academy of Sciences of the United States of America 97, 10739-10744. Pommier, Y., Minford, J.K., Schwartz, R.E., Zwelling. L.A., and Kohn, K.W. (1985). Effects of the DNA intercalators 4'-(9-acridinylamino)methanesulfon-m- anisidide and 2-methyl-9-hydroxyellipticinium on topoisomerase II mediated DNA strand cleavage and strand passage. Biochemistry 24, 6410-6416. Pouliot, J.J., Yao, K.C., Robertson, C.A., and Nash, H.A. (1999). Yeast gene for a Tyr- DNA phosphodiesterase that repairs topoisomerase I complexes. Science 286, 552-555. Radding. C.M. (1978). Genetic recombination: strand transfer and mismatch repair. Annual review of biochemistry 47, 847-880. Ross, W., Rowe, T., Glisson, B., Yalowich, J., and Liu, L. (1984). Role oftopoisomerase II in mediating epipodophyllotoxin-induced DNA cleavage. Cancer research 44, 5857-5860. Ross, W.E., Glaubiger, D., and Kohn, K. W. (1979). Qualitative and quantitative aspects of intercalator-induced DNA strand breaks. Biochimica et biophysica acta 562, 41-50. Ross, W.E., Glaubiger, D.L, and Kohn, K.W. (1978). Protein-associated DNA breaks in cells treated with adriamycin or ellipticine. Biochimica et biophysica acta 519, 23-30. Sakasai, R., and lwabuchi, K. (2016). The distinctive cellular responses to DNA strand breaks caused by a DNA topoisomerase I poison in conjunction with DNA replication and RNA transcription. Genes & genetic systems 90, 187-194. 210 K.W. Kohn Drugs Against Cancer CHAPTER 10 Schellenberg, M.J., Perera, L., Strom, C.N., Waters, C.A., Monian, B., Appel, C.D., Vilas, C.K., Williams, J.G., Ramsden, D.A., and Williams, RS. (2016). Reversal of DNA damage induced Topoisomerase 2 DNA-protein crosslinks by Tdp2. Nucleic acids research 44, 3829-3844. Stahelin, H. (1970). 4'-Demethyl-epipodophyllotoxin thenylidene glucoside (VM 26), a podophyllum compound with a new mechanism of action. Eur J Cancer 6, 303-311. Tewey, K.M., Chen, G.L., Nelson, E.M., and Liu, L.F. (1984a). Intercalative antitumor drugs interfere with the breakage-reunion reaction of mammalian DNA topoisomerase II. The Journal of biological chemistry 259, 9182-9187. Tewey, K.M., Rowe, T.C., Yang, L., Halligan, B.D., and Liu, L.F. (1984b). Adriamycin- induced DNA damage mediated by mammalian DNA topoisomerase II. Science 226, 466-468. Vos, S.M., Tretter, E.M., Schmidt, B.H., and Berger, J.M. (2011). All tangled up: how cells direct, manage and exploit topoisomerase function. Nature reviews Molecular cell biology 12, 827-841. Wang, J.C. (1969). Variation of the average rotation angle of the DNA helix and the superhelical turns of covalently closed cyclic lambda DNA. Journal of molecular biology 43, 25-39. Wang, J.C. (1971). Interaction between DNA and an Escherichia coli p rotein omega. Journal of molecular biology 55, 523-533. Wu, C.C., Li, Y.C., Wang, Y.R., Li, T.K., and Chan, N.L. (2013). On the structural basis and design guidelines for type II topoisomerase-targeting anticancer drugs. Nucleic acids research 41 , 10630-10640. Yang, S.W., Burgin, AB., Jr., Huizenga, B.N., Robertson, C.A., Yao, K.C., and Nash, H.A. (1996). A eukaryotic enzyme that ca n disjoin dead-end covalent complexes between DNA and type I topoisomerases. Proceedings o f the National Academy of Sciences of the United States of America 93, 11534-11539. Zwelling, L.A., Michaels, S., Erickson, L.C., Ungerleider, R.S., Nichols, M., and Kohn, K.W. (1981). Protein-associated deoxyribonucleic acid strand breaks in L1210 cells treated with the deoxyribonucleic acid intercalating agents 4'-(9- acridinylamino) methanesulfon-m-anisidide and adriamycin. Biochemistry 20, 6553-6563. 211 K.W. Kohn Drugs Against Cancer CHAPTER 11 Qoplu JJ 7111f TopoltolllMWISrory-a: ' peori\«111.fr-o ~OltiNwnwnomtd Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@oib gov CHAPTER 11 The Topoisomerase I Story: camptothecin, from a Happy Chinese Tree. In 1960, the National Cancer Institute (NCI) began a search for anticancer substances in extracts from plants and animals (so called "natural products"). That effort was added to the ongoing testing of large numbers of organic chemicals for anticancer activity. The work was being carried out under the auspices of the NCl's Cancer Chemotherapy National Service Center (CCNSC). Every substance tested in this system received an NSC number to code for it in the database, which had information about chemical structure, origin, and test results in animals and cancer cell lines. Among the most important discoveries by the natural products effort were camptothecin and taxol, both of which were isolated from plant material by Monroe Wall and Mansukh Wani at the Research Triangle Institute (RTI) in North Carolina (Kohn and Pommier, 2000). Here, I tell the story of camptothecin; the story of taxol is in the next chapter. According to Wani, when he arrived at RTI in 1962, there was nothing there except 4 walls, and it was only when the 5th 'Wall' joined him at RTI that things started to move. Wall and Wani worked together in a life-long collaboration that yielded some of the most important advances in the history of cancer chemotherapy (Figure 11.1). Before coming to RTI, Monroe Wall had directed a program at the U. S. Department of Agriculture (USDA) in a search for plant materials that could be used as a starting point for the synthesis of cortisone, which was at the time in short supply. Extracts from those materials, in addition to being tested for substances useful for cortisone synthesis, were also sent to NCI for testing against cancer in tumor-bearing mice. 212 K.W.Kohn Drugs Against Cancer CHAPTER 11 And so it was that an extract from the Chinese tree Camptotheca accuminata (known in China as Xi Shu, meaning "Happy Tree") (Figure 11.2) was found to have powerful anti-cancer activity. Anticancer search however did not fit in USDA's mandate, and Wall's desire to find the anticancer substance in those extracts had to wait a few years until he moved to RTL In 1963, Wall and Mani (Figure 11.1) began their attempt to isolate the anticancer substance from the "happy tree" (Figure 11.2). They began with 20 kg of bark and wood and made extracts using various solvent, which they then tested for anticancer activity in mice. They tested the most active samples at each purification step. The work was slow and painstaking. But by 1966, they had pure camptothecin and had determined its structure by x-ray diffraction (Wall, 1966). Figure 11.1. Monroe E. Wall (right) and Mansukh C. Wani (left), discoverers of camptothecin and taxol. 213 K.W. Kohn Drugs Against Cancer CHAPTER 11 Figure 11.2. Camptotheca accuminata (Xi Shu, "Happy tree") in the Chengdu Botanical Garden - Chengdu, China. It is native near the warm humid stream banks in Southern China and Tibet (Public domain, Wikipedia commons.) Camptothecin specifically inhibits topoisomerase I. The first clue that camptothecin targets a topoisomerase-like enzyme was unknowingly obtained by Susan B. Horwitz in 1973 in an early observation at a time when topoisomerases had not yet been discovered (Horwitz and Horwitz, 1973) (Figure 11.3). That was years before the name "topoisomerase" was invented. She had exposed human cancer cells to camptothecin, a novel anticancer drug. and observed that the cell's DNA strands were broken by the drug. When the drug was removed, the DNA strand breaks quickly reversed. It seemed that the drug caused repairable or reversible DNA strand breaks. However, there was an additional observation that was so bizarre that it was not mentioned in her paper, perhaps because the paper might then not have been accepted for publication. About the same time, a similar finding in cultured cancer cells was independently reported by Ann Spataro and David Kessel (Spataro and Kessel, 1972). Also about the same time, Rajalakshmi and Sarma (Rajalakshmi and Sarma, 1973) reported that camptothecin broke DNA strands in the liver of treated rats and that the DNA was repaired surprisingly quickly. According to Dr. Silvio Parodi, who worked with D.S. Sarma under the supervision of Emmanuel Farber at Fels Research Institute in Philadelphia, they were looking at anti-neoplastic agents (especially of natural origin) for their potential carcinogenicity, testing for induction of chromosomal 214 K.W.Kohn Drugs Against Cancer CHAPTER 11 aberrations and sister chromatid exchanges when they observed the unusual DNA breakage and repair by camptothecin. A few months before publication of Susan's paper, I visited her laboratory, which was then led by Arthur Grollman at Albert Einstein Medical Center in The Bronx. I had been studying DNA strand breakage and repair by various anticancer drugs, and she therefore told me about her findings with camptothecin. I then asked how long it took for the strands to be repaired. After some hesitation, Arthur Grollman said that the repair was very fast, so fast, even in the cold, that they could not measure it I asked how that could possibly be. After further hesitation, Arthur said he didn't know, but that maybe there was an enzyme right there by the breaks that resealed them immediately when the drug was removed. That speculation seemed so bizarre that I could not accept it. However, it turned out that Grollman's speculation was right on the mark, and the responsible enzyme was later identified as the then unknown topoisomerase I. Susan Horwitz had observed a new anticancer drug- induced mechanism of DNA breakage and repair that was to have major impact on cancer chemotherapy. Sus.an Band Horwftz, Figure 11.3. Susan Band Horwitz (1937- ), working at Albert Einstein Medical Center, discovered that camptothecin produced rapidly reversible DNA breaks. She also discovered that anticancer drug Taxol blocked microtubules (Chapter 12). In view of the early evidence that camptothecin caused DNA breaks and that inhibitors oftopoisomerase II caused protein-linked DNA stand breaks (see Chapter 10), Leroy Liu and his colleagues tested camptothecin against topoisomerase II. They were surprised to find that there was no effect on topoisomerase 11, but found that camptothecin induced topoisomerase I to produce both DNA strand breaks and DNA-protein crosslinks (Hsiang et al., 1985; Hsiang and Liu, 1988). Using the DNA filter elution methods (Chapter 9), Joe Covey, Christine Jaxel, Yves Pommier and I confirmed that indeed camptothecin produced typical protein-linked DNA strand breaks (Covey et al., 1989). As Susan Horwitz and Arthur Grollman had surmised, a DNA-associated enzyme (later identified as topoisomerase 1) rapidly reversed the strand breaks; they would have been amazed to know at the time that their postulated reversal enzyme also produced the breaks in the first place. 215 K.W.Kohn Drugs Against Cancer CHAPTER 11 Topoisomerase I solves the problem of over- and under-twisted DNA during transcription and replication. Figure 11.4 illustrates one of the cell's major DNA topology problems. As the paired DNA strands separate during transcription or replication, the DNA twists are pushed ahead and would become bunched up to an extent that strand-separation could not continue. In the case of transcription, there is an additional problem behind the bubble of separated strands. When the transcribed RNA emerges (diagram Bin Figure 11.4), the complementary DNA strands re-associate, but there are not enough twists to make the stable one twist per 10 base-pairs (Pommier, 2013). The problem is solved by type I topoisomerases that transiently cleave one DNA strand and allow the strands to swivel and remove the excessive or deficient twists as the DNA or RNA synthesis machinery marches on. After swiveling removes stress on the DNA helix, the topoisomerase rapidly reseals the break (Figure 11.5). A. DNA replication I Increasing twist B. Transcription __+- '1/' _ __,/••--l ' \-...- - - - \ _____,/ I I I Increasing decreasing twist twist Figure 11.4. The DNA twisting problem in replication (DNA synthesis, A) and transcription (RNA synthesis, B). As the strands separate, the twists are pushed ahead and would impede further strand separation. The excessive or deficient twists are resolved by topoisomerase 1, which is bound to and moves along with the replication and transcription machineries. Parallel line pairs represent double-stranded helix. In A, the red lines represent newly synthesized DNA In B, the red line represents newly synthesized RNA. The strand- separation forks are moving from right to left (fat blue arrows). 216 K.W. Kohn Drugs Against Cancer CHAPTER 11 S' :x: 1111 ·-1·1·1·1·1.. ....._ 11111_ '!1111 .,.,.... 5' 11111 :>C illl 11111 f"'IIII S' ....._ -,r1r1r1r,~ :X:/l liil .,...... IIIIIV ... 1111 y Figure 11.5. The t opological problem solved by type 1 topoisomerases. The enzyme breaks one DNA strand, allows the other strand t o pass through or swivel around the break, and finally reseals the break. The red strand is broken and the blue strand passes through. As the enzyme breaks the strand, it grabs hold of one end via the enzyme's tyrosine (Y) at the active site (Pommier, 2013) (from /ACS Chemical Biology). How camptothecin causes DNA damage that kills cancer cells. The earliest clue about the actions of camptothecin on DNA came in 1972, when Herbert E. Kann and I obtained evidence that camptothecin reduces the average length of newly synthesized RNA chains, suggesting that camptothecin was prematurely terminating the growth of the RNA chains (Kann and Kohn, 1972). [Herb and I were close friends. He went on to continue both clinical and laboratory cancer research at Emory University in Atlanta but died tragically at an early age of liver failure.] We had used the then available ultracentrifugation method, but later, in 1990, when electrophoretic methods had been perfected and purified topoisomerase I was available, Ole Westergaard and his colleagues at the University of Aarhus in Denmark showed that camptothecin stops RNA synthesis and that the RNA chains were terminated when they reached 10 base-pairs from the sites where topoisomerase I had become bound to the DNA (Bendixen et al., 1990). Then, in 1994, we found that camptothecin killed cells when they were in S phase of the cell cycle: the drug selectively killed cells when they were replicating their DNA (Goldwasser et al., 1996). We surmised that cells were selectively blocked and killed in S phase because of collisions between moving replication forks and sites on the DNA where topoisomerase 1 was trapped by camptothecin (Figure 11.6). Our view was based on what was then understood about how topoisomerase I (Topl ) operates: it binds DNA in front of moving replication forks and cycles through opening and closing of a DNA strand break, so as to allow the strands to swivel and relieve the accumulating supercoiling of the DNA helix. When topoisomerase I is in the state where it has cleaved the DNA, camptothecin binds and retards the further action of the topoisomerase that would close the DNA break. A double-strand end would form that looks to the cell like a DNA double-strand break (Figure 11.6) (Kohn and Pommier, 2000). 217 K.W. Kohn Drugs Against Cancer CHAPTER 11 We presumed that a similar, but less lethal, process occurred during the growth of RNA chains which were terminated as was found in the earlier experiments mentioned above (Kann and Kohn, 1972). Collisions due to progress of RNA synthesis, merely generating prematurely terminated RNA chains, apparently produced less toxicity than the DNA double-strand ends produced by collisions during the progress of DNA replication. Ca1nploLl1ccin / ,,-------3· s· - - ----, 3' _ _ _ _ _...;::c_ _ s· ---3·s· ' Replication ~ complex ~ '" 5- 3' 5' - - - - - 3' - -----==--~ - - ----3· -------s· Figure 11.6. How an encounter between a moving replication process and a camptothecin- blocked topoisomerase I (Topl) complex generated a potentially lethal DNA double-strand end, as we envisioned it in 1994 (Pommier et al., 1994). Cells that were not in the process of replicating their DNA, however, w ere still somewhat sensitive to camptothecin, because of analogous encounters of trapped Topl by a transcription process (Bendixen et al., 1990). Structure of the camptothecin/ topotecan-Top1-DNA cleavage complex. Although camptothecin by itself was not noted to insert or intercalate between DNA base-pair, the structure of the Topl -DNA complex trapped in the strand-cleavage state by camptothecin or topotecan nonetheless revealed an intercalation-like structure (Figure 1 1.6.1) ( Topotecan is a variant of camptothecin; DNA intercalation was explained in Chapter 4.) As Figure 11.6.2 shows, however, the base-pair on one side of the drug is displaced and does not lie flat against the drug as would be required by strict intercalation. Thus, only the base-pair on one side of the drug is flat against it, and that base-pair is preferentially a guanine-cytosine (G:C) pair. That structure was relatively stable, which is why camptothecin induced Topl -DNA cleavage preferentially at sites similar to the structure shown in Figure 1 1.6.2. 218 K.W.Kohn Drugs Against Cancer CHAPTER 11 But as just mentioned, the camptothecin or topotecan drug lies flat against only one base-pair: The G:C pair shown in Figure 11.6.2; the drug is as if only semi- intercalated. In the absence of Top1, the drug did not show evidence of DNA intercalation. Therefore, the drug must be stabilized by addition interactions with the Topl-DNA cleavage complex. Indeed, the crystallographic structures showed hydrogen bonds from amino acids of the Topl to the distorted DNA structure, as well as to the drug (Figure 11.7). Especially notable were the three hydrogen bonds between amino acids ofTopl and the oxygen atoms in the E ring of camptothecin (Figure 11.7). That likely accounted for the E ring structure being critical for camptothecin function. Moreover, the hydroxyl group at position 20 of camptothecin is asymmetric and only one of the two possible conformations gave an active drug. It seemed that only one of the conformations points the hydroxyl group at position-20 in the direction required for hydrogen bonding with the Asp533 amino acid ofTopl (Figure 11.7). 5'SH guanine cytosine T7 18 ~ + 1C threonine 1• Tyrosa,e nudeotide at end ofdeal,edstrand Figure 11.6.1. View of topotecan (a variant of camptothecin) stacked against a G:C base-pair in a complex with DNA and Topl. The cleaved DNA strand is on the left, where tyrosine-723 ofTopl is bound to an end of the cleaved strand. The DNA helix is altered by the inserted topotecan and the Topl-bound end of the cleaved strand. The altered DNA conformation is stabilized by hydrogen bonds to several amino acids of To pl. This structure was based on x- ray crystallography (Staker et al., 2002). Displaced base-pair Topotecan DNA stand break end 219 K.W.Kohn Drugs Against Cancer CHAPTER 11 Figure 11.6.2. Topotecan is stacked against the base-pair on topotecan's left. The base-pair to its right is displaced. In this view, the DNA strand near the upper end of the topotecan is intact, whereas the strand near the lower end of the topotecan is cleaved (Staker et al., 2002), + 1 Guanine +1 Cytosine on Scissile Strand on Non-Cleaved Strand > 0 , N 1 5' 0 Asn722 ~ 0 + 1 Deoxyribose on Scissile Strand Figure 11.7. ln addition to the hydrogen bonds stabilizing the altered DNA conformation shown in the previous Figure, additional hydrogen bonds stabilize the position of camptothecin in the complex. Of particular importance are the three hydrogen bonds involving camptothecin's E ring (Redinbo et al., 1998). How did camptothecin/ topotecan kill cancer cells? Was the lethal effect of camptothecin due to inhibiting the relaxing of the DNA supercoils that accumulate in front of a replication process? Or was it due to the consequences of the collision itself: the production of DNA double-strand end as shown in Figure 11.4? To address this question, we tested a Topl mutant that had a change in an amino acid at a critical site for the enzyme's function. In the presence of camptothecin, this particular mutant Topl functioned as it should in relieving stressful DNA twists but did not produce DNA-Topl trapped complexes. Camptothecin did not kill cells bearing this Topl mutation. We inferred, therefore, that the cells were most likely killed as a consequence of a collision of a moving replication fork with a trapped Topl-DNA-camptothecin complex (Pommier et al., 220 K.W.Kohn Drugs Against Cancer CHAPTER 11 1999; Urasaki et al., 2001). The potentially lethal effect probably came from the difficulty posed by the DNA double-strand end that is produced (Figure 11.4). Thus, the very transient camptothecin-induced DNA breakage, originally observed by Susan Horwitz and Arthur Grollman, was found to be due to the drug's effect on Topl (Hsiang et al., 1985) (Covey et al., 1989). As was the case with topoisomerase II targeted drugs, DNA strand breaks and DNA-protein crosslinks were produced in equal numbers, consistent with one protein bound consistently to one end of each DNA strand break (Mattern et al., 1987). The covalent association ofTopl at each camptothecin-induced DNA break was confirmed by Hsiang and Liu (Hsiang and Liu, 1988). Porter and Champoux then obtained evidence that the trapping of the Topl- DNA breaks was due to reduction by camptothecin of the rate at which the breaks reseal (Porter and Champoux, 1989). These studies clarified the essentials of how camptothecin traps DNA-Topl. Later studies, however, disclosed that the formation of the disastrous DNA double- strand end shown in Figure 11.4 in cells treated with a topoisomerase inhibitor could be avoided if the drug concentration was not too high. When the growing end of a replicating DNA encountered a drug-induced block, the growing replication fork, instead of proceeding into the blocked region, could temporarily invert, as shown in Figure 11.8 (Ray Chaudhuri et al., 2012). Figure 11.9 shows an electron microscope image of an inverted replication fork The new understanding in the 1980's of how camptothecin works greatly revived interest in testing the drug on cancer patients; camptothecin and related topoisomerase I inhibitors have since assumed an important role in cancer chemotherapy. The reversal of the replication fork was mediated in part by poly(ADPR) polymerase (PARP) (Ray Chaudhuri et al., 2012) (see Chapter 30). Figure 11.8. Inversion ofa replication fork when replicating DNA (purple lines) encountered a block, such as produced by a topoisomerase inhibitor. If the drug concentration was not too high, the replicating strands could invert temporarily until the block spontaneously reversed (Ray Chaudhuri et al., 2012). 221 K.W.Kohn Drugs Against Cancer CHAPTER 11 Figure 11.9. Electron microscope image of an inverted replication fork (Ray Chaudhuri et al., 2012). Notice the 4 DNA double-helices emerging from the inversion point (arrow). Early clinical trials of camptothecin. As prelude to trials of camptothecin in cancer patients, testing in animals showed that a dose-limiting toxicity was damage to the lower intestinal tract. The cells of the inner lining of the intestines multiply rapidly in order to renew cells that normally are continually sluffed off. In cancer patients, however, the dose-limiting toxicity was suppression of blood cell production in the bone marrow (Gottlieb et al., 1970). Nevertheless, the rapidly dividing cells, both in the intestines and in the bone marrow, were particularly sensitive to camptothecin. One of the problems with the early clinical trials of camptothecin was that they used the sodium salt form (Figure 11.10, right), which is inactive and its conversion to the active lactone form (left) in patients was erratic. The camptothecin lactone was the active form, but it was nearly insoluble and therefore difficult to prepare for clinical use. (The solubility problem was later solved by encapsulating the insoluble camptothecin lactone in gelatin capsules for oral administration). However, the sodium salt was soluble and readily administered. It was therefore used in the early studies when its lack of activity was yet unknown (Muggia et al., 1996). The early clinical experience with camptothecin was discouraging, and the clinical trials were therefore stopped. Camptothecin studies were resumed 15 years later when its action on topoisomerase I was discovered. Development of camptothecin as an anticancer drug then resumed with renewed intensity, although the laps of 15 years was unfortunate for a drug that was to become very useful for anticancer therapy. 222 K.W.Kohn Drugs Against Cancer CHAPTER 11 12 IL._1 0 Oli -ow --H• CAMPTOTHEON {IACTONE) CAMPTOTHECIN {SODIUM SALT) Figure 11.10. Chemical structure of camptothecin. The active form of camptothecin has a "lactone" structure in the E ring (left). Under alkaline conditions, the lactone ring opens to form the sodium salt (right), which was inactive. Under mild acidic conditions, the sodium salt slowly converted to the active lactone form. Notable also was that the natural active form had its OH group at position 20 pointing up, whereas the isomer whose OH pointed down was inactive. Thus the 3-dimensional structure around position-20 had to be just right for camptothecin to bind to the Topl protein. Modified camptothecins. In 1989, we collaborated with Monroe Wall and Mansukh Wani in testing a large number of modified camptothecins for their activity against topoisomerase I Oaxel et al., 1989) (Kohn and Pommier, 2000). The results showed where the camptothecin molecule could be modified to increase its potency and indicated where modifications abolished activity. We found out where the camptothecin molecule must remain unobstructed in order to fit into its binding site on the topoisomerase I protein, and where atoms could be added without loss of activity. For example, adding an NH2 group at position 9 on the A ring increased activity, whereas adding an NH2 group at position 12 destroyed activity (Figure 11.10 left shows position numbering). Thus, position 12 had to remain unobstructed to allow camptothecin to fit well into its binding site on topoisomerase I. Positions 10 and 11 were free for making small additions. In fact, adding an OH group, especially at position 10, substantially increased camptothecin potency (Jaxel et al., 1989). Among the modified camptothecins we examined, one of the most potent had a methylenedioxy (-O-CH2-0-) group added to form a 5-membered ring next to the A ring (Figure 11.11) (O 'Connor et al., 1990; O'Connor et al., 1991). Although this compound was not pursued for development at that time, it was later rediscovered and called "FL118" (Ling et al., 2012; Ling et al., 2015). 223 K.W.Kohn Drugs Against Cancer CHAPTER 11 Figure 11.11. 10,11-methylenedioxycamptothecin, a modified camptothecin having increased potency for inhibition of topoisomerase 1 (Jaxel et al., 1989). The addition to the camptothecin molecule is circled red. Topotecan became the most frequently used camptothecin in cancer therapy. Our structure-activity findings (laxel et al., 1989) helped to design the modified camptothecin, "topotecan", which became commonly used in cancer treatment. Topotecan has a positively charged methylamino group added at position 9 and an OH group added at position 10 (Figure 11.12). The positively charged group solved the solubility problem; its placement at position 9 was in accord with our finding that additions could be made at this position without interfering with the ability of the drug to block topoisomerase I. We had also found that adding an OH group at position 10, which is the case for topotecan, would increase the potency of the drug. Topotecan was relatively easy to make by chemical modification of camptothecin, and it was highly potent against experimental tumors in animals, as well as effective against topoisomerase I in cancer cells (Kingsbury et al., 1991). A potential drawback was that the charged group at position 9 would reduce the ability of the drug to penetrate the blood-brain barrier into the brain. That would be a disadvantage if there were cancer cells in the brain. On the other hand, it might be an advantage if it reduced toxicity induced by the drug action on normal brain cells. 224 K.W. Kohn Drugs Against Cancer CHAPTER 11 HO 0 0 ....__/ OH 0 Figure 11.12. Topotecan, a modified camptothecin became used in cancer therapy. The N- containing group added at position 9 became positively charged, and therefore improved the solubility of the drug. so that it could readily be administered to patients. The OH group added at position 10 increased the potency of the drug. lrinotecan Another modified camptothecin, irinotecan, also became commonly used in therapy. It was approved by the U. S. Food and Drug Administration in 1996 for the treatment of colon cancer; it was also active against several other types of cancer. Irinotecan is a "pro-drug": it was nearly inactive until a carboxyesterase enzyme, present in liver and other tissues, cut off an inactivating side-chain from the molecule (Figure 11.13A) (Ramesh et al., 2010). When combined with other drugs, such as 5-fluorouracil and oxaliplatin, it became a key drug for the treatment of metastatic colorectal cancer, and it was also useful against several other types of cancer (Fujita et al., 2015). 225 K.W.Kohn Drugs Against Cancer CHAPTER 11 A lrinotecan 0 B HOOC 8 91~0 OH Figure 11.13. A. Chemical structure of irinotecan. The side chain on the oxygen at position 10 conferred water-solubility but inactivated the drug. The drug was activated by an enzyme present in tissues that cleaved of the bond indicated by the red arrow (Ramesh et al., 2010). That left an OH group on position10, which increased the potency of the drug (Jaxel et al., 1989). 8. Chemical structure after replacement of the group at position 10 by a glucose-like (glucuronide) unit through the action of a UGT enzyme. This action by the enzyme inactivated the drug (Ramesh etal., 2010). Absence of this enzyme allowed the amount of available active drug to increase to higher levels and thereby made patients who lack active UGT unusually sensitive to the drug. lrinotecan produced unusually severe toxicity in some patients. Extensive studies were carried out to find out why that was the case. If the unusually sensitive patients could be identified, their drug dose could be reduced to a safe level. The studies revealed that a frequent cause of the unusual sensitivity was a particular isoform of a gene called UGTlAl that sensitive patients had in their genome. This gene was found to code for an enzyme called UDP-glucuronosyl-transferase (UGT), whose function will be explained shortly. Among the several genes that code for UGT enzymes, the most troublesome form was UGTlAl *28. People who had only that isoform of the UGTlAl gene were highly sensitive to irinotecan. The reason for 226 K.W.Kohn Drugs Against Cancer CHAPTER 11 that was that the UGT enzyme made by that isoform was nearly inactive (Schulz et al., 2009) (Fujiwara and Minami, 2010). According to Dr. Silvio Parodi, UGT (UDP-glucuronosyltransferase) is a cytosolic glycosyltransferase that catalyzes the transfer of the glucuronic acid component of UDP-glucuronic acid to a small hydrophobic molecule. This was a glucuronidation reaction. The reaction catalyzed by the UGT enzyme involved the addition of a glucuronic acid moiety to a variety of biologically active compounds found in nature. To understand all that, we have to know what the active UGT enzyme does. After irinotecan has been activated by cutting off the side chain from position-10 (Figure 11.13A), UGT inactivates it again by adding a glucuronide unit (Figure 11.13B). Without active UGT, therefore, the level of active irinotecan was elevated to unusually high levels after the customary dose of the drug (Schulz et al., 2009) (Fujiwara and Minami, 2010). The solution to the irinotecan dosage problem therefore was to determine the UGT status of the patient and adjust the drug dosage accordingly. A remarkable modification of irinotecan, called etirinotecan pegol, was designed that reduced toxicity and increased anti-tumor potency in mice by slowly releasing the active topoisomerase I inhibitor over long periods of time (Figure 11.14). The structure was designed to link irinotecan to long poly(ethylene glycol) chains in a manner that kept the drug inactive and to slowly and spontaneously release it in its active form (Hoch et al., 2014). Etirinotecan pegol was more effective than the bare irinotecan in suppressing the growth of tumors in mice (Figure 11.15), and clinical trials of this promising designer drug were begun (Alemany, 2014) (Jameson et al., 2013; Lopez-Miranda and Cortes, 2016). 227 K.W.Kohn Drugs Against Cancer CHAPTER 11 lrinotecan Cleavable Linker 20 kDa, 4-arm PEG Figure 11.14. Molecular structure of etirinotecan pegol, in which irinotecan molecules were tethered to the ends of poly(ethylene glycol) chains. The linker hydrolysed slowly to release active irinotecan (Hoch et al., 2014). H460 Lung Carcinoma ~ Control -T- IRN 40 mg/kg -+- IRN 90 mg/kg 9 -v- EP 40 mg/kg +- EP 90 mg/kg 0 0 7 M 21 U ~ q q ~ Days Post Initial Treatment Figure 11.15. Increased effectiveness of etirinotecan pegol (EP) relative to bare irinotecan (!RN) against human lung cancer cells growing as tumors in immune-deficient mice (Hoch eta!., 2014). Vertical axis: tumor volume; horizontal axis: time after treatment (arrows show times of EP injection). EP inhibited tumor growth for a much longer time than did !RN. Similar results were reported with several cell lines from other types of human cancer. Another way to make irinotecan more effective was to incorporate the drug in tiny, microscopic-sized lipid globules called nanoliposomes, from which the drug was 228 K.W.Kohn Drugs Against Cancer CHAPTER 11 slowly released. In addition, the idea was that the nanoliposomes would be small enough to exit from the tumor's abnormal blood vessels while being large enough to be retained in normal blood vessels. That would selectively deliver the drug to the tumor and reduce toxic effects to normal tissues. Another factor would be that drug within the tumor tissue would only slowly be flushed out, because of the poor lymphatic drainage that was common in tumors. Nanoliposomal irinotecan ("nalirinotecan") has already been approved for clinical use (Ko, 2016). References Alemany, C. (2014). Etirinotecan pegol: development of a novel conjugated topoisomerase I inhibitor. Current oncology reports 16, 367. Bendixen, C., Thomsen, B., Alsner, J., and Westergaard, 0 . (1990). Camptothecin- stabilized topoisomerase I-DNA adducts cause premature termination of transcription. Biochemistry 29, 5613-5619. Covey, J.M., Jaxel, C., Kohn, K.W., and Pommier, Y. (1989). Protein-linked DNA strand breaks induced in mammalian cells by camptothecin, an inhibitor of topoisomerase I. Cancer research 49, 5016-5022. Fujita, K., Kubota, Y., Ishida, H., and Sasaki, Y. (2015). lrinotecan, a key chemotherapeutic drug for metastatic colorectal cancer. World journal of gastroenterology 21, 12234-12248. Fujiwara, Y., and Minami, H. (2010). An overview of the recent progress in irinotecan phannacogenetics. Phannacogenomics 11, 391-406. Goldwasser, F., Shimizu, T., Jackman, J., Hoki, Y., O'Connor, P.M., Kohn, K.W., and Pommier, Y. (1996). Correlations between Sand G2 arrest and the cytotoxicity of camptothecin in human colon carcinoma cells. Cancer research 56, 4430- 4437. Gottlieb, J.A., Guarino, A.M., Call, J.B., Oliverio, V.T., and Block, J.B. (1970). Preliminary phannacologic and clinical evaluation of camptothecin sodium (NSC-100880). Cancer chemotherapy reports Part l 54, 461-470. Hoch, U., Staschen, C.M., Johnson, R.K., and Eldon, M.A. (2014). Nonclinical phannacokinetics and activity of etirinotecan pegol (NKTR-102), a long-acting topoisomerase 1 inhibitor, in multiple cancer models. Cancer chemotherapy and pharmacology 74, 1125-1137. Horwitz, S.B., and Horwitz, M.S. (1973). Effects of camptothecin on the breakage and repair of DNA during the cell cycle. Cancer research 33, 2834-2836. Hsiang, Y.H., Hertzberg, R., Hecht, S., and Liu, L.F. (1985). Camptothecin induces protein-linked DNA breaks via mammalian DNA topoisomerase I. The Journal of biological chemistry 260, 14873-14878. Hsiang, Y.H., and Liu, LF. (1988). Identification of mammalian DNA topoisomerase I as an intracellular target of the anticancer drug camptothecin. Cancer research 48, 1722-1726. Jameson, G.S., Hamm, J.T., Weiss, G.J., Alemany, C., Anthony, S., Basche, M., Ramanathan, R.K., Borad, M.J., Tibes, R., Cohn, A., et al. (2013). A multicenter, 229 K.W.Kohn Drugs Against Cancer CHAPTER 11 phase I, dose-escalation study to assess the safety, tolerability, and pharmacokinetics of etirinotecan pegol in patients with refractory solid tumors. Clinical cancer research : an official journal of the American Association for Cancer Research 19, 268-278. Jaxel, C., Kohn, K.W., Wani, M.C., Wall, M.E., and Pommier, Y. (1989). Structure- activity study of the actions of camptothecin derivatives on mammalian topoisomerase I: evidence for a specific receptor site and a relation to antitumor activity. Cancer research 49, 1465-1469. Kann, H.E., Jr., and Kohn, K.W. (1972). Effects of deoxyribonucleic acid-reactive drugs on ribonucleic acid synthesis in leukemia L1210 cells. Molecular pharmacology 8, 551-560. Kingsbury, W.D., Boehm, J.C., Jakas, D.R., Holden, K.G., Hecht, S.M., Gallagher, G., Caranfa, M.J., McCabe, F.L., Faucette, L.F., Johnson, R.K., eta/. (1991). Synthesis of water-soluble (aminoalkyl)camptothecin analogues: inhibition of topoisomerase I and antitumor activity. Journal of medicinal chemistry 34, 98- 107. Ko, AH. (2016). Nano medicine developments in the treatment of metastatic pancreatic cancer: focus on nanoliposomal irinotecan. International journal of nanomedicine 11, 1225-1235. Kohn, K.W., and Pommier, Y. (2000). Molecular and biological determinants of the cytotoxic actions of camptothecins. Perspective for the development of new topoisomerase I inhibitors. Annals of the New York Academy of Sciences 922, 11-26. Ling, X., Cao, S., Cheng, Q., Keefe, J.T., Rustum, Y.M., and Li, F. (2012). A novel small molecule FL118 that selectively inhibits survivin, Mcl-1, XlAP and cIAP2 in a p53-independent manner, shows superior antitumor activity. PloS one 7, e45571. Ling, X., Liu, X., Zhong, K., Smith, N., Prey, J., and Li, F. (2015). FL118, a novel camptothecin analogue, overcomes irinotecan and topotecan resistance in human tumor xenograft models. American journal of translational research 7, 1765-1781. Lopez-Miranda, E., and Cortes, J. (2016). Etirinotecan pegol for the treatment of breast cancer. Expert opinion on pharmacotherapy 17, 727-734. Mattern, M.R., Mong, S.M., Bartus, H.F., Mirabelli, C.K., Crooke, S.T., and Johnson, R.K. (1987). Relationship between the intracellular effects of camptothecin and the inhibition of DNA topoisomerase I in cultured L1210 cells. Cancer research 47, 1793-1798. Muggia, F.M., Dimery, I., and Arbuck, S.G. (1996). Camptothecin and its analogs. An overview of their potential in cancer therapeutics. Annals of the New York Academy of Sciences 803, 213-223. O'Connor, P.M., Kerrigan, D., Bertrand, R., Kohn, K.W., and Pommier, Y. (1990). 10,11-Methylenedioxycamptothecin, a topoisomerase I inhibitor of increased potency: DNA damage and correlation to cytotoxicity in human colon carcinoma (HT-29) cells. Cancer Commun 2, 395-400. O'Connor, P.M., Nieves-Neira, W., Kerrigan, D., Bertrand, R., Goldman, J., Kohn, K.W., and Pommier, Y. (1991). S-phase population analysis does not correlate with 230 K.W.Kohn Drugs Against Cancer CHAPTER 11 the cytotoxicity of camptothecin and 10,11-methylenedioxycamptothecin in human colon carcinoma HT-29 cells. Cancer Commun 3, 233-240. Pommier, Y. (2013). Drugging topoisomerases: lessons and challenges. ACS chemical biology 8, 82-95. Pommier, Y., Leteurtre, F., Fesen, M.R., Fujimori, A., Bertrand, R., Solary, E., Kohlhagen, G., and Kohn, K.W. (1994). Cellular determinants of sensitivity and resistance to DNA topoisomerase inhibitors. Cancer investigation 12, 530-542. Pommier, Y., Pourquier, P., Urasaki, Y., Wu, J., and Laco, G.S. (1999). Topoisomerase I inhibitors: selectivity and cellular resistance. Drug Resist Updat2, 307-318. Porter, S.E., and Champoux, J.J. (1989). The basis for camptothecin enhancement of DNA breakage by eukaryotic topoisomerase I. Nucleic acids research 17, 8521- 8532. Rajalakshmi, S., and Sarma, D.S. (1973). Rapid repair of hepatic DNA damage induced by camptothecin in the intact rat. Biochemical and biophysical research communications 53, 1268-1272. Ramesh, M., Ahlawat, P., and Srinivas, N.R. (2010). Irinotecan and its active metabolite, SN-38: review of bioanalytical methods and recent update from clinical pharmacology perspectives. Biomedical chromatography: BMC 24, 104- 123. Ray Chaudhuri, A., Hashimoto, Y., Herrador, R., Neelsen, K.J., Fachinetti, D., Bermejo, R., Cocito, A., Costanzo, V., and Lopes, M. (2012). Topoisomerase I poisoning results in ?ARP-mediated replication fork reversal. Nat Struct Mol Biol 19, 417- 423. Redinbo, M.R., Stewart, L., Kuhn, P., Champoux, J.J., and Hol, W.G. (1998). Crystal structures of human topoisomerase I in covalent and noncovalent complexes with DNA. Science 279, 1504-1513. Schulz, C., Boeck, S., Heinemann, V., and Stemmler, H.J. (2009). UGTlAl genotyping: a predictor of irinotecan-associated side effects and drug efficacy? Anti-cancer drugs 20, 867-879. Spataro, A., and Kessel, D. (1972). Studies on camptothecin-induced degradation and apparent reaggregation of DNA from L1210 cells. Biochemical and biophysical research communications 48, 643-648. Staker, B.L., Hjerrild, K., Feese, M.D., Behnke, C.A., Burgin, A.B., Jr., and Stewart, L. (2002). The mechanism oftopoisomerase I poisoning by a camptothecin analog. Proceedings of the National Academy of Sciences of the United States of America 99, 15387-15392. Urasaki, Y., Laco, G.S., Pourquier, P., Takebayashi, Y., Kohlhagen, G., Gioffre, C., Zhang, H., Chatterjee, D., Pantazis, P., and Pommier, Y. (2001). Characterization of a novel topoisomerase I mutation from a camptothecin-resistant human prostate cancer cell line. Cancer research 61, 1964-1969. Wall, M.E., Wani M. C., Cook, C. E., Palmer, K. H., McPhail, A. T., Sim, G. A. (1966). Plant Antitumor Agents. I. The Isolation and Structure ofCamptothecin, a Novel Alkaloidal Leukemia and Tumor Inhibitor from Camptotheca acuminata. Journal of the American Chemical Society 88, 3888. 231 K. W. Kohn Drugs Against Cancer CHAPTER12 Chapt,r 11. 17w J.lirotic bthibi10r Stor,v- ,'inra- ltWJI 111009,QJ Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@oib gov CHAPTER 12 The Mitotic Inhibitor Story: taxol and vinca. This chapter is about anticancer drugs that were discovered as toxins in certain plants or sea creatures and that were found to block the microtubules that pull the chromosomes apart during mitosis. Microtubules also convey essential molecules down the axons of nerve cells, which is why these same drugs can damage nerve cells. Anti-cancer drugs from natural products The natural world of animals, plants, and microorganisms is full of biological warfare agents in conflicts between various species. Natural poisons serve to ward off predators and competitors. Some were used by people over the ages, both to poison and to cure. A few became useful as medicines for treatment of cancer (Cragg and Newman, 2004; Vindya et al., 2015). Since those medicines are also poisons, the amounts given to patients, as with most drugs used in cancer chemotherapy, must be carefully adjusted to give significant effect against the cancer without producing too much toxicity. So, how do those microtubule poisons work? During mitosis, the newly formed chromosome pairs are pulled apart by fibers, called microtubules. Each daughter cell then gets one of the newly formed chromosome pairs, although cancer cells often have abnormal mitosis that yields cells with abnormal sets of chromosomes. A major action of anti-microtubule drugs is to impair cell division at mitosis. As with most cancer chemotherapy, however, those microtubule-binding drugs are effective only against those cancers that are more sensitive to them than critical normal tissues. I will tell the stories of two classes of anti-microtubule drugs that became 232 K. W. Kohn Drugs Against Cancer CHAPTER12 important in anticancer therapy and of a class of more recently discovered mitotic inhibitors. Drugs derived from plants and animals in nature often have more complicated chemical structures than what chemists can easily synthesize in the laboratory. Although making those complicated compounds artificially in the laboratory can be challenging, living creatures have enzymes that put together surprisingly complicated structures from simple building blocks using enzymes that occur naturally in their cells. Moreover, evolution provided enormous opportunity for selection of compounds that would help a species to survive and beat off the competition. That is why useful medicines can be derived from nature that our chemists would be unlikely to discover on their own. These drugs however evolved as poisons, and so it is not surprising that they have toxic side-effects in patients. But why they have anticancer activity is not entirely clear. This chapter is about the discovery, mechanism of action, therapeutic opportunities and toxic limitations of some of those complicated drug structures Many anticancer drugs act by damaging DNA or blocking its synthesis, but the mitotic inhibitors do not act on DNA; rather, they interfere with the process of cell division itself. They impair the function of the mitotic spindle, which is made up of microtubules that normally assure that each daughter cell gets precisely one pair of the newly made chromosomes. When the mitotic spindle is perturbed by these drugs, the cell cannot divide normally and often dies. The first two classes of mitotic inhibitors to be discovered and developed for cancer treatment were the vinca alkaloids and the taxanes. I will now begin their stories. The story of Vinca and the periwinkle. The renowned Madagascar periwinkle (Vinca rosea, also known as Catharanthus roseus) is a colorful flowering plant (Figure 12.1) that has a venerable provenance beginning with the writings of Albartus Magnus (circa 1200-1280), a German Catholic Dominican friar and philosopher. Here is a 16th century English translation of what Albartus wrote about the periwinkle in The Boke ofSecretes ofAlbartus Magnus of the Vertues of Herbs, Stones and certaine Beastes: "Perwynke when it is beate unto pouder with worms ofye earth wrapped about it and with an herbe called houslyke, it induceth love between man and wyfe if it bee used in their meales ... if the sayde confection be put in the fyre it shall be turned anone unto blue coloure". (From Botanical.com, A Modern Herbal). In Chaucer, we read: 233 K. W. Kohn Drugs Against Cancer CHAPTER12 And fresshe pervinke, riche of hewe, And floures ye/owe, whyte, and rede; The periwinkle also has world-wide medicinal traditions: "In India, they treated wasp sting with the juice from the leaves. In Hawaii they prescribed an extract of the boiled plant to arrest bleeding. In Central America and parts of South America, they made a gargle to ease sore throat and chest ailments and laryngitis. In Cuba, Puerto Rico, Jamaica and other islands, an extract of the flower was commonly administered as an eyewash for the eyes of infants. In Africa, leaves are used for menorrhagia and rheumatism. Surinamese boil ten leaves and ten flowers together for diabetes. Bahamians take flower decoction for asthma and flatulence, and the entire plant for tuberculosis. In Mauritius, the leaves infusion is given for dyspepsia and indigestion. In Vietnam, it is taken for diabetes and malaria. Curacao and Bermuda natives take the plant for high blood pressure. Indochinese use the stalks and leaves for dysmenorrhea." (From J. A Duke, Handbook of Medicinal Herbs, 1985; Magic and Medicine of Plants, 1993; cited by the National Tropical Botanical Garden.) Obviously, the periwinkle had worldwide reputations for medicinal use for treatment of many ailments. Therefore, it most likely was doing something useful in the sick body - but what, exactly? Among all of those cited uses there is no mention of cancer! Vinca rosea Figure 12.1. The Madagascar periwinkle (Vinca rosea, also known as Catharanthus roseus), the source of vinblastine and vincristine. Road to Discovery:from the periwinkle to an anticancer drug. We transition now from ancient lore and tradition to scientific knowledge and medical application. The fascinating history of the discovery of anti-cancer ingredients in the Madagascar periwinkle was summarized in 1958 by Noble (Noble 234 K.W.Kohn Drugs Against Cancer CHAPTER12 et al., 1958) and in 1968 by Johnson (Johnson, 1968). As early as 1910, Theodore Pickolt, a naturalist and pharmacist, had already described the medicinal use of the periwinkle in Brazil (Johnson et al., 1960). However, the story leading to anticancer drugs from the periwinkle began in a surprising way in 1949, when J. H. Cutts at the University of Western Ontario in Canada learned that in the West Indies a tea made from the leaves was used as a remedy for diabetes (Noble et al., 1958). When he investigated this in diabetic rats, however, there was no trace of any effect of the tea on diabetes. Undeterred, Cutts tried administering a stronger dose by injecting it instead of just giving it to the rats to sip. To his surprise, the tea-injected rats died within a week! The rats were found to be dying of infection, which was in tum caused by marked loss of infection-fighting white blood cells (leucocytes). Depletion of leucocytes was evident not only in the blood stream, but also in the bone marrow where these blood cells are made (Figure 12.2). Cutts may have known that depletion of white blood cells occurred in mustard gas-exposed sailors and a few years earlier was associated with the discovery of the anti-cancer activity of nitrogen mustard (see Chapter 1). Hence, it made sense to purify the white-count-suppressing ingredients from the periwinkle extracts and to test them against cancer. Normal Treated Figure 12.2. Bone marrow from a normal rat (left) and a rat treated with the active material purified from Vinca rosea (right). The bone marrow from treated rats (right) shows many red blood cells but very few of the large developing white blood cells such as those seen in normal bone marrow on the left (Noble et al., 1958). To give an idea of how R L Noble, C. T. Beer, and J. H. Cutts at the University of Western Ontario, Canada, purified the active compound from extracts of the plant material in 1953, here is a brief summary of their procedure: The plant material was first soaked in a solvent, such as alcohol or ether. The material in the resulting solutions was then separated into different "fractions". The researchers then inj ected samples of each fraction from various stages of purification into rats and monitored white blood cell counts in blood drawn daily from the rat's tail. Inj ecting the original extract caused the rat's leukocyte count to drop after 2-3 days and then to recover about a week 235 K. W. Kohn Drugs Against Cancer CHAPTER12 later. In the first step of purification, an acid extract (ethanol with 10% acetic acid) of the dried leaves was made alkaline, whereupon something precipitated that was more active than the original extract; thus, the active ingredient had been partially purified. Further steps of purification eventually yielded a pure highly potent needle-like crystalline compound, which they named vincaleukoblastin, later shortened to vinblastine (Figure 12.3). The chemical structure of vinblastine, like that of many natural products, took a lot of work to unravel, because it is very complicated with many interlocking rings and asymmetric centers (Figure 12.4). Even at this early stage on the road to vinca alkaloids as important anticancer drugs, the investigators already had clues to how the drug work and the drug's limiting toxicities. In addition to the blood count suppression, there was damage to the cells that line the inner surface of the intestines. We now know that these drugs tend to kill dividing cells and that active cell division is required to maintain blood counts and to replenish cells that are continually being sloughed off from the intestinal lining (mucosa). Blood count suppression and damage to the intestinal mucosa are two major toxicities of many anticancer drugs. Thus, the vinca story highlights the fact that attack on dividing cancer cells also impacts normal dividing cells and particularly normal tissues that critically rely on continual production of new cells. The damage to dividing cells in normal tissues is the main cause of dose-limiting toxicity by most anticancer agents. (Gastrointestinal toxicity in the case of vinca alkaloids turned out not to be a major clinical problem, most likely because other toxicities supervened, and patients rarely received dosage high enough to cause troublesome gastrointestinal toxicity.) In addition to vinblastine, several other alkaloids of related chemical structure were isolated, of which the most important, vincristine, differed from vinblastine only by addition of an oxygen atom at an important location in the molecule (Figure 12.4). Even though the change in chemical structure was tiny, the two drugs differed in the cancers they were most effective against. Most striking was the greater effectiveness of vincristine against acute leukemia Oohnson et al., 1963). Also, there was surprisingly little cross-resistance between the two vinca drugs; thus, when patients stopped responding to one of the vinca drugs, they sometimes subsequently responded to the other. In the 1980's Pierre Potier and his colleagues at the CNRS in France prepared another anti-cancer vinca drug, vinorelbine. The periwinkle makes several chemically related compounds ("alkaloids"). They started with one of those (not the one from which vinblastine was derived) and chemically modified it to obtain the active drug. Compared to vinblastine, vinorelbine lacks an oxygen and a carbon atom (CH2 group), indicated by the blue arrows in Figure 12.4. So, we see that the three vinca drugs are chemically very similar, and they were found to inhibit microtubules in essentially the same way. Nevertheless, they differed in the cancers for which they were most useful. 236 K. W. Kohn Drugs Against Cancer CHAPTER12 Figure 12.3. Needle-like crystals ofvinblastine purified from the periwinkle Vinca rosea (Noble et al., 1958). H "' I / 0 .... . ',?' y 0 0 / ?~ N\ H ,,,;,, )ro\ Vinblastine and the changes in Vincristine and Vinorelbine Figure 12.4. The vinblastine molecule consisted of two parts: a "catharanthine nucleus" to the upper left of the red line and a "vindoline nucleus" to the lower right of the line. Vincristine differed only in having an oxygen atom added at the red arrow (thus vincristine had a formyl group instead of a methyl group attached to the nitrogen at that position). The dashed bond that connects the 2 parts of the molecule indicated that the upper left half of the molecule is above the plane of the page, while lower right half is below the plane; only the correct 3D geometry worked. Vinorelbine was like vinblastine, except that it lacked the OH to which a blue arrow points and has only one instead of two carbons connecting two rings indicated by another blue arrow. 237 K. W. Kohn Drugs Against Cancer CHAPTER12 How the vinca drugs block mitosis. A major clue to how the vinca's work soon emerged, when it was found that vinblastine blocked the cell division process. Moreover, the block was not at the DNA duplication stage, where most of the previously known anti-cancer drugs, such as methotrexate and 5-fluorouracil, block cells. Instead, the block was at mitosis, during which the chromosomes segregate into the two daughter cells. More precisely, the block was at the metaphase stage of mitosis, where the chromosomes are fully condensed and line up, ready to separate into their respective daughter cells (Figures 12.5 and 12.6) (Palmer et al., 1960) Oohnson et al., 1960) (Cutts, 1961). The block was due to the fact that the vinca drugs impaired the microtubules that make up the mitotic spindle, the structure that lines up the chromosomes and separates them equally into the two daughter cells. The vinca drugs were found to bind to the microtubules and thereby to block mitosis. Figure 12.5. A vinblastine-treated cell arrested in a spindle-deficient metaphase. The chromosomes are condensed as in normal mitosis, but the microtubules of the mitotic spindle is absent; therefore, the chromosomes are scattered instead of being lines up as they would be in a normal meta phase (Palmer et al., 1960). 238 K. W. Kohn Drugs Against Cancer CHAPTER12 Figure 12.6. This beautiful picture shows a critical phase of a normal mitosis ("anaphase" - immediately after metaphase ), when the microtubules of the mitotic spindle pull the chromosomes apart so as to give to each daughter cell one copy of each duplicated chromosome. The chromosomes are stained red; the microtubules, green. A vinca-treated cell lacks microtubules and therefore could not proceed from meta phase to anaphase and would not show in this picture. Although the idea that mitotic block was plausible as the cause of cell killing by vinblastine, it was mere conjecture, because the drug had other actions as well. But the conjecture was soon supported by direct evidence: the extent of cell killing by various doses correlated quantitatively with the extent of microtubule inhibition; therefore the two effects were correlated and their causes were probably related (Said and Tsimberidou, 2014; Tucker et al., 1977). Further investigation revealed the molecular details of how the vinca alkaloids block mitosis. During mitosis, the chromosomes normally divide equally between the two daughter cells. Each chromosome set is pulled into its corresponding daughter cell through the action of the mitotic spindle (Figure 12.6). The mitotic spindle is made up ofmicrotubules, which in turn are made up of tubulin molecules (Figure 12.7). That is where the vinca's attack: they bind to the tubulins and prevent them from assembling into microtubules. Instead of adding to the microtubules, the tubulins become bound by vincristine or vinblastine and aggregate into "paracrystalline" structures that were ofno use to the cell (Na and Timasheff, 1982) (Figures 15.8 and 15.9). 239 K.W.Kohn Drugs Against Cancer CHAPTER12 -- r AJph;:i t11bulln • -- l Vinca alkaloids flftl tubulin • Taxanes prevent prevent microtubu le microtubule assembly. disassembly. Figure 12.7. Microtubule are made up of alpha and beta tubulin units (blue and red). The Vinca alkaloids, vinblastine and vincristine, were found to prevent tubulins from assembling into microtubules, whereas taxanes were found to lock the tubulins in place, so that the microtubules could not function. Either way, the progress of mitosis was blocked. Microtubule function required that tubulin units be able to add at one end and be removed at the other end. The vincas made microtubules shrink and disappear, whereas the taxanes caused microtubules to accumulate in functionless bundles (Lobert et al., 1996). Vlnca alkaklld """ ~~ ~4..- t, e - ◄·· () () .\ ....~. (I . .......... ~~() f} Microtuoole fJ t) ~ Free Vlnca-bouric:I Paracrysi:aline lubu1i'I lu:bu1i1 dimers aggregates dlmetS Figure 12.8. How vincristine or vinblastine sequestered tubulin into useless paracrystalline aggregates. The end of a microtubule is depicted on the left. It consists of alpha- and beta-tubulin units (blue and green), which pair up. Vinca molecules (orange crescents) bind tubulin pairs (heterodimers) and assemble them into useless paracrystalline aggregates (right). (Source: Wikipedia.) 240 K. W. Kohn Drugs Against Cancer CHAPTER12 Figure 12.9. Paracrystalline bodies formed by vinblastine-bound tubulins (Na and Timasheff, 1982) Vinblastine is found to be effective against ly mphomas. In the early 19S0's, Cutts had already noted that vinca extracts and the purified vinblastine -- the first vinca alkaloid to be isolated and tested -- inhibited the production of blood cells in the bone marrow ofrats (Figure 12.2). He thought it might therefore work against leukemias, which are malignancies arising in the bone marrow. When given to leukemic mice, the drug indeed extended their lifespan (Cutts et al., 1960; Johnson et al., 1960). In view of the impressive activity of vinblastine in mice, clinical investigators at the Ontario Cancer Institute in Toronto, Canada administered the drug to patients with advanced stages of leukemia, lymphoma, or other malignant tumors (Warwick et al., 1960). At about the same time, a preliminary clinical trial of vinblastine was also carried out at the Indiana University Medical Center (Hodes et al., 1960). In both studies, a few patients had partial remissions, but the results, although encouraging, were insufficient for firm conclusions. The main toxicity, as expected, was suppression of white blood cells (leukopenia). Since that time vinblastine has not been very useful in cancer treatment But its sister drug, vincristine, became very useful indeed. Vincristine is the star. Vinblastine and vincristine differed in how effective they were against different malignant tumor, possibly due to differences in how much of the drug enters particular types of tissues (Zhou et al., 1990). The most striking difference however was the extraordinary effectiveness of vincristine against acute leukemia. Even though vincristine differed from vinblastine only in the addition of an oxygen atom to an important part of the molecule (Figure 12.4), it became much more useful in cancer treatment, especially for acute leukemia. The first clinical study of vincristine for the treatment of acute leukemia in children was carried out in 1962 241 K. W. Kohn Drugs Against Cancer CHAPTER12 at the National Cancer Institute by Myron R Karon, Emil J Freireich, and Emil ("Tom") Frei (Karon et al., 1962). Karon unfortunately died of a cerebral hemorrhage in 1974 at the age of 42 at the height of his career as a leading researcher and pediatric oncologist (Hersh et al., 1975). Frei and Freireich went on to lead the development of cancer chemotherapy and the cure of childhood leukemia. In the initial study, with Myron Karon, they escalated the vincristine dose slowly while closely monitoring the blood counts and bone marrow of their patients. When dangerous toxicity threatened, they lowered the dose ofvincristine, and transfused whole blood, platelet-rich plasma, or leukocytes as needed. Of 12 children with acute lymphocytic leukemia who were treated with vincristine in that first study, 8 had a complete remission. This experience was the first indication that vincristine would become a major part of the cure of acute leukemia (Said and Tsimberidou, 2014). Vincristine had a leading role in the cure of acute lymphoblastic leukemia in children as part of the VAMP combination: Vincristine +Amethopterin (methotrexate) + 6-Mercaptopurine (6MP) + Prednisone. The VAMP story is told by John Laszlo in his book about the cure of childhood leukemia (Laszlo, 1995). But vincristine damages the nervous system. Researchers were pleased that vincristine rarely produced serious toxicity to the blood-cell-forming bone marrow. However, they were not at all pleased that the amount of drug given to patients had to be limited to avoid damage to the nervous system (neurotoxicity) (Legha, 1986; Rosenthal and Kaufman, 1974). The first sign of neurotoxicity was numbness and tingling in the fingers and toes (Legha, 1986). The reason was simple. In addition to depleting the microtubules of the mitotic spindle, vincristine attacked the microtubules that fill the nerve cell's axon; the nerve cell needs those axonal microtubules to function and survive. The axons long enough to reach the tips of fingers and toes contain the longest microtubules and therefore were the most vulnerable to the drug. Vincristine caused the axons to degenerate and the nerve cells to die. How mig ht the neurotoxicity of vincristine be avoided? Much effort was devoted to trying to reduce vincristine's neurotoxicity. At early stages of treatment, the ill-effects on peripheral nerves were reversible if the drug was discontinued; therefore, the dosage had to be kept within those limits. Several substances were investigated for possibly reducing this toxicity. Of those, glutamic acid received the greatest attention, and early studies were encouraging Qackson et al., 1988). However, recent studies unfortunately failed to confirm that hope (Bradfield et al., 2015). 242 K. W. Kohn Drugs Against Cancer CHAPTER12 In recent attempts to reduce toxicity in general and neurotoxicity in particular, vincristine has been incorporated into sub-microscopic fatty globules called liposomes (Raj et al., 2013). Toxicities may be less for the liposomal-vincristine, but the extent to which it reduced toxicity in relation to therapeutic efficacy remained uncertain. Within acceptable dosage, however, vincristine maintained an important role in cancer therapy, particularly in the cure of acute leukemia of children. Danger of inadvertent injection of vincristine into the spinal fluid. Even though air travel has become very safe, a serious accident sometimes happens. Similarly, much care is required to eliminate serious medical mistakes due to human error. Since vincristine attacks the microtubules in nerve axons, one of the worst medical mistakes would be if the drug were accidentally injected into the spinal fluid. How could that happen? Here is what happened not so long ago in Thailand to a 12-year old girl with acute lymphoblastic leukemia who was receiving treatment that would probably have cured her and saved her life (Chotsampancharoen et al., 2015). On the day of the error, she was to receive an intravenous injection of vincristine and a spinal injection of methotrexate as an essential part of the leukemia cure: methotrexate kills any leukemic cells in the central nervous system and does not damage normal brain cells. Vincristine fortunately is kept out of the central nervous system by the blood-brain barrier. Thus, vincristine acts safely against the leukemic cells outside of the central nervous system. Methotrexate is injected directly into the spinal fluid in order to kill any leukemic cells that may lurk within the central nervous system. But it is disastrous if the two drugs were accidentally mixed up as to which was injected where - which is what happened in this tragic case. The injection kit that was provided for treatment of the leukemic child was provided with 2 syringes, each properly labeled for what it contained: vincristine or methotrexate. Somehow, the administering team mixed up the 2 syringes and used the vincristine syringe for the spinal injection. The team realized their mistake almost immediately and tried to flush the drug out of the spinal canal. Nevertheless, despite all efforts for supportive care, the child suffered badly and died 5 days later. This case, as well as previous cases, were published with suggestions for additional safeguards to avoid such errors (Chotsampancharoen et al., 2015; Gilbar, 2012; Gilbar, 2014). Taxol and the Pacific yew tree. To find new anti-cancer drugs from nature, the National Cancer Institute began in 1960 under the direction of Jonathan L. Hartwell an ambitious program to collect natural products and screen them for their ability to kill cancer cells. If something kills cancer cells, however, it doesn't mean that it necessarily has anti-cancer activity, because the substance might kill normal cells just as well. Some of the toxic 243 K. W. Kohn Drugs Against Cancer CHAPTER12 extracts of plants or animals were selected to test whether they prolonged the life of tumor-bearing mice, which would suggest that the material was killing the cancer without killing the animal. The road from there to a useful anti-cancer drug however could be long and tortuous, and is well illustrated by the story of taxol, or "paclitaxel" as it is now called. It is remarkable that three scientists who discovered taxol and initiated an understanding of how it worked had also contributed similarly to the discovery of another natural product, the major anticancer drug, camptothecin, which was the subject or the previous chapter. The three were organic chemists Monroe E. Wall and Mansukh C. Wani, and biochemist Susan Band Horwitz (Figures 11.1 and 11.3). Wani and Horwitz described how they purified the drug and determined its chemical structure and mechanism of action (Wani and Horwitz, 2014). Other details of the story, including its political aspects, were told by Jordan Goodman and Vivien Waisch in The Story ofTaxol: Nature and Politics in the Pursuit ofan Anti- Cancer Drug (Goodman and Walsh, 2001). Figure 12.10. Left: The Pacific yew tree (Taxus brevifolia), the source of paclitaxel (Taxol). Right: Peeling the bark (image from the National Cancer Institute. Public domain.) The Taxol story began in 1962, when Arthur S. Barklay, a botanist working for the U.S. Department of Agriculture (USDA) collected bark from a Pacific yew tree (Taxus brevifolia) in Washington State (Figure 12.10). The USDA had been commissioned by the National Cancer Institute to collect samples of plants from which extracts were to be prepared and tested for activity against a cancer cell line. Extracts of the yew bark indeed killed cancer cells in culture. 244 K. W. Kohn Drugs Against Cancer CHAPTER12 There was not enough material to test it adequately in tumor-bearing mice, so they resorted to testing their materials for action against cells in culture. A natural products chemist however might be inclined to go after a biologically active compound even without knowing whether it may turn out be useful. Monroe E. Wall and his colleagues at Research Triangle Park in North Carolina however hoped that an anti-cancer agent was lurking within the bark extracts, and they had the skills, patience and determination to go after it. In 1964 they started a huge effort to carry this out It required testing on cancer cells at all stages of a seemingly endless separation sequence. After 2-3 years of painstaking work, they had a purified material that prolonged the lives of cancer-bearing mice. They named the new drug "taxol," in honor of the genus name of the tree it came from, and by 1971 they had determined its chemical structure (Figure 12.11) (Wani et al., 1971). The drug was later renamed "paclitaxel" when Taxol became the brand name. A glance at the complicated chemical structure of the taxanes (Figure 12.11) gives an appreciation of the difficulty of solving the structure. Not only does the molecule have a great many atoms, but the atoms are arranged in a complicated way with interlocking rings and many asymmetric centers (indicated by the thick and hatched bonds). ;1/ ~ I 6H ~ o }- ~ -0 0 Paclitaxel (Taxol) A, microtubule binding site B, replaced by (CH 3),CO- in docetaxel C, removed in docetaxel Figure 12.11. Structure ofpaclitaxel (Taxol) and the changes in docetaxel (Taxotere ). The part of the molecule in the red circle (A), which is the same in paclitaxel and docetaxel, is where microtubules bind (de Weger et al., 2014). The blue circles show the parts of the paclitaxel molecule that were changed in going from paclitaxel to docetaxel. Removing the atoms in C left: a hydrogen attached to the oxygen; that H can dissociate, leaving a negative charge on the oxygen, which can be shared with the oxygen to its right. The resulting negative charge makes docetaxel more water-soluble than paclitaxel. 245 K. W. Kohn Drugs Against Cancer CHAPTER12 More studies of paclitaxel and a new taxane: docetaxel. Taxol (paclitaxel), like the vinca drugs, bound and blocked microtubules (Figure 12.7). Although the mechanism of the blockage was different, the net effect was to prevent microtubule function in the mitotic spindle and in the long axons of neurons. Taxol, like the vincas, was toxic to the central nervous system. Moreover, very little of the drug would dissolve to allow it to be injected. These problems spurred the search for new paclitaxel-like drugs that would have less toxicity and better solubility. A promising candidate was developed in 1981 from a compound extracted from another species of yew tree, the European yew Taxus piccata (de Weger et al., 2014). The extracted compound was itself inactive, but chemists had the insight to modify its structure in a way that conferred paclitaxel-like actions. The new drug, docetaxel (brand name Taxotere) had a chemical structure like paclitaxel with two simple modifications (Figure 12.11) that gave the drug better solubility and better clinical results: thus, it became mainstream in cancer treatment and research. Clinicians did not give up on paclitaxel, however, because when one of the taxanes didn't work, sometimes the other did. Also, the paclitaxel solubility problem was attacked in another way that turned out had other advantages. The drug was combined with albumin in aggregates of sub-microscopic clusters called nanoparticles (Henderson and Bhatia, 2007). Albumin is a blood protein having high solubility that also has the ability to bind many kinds of low-solubility molecules and distribute them through the blood stream. The new advantage was that these nanoparticles tended to leak out of blood vessels in tumors more easily than from blood vessels in normal tissues; therefore, the drug was somewhat selective for delivery into cancer tissues. Nab-paclitaxel, as the albumin conjugated form was called, was found to be clinically more effective than conventional paclitaxel (Henderson and Bhatia, 2007). Nab-paclitaxel became first-line treatment, for example, of metastatic pancreatic cancer. But it extended life expectancy by a mere 2 to 8 months and was more toxic (peripheral neurotoxicity and fatigue in about 17% of patients) relative to previously available treatment (Hoy, 2014). This disease was extremely deadly and difficult to treat, so any positive effect was seen as a clue to better therapy. How the taxanes produced their anticancer effects. Interest in taxol languished for several years because it was difficult to acquire enough bark material from which to isolate the drug. and because the drug's activity in leukemia test systems in mice was deemed mediocre (Wani and Horwitz, 2014). The chemical structure of taxol (Figure 12.11) was too complicated for chemists to 246 K.W.Kohn Drugs Against Cancer CHAPTER12 synthesize routinely in the laboratory. However, interest in the drug mounted when taxol was found to work unusually well against a mouse melanoma called B16. Susan Horwitz then made an important discovery. Working at the Albert Einstein College of Medicine in New York, she found that taxol blocked cells in mitosis by perturbing the function of the microtubules that make up the mitotic spindle; moreover, the way in which the drug affected microtubules was novel (Figure 12.7) (Schiff et al., 1979; Schiff and Horwitz, 1980). (For many years, we had a small Taxol tree, acquired I think by our lab chief David Rall, growing in the hall of our laboratory building - 5th floor of building 37 on the NIH campus in Bethesda. The tree was located in the Northwest corner of the hall that went all around the exterior next to the windows before the interior was rebuilt. That exterior hall had added cheer to our windowless labs and allowed a pleasant walk around as brief relief from long laboratory hours; some of us used to - gather to view the sunset and share our latest ideas. The redesign was in part fired by the misguided notion that the exterior hall was wasted space.) Microtubules are composed of two types of subunits: the protein molecules alpha and beta-tubulin, which associate in pairs in a manner that produces alternating alpha-beta pairs in intact microtubules (Figures 15.7). Taxol binds to a specific site on beta-tubulin in the intact microtubule (Figure 12.12). The microtubules of the mitotic spindle are a framework on which various motor proteins apply forces to move the chromosomes appropriately during mitosis. It is a complicated process in which tubulin molecules are added to one end of a microtubule and removed from the other end. That is how a microtubule grows and shrinks and moves to its proper place in the spindle. Essentially, the way taxol blocks microtubule function and the difference from the vincas is shown in Figure 12.7. But, anyone brave enough to read all the details -- including the role of GTP in microtubule function, which is not shown in the Figure -- could find them in a comprehensive review article by Walczak and Heald (Walczak and Heald, 2008). More details can also be found in the early articles by Susan Horwitz and coworkers (Schiff et al., 1979; Schiff and Horwitz, 1980) and in more recent review articles (Orr et al., 2003) and (Wani and Horwitz, 2014). 247 K. W. Kohn Drugs Against Cancer CHAPTER12 alpha lubulin beta tli>ui n Figure 12.12. Structure ofan alpha and beta tubulin pair (dimer) showing the backbones of those protein molecules. The place where Taxol binds, which is on the beta subunit, is shown in black. Red, alpha-tubulin; blue, beta-tubulin. In yellow, are GDP and GTP, which are small energy-bearing molecules that are essential for the structure and function of microtubules. Therapy with vinca and taxane drugs. Paclitaxel (as Taxol had by then been renamed) was becoming promising for treatment of several common cancers that were unresponsive or had developed resistance to other drugs. But two problems emerged that limited the dose levels that could be given to patients (Rowinsky and Donehower, 1995). First, was neurotoxicity: nerve damage, especially to the long nerves leading to the tips of fingers and toes. As already noted for vincristine, those long nerves are especially sensitive to anti-microtubule drugs, because long nerve axons contain long microtubules (Kudlowitz and Muggia, 2013) (Kudlowitz and Muggia, 2014). Those microtubules carry essential molecules from the neuron's cell body where they are made all the way to synapses at the end of the axon. The second problem was more specific to paclitaxel. Many patients developed hypersensitivity akin to the severe reactions that some people have to shellfish or beestings. The reactions to paclitaxel were sometimes even life-threatening. Here is how that problem arose: Paclitaxel was so insoluble that an injectable preparation could not be made directly. Therefore, the drug was mixed with an oily substance called cremaphore, which is a chemical modification of castor oil. That solved the solubility problem, but created another problem: the hypersensitivity reactions that 248 K. W. Kohn Drugs Against Cancer CHAPTER12 patients were experiencing turned out not to be due to paclitaxel itself, but to the cremaphore additive (Rowinsky and Donehower, 1995). Taxanes in combination chemotherapy. Cancer of the ovary had shown, albeit meager, responses to alkylating agents such as chlorambucil, and the addition of cisplatin was found improve the responses. In the mid-1990's, the standard of treatment for advanced ovarian cancer was the combination of cisplatin and the alkylating agent, cyclophosphamide. However, the outlook for the patients improved when paclitaxel replaced cyclophosphamide in the combination (McGuire et al., 1996) (Figure 12.13). There was however a caveat to this kind of clinical trial. The possibility was not excluded that increasing the dosage of cyclophosphamide a little or administering it on a different time schedule might have given results as good as paclitaxel in the cisplatin-combination. That was one reason that multiple trials under slightly different conditions would be needed for firm conclusions. Although early phase I trials that combined paclitaxel with topotecan suggested that this combination merited further investigation (Lilenbaum et al., 1995; O'Reilly et al., 1997), adding topotecan to a paclitaxel-carboplatin combination in treatment of phase Ill ovarian cancer gave no benefit (Bookman et al., 2009). (Carboplatin equaled cisplatin in effect against the cancer, but it was less toxic.) The standard of care of advanced ovarian cancer became surgery followed by treatment with a combination of paclitaxel and carboplatin. After that drug treatment had been established, clinical researchers searched for and determined the best time schedule by which to administer the drugs (Lee and Tan, 2018). That was one of the better results of a 2-drug combination with a taxane. Evidently, there was a long way to go in the treatment of advanced cancers. 249 K. W. Kohn Drugs Against Cancer CHAPTER12 O> C ·:, -~ :;) (/) 1.0 0.9 0.8 0.7 0.6 ...__ .. -------/ _z_.-~Cisplatin + cyclophosphamide 0.5 ~......... C 0 0.4 'i: 0.3 0 a. 0.2 0 ~ [L 0.1 0.0 0 6 12 18 24 30 36 42 48 Months after Entry into Study Figure 12.13. Paclitaxel + cisplatin was better than cyclophosphamide + cisplatin for treatment of ovarian cancer that had spread locally or metastasized to distant sites (stage III or IV) (from (McGuire et al., 1996), annotated). When paclitaxel replaced cyclophosphamide in the combination, the median survival time (the time when 50% of the patients were still living) went from 24 months to 38 months. Their life- expectancy went from 2 years to more than 3 years. In retrospect, much was accomplished in discovering and developing the taxanes. As with many other conventional anticancer drugs, the taxanes also attacked dividing normal cells. Thus, paclitaxel and docetaxel, either as single agent or in combination with other drugs, sometimes slowed the progression of some of the most common cancers. But the benefit was to prolong life usually only for a few months and only in a fraction of the patients. The tumors soon stopped responding to the drug by acquiring mutations that lowered the ability of the drugs to bind microtubules or to inhibit their function (Orr et al., 2003). Moreover, the quality of life during those few added months was often degraded by the toxicity and side effects of the treatment. Much was learned about the taxanes and how they affect cancer as well as normal cells. But the impact on the most common cancers was painfully limited. That might not be surprising. considering that these materials evolved as biological warfare poisons in nature. From a marine sponge, a new microtubule inhibitor, halichondrin B, and from it a synthetic derivative, eribulin. In the quest for new and better anti-cancer drugs, the National Cancer Institute began collecting invertebrate marine animals and tested extracts for ability to kill cancer cells (Vindya et al., 2015). They hoped that some of the toxins made by those creatures in their natural biological warfare might be useful against cancers. One of 250 K. W. Kohn Drugs Against Cancer CHAPTER12 the most promising came from a rare Japanese sponge called Halichondria okadai (Figure 12.14) (Hirata and lJemura, 1986; Swami et al., 2015). Extracts from this organism were extraordinarily potent in killing cancer cells in culture, and the active component became a new anticancer drug, halichondrin B. The isolation of this rare molecule and the determination of its complicated chemical structure were themselves a tour de force (Figure 12.15). But, coupled with the novel way its mechanism of action was unraveled makes this a truly remarkable achievement. The NCI team credited with this work is depicted in Figure 12.16. Chemists isolated the most active toxin, which they named halichondrin B, and determined its complicated chemical structure (Figure 12.15). The drug held promise, because it suppressed several human tumors transplanted into immune- deficient mice ("xenograft tumors") (Fodstad et al., 1996). Further progress was hampered however, because it was difficult to obtain enough material from that rare sponge, and the chemical structure was too complex to prepare routinely in the laboratory. Therefore, chemists prepared simpler structures by leaving out parts of the full Halichondrin B molecule, hoping to hit upon a compound that had the desired activity and that was feasible to synthesize in sufficient quantity. That effort yielded a promising new anti-cancer drug: eribulin (Figure 12.15) (Dybdal- Hargreaves et al., 2015) (Thara and Gitlitz, 2014). The chemists could be congratulated for having the insight that allowed them to select a small part of the halichondrin molecule that was active and that they could synthesize in the laboratory. Moreover, the new synthetic drug. eribulin, had better solubility than the parent halichondrin. Halichondria Figure 12.14. Halichondria, the type of marine sponge from which halichondrin B was extracted. (From Wikipedia.) 251 K.W.Kohn Drugs Against Cancer CHAPTER12 Meq Haliehonciin B eribuln mesylate Figure 12.15. Chemical structures of the natural product Halichondrin B (left) and its synthetic derivative Eribulin (right). The latter has a positive charge on the NH2 group (red circle), which is paired with the negatively charged mesylate (MsOH) ion . The additional charged group makes eribulin more soluble than halichondrin. The part of the halichondrin B molecule that is preserved in eribulin is shown in blue (Swami et al., 2015). (Permission needed.) Figure 12.16. The National Cancer lnstitute's halichondrin team in 1992. From left to right: Robert Shoemaker, Ernest Hamel, George Pettit, Kenneth Paull, Michael Boyd (Shoemaker, 2006). (All were NCI staff, except for George Pettit who was Professor of Chemistry at Arizona State University and worked under NCI contract) How Halichondrin B was found to be a microtubule inhibitor. That halichondrin was a mitotic inhibitor was first indicated by cell toxicity assays in the National Cancer lnstitute's 60 human cell lines (NC l-60) (see Chapter 20) (Figure 12.17). The pattern of toxicity (inhibition of cell growth and/or increase in cell killing) among the cell lines showed that halichondrin B had similar effects to 252 K. W. Kohn Drugs Against Cancer CHAPTER12 those of maytansine, a known microtubule inhibitor. This mechanism of action was confirmed by Earnest Hamel at the NCI, who showed that halichondrin B binds tubulin and inhibits its assembly into microtubules in a manner similar to vinca alkaloids (Bai et al., 1991 ). HUMAN TUMOR MAYTANSINE KAL.ICHOND!Wf 8 HOMOHALICHCINOAW 8 VM -26 •L -e=- CELL LINES TESTED Leuhrni. :.-- t I t• - = ~ :I Non-Smell Cell Lung ~ • ~ • Smll C.11 luno ,::- ,I' ~ I='" Cdon ~ ~ ~ ~ Cerrtrel Nervo~ System ~ f ~ -;;- -= } ~ " .........,. ~ ~ <N•i.-. ~ r t- -:i._ --- • ~ ..... -I } } ___,.b + + + OIFFEJIIEHTIAL CY'TOTOXICrrY lflet-.V. Otvleon ft'Om Mt.-i> Figure 12.17. Toxicity patterns (inhibitions of cell growth and/or increase in cell killing) in the National Cancer lnstitute's 60 cell line panel (NCl-60) (Chapter 20). The pattern for halichondrin (2 nd from the left) resembled that ofmaytansine (a known microtubule-targeted drug) (left-most) and differed from the pattern of VM26 (teniposide, a topoisomerase II targeted drug) (Bai et al., 1991). This was the first clue that halichondrin targets mitotic microtubules. This method of comparing anticancer cell activities was developed by NCl's Kenneth Paull. Other mitotic inhibitors: prosp ects and surprises Many other microtubule-targeted inhibitors were isolated from plants and animals or made in the laboratory by chemists who continued to modify their structures in hope of finding new useful drugs (Jiang et al., 2006; Jordan et al., 1998). Several have been tested in early clinical trials (phase 1 or 2), but none of them have, as of this writing, become part of our anti-cancer armamentarium. 253 K. W. Kohn Drugs Against Cancer CHAPTER12 Aside from microtubules, there are other essential mitotic spindle components that are being investigated as potential anti-cancer targets (Jiang et al., 2006). Although none have yet been demonstrated to have useful anti-cancer action, we can briefly state what those targets are inhibitors that target (1) kinesin motor proteins that move chromosomes during mitosis by attaching to and pulling on spindle microtubules, (2) aurora kinases that are required to initiate the formation of the mitotic spindle, and (3) polo-like kinases that are required to turn on the machinery that initiates mitosis. This gives an idea of the extensive terrain remaining for researchers to explore for new mitotic inhibitors. A small change converts a microtubule blocker into a topoisomerase blocker. Podophyllin has a long history of medicinal use by Native Americans -- it was used as a suicide agent by the Iroquois (Kelly and Hartwell, 1954). In recent times, it became the source of two very different types of drugs, although, remarkably, both types derived from the same chemical structure. First, podophyllin (podophyllotxin) was found to be a mitotic inhibitor like colchicine and vinblastine. A small change in chemical structure however yielded the important anti-cancer epipodophyllotoxins etoposide and teniposide, which work in an entirely different way: they target topoisomerase II (Chapter 10). Podophyllin resin is made from the mayapple, also known as the American mandrake. The active podophyllotoxin is in the plant's roots, leaves, and creeping underground rhizomes, which were used by Native Americans as an emetic and cathartic and to expel parasitic worms from the intestinal tract (Small and Catling, Canadian Medicinal Crops, NRC Research Press 1999, cited by Wikipedia). The first U.S. Pharmacopeia (1820) listed podophyllin as a cathartic but was dropped in the 12th revision (1942). Interest revived when it as found an effective dermatologic treatment of condyloma acuminata (genital warts now known to be caused by certain human papilloma viruses (HPV)) (Kelly and Hartwell, 1954; King and Sullivan, 1946). Investigating why podophyllin was effective against genital warts, King and Sullivan (King and Sullivan, 1946) applied it to the skin of rabbits and noted unusual changes in the nuclei of the skin cells. They thought that many of the cells were in a distorted state of mitosis, similar to that produced by colchicine. These simple observations, reported in a brief note in Science in 1946, led the researchers to a correct idea about what the drug does to cells. In a subsequent note in Science, Sullivan and Wechsler (Sullivan and Wechsler, 1947) looked at the effects of podophyllin in onion root tips, a convenient tissue for study of mitosis in the rapidly proliferating cells of the growing root They 254 K. W. Kohn Drugs Against Cancer CHAPTER12 confirmed the colchicine-like block of mitosis and thought podophyllin useful for cell division studies; they noted that "podophyllin is readily available at pharmaceutical supply houses and may be obtained at approximately 90 cents for four ounces." Podophyllin was tested in a variety of experimental systems, including tumors in mice, and clinically, especially in treatment of various skin conditions, but the main lasting clinical application has been for genital warts. The extensive history of podophyllin studies and trials was compiled by Margaret Kelly and Jonathan Hartwell in the NCl's former Laboratory of Chemical Pharmacology (Kelly and Hartwell, 1954). Colchicine revisited. Colchicine, a product of Colchicum automnale, a plant of the Lily family, has long been used in the treatment of gout, tracing back to the Byzantine physician Alexander of Tralles in 550 AD (Kumar et al., 2017). It has many actions in therapy of inflammatory diseases, largely based on its ability to inhibit the polymerization of tubulin by binding to a site at the interface between the alpha and beta subunits (Cheng et al., 2020). The colchicine binding site is distinct from the binding sites of taxol and the vincas. The drug has been reported to bind at a terminus of the microtubule, thereby preventing elongation at that end and leading to its depolymerization. The higher dose required for anti tumor activity, however, is close to the toxicity level: an unfavorable therapeutic ratio. The drug therefore has not been useful in cancer therapy. In view of its unique microtubule binding site, the colchicine molecule has been synthetically modified in attempts to create new anticancer drugs, but so far without clinical success. Mitotic inhibitors: overview The unusual sources and chemistry of the major mitotic inhibitor drugs may at first be puzzling. Unlike DNA damaging agents, they do not have highly reactive (covalent bond-forming) chemistries. Unlike the DNA synthesis inhibitors, they are not analogs of vitamins or molecules of the cell's normal biochemistry. They are not antibiotics such as are produced by microorganisms. Instead, they are complicated molecules that do not at all resemble any of the cell's normal molecules. Also, they derive almost exclusively from plants or animals, including marine invertebrates. They almost all come from creatures whose cells have a nucleus ("eukaryotes") that undergoes mitosis; in other words, mitotic inhibitors are made almost exclusively by organisms that engage in mitosis. They seem to be poisons used in the competition (biological warfare if you will) among eukaryotes in nature. 255 K. W. Kohn Drugs Against Cancer CHAPTER12 This chapter was about 3 classes of mitotic inhibitors: vinca alkaloids, taxanes, and halichondrins that have established roles in cancer chemotherapy. However, there are other classes of mitotic inhibitors, most of them from eukaryotic animals or plants Qordan et al., 1998). Several are or have been in clinical development; some have been discarded as ineffective or too toxic, but several remain promising and are still being studied. Mitotic inhibitors bind and disable the microtubules whose function is required for cell division. But microtubules also have other important functions in the cell, functions that do not involve cell division. Inhibition of mitosis however seems to be the major anti-cancer action of these drugs. We have seen that some of these drugs disrupt microtubules in the axons of nerve cells, resulting in sometimes severe neurotoxicity. 256 K.W.Kohn Drugs Against Cancer CHAPTER12 References Bai, R. L., Paull, K. D., Herald, C. L., Malspeis, L., Pettit, G. R., and Hamel, E. <1991 >· Halichondrin Band homohalichondrin B, marine natural products binding in the vinca domain of tubulin. Discovery of tubulin- based mechanism of action by analysis of differential cytotoxicity data. The Journal of biological chemistry 266, 15882- 15889. Bookman, M.A., Brady, M.F., McGuire, W. P., Harper, P. G., Alberts, D.S., Friedlander, M., Colombo, N., Fowler, J.M., Argenta, P.A., De Geest, K., eta/. <2009 ). Evaluation of new platinum-based treatment regimens in advanced- stage ovarian cancer: a Phase III Trial of the Gynecologic Cancer Intergroup. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 27, 1419- 1425. Bradfield, S. M., Sandler, E., Geller, T., Tamura, R.N ., and Krischer, J. P. <2015 >· Glutamic acid not beneficial for the prevention of vincristine neurotoxicity in children with cancer. Pediatric blood & cancer 62, 1004- 1010. Cheng, Z., Lu, X., and Feng, B. <2020 ). A review of research progress ofantitumor drugs based on tubulin targets. Transl Cancer Res 9, 4020-4027. Chotsampancharoen, T., Sripornsawan, P., and Wongchanchailert, M. <2015 >· Two Fatal Cases of Accidental lntrathecal Vincristine Administration: Learning from Death Events. Chemotherapy 61, 108- 110. Cragg, G. M., and Newman, D.J. c2004 ). A tale of two tumor targets: topoisomerase I and tubulin. The Wall and Wani contribution to cancer chemotherapy. Journal of natural products 67, 232- 244. Cutts, J. H. <1961 >· The effect ofvincaleukoblastine on dividing cells in vivo. Cancer research 21, 168-1 72. Cutts, J. H., Beer, C. T., and Noble, R.L. <1960 >· Biological properties of Vincaleukoblastine, an alkaloid in Vinca rosea Linn, with reference to its antitumor action. Cancer research 20, 1023- 1031. de Weger, V.A., Beijnen, J. H., and Schellens, J. H. <2014 ). Cellular and clinical pharmacology of the taxanes docetaxel and paclitaxel- a review. Anti- cancer drugs ZS, 488- 494. Dybdal- Hargreaves, N. F., Risinger, A. L., and Mooberry, S. L. <2015 >· Eribulin mesylate: mechanism of action of a unique microtubule- targeting agent. Clinical cancer research: an official journal of the American Association for Cancer Research 21, 2445- 2452. 257 K. W. Kohn Drugs Against Cancer CHAPTER12 Fodstad, 0 ., Breistol, K., Pettit, G. R., Shoemaker, R.H., and Boyd, M. R. <1996 ). Comparative antitumor activities of halichondrins and vinblastine against human tumor xenografts. Journal of experimental therapeutics & oncology 1, 119- 125. Gilbar, P. <2012 l· Inadvertent intrathecal administration ofvincristine: has anything changed? Journal of oncology pharmacy practice : official publication of the International Society of Oncology Pharmacy Practitioners 18, 155- 15 7. Gilbar, P.J. <2014 ), lntrathecal chemotherapy: potential for medication error. Cancer nursing 37, 299- 309. Goodman, J., and Walsh, V. <2001 ), The Story ofTaxol. Nature and Politics in the Pursuit ofan Anti- cancer Drug. (Cambridge, UK; New York, NY: Cambridge University Press l· Henderson, I. C., and Bhatia, V. <2007 ), Nab-paclitaxel for breast cancer: a new formulation with an improved safety profile and greater efficacy. Expert review of anticancer therapy 7, 919- 943. Hersh, E.M., Freireich, E.J., Frei, E., and Hammond, D. ( 1975 ). Myron r. karon-a tribute. Cancer research 35, 2220- 2221. Hirata, Y., and IJemura, D. <1986). Halichondrins- antitumor polyether macrolides from a marine sponge. Pure & Appl Chem 58, 701- 710. Hodes, M. E., Rohn, R.J., and Bond, W. H. <1960 ), Vincaleukoblastine. I. Preliminary clinical studies. Cancer research 20, 1041- 1049. Hoy, S. M. <2014 ), Albumin- bound paclitaxel: a review of its use for the first- line combination treatment of metastatic pancreatic cancer. Drugs 74, 175 7- 1 768. Jackson, D. V., Wells, H. B., Atkins, J. N., Zekan, P. J., White, D. R., Richards, F., 2nd, Cruz, J. M., and Muss, H.B. <1988), Amelioration ofvincristine neurotoxicity by glutamic acid. The American journal of medicine 84, 1016- 1022. Jiang, N., Wang.X., Yang. Y., and Dai, W. (2006 ). Advances in mitotic inhibitors for cancer treatment. Mini reviews in medicinal chemistry 6, 885- 895. Johnson, I. S. <1968 ), Historical background ofVinca alkaloid research and areas of future interest. Cancer chemotherapy reports Part 1 52, 455- 461. Johnson, I.S., Armstrong. J. G., Gorman, M., and Burnett, J. P., Jr. <1963). The Vinca Alkaloids: A New Class of Oncolytic Agents. Cancer research 23, 1390- 1427. Johnson, I.S., Wright, H. F., Svoboda, G. H., and Vian tis, J. <1960), Antitumor principles derived from Vinca rosea Linn. I. Vincaleukoblastine and leurosine. Cancer research 20, 1016- 1022. 258 K. W. Kohn Drugs Against Cancer CHAPTER12 Jordan, A., Hadfield, J.A., Lawrence, N.J., and McGown,A. T. c1998 ). Tubulin as a target for anticancer drugs: agents which interact with the mitotic spindle. Medicinal research reviews 18, 259- 296. Karon, M., Freireich, E.J., and Frei, E., 3rd c1962 >· A preliminary report on vincristine sulfate - a new active agent for the treatment of acute leukemia. Pediatrics 30, 791- 796. Kelly, M., and Hartwell, J. L. c1954). The biological effects and the chemical composition ofpodophyllin: a review. Journal of the National Cancer Institute 14, 967- 1010. King. L. S., and Sullivan, M. c1946). The similarity of the effect of podophyllin and colchicine and their use in the treatment of condylomata acuminata. Science 104, 244. Kudlowitz, D., and Muggia, F. c 2013 >· Defining risks of taxane neuropathy: insights from randomized clinical trials. Clinical cancer research : an official journal of the American Association for Cancer Research 19, 4570- 4577. Kudlowitz, D., and Muggia, F. c 2014 ). Clinical features of taxane neuropathy. Anti- cancer drugs 25, 495- 501. Kumar, A., Sharma, P.R., and Mondhe, D. M. c 2017). Potential anticancer role of colchicine- based derivatives: an overview. Anti- cancer drugs 28, 250- 262. Laszlo,J. c1995 ). The Cure of Childhood Leukemia: into the Age of Miracles. c New Brunswick, New Jersey, USA: Rutgers University Press ). Lee, M . X., and Tan, D.S. c2018 >· Weekly versus 3- weekly paclitaxel in combination with carboplatin in advanced ovarian cancer: which is the optimal adjuvant chemotherapy regimen? J Gynecol Oncol 29, e96. Legha, S.S. c1986 >· Vincristine neurotoxicity. Pathophysiology and management. Medical toxicology 1, 421- 427. Lilenbaum, R. C., Ratain, M.J., Miller, A.A., Hargis, J. B., Hollis, D.R., Rosner, G. L., OBrien, S. M., Brewster, L., Green, M. R., and Schilsky, R. L. c1995). Phase I study of paclitaxel and topotecan in patients with advanced tumors: a cancer and leukemia group B study. Journal of clinical oncology: official journal of the American Society of Clinical Oncology 13, 2230- 2237. Lobert, S., Vulevic, B., and Correia, J.J. c1996 >· Interaction ofvinca alkaloids with tubulin: a comparison ofvinblastine, vincristine, and vinorelbine. Biochemistry 35, 6806-6814. McGuire, W.P., Hoskins, W.J., Brady, M.F., Kucera, P.R., Partridge, E. E., Look, K. Y., Clarke- Pearson, D. L., and Davidson, M. c1996 >· Cyclophosphamide and cisplatin 259 K. W. Kohn Drugs Against Cancer CHAPTER12 compared with paclitaxel and cisplatin in patients with stage III and stage IV ovarian cancer. The New England journal of medicine 334, 1- 6. Na, G. C., and Timasheff, S.N. <1982 l· In vitro vinblastine- induced tubulin paracrystals. The Journal of biological chemistry 257, 10387- 10391. Noble, R. L., Beer, C. T., and Cutts, J. H. <1958 ), Role of chance observations in chemotherapy: Vinca rosea. Annals of the New York Academy of Sciences 76, 882- 894. OReilly, S,, Fleming, G, F., Barker, S.D., Walczak, J.R., Bookman, M.A., McGuire, W.P., 3rd, Schilder, R. J., Alvarez, R. D., Armstrong, D. K., Horowitz, I. R., et al. <1997 ), Phase I trial and pharmacologic trial of sequences of paclitaxel and topotecan in previously treated ovarian epithelial malignancies: a Gynecologic Oncology Group study. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 15, 177- 186. Orr, G.A., Verdier- Pinard, P., McDaid, H., and Horwitz, S.B. (2003 ), Mechanisms of Taxol resistance related to microtubules. Oncogene 22, 7280- 7295. Palmer, C. G., Livengood, D., Warren, A. K., Simpson, P.J., and Johnson, I.S. <1960 l· The action of the vincaleukolastine on mitosis in vitro. Experimental cell research 20, 198-201. Raj, T. A., Smith, A. M., and Moore, A.S. <2013 l· Vincristine sulfate liposomal injection for acute lymphoblastic leukemia. International journal of nanomedicine 8, 4361- 4369. Rosenthal, S., and Kaufman, S. <1974), Vincristine neurotoxicity. Annals of internal medicine 80, 733- 737. Rowinsky, E. K., and Donehower, R. C. <1995 ), Paclitaxel (taxol l· The New England journal of medicine 332, 1004- 1014. Said, R., and Tsimberidou, A.M. <2014 ), Pharmacokinetic evaluation ofvincristine for the treatment of lymphoid malignancies. Expert opinion on drug metabolism & toxicology 10, 483-494. Schiff, P. B., Fant, J., and Horwitz, S. B. <1979 l· Promotion of microtubule assembly in vitro by taxol. Nature 277, 665- 667. Schiff, P. B., and Horwitz, S.B. ( 1980). Taxol stabilizes microtubules in mouse fibroblast cells. Proceedings of the National Academy of Sciences of the United States of America 77, 1561- 1565. Shoemaker, R.H. <2006 ), The NCl60 human tumour cell line anticancer drug screen. Nature reviews Cancer 6, 813-823. 260 K. W. Kohn Drugs Against Cancer CHAPTER12 Sullivan, B.J., and Wechsler, H. I. <1947 l· The Cytological Effects of Podophyllin. Science 105, 433. Swami, U., Shah, U., and Goel, S. <2015 l· Eribulin in Cancer Treatment. Marine drugs 13, 5016-5058. Thara, E., and Gitlitz, B.J. <2014 ). Eribulin: a new- generation antimicrotubule agent in lung cancer therapy. Future Oncol 10, 1913- 1924. Tucker, R. W., Owellen, R. J., and Harris, S.B. <1977 ), Correlation of cytotoxicity and mitotic spindle dissolution by vinblastine in mammalian cells. Cancer research 37, 4346-4351. Vindya, N. G., Sharma, N., Yadav, M., and Ethiraj, K.R. <2015 ), Tubulins - the target for anticancer therapy. Current topics in medicinal chemistry 15, 73-82. Walczak, C. E., and Heald, R. <2008 ), Mechanisms of mitotic spindle assembly and function. International review of cytology 265, 111- 158. Wani, M. C., and Horwitz, S. B. (2014 ), Nature as a remarkable chemist: a personal story of the discovery and development ofTaxol. Anti-cancer drugs ZS, 482- 487. Wani, M. C., Taylor, H. L., Wall, M. E., Coggon, P., and McPhail, A. T. ( 1971 ), Plant antitumor agents. VI. The isolation and structure oftaxol, a novel antileukemic and antitumor agent from Taxus brevifolia. Journal of the American Chemical Society 93, 2325- 2327. Warwick, 0 , H., Darte, J.M., Brown, T.C., Beer, C, T., Cutts, J. H., and Noble, R.L. <1960 l· Some biological effects ofVincaleukoblastine, an alkaloid in Vinca rosea Linn in patients with malignant disease. Cancer research 20, 1032- 1040. Zhou, X. J., Martin, M., Placidi, M., Cano, J. P., and Rahmani, R. <1990). In vivo and in vitro pharmacokinetics and metabolism ofvincaalkaloids in rat. II. Vinblastine and vincristine. European journal of drug metabolism and pharmacokinetics 15, 323- 332. 261 K. W. Kohn Drugs Against Cancer CHAPTER!3 Chapt,er- l.1 1',~ Ble.omydn SU>ry: onandaJtteff'drug with a unlqu~ tmxkofoction ZZOl130od Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@oib gov CHAPTER 13 The Bleomycin Story: an anticancer drug with a unique mode of action. In 1957, Hamao Umezawa at the Institute for Microbial Chemistry in Tokyo was running a program to discover new anticancer drugs that microbes might produce. His group was testing samples for their ability to act against sarcoma tumors transplanted in mice. In 1962, they detected an active material produced by Streptomyces verticillus bacteria of the actinomyces family that had previously been a source of several antibiotics. The purified active material became the anticancer drug bleomycin, which actually is a mixture of chemical cousins with bleomycins A2 and B2 as the major components (Umezawa et al., 1967). Bleomycin became important in the cure of Hodgkins lymphoma and germ cell cancers of testis and ovary. As of 1967, what was known about bleomycin was that it inhibited the growth of a sarcoma tumor in mice and inhibited the synthesis of DNA, but not RNA. It went to the lung and kidney, but not liver or brain. Chemically, it was water-soluble and had a tightly bound copper atom (Umezawa et al., 1967). The surprising significance of the bound copper atom only came to light years later. Throughout the next decade, it became evident that bleomycin could break DNA strands, but that some kind of chemical activation was required. But the data on the chemicals that activated bleomycin were very confusing and contradictory. A breakthrough came at last around 1975, when Susan Horwitz and her colleagues at the Albert Einstein College of Medicine in The Bronx, New York discovered that as long as the copper atom remained bound to bleomycin, no DNA degradation occurred. But, when the copper atom was replaced by iron (ferrous ion), which could bind at the same bleomycin site as copper, then bleomycin became highly active in degrading DNA (Sausville et al., 1976). Another requirement was the presence of oxygen. The light then dawned as the so- 262 K. W. Kohn Drugs Against Cancer CHAPTER!3 called Fenton reaction came to mind, a reaction in which ferrous ion reacts with hydrogen peroxide to generate highly reactive hydroxyl radicals that were known to break up DNA strands. Indeed, the Horwitz group showed that bleomycin, ferrous ion, and oxygen bound together in a complex (Burger et al., 1979). Moreover, the action ofbleomycin on DNA was enhanced by the drug's ability to bind DNA by intercalation. Figure 13.1 shows bleomycin's chemical structure as Horwitz understood it in 1979. The structure was later modified slightly and confirmed by total synthesis as the reactive parts of the molecule became better defined (Figure13.2) (Hecht, 2000). A unique feature ofbleomycin's action on DNA was that it produced almost entirely double- strand breaks, rather than single-strand breaks, as shown by a beautiful experiment carried out in 1977 by Larry Povirk and his colleagues at Yale (Figure 13.3) (Povirk et al., 1977). Double-strand breaks are much more lethal than single-strand breaks, although the latter produce more mutations. Figure13.l. Chemical structure ofbleomycin A2 (later corrected slightly (Hecht, 2000)). Asterisks show possible metal binding sites proposed by Susan Horwitz and her coworkers (Burger et al., 1979). 263 K.W.Kohn Drugs Against Cancer CHAPTER!3 metal bmdmg domain H2N 0 NH 1 H • 2 HN~NH2 hnker domain N , N O ~H3 H H O /-0 ~ O O"N N ,....R H,N H HNH JI y"',,(' r .NH S~ I c' N N)...cH?Ho, cH~ N 5 ~ H DN A binding domain OH I ') HO~O NH HO ~o--.-OH ~OH pos,llvely charged tail (R) OH 0 ~ H2 Ao= - \ ~SMe,· carbohydrate domain NH,' •\~~JJ...NH2 Figure 13.2. Structure ofbleomycin, showing its functional parts (Yu et al., 2016). The positively charged tails that distinguish the 2 major forms ofbleomycin are shown in the box at the lower right These tails attach at the right end of the structure, where they are marked "R". .3~-----------------, "' -"' ., 0 ~ m .2- ' a, .I - :0 :, 0 r-roys 0 ' I 2 3 Single - strand Breaks Figure 13.3. Bleomycin produced almost entirely double-strand breaks rather than single-strand breaks in DNA. The number of double strand breaks produced by bleomycin (upper curve) was far greater than the number that would have been expected from randomly close pairs of opposed single-strand breaks (curve labeled "Theory"). In these experiments, small double-stranded circular DNA was treated with bleomycin. A double-stand break would convert the circular DNA to linear DNA, whereas a single-strand break would not do so. The fraction of the DNA molecules linearized were measured and made for a highly sensitive and quantitative assay (vertical axis) (Povirk et al., 1977). x-rays on the other hand produced mainly single-strand breaks. 264 K. W. Kohn Drugs Against Cancer CHAPTER!3 Effect ofpH on the action of bleomycin. My laboratory began studying bleomycin 1975, shortly after we had developed the alkaline filter elution method for quantitating DNA breaks in mammalian cells (Chapter 9). We didn't yet know about the role of ferrous iron - but that didn't bear directly on our studies ofbleomycin-induced DNA damage in intact cells. The elution method was more sensitive and quantifiable than previous methods and allowed us to measure bleomycin's effects at low dosage, where cells began to be killed. Indeed, the extent of cell killing was precisely related to the amount of DNA strand breakage (Figure 13.4) (Iqbal et al., 1976; Kohn and Ewig, 1976). Looking back on those experiments of more than 40 years ago, it seems that they might have unexpected current clinical bearing on the toxic effect ofbleomycin on the lung, as I will explain after relating the experimental problem we first had to solve. We had to overcome two vexing problems that had given widely different results from one experiment to another. These inconsistencies had given Zafar Iqbal, a post-doctoral fellow in our lab who was carrying out those experiments, much anxiety and consternation. Looking over the data from the many experiments he had carried out, we noticed that experiments carried on Tuesdays seemed to give different results from those carried on Thursdays. It turned out that the different results seemed to depend in part on whether the experiments were carried out in glass as opposed to plastic tubes. Further tests showed that bleomycin indeed had a marked tendency to stick to some surfaces and not to others. If much of the bleomycin stuck to the tube surface, it would not be available to enter the cells. When we were careful to avoid the surface-sticking effect, much of the variability between experiments disappeared. But not all of it! Some experiments still gave deviant results. Becoming aware of our consternation, our keen cell culture technician, Irene Clark, said that she thought the color of the medium in which the cells were growing was sometimes more yellowish at the time of the experiment than the usual red of fresh medium. The color indicated the pH of the medium. So ... when we carried out all experiments in fresh, nicely red medium and avoided the surface-sticking effect, all variability vanished and the results became quantitatively beautiful (Figures 13.5 and 13.6) (Iqbal et al., 1976; Kohn and Ewig, 1976). So, what about the color of the medium, you ask. Well, the color of the medium is due to a pH indicator added to help see the state or the culture: as cells grow, they make the medium more acidic, which changes the color from red to yellowish. Experiments carried out in yellowish medium would have been under more acidic (lower pH) conditions, suggesting that the toxic effect ofbleomycin might depend on pH. That turned out indeed to be the case. Moreover, both cell killing and DNA damage were greater when cells were treated under mildly alkaline conditions (pH 7.5) than when treatment was under slightly acid conditions (pH 6.8) (Figures 13.5 and 13.6) (Kohn and Ewig, 1976). But why is the action ofbleomycin dependent on pH? One possibility that we considered was that bleomycin's histidine became positively charged at the lower pH, as would be 265 K.W.Kohn Drugs Against Cancer CHAPTER!3 expected from the known chemistry of histidine. The additional charge could impair the ability ofbleomycin to enter cells. Alternatively, it might impair its metal-binding activation. M • Ul I "§ • e i 0.1 "• ~ l!s • i • • • 0.01,.0 u ~ OF DNA ... "'Vt••• •o... AFTIR 10Hlll l;UJ'f1(JN AT df 12.2 Figure 13.4. Killing of cells by bleomycin depended on the amount of DNA damage, which was higher when there was less DNA remaining on the filter. The cell killing (loss of the cells' ability to grow into colonies) had the same dependance on DNA damage regardless of pH or bleomycin concentration or duration of treatment Mouse leukemia cells in suspension culture were exposed to bleomycin for various amount of time in medium buffered at pH 6.7 (open symbols) or pH 7.5 (filled symbols) and bleomycin concentrations of 13 (triangles) or 50 µg/ml (circles) (Kohn and Ewig. 1976). 266 K. W. Kohn Drugs Against Cancer CHAPTER!3 5 µg/ml pH - Figure 13.5. At pH above 7.0, bleomycin became increasingly effective in killing cells. The graph shows the change in number of mouse leukemia cells after 20 hours of incubation in suspension culture at the indicated pH containing the indicated concentration ofbleomycin (Kohn and Ewig. 1976). 1.0 .. o.a u J! .E IIO .E 0.6 C ii E ... QI c( z - 0.4 Q 0 C 0 tl ... I! 0 2 4 6 HOURS AFTER ADDIT10N Of 133 loQ'ml BLEOMYCIN Figure 13.6. Bleomycin degraded the cells' DNA faster at pH7.5 (lower curve) than at pH6.7 (upper curve). Cells were treated in growth medium at the indicated pH. The vertical axis shows the fraction of the DNA remaining free of strand breaks at various times after addition of bleomycin to the medium (Kohn and Ewig. 1976). 267 K. W. Kohn Drugs Against Cancer CHAPTER!3 Bleomycin in cancer treatment. Choice of drug combinations for cancer chemotherapy aimed to match drugs whose toxicities affected different organs or tissues. Bleomycin was often chosen because it was not very toxic to the bone marrow or heart. Its toxicity was instead mainly to the lung. That led to bleomycin becoming an effective part of the treatment of Hodgkin lymphoma and of germ cell cancers of the testis and ovary. Hodgkin lymphoma accounted for about 10% of lymphomas and about 1-2% of all cancers (Cuceu et al., 2018). However, it often occurred in younger people, which made it important to have treatments that allow long-term survival. Bleomycin had an important role alone the road to effective treatment that eventually cured about 80% of patients, although its contribution was limited by its unusual toxicity to the lungs. Patients who at first appeared to be cured had a shorter survival time than would have been expected. Also, about 20% of patients did not respond to the drugs. The treatments had dramatic successes but failures as well. In his book "The Death of Cancer", Vincent DeVita tells the remarkable story of how Hodgkin lymphoma (originally called Hodgkin's disease), was cured by a combination of 4 anticancer drugs . Hodgkin lymphoma is a malignant tumor composed of particular kinds of lymphatic cells. It is a special kind of lymphoma. DeVita and his colleagues at the National Cancer Institute had chosen to attack this cancer because of its known growth characteristics that indicated that most or all of its malignant cells would be in the cell division cycle and therefore vulnerable to chemotherapy drugs. Moreover, Hodgkin lymphoma tumors had been seen to shrink in some patients treated by nitrogen mustard (nitrogen mustard is a topic in Chapter 1). Therefore, nitrogen mustard was one of the drugs DeVita and his colleagues chose in their 4-drug combination. They thought that a combination of 4 drugs would be needed to clear out all of the malignant cells, based Howard Skipper's quantitative findings in leukemia in mice. They combined drugs that had different modes of action and different toxicities, so as to attack the tumor from different angles and minimizing the net toxicity. They called their combination, MOPP, for nitrogen mustard (M for mustard), vincristine (0 for its other name, oncovine, see Chapter 12), procarbazine (another kind of alkylating agent, see Chapter 2) and prednisone. Each of these drugs were known to kill or inhibit malignant white blood cells and had a different mechanism of action. Their attack on Hodgkin lymphoma was similar to the 4-drugs combination, VAMP, that had successfully cured childhood leukemia. That story was told by John Laszlo in his book "The Cure of Childhood Leukemia". However, there were two crucial differences that had to be considered when it came to choosing a drug combination to combat Hodgkin lymphoma. First, leukemia cells populate the bone marrow, and to get cures the bone marrow had be cleared out of all cells. Hodgkin lymphoma, on the other hand, consists of solid tumors that 268 K. W. Kohn Drugs Against Cancer CHAPTER!3 usually do not harm the bone marrow. Therefore, contrary to drugs for leukemia, drug choice for Hodgkin lymphoma did not try to eliminate the blood cells in the bone marrow. Second, although most leukemia cells undergo cell division within a couple of weeks, the malignant cells in Hodgkin lymphoma often wait for a longer time before they divide. Therefore, the drugs that kill the malignant cells during their cell division period had to be administered to patients over a longer period of time than was the case with leukemia. The 4-drugs combination, MOPP, had been successful in treatment of Hodgkin lymphoma for several years, when bleomycin came along, featuring low bone marrow toxicity. In 1975, Gianni Bonadonna at the Instituto Nationale Tumori in Milan, Italy used bleomycin to replace the DNA-damaging nitrogen mustard in new drug combinations. The combination ofbleomycin, doxorubicin, vinblastine, and dacarbazine (ABVD), was eventually found to be at least as effective as MOPP or its variants, and became the standard treatment for Hodgkin lymphoma (Canellos et al., 2014). Initially, the Bonadonna group continued to use the MOPP combination, but added the new bleomycin and doxorubicin-containing combination (ABVD). Their new treatment of Hodgkin lymphoma alternated MOPP and ABVD. MOPP alone cured about 45% of the patients, whereas the MOPP/ ABVD alternating treatment increased the cure rate to about 73% (Bonadonna et al., 1986) (Figure 13.7). It was later found that treatment with ABVD by itself was at least as effective, and gave less toxicity (Canellos et al., 2014). 269 K. W. Kohn Drugs Against Cancer CHAPTER!3 FREEDOM FROM PROC.RESSIC>N ', \ '',, ,....._ - -• • •• , MOPP/ABVD '----------- M/A 50 p < 0 005 . __ _ _ _ __ MOPP only 3 Figure 13.7. Survival of patients with advanced Hodgkin lymphoma was increased by adding a bleomycin- and doxorubicin-containing drug combination (ABVD) to the MOPP treatment (dashed curve). The curves show the fraction of patients free of progression or relapse of the tumor as a function of the number of years after treatment. Once patients had survived for 5 years, they were cured: the curves flattened out after 5 years, and there was no further progression of the tumor for at least up to 8 years. Patients who were treated with the then standard treatment with the MOPP drug combination (lower curve) had a 45% chance to cure. However, when the bleomycin- doxorubin-vinblastine-dacarbazine combination (ABVD) was added in alternation with MOPP, the cure rate increased to 73% (upper curve) (Bonadonna et al., 1986). Later, it was found that ABVD by itself was as effective and less toxic than the alternating MOPP / ABVD treatments. Hodgkin lymphoma contains unusual giant cells, called Reed-Sternberg cells (Figure 13.8). These cells are giants, much larger than other cells in the tumor; they often have more than one nucleus and the nuclei have prominent nucleoli where ribosomal RNA is made. Although there are relatively few of those cells in the tumor, they seemed to be the malignant cells that divided to produce the other cells that make up the bulk of the tumor. It became possible to identify those truly malignant cells in Hodgkin lymphoma tissues by means of antibodies that bind to specific "marker" proteins, called CD15 and CD30, on the surface of those cells (Cuceu et al., 2018). The Reed-Sternberg cells seemed to arise by cell fusion from B-lymphocytes. Their chromosomes were particularly unstable, in that the number of chromosomes in each cell tended to vary and there often were recombinations between different chromosomes. Most of the cells in a Hodgkin lymphoma tumor were killed by anticancer drugs. But the Reed-Sternberg cells in the tumor were harder to kill and grew new tumors after drugs had shrunken much of the tumor bulk Thus, the tumors would seem to have been almost eliminated, but would soon grow again. The treatment therefore had to be strong enough (despite greater toxicity) to eliminate those truly malignant cells, which are like "stem cells" that gave rise to the bulk of the tumors. 270 K. W. Kohn Drugs Against Cancer CHAPTER!3 Figure 13.8. A Reed-Sternberg cell in a Hodgkin lymphoma tumor (arrow) (from (Cuceu et al., 2018) (permission needed to use their figure 18). The cell looks pink, because of staining for the Li- Fraumeni protein, p53, which is often prominent in these cells. Bleomycin contributed also to the effective treatment of germ cell cancers of testis and ovaries as well as of teratomas that occur usually in those organs and of the formerly deadly choriocarcinomas that form in the placenta during pregnancy. Germ cells in humans are the sex cells that develop into sperm and ova. In 1974, a group at Indiana University cured advanced testicular cancer in 57% of patients with bleomyin in combination with cisplatin and vinblastine. Investigators eventually found an optimum dosage and treatment schedule that cured nearly all of the patients (Einhorn, 1981). Although rare, testicular cancer was the most common cancer in the 15 to 35 age group, and it was one of the first cancers at an advanced stage that could usually be cured by chemotherapy. (Choriocarcinoma was the first cancer to be cured by chemotherapy (Chapter 5)). Lung toxicity The bleomycin-containing ABVD drug combination cured 80% of Hodgkin lymphoma patients. A major problem however was bleomycin-induced lung toxicity, which was an inflammation of the bronchial tree that sometimes led to irreversible fibrosis and permanently reduced lung function (Figures 13.9 and 13.10). The huge amount of fibrosis produced in the lung by bleomycin can be appreciated by comparing with normal lung in Figure 13.11. The lung toxicity ofbleomycin was of particular concern, because many of the cured patients were relatively young and could have breathing problems for the rest of their lives. Despite much effort to find ways to counter bleomycin's lung toxicity, the most reliable way to avoid the toxicity was to reduce the dose of the drug, which however reduced its therapeutic effect. Early detection ofbleomycin's lung toxicity was important, so that the dose could be reduced promptly. But the lung inflammation caused by bleomycin was difficult to distinguish from a bacterial lung infection. An early clinical trial 271 K.W.Kohn Drugs Against Cancer CHAPTER!3 already indicated that, ifbleomycin lung toxicity could be avoided, 90% of Hodgkin lymphoma patients treated with ABVD would be cured (Figure 13.12) (Martin et al., 2005). The main lung abnormalities caused by bleomycin was damage to the cells of the inner linings of the bronchi, alveoli, and blood vessels (Sleijfer, 2001). These cells would be at the interface of the transfer of carbon dioxide to the expired air, and consequently might undergo a pH change that could make the cells more sensitive to bleomycin, as suggested above. Since bleomycin is activated by oxygen, the freshly oxygenated parts of the lung may be particularly sensitive to damage by the drug. Figure 13.9. Lung biopsy tissue sample from the lung of a patient with bleomycin-induced toxicity, showing large amounts of inflammatory cells and fibrosis between the alveoli that greatly impaired the alveolar air spaces from opening (Reinert et al., 2013) (permission need to use their figure 1). Figure 13.10. Lung tissue of a 20-year-old man who was cured of metastatic testicular cancer by a combination ofbleomycin, etoposide, and cisplatin. However, he developed severe lung toxicity that required transplantation of both of his lungs, which was successful. The walls of the alveoli of the removed lung are thickened or completely overgrown by inflammatory cells, especially in the upper tight of the slide. The overgrown cells and fibrosis prevented opening of most of the air spaces of the alveoli. This was the first reported case oflung transplantation for bleomycin toxicity (Narayan eta!., 2017) (permission need to use their figure 28). (Permission needed.) 272 K.W.Kohn Drugs Against Cancer CHAPTER!3 Figure 13.11. Normal lung for comparison with the previous two Figures. Note the thin epithelial linings of the air-filled alveoli. (Freely available from Wikipedia.) ~ 1.0 0.8 "' Patients without lung toxicity l 0.6 Patients with lung toxicity l 1 :> 1/) 0.4 S·Year Overall Survival L>No BPT (n = 116) 90% 0.2· 0 BPT (n = 25) 63% . . . . 0 25 SO 75 100 125 150 175 200 225 Months Figure 13.12. Survival of Hodgkin lymphoma patients treated with a bleomycin-containing combination (usually ABVD or MOPP / ABVD). The graph shows the percent of patients surviving versus the number of months after treatment After 5 years, 90% of patients who did not have lung toxicity were still alive (upper curve), whereas only 63% of patients who did have lung toxicity survived (lower curve) (Martin et al., 2005). If its lung toxicity could be controlled, bleomycin could help cure many more cancer patients. Many attempts were made to control the lung toxicity ofbleomycin. The receptor kinase inhibitor, nintedanib, reduced bleomycin-induced lung damage in mice. The fibroblast inhibitor, pirfenidone, reduced bleomycin-induced in the lung of rats. But how effective these small-molecule drugs were in bleomycin-treated patients was uncertain. The large cyclic peptide, everolimus, inhibits the proliferation of fibroblasts, suggesting that it might inhibit fibrosis in the lungs ofbleomycin-treated patients. Clinical trials of this 273 K. W. Kohn Drugs Against Cancer CHAPTER!3 drugs however were disappointing. Drugs that tended to keep the bronchi open were also ineffective (Della Latta et al., 2015). References Bonadonna, G., Valagussa, P., and Santoro, A. (1986). Alternating non-cross-resistant combination chemotherapy or MOPP in stage IV Hodgkin's disease. A report of 8-year results. Annals of internal medicine 104, 739-746. Burger, RM., Peisach, J., Blumberg, W.E., and Horwitz, S.B. (1979). lron-bleomycin interactions with oxygen and oxygen analogues. Effects on spectra and drug activity. The Journal of biological chemistry 254, 10906-10912. Canellos, G.P., Rosenberg, S.A., Friedberg, J.W., Lister, T.A., and Devita, V.T. (2014). Treatment of Hodgkin lymphoma: a SO-year perspective. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 32, 163-168. Cuceu, C., Hempel, W .M., Sabatier, L., Bosq, J., Carde, P., and M'Kacher, R. (2018). Chromosomal Instability in Hodgkin Lymphoma: An In-Depth Review and Perspectives. Cancers (Basel) 10. Della Latta, V., Cecchettini, A., Del Ry, S., and Morales, M.A. (2015). Bleomycin in the setting of lung fibrosis induction: From biological mechanisms to counteractions. Pharmacological research 97, 122-130. Einhorn, L.H. (1981). Testicular cancer as a model for a curable neoplasm: The Richard and Hinda Rosenthal Foundation Award Lecture. Cancer research 41, 3275-3280. Hecht, S.M. (2000). Bleomycin: new perspectives on the mechanism of action. Journal of natural products 63, 158-168. Iqbal, Z.M., Kohn, K.W., Ewig, R.A., and Fornace, A.J., Jr. (1976). Single-strand scission and repair of DNA in mammalian cells by bleomycin. Cancer research 36, 3834-3838. Kohn, K.W., and Ewig, RA. (1976). Effect of pH on the bleomycin-induced DNA single- strand scission in L1210 cells and the relation to cell survival. Cancer research 36, 3839-3841. Martin, W.G., Ristow, K.M., Habermann, T.M., Colgan, J.P., Witzig, T.E., and Ansell, S.M. (2005). Bleomycin pulmonary toxicity has a negative impact on the outcome of patients with Hodgkin's lymphoma. Journal of clinical oncology: official journal of the American Society of Clinical Oncology 23, 7 614-7 620. Narayan, V., Deshpande, C., Bermudez, C.A., Golato, J.M., Lee, J.C., Diamond, J., and Vaughn, D.J. (2017). Bilateral Lung Transplantation for Bleomycin-Associated Lung Injury. Oncologist 22, 620-622. Povirk, LF., Wubter, W., Kohnlein, W., and Hutchinson, F. (1977). DNA double-strand breaks and alkali-labile bonds produced by bleomycin. Nucleic acids research 4, 3573- 3580. Reinert, T., da Rocha Baldotto, C.S., Pereira Nunes, F.A., and de Souza Scheliga, A.A. (2013). Bleomycin-Induced Lung Injury. J Cancer Res 2913, 1-9. Sausville, E.A., Peisach, J., and Horwitz, S.B. (1976). A role for ferrous ion and oxygen in the degradation of DNA by bleomycin. Biochemical and biophysical research communications 73, 814-822. 274 K. W. Kohn Drugs Against Cancer CHAPTER!3 Sleijfer, S. (2001). Bleomycin-induced pneumonitis. Chest 120, 61 7-624. Umezawa, H., Ishizuka, M., Maeda, K., and Takeuchi, T. (1967). Studies on bleomycin. Cancer 20, 891-895. Yu, Z., Yan, B., Gao, L., Dong. C., Zhong. J., M, D.O., Nguyen, B., Seong Lee, S., Hu, X., and Liang. F. (2016). Targeted Delivery o f Bleomycin: A Comprehensive Antica ncer Review. Curr Cancer Drug Targets 16, 509-521. 275 K. W. Kohn Drugs Against cancer CHAPTER 14 Chapt,er-14. '"'~ N,ikuklphio ChromoJOm~ SLo,yond o MW4':l'O o/ta~c«I CClllffl" thffllpy ZZOBZ6a$3 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@oib gov CHAPTER14 The Philadelphia Chromosome Story and a new era of targeted cancer therapy. Normal cells have control systems that keeps cells well behaved; that is, they keep the cells from proliferating excessively. Malignant tumors, however, are often defective in those controls. We have learned a great deal about how those controls work, the molecules that carry them out, and the way they are defective in cancer. The challenge was how to take advantage of that knowledge for therapy. In order to do that, a molecular diagnosis was needed to tell physicians what molecular defects are driving the malignant process in each particular patient and what drugs may provide a remedy. This chapter is about an unexpected observation more than 60 years ago that became a harbinger of this approach long before the idea of targeted cancer therapy was even conceived. It came from noticing under the microscope something strange that came to be known as the "Philadelphia chromosome" in honor of the city where it was discovered. In 1960, Peter C. Nowell and David A. Hungerford at the University of Pennsylvania noticed something strange about the chromosomes in the leukemia cells of 7 patients with chronic myelogenous leukemia (CML). One of the smallest of the 46 chromosomes that human cells normally have was even smaller than usual. They published their observation as a brief note in Science (Nowell and Hungerford, 1960). Even though their brief report was immersed among other small reports, it did not go unnoticed. In modern parlance, one would say it went viral. Cancer cells were long known to have scrambled and abnormal chromosomes, but the observation of a specific chromosome change in a particular type of malignancy was so novel and remarkable that it was soon confirmed in many laboratories, and the strange 276 K. W. Kohn Drugs Against cancer CHAPTER 14 little one then came to be known as the Philadelphia chromosome (Figure 14.1). Nowell and Hungerford surmised correctly that the novel little chromosome was somehow connected with the cause of the disease (CML), but they had no clue just how. The Ph• chromosome, as it is now designated, is nearly diagnostic for chronic myelogenous leukemia (95% of patients with CML have it), although it also occurs occasionally in some other types of cancer. At last, in 1973, Janet D. Rowley at the University of Chicago figured out what the Ph• chromosome was (Rowley, 1973). Using new staining techniques, she saw that it consisted of a piece of chromosome 9 (a moderately long chromosome) stuck to a piece of chromosome 22 (one of the smallest chromosomes). A combination of parts from those two chromosomes is what the tiny Ph• chromosome was. Still, she had no idea how the Ph• chromosome caused the disease. She did however know how such abnormal chromosomes form: by a phenomenon called "chromosome translocation" that tends to occur in many types of cancer cells, as well as in cells exposed to radiation or mutagens that break chromosomes. Broken ends of chromosomes often stick to each other forming abnormal chromosomes from the joined-up pieces. Thus Ph• results from a t(9;22) translocation between chromosomes 9 and 22 (Figure 14.2). Chromosome translocations are common in cancer cells, but this particular one is closely associated with and by far the most frequent cause of CML. Ph• was noticed because of its unusually small size and frequent occurrence in CML. But it was still puzzling why this particular translocation tends to occur, considering the huge number of different translocation possibilities that might exist among the chromosomes of a cell. And why was it specifically associated with CML? It took another decade to work out what was going on in the Philadelphia chromosome (Ph•). In chromosome 9, there is a gene called ABL (or recently denoted ABLl) that tends to push cells to divide and multiply. Normally, ABL is kept under control, so that it doesn't cause cells to keep on dividing like the brooms in The Sorcerer's Apprentice (from the film "Fantasia"). In chronic myelogenous leukemia (CML), an unregulated ABL keeps immature white blood cells dividing until they eventually overwhelm the body. ABL remains active and out of control in CML because of a gene region in a part of the Ph• chromosome, called BCR, that comes from chromosome 22 (Figure 14.2). The BCR-ABL combination is what causes the trouble: the BCR in the piece from chromosome 22 is right next to the greater part of the ABL gene that comes from chromosome 9. The BCR part stimulates the ABL part to produce a large amount of an abnormal ABL protein that continually pushes the cell into the cell division cycle. (Italics are commonly used when the name refers to a gene, as opposed to its protein product that may go by the same name.) On the positive side, however, it gave oncologists a target, namely the abnormal BCR-ABL protein, which only CML cells need to stay alive and actively dividing. It looked like a perfect chance to kill those malignant cells without harming normal cells. To understand how that therapy works and why it is not by itself the whole solution, we must delve a little deeper into how BCR-ABL causes its effects: how it induces cells to keep dividing. 277 K. W. Kohn Drugs Against cancer CHAPTER 14 To summarize to this point: what is important about the Ph• chromosome is not its small size, but the way the two chromosomes, 9 and 22, join so as to connect the ABL gene from chromosome 9 directly to the BCR gene from chromosome 22 (Figure 14.2). That rearrangement caused an abnormal protein to be made that included almost all of the normal ABL protein plus a piece that is coded by the BCR gene (Ben-Neriah et al., 1986) (Heisterkamp et al., 1983). The attached piece of BCR stimulated the action of the slightly truncated ABL gene, thereby producing an abnormal ABL protein that induced the cells to divide without end: the attached BCR piece prevented the ABL part from being turned off. Consequently, the malignant cells continued to proliferate without control (Wang and Pendergast, 2015). The defective control however provided an opportunity for therapy. The breakage and rejoining points on chromosomes 9 and 22 (a bit of minor detail here) were not always in exactly the same place, which means that the resultant BCR-ABL protein varied somewhat from one patient to another. The clinical picture of the disease therefore varied somewhat (Lugo et al., 1990). In fact, the reason that the break was in approximately the same place in chromosome 22 is that BCR, which stands for "breakpoint cluster region" was a region of the chromosome that, as its name implies, was prone to break. Figure 14.1. The Philadelphia chromosome (Ph 1) in a chronic myelogenous leukemia (CML) cell. The patient's CML cells were cultured, blocked in meta phase with vinblastine (see Chapter 10) and then stained to allow chromosomes to be identified. The identifiable chromosomes in this image w ere numbered. (From (Rowley, 1973) with red oval added}. 278 K. W. Kohn Drugs Against cancer CHAPTER 14 Philadelphia chromosome :-·· BCRQ AJL ABL~ 9 22 Normal Translocated Figure 14.2. How a Philadelphia chromosome forms by translocation of parts from chromosomes 9 and 22. The translocation puts the ABL gene from chromosome 9 next to the BCR gene from chromosome 22 in a new chromosome, called the Philadelphia chromosome. The dotted line show where chromosomes 9 and 22 break; the parts of the two chromosomes then join (translocate) as shown on the right. How the ABL gene was discovered. In the 1960's several cancer-causing viruses had been discovered. Each of them had somehow incorporated a gene that drove the cancer, which was usually a leukemia, lymphoma or sarcoma. The viruses induced cancers in various strains of mice. In 1970, Herbert Abelson and Louise Rabstein reported a new cancer-causing virus in mice that was unusual in that it did not involve the thymus (Abelson and Rabstein, 1970). The new virus acquired the name Abelson leukemia virus and the cancer-causing gene (oncogene) harbored by the virus was duly dubbed ABL ( or v -ABL to distinguish from c-ABL for the gene in normal cells). That was during an era when much effort was directed to the idea that viruses were the major cause of cancer in humans, an erroneous idea that only slowly died; a large story could be told about that era. Nevertheless, the oncogenes in those viruses were major causes of human cancers (Chapter 15). The oncogenes were in fact mutated versions of normal genes that the viruses had picked up during their transfer in mice. The normal versions of those genes were capable of driving cell division, but that action was normally under control, so as to limit how often a cell would divide. Control was lost when the gene became mutated or otherwise altered; without the control, the excessive cell division led to cancer. Taking advantage of the control defect caused by the translocation. 279 279 K. W. Kohn Drugs Against cancer CHAPTER 14 It was long known that the ABL gene codes for a protein tyrosine kinase. Tyrosine kinases are proteins (enzymes) that can stick phosphate groups onto tyrosine amino acid units of other protein molecules. Those kinases each act on particular proteins that convey signals to tum particular cell functions on or off. The proteins phosphorylated by the ABL tyrosine kinase sends signals -- mostly via chains of phosphorylation events -- to the system that initiates processes leading to cell division. That's how the overactive ABL gene in the Philadelphia chromosome caused the malignant disease. (The important thing about a phosphate group, by the way, is that it has a negative charge. An electric charge on a protein can have a big effect on its structure and function. A cell's regulatory network is in that respect somewhat like an electronic computer: presence or absence of a phosphate on a particular protein is like on/off in an electronic unit. Phosphorylations on serine or threonine units, too, can regulate protein functions, but, for many initiators of cell division, it is usually the phosphorylations on tyrosines that are most important.) It turned out that, not only did inhibitors of the BCR-ABL protein stop the uncontrolled cell division, but they caused the malignant cells to die. It was as if the malignant cells had become addicted to the abnormally high levels of ABL tyrosine kinase activity; when that activity was cut off, the cells died. To recapitulate the important point: when ABL's tyrosine kinase activity was continually on, as in the BCR-ABL fusion protein in CML, signals were continually sent to activate the proteins that initiated cell division. An inhibitor of the tyrosine kinase function of the BCR- ABL protein would therefore halt the malignant cell division process (Druker, 2002). (When we say "BCR-ABL", we have to specify whether we mean the fusion protein or the gene that codes for it I use the convention of gene names in italics.) It was a long way to go from the discovery of the Philadelphia chromosome to effective treatment of CML that stretched to nearly 4 decades. Selective inhibitors of ABL tyrosine kinase At this point in the story, the basic science information indicated that, if oncologists could inhibit the tyrosine kinase activity of the ABL part of the BCR-ABL fusion protein, the uncontrolled proliferation of CML cells would be stopped, and the malignant cells might even die. Several tyrosine kinase inhibitors were already known in the 1980's (Druker, 2002), but the problem was that the cell has many different tyrosine kinase proteins that it needs in order to regulate many essential processes. The previously known inhibitors of tyrosine kinases were non-selective: they inhibited a great many of them of them. A normal cell can tolerate inhibition of the tyrosine kinase activity of its ABL protein, but inhibiting many of the cell's other tyrosine kinases would not be good. It was necessary to find inhibitors that mostly inhibited the tyrosine kinase activity of only the ABL protein. 280 K. W. Kohn Drugs Against cancer CHAPTER 14 An enormous effort was made to find the right kind of selective tyrosine kinase inhibitors. Many compounds were synthesized by chemists or obtained from natural sources and tested for their abilities to inhibit different tyrosine kinases needed by the cell. In addition, researchers tested the ability of their potential drugs to selectively kill or inhibit the proliferation of cells that had the BCR-ABL fusion protein. To guide the search for the best chemical structure, researchers studied the relationship between the chemical structures of the compounds and their abilities to selectively inhibit the ABL tyrosine kinase or to kill only cells that have the BCR-ABL fusion protein. The first drug to come out of that endeavor and promising enough to put into clinical trial against CML was imatinib, also called Gleevec (Figure 14.3) (Druker, 2002; Druker et al., 1996). Researchers at Oregon Health Sciences University and Ciba-Geigy of Basel, Switzerland found a specific ABL inhibitor in 1996 that they called CGP57148 and which became known as imatinib or Gleevec. Importantly, the new drug inhibited the tyrosine kinase activity equally well of BRC-ABLand of normal ABL (c-ABL) and was inactive against a panel of other tyrosine kinase, as well as serine/threonine kinases (activity was also noted against the receptor of platelet-derived growth factor receptor (PDGF-R)) (Table 1). Tab le 1 Profile of inhibition of p rotein l<lnases by CGP 57148 Protein kinase Substrate Cellular tyrosine phospho,ytation phosphorylation IC~ value (µM ) IC,. value (µM) v-Abl 0.038 0.25 Bcr-Abl 0.025 0.25 c-Abl 0.025 EGFR,R- ICD >100 >100 Her-2/neu >100 Insulin receptor >1 00 IGF-lR >100 PDGF-R 0. 3 c-5rc , 100 v-Src >100 c-Fgr >100 c-Lyn >100 v-Fms >100 TPK-118 >100 PKA >500 PPK >100 PKC o, Pl, p2, y, e, a,~-< >100 Casein klnases - 1 and 2 >100 cdc2/cyclin >100 Abbreviations: EGF-R- ICO, epidermal growth factor receptor-intra- cellular d omain; IGF-1 R, insulin like growth factor-1 receptor; TPK, tyrosine protein kinase; PKA, protein kinase A; PPK, phosphorylase kinase; PKC, protein kinase C. Table 1. Specificity of Ciba-Geigy CGP57148 for inhib ition of the protein kinase activity of ABL. The drug became known as imat inib or Gleevec (Druker et al., 1996). 281 K. W. Kohn Drugs Against cancer CHAPTER 14 lmatinib (Gleevec) Figure 14.3. Chemical structure of imatinib (Gleevec), the first clinically effective inhibitor of the BCR-ABL tyrosine kinase in the treatment of chronic myelogenous leukemia (CML). The drug inhibited BCR-ABL, as well as the normal ABL tyrosine kinase, but had no effect on a panel of other protein tyrosine kinases (Buchdunger et al., 1996; Druker et al., 1996). Results in the early clinical trials were spectacular: in a phase III trial of 553 newly diagnosed CML patients, 96% of those treated with imatinib had a complete disappearance of visible CML cells from the blood and bone marrow, and in 68% there was no longer any trace of the Philadelphia chromosome. These remissions of the disease lasted more than 14 months, which was the time limit of that trial. Compared to an equal number of patients who received the previous standard treatment with interferon plus cytosine arabinoside, treatment with imatinib was much superior (Druker, 2002). The success of the treatment of CML with imatinib opened the door to the era of targeted cancer therapy: it was the first time that a successful drug was designed to act on a specific protein target The clinical researchers were impressed by the low toxicity of effective imatinib treatments, which was very different from the experience with other cytotoxic chemotherapy. Less that 1 % of patients had side effects severe enough to limit treatment with imatinib (Druker, 2002). But why did imatinib give long-term survival of most chronic myelogenous leukemia (CML) patients, while it was not nearly as effective in treatment of other malignancies? One possibility that was considered was that the ABL gene is mainly needed only during development of the embryo and is dispensable in adults (Wang. 2014). Therefore, a drug that specifically targeted ABL could be given at high enough dosage to completely block the ABL component of the abnormally active BCR-ABL fusion protein that drives the disease. Many tyrosine kinases are located in the cell surface membrane and convey signals from receptors in the outside to actions in the inside of the cell (Chapter 17). ABL however is a non-receptor tyrosine kinase, not localized to the cell surface. Instead, it shuttles information between cytoplasm and nucleus. It can be activated by certain receptor 282 K. W. Kohn Drugs Against cancer CHAPTER 14 tyrosine kinases, from which it then transmits signals to the nucleus to activate genes for cell division. When not engaged in this signal transmission task, ABL normally is self- inactivated. The BCR-ABL combination gets around this self-inactivation and causes the ABL signaling to continue non-stop, thereby inducing non-stop cell division and cancer. ABL-inhibitor drugs, such as imatinib/Gleevec, blocked ABL's tyrosine kinase activity, thereby blocking its ability to signal genes in the nucleus to initiate cell division. How imatinib inhibits the ABL tyrosine kinase activity. The molecular details of how imatinib inhibits ABL was revealed by crystallographic analysis that showed the structure of the protein and how imatinib binds to it (Figure 14.4). ABL was shown to work by first binding ATP within a pocket in the protein and then transferring the high energy phosphate bond from ATP to phosphorylate the substrate proteins. Imatinib binds to ABL in the pocket where ATP ought to bind but cannot because the drug is already bound there. Figure 14.4. The crystallographic structure of the ABL protein, showing the pocket where ATP would bind. The structure shows imatinib binding within the pocket, thereby preventing ATP from binding there, and thereby inhibiting ABL's tyrosine kinase activity. (From (Patel et al., 2017) with labels added.) Ty rosine kinase inhibitors FDA-approved fo r treatment ofCML. The large majority of chromic myelogenous leukemia (CML) patients treated with imatinib had long-lasting remissions. Nearly 90% of the patients survived more than 5 years without signs of any return of the disease (Eide and O'Hare, 2015). However, patients 283 K. W. Kohn Drugs Against cancer CHAPTER 14 eventually relapsed and became resistant to imatinib. Therefore, new drugs were sought for patients who had become resistant to imatinib. By 2015, four new tyrosine kinase inhibitors were approved by the U.S. Food and Drug Administration (FDA) for treatment of malignancies driven by chromosome translocations yielding a BCR-ABL fusion (Eide and O'Hare, 2015) (Figure 14.5). Binding site/ Regulatory status/ Inhibitor Chemical structure Inhibitor type approval lmatinib (Gleevec) t;()' • 'S • ,< yCr''l 0 V, ATP-bind ing site/ ATP-competitive FDA approved/ frontline therapy rt• Nilotinib (Tasigna) F~ H 'S ATP-bind ing site/ ATP-competi t ive FDA approved/ frontline therapy Dasat inib .J-.. ATP-bind ing site/ FDA approved/ (Sprycel) &:y(½~----i v"""'" 0 o ATP-com petit ive front line t herapy ox,: Bosuti nib (Bosulif) •·<w . ("'t:.............,.o I" ,..__,) " ~ ( I _,. ATP-bind ing site/ ATP-competitive FDA approved/ 2nd-line therap•f 0:,, Ponati nib (lclusig) l I I o ~'Q-J•' I _,. ", -.,) ATP-binding site/ ATP-competitive FDA approved/ 2nd-line t herap·f Figure 14.5. The FDA had by 2015 approved these five tyrosine kinase inhibitors to treat malignancies driven by BCR-ABL (Eide and O'Hare, 2015). How the function of the ABL protein is regulated. I begin with the help of Figure 14.6 to relate what was learned about the function of the different parts of the ABL protein. It consists of a series of domains and motifs lined up from the amino-terminus to the carboxy-terminus of its amino sequence (Figure 14.6). This 284 K. W. Kohn Drugs Against cancer CHAPTER 14 kind of cobbling together of domains and motifs, as well as phosphorylation sites, is typical for many proteins that function in regulatory pathways. Starting at the amino end (left), we find an SH2 domain that is notable for its ability to bind phosphorylated tyrosines of proteins. It is followed by an SH3 domain that binds to some amino acid sequences that have two prolines separated by two other amino acids (PXXP). Then, there is a tyrosine (Y) that can be phosphorylated; it is within a short sequence of amino acids that links to the next domain where the tyrosine kinase enzyme function of the protein resides. We come next to an amino acid stretch that contains three PXXP motifs that can serve to bind to SH3 domains of various other proteins. Next is the DNA-binding domain that binds to certain gene promoters and turns on the transcription of those genes. Interestingly, ABL is both an enzyme and a gene regulator. The latter activity depends on the amount of ABL that is in the cell nucleus, which is regulated by two different motifs: one controls its entry into the nucleus (NLS, nuclear localization signal); the other controls its exit (NES). Finally, we come to two domains that bind to actin cytoskeleton; this binding tends to keep ABL in the cytoplasm and out of the cell nucleus. Such an arrangement of SH2, SH3, PXXP, and phospho-tyrosine domains and motifs allowed ABL to link to other proteins to form multi-protein integrated networks of regulated functions. It turned out, however, that those domains and motifs formed bonds within the same protein molecule as well, forming an internal clamp that keeps the ABL protein inactive until the clamp was relieved by external interactions (Hantschel and Superti-Furga, 2004). This situation is the same in a closely related tyrosine kinase, SRC, for which I had some years ago prepared a molecular interaction map (Figure 14.7). The legend to the Figure describes the essential features of the regulation of this internal clamp. ABL in cancers. ABL was found to promote the development of several types of cancers other than leukemias. The activation of ABL in the solid-tumor-type cancers, however, was often not due to chromosome translocations. Sometimes the ABL gene in these cancers was amplified or mutated, thereby increasing its promoting of the cancers. An amplified gene has multiple copies of the gene that work together and increase the net expression of the gene. The possible roles of those actions on cancers however was not fully established (Wang and Pendergast, 2015). Some cases of ABL gene mutations were found but their role in cancer also was not entirely clear. Nonetheless, some patients benefitted from treatment with the ABL tyrosine kinase inhibitors dasatinib, bosutinib or nilotinib Oones and Thompson, 2020). Clearly, overactive ABL promoted the development of malignancies, particularly chronic myeloid leukemia (CML). As already explained above, the activity of ABL was normally kept in check by internal bindings: the SH3 domain with the PXXP motif and the SH2 domain with phospho-tyrosines Y245 and Y412 (Figure 14.6). When ABL was activated in the course of the normal functioning of the cell, this happened by controlled external interactions that competed with the internal inhibitory bindings (Figure 14 .7). 285 K. W. Kohn Drugs Against cancer CHAPTER 14 Uncontrolled activation of ABL was found to occur when a recombination deleted or disrupted its SH3 domain, which is at the N-terminal end of the protein (Figure 14.8). Nuclear Nuclear Binds to impon eJCPOrt Tyrosine-P Tyrosine-P signal signal Binds to SH2 Binds to PXXP motifs Binds to .. cop ABLl •• Tyrosine kinase domain Figure 14.6. Domain structure of the ABL protein showing the motifs and domains and whatthey bind to. Like many signaling proteins, ABL is made up of a number of binding motifs and domains that are cobbled together into an integrated functional unit Y = tyrosine; P = phosphate. ABLl = ABL. (ABL2 = ARG, which is not included here.) (From (Wang and Pendergast, 2015) with labels in red added.) 286 K. W. Kohn Drugs Against cancer CHAPTER 14 EGF IJ P las m a memb rane My ri styl Cbp 20 J pY ....,,........-1-!~ p85-Pl3K SH3 Pro PTPa ? 13 8 74 pY314 L 7 6 3 11 pY416 pYS27 18 17 19 PEP ('r O"t~bc;;;r~ S.;;,:;i , b;-;;s:;;f,.::: a:i':tc ;:-;s:'l)+(~ '..... pY Src3e Figure 14.7. Molecular interaction map (Kohn, 1999) of the SRC tyrosine kinase, showing the internal clamp and its release, which are very similar in ABL. The double-arrowed black lines point between elements that bind to each other. Bindings between different parts of the protein form an internal clamp consisting of SH3 bound to Pro (PXXP) and SH2 bound to a phospho-tyrosine (PY); this is the same in SRC and ABL. The internal clamp can be released by the combined actions of (1) a Pro (PXXP) domain of p85-PI3K binding to SRC's SH3, displacing the internal bond from the SH3 to Pro; (2) a phospho-tyrosine of EGFR binding to SRC's SH2, displacing the internal bond from the SH2 to a phospho-tyrosine near the carboxy end of the protein. These interactions are similar for SRC and ABL, except that the clamp release proteins may differ. The two steps can happen in concerted fashion, because the p85-Pl3K and EGFR are bound to each other. (The myristyl group that is linked to the amino-terminal region of ABL, rather than being bound to the cell membrane as shown in the diagram, actually binds to a hydrophobic pocket in the protein itself. This intramolecular binding further stabilizes the clamp that inhibits the kinase.) Chromosome translocations drive leukemias. It turned out that BCR on chromosome 22 (Figure 14.2) was not the only translocation that activated ABL in leukemias. Alternative translocations were found that occasionally drove 287 K. W. Kohn Drugs Against cancer CHAPTER 14 leukemias. In those cases, there was no Philadelphia chromosome, because the translocated chromosomes were not so tiny. Figure 14.8 shows some of those translocations. The break site of the translocation was often at a place that retained at least much of the SH3 domain at the amino end of the protein (upper part of Figure 14.8). @) ETV6 EMLl ABLl NUP214 ZIMZl SEPT9 RCSDl SNX2 ABLl FOXPl SFPQ Figure 14.8. Alternative chromosome translocations in leukemias (Wang and Pendergast, 2015). The names on the left are of the genes that became placed next to theABL gene. These translocations did not result in a Philadelphia chromosome. In the cases shown in the lower part of the Figure, the translocation cuts the ABL after the SH3 domain, which therefore is not included in the product of those translocations, and internal self- inactivation cannot occur. (From (Wang and Pendergast, 2015) with red oval added.) Chronic myelogenous leukemia (CML} becomes resistant to imatinib. Although the great majority of CML patients (as well as occasional BCR-ABL-positive acute lymphatic leukemia (ALL) patients) responded well to imatinib for several years, the disease eventually recurred and was then resistant to the drug. The resistance was usually due to a mutated BCR-ABL that did not bind the drug. Two new ABL inhibitors, nilotinib and dasatinib, were active against some of those mutants; the chemical structure of nilotinib closely resembles imatinib, while dasatinib is more different (Figure 14.5). However; there was a particular mutant, BCR-ABL13151 (threonine at position 315 of ABL replaced by isoleucine) that was resistant to all three drugs. To meet this problem, a new inhibitor, ponatinib (Figure 14.5; originally AP24534), was developed that worked against that mutant, as well as against other forms of BCR-ABL (O'Hare et al., 2009). Another drug, axitinib, found to work against CML cells harboring the BCR-ABLT3151 mutation is discussed in the next section. The most common way that resistance developed was by a mutation in the cell's ABL protein that altered the drug-binding pocket in a manner that prevented the drug from 288 K. W. Kohn Drugs Against cancer CHAPTER 14 binding there (Greuber et al., 2013). Many mutations around the pocket region of the ABL protein were discovered that reduced the effectiveness of imatinib. Drugs were developed that could bind to ABL despite the most common mutations, and some of those single mutations could be circumvented by one or another of the new inhibitors listed in Figure 14.5. There were however cases of double mutations for which no drug therapy was available (Eide and O'Hare, 2015). Other albeit less common mechanisms of resistance were discovered that had to do with the fact that ABL acts inside the cell nucleus. The membrane that encompasses the nucleus has channels that pump ABL into the nucleus or that pump it out. Resistance occasionally developed due to defective exit pumps or to overactive input pumps. Either way, there would be too much ABL in the nucleus for the drugs to inhibit it all (Yaghmaie and Yeung, 2019). These pathways to drug resistance remained a challenge for medicinal chemists. New drug combinations f or chronic myelogenous leukemia (CML). Since 1986, the National Cancer Institute (NCI) and cooperating institutions have been accumulating vast amounts of data on gene expression and drug sensitivities of many human patient-derived cell lines. The data contain much information about correlations and patterns that could be used in clinical and experimental studies but accessing and making sense of the vast data was a big challenge. Software tools to access and analyze the data were therefore developed, led by a Genomics and Pharmacology group within NCl's Developmental Therapeutics Branch. As the final part of this chapter, I used one of the tool sets, CellMinerCDB (Luna et al., 2021), to look for relationships between the expression of the ABL gene and the drug-sensitivities of various human cancer cell lines, with special attention to lines from chronic myelogenous leukemia (CML) patients (Figures 14.9 -14.11 and Table 14.2). CellMinerCDB is freely accessible at https://discover.nci.nih.gov /rsconnect/cellminercdb/. CellMinerCDB has data for several datasets from different institutions. The most useful for the current analyses were the CTRP-Broad-MIT and the GDSC-MGH-Sanger datasets. Figure 14.9 shows that results from these two datasets agreed with each other quite well. They show that most of the CML cell lines (red dots) expressed the ABL gene (ABLl) more than did the great majority of other leukemia cell lines (Figure 14.9, left) or of all the other lines in the datasets (right). It is interesting, as well as reassuring. that these cell line data gave results consistent with the clinical and experiment findings: namely, that CML cells expressed ABLl to an unusually high degree. However, they also showed that there were a few non-CML lines that also exhibited high ABLl expression; this too was consistent with the clinical finding of occasional cancers other than CML that had high ABLl expression. In further exploration, I focused on the CTRP-Broad-MIT dataset, because it had the larger number of CML cell lines. Figure 14.10 shows that CML cell lines, which had relatively high ABLl expression, were highly sensitive to imatinib (Gleevec), as expected since the drug 289 K. W. Kohn Drugs Against cancer CHAPTER 14 was developed as a specific ABL inhibitor (Table 1). The other two drugs at the top of Figure 14.5 and Table 14.2, nilotinib and dasatinib, gave results similar to imatinib. The next step was to use CellMinerCDB to get a list of drugs whose action against leukemia cell lines was most highly correlated with ABLl expression (Table 14.2). We see that the top three entries in the table, as well as one at the bottom, were among the five drugs approved for treatment of ABL-positive CML, and whose chemical structures were shown in Figure 14.5. However, Table 14.2 also included several drugs that were approved for treatment of other cancers and that were reported to act on molecular targets other than or in addition to ABLl. Combining these different targets by combing the drugs could perhaps improve treatment The first of these possible combinations was imatinib (or dasatinib) together with axitinib (Figure14.11A and Table 14.2). We see that the CML cells were substantially more sensitive to both drugs than nearly all the other lines in the database. A literature search then disclosed a report that axitinib could overcome resistance mediated by a mutation of BCR- ABL (Halbach et al., 2016), which followed up on a brief letter that this might be the case (Okabe et al., 2015). There was also a more recent report of a CML patient who had become resistant to imatinib and dasatinib who then responded to a combination of axitinib and dasatinib (Deng et al., 2020). Other than those reports, I found nothing in the literature to relate axitinib to ABL or to CML, findings that came independently from CellMinerCDB (Figure14.11A and Table 14.2). Axitinib had been extensively investigated as an inhibitor of vascular endothelial growth factor receptors (VEGFR) that nourish cancers by stimulating blood vessel production in the tumors, and it had been combined with other drugs for treatment of several cancers other than leukemias. Clinical trials of axitinib in the treatment of BCR-ABL-mutated CML were planned. Another potentially effective combination in the findings with CellMinerCDB was imatinib together with crizotinib (Figure 14.11 B and Table 14.2). A literature search then disclosed a recent report that crizotinib inhibited resistant mutants including BCR-ABLn i si (Mian et al., 2021). The drug was known to block several protein kinases, most notably hepatocyte growth factor receptor (HGFR, also known as MET) and anaplastic lymphoma kinase (ALK) and was approved for treatment of advanced lung cancers that had an ALK fusion protein that was continually active. Here again, CellMinerCDB independently predicted this effective drug combination. Tivozanib, which appears below axitinib in Table 14.2, received FDA approval in March 2021 for treatment of advanced renal cell carcinoma (Chang et al., 2022). The drug is a tyrosine kinase inhibitor that targets vascular endothelial growth factor receptors (VEGFR), a platelet-derived growth factor (PDGFR), and KIT. However, there were no papers relating tivozanib to leukemia, CML, ABLl, or BCR-ABL in the PubMed literature. Figure 14.llC shows the selective responses of CML cells to tivozanib and imatinib. This is very much like the results with axitinib and crizotinib and suggests that the combination of tivozanib with imatinib or one of its approved relatives (nilotinib or dasatinib) would be another useful treatment ofCML with a different range of kinase targets upon which the combination would act and counter drug resistance. 290 K. W. Kohn Drugs Against cancer CHAPTER 14 Below tivozanib in Table 14.2, we come to a drug called pluripotin. Although it has received little or no attention in the recent literature and we still do not know its targets of action, a report suggested that pluripotin may affect cancer stem cells in culture (Mertins et al., 2013). In view of its selective action on CML cells (Figure 14.11D), it would be interesting to explore its targets of action on the possibility that it may offer a novel therapy for drug- resistant CML. Notably, CML cells were the only type of cells that showed a selective response to pluripotin. Next in Table 14.2, we come to masitinib, a tyrosine kinase inhibitor that targets PDGFR, FGFR, and KIT and that was in clinical trial against pancreatic cancer but showed little benefit (Aljoundi et al., 2019; Waheed et al., 2018). Masitinib was another drug that inhibited CML cells specifically and that could be considered for testing in combination with imatinib, nilotinib or dasatinib for treatment of drug-resistant CML (Figure 14.llE). Finally worth noting is that Nilotinib showed a remarkably strong selectivity for action on CML cells (Figure 14.2F) and might be particularly effective in drug combinations. Thus, CellMinerCDB can help reveal possible cancer targets, such as CML, including drugs already approved for treatment of other cancers. 291 K. W. Kohn Drugs Against cancer CHAPTER 14 ABLl (exp, CTRP•Broad•MJT) vs. ASL! (exp, GOSC•MGH•San9er) ABLl (exp, CTRP-Broad•MIT) V'S. ABU (exp, GDSC-MGH-Sanger) Pearson correlatlon (r) • 0.$7, p•value• 4 .9e• 1S Pearson correlatson (r)=0.79, p-vatue=◄ . 2e• 124 • •• • •• • 10 · ,.. . • •• • ••• ,.s • ti- • : E :r .," C e ~ 9.0 • • •• • •• • ...• c • •• • ~ •• • iii <( • • • 8.S ••• • • • • ••• • s.o • 7 S 6 7 4 ~ S 6 ~ , ABLl (exp, GOSC- MGH-Sanger) ABLI (exp, GDSC·MGH·S&ng,er) Figure 14.9. Six of eight chronic myelogenous leukemia (CML) cell lines (red) had high ABLl expression relat ive t o other cell lines. The cell lines shown were those for w hich there was dat a in both the CTRP-Broad-M IT (vertical axis) and t he GDSC-MGH-Sanger (horizonta l axis) data sets, showing the consistency between t he two dat a sets. Left, data for CML relative to other leukemia cell lines. Right, data for CML relative t o all other cell lines in bot h dat a set s. (I creat ed the graphs using CellM inerCDB version 1.4 (Luna et al., 2021). Sca les are in log2 units.) https://discover.nci.nih.gov/rsconnect/cellminercdb/ 292 K. W. Kohn Drugs Against cancer CHAPTER 14 lmatlnlb {act, CTRP· Broad-MIT) vs. ABU (exp, CTRP·Broad-MIT) lma tlnlb ( act, CTRP-BtOad- MIT) V$. ABL I ( exp, CTRP· Bl'Oad-MIT) __ Pearson correlation ( r)=0.59, p-value• l. le-0 1 Pearson correlation (r) • 0. 12, p -value• 0.000S8 2S • • •• • • • • •• • • • • , . • , •• ••• • • ~ 20· -g • •• • .,e #. "'• • ,... .-,.,. Q. • e tr .!!. .0 ·a • ,__,,.~ -4• ••--. . - • • • "'• ,$ 15 • • • • • 10 12 ~--------------~ 7.S 8.0 8.S 9 .0 9.S 10.0 6 1 8 9 W ABLl (exp, CTRP-8roa d-MIT) ABU (exp, CTRP·Broad· MIT) Figure 14.10. Response to imatinib (vertical axis) versus expression of ABLl (horizontal axis) of cell lines in the CTRP-Broad-MIT data set. Left, CML cell lines (red) relative to other leukemia cell lines. Right, CML cell lines (red) relative all other cell lines. (Graphs crea ted using CellMinerCDB version 1.4) 293 K. W. Kohn Drugs Against cancer CHAPTER 14 Table 14.2. Correlation of drug activities with ABLl gene expression. (Table created using CellMinerCDB version 1.4, data set CTRP-Broad-MlT for all leukemia cell lines.) Drug name Clinical status Correlation P-Value Nilotinib• FDA approval 0.636 4.19E-09 Dasat inib• FDA approval 0.590 9.33E-08 lmatinib* FDA approval 0.591 1.14E-07 saracatinib Clinical t rial 0.567 3.86E-07 crizotinib FDA approval 0.521 6.14E-06 axit inib FDA approval 0.512 7.98E-06 t ivozan ib FDA approval 0.501 2.09E-05 pluripotin 0.470 4.13E-05 masit inib Clinical t rial 0.445 2.07E-04 vandetanib FDA approval 0.356 3.87E-03 GW-843682X 0.381 8.26E-03 Bosutinib• FDA approval 0.329 9.02E-03 * ABLl-inhibitordrugs whose structure is shown in Figure 14.5. f w!lfflil> ( ~ CTRP·8•0kl•MIT) VS. lm;ll,lf'liO (kt, CTIU•·6•0ict-NJT) <rlt<ilinit> { llet", CTRP•8t0-'d,MIT) W. iM.:t!.it'lib (at.t, CTRP.8rQ.kl,MIT) Pi:or,on CIOoM'deltlon (rJ•0 ,6 5, i,-o;el...,.• l,6e·89 """"°" ~ <~)100,'9, p-~w o:6,te•28 " A B •• .·, • .•,,, • • ,.. • • • • .,,,, . ! -· ..\.,·.. • • • • • •• • 15 ,. • • • l ,o 15 lO •~unti («t, crRP ·6rooo-HlT) " ,o • 15 20 lm.aunlb ( ~. CTRP-.8t0"<f-MJ'T) " 294 K. W. Kohn Drugs Against cancer CHAPTER 14 t1VOZ¥Olb (ea, CTR.P·6roed ·MIT) vs. lmotlnlb («t, CTRP+tlt'ood·M IT) p1ut'll)01fn ( ~ CTII.P •Brolld· MtT) \15, ll!'llltlnil> (act, ctRP·&OIH:1-fflT) • ~ COff'f:'latton (r): 0.•11. p-v,a1uc=6.St'· 32 Pffrton com!lation {r)• O,S9, 0•11eh,, e• 1, Hl·7'0 30 C D ,s • • ,,., . .• .: • .,• .. . .•.. 25 '/ ' • •• • •• ,o l .. IS 20 ftn;Ull'litl (kt, cn:tP<sBt'OIHl •M.lT) 2S 10 IS 2<) fm,ttjnib (•ct.. CTII.P•8ro.d•MIT) 2S MMitinib (iCl, CTIV'- ~(1 -MIT} 11$, im;ninib (kl, CTllP-6 roecMolJT) n ilotinib { ctct, CTRP..6.-o.d•MJT) .,.,. .ncltinib (Clct. CfRP•6tWd·Mll ) Pleorsoncorrelotkln (r) • 0.78, ~ • 2. 7c·1Sl Pcor:1on con-eh,,tlOn ( r) • 0.6 6, p•vol1.te• 3 . lc·94 E • F .·- :- •• •• •• 2.5 •• • • • ......···"./ .. .. ••• •' • E ..i "/ / • 5 20 ! 7.• ~ ! lS " • " ,o / 1$ lmatinlb (~cc, cn:u>- ~ 20 -MIT) " ,ol,o - lS 20 1m1utn1b ( llct, CTRP·8ro11d·M.IT) 25 Figure 14.11. Chronic myelogenous leukemia (CML) cell lines (red) responded strongly to drugs from Table 14.2. The data were from the CTRP-Broad-MIT data set; CML cell lines in red, other cell lines in blue. The response levels shown numerically on the axes are in log2 units. Horizontal axes: response to imatinib. Vertical axis: response to A, axitinib; B, crizotinib; C, tivozanib; D, pluripotin; E, masitinib; F, nilotinib. These represent possible 295 K. W. Kohn Drugs Against cancer CHAPTER 14 drug combinations with imatinib ( or with nilotinib or dasatinib). (Graphs created using CellMinerCOB version 1.4.) Summary It was a long path from the time that Peter Nowell and David Hungerford first noticed the tiny Philadelphia chromosome in patients with chronic myelogenous leukemia (CML) in 1960. The next landmark did not occur until 1973, when Janet Rowley figured out that the Philadelphia chromosome resulted from a translocation between chromosomes 9 and 22. Further elucidation came indirectly from an unusual mouse leukemia virus discovered by Herbert Abelson and Louise Rabstein. The virus was found to harbor a cancer-causing gene - a mutated normal gene - that came to be known as the ABL oncogene. ABL was found to be a tyrosine kinase: an enzyme that puts phosphate groups onto particular tyrosines in proteins. Hard work by medicinal chemists then came up with the selective ABL tyrosine kinase inhibitor, imatinib, popularly known as Gleevec. lmatinib changed the world of patients suffering from chronic myelogenous leukemia (CM L): 90% of them had long-term survival and appeared to be cured without having to endure severe toxicity. Eventually however, mutant CML cells appeared that were resistant to the drug. Medicinal chemists then went to work and developed several drugs that were effective against some of the resistant cases and that received FDA approval. But certain of the CML mutations resisted all of the approved drugs. Finally, a study using a software tool, called CellMinerCDB, of the selective responses of CML cell lines to the drugs found several drug combinations for possible testing against drug-resistant CML. The development of a drug. imatinib (Gleevec), that was specific for cancers caused by a specific oncogene (ABLl) issued in the new era of targeted cancer therapy. References Abelson, H.T., and Ra bstein, LS. {1970). Lymphosa rcoma: virus-induced thymic-independent disease in mice. Cancer research 30, 2213-2222. Aljoundi, A.K., Agoni, C., Olotu, F.A., and Soliman, M.E. {2019). 'Piperazining' the catalytic gatekeepers: unraveling the pan-inhibition of SRC kinases; LYN, FYN and BLK by masitinib. Future Med Chem 11, 2365-2380. Ben-Neriah, Y., Daley, G.Q., Mes-Masson, A. M., Witte, O.N., and Baltimore, D. {1986). The chronic myelogenous le ukemia-specific P210 protein is the product of the bcr/abl hybrid gene. Science 233, 212-214. Buchdunger, E., Zimmermann, J., Mett, H., Meyer, T., Muller, M., Druker, BJ., and Lydon, N.B. {1996). Inhibition of the Abl prote in-tyrosine kinase in vitro and in vivo by a 2- phenylaminopyrimidine derivative. Cancer resea rch 56, 100-104. Cha ng, E., Weinstock, C., Zhang, L, Fiero, M.H., Zhao, M., Zahalka, E., Ricks, T.K., Fourie Zirkelbach, J., Qiu, J., Yu, J., et al. {2022). FDA Approval Summary: Tivozanib for Relapsed or Refractory Renal Cell Carcinoma. Clinica l cancer resea rch : an official journal of the American Association for Cancer Research 28, 441-445. 296 K. W. Kohn Drugs Against cancer CHAPTER 14 Deng, Q , Wang, E., Wu, X., Cheng, Q, Liu, J., and Li, X. (2020). Combination of axitinib w ith dasatinib improves the outcome of a chronic myeloid leukemia patient with BCR-ABLl T315I mutat ion. Zhong Nan Da Xue Xue Bao Yi Xue Ban 45, 874-880. Druker, BJ. (2002). Inhibition of t he Bcr-Abl tyrosine kinase as a t herapeutic strat egy for CML. Oncogene 21, 8541-8546. Druker, BJ., Tamura, S., Buchdunger, E., Ohno, S., Sega l, G.M., Fanning, S., Zimmermann, J., and Lydon, N.B. (1996). Effects of a selective inhibitor of t he Abl tyrosine kinase on t he growth of Bcr-Abl positive cells. Nature medicine 2, 561-566. Eide, C.A., and O'Hare, T. (2015). Chronic myeloid leukemia: advances in understanding disease biology and mechanisms of resist ance to tyrosine kinase inhibit ors. Curr Hematol Malig Rep 10, 158-166. Greuber, E.K., Smith-Pea rson, P., Wang, J., and Pendergast, A.M . (2013). Role of ABL family kinases in cancer: from leukaemia t o solid t umours. Nat ure reviews Ca ncer 13, 559-571. Halbach, S., Hu, Z., Gretzmeier, C., Ellermann, J., Woh rle, F.U., Dengjel, J., and Brummer, T. (2016). Axitinib and sorafenib are pot ent in tyrosine kinase inhibit or resistant chronic myeloid leukemia cells. Cell Commun Signal 14, 6. Hantschel, 0., and Superti-Furga, G. (2004). Regulation of t he c-Abl and Bcr-Abl tyrosine kinases. Nat ure reviews M olecular cell biology 5, 33-44. Heisterkamp, N., St ephenson, J.R., Groffen, J., Hansen, P.F., de Klein, A., Bartram, C.R., and Grosveld, G. (1983). Localization of t he c-abl oncogene adjacent to a translocation break point in chronic myelocytic leukaemia. Nature 306, 239-242. Jones, J.K., and Thompson, E.M. (2020). Allosteric Inhibit ion of ABL Kinases: Therapeutic Potential in Cancer. M olecular cancer t herapeutics 19, 1763-1769. Kohn, K.W. (1999). Molecular interaction map of t he mammalian cell cycle control and DNA repa ir systems. Mol Biol Cell 10, 2703-2734. Lugo, T.G., Pendergast, A. M., M uller, A.J., and Witte, O.N. (1990). Tyrosine kinase activit y and transformation pot ency of bcr-abl oncogene products. Science 247, 1079-1082. Luna, A., Elloumi, F., Varma, S., Wa ng, Y., Rajapakse, V.N., Aladjem, M.I., Robert, J., Sa nder, C., Pommier, Y., and Reinhold, W .C. (2021). Cell M iner Cross-Database (CellM inerCDB) version 1.2: Exploration of patient -derived ca ncer cell line pharmacogenomics. Nucleic acids resea rch 49, D1083-D1093. M ertins, S.D., Scud iero, D.A., Hollingshead, M.G., Divelbiss, R.D., Jr., Alley, M .C., Monks, A., Covell, D.G., Hite, K.M., Sa lomon, D.S., and Niederhuber, J.E. (2013). A small molecule (pluripoti n) as a tool for st udying ca ncer st em cell biology: proof of concept . PloS one 8, e57099. M ian, A.A., Haberbosch, I., Khamaisie, H., Agbarya, A., Pietsch, L., Eshel, E., Najib, D., Chiriches, C., Ottmann, O.G., Hantschel, 0 ., et al. (2021). Crizot inib acts as ABLl inhibitor combining ATP-binding wit h allosteric inhibition and is active against native BCR-ABLl and its resist ance and compound mutants BCR-ABLl (T315I) and BCR-ABLl (T315I-E255K). Ann Hemat ol 100, 2023-2029. Nowell, P.C., and Hungerford, D.A. (1960). A minute chromosome in human chronic granulocytic leukemia. Science 132, 1497. O'Hare, T., Shakespeare, W.C., Zhu, X., Eide, C.A., Rivera, V.M ., Wang, F., Adrian, LT., Zhou, T., Huang, W.S., Xu, Q ., et al. (2009). AP24534, a pan-BCR-ABL inhibitor for chronic myeloid 297 K. W. Kohn Drugs Against cancer CHAPTER 14 leukemia, potently inhibits the T315I mutant and overcomes mutation-based resist ance. Cancer cell 16, 401-412. Oka be, S., Tauchi, T., Tanaka, Y., Sakuta, J., and Ohyashiki, K. {2015). Ant i-leukemic activity of axitinib against cells harboring t he BCR-ABl T315I point mutat ion. J Hematol Oncol 8, 97. Pat el, A.B., O'Ha re, T., and Deininger, M.W. {2017). Mechanisms of Resist ance to ABl Kinase Inhibition in Chronic Myeloid leukemia and the Development of Next Generation ABl Kinase Inhibit ors. Hematol Oncol Clin North Am 31, 589-612. Rowley, J.D. {1973). letter: A new consist ent chromosomal abnormality in chronic myelogenous leukaemia identified by quinacrine fluorescence and Giemsa staining. Nature 243, 290- 293. Waheed, A., Purvey, S., and Saif, M.W. {2018). Masitinib in t reatment of pancreatic cancer. Expert opinion on pharmacotherapy 19, 759-764. Wang, J., and Pendergast, A.M . {2015). The Emerging Role of ABl Kinases in Solid Tumors. Trends in cancer 1, 110-123. Wang, J.Y. {2014). The capable ABl: what is its biological function? Molecular and cellular biology 34, 1188-1197. Yaghmaie, M ., and Yeung, C.C. {2019). Molecular Mechanisms of Resist ance to Tyrosine Kinase Inhibitors. Curr Hematol Malig Rep 14, 395-404. 298 K. W. Kohn Drugs Against cancer CHAPTER 15 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@oib gov CHAPTER 15 The Oncogene Discovery Story. Introduction The previous chapter was about the ABL oncogene, one of the first cancer genes to be discovered. The oncogene discovery story however converged from several different directions. Pathologists had puzzled about the origin of cancer for more than a century -- until an answer emerged from out of the genetic shadows. There were genes known to predispose to cancers in families - yes, but how did they work? An answer to that old complicated and controversial question came from a remarkable discovery - the discovery of "oncogenes", literally "cancer genes," genes associated with cancer causation. As in the case of ABL, an oncogene was often found to be a mutated version of a normal gene, a "proto-oncogene," that had become overactive and consequently pushed the cell to multiply without control. In the previous chapter, we saw an example in the ABL gene that was driven to become overactive by a gene from another chromosome, the gene coming to lie next to the ABL gene as a consequence of a recombination event between the two chromosomes. Another way that cancer may be driven was found to be by inhibition of a "tumor suppressor" gene that normally held cell division in check. The most famous tumor suppressor was TP53, whose story will be the subject of a later chapter. The full story of RAS oncogenes, one of the most important class of oncogenes, will be told in Chapter 18. Here I will begin the story with how the concept of "oncogene" actually emerged in the context of the first discovered oncogene, RAS. 299 K. W. Kohn Drugs Against cancer CHAPTER 15 Here is an overview of the various ways oncogenes can arise, as revealed by long and weary studies: A normal gene, a proto-oncogene, can become an oncogene by way of (1) a mutation in the gene; (2) reduced methylation of the gene's promoter region (methylation normally suppresses the gene's activity); (3) increase in the number of copies (amplification) of the gene in the cell, due to an increase in the number of chromosomes, or in chromosome sections (as in homogeneously staining regions, Figure 5.7 in Chapter 5), that have many copies of the gene; or (4) a recombination event where pieces of two different chromosomes become stuck together in such a manner that an activator region in one chromosome piece abnormally activates a proto-oncogene in the other chromosome piece, thereby making that gene an oncogene. An example, as already noted, was the story of the BCR-ABL gene recombination that caused chronic myelogenous leukemia (Chapter 14). (5) A proto-oncogene can become an oncogene when it becomes overactive due to damage or deletion of a tumor suppressor gene that normally limits the activity of the oncogene. These are the most frequent among the many ways that a normal gene, a proto-oncogene, can become an oncogene. In short, anything that causes a gene to send too many "divide, p/ease!"messages to the nucleus causes that gene to become an oncogene. How the first oncogenes were discovered. The modern story of cancer biology began with the discovery that there were such things as oncogenes. Three separate stories came together in the discovery of the first oncogenes: the RAS genes. Three different paths from surprisingly different sources converged in this seminal discovery: (1) mouse leukemia viruses; (2) mutant fruit fly eyes; and (3) gene transfer in human cells. Each of those stories is remarkable in its own way. Each of them came from a different experimental and conceptual background, and their profound relevance to human cancer led to surprising and dramatic changes in cancer cause and treatment concepts. One of the most astonishing discoveries came from the studies of mutations in eyes of fruit flies: who would have guessed that those studies (which might have qualified for Senator Proxmire's "Golden Fleece Award") would lead to discovery of a major human oncogene? (Senator William Proxmire gave 168 or those dubious awards from 1975 to 1988 for projects that he considered to be a gross waste of taxpayers' money.). The remarkable paths from mutant fruit fly eyes and from mouse leukemia viruses, however, will be told in the chapter about the RAS oncogenes (Chapter 18). This chapter begins the story of how the oncogene concept came to be discovered. Discovery of how to transfer genes from one human cell to another. It used to seem incredible that it would some-day become possible to transfer genes from one human cell to another. Gene transfer, both natural and experimenter-induced, had been well-known in bacteria and bacterial viruses (bacteriophages) for many years, but attempts to produce it in human or other mammalian cells all failed, until --- 300 K. W. Kohn Drugs Against cancer CHAPTER 15 In 1962, a paper appeared that reported that feat (Szybalska and Szybalski, 1962). It seemed astounding the first time I heard of it at a conference in 1961. Waclaw Szybalski and his wife, Elizabeth Hunter Szybalska (Figure 15.1), reported that they had transferred gene DNA extracted from one culture of human cells into the genome of a different culture of human cells. They succeeded in transmitting a gene that coded for an enzyme in the donor cells to recipient cells that were deficient in that function. Although this phenomenon of gene transfer was well known in bacteria, it was a tour-de-force to demonstrate it in mammalian cells (Szybalska and Szybalski, 1962). A few years after that report, the literature had still remained silent: there were no further reports to confirm that astonishing result I asked Waclaw about it at a conference. He replied (in effect), "Well, it was a difficult experiment and hard to get reproducible data." Years later, it turned out that the difficulty was a surprising detail that caused the method they used to fail much of the time. When researchers prepared DNA from donor cells for gene transfer, the solutions often became cloudy, because some of the DNA precipitated. The investigators did not like those cloudy solutions and either clarified them or discarded them. It came as a big surprise when researchers eventually noted that the more cloudy the DNA solution, the better the gene transfer worked. When the method was optimized to make the most effective DNA precipitate for uptake into cells, the method became routinely successful. It was the tiny DNA particles of the precipitate that were taken up by the cells, allowing the DNA to enter the recipient cell's genome. It was that long-time prejudice that chemists had against cloudy solution that actually impeded an important discovery. I am reminded now of an NIH lecture by Nobel Prize winner Albert Szent-Gyorgyi (1 think it was in 1958), who mentioned his discovery of the chemical basis of muscle contraction. The discovery, he said, had been hampered for many years by scientists' prejudice against cloudy solution: when researchers added calcium (which was known to initiate muscular contraction) to an extract from muscle, the solution became murky and the scientists would throw it away in disgust. "But," said Szent-Gyorgyi (in effect), "when I saw that precipitation in 1938, I imagined a muscle contracting," which led him to discover the chemical key to muscular contraction: the actin and myosin proteins, and their calcium- induced binding, which is what causes muscles to contract, but in solutions of actin plus myosin produced those precipitates. It took 16 years after the Szybalska and Szybalski report before DNA transfer between human cells became routinely successful. Michael Wigler, who was then still a graduate student at Columbia University, used a calcium phosphate-DNA co-precipitation technique that worked well in transferring DNA into recipient cells and into their chromosomes (Wigler et al., 1978). Wigler and his coauthors however seem to have been unaware of the earlier work by Szybalska and Szybalski, which my account here may help preserve in the historical record. After Wigler's report in 1978, many laboratories started using that method to transfer genes from one cell type to another (Malumbres and Barbacid, 2003), which opened bright new vistas for research, including the direct identification of oncogenes. 301 K. W. Kohn Drugs Against cancer CHAPTER 15 Figure 15.1. Waclaw Szybalski (1921-2020) and Elizabeth Hunter Szybalska (1927-2015), the husband-and-wife team who for the first time accomplished gene transfer between human cells (Szybalska and Szybalski, 1962). Discovery of oncogenes by gene transfers. In 1978, about the same time that Wigler's method of gene DNA transfer was developed. Robert A. Weinberg (Figure 15.2), at the time a new faculty member in MIT's Biology Department, had a bold vision (one of many bold visions during his career). He thought that DNA extracted from cancer cells could be transferred into non-cancer cells and cause them to become cancer-like (Shih et al., 1979). He thought that this seemingly far-out idea might succeed using the calcium phosphate DNA co-precipitation technique to transfer DNA and its specific function from one cell to another. The experiment was designed to test whether a gene from a malignant cell could cause cancer in a non-cancer cell. If the test succeeded, it could lead to a major advance in understanding cancer. However, the experiment was difficult to carry out and was thought to have a low chance of success. No one in Weinberg's lab was willing to undertake it because success seemed unlikely, and his junior doctoral level staff members needed a research success to propel them to their next jobs. I guess his junior researchers thought their mentor's idea was far- fetched. Then, it so happened that a pre-doctoral student Chiaho Shih appeared, looking for a new research project. Shih undertook to carry out Weinberg's idea, and within a few months succeeded in this world-famous accomplishment (Shih et al., 1979; Weinberg, 2011). The gene transfer from cancer cells to non-cancer cells caused the latter to grow excessively on the surface of a glass dish, producing areas of piled up cells (Figure 15.3). Weinberg was astonished that the experiment actually worked and that a tiny amount DNA from cancer cells could induce cancer-like behavior in non-cancerous cells. It seems 302 K. W. Kohn Drugs Against cancer CHAPTER 15 that Weinberg himself may have thought his own idea to be far-fetched and was actually surprised by the astonishing result. After that ground-breaking success, there was no difficulty finding young researchers enthusiastic about carrying these studies forward, which led to their finding the notorious KRAS oncogene in a human colon carcinoma (McCoy et al., 1983). The KRAS oncogene will be a protagonist in Chapter 18. Weinberg and his colleagues also showed that DNA from cells that were made cancerous by treatment with a chemical carcinogen could be transferred to non-cancer cells and make them cancer-like. It seemed, therefore, that carcinogens caused changes in some gene or genes that made cells cancerous. In modern language: the discovery was that carcinogens could mutate certain normal genes and cause them to make the cell cancer- like: carcinogens seemed capable of converting proto-oncogenes to oncogenes. Weinberg tells how the oncogene work got started (Weinberg, 2011). He had returned to MIT, where he had obtained his undergraduate and doctoral degrees -- in part because he would learn much there by working with Nobel Prize winner David Baltimore. Weinberg tells the story of how the critical experiments were carried out by graduate student, Chia- Ho Shih, who came to his lab and agreed to undertake this challenging work that Weinberg's other junior scientists were reluctant to do, because they felt it was not likely to be successful. When these high-stakes experiments actually seemed to work, they aimed for definitive proof by carrying out the experiments in double-blind fashion that, in Weinberg's words, "yielded unequivocal evidence of transforming sequences in the DNA of chemically transformed cells and later in the DNA of a variety of human tumor cells - results that turned out to be most consequential in my own research career." (When I use the word "proof' in these writings, I am aware that science cannot prove anything beyond doubt. "Proof' comes from Latin probare or proba to test. Thus "proof' can be taken to mean crucial confirmatory evidence.) Figure 15.2. Robert A Weinberg (1942- ). discoverer of oncogenes. (Picture from MIT website.) 303 K. W. Kohn Drugs Against cancer CHAPTER 15 Message of that early success flew rapidly across the Charles River from MIT to the Sidney Farber Cancer Institute of Harvard Medical School, where the new oncogene methodology was enthusiastically taken up in the laboratory of Geoffery M. Cooper (Cooper, 1982; Cooper et al., 1980; Krontiris and Cooper, 1981). The methodology was applied to carcinogen-induced mouse cancers, human cancers, and cancer cell lines. The researchers purified DNA from cancers or cancer cell lines and applied it as a DNA-calcium phosphate co-precipitate to recipient non-cancer cells growing on the surface of a plate. Normal cells limited their growth to a single layer of cells. The DNA from cancer cells, as already mentioned, caused some of the recipient cells to multiply excessively (Figure 15.3). As evidence that the foci of piled-up cells were cancer-like, the researchers showed that foci were produced by DNA from cells that had been exposed to carcinogens and not from unexposed cells. Moreover, DNA from cells of a focus was highly effective in producing foci in cultures of non-cancer cells. The cancerous nature of the foci cells was later confirmed by showing that cells from those foci produced tumors when injected into mice. That led cancer researchers to think that at least some and perhaps most cancers arose from one or more mutations in the cell's genome. Geoffrey Cooper and his colleagues soon found "transforming" genes in several animal and human cancers (Cooper, 1982). ("Transforming gene" meant that DNA containing the gene produced foci of over-growing cells, and that cells from such foci caused cancer in animals.) They found that some of the transforming genes in the recipient cells had DNA sequences resembling the RAS genes of Harvey and Kirsten sarcoma viruses, as will be related in the Chapter 18 (Cooper, 1984; Cooper and Lane, 1984) (Der et al., 1982). Further studies in several laboratories identified three RAS genes (HRAS, KRAS, and NRAS) in normal and cancer cells of humans and rodents. They found altered or mutated forms of those genes in many different types of cancer and suspected that the genes contributed to the cause of the cancers. RAS genes were found in all vertebrate cells examined, and also in yeast, highlighting the importance of these genes in the control of cell proliferation of very different creatures. Studies in many laboratories over many years disclosed a large number of different oncogenes responsible for many kinds of human cancers. Those cancer-causing oncogenes evidently were mutated versions of the normal RAS genes. 304 K. W. Kohn Drugs Against cancer CHAPTER 15 Figure 15.3. Example of an area of cell pile-up caused by excessive multiplication of cells that had received DNA from cancer cells. Normal cells on the surface of a dish grew to form a single layer of cells. When DNA from cancer cells was added to the plate, some of the cells took up the cancer DNA, which induced them to overgrow and pile up in multi- cell layers. One such pile up focus is shown here as an area of high cell density. Cells from such foci produced cancers in animals, confirming that the DNA had caused the cells to become cancerous (Krontiris and Cooper, 1981). References Cooper, G.M. (1982). Cellular transforming genes. Science 217, 801-806. Cooper, G.M. (1984). The 1984 Walter Hubert lecture. Activation of transforming genes in neoplasms. British journal of cancer 50, 137-142. Cooper, G.M., and Lane, M.A. (1984). Cellular transforming genes and oncogenesis. Biochimica et biophysica acta 738, 9-20. Cooper, G.M., Okenquist, S., and Silverman, L. (1980). Transforming activity of DNA of chemically transformed and normal cells. Nature 284, 418-421. Der, C.J., Krontiris, T.G., and Cooper, G.M. (1982). Transforming genes of human bladder and lung carcinoma cell lines are homologous to the ras genes of Harvey and Kirsten sarcoma viruses. Proceedings of the National Academy of Sciences of the United States of America 79, 3637-3640. Krontiris, T.G., and Cooper, G.M. (1981). Transforming activity of human tumor DNAs. Proceedings of the National Academy of Sciences of the United States of America 78, 1181-1184. Malumbres, M., and Barbacid, M. (2003). RAS oncogenes: the first 30 years. Nature reviews Cancer 3, 459-465. McCoy, M.S., Toole, J.J., Cunningham, J.M., Chang, E.H., Lowy, D.R., and Weinberg. R.A. (1983). Characterization of a human colon/lung carcinoma oncogene. Nature 302, 79- 81. Shih, C., Shilo, B.Z., Goldfarb, M.P., Dannenberg, A., and Weinberg. RA (1979). Passage of phenotypes of chemically transformed cells via transfection of DNA and chromatin. 305 K. W. Kohn Drugs Against cancer CHAPTER 15 Proceedings of the National Academy of Sciences of the United States of America 76, 5714-5718. Szybalska, E.H., and Szybalski, W. (1962). Genetics of human cess line. IV. DNA-mediated heritable transformation of a biochemical trait Proceedings of the National Academy of Sciences of the United States of America 48, 2026-2034. Weinberg, R.A. (201 1). Hunting the elusive oncogene: a stroke of good luck. Nat Cell Biol 13, 876. Wigler, M., Pellicer, A., Silverstein, S., and Axel, R (1978). Biochemical transfer of single- copy eucaryotic genes using total cellular DNA as donor. Cell 14, 725-731. 306 K. W. Kohn Drugs Against cancer CHAPTER 16 Chapt,er- 16. Th~ Oncogest~ Addlt1ion Sro,y ZZ07Z3%3 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@n ih.gov CHAPTER 16 The Oncogene Addiction Story: a conceptual basis for cancer therapy. The idea that cancer cells could become addicted to an oncogene came to mind from observations far removed from anything that cancer researchers ordinarily thought about. It came from observations by Michael Yarmolinsky of bacteria that became addicted to viruses infecting them. Curing the infection caused the bacteria to die, because the virus generated both a toxin and an antidote. When the virus was eliminated, the toxin survived longer than the antidote and the bacterial cell died (Lehnherr et al., 1993). It was as if the bacteria had become addicted to their infection. Yarmolinsky (Figure 16.1) may have been the first to apply the epithet "addicted" to a living cell. As explained in the preceding chapter, cancer is often promoted by abnormal activation of a gene that pushes the cells to keep dividing without the normal constraints. The normal version of the gene (proto-oncogene) that stimulates cell division in a controlled fashion sometimes becomes mutated or otherwise altered genetically in a manner that stimulates uncontrolled cell division. The altered gene thus becomes an oncogene: a gene that pushes cells to become cancerous. A rather far-out notion came to mind that an oncogene-addicted cancer cell might be thought of as analogous to a virus-addicted bacterium. Inhibiting the oncogene in a cancer cell addicted to it might then cause the cell to die. Indeed, drugs that inhibited an oncogene in cancer cells sometimes, not only stopped the cells from dividing, but caused the cancer cells to die. It seemed that cancer cells sometimes became dependent on the function an oncogene. As the cell adapted to being continually stimulated by an uncontrolled oncogene, other changes would develop that the cancer cell could tolerate only while the oncogene is functioning. 307 K. W. Kohn Drugs Against cancer CHAPTER 16 Yarmolinsky and I had met during a course on mathematical probability at Harvard College, and we were now working in the same cancer research building at NIH. He told me about his remarkable bacteria-virus addiction observations when I stopped at his lab one day to pick him up for our frequent lunch breaks together. Some months later, I attended a lecture by Bernard Weinstein. He and I had come to NCI at the same time and were in the same group of Clinical Associates. At the end of the lecture, I told him about Yarmolinsky's observations and suggested that cancer cells might in somewhat analogous fashion become addicted to their oncogenes. Weinstein then assembled evidence from a variety of previous reports to crystallize the concept of oncogene addiction (Weinstein, 2002) (Weinstein and Joe, 2008) (Figure 16.2). The virus infecting Yarmolinsky's bacteria produced both a toxin and an antidote. If you cured the infection, the antidote disappeared first and the toxin then killed the host cell. Somewhat analogously, deleterious effects of an abnormally expressed cancer gene, an "oncogene", could be compensated by altered activity of other genes. The cancer cells that had those altered gene activities could survive the over-activity of the oncogene and thrive - provided that the oncogene continued to be overactive. The cell would then be dependent on the high activity of the oncogene. Block the activity of the oncogene, and the cell would die. Why? Because the gene activity changes that are protective when the oncogene is over- active are lethal when the oncogene is blocked. That was the original form of the oncogene addiction concept. i\lOR,\ Nl)l Figure 16.1. Michael B. Yarmolinsky (1931- ), discoverer of gene addiction in bacteria. 308 K. W. Kohn Drugs Against cancer CHAPTER 16 Figure 16.2. I. Bernard Weinstein (1930-2008), who crystallizes the concept of oncogene addiction. Weinstein's reasoning began by noting a well-established pattern in the way most cancers become malignant: cancer cells develop numerous abnormalities in the amounts of a variety genes that they express, the number of abnormalities increasing as the cells become increasingly malignant (Weinstein, 2002) (Weinstein and Joe, 2008). Many genes become overactive. The essential fact, he reasoned, was that, despite the large number changes, a cancer can sometimes be suppressed by blocking just one of the overactive genes. He argued that cancer cells often become dependent on ("addicted to") a gene (an "oncogene") that drives the cancer process, and that suppressing that oncogene's function with a suitable drug could suppress the cancer. This has become known as "oncogene addiction," analogous to Yarmolinsky's "viral addiction" observation. In many experiments, switching on an oncogene in a mouse, for example by genetic engineering, caused a malignant tumor to appear. Then, treating the mouse with a drug that suppressed the oncogene caused the tumor to regress. Most of the tumor cells die, but some "differentiate" and assume the guise of normal cells of the kind that initially gave rise to the tumor. In some cases, particularly in tumors of lymphocytes, differentiation of the tumor cells into seemingly normal lymphocytes was often the main thing that happened. Since lymphocytes normally have a short lifespan, the drug-treated malignant cells that differentiated into normal-seeming lymphocytes tended to be eliminated. Observations indicating oncogene addiction. An early experiment indicating oncogene addiction utilized a genetically engineered chronic myelogenous leukemia (CML), a disease discussed in Chapter 14, where a BCR-ABL translocation drives the overexpression oftheABL oncogene. In 1999, Claudia Huettner, Daniel Tenen and their colleagues at the Harvard Medical School (Huettner et al., 2000) showed that leukemic cells could in fact become addicted to BCR-ABL. They induced 309 K. W. Kohn Drugs Against cancer CHAPTER 16 leukemia in mice by inserting a BCR-ABL genetic element into their genome in a manner that allowed the researchers to control whether or not the inserted BCR-ABL was active. When they suppressed the BCR-ABL, the leukemia went away. Moreover, when they caused BCR-ABL activity to resume, the leukemia came back, only to disappear again when they again suppressed the BCR-ABL. That demonstrated that the leukemic cells needed the BCR- ABL function in order to survive and proliferate. The leukemic cells had become addicted to the effect that BCR-ABL had on them. Drugs, such as Gleevec, that blocked the overexpressed ABL tyrosine kinase in CML patients, suppressed the disease to such an extent that patients often seemed cured. The malignant CML cells had become addicted to the oncogene. When the drug suppressed the ABL tyrosine kinase that was produced by the overactive ABL gene, the malignant cells reverted to their normally brief lifespan and died by apoptosis (programmed cell death; cell suicide). However, the disease sometimes recurred because the apoptosis required TPS3 whose gene is frequently mutated in cancers. The mutation often inactivated the TP53 gene, and the CML cells could no longer die by apoptosis. The disease, after remaining dormant for years, could then resume its malignant course impervious to the drug treatment (Sawyers, 2003). Another early experiment pointing to oncogene addiction involved the MYC oncogene. In 1999, Dean Felsher and Michael Bishop at the University of California San Francisco inserted into mice a MYC gene that they (the researchers) could control at will. When MYC was active at a high level, it behaved like an oncogene, and the mice developed leukemia. However, when the activity of the gene was stopped, the leukemia went away due to arrest and apoptosis of the leukemia cells. As well as demonstrating oncogene addiction, the experiment showed that the appearance and disappearance of a malignancy could depend on the activity of a single gene (Felsher and Bishop, 1999). A further example of oncogene addiction was the HER2 oncogene that is overactive in some types of breast cancer (discussed in Chapter 17). The normal HER2 gene (the proto- oncogene) became overactive due to mutation or gene amplification, which was what made the gene an oncogene. Inhibitors of the overactive HERZ caused those types of breast cancer to shrink That is part of the EGFR oncogene story related in the next chapter. Many additional examples of oncogene addiction were discovered, where a cancer driven by an oncogene is largely, albeit temporarily, eliminated by a drug that suppresses the oncogene's activity (Weinstein and Joe, 2008). The RAS and MYC oncogenes were important examples. 310 K. W. Kohn Drugs Against cancer CHAPTER 16 Synthetic lethality, a log ic-based therapy. An ideal cancer therapy would be a drug that kills cancer cells but spares the cells of normal tissues. In 2005, Bill Kaelin at the Howard Hughes Institute and Harvard University described situations where that ideal might be achieved (Kaelin, 2005). Called "synthetic lethality," it is where two genes, let's call them A and B, determine whether a cell lives or dies. If both A and Bare inactivated, then the cell dies. However, if either A or B or both are active, then the cell survives. How does that relate to cancer therapy? Well, it happens, first of all, when the cancer has a mutation that inactivates a gene - like the above gene A or B. Let's say that the mutation inactivates the cancer's gene A. Then a drug that inactivates gene B would kill the cancer cells, while normal cells would be protected by their non-mutated gene A. That is like the cancer cells becoming addicted to gene B, because - already having a defective A -- they can't survive without B. Might it actually be a major way that oncogene addiction works? If a cancer had a defective (mutated) A, then the cancer may need a high expression of B to survive. If B is then inhibited by a drug, the cancer cell dies - as if it were addicted to B. Although one might think that instances of therapeutically useful synthetic lethality would not be difficult to find, great effort has so far led to only one case. That one instance, based on combined inhibition of the BRCA and PARP genes, proved very effective and is a topic in a later chapter. References Felsher, D.W., and Bishop, J.M. (1999). Reversible tumorigenesis by MYC in hematopoietic lineages. Mo) Cell 4, 199-207. Huettner, C.S., Zhang, P., Van Etten, RA., and Tenen, D.G. (2000). Reversibility of acute B- cell leukaemia induced by BCR-ABLl. Nat Genet 24, 57-60. Kaelin, W.G., Jr. (2005). The concept of synthetic lethality in the context of anticancer therapy. Nature reviews Cancer 5, 689-698. Lehnherr, H., Maguin, E., Jafri, S., and Yarmolinsky, M.B. (1993). Plasmid addiction genes of bacteriophage Pl: doc, which causes cell death on curing of prophage, and phd, which prevents host death when prophage is retained. Journal of molecular biology 233, 414- 428. Sawyers, C.L. (2003). Opportunities and challenges in the development of kinase inhibitor therapy for cancer. Genes Dev 17, 2998-3010. Weinstein, J.B. (2002). Cancer. Addiction to oncogenes--the Achilles heal of cancer. Science 297, 63-64. Weinstein, J.B., and Joe, A. (2008). Oncogene addiction. Cancer research 68, 3077-3080; discussion 3 080. 311 K. W. Kohn Drugs Against cancer CHAPTER 17 Chapt,er- 11. Th~ EGFR o,xoge-M tlJJ,Y ZZ07Uds3 Drugs Against Cancer: Stories of discovery and the quest for a cure. Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@nih.gov CHAPTER 17 The EGFR Oncogene story. Prolog: the ErbB oncogene story In the 1970's, researchers had been investigating certain viruses common in birds. These "avian erythroblastosis retroviruses" had an RNA genome and a "reverse transcriptase" that copied its RNA into DNA, which then became incorporated into the host's DNA, from which it was recopied into viral RNA for the next cycle of virus production. When injected into susceptible chickens, the viruses caused an overproduction of red blood cells (erythroblastosis), but unexpectedly sometimes also produced cancers (sarcomas). Tracking down the cause of the cancers, investigators found, in the RNA of the cancer- producing retroviruses, a nucleotide sequence that they thought to be the culprit and surmised it to be an "oncogene" - - a gene that caused the cancers. They dubbed the oncogene erb for erythroblastosis; there were two types: ErbA and ErbB. Amazingly, genes with nucleotide sequence similarities to ErbB were found in the genomes of vertebrate animals from fish to humans; moreover, those oncogenic sequences in the ErbB gene resembled sequences found in the human epidermal growth factor receptor gene (EGFR) (Downward et al., 1984; Saule et al., 1981). Incredibly, the erb oncogene in an avian retrovirus had nucleotide sequence similarity to a normal human gene! It was the epidermal growth factor receptor (EGFR) gene, which was found to become an oncogene when mutated or amplified -- which is the topic of this Chapter. That summarizes a complicated and confusing history of experiments. However, it led to a dual terminology that one had to become accustomed to, because EGFR was discovered by a totally independent route -- which is the topic of the next sections. 312 K. W. Kohn Drugs Against cancer CHAPTER 17 For future reference, the terminology for the 4 members of the EGFR family is as follows: ErbBl=EGFR=HERl; ErbB2=HER2; ErbB3=HER3; ErbB4=HER4. (HER stands for "human epidermal growth factor receptor.") The HER terminology is specific for the human genes, whereas the ErbB and EGFR terms are used more generally. Receptor tyrosine kinases (RTK's): how signals are transmitted from outside to inside the cell. I'll begin with an overview of how RTK's work by transmitting signals from outside to inside the cell. Chapter 14 told how specific inhibitors of the ABL tyrosine kinase provided effective treatment for chronic myelogenous leukemia (CML) patients. That story was a striking example of how a specific molecular abnormality in a certain type of cancer led to a cancer cure. There are many tyrosine kinases that transmit signals to the cell nucleus. There are two general types: (1) receptor tyrosine kinases (RTK's) that transmit signals from outside the cell -- the EGFR family are among these; and (2) non-receptor kinases that may roam the cytoplasm and enter the nucleus to affect gene expression; ABL is one of these. To transmit signals from outside to inside the cell, the receptor part of the RTK molecule sticks out of the cell, ready to bind a signaling molecule, such as a "growth factor", that may be drifting about in the exterior. Most cell types will grow and divide only in the presence of appropriate growth factor molecules. The receptor part of the RTK molecule that binds the growth factor outside the cell connects to a narrow piece that passes through the plasma membrane and, in turn, connects to a large part that is inside the cell. That is the business part: triggered by the extracellular domain of the RTK, the intracellular domain of the RTK engages in a complicated set of interactions inside the cell that stimulate the growth and division of the cell. I go on now to tell the story of how the first RTK was discovered. The story begins however with the discovery of a signaling molecule, a small protein called epidermal growth factor (EGF) that was later found to bind to the exterior or extracellular part of a membrane protein that became known as epidermal growth factor receptor (EGFR). Discovery of epidermal growth factor (EGF). One the most important developments in cancer biology and therapeutics was the discovery of the growth factors that were found to bind to the extracellular domain of the epidermal growth factor receptor. The story began in the early 1960's, when Stanley Cohen (Figure 17.1) at Vanderbilt University in Nashville, Tennessee isolated a small protein that 313 K. W. Kohn Drugs Against cancer CHAPTER 17 stimulated the proliferation of skin cells (Cohen, 1965) (Figure 17.2). That small protein was later to be called epidermal growth factor, EGF. Figure 17.1. Stanley Cohen (1922-2020) was awarded the Nobel Prize in Physiology and Medicine, along with Rita Levi-Montalcini, in 1986 for the discovery of epidermal growth factor and the purification of nerve growth factor. A son of Jewish immigrants, Cohen was born in Brooklyn, New York and received his bachelor's degree from Brooklyn College in 1943. He carried out his major research at Vanderbilt University in Nashville, Tennessee from 1959 until his retirement in 1999. He contributed much to the story told in the early part of this chapter. proliferating } skin cells 314 K. W. Kohn Drugs Against cancer CHAPTER 17 Figure 17.2. Cross-sections of skin, showing one of the first demonstrations of the growth- promoting effect of a small protein that was later named epidermal growth factor (EGF). This experiment, carried out by Stanley Cohen in 1965, shows how EGF causes skin cells to proliferate (bottom), compared with the cells of skin cultured in the absence of EGF (top). The skin was dissected from chick embryos and grown in a solution containing essential ingredients with (bottom) or without (top) EGF (Cohen, 1965). In the early part of the work, Cohen collaborated with Rita Levi-Montalcini, who had devised a method to grow skin from chick embryos in a culture medium, where it could be studied under well-defined conditions. They shared the Nobel Prize in Physiology and Medicine in 1986 for the discovery of epidermal growth factor and the isolation of nerve growth factor. Here I cannot resist recounting, even though embarrassing to me, my one-time encounter with Rita Levi-Montalcini, as it shows her delightful and indomitable personality (she lived to be 103). We were seated next to each other at dinner at a conference and had just begun to chat, when she asked me right-out whether I knew who she was and about her work. Perhaps more indignant than amused by my nonplussed expression, she took some time before admitting at last that she was in fact the famous discoverer of nerve growth factor -- which was an area of research about which I was at the time unfortunately abysmally ignorant To continue: Stanley Cohen's small protein, later dubbed "epidermal growth factor" (EGF), was first purified from the salivary glands of mice -- Cohen and Carpenter isolated the human version in 1975; the two EGF's, mouse and human, had very similar physical and biological properties -- the same growth-promoting effects on cells (Cohen and Carpenter, 1975). EGF stimulated the proliferation, not only of skin epidermal cells, but also of cells from many epithelial tissues and epithelial cancers. (An epithelium consists of one or more layers of cells that separate the outside from the inside of a tissue. Most cancers develop from epithelial cells.) Chemical s tructure of EGF. The first step in analyzing the chemical structure of EGF was to determine its sequence of amino acids. The molecule was found to consist of a chain of 53 amino acids, with three crosslinks (Figure 17.3). Each crosslink connected two cysteine (CYS) amino acids located at different places in the chain. (Cysteines have a sulfur atom at the end; two sulfur atoms can bind to each other to form a "disulfide" crosslink that connects between different parts of the amino acid chain. This, by the way, can only happen outside the cell, because the "reducing" conditions inside cells keep the sulfur atoms from binding to each other. Growth factors are located outside the cell, where disulfide bonds are stable.) The next step was to determine the 3-dimensional structure. That is important, because the EGF molecule must have the right shape to fit into a receptor site and to exert its effects. 315 K. W. Kohn Drugs Against cancer CHAPTER 17 The structure was determined by x-ray crystallography in 2001 (Lu et al., 2001), and is shown in Figure 17.4. The EGF amino acid chain is not long enough to maintain a 3- dimensional structure all by itself (it wiggles around too much); the three disulfide crosslinks help to keep its shape. Figure 17.3. The amino acid chain structure of epidermal growth factor (EGF), showing the 3 cysteine (CYS) pairs that hold different parts of the chain together (Carpenter and Cohen, 1979). The 3-dimensional structure of the amino acid chain is shown in Figure 17.4. C-term Figure 17.4. The 3-dimensional structure of human epidermal growth factor (EGF) protein (Lu et al., 2001). The shape of the structure is held securely in place by the three Cys-Cys crosslinks (shown in yellow - shown also in Figure 17.3). (The fat green arrows show the direction of the amino acid chain from the N-terminus to the C-terminus.) 316 K. W. Kohn Drugs Against cancer CHAPTER 17 How does EGF trigger a growth signal to be sent from outside to ins ide the cell? A puzzle: EGF was found to transmit a growth signal to the cell nucleus -- but how was the signal transmitted, since EGF is located outside the cell and cannot get in? Somehow, EGF transmits its signal right through the intact cell surface membrane without the EGF molecule itself, or any part of it, going through. The first clue to how that happens came from a finding by Hollenberg and Cuatrecasas, who, in 1973, demonstrated that EGF binds to specific receptors on the cell surface (Hollenberg and Cuatrecasas, 1973). Using radioactively-tagged EGF, Stanley Cohen and his colleagues determined that there was a limited number of such receptor sites on the surface of various types of cells (Carpenter and Cohen, 1979). To give an idea of numbers, there are typically about 70,000 EGF receptor sites on the surface of a cell -- which is not very many, considering how tiny the molecules are, compared to the size of a cell. Figure 17.5 showed that EGF molecules were bound to the surface of cells; to make the molecules visible, a fluorescent chemical group was attached to the EGF, which made is glow under ultraviolet light. The receptors to which EGF bound on the cell surface were later found to be the extracellular part the epidermal growth factor receptor (EGFR), whose story follows. Figure 17.5. Epidermal growth factor (EGF) molecules bound to the cell surface. A fluorescent tag was attached to the EGF molecules to make them glow when viewed with a fluorescence microscope. When mixed with cells, the fluorescence-labeled EGF was seen bound to the surface of the cells, as shown here by the bright edges where the surfaces of adjacent cells meet. This experiment was carried out in Stanley Cohen's laboratory and published in 1978 (Haigler et al., 1978). Discovery of the EGF receptor (EGFR). One of the first clues to the existence of receptors for epidermal growth factor and how the receptor transmits a signal from outside to inside the cell came from experiments 317 K. W. Kohn Drugs Against cancer CHAPTER 17 conducted by Stanley Cohen and his colleagues in 1980 (Cohen et al., 1980). Using isolated cell membranes, they found that epidermal growth factor (EGF) bound to the membranes and caused a large protein in the membranes to become phosphorylated (i.e., EGF caused phosphate groups to be added to a large membrane-associated protein). The large protein turned out to be the receptor for EGF (EGFR). EGFR was estimated to be about 150,000 Dal tons in molecular weight, compared to only about 6,000 Daltons for EGF. The discovery of EGFR was a major breakthrough for cancer therapy. Excessive function of EGFR was later found to send abnormally strong signals that push cells to divide excessively and without control. This was found to be an important contributory cause of about 30% of all cancers (malignant tumors arising from epithelia). Quite remarkably, the cancer cells often became addicted to the high EGFR activity: when EGFR activity was inhibited by means of a drug, the cancer cells tended to die! However, much first had to be discovered about EGFR and how it worked. By 1987, YosefYarden and Joseph Schlessinger together with other researchers purified EGFR and determined that the molecule consists of 3 parts: an extra-cellular domain that binds EGF, a trans-membrane domain, and an intra-cellular domain that has tyrosine kinase activity (Yarden and Schlessinger, 1987a, b). They also showed that the active unit consisted of two EGFR molecules bound together, and that each member of the pair phosphorylated the other; the binding of EGF caused the pairing and stimulated each to phosphorylate the other. The magic of how EGFR trans mits signals from outside to inside the cell. EGFR was found to be one of a great number ofreceptortyrosine kinases (RTKs) that function to pass extracellular signals of various kinds from outside the cell to molecules inside the cell; the signals then pass to intracellular protein molecules to eventually reach the cell nucleus, where the signals affect which genes will be expressed. The RTKs passed nothing physical through the surface membrane of the cell, only information - but how could they do that? The epidermal growth factor receptors (EGFR-family) were the first and most intensively studied receptor tyrosine kinases and were found to have major roles in cancer cause and treatment (Endres et al., 2014; Lemmon et al., 2014). The EGFR family has four members, Erb1, Erb2, Erb3, and Erb4; the human versions are designated HERl, HERZ, HER3, and HER4. The first member to be discovered, EGFR, has the alternative names Erbl and HERl. It seemed magical that binding to something outside the cell caused something to happen inside the cell without any substance moving through the cell surface membrane. The details of how that is accomplished took much time and effort to be revealed, and there is much to tell about how it works and about drugs designed to block those actions - and how those discoveries were made. 318 K. W. Kohn Drugs Against cancer CHAPTER 17 Here is how it works: The binding of an EGF to an EGFR causes two EGFR molecules to bind to each other to form a homodimer ("dimer"= "two-part"; "homo" indicates that the 2 parts are the same). If different members of the EGFR family bind to each other, they form a "heterodimer" ("hetero" meaning "different'). The four EGFR family members can bind to each other in all possible binary combinations to form homodimers and heterodimers. A heterodimer particularly important for human cancer was found to be the EGFR-HER2 pair, more generally known as the ErbB1-ErbB2 pair. The binding of EGF causes two EGFR-family members to come together to form a homo- or hetero-dimer. The dimer formation then causes the two EGFR-family molecules to change shape in such a way that it enables the intracellular parts of the two molecules of the dimer to add phosphate groups to each other at certain of their tyrosine amino acids. This happens because the intracellular part of each EGFR-family molecule has tyrosine kinase enzyme activity. The phosphate groups at specific places on the intracellular domains then bind particular molecules inside the cell (helped by phosphate's negative charge). That leads to several reaction paths, one of the most important being a chain of reactions that proceeds by way of one or another member of the RAS-family (which is the subject of the next chapter). How this amazing signal transmission is accomplished is diagrammed in Figure 17.6- for the case where EGFR (ErbBl) interacts with HERZ (ErbB2). The signaling process is also portrayed as a whimsical story in Textbox 1. The ErbB2/HER2 pairing with ErbBl/EGFR or other ErbB's is especially relevant to cancers of breast and ovary, where the cancer cells produce too much ErbB2/HER2, and where drugs were being developed to inhibit its tyrosine-phosphorylation activities (Fabi et al., 2014). ErbB2/HER2 is unique among the 4 family members in that its extracellular domain lacks the ability to bind any growth factor; it is activated when it binds to growth factor-activated ErbBl /EGFR); the details are described in Textbox 1, wherein I have taken some liberties to lighten the complexities that are diagrammed in Figures 17.6 and 1 7.7 and described in the legends. 319 K. W. Kohn Drugs Against cancer CHAPTER 17 Outside Surface membrane Inside Gefitinib 7 Symbols --I> Stimulation EGFR HER2 --f Inhibition (ErbBl) (ErbB2) , • , Binding • I I I The entity that results Regulators of cell from the binding growth and division Figure 17.6. Molecular interaction map showing how epidermal growth factor (EGF) binds epidermal growth factor receptor EGFR, allowing it to bind HER2, resulting in a growth signal sent into the cell. The steps are as follows: (1] EGF binds EGFR outside the cell; [2, 3] the EGF binding allows EGFR to bind HER2; (4, 5] the EGFR and HER2 then phosphorylate each other (only one phosphorylation of EGFR is shown; actually, many sites on both EGFR and HER2 become phosphorylated; [6] the phosphorylated EGFR-HER2 heterodimer then sends signals to the cell nucleus via complicated steps that are not shown in this diagram. (7) Drugs, such as gefitinib, inhibit the kinase domains of the ErbB's, thereby inhibiting the phosphorylations. The actions of three monoclonal antibodies are shown: [BJ pertuzumab inhibits the ErbB's from binding to each other; (9] cetuximab may prevent the binding of a growth factor; and [10] trastuzumab inhibits multiple functions of the extracellular domain. (A note about the notation: the small, filled circles on the interaction lines represent the product of the interaction. For example, small, filled circles represent EGF-bound EGFR in step (1] and phosphorylated EGFR in step [5]). The molecular interaction map notation is fully described in (Kohn et al., 2006). 320 K. W. Kohn Drugs Against cancer CHAPTER 17 Domain 1 Domain II Ooma.nm Oomainr+J Out ™ In UM Tyrosioo TK kinase domain '( '( '( '( '( Carboxytail '( '( '( '( '( Figure 17.7. Another depiction of how a receptor tyrosine kinase (RTK) transmits a signal across the cell surface membrane, showing some recently discovered details (Sigismund et al., 2018). (Permission needed.) In the case shown, the 2 RTK's are of the same type, forming. for example, an EGFR homodimer. Upon binding of a growth factor, such as EGF, the 4 extracellular domains of the RTK (shown in different colors) rearrange in a manner that allows 2 RTK molecules to bind to each other (via subdomain II). This brings together their tyrosine kinase domains (TK), allowing each of them to phosphorylate several tyrosines (Y's) on the intracellular tail of the other. The resulting phosphorylated tyrosines (encircled P's) then bind intracellular molecules that convey signals for cell division or other cell functions (not shown). In summary up to this point, the vast realm of receptor tyrosine kinases (RTK's) tells cells what to do in response to a wide variety of signals from outside the cell. They are amazingly well-designed molecular machines that transmit information from outside to inside the cell without transmitting anything material. The four members of the epidermal growth factor receptor (EGFR) family are particularly important in cancer. They pair up in all pairs to form homodimers and heterodimers, each combination having its own characteristics and functions. HERZ (the human version of the ErbB2 gene) was found to be an important driver of some breast cancers. I selected the EGFR-ErbB2 pair for illustration in Figure 17.6, because, aside from being medically important, it gives a simpler diagram. The diagram uses just four symbols, which are defined in the Figure: stimulation, inhibition, binding, and entity that is the product of a binding (such as a dimer or a phosphorylated molecule). To summarize how it works (also explained in the Figure 17.6 legend): The part that is outside (the receptor domain) binds a growth factor molecule -- small protein molecules floating around between the cells. For example, EGFR can bind 321 K. W. Kohn Drugs Against cancer CHAPTER 17 epidermal growth factor (EGF) [1 ] (the numbers in brackets correspond to the numbers next to the interactions in Figure 17.6). There are several other growth factors that EGFR can bind, but EGF is the most common. After binding EGF, EGFR can go on to bind HERZ [Z, 3) (or another member of the EGFR-family). HERZ is the only member of the EGFR-family that lacks the ability to bind growth factors from outside of the cell. However, when bound to EGFR, HERZ phosphorylates and activates the EGFR to which it is bound. When EGFR and HERZ are bound together, the shapes of the two molecules change in a way that brings their intra-cellular domains together, which allows the tyrosine kinase domain of one member of the pair to add phosphate groups to the intracellular part of the other member of the pair (4, 5]. (Although Figure 17.6 shows only one phosphate added to EGFR, several phosphates actually are added to both EGFR and HERZ.) The anticancer drug, gefitinib, inhibits these phosphorylations [7]. The phosphorylations require ATP, and each kinase domain has a cleft where the ATP binds to do its work. It is in this cleft where gefitinib binds and prevents ATP from coming in. That is how gefitinib inhibits the EGFR receptors. The phosphorylated sites on EGFR and HERZ then bind and activate a host of different molecules in the cytoplasm that convey a cascade of signals into the depths of the cell to get it ready to divide [6]. Even though these down-stream signals are conveyed by many different molecules interacting in a complex network, the signals are transmitted remarkably quickly, mostly within a fraction of a minute (Reddy et al., 2016). Anti-cancer drugs such as gefitinib (also known as lressa) inhibit the tyrosine kinase activities, so that the phosphorylation process is blocked [7]. Also depicted are the actions of three monoclonal antibodies that were later developed as promising anti-cancer drugs [BJ, (9), and (10). The elegant beauty of how receptor tyrosine kinases work their magic is described also, and in livelier fashion, in Textboxl. Textbox 1 A Tale of Two ErbB's. ErbBl (another name for EGFR) is floating within the membrane at the surface of a cell. He is hoping for a signal that he could send to his people down in the innards of cell to tell them it's ok to get the cell ready to divide. He has a head that sticks out from the cell surface, a thin neck that passes through the membrane and connects to a big body he has inside the cell. The signal he is waiting for is brought by a small molecule called EGF that floats around outside the cell. Various tissue cells make EGF, but only in small amounts, so that cells having EGFR-family receptors on their surface will divide only occasionally. If he is lucky enough to catch one of these rare EGF's, he binds it tightly [1] (the numbers in brackets refer to reaction steps in Figure 17.6). This causes his head to change shape, which makes him ready for an encounter with one of his potential mates, such as ErbB2/HER2. Now ErbB2 is very much like ErbBl, except that her head is smaller, which actually makes her smarter, because now she doesn't have to bother with any signal 322 K. W. Kohn Drugs Against cancer CHAPTER 17 molecule from the outside. She just looks for an ErbBl that already has an EGF stuck to his head and therefore is ready to mate without further ado [Z, 3). The rest of the story is pretty dull. The two of them phosphorylate each other (4, 5) and create lots of children in the form of information that they spill into the depths of cell [6). How that information is processed inside the cell is rather complicated, and only those who absolutely need those details bother learning them (Roskoski, 2016). How do receptor tyrosine kinases (RTKs) stimulate cells to become cancerous, and could they be targeted for therapy? To manage the great variety of signals arriving from outside, a cell has many types of receptor tyrosine kinases (RTKs) on its surface, each of which activates molecules that transmit a signal from outside to inside the cell. As of 2016, there were 58 known receptor tyrosine kinases, as well as 32 non-receptor tyrosine kinases, most of which transmitted a signal from the inner side of the cell surface to the nucleus (Roskoski, 2016). Many of those signals stimulated the cell to grow and divide, and several of them were implicated in cancer causation. For RTKs of the EGFR-family, four processes were identified that direct cells on a path to cancer: (1) mutation of the RTK that causes it to emit strong persistent signals to genes in the cell nucleus, stimulating genes that promote cell growth and division; (2) increase in the number RTK gene copies, thereby producing a strong chorus of such growth and cell division signals; (3) reduced destruction of RTK molecules, thereby allowing the newly synthesized RTKs to accumulate excessively in the cell surface membrane; ( 4) excessive activity of one the molecules in the pathway that conveys the signals from the RTKs to the cell nucleus. Efforts were made to block each of those steps with drugs or antibodies. Blocking the action at any of those steps often caused cancer cells, not only to stop dividing. but to die. It was as if the cancer cells had become addicted to increased levels of RTK activity. Particular attention was given to the EGFR family of RTKs. Mutations of the epidermal growth factor (EGFR) gene. If EGFR emits growth-promoting signals without control, it pushes cells to divide excessively. Such "over-expression" of EGFR was found to be a major factor promoting some cancers to develop and progress. One of the reasons for the excessive signaling. particularly in lung cancers, was found to be a mutation in the EGFR gene, for example a mutation that causes amino acid number 790 in the EGFR protein to be changed from threonine to methionine. This T790M mutation caused EGFR to send growth-promoting signals into the cells, even when there was no signal from the outside, e.g., even in the absence of binding of a growth factor, such as EGF. 323 K. W. Kohn Drugs Against cancer CHAPTER 17 Many different mutations were mapped in the EGFR gene in different cancers (Figure 17.8) (Sigismund et al., 2018). Several of those mutations made EGFR active even without binding a growth factor and therefore made the mutated EGFR an oncogene: an uncontrolled gene that continually sends growth signals to the cell nucleus pushing the cell to divide excessively. Would inhibiting EGFR's tyrosine kinase stop the excessive cell division in cancers? That thought stimulated a search for drugs that inhibit EGFR's tyrosine kinase. Glioblastoma NSCLC CRC 06-~EGFR~II) (06-185) .., ·· R451C, $46-11.. .,,,.. G465R G465E. K467T ·- •.. '491M,S492R M21-603 ........-"(,1195. G719C ....• ti.E7~A750 ll.L747-P753in$S "-- 4.L747-T751insS ~ '-.., I 1!:IUM y y '-. l.858R, l.8610 y y y y y ., " Figure 17.8. Common sites of mutation of EGFR gene in cancers of brain (glioblastoma), lung (NSCLC), and colon (CRC). ('16-273 means deletion of amino acids 6 through 273; G719S means mutation that replaces glycine at position 719 with serine, using single-letter amino acid symbols: R=arginine, E=glutamic acid, Q=glutamine, Y=tyrosine.) Mutations indicated in red made the mutated EGFR resistant to inhibitor drugs (Sigismund et al., 2018). (Why is itthat different cancers are associated with different mutation sites?) Finding inhibitors ofEGFR's ty rosine kinase. The focus on EGFR began in 1985 when Towia Libermann, Joseph Schlessinger and their coworker reported that the brain tumors of some patients had too many copies of the EGFR gene in their cancer cells (Libermann et al., 1985). Such amplification of the EGFR genes was found in about 40% of malignant brain tumors. The search for EGFR tyrosine kinase inhibitors then began in 1987 with the work of Gazit, Yaish, and Levitsky (Gazit et al., 1991; Gazit et al., 1989). The first inhibitors they found however were not specific for EGFR but inhibited the tyrosine kinase of many RTK types. 324 K. W. Kohn Drugs Against cancer CHAPTER 17 By 2006, it was clear that mutations, as well as amplifications, of the EGFR gene occurred in malignant brain tumors; investigators at Dana Farber Institute and Harvard Medical School found that 18 of 132 patients had missense mutations in their tumor's EGFR protein (missense mutations cause an improper amino acid to replace the normal one in the protein) (Lee et al., 2006). The mutations discovered at that time were in the part of the gene that codes for the extracellular domain of the EGFR protein. Cultured brain tumor cell lines bearing such mutations had increased growth capabilities and their growth was later found to be retarded by EGFR-inhibiting drugs. This strengthened researchers' conviction that such drugs could be clinically useful. The search for inhibitors of EGFR was driven, first of all, by evidence that cancers that expressed high levels of EGFR's tyrosine kinase tended to grow faster and metastasize more frequently. Such cancers thus were more malignant and patient survival was poor. Secondly, most critical normal tissues had no obvious need for EGFR, suggesting that EGFR- inhibitors would not cause major toxicity. Moreover, methods were available to determine the degree of expression of EGFR in the cancer of a particular patient. Finally, EGFR inhibitors were expected to be especially effective in cancers that produced both EGF (the ligand) and EGFR (the receptor for the ligand); the stimulator (EGF) and the effector upon which the stimulator acts (EGFR) would then be produced by the very same tumor: the EGF would be produced in the very vicinity where the EGFR receptors are located on the tumor cells (Arteaga, 2003). That circumstance is known as "autocrine" if the same cell produces both ligand and receptor. Or "paracrine" if ligand and receptor are produced by different cells in the same tumor. In either case, the tumor would self-stimulate and could grow without control. It turned out, however, that most EGFR mutations in human cancers inactivated the tyrosine kinase - so, using a tyrosine kinase inhibitor was pointless and useless in those cases. Most of the cancers, however, did not have EGFR mutations, but rather had an amplified EGFR gene, and they did respond to the tyrosine kinase inhibitors (Tsao et al., 2005). Regardless of whether the cause was mutation or gene amplification, excessive growth signals would be sent to the cancer cell nucleus. The path to the nucleus went by way of signaling molecules that could bear mutations causing effects similar to EGFR family mutations per se. Drugs that specifically inhibited this signaling pathway could cause the cancer cells to die, because those cells would have become addicted to the presence of the strong growth signals. An example was a mutation of BRAF, which was one of molecules in the signaling chain from EGFR to the nucleus; the mutation made BRAF abnormally active and uncontrolled: it sent growth-signal barrages to the nucleus, even without stimulation from EGFR. This was the cause of about half of melanoma cases, and these melanomas responded well, albeit only for several months, to BRAF inhibitors. This story is the subject of Chapter 19. 325 K. W. Kohn Drugs Against cancer CHAPTER 17 EGFR and its 3 close relatives were the first receptor tyrosine kinases to be studied intensively (Carpenter, 1987; Varden and Ullrich, 1988). At that time, over-expression of EGFR was already suspected to be a driver of the malignant cell division process in cancer. The EGFR family of receptor tyrosine kinases were also among the first for which inhibitor drugs were developed and approved by the U.S. Food and Drug Administration for cancer treatment. After imatinib/Gleevec (an inhibitor of the non-receptor tyrosine kinase BCR- ABL approved in 2001, see Chapter 14), the next to be approved, in 2003, was gefitinib/Iressa (Figure 17.9). Over the next 13 years, more than 20 tyrosine kinase inhibitors were approved, burgeoning an extremely active area of anticancer drug development (Roskoski, 2016). Thus: EGFR (Epidermal Growth Factor Receptor) and their family members are receptor tyrosine kinases: they are located in the surface membrane with an extracellular part that binds (is a receptor for) small regulatory molecules, such as epidermal growth factor (EGF); they have an intracellular part that has tyrosine kinase activity that stimulates certain proteins to signal cell division. HERZ is similar to EGFR, except that it lacks an extracellular growth factor receptor (Figure 17.6 and Textbox 1). HER2 and breast cancer. The importance of the HER2 gene in breast cancer became apparent when it was found that patients whose cancer cells had an excess number of HER2 genes had a relatively poor prognosis (Slamon et al., 1987). The HER2 gene was found to be amplified in some cases of breast, ovarian, and occasionally other types of cancer. A search therefore began for specific inhibitors of the HERZ protein's function. Two kinds of HERZ-inhibitors were developed: monoclonal antibodies that bound to the extracellular part of the HERZ molecule and drugs that bound and blocked HER2's tyrosine kinase activity in the intracellular part of the molecule. The cancers were, not only driven by abnormally high expression of the HER2 gene, but became dependent on (addicted to) that high degree of expression, such that inhibition of the HERZ protein caused the cancer cells to die. How inhibitors block EGF/t Crystallographic analysis of EGFR's protein structure showed how these drugs bound in a cleft at the active site of EGFR's tyrosine kinase domain, as shown in Figure 17.10. The tyrosine kinase reaction was found to take place within this cleft where ATP had to be present to contribute the phosphate group that the enzyme pushed onto the tyrosines of the EGFR protein. A drug molecule that entered and bound in the cleft would prevent ATP from entering. which is how those drugs blocked EGFR's function. 326 K. W. Kohn Drugs Against cancer CHAPTER 17 H~ C I ~ Gefitinib / iressa Erlotinib lapatinib Figure 17.9. Three inhibitors EGFR's tyrosine kinase approved for cancer treatment by the U.S. Food and Drug Administration (FDA). Gefitinib, approved in 2003 for treatment of lung cancer, inhibited platelet-derived growth factor (PDGF) as well as EGFR. Erlotinib, approved in 2004 for treatment of lung and pancreas cancers, was a more specific inhibitor of EGFR; Lapatinib, approved in 2007 for treatment of breast cancer, inhibited both EGFR and HER2 (Roskoski, 2016). K72 in h ibitor Figure 17.10. The active site of epidermal growth factor receptors, where phosphorylation reactions take place. Inhibitors, such as gefitinib bind in the cleft where ATP normally binds. The ATP molecule contributes the phosphate group used in the phosphorylation reactions. Some of the amino acids involved in catalyzing the phosphorylation reaction are labeled (Lynch et al., 2004). The cysteine (C) and the two methionines (M) each contribute a sulfur atom to the reaction, and the lysine (K) contributes a positive charge that stabilizes the ATP or inhibitor in the cleft. 327 K. W. Kohn Drugs Against cancer CHAPTER 17 How EGFR inhibitors kill cancer cells. After the discovery of the first of those drugs, it seemed that the excessively dividing cells driven by an overactive EGFR oncogene often became addicted and dependent on the oncogene's tyrosine kinase overactivity. Indeed, the tyrosine kinase inhibitor drugs caused the addicted cells to suicide ("apoptosis"). It was as if the cells could not tolerate interference with their addiction. Thus, tyrosine kinase inhibitor drugs became useful treatment for cancers that were driven by an overactive gene of the EGFR family. The response to a given inhibitor however depended on the type of mutation in the particular cancer it was intended to treat; some mutations even made cells resistant to the drugs (shown in red in Figure 17.8). It turned out, however, that overactive EGFR was not very often due to EGFR mutation and was more often caused by an abnormally high number (amplification) of normal EGFR genes, and that those were the cancers that responded to the tyrosine kinase inhibitors. Amplified EGFR genes and response to EGFR inhibitors. Increased gene copy number (gene amplification) had been observed in cancers that acquired resistance to antifolate drugs due to amplification the dihydrofolate reductase gene (Chapter 5). In the same vein, some cancers had amplified EGFR genes that were drivers of the malignancies, and those cancers were found to respond to EGFR inhibitors. Early studies revealed that malignant brain tumors (glioblastomas) often had an amplified EGFR gene as the probable cause of unusually high levels of EGFR protein that seemed to drive that cancer. The first report came in 1985 from Libermann and coworkers, who observed that 3 out of the 12 glioblastomas that they studied had more than 20 copies of the EGFR gene in their DNA and an increased amount of the EGFR protein in their cancer's cells (Libermann et al., 1985). Then, in 1996, Sauter and coworkers used a more precise technique to gauge gene amplification, which enabled them to count the number of EGFR genes per chromosome in a cell. They found EGFR amplified in about 40% of glioblastomas, although the number of amplified genes varied among the cells of a given tumor (Sauter et al., 1996). As we now understand it, only the EGFR-overexpressing cells would be EGFR-dependent and killed by an EGFR inhibitor. Some cells in the same tumor may have little or no EGFR-amplification, and those cells would survive and regrow the tumor. The newly grown tumor would then be resistant to the drug. That may be an important reason for the usually brief response of most cancers to EGFR inhibitors -- which gave impetus to studies to figuring out how to get around this difficulty. EGFR inhibitors and treatment of lung cancer. 328 K. W. Kohn Drugs Against cancer CHAPTER 17 Lung cancers were a notoriously difficult problem. Cisplatin and docetaxel improved survival a little, but after that there was no further therapy available - until, in 2003, early clinical trials tested tyrosine kinase inhibitors in lung cancer patients whose previous chemotherapy had failed. Both gefitinib and erlotinib seemed to improve the survival of these patients. In 2005, a larger randomized double-blind study was reported of advanced lung cancer patients whose previous chemotherapy had failed (Shepherd et al., 2005). The study showed that erlotinib improved the survival of some of the patients. However, the clearest measure of response w as the length of time that the treatment held the cancer in check, before it resumed growing (Figure 17.11). The graph indicated that about half of the patients did not respond to erlotinib. But the remaining half had a clear response: erlotinib extended the length of time before those cancers progressed. It may be that the responding patients were those whose cancers had an amplified EGFR gene -- but that possibility was not tested in this study. That idea - that only lung cancers that had amplified EGFR genes responded to erlotinib - was supported by further studies from the laboratory of Frances A. Shepherd at the University of Toronto, Canada (Figures 17.12 and 17.13) (Tsao et al., 2005). Progression-free Survival 100 P<0.001 by stratified log-rank test Hazard ratio, 0.61 (95% Cl, 0.5 1-0.74) 80 ~ 60 l!! ,:: ..," l 40 ··-·~ 20 Place~~\ Erlotinib 0 ·-···-··----· ------ 0 6 12 18 24 30 Months Figure 17.11. Progression-free survival of advanced lung cancer patients who were randomized as to whether or not they received erlotinib treatment The patients had received previous chemotherapy, which was not (or no longer was) effective. The graph shows that some of these patients responded to erlotinib: the drug increased the length of time before their tumors progressed (Shepherd et al., 2005). The upper part of the graph shows that about half the patients did not respond at all, while the lower part of the graph shows that the remaining halfof the patients had a clear response. 329 K. W. Kohn Drugs Against cancer CHAPTER 17 Figure 17.12. Increased number of copies (amplification) of the EGFR gene in cells of a lung cancer (right), compared with a lung cancer that lacked the gene amplification (left). This experiment used a double-staining method: a red spot shows where a DNA sequence of the EGFR gene is located, a green spot shows the centromere of chromosome 7, which is the chromosome in which EGFR is located. The cells of the cancer without the EGFR amplification show a single EGFR gene near the centromere of its host chromosome (left), while the cancer with the gene amplification shows an excessive number of EGFR genes (right). Some of cells had many EGFR copies (red spots) while some had relatively few, indicating that the number of copies varied among the cells of the same tumor (Tsao et al., 2005). 100 '• P=0.59 Hazard ratio, 0.85 100 . P=0.008 Hazard ratio, 0.44 -- 80 '• -~ (95%Cl, 0.48-1.51) -- 80 ,;·:.. : .,___ ~:. (95% Cl, 0.23-0.82) *c,i .:!: 60 '\_ * ti > 60 •, ·-. Erloti nib ~ :I 40 -~ :I 40 ' ---------· ·------L.--. VI 20 --°L~!!?_linib VI 20 0 0 0 6 12 18 24 30 0 6 12 18 24 30 Months Months Figure 17.13. An early clinical trial, reported in 2005, showed that the tyrosine kinase inhibitor, erlotinib, increased the survival of lung cancer patients whose cancers had amplified EGFR genes (right), but had no effect on the survival of the patients whose cancers lacked the amplification (left) (Tsao et al., 2005). 330 K. W. Kohn Drugs Against cancer CHAPTER 17 How HERZ g enes were discovered to drive the malignancy of many cancers. It had already been found out by 1995 that the cancer cells of advanced breast cancer patients often overexpressed either EGFR or of its relative, HER2, and that those patients had a relatively poor prognosis (Earp et al., 1995). Then by 2002, when the molecular structure of EGFR's tyrosine kinase domain had been determined, it became possible to design drug molecules that would bind to a pocket in the protein structure where the enzyme action occurs and thereby inhibit it (Figure 17.10). As already mentioned, the first inhibitors to be developed were designed to bind in a pocket in the tyrosine kinase domain of both EGFR and HER where ATP normally binds and where the enzyme action takes place (Figure 17.10). The first such inhibitor to become a useful anticancer drug was gefitinib (Iressa) (Figure 17.9) (Wakeling et al., 2002). Here is how the discovery of HER2 came about: In 1984, Alan Schechter, Robert Weinberg and their colleagues discovered an oncogene in a rat brain cancer called neuroglioblastoma. Because of where it was found, they called the new oncogene neu (Schechter et al., 1985; Schechter et al., 1984). (How they discovered oncogenes was related in Chapter 15.) They observed that DNA sequences of their neu gene were similar to those found in the previously discovered erbB oncogene of avian erythroblastosis virus. The neu gene therefore acquired the name erb82 - because the original erbB gene had been found to be similar to EGFR, which then became erbBl . The two genes, EGFR/ErbBl and neu/Erb82/HER2 were found to be closely related and they became the first of the 4 members of the EGFR family. The HER designation was for the human versions of those epidermal growth factor receptors. Despite this confusion of names, the prominence of the HER2 gene in human cancer perhaps merits its ownership of 3 names: neu, erb82, and HER2. Then in 1987, Dennis Slamon and his collaborators at the UCLA and University of Texas Medical Center reported that patients with breast cancers that had an abnormally high number of HER2 genes often had a relatively short survival time (Slamon et al., 1987) (Figure 17.14). They found that approximately 25% of breast cancers had an amplified number of HER2 genes (Slamon et al., 2001). These cancers appeared to be driven by the increased number of HER2 genes in the cell. But the HERZ-positive breast cancers did respond (albeit only transiently) to inhibitors of the tyrosine kinase function of HER2. 331 K. W. Kohn Drugs Against cancer CHAPTER 17 .C ·;; Not amplified (n :52) -~ 0.8 :, ~ ~ f 0.6 . ·.:; a. 0 0.4 C Amplified (n aa11) 0 >5 copies -~ 0.2 ~ 0,0-+--.....---....--.....---.-- -.----.-----. 0 12 24 36 48 60 72 84 Time (months) Figure 17.14. Amplification of the HER2 gene (more than 5 copies of the gene per cancer cell) impaired the survival of breast cancer patients (among patients who had positive lymph nodes for cancer) (Slamon et al., 1987). This was among the first findings to show the importance of the HER2 gene in breast cancer. Antibodies that inhibit EGFR and/or HERZ. Instead of targeting the tyrosine kinase activity of EGFR or HER2, another idea was to create antibodies that would bind to the extracellular domains of those receptor tyrosine kinases (RTKs). The idea was that the antibodies would block the binding of growth factors, such as EGF, to the extracellular domains or prevent the extracellular domains of two RTKs from coming together to form an active dimer (dimer formation and its consequences are diagrammed in Figures 17.6 and 17.7). Either of those inhibitions would prevent signaling to the nucleus to initiate DNA replication and cell division. Antibodies cannot penetrate directly through the cell surface membrane, but they could bind and inhibit the extracellular domains without entering the cell. In the early 1990s, John Mendelsohn and his coworkers at the Memorial Sloan-Kettering Cancer Center in New York developed monoclonal antibodies targeted to EGFR's extracellular domain. One of their antibodies (MAb528) inhibited the binding of growth factor EGF to the extracellular domain of EGFR and inhibited the growth of a human cancer in immune-deficient mice (Figure 17.15) (Baselga et al., 1993). They grew the tumor from cells of a human cancer cell line, which they injected into immune-deficient mice (that did not reject the cells of another species). When the tumor-bearing mice were treated with both MAb528 and doxorubicin, the growth of the tumor was completely inhibited, whereas each of those treatments by themselves only slowed the tumor growth to some degree (Figure 17.15) (Baselga et al., 1993). 332 K. W. Kohn Drugs Against cancer CHAPTER 17 14 MAb 528 12 •• ••• ••••• « ooxo control ~ 10 "'E .!:!. Q) 8 OOXO .tl - "' 0 E 6 .2 4 2 OOXO + MAb 528 0 0 20 30 40 50 days Figure 17.15. One of the first monoclonal antibodies (MAb528) that targeted EGFR's extracellular domain was shown in this experiment to inhibit the growth of a cancer. Combining the antibody with doxorubicin (DOX) completely inhibited the growth of a human tumor in immune-deficient mice. MAb528 bound the extracellular domain of EGFR and blocked the binding of growth factor EGF, thereby inhibiting the activation of EGFR's tyrosine kinase domain. MAb528 and doxorubicin each slowed the growth of the tumor compared with the growth of the tumor in untreated control mice. The combination of both treatments, however, completely inhibited the growth of the tumor. The arrows at the top show when the treatments were administered (Baselga et al., 1993). In 1989, researchers at Genentech Inc in San Francisco, California had reported the effects of several monoclonal antibodies directed against the extracellular domain of HERZ. They found that one of their antibodies, which they called 4D5, had the hoped-for specific action against HERZ-overexpressing breast cancer cells. After confirming their antibody's ability to bind HERZ, they found that the antibody, unlike other antibodies they had prepared, did in fact inhibit the growth of a HERZ-overexpressing human breast cancer cell line (adenocarcinoma SK-BR-3), but had little or no effect on several other cell lines (Hudziak et al., 1989). Their 4D5 antibody, however had a drawback: it was a mouse protein, which would likely be rejected by the human immune system. They therefore engineered a hybrid antibody that retained only the part of the mouse antibody that recognized the HERZ domain and replaced the rest of the mouse antibody with corresponding human antibody parts. The resulting humanized antibody became known as trastuzumab or Herceptin, and became a 333 K. W. Kohn Drugs Against cancer CHAPTER 17 useful addition to cancer treatment (Carter et al., 1992). When combined with standard chemotherapy, trastuzumab improved the prognosis of those breast cancer patients whose tumors produced high amounts ofHER2 (Figure 17.16 and 17.17) (Moasser, 2007a, b; Romond et al., 2005; Stebbing et al., 2000). These successes by scientists at Memorial Sloan-Kettering Cancer Center in New York and Genentech Inc in San Francisco opened a new era of cancer treatment using EGFR- and HERZ-directed antibodies. Genentech then developed another HERZ-directed antibody called 2C4, which prevented the dimer formation between EGFR and HERZ (Franklin et al., 2004). They engineered a humanized version of the antibody which became known as pertuzumab (Adams et al., 2006). Pertuzumab differed from trastuzumab by targeting subdomain II rather than subdomain IV of the extracellular domains of EGFR and HERZ (Figures 17.18 and 17.7). Subdomain II is the part of EGFR and HERZ that allows dimer formation between them, which has to happen before they can phosphorylate each other and send signals downstream; pertuzumab prevents that from happening (Figure 17.18). Did pertuzumab have a role in the treatment of cancer? In a phase II clinical trial, pertuzumab gave promising results in the treatment of advanced ovarian cancer. Ovarian cancer frequently had active, albeit not overexpressed, HERZ. Although there were no complete remissions in these patients, who had previously been extensively treated with chemotherapy, there were several partial remission or stable disease where the cancer was held in check for several months (Gordon et al., 2006). Although the early tests of pertuzumab failed to confer enthusiasm among clinicians, preclinical studies indicated that pertuzumab increased the effectiveness of trastuzumab, when the two antibodies were used together. The two antibodies differed in where on HER2's extracellular domain they bound and the manner in which they blocked HER2's function (Figure 17.18). The effect of the two antibodies used together was substantially greater than the effect of each of them by itself: they were synergistic. The synergism was first demonstrated in cultures of HER2-overexpressing breast cancer cells, conducted by researchers at M.D. Anderson Cancer Center in Houston, Texas. They showed that the two antibodies acted synergistically to kill the cancer cells (Nahta et al., 2004) (Figure 1 7.9). A large randomized clinical trial of previously untreated metastatic HER2-overexpressing breast cancer patients was reported in January 2012 by a large international group of clinical investigators led by Jose Baselga of Massachusetts General Hospital and Harvard Medical School (Baselga et al., 2012). They randomly assigned patients to be treated with trastuzumab plus docetaxel with or without the addition of pertuzumab. The results showed definitively that the addition of pertuzumab improved progression-free survival of the patients (Figure 17.19). The trastuzumab-docetaxel-pertuzumab triplet extended the time by a median of six months before the cancer progressed. Although that was far from a cure, it indicated a possible path toward that goal. 334 K. W. Kohn Drugs Against cancer CHAPTER 17 By 2012, several clinical trials indicated that pertuzumab by itself did little to benefit patients with HERZ-positive metastatic breast cancer. However, when combined with trastuzumab plus docetaxel, the three-drug combination gave favorable responses. That same year, the FDA approved the three-drug combination for the treatment of metastatic breast cancer (Hubalek et al., 2012; Keating, 2012; Traynor, 2012). 100 - cf?.. co 90 > 80 ~ 70 :::J C/) 60 - Q) Paclitaxel and trastuzumab .... Q) 50 C' 40 0 (/) 30 (/) .... Q) 20 0) 0 10 .... a. 0 0 5 10 15 Months after Enrollment Figure 17.16. Women with HER2-amplified breast cancer that had already metastasized were helped by adding trastuzumab (Herceptin) to their chemotherapy. In this randomized clinical trial, adding trastuzumab to paclitaxel lengthened the time that the cancer remained dormant and did not progress, compared with treatment with paclitaxel alone (Slamon et al., 2001). Other studies showed that trastuzumab also increased the progression-free time when added to other chemotherapy drugs. 335 K. W. Kohn Drugs Against cancer CHAPTER 17 100 I Trast'uzumab _ % (133: events) 90 87 1 85i3% 'if/. ' 'ii ~~ 80 ' Co nt,ol ' \Cherj,otherapy ! plus trastuzumab 1261 even ts) ~ J....-chemotherapy only "'" iN ~ 70 75.4% -..,; 67rr · i5 60 P<0.0001 Hazard ratio, 0.48 50 0 0 I 2 3 4 5 nme {years after start of drug therapy) Figure 17.17. Trastuzumab (Herceptin) increased the length of time before the cancer progressed in breast cancer patients whose cancer had had amplified HER2 genes and no evidence of distant metastases. All patients had surgery to remove their primary tumors followed by combination chemotherapy with or without trastuzumab; the control group received chemotherapy only (Romond et al., 2005). Dlmeri1atlon domain Figure 17.18. The antibodies trastuzumab and pertuzumab bind to different subdomains of HER2 (structure on the left; Figure 17.7 shows the subdomain structure.) (Harbeck et al., 2013). Trastuzumab binds to subdomain IV and thereby inhibits HER2's ability to emit growth signals. Pertuzumab also inhibits HER2 from emitting signals but does so by binding to subdomain II and thereby blocking HER2 dimerization and ability to bind growth factors. Trastuzumab and pertuzumab thus act in complementary fashion to block HER2 function. EGFR (HERl) must bind a growth factor, such as EGF (red oval) to achieve the correct conformation of the subdomains to permit dimer formation by way of subdomain II. However, HER2 already has the correct subdomain II conformation for dimer formation and does not have to bind growth factor. 336 K. W. Kohn Drugs Against cancer CHAPTER 17 100 - - Pertuzumab (medi8.l'I, 18.5 mo) 90 - - Control (median. lVI mo) it ;; 80 Trastuzumab plus docetaxel plus pertuzumab 70 ..·E, 8 tiO 50 / Trastuzumab plus docetaxel "'e 0 40 ·a ~ Hazard ratio, 0.62 30 (95%Cl, 0.Sl- 0,75) i ,I: 20 1><0.001 10 0 0 10 20 25 30 JS ,o Months Figure 17.19. Trastuzumab plus docetaxel had previously been shown to extend the progression- free survival of patients who had metastatic HER2-overexpressing breast cancers. This random clinical trial showed that adding pertuzumab to the therapy further extended the median progression-free survival of the patients for an additional 6 months (Baselga et al., 2012). How do cancer cells become resistant to those antibodies? Cancer cells were remarkably capable of becoming resistant, not only to all kinds of chemotherapy but to antibody agents as well. Although trastuzumab and pertuzumab were known to bind to HER2, the effects of that binding on the molecular and physiologic events in the cell remained enigmatic. The more researchers investigated, the more consequences of the antibody treatments they discovered. Of the multitude of effects that were discovered, it remained uncertain which of them were relevant to the resistance. Our state of knowledge about that in 2018 was summarized by (Derakhshani et al., 2020). Yet another therapeutic antibody. A humanized monoclonal antibody, cetuximab, was developed that targeted the extracellular domain of epidermal growth factor (EGFR) rather than of HERZ. It improved the therapy of colon cancer and was approved in 2004 for treatment of EGFR-expressing colon cancer. Like trastuzumab and pertuzumab, cetuximab was made up of a combination of human and murine parts; only the antigen-recognition parts were non-human (Figure 1 7.20) (Brand etal., 2011). Since most of the molecule was human, it was hoped that the human immune system would not react against it Among several early clinical trials of cetuximab, one that is notable was a collaboration of Canadian and Australian clinical research groups in 2007 that reported that, among 337 K. W. Kohn Drugs Against cancer CHAPTER 17 patients with advanced colon cancer whose cancers expressed EGFR, about half responded to cetuximab (even after chemotherapy had failed) (Jonker et al., 2007). Particularly interesting was the effect of cetuximab on the time until progression of the tumor (Figure 17.21 ). Its response curve was remarkably similar to that of the earlier study by (Shepherd et al., 2005) of the response of lung cancer patients to erlotinib (Figure 17.11). Even though the two studies differed in the type of cancer studied (lung versus colon cancer) and in the type of EGFR-targeted drug used (cetuximab versus erlotinib), the progression-free survival curves were amazingly similar. Both curves suggest that only about half the patients responded, as if there were two distinct groups of patients in each of the two studies. The patients whose cancers responded may have been those whose cancers had amplified EGFR genes, but that possibility was not investigated in these studies. A caveat that should be mentioned here is that targeting EGFR with either tyrosine kinase inhibitor or antibody was ineffective if the patient's cancer had an activating mutation in KRAS (the most important member of the RAS family) . That is because KRAS is downstream from EGFR: EGFR activates KRAS -- but, if KRAS was already activated by a mutation, it did no good to inhibit EGFR, because KRAS would drive the malignancy regardless (Li et al., 2015). KRAS is a topic in the next chapter. human: murine chimeric F Fe lgG1 Human Murine Figure 17.20. Development of the EGFR-targeted antibody, cetuximab. First, several antibodies against EGFR were generated in mice Qeft), and the most promising one of them was tested, but ran into the problem that patient's immune system reacted against this mouse protein. Most of the mouse protein was therefore replaced by means of genetic engineering with the corresponding human sections (right, red parts). The only murine part that was retained was the part that recognized the EGFR (Brand et al., 2011). 338 K. W. Kohn Drugs Against cancer CHAPTER 17 100 *t <I: 80 c0 ·;;; ., 60 f "" 0 ~ ll. 40 C 0 '€ 0 Q. e ll. 20 "'· Best supporti~ 0 care alone ···-·····- 0 2 4 6 8 10 12 Months since Randomization Figure 17.21. Response of patients with advanced colon cancer to cetuximab. The patients selected for this study had cancers that failed to respond or no longer responded to chemotherapy and whose cancers expressed EGFR Oonker et al., 2007). The curves indicate that about 4-0% of the patients responded to cetuximab, whereas 60%, whose disease progressed within 2 months, did not respond at all. Recycling and destruction of EGFR proteins. Although antibodies cannot penetrate through the cell surface membrane, they can be taken up into the cytoplasm when bound to a receptor, such as EGFR or HERZ, by a process known as endocytosis: the cell surface membrane together with attached EGFR-bound antibody is engulfed into vesicles. The antibodies are then located within the vesicles in the cytoplasm and are subject to processes that recycle EGFR (Figure 17.22). As EGFR molecules are made and sent to the surface membrane, they couldn't accumulate there indefinitely. A solution to this problem was found in 1998 by YosefYarden and his colleagues (Lenferink et al., 1998; Levkowitz et al., 1998; Waterman et al., 1999) at the Weizmann Institute in Israel. They found that EGFR molecules are sucked back from the cell surface membrane within vesicles that move into the cytoplasm. How that happens and what then happens to EGFR in the vesicles is pictured in Figure 17.22 as conceptualized by Sigismund et al (Sigismund et al., 2018). What ensues is a kind of choreography of vesicles, which can move and merge in various ways. The EGFR molecule then has one of two possible fates. It can be destroyed in special vesicles (lysosomes) that contain digestive enzymes. Or it can be recycled as the vesicle in which it resides merges with the surface membrane. All that is pictured in Figure 17.22. 339 K. W. Kohn Drugs Against cancer CHAPTER 17 The balance between the two pathways that lead to these alternative fates affects how many EGFR molecules would be on the surface at any one time, thereby affecting the strength of the EGFR signaling. It seemed that control of the balance between those pathways would be one way the cell could regulate EGFR activity. It would also affect the amount EGFR-bound antibody displayed on the cell surface. A beautiful example of control of how much EGFR is displayed on the cell surface was the effect of growth factor EGF. When there was little or no EGF available, the amount of EGFR on the cell surface accumulated and made the cell sensitive to detecting the rare EGF molecules. If EGF became abundant, there was the danger of EGFR overactivity. The cell could solve this dilemma by simply sucking EGF-bound EGFR out of the surface membrane 10-times more quickly than unbound EGFR. Excessive expression of EGFR would push cells toward becoming cancerous. The most common way that happened was by mutation of EGFR or one of its family members. But another way would be for the EGFR destruction machinery to be defective. In 2001, Waterman and Varden suggested that it might be possible to develop drugs to inhibit the recycling pathway, thereby perhaps enhancing the destruction pathway or to develop antibodies that, after binding to the EGFR extracellular domain, would enhance the uptake of the EGFR-antibody complex into cytoplasmic vesicles headed for destruction (Waterman and Varden, 2001). Cancer cells that had become addicted to a high level of EGFR might then die for lack of it. They proposed that this might be a way to treat overactive HERZ in aggressive forms of breast, ovary, and lung cancers. Evidence supporting that idea was reported in 2009 by Tsipi Ben-Kasus, Yoseph Varden, Michael Sela and their colleagues at the Weitzman Institute, Jerusalem. They found that certain monoclonal antibodies increased the rate of removal of HERZ from the surface membrane into endosomes (Ben- Kasus et al., 2009). They accomplished this by means of a pair of monoclonal antibodies that attached to different parts of the HERZ molecule, thereby producing a large complex with a strong tendency to be sucked into endosomes (Marmor and Varden, 2004; Mellman and Varden, 2013; Mosesson et al., 2008). 340 K. W. Kohn Drugs Against cancer CHAPTER 17 Cell surface ,._. EGFR recycled r EGFR dimer --- Ubiquitylated EGFR taken into clathrin- coated pit. rt~lng tndot0m• '~,,.-_.._-c.--.-.. ...-,1-y .-.-_) ....... , ......... Late endosome .., e ... EGFR in early endosome •~~7 ... late tMo.omeMVII .. .... oc•TU.1 EGFR decraded _ _ __ _ _ _ --_ .._ _ _ . , in lysosome .... Figure 17.22. EGFR degradation and recycling by way of vesicles (endosomes and lysosomes) in the cytoplasm. The cycle begins with an EGFR homodimer in the cell surface membrane (top, left); in the first step, the enzyme Chi adds ubiquitin (Uh) molecules to the EGFR That causes the EGFR to be taken into pits in the membrane, which is brought about by clathrin molecules that coat the cytoplasmic side of the membrane. The pits then close to form vesicles, called endosomes. The EGFR is then transferred, either to lysosomes, where it is degraded, or to other endosomes that recycle the EGFR to the cell surface membrane. From (Marmor and Varden, 2004) with labels in red added. Combining an EGFR a ntibody a nd a cytotoxic drug in the same molecule. In 2008, researchers at Genentech Inc. reported their development of an antibody-drug conjugate that takes advantage of the ability of cells to slurp from the cell surface EGFR- bound antibodies. Their new antibody-drug conjugate carried with it into the cytoplasm a toxic drug that killed or disabled the cell. Several antibody-drug conjugates targeted against various types of cancer cells had previously been developed, but Genentech's trastuzumab-emtansine or trastuzumab-DMl, specifically targeted HERZ-expressing cancers, including about one quarter of breast cancers. The conjugate consisted of the HERZ-targeted antibody, trastuzumab (Herceptin), chemically linked to the cell-killing drug maytansine, which bound and inhibited the function of the cell's microtubules (Chapter 12) (Figure 17.23) (Lewis Phillips et al., 2008). 341 K. W. Kohn Drugs Against cancer CHAPTER 17 The idea was that the trastuzumab part of the conjugate would bind exclusively to HER2 on the cell surface and would carry toxic maytansine into the cancer cell, whereupon the cancer cell would die. That scheme was targeted against cancers that had an amplified HER2 gene, such as HERZ-positive breast cancers. Critical normal tissue cells presumably would be spared due to their having few, if any, HER2 molecules on their surface. In their studies in 2008, the researchers reported evidence that the trastuzumab- emtansine conjugate indeed was effective against human HER2-overexpressing breast cancers grown as xenografts in immune-deficient mice, and moreover that it was more effective than trastuzumab by itself (Figure 17.24) (Lewis Phillips et al., 2008). But how effective would trastuzumab-emtansine be against HER2-overexpressing breast cancers in patients? In 2012, an international group of researchers randomly assigned 991 patients to treatment with trastuzumab-emtansine or to the standard therapy of a combination oflapatinib plus capecitabine (Verma et al., 2012). Lapatinib inhibited the tyrosine-kinase activities of HER2 and EGFR, and capecitabine was metabolized in the cell to release 5-fluorouracil (Chapter 6). The selected patients were in an advanced stage of their disease and had already been treated with trastuzumab plus a taxane (a microtubule inhibitor; Chapter 12). Thus, the test was whether trastuzumab and a microtubule inhibitor (maytansine) as a conjugate (trastuzumab-emtansine) would be effective after tastuzumab and a microtubule inhibitor given separately had failed. Analysis of the results showed that the patients did indeed respond to the trastuzumab-emtansine conjugate (Figure 17.25). Although the response was significantly better than a standard therapy (p<0.001), the addition of only 3 months to the time before the cancer progressed remained a bleak outlook for the patients. The clinical activity of trastuzumab emtansine was limited by the development ofresistance. Although several possible causes of resistance were suggested, which of them might be relevant remained uncertain (Hunter et al., 2020) (Garcia-Alonso et al., 2020). 342 K. W. Kohn Drugs Against cancer CHAPTER 17 ,,..o H ..::?o o I ~ O~N 0 CIO ,,..o ~ -NH 0 Maytansine Figure 17.23. Genetec's antibody-toxin conjugate drug. named trastuzumab-emtansine or trastuzumab-DMl (abbreviated T-DMl) consisted of trastuzumab connected by way of a linker to the microtubule blocker maytansine. The conjugate drug targeted HERZ-positive cancers, particularly breast cancers that had amplified HER2 genes (Lewis Phillips et al., 2008). The trastuzumab (Herceptin) part of the conjugate molecule was an antibody whose structure was similar to the structure shown on the left side of Figure 17.18. It bound to HER2 on the surface of HERZ-positive cancer cells, which took up the conjugate drug and delivered it into the cytoplasm. The toxin part of the conjugate, maytansine, then bound and inactivated the cell's microtubules. The linker between the trastuzumab and the maytansine was stable and did not release free maytansine. From (Lewis Phillips et al., 2008) modified in red to show the parts of the molecule. 343 K. W. Kohn Drugs Against cancer CHAPTER 17 KPL-4 1,200 1,000 ".s E 800 Q) E ::, 0 > 600 0 E .3 C 400 "' Q) ::; ---W- Vehicle --.- Trasa:uzumab (15 mgl1<g) 200 ~ Trastuntmab-MCC•DM 1 (15 mgikg, 750µg/~ DM1) oL..:::::~!::;!;:~~~11::11=----. 0 10 20 30 40 50 60 70 80 90 100 110 120 130 ,J';'.!J,.,♦ ♦ ♦ Time (d) "°"' ..., ♦ Figure 17.24. The antibody-drug conjugate trastuzumab-DMl (trastuzumab emtansine) cured human HER2-overexpressing cancers implanted as xenografts in immune deficient mice (blue squares) (Lewis Phillips et al., 2008). Trastuzumab by itself transiently inhibited the tumors, which however recovered and grew exponentially (red circles). In untreated mice, the tumors grew rapidly without delay (black x). 100 Median No. No.of of Months Events 80 Lapatinib-Capecitabine 6.4 304 T-DMl 9.6 265 Trastuzumab- Stratified hazard ratio, 0.6S 60 Emt1ns;ne (T-DM1) (95% Cl, 0.55- 0.77) / P<0.001 40 20 Capecitabine Lapatinib- c.apecitabine 0+--.--,,--,--,--,--.---r--'-,--,--',--r-r-"T'"-r--, 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 JO Months Figure 17.25. The trastuzumab-emtansine (T-DMl) drug conjugate oftrastuzumab linked to maytansine was better than standard chemotherapy with lapatinib plus capecitabine in prolonging the time before the cancers progressed (9.6 versus 6.4 months) . The patients had advanced HER2- overexpressing breast cancers who did not or no longer responded to trastuzumab and a taxane given separately ((Verma et al., 2012) with colored labels added). 344 K. W. Kohn Drugs Against cancer CHAPTER 17 Another type of HERZ-targeted antibody-toxin conjugate, trastuzumab-deruxtican, was made in 2016 by researchers in Japan. This conjugate consisted of trastuzumab linked to a camptothecin-related inhibitor oftopoisomerase I (Chapter 11). The HERZ humanized antibody (trastuzumab) was linked to a camptothecin-related topoisomerase I inhibitor (deruxtecan) by way of a peptide chain (five amino acids) that would be cleaved by a protease in the cytoplasm to release free deruxtecan (Figure 17.26) (Ogitani et al., 2016a; Ogitani et al., 2016b). The free deruxtecan has no electric charge and therefore could penetrate the blood-brain barrier. This is important, because patients with advanced HERZ-positive breast cancer often had brain metastases. Moreover, the released deruxtecan would perhaps have a relatively high concentration within the bulk of the tumor and would kill also the fraction of cancer cells that lacked high HERZ expression (a so-called bystander effect). That theoretically would confer an advantage, because HERZ-positive breast cancers often consisted of cells that expressed HERZ at various levels. Some of the malignant cells had relatively low levels of HERZ expression, and the released deruxtecan could kill these bystander cells. In 2019, researchers in Japan reported a non-randomized phase I study of trastuzumab- deruxtican in advanced HERZ-positive breast cancer patients who had previous treatment with trastuzumab-emtansine. They found significant benefit of trastuzumab-deruxtican in the patients beyond the benefit of the trastuzumab-emtansine treatment, and they concluded that the new drug should go on to phase 2 and phase 3 studies (Tamura et al., 2019). In December 2019, this new antibody-drug conjugate received approval for treatment of HERZ-expressing metastatic breast cancers that had failed previous HERZ- directed treatments (Kearn, 2020). Deruxtican by itself has interesting potential as a drug. It is similar to the commonly used topoisomerase I blocker topotecan (see Figure 11.11 in Chapter 11), the essential difference being that topotecan has a positive charge, whereas deruxtican is uncharged. Therefore, deruxtican may pass through membranes and enter the brain, whereas topotecan may not be able to do so. Moreover, deruxtican could enter cells more easily. However, as of April 2020, I could not find any reports of deruxtican as a drug in its own right. 345 K. W. Kohn Drugs Against cancer CHAPTER 17 Protease digestion Trastuzum ab F ......... OH 0 camptothecin-related topoisomerase 1 inhibitor. Figure 17.26. A HERZ-targeted antibody-toxin conjugate, trastuzumab-deruxtecan, linked trastuzumab to a camptothecin-related topoisomerase I inhibitor (deruxtecan). The linker consisted of a chain of five amino acids that would be digested by enzymes in the cytoplasm to release free deruxtecan (right) (Ogitani et al., 2016a). Targeting EGFR exon 20 insertion mutations in lung cancers. Non-small cell lung cancer (NSCLC) accounts for 80% to 85% of all lung cancers. Approximately 2% to 3% of those patients have EGFR exon 20 insertion mutations that promote rapid cancer cell growth and spread (Figure 17.27). Lung cancer was the leading cause of cancer mortality worldwide. Hence the 2% to 3% still added up to a substantial number of people. These mutations have not responded to current EGFR-targeted drugs, but today (May 21, 2021) the FDA gave accelerated approval to amivantamab, a monoclonal antibody targeted against those EGFR exon 20 mutations. Patients with lung cancers driven by that mutation will now for the first time have targeted therapy available. The accelerated approval of amivantamab was based on a clinical trial of 81 patients with lung cancers with EGFR exon 20 insertion mutations whose cancers were progressing despite treatment with platinum drugs (Chapter 3). 40% of the patients then responded to amivantamab with a median response duration of 11 months. This was impressive progress, yet less than half the patients responded to the drug and their responses usually lasted less than a year. 346 K. W. Kohn Drugs Against cancer CHAPTER 17 'Classical' EGFR mutations Exon 19 deletion$ L858R ' i Trans- Extracellular membrane ~'. Tyrosine kinase domain d omain domain EGFR Exon Exon Exon Exon Exon Exon • 1/11 18 19 , 20 -~ ........ - Exon 20 insertions ' , .................... ............. _.........- . ........ __ ....., ........ -- ,,,,,- C-helix 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 Loop following C-helix ',, ,, D E A y V M A s V D N p H V C ... .... ....-< <... ....... 'I'.... "'<p0.... '?........z ::t;,... :i:...... ....... ....f-... 0 )> )> C/)< 0 z "O :r < ~ (1) (1) (1) (1) (1) f- N ~ ... ...< ...... ... ::j ...... ~ ....... '{' ,;, .... ...< s::a,... ... a, ~ (/) . () ~ $ (1) ::t w "' 5· ;r "'5· "'5· 5· 5· 0 5· 5 "'5· 5 5· "' 5· "'X ~ "'X "'X "'X "'X "'X "'X "'X "'X "'X "'X 0 0 "'~ C) • . 0 .,, 0 "' ;;e "'..,, :., ~ ::, :ti:- * O> * . ,.""• !"' w ~ n 5' a -. .. "'<o ::i"'en ..- )( .... 0 " a"' Figure 17.27. Insertion mutations in EGFR exon 20, which is within the kinase domain in the intracellular region of the EGFR protein (Vyse and Huang. 2019). Most of the insertion mutations w ere found to be at three sites in the amino acid chain. For example, 24% of the insertions were between amino acids 769 and 770 and 25% w ere between amino acids 770 and 771. Synopsis It was a long road from the discovery epidermal growth factor (EG F) by Stanley Cohen in 1 965 to the development of targeted drugs and antibodies that improved the prognosis of patients with metastatic cancers of breast, lung, and others, as of the time of this writing in April 2020. A large number of tyrosine kinase inhibitor drugs were synthesized targeted to the enzyme activity of the epidermal growth factor receptor (EGFR) family, some of which became clinically effective. Another class of EGFR-targeted agents were antibodies targeted to extracellular domains of EGFR or its relative, HERZ. Improved outcomes were achieved by combining the inhibitors and antibodies in various ways, including combinations of 347 K. W. Kohn Drugs Against cancer CHAPTER 17 antibody and drug within the same molecule. Useful and promising clinical results were achieved against some of the common metastatic cancers, but it remained far from a cure of any of them. These achievements were made possible by much basic research that revealed molecular mechanisms of how mutation and amplification of the EGFR or HER2 genes drive the cancer process and how treatment could take advantage of the molecular vulnerabilities of those cancers. It was a long, complex, and fascinating story that still remains in progress. The recent developments of EGFR-targeted drugs and antibodies shows the lightning speed of current progress in biomedical research, while there evidently remains a long way to go. References Adams, C.W., Allison, D.E., Flagella, K., Presta, L., Clarke, J., Dybdal, N., McKeever, K., and Sliwkowski, M.X. (2006). Humanization of a recombinant monoclonal antibody to produce a therapeutic HER dimerization inhibitor, pertuzumab. Cancer lmmunol lmmunother55, 717-727. Arteaga, C.L. (2003). ErbB-targeted therapeutic approaches in human cancer. Experimental cell research 284, 122-130. Baselga, J., Cortes, J., Kim, S.B., Im, SA, Hegg, R, Im, Y.H., Roman, L., Pedrini, J.L., Pienkowski, T., Knott, A., et al. (2012). Pertuzumab plus trastuzumab plus docetaxel for metastatic breast cancer. The New England journal of medicine 366, 109-119. Baselga, J., Norton, L., Masui, H., Pandiella, A., Coplan, K., Miller, W.H., Jr., and Mendelsohn, J. (1993). Anti tumor effects of doxorubicin in combination with anti-epidermal growth factor receptor monoclonal antibodies. Journal of the National Cancer Institute 85, 1327-1333. Ben-Kasus, T., Schechter, B., Lavi, S., Varden, Y., and Sela, M. (2009). Persistent elimination of ErbB-2/HER2-overexpressing tumors using combinations of monoclonal antibodies: relevance of receptor endocytosis. Proceedings of the National Academy of Sciences of the United States of America 106, 3294-3299. Brand, T.M., Iida, M., and Wheeler, D.L (2011). Molecular mechanisms ofresistance to the EGFR monoclonal antibody cetuximab. Cancer biology & therapy 11, 777-792. Carpenter, G. (1987). Receptors for epidermal growth factor and other polypeptide mitogens. Annual review of biochemistry 56, 881-914. Carpenter, G., and Cohen, S. (1979). Epidermal growth factor. Annual review of biochemistry 48, 193-216. Carter, P., Presta, L., Gorman, C.M., Ridgway, J.B., Henner, D., Wong, W.L., Rowland, A.M., Kotts, C., Carver, M.E., and Shepard, H.M. (1992). Humanization of an anti-p185HER2 antibody for human cancer therapy. Proceedings of the National Academy of Sciences of the United States of America 89, 4285-4289. Cohen, S. (1965). The stimulation of epidermal proliferation by a specific protein (EGF). Developmental biology 12, 394-407. Cohen, S., and Carpenter, G. (1975). Human epidermal growth factor: isolation and chemical and biological properties. Proceedings of the National Academy of Sciences of the United States of America 72, 1317-1321. 348 K. W. Kohn Drugs Against cancer CHAPTER 17 Cohen, S., Carpenter, G., and King, L., Jr. (1980). Epidermal growth factor-receptor-protein kinase interactions. Co-purification of receptor and epidermal growth factor- enhanced phosphorylation activity. The Journal of biological chemistry 255, 4834- 4842. Derakhshani, A., Rezaei, Z., Safarpour, H., Sabri, M., Mir, A., Sanati, M.A., Vahidian, F., Gholamiyan Moghadam, A., Aghadoukht, A., Hajiasgharzadeh, K., et al. (2020). Overcoming trastuzumab resistance in HERZ-positive breast cancer using combination therapy. J Cell Physiol 235, 3142-3156. Downward, J., Varden, Y., Mayes, E., Scrace, G., Totty, N., Stockwell, P., Ullrich, A., Schlessinger, J., and Waterfield, M.D. (1984). Close similarity of epidermal growth factor receptor and v-erb-B oncogene protein sequences. Nature 307, 521-527. Earp, H.S., Dawson, T.L., Li, X., and Yu, H. (1995). Heterodimerization and functional interaction between EGF receptor family members: a new signaling paradigm with implications for breast cancer research. Breast cancer research and treatment 35, 115-132. Endres, N.F., Barros, T., Cantor, A.J., and Kuriyan, J. (2014). Emerging concepts in the regulation of the EGF receptor and other receptor tyrosine kinases. Trends Biochem Sci 39, 437-446. Fabi, A., Mottolese, M., and Segatto, 0 . (2014). Therapeutic targeting of ERBB2 in breast cancer: understanding resistance in the laboratory and combating it in the clinic. J Mol Med (Berl) 92, 681-695. Franklin, M.C., Carey, K.D., Vajdos, F.F., Leahy, D.J., de Vos, A.M., and Sliwkowski, M.X. (2004). Insights into ErbB signaling from the structure of the ErbB2-pertuzumab complex. Cancer cell 5, 317-328. Garcia-Alonso, S., Ocana, A., and Pandiella, A. (2020). Trastuzumab Emtansine: Mechanisms of Action and Resistance, Clinical Progress, and Beyond. Trends in cancer 6, 130-146. Gazit, A., Osherov, N., Posner, I., Yaish, P., Poradosu, E., Gilon, C., and Levitzki, A. (1991 ). Tyrphostins. 2. Heterocyclic and alpha-substituted benzylidenemalononitrile tyrphostins as potent inhibitors of EGF receptor and ErbB2/neu tyrosine kinases. Journal of medicinal chemistry 34, 1896-1907. Gazit, A., Yaish, P., Gilon, C., and Levitzki, A. (1989). Tyrphostins I: synthesis and biological activity of protein tyrosine kinase inhibitors. Journal of medicinal chemistry 32, 2344- 2352. Gordon, M.S., Matei, D., Aghajanian, C., Matulonis, U.A., Brewer, M., Fleming, G.F., Hainsworth, J.D., Garcia, A.A., Pegram, M.D., Schilder, RJ., et al. (2006). Clinical activity ofpertuzumab (rhuMAb 2C4), a HER dimerization inhibitor, in advanced ovarian cancer: potential predictive relationship with tumor HERZ activation status. Journal of clinical oncology : official journal of the American Society of Clinical Oncology 24, 4324-4332. Haigler, H., Ash, J.F., Singer, S.J., and Cohen, S. (1978). Visualization by fluorescence of the binding and internalization of epidermal growth factor in human carcinoma cells A- 431. Proceedings of the National Academy of Sciences of the United States of America 75, 3317-3321. Harbeck, N., Beckmann, M.W., Rody, A., Schneeweiss, A., Muller, V., Fehm, T., Marschner, N., Gluz, 0 ., Schrader, I., Heinrich, G., et al. (2013). HERZ Dimerization Inhibitor 349 K. W. Kohn Drugs Against cancer CHAPTER 17 Pertuzumab - Mode of Action and Clinical Data in Breast Cancer. Breast Care (Basel) 8, 49-55. Hollenberg. M.D., and Cuatrecasas, P. (1973). Epidermal growth factor: receptors in human fibroblasts and modulation of action by cholera toxin. Proceedings of the National Academy of Sciences of the United States of America 70, 2964-2968. Hubalek, M., Brantner, C., and Marth, C. (2012). Role ofpertuzumab in the treatment of HERZ-positive breast cancer. Breast Cancer (Dove Med Press) 4, 65-73. Hudziak, R.M., Lewis, G.D., Winget, M., Fendly, B.M., Shepard, H.M., and Ullrich, A. (1989). p185HER2 monoclonal antibody has anti proliferative effects in vitro and sensitizes human breast tumor cells to tumor necrosis factor. Molecular and cellular biology 9, 1165-1172. Hunter, F.W., Barker, H.R., Lipert, B., Rothe, F., Gebhart, G., Piccart-Gebhart, M.J., Sotiriou, C., and Jamieson, S.M.F. (2020). Mechanisms of resistance to trastuzumab emtansine (T- DMl) in HERZ-positive breast cancer. British journal of cancer 122, 603-612. Jonker, D.J., O'Callaghan, C.J., Karapetis, C.S., Zalcberg, J.R., Tu, D., Au, H.J., Berry, S.R., Krahn, M., Price, T., Simes, R.J., et al. (2007). Cetuximab for the treatment of colorectal cancer. The New England journal of medicine 357, 2040-2048. Kearn, S.J. (2020). Trastuzumab Deruxtecan: First Approval. Drugs 80, 501-508. Keating, G.M. (2012). Pertuzumab: in the first-line treatment of HERZ-positive metastatic breast cancer. Drugs 72, 353-360. Kohn, K.W., Aladjem, M.I., Weinstein, J.N., and Pommier, Y. (2006). Molecular interaction maps ofbioregulatory networks: a general rubric for systems biology. Mol Biol Cell 17, 1-13. Lee, J.C., Vivanco, I., Beroukhim, R., Huang, J.H., Feng, W.L., DeBiasi, R.M., Yoshimoto, K., King. J.C., Nghiemphu, P., Yuza, Y., et al. (2006). Epidermal growth factor receptor activation in glioblastoma through novel missense mutations in the extracellular domain. PLoS Med 3, e485. Lemmon, M.A., Schlessinger, J., and Ferguson, K.M. (2014). The EGFR family: not so prototypical receptor tyrosine kinases. Cold Spring Harb Perspect Biol 6, a020768. Lenferink, A.E., Pinkas-Kramarski, R., van de Poll, M.L., van Vugt, M.J., Klapper, L.N., Tzahar, E., Waterman, H., Sela, M., van Zoelen, E.J., and Varden, Y. (1998). Differential endocytic routing of homo- and hetero-dimeric ErbB tyrosine kinases confers signaling superiority to receptor heterodimers. The EMBO journal 17, 3385-3397. Levkowitz, G., Waterman, H., Zamir, E., Kam, Z., Oved, S., Langdon, W.Y., Beguinot, L., Geiger, B., and Varden, Y. (1998). c-Cbl/Sli-1 regulates endocytic sorting and ubiquitination of the epidermal growth factor receptor. Genes Dev 12, 3663-3674. Lewis Phillips, G.D., Li, G., Dugger, D.L., Crocker, L.M., Parsons, K.L., Mai, E., Blattler, W.A., Lambert, J.M., Chari, R.V., Lutz, R.J., et al. (2008). Targeting HERZ-positive breast cancer with trastuzumab-DMl, an antibody-cytotoxic drug conjugate. Cancer research 68, 9280-9290. Li, X.X., Liang. L., Huang. L.Y., and Cai, S.J. (2015). Standard chemotherapy with cetuximab for treatment of colorectal cancer. World journal of gastroenterology 21, 7022-7035. Libermann, T.A., Nusbaum, H.R., Razon, N., Kris, R., Lax, I., Soreq, H., Whittle, N., Waterfield, M.D., Ullrich, A., and Schlessinger, J. (1985). Amplification, enhanced expression and possible rearrangement of EGF receptor gene in primary human brain tumours of glial origin. Nature 313, 144-147. 350 K. W. Kohn Drugs Against cancer CHAPTER 17 Lu, H.S., Chai, J.J., Li, M., Huang, B.R., He, C.H., and Bi, R.C. (2001). Crystal structure of human epidermal growth factor and its dimerization. The Journal of biological chemistry 276, 34913-34917. Lynch, T.J., Bell, D.W., Sordella, R., Gurubhagavatula, S., Okimoto, R.A., Brannigan, B.W., Harris, P.L., Haserlat, S.M., Supko, J.G., Haluska, F.G., et al. (2004). Activating mutations in the epidermal growth factor receptor underlying responsiveness of non-small-cell lung cancer to gefitinib. The New England journal of medicine 350, 2129-2139. Marmor, M.D., and Varden, Y. (2004). Role of protein ubiquitylation in regulating endocytosis ofreceptor tyrosine kinases. Oncogene 23, 2057-2070. Mellman, I., and Varden, Y. (2013). Endocytosis and cancer. Cold Spring Harb Perspect Biol 5, a016949. Moasser, M .M. (2007a). The oncogene HERZ: its signaling and transforming functions and its role in human cancer pathogenesis. Oncogene 26, 6469-6487. Moasser, M .M. (2007b ). Targeting the function of the HERZ oncogene in human cancer therapeutics. Oncogene 26, 6577-6592. Mosesson, Y., Mills, G.B., and Varden, Y. (2008). Derailed endocytosis: an emerging feature of cancer. Nature reviews Cancer 8, 835-850. Nahta, R., Hung, M.C., and Esteva, F.J. (2004). The HER-2-targeting antibodies trastuzumab and pertuzumab synergistically inhibit the survival of breast cancer cells. Cancer research 64, 2343-2346. Ogitani, Y., Aida, T., Hagihara, K., Yamaguchi, J., Ishii, C., Harada, N., Soma, M., Okamoto, H., Oitate, M., Arakawa, S., et al. (2016a). DS-8201a, A Novel HERZ-Targeting ADC with a Novel DNA Topoisomerase I Inhibitor, Demonstrates a Promising Anti tumor Efficacy with Differentiation from T-DMl. Clinical cancer research : an official journal of the American Association for Cancer Research 22, 5097-5108. Ogitani, Y., Hagihara, K., Oitate, M., Naito, H., and Agatsuma, T. (2016b). Bystander killing effect of DS-8201a, a novel anti-human epidermal growth factor receptor 2 antibody- drug conjugate, in tumors with human epidermal growth factor receptor 2 heterogeneity. Cancer Sci 107, 1039-1046. Reddy, R.J., Gajadhar, A.S., Swenson, E.J., Rothenberg, D.A., Curran, T.G., and White, F.M. (2016). Early signaling dynamics of the epidermal growth factor receptor. Proceedings of the National Academy of Sciences of the United States of America 113, 3114-3119. Romond, E.H., Perez, E.A., Bryant, J., Suman, V.J., Geyer, C.E., Jr., Davidson, N.E., Tan-Chiu, E., Martino, S., Paik, S., Kaufman, P.A. et al. (2005). Trastuzumab plus adjuvant chemotherapy for operable HERZ-positive breast cancer. The New England journal of medicine 353, 1673-1684. Roskoski, R., Jr. (2016). Classification of small molecule protein kinase inhibitors based upon the structures of their drug-enzyme complexes. Pharmacological research 103, 26-48. Saule, S., Roussel, M., Lagrou, C., and Stehelin, D. (1981). Characterization of the oncogene (erb) of avian erythroblastosis virus and its cellular progenitor. Journal of virology 38, 409-419. Sauter, G., Maeda, T., Waldman, F.M., Davis, R.L, and Feuerstein, B.G. (1996). Patterns of epidermal growth factor receptor amplification in malignant gliomas. Am J Pathol 148, 1047-1053. 351 K. W. Kohn Drugs Against cancer CHAPTER 17 Schechter, A.L, Hung. M.C., Vaidyanathan, L, Weinberg. R.A., Yang-Feng. T.L., Francke, U., Ullrich, A., and Coussens, L. (1985). The neu gene: an erbB-homologous gene distinct from and unlinked to the gene encoding the EGF receptor. Science 229, 976-978. Schechter, A.L, Stern, D.F., Vaidyanathan, L., Decker, S.J., Drebin, J.A., Greene, M.I., and Weinberg. R.A. (1984). The neu oncogene: an erb-B-related gene encoding a 185,000- Mr tumour antigen. Nature 312, 513-516. Shepherd, F.A., Rodrigues Pereira, J., Ciuleanu, T., Tan, E.H., Hirsh, V., Thongprasert, S., Campos, D., Maoleekoonpiroj, S., Smylie, M., Martins, R., et al. (2005). Erlotinib in previously treated non-small-cell lung cancer. The New England journal of medicine 353, 123-132. Sigismund, S., Avanzato, D., and Lanzetti, L. (2018). Emerging functions of the EGFR in cancer. Mol Oncol 12, 3-20. Slamon, D.J., Clark, G.M., Wong. S.G., Levin, W.J., Ullrich, A., and McGuire, W.L. (1987). Human breast cancer: correlation of relapse and survival with amplification o f the HER-2/neu oncogene. Science 235, 177-182. Slamon, D.J., Leyland-Jones, B., Shak, S., Fuchs, H., Paton, V., Bajamonde, A., Fleming. T., Eiermann, W., Wolter, J., Pegram, M., eta/. (2001). Use of chemotherapy plus a monoclonal antibody against HER2 for metastatic breast cancer that overexpresses HER2. The New England journal of medicine 344, 783-792. Stebbing, J., Copson, E., and O'Reilly, S. (2000). Herceptin (trastuzamab) in advanced breast cancer. Cancer treatment reviews 26, 287-290. Tamura, K., Tsurutani, J., Takahashi, S., Iwata, H., Krop, I.E., Redfern, C., Sagara, Y., Doi, T., Park, H., Murthy, R.K., et al. (2019). Trastuzumab deruxtecan (DS-8201a) in patients with advanced HERZ -positive breast cancer previously treated with trastuzumab emtansine: a dose-expansion, phase 1 study. The Lancet Oncology 20, 816-826. Traynor, K. (2012). FDA approves pertuzumab for breast cancer. Am J Health Syst Pharm 69, 1178. Tsao, M.S., Sakurada, A., Cutz, J.C., Zhu, C.Q., Kamel-Reid, S., Squire, J., Lorimer, I., Zhang, T., Liu, N., Daneshmand, M., et al. (2005). Erlotinib in lung cancer - molecular and clinical predictors of outcome. The New England journal of medicine 353, 133-144. Verma, S., Miles, D., Gianni, L., Krop, I.E., Welslau, M., Baselga, J., Pegram, M., Oh, D.Y., Dieras, V., Guardino, E., et al. (2012). Trastuzumab emtansine for HERZ-positive advanced breast cancer. The New England journal of medicine 367, 1783-1 791. Vyse, S., and Huang, P.H. (2019). Targeting EGFR exon 20 insertion mutations in non-small cell lung ca ncer. Signal Transduct Target Ther 4, 5. Wakeling. A.E., Guy, S.P., Woodburn, J.R., Ashton, S.E., Curry, B.J., Barker, A.J., and Gibson, K.H. (2002). ZD1839 (Iressa): an orally active inhibitor o f epidermal growth factor signaling with potential for ca ncer therapy. Cancer research 62, 5 749-5754. Waterman, H., Alroy, I., Strano, S., Seger, R., and Varden, Y. (1999). The C-terminus of the kinase-defective neuregulin receptor ErbB-3 confers mitogenic superiority and dictates endocytic routing. The EMBO journal 18, 3348-3358. Waterman, H., and Varden, Y. (2001). Molecular mechanisms underlying endocytosis and sorting of ErbB receptor tyrosine kinases. FEBS Lett 4 90, 142-152. Varden, Y., and Schlessinger, J. (1987a). Epidermal growth factor induces rapid, reversible aggregation o f the purified epidermal growth factor receptor. Biochemistry 26, 1443- 1451. 352 K. W. Kohn Drugs Against cancer CHAPTER 17 Varden, V., and Schlessinger, J. (1987b). Self-phosphorylation of epidermal growth factor receptor: evidence for a model of intermolecular allosteric activation. Biochemistry 26, 1434-1442. Varden, V., and Ullrich, A. (1988). Growth factor receptor tyrosine kinases. Annual review of biochemistry 57, 443-478. 353 K. W. Kohn Drugs Against cancer CHAPTER 18 QopluJB. fllcMS~stoty12l007b.3 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@oib gov CHAPTER 18 The RAS oncogene story RAS oncogenes in viruses. A particularly important family of genes or oncogenes in human cancer, the RAS genes, were first discovered through studies of cancer-causing viruses. Sometime in 1963, Jennifer Harvey, working at the Cancer Research Department of the London Hospital Research Laboratories, was inoculating mice and rats with plasma from a rat that had a virus-induced leukemia. She was routinely transferring the virus from one animal to another, inducing leukemia in each of them. However, on one occasion that year, she noted something unusual that was to open a new window to cancer cause and treatment (Harvey, 1964). Mice that were inoculated with virus from one of her leukemic rats unexpectedly developed solid tumors in addition to the usual leukemia (which have malignant cells in the blood and lymph nodes instead of in lumps in various tissues). Her leukemia virus was later shown to have picked up (spliced into its genome) a DNA fragment from the rat's own genome. That piece of DNA, which was now part of the genome of the new virus, caused the solid-tumor-type cancer lumps in her mice. Moreover, the new cancer gene was found to be a mutated version of a normal gene, RAS (probably for rat.s.arcoma, where a mutated version was first discovered). Harvey's name was to become immortalized by the letter Hin the newly discovered HRAS oncogene, which was a mutated form of a normal HRAS gene. Harvey's new virus caused cells on the surface of a dish to overgrow to form "foci" (Figure 18.1) in a manner similar to what Weinberg's group later observed in their oncogene studies (Figure 15.3 in Chapter 15). Harvey's virus particles seen in electron microscope images had a remarkable unusual structure resembling spoked wheels (Figure 18.2). 354 K. W. Kohn Drugs Against cancer CHAPTER 18 In 1967, W. H. Kirsten and L. A. Mayer detected another virus that produced solid tumors in mice. That virus was later found to have picked up a mutated version of another gene of the RAS family, which became known as KRAS (K for Kirsten) (Kirsten and Mayer, 1967). KRAS became one of the most important cancer genes and was discovered to be mutated in nearly all cases of pancreatic cancer. These early observations led to enormous research efforts that gave much detailed information about the RAS genes and their cancer- inducing mutations. .- -~ ..,..,q • . • .,,.. ••.," . t ., , • • • - ' .. Figure 18.1. Foci of high cell density caused by Harvey's new cancer virus that contained the mutated HRAS gene in its genome. Left, normal cells growing on a surface; right, foci of excessive cell multiplication caused by the virus (Simons et al., 1967). 355 K. W. Kohn Drugs Against cancer CHAPTER 18 Figure 18.2. Electron microscope images of Harvey's new cancer virus. Each cell sometimes had hundreds of these unusual particles whose structure differed from all previously known viruses. The virus structure resembled spoked wheels within a vesicle membrane that sometimes had ribosomes attached (dark bodies in figure f (lower left), showing that the membrane came from the cell's "rough endoplasmic reticulum" where proteins are made) (De Petris and Harvey, 1969). But where did the Harvey and Kirsten tumor-producing viruses come from? In 1973, Ed Scolnick and his colleagues at NCI reported that the Kirsten sarcoma virus arose from the Kirsten leukemia virus by genetic recombination with sequences present in the rat cells (Scolnick et al., 1973). The Kirsten leukemia virus, while growing in rat cells, had picked up sequences that were already present in those cells, with the result that the new virus was then able to form tumors. In 1974, Scolnick found that the Kirsten and Harvey viruses had picked the same sequences, which is what made them tumor-producing viruses, and which may have come from the rat genome itself (Scolnick and Parks, 1974)! Finally in 1979, after restrictions on cloning had been lifted in 1977, Gordon Hager, Ed Scolnick, Doug Lowy, and their colleagues at NIH cloned the Harvey sarcoma virus genome (Hager et al., 1979). 356 K. W. Kohn Drugs Against cancer CHAPTER 18 What do RAS genes do in cells? Since a version of the RAS oncogene caused or triggered the development of cancer, researchers were anxious to find out what the RAS protein does in cells. An important observation about the proteins derived from RAS genes was reported in 1980 by NIH researchers Mark Willingham, Ira Pastan, Thomas Shih, and Ed Scolnick (Willingham et al., 1980). They found RAS-like proteins at the inner surface of the plasma membrane of cells that had been transformed by Harvey sarcoma virus (Figure 18.3). The result was similar to the observation that epidermal growth factor (EGF) also bound to the cell surface membrane (Figure 17.5 in Chapter 17). The importance of these observations became evident when the role of RAS in the signaling network from receptor tyrosine kinases, such as epidermal growth-factor receptor (EGFR), was worked out -- and when it was discovered that receptor tyrosine kinases, such as EGFR, as well as the RAS proteins were attached to the cell surface membranes of the cells. As we will see, RAS turned out to be directly in the signaling path from EGFR. (The EGFR story was told in Chapter 17.) Proteins with structure and function similarities to mammalian RAS were found in a remarkably wide variety of organisms from yeast to worms to insects, which highlighted their central role in the life of many kinds of cells (Sigal et al., 1988) (Lowenstein et al., 1992). The fact that the cancer-driving RAS oncogenes are mutated versions of the normal RAS genes was reported in 1982 by M. Barbacid and his coworkers in the National Cancer Institute (Santos et al., 1982). In 1984, Raymond Sweet and his colleagues at Cold Spring Harbor Laboratory injected the mutated HRAS gene into a variety of cells and found that it increased the proliferation of the cells in cancer-like fashion (Feramisco et al., 1984). The mutated RAS protein (product of a mutated RAS gene) was later found to be a rogue molecule that sent its growth-promoting signal downstream without control and without requiring input from receptor tyrosine kinases. Overview ofRAS in the s ignaling path from EGF/t After receiving activating signals from EGFR (or from other receptor tyrosine kinases), RAS transmits the signal to the cell nucleus, telling the machinery therein to activate cell division. For RAS to receive signals from EGFR, it helped for the two to be located in the same neighborhood. Since EGFR transmits signals from outside to inside the cell, the EGFR molecule is in the cell surface membrane with part of the molecule outside and part inside the cell (Chapter 17). The location of RAS at the inner surface of the membrane is therefore ideal for efficient interaction with EGFR. It was indeed found that the ability of the RAS protein to bind to the inner surface of the membrane was required for RAS to receive signals from the receptor tyrosine kinases. However, RAS did not bind directly to EGFR. Instead, there was a protein that connected between the two. This EGFR-to-RAS connector protein came to have a strange name: SOS, 357 K. W. Kohn Drugs Against cancer CHAPTER 18 standing for "sister of sevenless." The discovery of SOS and the reason for its strange name story came from research on fruit fly eyes, a remarkable story that I will tell next. Figure 18.3. An experiment showing that RAS proteins are located at the inner surface of the cell surface membrane (arrow). This experiment was reported in 1980 by NIH scientists Mark Willingham, Ira Pastan, Thomas Shih, and Ed Scolnick (Willingham et al., 1980). They used an antibody that bound to the RAS protein specifically. The antibody's fluorescence under ultraviolet light showed up bright in this image. They also showed that the RAS protein was not on the external surface of the cell: there was no fluorescence when the antibody was applied to intact cells rather than to the fixed cells in the experiment shown here. (The antibody could not penetrate into cells unless the cells were opened up by chemical fixation.) From viruses and fruit fly eyes to RAS and cancer-driver genes. Three seemingly unrelated and arcane investigations converged to one of the most important discoveries about cancer: the discovery of the RAS oncogenes, which paved the way for the development of targeted anticancer drugs: • A virus unexpectedly produced malignant tumors in mice. • Peculiar mutations in the eyes offruit flies disclosed genes that were similar to previously unidentified human genes. • DNA from human cancer cells transformed non-cancerous cells to become cancerous. I have already told the first and third of those stories; this section is about the second - an arcane and indeed amazing story about mutations of the eyes of fruit Hies. Who would have imagined that studies of genetic alterations in fruit fly eyes would lead to the discovery of cancer-causing genes and to therapies designed to block those over-active 358 K. W. Kohn Drugs Against cancer CHAPTER 18 mutated genes in cancers? The story of how that happened is both fascinating and enlightening. From fruit fly eyes to human RAS genes. In order to probe the unknown, a key is needed to unlock a door. A key can be found in the most unlikely of places -- which, in this case, was memorialized by an unknown (to me) author: 3 blind flies, see how they fly one was missing the seventh cell another lost its daughter cell the third had no mother cell but it a/I led to a cancer cure and never got a golden fleece prize for 3 blind flies, 3 blind flies. So, let's have a look at the fruit fly eye and what those missing eye cells were all about. The compound eye of a fruit fly consists of several hundred small eye units, called "ommatidia", each of which has 8 photoreceptor cells arranged in a strict geometric order. Each of those photoreceptor cells was designated by a number, based on its position (Figure 18.4). A mutation was found in a blind fly whose photoreceptor cell number 7 was missing in every little eye unit (ommatidium) (Figure 18.4). Geneticists dubbed the mutation sevenless, in line with the usual whimsy of those researchers. To have a normal eye, the fly had to have a normal sevenless gene. If its sevenless gene was mutated, photoreceptor cell number 7 was missing. and the fly did not see well. To see the drastic effect that a mutation of its sevenless gene has on the structure of a fly's eye, have a look at Figure 18.5. However, geneticists as usual were not content with discovering just one interesting mutation. They observed that the normal development of receptor cell number 7 was defective if there was a mutation in a different gene, which their whimsy dubbed bride of sevenless. That name reflected their finding that the protein coded by that gene binds to and is required for the function of the sevenless protein. But the process of finding mutations in fruit fly eye cells did not end there. They found yet another gene whose mutation caused problems with receptor cell number 7. They dubbed that gene son ofsevenless (SOS). To everyone's astonishment, that SOS gene of the fruit fly had a DNA sequence that resembled a human gene that was implicated in the function of the RAS genes (Raabe, 2000). After much investigation, the human version of the SOS gene was found to fit in the pathway that leads from a variety of receptor tyrosine kinases -- most notably EGFR-- to RAS. The EGFR story was related in Chapter 17. Figure 18.6 shows the remarkable similarity of the pathways where SOS has a role in transmitting 359 K. W. Kohn Drugs Against cancer CHAPTER 18 signals from outside the cell to genes in the cell nucleus. The pathways from EGF via SOS and RAS, to RAF, MEK, and ERK were found to be the same in the different species. Figure 18.4. Eye units ("omatidia") of a normal fly (left) and a sevenless mutant (right). As you can see by counting the dark blobs in each group, the normal fly had 7 photoreceptor cells visible in each omatidlum, whereas the mutant had only 6. Photoreceptor cell number 7 was missing in the mutant. (An 8th photoreceptor is not visible in this section and was unaffected by this mutation.) (From (Raabe, 2000).) Figure 18.5. How mutation of the SOS gene affects the eye of a fruit fly. Left, eye of a normal fruit fly; right, eye of a fruit fly that had an SOS mutation (Rogge et al., 1991). 360 K. W. Kohn Drugs Against cancer CHAPTER 18 RB ce ll Bo ss EGF EGFR Seuenless RAS-GAP R7 cell - ( STY ) ~ AS-GDP ( ~ sos ) / GRB2 sos ' ~ RAF i p --- -- ..}.: ♦ --- -- Figure 18.6. The role of SOS in the pathway from EGF to RAS was found to be similar in the fruit fly and in humans, as well as other animals. This diagram shows the pathway in the fruit fly proposed by Thomas Raabe in 2000 (Raabe, 2000). I have added the corresponding human gene names in red. The DNA sequences of the fruit fly genes and the corresponding mammalian genes were s imilar, although not identical. SOS in both species stimulates the conversion the inactive form of RAS (RAS-GDP) to the active form (RAS-GTP). In humans, the input to the pathway is EGF (epidermal growth factor), which corresponds to the fruit fly's Boss gene ("bride of sevenless"). The output of the pathway from RAS via RAF, MEK, and ERK was also similar in the fruit fly and humans (compare with Figure 18.7). The known functions of the genes at the end of the pathway, however, were different: eye development in the fruit fly versus cell divis ion in humans. 361 K. W. Kohn Drugs Against cancer CHAPTER 18 The critical role of RAS genes in trans mitting s ig nals from growth factor receptors, s uch as EGFR. The RAS story expanded enormously as its role in stimulating uncontrolled division of cancer cells gradually emerged from the mist The strength of the cell division signal from RAS obviously had to be strictly controlled, because excessive cell division could lead to cancer. The control of RAS via positive and negative influences were discovered, and how it all works to control cell division gradually unfolded. The unravelling of the story began with the fruit fly eye mutation studies described above. The mutated genes were then isolated and their DNA sequenced, which revealed the amino acid sequences of the proteins encoded in the genes. In 1987, Ernst Hafen, Gerald Rubin and their coworkers at the University of California at Berkley located the sevenless gene on the fruit fly chromosomes (Hafen et al., 1987). They isolated the gene and determined its DNA sequence, from which they surmised that the gene coded for a receptor tyrosine kinase that had the structure of a trans-membrane protein. In the fruit fly eye, the sevenless protein (corresponding to EGFR in humans) on photoreceptor cell R7 bound the bride ofsevenless (Boss) protein on the adjacent cell RB. In that way, the RB cell controlled the behavior of the R7 cell. The sevenless protein in the R7 cell then signaled, by way of son ofsevenless (SOS}, down the chain to ERK, which entered the cell nucleus to activate genes. If that control was in any way defective due a mutation, the development of the eye was defective and produced abnormal structures, such as shown in Figure 18.5. Understanding of the fruit fly's signaling from sevenless accelerated in the 1990's, particularly in the laboratory ofUptal Banerjee at the University of California in Los Angeles. In 1991, they reported studies of SOS mutants that pointed to SOS being an intermediary between sevenless (corresponding to EGFR) and RAS (Rogge et al., 1991). Then in 1992, they sequenced the SOS gene and inferred that it served to activate RAS (Bonfini et al., 1992). By 1993, the chain from sevenless/EGFR via GRB2 and SOS to RAS had been worked out (Karlovich et al., 1995) (Figures 18.6). The parts (domains) of those proteins that carried out their respective bindings had also been worked out. The GRB2 protein was found to serve only as a linker between EGFR and SOS. One end of the GRB2 molecule had an 'SH2' domain that was noted to bind to phosphate groups on tyrosine amino acids of proteins. Thus, there was a sequence of links from EGFR to GRB2 to SOS to RAS. When EGFR bound to EGF, a pair of EGFR protein molecules paired up and added phosphate groups to each other's tyrosines at specific places on the proteins (described in Chapter 17). Those phosphotyrosines then bound the SH2 end of a GRB2 protein. The other end of GRB2 had an 'SH3' domain that bound a particular amino acid arrangement on SOS. That's all that GRB2 was responsible for doing. SOS, on the other hand, not only linked between GRB2 and RAS, but also stimulated the activity of RAS by facilitating the 362 K. W. Kohn Drugs Against cancer CHAPTER 18 replacement of GDP by GTP on the RAS molecule (Figure 18.6). That chain of proteins then sent signals to the R7 cell urging it to become a photoreceptor in the fruit fly eye. It is mind-blowing how nearly the same network of protein interactions in a critical control pathway exists in humans and in fruit flies. In the fruit fly, the network controls the development of the eye, whereas in humans it controls cell division. I don't know whether the fruit fly perhaps has another similar network that controls cell division, or whether humans have other networks of this kind that function in the development of the eye or other anatomical structure. Interestingly, the same network arrangement served quite different purposes. That fact of nature enabled the extraordinary connection from of fruit fly eyes to human cancer. How the receptor tyrosine kinase EGFR connects to RAS via SOS and stimulates RAS to signals the cell to divide is shown by the molecular interaction map in Figure 18.7, which builds on the map in Figure 17.6 of Chapter 17. The signal from RAS goes to the cell nucleus by way of a chain of kinase proteins (RAF, MEK, and ERK) that are used by many signaling systems in the cell. The interesting way that RAS itself is regulated was shown in Figure 18.6. That regulation is based on the fact that the RAS protein has on it a site that can bind either GTP or GDP (guanosine triphosphate or guanosine diphosphate). When RAS has GTP bound to the site, it is active and sends signals down the pathway to the cell nucleus. When, instead, GDP is bound to the site, RAS is inactive and does not send signals. SOS activates RAS by stimulating the replacement of GDP by GTP on the RAS protein. In the opposite direction, a RAS-GAP protein inactivates RAS by stimulating the conversion of the bound GTP to GDP. This balance between activation and inactivation regulates RAS and thereby regulates the strength of the signals sent down the pathway to the cell nucleus. 363 K. W. Kohn Drugs Against cancer CHAPTER 18 EGFR ErbBn (n=l,2,3,4 ) (ErbBl) L ""''~ ~vGRB2 2 ~ ~ St!muf.a tlon --I Inhibition ~ Blndlng • The entity that results from the binding: ICell Division I Figure 18.7. How SOS connects growth factor receptors with RAS in the activation of cell division. The epidermal growth factor receptor (EGFR, also known as ErbBl, see Chapter 17) becomes activated when it binds an epidermal growth factor. EGFR can then bind to another ErbB family member. The two ErbB's then phosphorylate each other's intracellular parts (domains). Many s ites are phosphorylated, but only one is shown. These events are in a red box, because some details are omitted (see Figure 17.6 in Chapter 17). The phosphorylated EGFR site then binds the adapter protein, GRB2 {1), which then binds SOS [2]. That brings SOS to the cell membrane, where both EGFR and RAS [3) are located. The combination of SOS and RAS {4) then activates RAS [SJ to send a s ignal down the RAF-MEK-ERK pathway [6, 7,8) that simulates cells to enter the cell division cycle [9). In 1984, an important discovery connected that story to human cancers. It was found that the RAS genes were often mutated in cancer and, furthermore, that the mutation blocked the conversion of the bound GTP to GDP, thereby preventing the inactivation of active RAS (Gibbs et al ., 1984). Consequently, the mutated RAS was active all the time and sent excessively strong cell division signals. Thus, when a mutant RAS gene was injected into cells, the cells divided without control, as they do in cancer (Feramisco et al ., 1984). But the question remained: why was the conversion GTP to GDP defective in the mutant RAS p rotein? The reason turned out to be that this GTPase activity, which is an integral part of the RAS p rotein, is normally activated by another protein, RAS-GTPase-activating- protein (RAS-GAP, for short). The defect in the mutant RAS was that it did not respond to RAS-GAP (Trahey and McCormick, 1987) (Vogel et al., 1988). (Like RAS, the RAS-GAP 364 K. W. Kohn Drugs Against cancer CHAPTER 18 protein binds to the inner surface of the cell surface membrane, thus localizing it to where it can efficiently interact with RAS.) HRAS and KRAS, together with NRAS, constituted the RAS family of genes of very similar DNA sequences. Taken together, mutations in one or another of the RAS genes was found in about 10% of all cancers. Of the three RAS genes, KRAS was found to be by far the most frequently mutated in cancer. Remarkably, there was one type of cancer that nearly always had a KRAS mutation: cancer of the pancreas. Other types of cancer that frequently had KRAS mutations were about 45% of colorectal cancers and about 35% of lung adenocarcinomas. HRAS was mutated in about 10% of lung adenocarcinomas. NRAS was mutated in about 15% of melanomas. I don't know (and perhaps no one knows) why RAS mutations are common in only certain types of cancer. In particular, why do pancreatic cancers almost always have a KRAS mutation? Almost all of the oncogenic mutations of RAS were at only three position in the amino acid chain of the protein (Cox et al., 2014). Moreover, the three changes each had the same effect: they prevented RAS-GAP from interacting with RAS, thereby keeping the RAS protein continually in its active GTP-bound state. In other words, the GTPase activity of RAS was unable to convert its bound GTP to GDP. Therefore, since RAS was active in its GTP-bound state, the mutated RAS protein remained active all the time and continually sent signals to the nucleus to stimulate the cell to divide. A RAS mutation by itself, however, was not enough to cause cancer -- because other proteins, particularly TP53, could stop the malignancy. To become malignant, a tumor needed one or more other defects, such as an inactivating TP53 mutation. Although we understood how these oncogenic mutations induced cells to grow into cancers, how to interfere with that process so as to provide therapy for the 10% of patients whose cancer was driven by a RAS mutation remained a big problem. It was a complex problem, in part because the RAS proteins have several important functions in the cell. Efforts to find a solution were in progress at the time of this writing. Failure of efforts to find RAS-inhibiting anticancer drugs. There were several possible ways to suppress the overactivity of mutated RAS. A drug that inhibited any of the many factors required by RAS to be active might be effective. Despite decades of efforts, however, medicinal chemists had not come up with a clinically approved drug (Cox et al., 2014). Some cancers became addicted to high RAS activity. A drug that inhibited RAS, either directly or in a downstream pathway, might be effective against those cancer cases. Research became directed mainly on KRAS-dependent cancer - where the cancer cells were addicted to high expression of KRAS. One of the first approaches was to look for drugs that would compete with GTP for binding to the mutant RAS protein. That effort failed, however, because the affinity of RAS for GTP 365 K. W. Kohn Drugs Against cancer CHAPTER 18 was too high: chemists could not find a drug molecule that could compete with that high affinity. Blocking the GTPase activity of the RAS protein was not a good idea, because it would maintain RAS in its high-activity GTP-bound state. On the other hand, a drug that worked like RAS-GTPase to convert the RAS-bound GTP to GDP would inhibit RAS activity, but attempts to find such a drug also failed. Another idea was to inhibit the binding of RAS to the cell surface membrane, because that would hinder RAS from receiving signals from EGFR, which was located in the cell surface membrane. Well then, what causes RAS to become bound to the membrane, and could that be inhibited? To enable RAS binding to the cell surface membrane, the cell has an enzyme that adds a long hydrocarbon chain to the RAS protein. The hydrocarbon chain is lipid-like and tends to merge with the lipid part of membranes, thereby carrying the RAS protein along with it to the cell surface. Inhibitors of that enzyme were therefore considered as drugs that might suppress RAS activity. The problem was that many other essential molecules rely on the same chemistry to carry them to the cell surface, and it was difficult to find a drug specific for the RAS protein. Another problem was that there were different enzymes that linked different kinds of hydrocarbon chains onto RAS and inhibiting any one of those enzymes would still allow a different enzyme to link a similarly effective hydrocarbon chain. Efforts to use this approach were rekindled based on deeper understanding of the relevant molecular complexities (Cox et al., 2015). In the face of all those difficulties and failures, RAS had become considered to be "undruggable." New technology, however, restored hope that direct targeting of RAS may yet succeed (Cox et al., 2015) (Ryan et al., 2015). The most frequent oncogenes whose over-activity drove perhaps as many as 20% of human cancers were the closely related members of the RAS family: KRAS, HRAS, and NRAS (Downward, 2015). Of those, KRAS mutations were extraordinarily common in cancers. Most remarkable was that a KRAS mutation was found in as many as 95% of patients with pancreatic cancer. In addition, such mutations were found in about 40 % of patients with colorectal cancer and in 20% to 25% of patients with adenocarcinoma of the lung. The KRAS story. A KRAS oncogene was discovered in 1983 by Manning Der and Geoffrey Cooper of Harvard Medical School. They discovered an abnormal protein in cancer cells, made by a mutated gene that produced cancer upon transfecting the gene into non-cancer cells. The mutated gene thus was an oncogene -- which they identified as a mutant KRAS (Der and Cooper, 1983). Much time and effort was needed, however, to find out what overactive KRAS did to make cells cancerous. In 2009, Jeff Settleman and his colleagues showed that cell lines derived from human lung or pancreas cancers differed in the degree to which they were addicted to KRAS (Singh et al., 2009). Thus, if KRAS or its downstream pathway were blocked by a drug or other 366 K. W. Kohn Drugs Against cancer CHAPTER 18 means, the cancer cell should usually die. They thought that the addiction might make those cancers vulnerable to specific drugs, and they set about investigating whether that approach could lead to drugs that were effective against cancers whose KRAS was overactive, and the cells had become addicted to it. Figure 18.8 shows how they identified cell lines that were highly addicted and that could perhaps be targeted by specific drugs. In order to determine the degree of addiction, they first suppressed the production of KRAS by inserting into the cells a small hairpin RNA (shRNA) that specifically blocked the KRAS messenger-RNA, thereby blocking the production of KRAS protein. Then, they looked to see whether the cells were dying. which would indicate that the cells were addicted and would not be able to survive without KRAS. They did that by measuring the amount of cleaved caspase-3 protein that was produced when KRAS was suppressed. A central feature of cell death by apoptosis was the cleavage of the caspase-3 protein (it is broken into two pieces that then come together in a new configuration to generate an active caspase-3 enzyme that starts the apoptosis process). Since attempts to develop a KRAS-inhibiting drug had failed, the investigators thought that inhibiting a step downstream from KRAS might work. They therefore set out to investigate the molecular changes occurring when KRAS was artificially suppressed using an sh RNA Although such RN A's may not become useful drugs, researchers did not give up trying to target RAS. Among many efforts to apply new molecular techniques was the possibility of engineering antibodies that would specifically target mutant KRAS protein inside the cell (Shin et al., 2020). ., .c When KRAS .,::; When KRAS 0 was suppressed ~ was suppressed !>sh-K-Ras gsh-K-Ras ~ A BC 8ABC < 1- - I E - - - !K-Ras AmountofKRAS I - - - - I 1- - - !c1eaved Casp-3 Cell li ne NOT Cell line addicted Sign of apoptosis addicted to KRAS to KRAS Figure 18.8. An example of two human cancer cell lines that differed in whether addicted to KRAS (Singh et al., 2009). The cell line on the right was addicted to KRAS: when the experimenters suppressed KRAS, the cells died by apoptosis. The cell line on the left was not addicted to KRAS: when the experimenters suppressed KRAS, the cells did not die. In order to tell whether or not the cells were KRAS-addicted, they measured the caspase-3 cleavage product If the amount of cleaved Casp-3 protein increased upon addition of sh-KRAS to suppress KRAS production (right), it indicated that the cells were dying because of addiction to KRAS. If there was no increase in Casp-3 cleavage (left), it indicated that the cells were not addicted 367 K. W. Kohn Drugs Against cancer CHAPTER 18 How the activity ofKRAS is regulated. Knowledge of how KRAS is regulated was thought to open new opportunities for therapy. Since the RAS proteins stimulate important processes, such as cell division, the cell must regulate their activity. Without RAS regulation, cancer may ensue. Most attention was given to the one that is most prominent in cancer, KRAS. What had to be regulated was the balance between KRAS-GTP and KRAS-GDP, where the former was active and the latter was inactive. The active KRAS-GTP would stimulate the first member of the downstream pathway, which is the protein RAF, from which the stimulation signal may proceed all the way to the genes that promote cell division (Figure 18.7). A mutation of RAF could, by the way, stimulate the downstream pathway to cell division independent of RAS - in fact, it leads to melanoma, a story that is told in the next chapter. It turned out that the controlling factor for the GTP /GDP regulation of KRAS was the son of sevenless (SOS) discovered in fruit flies but of similar function in humans. How it works was nicely shown in 2015 by Channing Der as a cycle that he referred to as the beating heart of cancer (Figure 18.9) - reflecting that about 1 in 7 cancers were driven by dysregulation of KRAS (Kessler et al., 2021). SOS would be the cycle's pacemaker (Figure 18.9). The dominant role of the SOS protein in the cycle is shown by it being about 7 times as large as KRAS, consisting of 1333 amino acid, compared with only 189 for KRAS - thus KRAS is only about the size if a typical SOS domain (Figure 18.10). According to Der's model (Figure 18.9), the cycle begins with GDP-bound KRAS (KRAScoP), which is the "offstate". KRAScopthen binds to the CDC25H domain of SOS (Figure 18.10), where the bound GDP (guanosine diphosphate) is replaced by GTP (guanosine triphosphate), which yields the "on state" KRAScrP . The replacement of GDP by GTP is accelerated when another KRAScrp molecule is bound to an allosteric site of SOS at its REM domain (Figure 18.10). ("Allosteric" is a change in a protein's shape that affects a distant site on the same molecule.) Finally, KRAScrp slowly removes the high energy phosphate at the end of the chain of three phosphates of the GTP, converting it to GDP: back to the "off state". KRAS has an intrinsic GTPase activity that slowly does that. But the rate of GTP-to- GDP conversion is greatly increased by a GTPase-accelerating protein (GAP) (Figure 18.9). Much research effort went into elucidating what SOS does and the conformational changes this large protein undergoes (Figure 18.10). SOS was thought a potential target for the development of inhibitor drugs (Hofmann et al., 2020). 368 K. W. Kohn Drugs Against cancer CHAPTER 18 KRASGDP "offstate H KRASGr, "on state" o.irren1Opinion rl Qlomcel 8iology Figure 18.9. Channing Der's KRAS cycle that he thought of as the beating heart ofcancer, with SOS as pacemaker. See text for explanation. From (Kessler et al., 2021) with additional labels. "" 1150S1 Figure 18.10. The domains of SOS and how they may interact with RAS and elements in the cell surface membrane (Baltanas et al., 2020). Starting at the carboxy (C) end of SOS (right end of the upper diagram), we come first to a proline-rich domain (PR), which is shown in the lower diagram as bound to the GRB2 protein, which in turn is bound to EGFR. (GRB2 has an SH3 domain that binds PR and an SH2 motif that binds a tyrosine-phosphate on EGFR) We come next to a CDC25H domain that binds RAScoP and replaces the GDP with GTP. Then, there is a RAS-exchanger motif (REM) where a RAScTP can bind and accelerate the replacement of GDP with GTP. Finally, there are some positively charged domains that bind to negatively charged placed on the membrane, thereby stabilizing the membrane binding of SOS. From (Baltanas et al., 2020). 369 K. W. Kohn Drugs Against cancer CHAPTER 18 How RAS mutations lead to cancer. How marvelous that human evolution of large brain and communal societies building knowledge over generations has already led us to glimpse the molecular functioning of our own bodies, their disorders and remedies. Those thoughts came to mind on contemplating the KRAS cycle, how its derangement leads to cancer, and at least in one limited circumstance to a road to therapy (Figure 18.9). A path to cancer happens when cells have excessive amounts of KRAS in its "on state", KRAScrP, where KRAS has GTP bound -- because KRAScTP signals cell division and must be controlled to avoid the excessive cell growth of cancer. Excessive KRAScrp could accumulate if the rate of KRAScrPproduction is too high or if the rate of its loss by conversion to the "off state", KRAScoP, is too low. It turns out that it is the latter case that is most often the trouble. In particular, it is because the mutant KRAS has lost its ability to bind well to "GTPase accelerating protein" or GAP that accelerates the conversion of KRAScrP to KRAScop. The mutant KRAS therefore accumulates in its "on state", KRAScTP, and stimulates excessive cell division. A drug that targets a particular KRAS mutation was designed using detailed knowledge of the chemistry and molecular structure of the mutant protein. The new drug combined covalent and non-covalent binding designed specifically to fit the mutant protein's structure and may be a step forward in the design of targeted drugs. Its story follows. Mutant KRAS as anticancer drug target. I have for the most part focused on history because current events often soon become obsolete, but make an exception now because of the recent molecular design of a drug that targets specifically a particular mutation of the KRAS protein and binds tightly, apparently covalently, only with the protein that has that mutation and only when KRAS is in the GDP- bound state. The drug was deemed so promising that just two months ago, in March 2021, the U.S. Food and Drug Administration (FDA) granted it a Breakthrou,qh Therapy designation as the first promising anticancer drug targeting a KRAS mutation. The drug is Amgen's AMGSl 0 "Sotorasib" that specifically inhibits the KRAS G12C mutant The preliminary approval was for treatment of patients with locally advanced or metastatic non-small cell lung cancer (NSCLC) with the KRAS G12C mutation who had received at least one prior systemic therapy. By blocking the KRAsc12c mutant in its GDP-bound state, the drug prevented the replacement of GDP with GTP (Figure 18.9, the part in blue labelled SOSl). The drug thus traps KRAsc12c in an inactive state and prevents it from sending signals that would stimulate cell division. The remarkable specificity of the drug comes from a combination of two factors (Canon et al., 2019). First, it fits and binds in a hydrophobic groove in the protein with a geometry 370 K. W. Kohn Drugs Against cancer CHAPTER 18 specific to the mutant protein in its GDP state (Figure 18.11). Second, the drug can bind covalently to a sulfur atom of a cysteine that is only present in the mutant protein - because the mutation puts a cysteine in place of a glycine at position 12 in the amino acid sequence (Figure 18.12). A phase I trial of sotorasib was conducted in 129 patients who had advanced cancers with a KRAS G12C mutation (non-small-cell lung (59), colorectal (42), or other (28)) (Hong et al., 2020). Of the 129 patients in the study, 88% had evidence of response or had stable disease. The cancer was held in check (progression-free survival) for a median of 6 months. However, 12% of the patients had serious side-effects, perhaps due in part to the alkylation and hydrophobic binding abilities of the drug to attack sites on normal cell components - a side-effect that might be reduced by modifying the drug's structure. Better results might be expected after further studies to determine optimal dosage and to add drugs that could prevent resistance to the drug. Still, the drug would be effective only in the relatively low, albeit significant, fraction of patients who have cancers with that particular KRAS mutation. AMG 510 Figure 18.11. The structure of sotorasib (AMG510) and how it fits in a groove in the G12C mutant KRAS protein (Canon et al., 2019). The cysteine that replaces the glycine at position 12 in the mutant protein is shown in yellow. The carbon atom double-bonded shown at the upper end of the structure (left) is close to the sulfur atom of cysteine-12 (right) . The hydrophobic part of the drug fits nicely in a hydrophobic groove in the protein. 371 K. W. Kohn Drugs Against cancer CHAPTER 18 KRA5Gl2C Cys12 S--H Q mutant protein :~ .H H-(cl_j .. ~) ) -~ ~ I w,e ~/ ~ 'f\O 'fr~ w,e Sotorasib o~~ I w,e \J'.8 ~ Figure 18.12. How sotorasib could bind covalently to the sulfur of the cysteine that replaces the glycine in the G12C mutation of KRAS. A hydrogen bond from the KRAS protein (red color) helps sotorasib to suck in an electron from the sulfur atom (yellow color) to from a covalent bond between drug and protein. Summary The three RAS genes are the most frequently mutated genes that drive human cancer - they are the most frequent oncogenes activated by mutation. Their importance drove strong efforts to develop inhibitors of the overactive RAS functions. However, these efforts, extending over more than three decades, were disappointing, giving rise to the opinion that mutant RAS proteins were "undruggable." Armed with new technology and deeper understanding of the complexities of RAS functions, attempts to develop therapy targeted against RAS oncogenes were renewed (Papke and Der, 2017). Earlier studies - before 2015 -- had revealed that the strength of signals from RAS proteins depended on control of RAS activity. RAS proteins send signals to the cell n ucleus to initiate cell division, but this happens only when RAS is in its GTP-bound state. Importantly, the amount of RAS-GTP was tightly controlled, so that cells did not divide too often. That was accomplished by control of RAS cycling between the active GTP-bound state and the inactive GDP-bound state. This Chapter looked back at how mutations of fruit fly genes led to the discovery of h uman versions of genes functioning in an analogous pathway. The fruit fly protein altered by the sevenless mutation was found to be a receptor tyrosine kinase that corresponded to 372 K. W. Kohn Drugs Against cancer CHAPTER 18 human EGFR (Simon et al., 1991). The Son ofsevenless (SOS) mutation was especially revealing, because it disclosed previously unknown genes that turned out to be central to the cause and treatment of many human cancers. Particularly important was the discovery of the RAS genes. The relevance of SOS to cancer was shown by finding that it transmits signals from EGFR that activate RAS. RAS in turn activates RAF (the topic of Chapter 19). It is remarkable how that arcane route from fruit fly eye mutations to the RAS oncogenes, together findings about cancer-causing viruses, led to the discovery of human oncogenes and their importance in cancer cause and treatment. Who would have imagined that the SOS gene of fruit flies would become thought of as the pacemaker of a beating heart of cancer? The most important of the RAS proteins was found to be KRAS whose mutations are prominent in several common cancers. KRAS therefore became a major focus of research, and efforts were made to find drugs that would be effective against KRAS-mutated cancers. References Baltanas, F.C., Zarich, N., Rojas-Cabaneros, J.M., and Santos, E. (2020). SOS GEFs in health and disease. Biochim Biophys Acta Rev Cancer 1874, 188445. Bonfini, L, Karlovich, C.A., Dasgupta, C., and Banerjee, U. (1992). The Son of sevenless gene product: a putative activator of Ras. Science 255, 603-606. Canon, J., Rex, K., Saiki, A.Y., Mohr, C., Cooke, K., Bagal, D., Gaida, K., Holt, T., Knutson, C.G., Koppada, N., et al. (2019). The clinical KRAS(G12C) inhibitor AMG 510 drives anti- tumour immunity. Nature 575, 217-223. Cox, A.D., Der, C.J., and Philips, M.R (2015). Targeting RAS Membrane Association: Back to the Future for Anti-RAS Drug Discovery? Clinical cancer research : an official journal of the American Association for Cancer Research 21, 1819-1827. Cox, A.D., Fesik, S.W., Kimmelman, A.C., Luo, J., and Der, C.J. (2014). Drugging the undruggable RAS: Mission possible? Nature reviews Drug discovery 13, 828-851. De Petris, S., and Harvey, J.J. (1969). Presence of Unusual Virus Particles in Two Hamster Tumour Tissue culture cell lines Induced by Murine Sarcoma Virus. J Gen Virol 5, 561- 54. Der, C.J., and Cooper, G.M. (1983). Altered gene products are associated with activation of cellular rasK genes in human lung and colon carcinomas. Cell 32, 201-208. Downward, J. (2015). RAS Synthetic Lethal Screens Revisited: Still Seeking the Elusive Prize? Clinical cancer research : an official journal of the American Association for Cancer Research 21, 1802-1809. Feramisco, J.R, Gross, M., Kamata, T., Rosenberg. M., and Sweet, R W. (1984). Microinjection of the oncogene form of the human H-ras (T-24) protein results in rapid proliferation of quiescent cells. Cell 38, 109-117. Gibbs, J.B., Sigal, I.S., Poe, M., and Scolnick, E.M. (1984). Intrinsic GTPase activity distinguishes normal and oncogenic ras p21 molecules. Proceedings of the National Academy of Sciences of the United States of America 81, 5704-5708. 373 K. W. Kohn Drugs Against cancer CHAPTER 18 Hafen, E., Basler, K., Edstroem, J.E., and Rubin, G.M. (1987). Sevenless, a cell-specific homeotic gene of Drosophila, encodes a putative transmembrane receptor with a tyrosine kinase domain. Science 236, 55-63. Hager, G.L., Chang. E.H., Chan, H.W., Garon, C.F., Israel, M.A., Martin, M.A., Scoinick, E.M., and Lowy, D.R. (1979). Molecular cloning of the Harvey sarcoma virus closed circular DNA intermediates: initial structural and biological characterization. Journal of virology 31, 795-809. Harvey, J.J. (1964). An Unidentified Virus Which Causes the Rapid Production of Tumours in Mice. Nature 204, 1104-1 105. Hofmann, M.H., Gmachl, M., Ramharter, J., Savarese, F., Gerlach, D., Marszalek, J.R., Sanderson, M.P., Kessler, D., Trapani, F., Arnhof, H., et al. (2020). BI-3406, a Potent and Selective SOSl -KRAS Interaction Inhibitor, Is Effective in KRAS-Driven Cancers through Combined MEK Inhibition. Cancer Discov. Hong. D.S., Fakih, M.G., Strickler, J.H., Desai, J., Durm, G.A., Shapiro, G.I., Falchook, G.S., Price, T.J., Sacher, A., Denlinger, C.S., et al. (2020). KRAS(G12C) Inhibition with Sotorasib in Advanced Solid Tumors. The New England journal of medicine 383, 1207-1217. Karlovich, C.A., Bonfini, L., Mccollam, L., Rogge, R.D., Daga, A., Czech, M.P., and Banerjee, U. (1995). In vivo functional analysis of the Ras exchange factor son of sevenless. Science 268, 576-579. Kessler, D., Gerlach, D., Kraut, N., and McConnell, D.B. (2021). Targeting Son of Sevenless 1: The pacemaker of KRAS. Curr Opin Chem Biol 62, 109-118. Kirsten, W.H., and Mayer, L.A. (1967). Morphologic responses to a murine erythroblastosis virus. Journal of the National Cancer Institute 39, 311-335. Lowenstein, E.J., Daly, R.J., Batzer, A.G., Li, W., Margolis, B., Lammers, R., Ullrich, A., Skolnik, E.Y., Bar-Sagi, D., and Schlessinger, J. (1992). The SH2 and SH3 domain-containing protein GRB2 links receptor tyrosine kinases to ras signaling. Cell 70, 431-442. Papke, B., and Der, C.J. (201 7). Drugging RAS: Know the enemy. Science 355, 1158-1 163. Raabe, T. (2000). The sevenless signaling pathway: variations of a common theme. Biochimica et biophysica acta 1496, 151-163. Rogge, R.D., Karlovich, C.A., and Banerjee, U. (1991 ). Genetic dissection of a neurodevelopmental pathway: Son of sevenless functions downstream of the sevenless and EGF receptor tyrosine kinases. Cell 64, 39-48. Ryan, M.B., Der, C.J., Wang-Gillam, A., and Cox, A.D. (2015). Targeting RAS-mutant cancers: is ERK the key? Trends in cancer 1, 183-198. Santos, E., Tronick, S.R., Aaronson, S.A., Pulciani, S., and Barbacid, M. (1982). T24 human bladder carcinoma oncogene is an activated form of the normal human homologue of BALB- and Harvey-MSV transforming genes. Nature 298, 343-347. Scolnick, E.M., and Parks, W.P. (1974). Harvey sarcoma virus: a second murine type C sarcoma virus with rat genetic information. Journal of virology 13, 1211-1219. Scolnick, E.M., Rands, E., Williams, D., and Parks, W.P. (1973). Studies on the nucleic acid sequences of Kirsten sarcoma virus: a model for formation of a mammalian RNA- containing sarcoma virus. Journal of virology 12, 458-463. Shin, S.M., Kim, J.S., Park, S.W., Jun, S.Y., Kweon, H.J., Choi, D.K., Lee, D., Cho, Y.B., and Kim, Y.S. (2020). Direct targeting of oncogenic RAS mutants with a tumor-specific cytosol- penetrating antibody inhibits RAS mutant-driven tumor growth. Sci Adv 6, eaay2174. 374 K. W. Kohn Drugs Against cancer CHAPTER 18 Sigal, LS., Marshall, M.S., Schaber, M.D., Vogel, U.S., Scolnick, E.M., and Gibbs, J.B. (1988). Structure/function studies of the ras protein. Cold Spring Harbor symposia on quantitative biology 53 Pt 2, 863-869. Simon, M.A., Bowtell, D.D., Dodson, G.S., Laverty, T.R., and Rubin, G.M. (1991). Rasl and a putative guanine nucleotide exchange factor perform crucial steps in signaling by the sevenless protein tyrosine kinase. Cell 67, 701-716. Simons, P.J., Bassin, R.H ., and Harvey, J.J. (1967). Transformation of hamster embryo cells in vitro by Murine Sarcoma Virus (Harvey). Proc Soc Exp Biol Med 125, 1242-1246. Singh, A., Greninger, P., Rhodes, D., Koopman, L., Violette, S., Bardeesy, N., and Settleman, J. (2009). A gene expression signature associated with "K-Ras addiction" reveals regulators of EMT and tumor cell survival. Cancer cell 15, 489-500. Trahey, M., and McCormick, F. (1987). A cytoplasmic protein stimulates normal N-ras p21 GTPase, but does not affect oncogenic mutants. Science 238, 542-545. Vogel, U.S., Dixon, RA, Schaber, M.D., Diehl, R.E., Marshall, M.S., Scolnick, E.M., Sigal, LS., and Gibbs, J.B. (1988). Cloning of bovine GAP and its interaction with oncogenic ras p21. Nature 335, 90-93. Willingham, M.C., Pastan, I., Shih, T.Y., and Scolnick, E.M. (1980). Localization of the src gene product of the Harvey strain of MSV to plasma membrane of transformed cells by electron microscopic immunocytochemistry. Cell 19, 1005-1014. 375 K.W.Kohn Drugs Against Cancer CHAPTER 19 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@nih gov CHAPTER 19 The BRAF-melanoma story. Introduction Melanoma is a dangerous skin cancer often caused by excessive exposure to the sun. Unless surgically removed while the tumor is still small, the malignancy eventually spreads, leading to a fatal outcome. The lifetime risk of developing melanoma in the United States in 1999 was about one chance in 75 (Atkins et al., 1999). Early attempts to treat advanced melanoma with chemotherapy or immunotherapy only produced responses in about one-fifth of the patients and the responses rarely lasted more than a few months. Later years unveiled dramatic new ways to attack the disease. One of the most important was the discovery of the role of mutations of the BRAF oncogene. Equally important were discoveries about how the anti-cancer immune system is controlled. In 2012, both areas of investigation Jed to dramatic new paths of treatment for melanoma and other malignant tumors. The state of knowledge as of 2002, which is where this chapter begins, is summarized in Figures 19.1 and 19.2. Before relating how all that developed over the years, however, here is how the story of BRAF mutation fits in the current sequence of chapters. Chapter 17 told the story of epidermal growth factor receptor (EGFR) mutations and signaling to RAS, and Chapter 18 told the story of RAS mutations and signaling to RAF. The current chapter (Chapter 19) focusses on RAF mutations. Researchers found opportunities for therapy all along the signaling chains from EGFR (epidermal growth factor receptor) to the molecules that initiate cell proliferation (Figures 19.1 and 19.2). As information about this signaling accrued 376 K.W. Kohn Drugs Against Cancer CHAPTER 19 over the years, it became more and more complex, as might be expected for a crucial biological system whose activity takes account of many conditions in the cell. A more recent view of the main part of the pathway and the mutations common in malignant melanoma is shown in Figure 19.3. Cell surface membrane Cell MYC division Figure 19.1. Pathway from receptor tyrosine kinase, EGFR, via the MAP kinase (mitogen- activated protein kinase) cascade to cell division-enhancing genes, as understood in 2002. A growth factor, such as epidermal growth factor, EGF, binds and stimulates EGFR, which in turn stimulates RAS via SOS (the human version of the fruit fly's "son of sevenless• - related in Chapter 18). RAS then stimulates BRAF (the most prominent member of the RAF family), which is mutated in about half of melanoma cases and is what drives the malignancy in those cases. Mutant BRAF is many times more active than the normal BRAF, and the mutant's activity is independent of RAS. BRAF then stimulates MEK and ERK, which are MAP kinases (mitogen-activated protein kinases). The signal is amplified at each step and then finally reaches growth factors, such as MYC, which stimulate the expression of cell division genes. 377 K.W. Kohn Drugs Against Cancer CHAPTER 19 RAS RAF Ki:nases e.g . Src, PAK EGF rece EGFR Substrates in cytosol and cyt0$kele1on "" Nucleus Figure 19.2. The state of knowledge in 2002 about how the EGFR-RAS-RAF-MEK-ERK pathway stimulates cell d ivision. Signaling through the pathway was made efficient by all the components being held near each other. EGFR, SOS, RAS, and RAF are kept close to each other by binding to the cell surface membrane. RAF, MEK, and ERK are kept adjacent to each other through binding to a scaffold protein (KSR). (From (Kolch et al., 2002) with labels in red added.) 378 K.W.Kohn Drugs Against Cancer CHAPTER 19 RTK Ligand Receptor Tyrosine Kinase Plasma M embrane l 10-15% . __, ◄Hk.P- 20-30% l 40-50% ~ l ~ l ~ l Cell growth and proliferation Figure 19.3. A recent overview of the pathway from receptor tyrosine kinases ( such as EGFR) to cell proliferation, showing the frequencies of the mutations in melanomas Qenkins and Fisher, 2020). The large majority of melanoma cases had one or another of these mutations. A recent finding was that the RAF proteins function as homo- or heterodimers. The COSMIC database of the Sanger Center listed the following frequencies of mutations in malignant melanoma (courtesy of Dr. Silvio Parodi): BRAF 45%, NF1 20%, and NRAS 17%. Also frequently mutated were LRP18 36 %, FAT4 (29 %), PTPRT 25 %, GRIN2A 23 %, and others that may confer resistance to inhibitors of the NRAS-BRAF pathway. Discovery of the BRAF oncogene and its role in melanoma. In 2002, an international group Jed by researchers in the UK aimed to discover new cancer-causing mutations, i.e., oncogenes (Davies et al., 2002). They did so by looking for DNA sequence differences between cancer and normal cells, focusing on the pathway from EGFR via RAS, RAF and MEK to ERK, a pathway that was known to push cell proliferation, a hallmark of cancer (Figure 19.1). Since RAS-mutant genes had been discovered to be oncogenes, they thought that other genes in the pathway might also sometimes be mutated to become oncogenes. The UK investigators then went on to make a remarkable discovery: one of the three RAF genes, namely BRAF, was mutated in about half the cases of malignant melanomas (Davies et al., 2002)! Moreover, all of the mutations affected amino acids 379 K.W.Kohn Drugs Against Cancer CHAPTER 19 in the protein kinase region of the BRAF protein, the region that activates MEK by adding phosphate groups to it And 80% of the BRAF mutations were at a single site (V600E), where the mutation replaced an uncharged valine (V) with a negatively charged glutamate (E). They also found that the mutated BRAF protein was continually overactive and did not need activation signals from RAS or anything else. The negatively charged glutamate was thought to mimic a negatively charged phosphate that normally activated BRAF (Vogelstein and Kinzler, 2004) (Sala et al., 2008). Since the glutamate was an integral part of the mutant BRAF protein, it was presumed to continually activate the protein's enzyme (kinase) activity. BRAF was known to be a protein kinase that phosphorylates and thereby activates MEK in the pathway to ERK and cell division, thereby over-stimulating the pathway to cell division and cancer (Figure 19.1). Research then focused on mutations of BRAF, the first discovered melanoma- associated mutation. It was reasonable to suppose that BRAF-inhibitor drugs would have therapeutic potential, particularly against those melanomas that were dependent on or addicted to an overactive BRAF. But many questions had to be considered about whether or under what circumstances targeting BRAF mutation might become effective treatment BRAFmutation was found in about half of melanoma cases. But many questions remained. Was the mutation a cause of the malignancy, or an innocent bystander? Melanoma was known to be induced by sun exposure, but did sun exposure cause the mutation? Did the mutation occur also in benign melanocytic nevi (black birthmarks)? Did the mutation sometimes arise during the progress of the malignancy? In the tumors that had the BRAF mutation, did all of the cells have the mutation? These questions were debated for the first few years after the discovery of the BRAF mutation, until evidence and answers to most of them gradually emerged. The V600E mutation of the BRAFgene was of the kind expected from exposure to sunlight because it entailed replacement of a T (thymine) by an A (adenine) in DNA, T being the nucleotide most susceptible to chemical change by ultraviolet light BRAFmutation was already present in many benign congenital melanocytic nevi and in benign nevi acquired due to sun exposure. The nevi usually remained benign, although they occasionally did go on to malignant melanoma (Roh et al., 2015). BRAFmutation therefore appeared to predispose a benign nevus to become malignant but did not by itself cause the malignancy. Some melanoma tumors seemed to be composed of a mixture of cells that did and did not have the BRAF mutation (Helias-Rodzewicz et al., 2015). Researchers at the Cancer Institute in Santa Monica California found that the frequency of BRAF mutation was often higher in melanoma metastases than in the primary tumors 380 K.W.Kohn Drugs Against Cancer CHAPTER 19 (Shinozaki et al., 2004). Investigating further, they were able to determine in 13 patients whether the mutation was present in the primary and metastatic tumors in the same patient. Of the 13 patients, 4 had the mutation in both the primary and metastatic tumors; however, 4 other patients had the mutation in a metastasis, but not in the primary tumor (Figure 19.4). It seemed that a malignant melanoma was sometimes initiated by sun exposure producing a BRAF mutation in a benign nevus. Surprisingly, however, the mutation was occasionally found in the primary tumor but not in a metastasis (Sakaizawa et al., 2020) (Figure 19.5). In such cases, the primary tumor (where the malignancy began) might have had a mixture of BRAF mutant cells and cells were driven to malignancy by a different mutation. Melanomas sometimes occurred in members of families predisposed to the disease. In 2003, Peter Meyer and his colleagues in Tuebingen, Germany, reported that there was no RAF mutation in the melanomas of patients who had close relatives who had developed melanomas (Meyer et al., 2003). Instead, their inherited predisposition to melanoma may have come from a different mutation. So, we see that the BRAF mutation story had a number of variations and complications. Nevertheless, the main picture that emerged was that the malignancy of melanomas was driven by mutated genes, about half of them mutated BRAF. That gave strong incentive to develop inhibitors of BRAF's protein kinase activity. BRAF mutation o f BRAF mutation Patient primary tumor in metastasis I Mutant Mutant 2 Mutant Mutant 3 Mutant Mutant 4 Mutant Mutant 5 Wild t)•pe Mutant 6 Wi ld type Mutant 7 Wi ld type Mutant 8 Wild type M utant 9 Wi ld type M utant 10 Wi ld type Wild type 11 Wi ld type W ild type 12 Wild type W ild type 13 Wild type Wild type Figure 19.4. Shinozaki and coworkers determined whether a BRAF mutation was present in the primary tumor and in a lymph node metastasis of the same melanoma patient (Shinozaki et al., 2004). Cases 5-9 suggested that the mutation arose in cells that produced or arose in the metastases. ("Wild type" is jargon from bacterial genetics. Here it means the normal case.) 381 K.W.Kohn Drugs Against Cancer CHAPTER 19 A C A G T G AAA . Mutation present in the melanoma primary tumor. Mutation absent in a metastasis. Figure 19.5. A case of malignant melanoma where the primary tumor had the BRAFV600E mutation (both the red and green peaks elevated), but a metastasis did not have the mutation (only the red peak elevated) (Sakaizawa et al., 2020). The arrow points to the mutated nucleotide at position 600: red, the normal nucleotide (T); green, the mutated nucleotide (A). The primary tumor had a both a mutated and a wild-type allele, whereas the metastasis had only wild-type alleles. This patient's metastases might have arisen from a different mutation. Chemotherapy of metastatic melanoma. Chemotherapy of metastatic melanoma began Jong before anything was known about the mutations. The earliest chemotherapy of melanoma began with a drug, initially called DTIC, later renamed dacarbazine (see Figure 2.5 in Chapter 2). It was synthesized in 1961 by Y. F. Shealy, John A. Montgomery and their coworkers at the Southern Research Institute and drew the attention of investigators when studies at the National Cancer Institute found the drug to be active against several mouse tumors. Initial clinical studies suggested that dacarbazine might be active against malignant melanoma (Carter and Friedman, 1972). Several further clinical trials, however, found that only about 20% of the patients responded to the drug, the responses lasted only a few months, and long survival was rare (Gerner et al., 1973). Dacarbazine, however, was the standard of care for metastatic melanoma, because there was nothing better. By the end of the 20 th century, most patients with metastatic melanoma, despite treatment with dacarbazine, survived less than one year, and combinations of dacarbazine with other drugs were no better than dacarbazine alone (Sirott et al., 1993) (Chapman et al., 1999). 382 K.W.Kohn Drugs Against Cancer CHAPTER 19 Before 2011, the only FDA-approved drugs for the treatment of metastatic melanoma were dacarbazine and immune system regulators of the interferon type. Few patients responded, toxicities were high, and there was little increase in survival time. However, in 2011 and the decade following, therapeutic options increased dramatically, and metastatic melanoma patients began to have a more optimistic outlook. Two major advances were central to this advance. First, new chemotherapy drugs specifically targeted to the RAF and MEK components of the pathway from RAS to the cell proliferation stimulators in the cell nucleus. Second, new immunotherapy that increased the ability of cells of the immune system to act against the tumors. Search for a specific inhibitor of the V600E mutant BRAF. Researchers knew that cancers are often driven by overactivity of the pathway from receptor tyrosine kinases such as EGFR via RAS, RAF, MEK, and ERK to stimulation of cell division (Figure 19.1). An early attempt to inhibit this pathway in cancer patients used a drug called sorafenib that was originally aimed to inhibit RAF, but turned out to inhibit several different kinases as well. The drug did help against liver and kidney cancers, but, disappointingly, was useless for melanoma (Pratilas and Solit, 2010). What was needed for melanoma was a drug that inhibited specifically mutant BRAF and nothing else. That was asking for a lot! The drug would have to distinguish between similar sites in many protein kinases and also to distinguish between mutant and normal BRAF that differed by only a single amino acid change. The challenge was undertaken by a consortium of investigators who reported their work in 2008 (Tsai et al., 2008). It was an extensive and intensive investigation with several stages of screening and molecular characterization that eventually yielded a highly specific inhibitor of the V600E BRAF mutant (which, as a reminder, had glutamate in position 600 in place of valine and was the most common mutation in melanoma). The success of this project showed how specific inhibitor drugs can be designed based on molecular structure analyses of the binding-interactions between drug and the amino acids at the active site of the enzyme and screening large numbers of candidates. The researchers began by screening 20,000 small molecules for their ability to inhibit protein kinases among a large panel of both tyrosine- and serine/threonine- protein kinases. Of that large set of compounds, they found 238 that inhibited three of the kinases by about 30%. They then co-crystallized those compounds with the kinases to determine the molecular structure of how the compounds bound to the kinases (they used certain kinases that they found were relatively easy to crystallize). About half of the structures bound at the kinase site on the protein and revealed two hydrogen bonds between inhibitor and amino acids at the kinase site. 383 K.W.Kohn Drugs Against Cancer CHAPTER 19 The structures of the active sites of the different kinases were similar enough to draw conclusions that would apply to many kinases. At this early stage, specificity for a particular kinase was not yet an aim. In order to obtain specificity, a compound that seemed suitable as an initial structure was modified by adding molecular groups to optimize how an inhibitor would fit at the active site of a particular kinase, aiming for best fit to the mutant BRAF. The binding of one of the intermediate compounds to the active site of a kinase is shown on the left panel of Figure 19.6. The Figure shows two of the compounds and how they bound to kinase domains along the way to the specific mutant-BRAF inhibitor vemurafenib. The paper reporting this long and tedious effort by many researchers had as many as 38 coauthors (Tsai et al., 2008). Vemurafenib did well in clinical trials against advanced melanomas having the V600E-mutated BRAF and became standard treatment, although bedeviled by the development of resistance to the drug. The progression-free survival of the patients was usually about 6 months, which was better than the previously standard treatment with dacarbazine, for which progression-free survival was only 1 or 2 months (Figure 19.7) (Chapman et al., 2011). Clinical trials showed that vemurafenib held BRAF-mutated metastatic melanomas in check for a median of 6 months, and overall survival was a median of about 14 months (Dossett et al., 2015). Vemurafenib was the first drug that targeted BRAF (V200E) mutant metastatic melanoma and was approved for treatment of those cancers by the U.S. Food and Drug Administration (FDA) in 2011. Although the median extension ofprogression- free survival was only 6 months, some patients survived up to 18 months. The cancer then became resistant to the drug and resumed its growth. Something obviously had to be done to counter the resistance to vemurafenib that limited the effectiveness of the drug. The first step, however, was to find out what caused the resistance. 384 K.W. Kohn Drugs Against Cancer CHAPTER 19 Compound 2 PLX4720 bound bound to FGFRl to BRAF F 2 Vemurafenib Figure 19.6. Two of the candidate structures on the way to vemurafenib, the specific inhibitor of the V600E mutant form ofBRAF. Upper left shows the authors' compound 2 (yellow) and how it bound to the kinase domain of the fibroblast growth factor receptor, FGFRl. Upper r ight shows their compound PLX4720 and how it bound to the kinase domain of BRAF. (Notice the s im ilarity of the protein structure around the kinase s ite of the two d ifferent kinases.) Below is the structure ofvemurafenib. The red ovals show the SO2- containing group that was added to compound 2 to make PLX4720. To make vemurafenib, an additional benzene r ing (blue oval) was added to PLX4720: that is what made the drug specific for the V600E mutation. ("Vemurafenib" can be parsed as follows: VE[valine- glutamate)-mu[mutant]-raf[RAF)-enib[inhibitor).) (From (Tsai et al., 2008) with colored ovals added.) 385 K.W.Kohn Drugs Against Cancer CHAPTER 19 Progrtssion,free Survival 100-k:----''-. Haurd r31io, 0.26; 9S% 0, 0.2010 0.3l; 90 P<0.001 g 30 •.,~ ~ 70 60 .. "'• ~o 0 < ,o ·i [ 30 .. e 20 10 0 0 6 8 910 11 12 Months Figure 19.7. Vemurafenib was far better than standard dacarbazine therapy for treatment of metastatic melanomas that had the V600E BRAF mutation (phase III study), although the median time before the disease progressed was only 6 months. The graph shows the fraction of patients that remained free of progression of the cancer as a function time (Chapman et al., 2011). What caused resistance to BRAF-targeted drugs? BRAF-mutated metastatic melanomas responded to the BRAF inhibitor, vemurafenib, in over half the cases. After holding the cancer in check, typically for 5 to 7 months, however, the cancer became resistant to the drug, and the treatment was no longer effective. Very soon after those findings were reported, a flurry of papers appeared in high-profile journals in 2010 addressing this conundrum, and the findings were full of surprises (Flaherty et al., 2010) (Hatzivassiliou et al., 2010; Heidorn et al., 2010; Livingstone et al., 2010; Nazarian et al., 2010; Poulikakos et al., 2010; Smalley, 2010; Sondergaard et al., 2010; Yang et al., 2010). The essential findings were summarized by Solit and Sawyers (Solit and Sawyers, 2010). The first idea about the cause of the resistance was that -- as was the case in resistance to many other drugs -- the mutated target (BRAF in this case) would acquire an additional mutation that prevented the drug's effect. Shockingly however, vemurafenib did inhibit mutant BRAF in resistant melanoma cells even though the overactive cell division continued unabated. Moreover, DNA sequencing showed that there was no second mutation in the BRAF gene in drug-resistant melanomas (Nazarian et al., 2010). This finding was even more surprising, because engineered mutations at the binding pocket of mutant BRAF did in fact confer resistance to vemurafenib (Solit and Sawyers, 2010). Therefore, something else was going on to explain why vemurafenib remained fully effective in inhibiting the mutated BRAF, whereas the melanomas no longer responded. 386 K.W.Kohn Drugs Against Cancer CHAPTER 19 But yet another surprise was in store: BRAF inhibitors actually sim ulated the MEK- ERK pathway in BRAF-mutated melanoma cells. This was found to be due to activation of another member of the RAF family, CRAF (also known as RAF1 ), which, like BRAF, signals cell division via the MEK-ERK pathway (Solit and Sawyers, 2010). Thus, resistance to inhibitors of mutant BRAF developed when CRAF became activated and bypassed the inhibition of mutant BRAF (Figure 19.3). How all that happens was elucidated in 2013 by a large consortium of clinical and basic science investigators (Trunzer et al., 2013). They discovered that resistance to inhibition of mutant BRAF by vemurafenib was usually caused by one of two events: (1) activating mutation in MEK, or (2) activating mutation in a RAS gene, particularly NRAS (Figure 19.3). How overactive MEK would bypass the inhibition of mutant BRAF was obvious, because MEK was downstream from BRAF in the signaling cascade (Figure 19.1). But how would an activating mutation of NRAS do that? The process was found to be quite simple: NRAS stimulates the activity of another member of the RAF family, the already mentioned CRAF, which bypasses mutant BRAF and stimulates MEK, which in turn directly stimulates proliferation of the melanoma cells (Trunzer et al., 2013). Activation ofCRAF also accounted for most of the melanoma cases that had normal BRAF genes. (Some facts about protein kinases, such as BRAF, CRAF, MEK, and ERK, are summarized in Box 19.1.) That was the state of understanding of vemurafenib resistance as of 2013. During the next few years, several new inhibitors of BRAF and MEK were developed. Additional routes to resistance were uncovered and the full story became ever more complicated (Luebker and Koepsell, 2019). Resistance thus developed in several different ways that became well understood, and countermeasures based on that knowledge were tested. The new treatments that prevented or delayed the onset of resistance involved combining a BRAF- inhibitor with another drug. How it all worked was based on a signaling cascade that went as follows: The main parts of the signaling cascade are diagrammed, as they were understood at various levels of detail, in Figures 19.1, 19.2, and 19.3. It all begins with the activity of a receptor tyrosine kinase, such as epidermal growth factor receptor (EGFR, Chapter 17). EGFR is in the cell surface membrane and is activated when its extracellular part binds a growth factor that is floating around outside the cell. Several different receptor tyrosine kinases were known to funnel into the same signaling pathway via RAS, RAF, MEK, and ERK, and drive cell proliferation, although EGFR seemed to be the most important. The signal from EGFR was known to go through a series of molecular steps that Jed to genes in the cell nucleus to become activated to promote cell division and to progress of cancer. First, the signal from EGFR activated, via SOS, members of the RAS family. (SOS is a human version of "son of sevenless" that was originally 387 K.W.Kohn Drugs Against Cancer CHAPTER 19 discovered in a mutation of the fruit fly eye, as described in Chapter 18). From RAS, the signal activates a cascade of kinases, leading to transcription factors, such as MYC, that activate genes stimulating cell division and cancer progression. So, how does resistance to BRAF inhibitors, such as vemurafenib, develop? First, a brief review: About half of melanoma cases have a BRAF mutation, over 80% of which are changes in a single amino acid in which valine is replaced by glutamic acid at amino acid position 600 (V600E; Eis single-letter code for glutamic acid). The V600E mutation greatly increased the kinase activity of BRAF and promoted the growth of cancers that were dependent on signaling via MEK and ERK. Therefore, inhibiting the kinase activity of BRAF with vemurafenib was effective against cancers, particular V600E-mutated melanoma. Resistance developed when the cancer's dependence on the RAF-MEK-ERK pathway was overcome or circumvented (Trunzer et al., 2013). Overcoming the resistance was a difficult problem, however, because there were several different ways that resistance could develop. More than half a dozen molecular pathway changes were described, each of which could cause resistance to BRAF inhibitors (Johnson et al., 2015). Many studies were initiated to test various hypotheses about how to counter the different pathways to resistance. Box 19.1. Some facts about protein kinases. A protein kinase is an enzyme that adds phosphate groups to certain amino acids in proteins. There are two major classes of protein kinases: those that add a phosphate to tyrosine and those that add a phosphate to serine and/or threonine (serine and threonine are closely related amino acids, while tyrosine is special). One of the protein kinases in the cascade, MEK, adds phosphates to both a tyrosine and a serine in the next protein kinase in the cascade, but such dual tyrosine-serine kinases are uncommon. When a protein kinase adds one or more phosphates to the next protein kinase in the cascade, the latter becomes activated so as to enable it to phosphorylate the protein kinase that comes next in the chain. The major steps in the MAP kinase cascade are diagrammed in Figure 19.1. Each kinase in the sequence can activate several molecules of its target in the next step, thereby functioning as an amplifier of the signal, which is why the chain of protein kinases is aptly called a cascade. How might the resistance of BRAF-mutated melanomas be overcome? The most common cause of the resistance of BRAF-mutated melanomas to BRAF inhibitors was acquisition by the tumor of an activating mutation in the MEK gene. Since MEK is downstream of BRAF in the signaling sequence, overactive MEK would bypass BRAF in stimulating cell division (Figure 19.1 and Box 19.2). An obvious way 388 K.W.Kohn Drugs Against Cancer CHAPTER 19 to counter or delay resistance due to mutation of MEK, therefore, was to add an inhibitor of MEK. When that was tested in a large randomized trial, the combination of vemurafenib and a MEK inhibitor added about 4 months (compared to vemurafenib alone) to the length of time that the malignancy was held in check (Figure 19.8) (Larkin et al., 2014). The combination therapy halted the progression of the disease by a median of 10 months, but the cancer would then resume its growth. Therefore, additional paths to resistance were sought One possibility that was investigated was activation of a pathway from EGFR to mTOR (mammalian target of rapamycin), which, like the pathway to ERK, stimulated genes that promoted cell proliferation (Figure 19.9). Drug combinations that would inhibit both the pathway to mTOR and the pathway to ERK were therefore proposed for treatment ofmutant- BRAF cancers, (Taieb et al., 2019). The pathway to mTOR branches at the EGFR level as an alternative to the branch leading to ERK (Figure 19.9). The first step in this alternative branch is binding and activation of Pl3K (phosphatidylinositol-3'-kinase). Pl3K binds to EGFR (at phosphotyrosines that are produced when EGFR is activated); from that perch, Pl3K phosphorylates a membrane lipid (phosphatidylinositol) that binds and activates AKT (also known as protein kinase B). AKT then phosphorylates and activates mTOR, which proceeds to stimulate cell proliferation. This pathway was entirely independent of mutant-BRAF and therefore was considered as a possible way that resistance to inhibitors of BRAF and MEK could develop. A further detail was that Pl3K can be stimulated by EGFR via NRAS (this role of NRAS was omitted in Figure 19.9). An activating mutation of NRAS, which was sometimes found in malignant melanomas, therefore could stimulate Pl3K and drive cell division via mTOR. In order to moderate the activity of Pl3K, cells have evolved PTEN as a regulator (Figure 19.9). PTEN, however, was subject to inactivation by mutation, leading to enhanced cell division and enhanced malignancy (Nogueira et al., 2010). Various mutations in the positive and negative controllers of the pathway from EGFR to mTOR were being considered as possible causes of resistance that might be countered by adding inhibitors of NRAS and/or Pl3K. That is where the investigations stood at the time of this writing. BRAF mutation in other cancers. The V600E-mutation of BRAF was occasionally found in cancers other than melanomas. (Why it was so common in melanomas - about 50% of cases - was unknown.) The mutation was found in 10% of metastatic colon cancers and portended a poor prognosis (Ducreux et al., 2019). Although that was a relatively low percentage, it amounted a large number of life-threatening cases, because colon cancer was the second most frequent cause of cancer deaths. 389 K.W.Kohn Drugs Against Cancer CHAPTER 19 BRAF-mutated colon cancers, in contrast to melanomas, were not driven by mutated BRAF and did not respond to BRAF and MEK inhibitors. It seemed likely, therefore that colon cancers were often driven by a different pathway to uncontrolled cell proliferation. The pathway to mTOR was thought a good possibility (Taieb et al., 2019) (Ducreux et al., 2019) (Figure 19.9). Early clinical trials however were disappointing. Researchers therefore had to go back to the drawing board to find other routs to therapy. An important consideration was that BRAF-mutated colon cancers were generally found to occur in the ascending part of the colon on the right side of the body, which is where about 15% of colon cancers happen. These cancers usually have defects in DNA mismatch repair, which causes instability of certain DNA sequences and leads to cancer. This defect was a possible lead to new therapy. The DNA mismatch repair story will be told in Chapter 25. 100 90 ,. 80 -~ * ...,""1 l Vemurafenib + cobimetinib (N=247) "-~ > 70 60 "'al" --------- -------- ·~----- ------ 50 ... -- <If "-•1_ C 0 ·.;; "' 40 Vemurafenib + placebo (N =248) , __ ~ 30 ¼---·· 0 ~ 20 I ·----•♦ 10 Hazard ratio, 0.5 1 (95% Cl, 0 .39- 0.68) P<0.001 0+ - - - - - - - - ~ - - - - - ~ - - - - - - - ~ 0 I 3 5 7 9 11 13 15 Months Figure 19.8. Metastatic melanoma patients whose tumor had a V600 BRAF mutation were randomized for treatment with vemurafenib with or without the addition of a MET inhibitor (cobimetinib) (Larkin et al., 2014). The combination of the two drugs, compared with vemurafenib alone, added about 4 months to the length of time before the malignancy progressed. The median progression-free survival was about 10 months for the combination, as opposed to only about 6 months for vemurafenib alone. 390 K.W.Kohn Drugs Against Cancer CHAPTER 19 Antl·EGFA monoclonal antibody 8 ~► 7 \ p ✓ P13K lnhlbilo, -i 0 8RAF - i 0 lnh1bl1or ~ / PIP2 '\ •/ r PIP3 ' \ MEK - i Q 'J • lnhlbftor ""- 8 .--G 8 Figure 19.9. Two s ignaling pathways from EGFR that stimulated cell proliferation and caused resistance to BRAF inhibitors. In addition to the pathway leading to ERK, there was a separate branch from EGFR that led to mTOR. Cell proliferation was stimulated by mTOR as well as by ERK. This diagram is from a recent review by (Taieb et al., 2019), who proposed treatment of BRAF-mutant metastatic cancers (including some colon cancers) with drug combinations that would block both pathways and avoid or overcome drug-resistance: EGFR (by a monoclonal antibody), BRAF and MEK inhibitors, plus an inhibitor of PI3K. What could be done for melanoma patients whose tumors had normal BRAF? The question was what drives the malignancy of melanomas whose BRAF was normal, not mutated? Investigators found that the culprit was most often an activating mutation of one of the RAS genes, namely NRAS. Mutant NRAS was found to stimulate CRAF and the pathway to ERK and uncontrolled cell proliferation, as shown in Figures 19.3 and 19.9. In that case, drugs that inhibited BRAF were of no use - and, surprisingly, they might actually make things worse by stimulating the pathway through CRAF (Halaban et al., 2010). Activation of the pathway through CRAF was also how mutant-BRAF melanomas could become resistant to drugs such as vemurafenib. These melanomas may initially respond to the drug, but might soon develop an activating mutation in NRAS, which drives uncontrolled cell proliferation via CRAF (Dossett et al., 2015). To make matters even more complicated, there were pathways from receptor 391 K.W.Kohn Drugs Against Cancer CHAPTER 19 tyrosine kinases, other than EGFR, that could become mutated and stimulate cell proliferation by way of various other pathways The treatment of metastatic melanoma by inhibitors of specific molecular targeting therefore remained a work- in-progress in the quest for a cure. A different approach was also being investigated with similar or even greater intensity: immunotherapy, which is the subject of the next chapter. Summary About 50% of melanomas have a particular oncogenic mutation in BRAF, which is an unusually high incidence for a particular mutation in a particular disease. By far the most common mutation was a change in a single amino acid at position 600 in the amino acid chain of BRAF, a valine being replaced by a glutamic acid at that position (more rarely, the valine was replaced by aspartic acid, lysine or arginine). Thus, the mutation that made BRAF oncogenic was exquisitely specific. The mutation increased by a factor of 10 the ability of BRAF to signal faster cell proliferation (Dossett et al., 2015). The signal passed to MEK and from there down the chain to transcription factors that increase the production of proteins for cell proliferation. However, the multiple pathways, mutations, and by-pass possibilities that could occur presented a highly complicated picture that remained a great challenge. References Atkins, M.B., Lotze, M.T., Dutcher, J.P., Fisher, R.I., Weiss, G., Margolin, K, Abrams, J., Sznol, M., Parkinson, D., Hawkins, M., et al. (1999). High-dose recombinant interleukin 2 therapy for patients with metastatic melanoma: analysis of270 patients treated between 1985 and 1993. Journal of clinical oncology: official journal of the American Society of Clinical Oncology 17, 2105-2116. Carter, S.K., and Friedman, M.A. (1972). 5-(3,3-dimethyl-l-triazeno)-imidazole-4- carboxamide (DTIC, DIC, NSC-45388)--a new antitumor agent with activity against malignant melanoma. Eur J Cancer 8, 85-92. Chapman, P.B., Einhorn, L. H., Meyers, M.L., Saxman, S., Destro, A.N., Panageas, K.S., Begg, C.B., Agarwala, S.S., Schuchter, LM., Ernstoff, M.S., et al. (1999). Phase III multicenter randomized trial of the Dartmouth regimen versus dacarbazine in patients with metastatic melanoma. Journal of clinical oncology: official journal of the American Society of Clinical Oncology 17, 2745-2751. Chapman, P.B., Hauschild, A., Robert, C., Haanen, J.B., Ascierto, P., Larkin, J., Dummer, R, Garbe, C., Testori, A., Maio, M., eta/. (2011). Improved survival with vemurafenib in melanoma with BRAF V600E mutation. The New England journal of medicine 364, 2507-2516. Davies, H., Bignell, G.R., Cox, C., Stephens, P., Edkins, S., Clegg, S., Teague, J., Woffendin, H., Garnett, M.J., Bottomley, W., eta/. (2002). Mutations of the BRAF gene in human cancer. Nature 417, 949-954. 392 K.W.Kohn Drugs Against Cancer CHAPTER 19 Dossett, L.A., Kudchadkar, R.R., and Zager, J.S. (2015). BRAF and MEK inhibition in melanoma. Expert opinion on drug safety 14, 559-570. Ducreux, M., Chamseddine, A., Laurent-Puig, P., Smolenschi, C., Hollebecque, A., Dartigues, P., Samallin, E., Boige, V., Malka, D., and Gelli, M. (2019). Molecular targeted therapy of BRAF-mutant co lo rectal cancer. Therapeutic advances in medical oncology 11, 1758835919856494. Flaherty, K.T., Puzanov, I., Kim, K.B., Ribas, A., McArthur, G.A., Sosman, J.A., O'Dwyer, P.J., Lee, R.J., Grippo, J.F., Nolop, K., eta/. (2010). Inhibition of mutated, activated BRAF in metastatic melanoma. The New England journal of medicine 363, 809-819. Gerner, R.E., Moore, G.E., and Didolkar, M.S. (1973). Chemotherapy of disseminated malignant melanoma with dimethyl triazeno imidazole carboxamide and dactinomycin. Cancer 32, 756-760. Halaban, R., Zhang, W., Bacchiocchi, A., Cheng. E., Parisi, F., Ariyan, S., Krauthammer, M., Mccusker, J.P., Kluger, Y., and Sznol, M. (2010). PLX4032, a selective BRAF(V600E) kinase inhibitor, activates the ERK pathway and enhances cell migration and proliferation of BRAF melanoma cells. Pigment Cell Melanoma Res 23, 190-200. Hatzivassiliou, G., Song, K., Yen, I., Brandhuber, B.J., Anderson, D.J., Alvarado, R., Ludlam, M.J., Stokoe, D., Gloor, S.L., Vigers, G., et al. (2010). RAF inhibitors prime wild-type RAF to activate the MAPK pathway and enhance growth. Nature 464, 431-435. Heidorn, S.j., Milagre, C., Whittaker, S., Nourry, A., Niculescu-Duvas, I., Dhomen, N., Hussain, J., Reis-Filho, J.S., Springer, C.J., Pritchard, C., et al. (2010). Kinase-dead BRAF and oncogenic RAS cooperate to drive tumor progression through CRAF. Cell 140, 209-221. Helias-Rodzewicz, Z., Funck-Brentano, E., Baudoux, L., Jung. C.K., Zimmermann, U., Marin, C., Clerici, T., Le Gall, C., Peschaud, F., Taly, V., et al. (2015). Variations of BRAF mutant allele percentage in melanomas. BMC cancer 15, 497. Jenkins, R.W., and Fisher, D.E. (2020). Treatment of Advanced Melanoma in 2020 and Beyond. J Invest Dermatol. Johnson, D.B., Menzies, A.M., Zimmer, L., Eroglu, Z., Ye, F., Zhao, S., Rizos, H., Sucker, A., Scolyer, R.A., Gutzmer, R., et al. (2015). Acquired BRAF inhibitor resistance: A multicenter meta-analysis of the spectrum and frequencies, clinical behaviour, and phenotypic associations of resistance mechanisms. Eur J Cancer 51 , 2792-2799. Kolch, W., Kotwaliwale, A., Vass, K., and Janosch, P. (2002). The role of Raf kinases in malignant transformation. Expert Rev Mo! Med 4, 1-18. Larkin, J., Ascierto, P.A., Dreno, B., Atkinson, V., Liszkay, G., Maio, M., Mandala, M., Demidov, L., Stroyakovskiy, D., Thomas, L., eta/. (2014). Combined vemurafenib and cobimetinib in BRAF-mutated melanoma. The New England journal of medicine 371, 1867-1876. Livingstone, E., Zimmer, L., Piel, S., and Schadendorf, D. (2010). PLX4032: does it keep its promise for metastatic melanoma treatment? Expert opinion on investigational drugs 19, 1439-1449. 393 K.W.Kohn Drugs Against Cancer CHAPTER 19 Luebker, S.A., and Koepsell, S.A. (2019). Diverse Mechanisms of BRAF Inhibitor Resistance in Melanoma Identified in Clinical and Preclinical Studies. Front Oneal 9, 268. Meyer, P., Klaes, R., Schmitt, C., Boettger, M.B., and Garbe, C. (2003). Exclusion of BRAFV599E as a melanoma susceptibility mutation. Int J Cancer 106, 78-80. Nazarian, R., Shi, H., Wang, Q., Kong, X., Kaya, R.C., Lee, H., Chen, Z., Lee, M.K, Attar, N., Sazegar, H., et al. (2010). Melanomas acquire resistance to B-RAF(V600E) inhibition by RTK or N-RAS upregulation. Nature 468, 973-977. Nogueira, C., Kim, KH., Sung, H., Paraiso, KH., Dannenberg, J.H., Bosenberg, M., Chin, L., and Kim, M. (2010). Cooperative interactions of PTEN deficiency and RAS activation in melanoma metastasis. Oncogene 29, 6222-6232. Poulikakos, P.I., Zhang, C., Bollag, G., Shokat, KM., and Rosen, N. (2010). RAF inhibitors transactivate RAF dimers and ERK signalli ng in cells with wild-type BRAF. Nature 464, 427-430. Pratilas, C.A., and Solit, 0 .8. (2010). Targeting the mitogen-activated protein kinase pathway: physiological feedback and drug response. Clinical cancer research : an official journal of the American Association for Cancer Research 16, 3329- 3334. Roh, M.R., Eliades, P., Gupta, S., and Tsao, H. (2015). Genetics ofmelanocytic nevi. Pigment Cell Melanoma Res 28, 661-672. Sakaizawa, K, Ashida, A., Kiniwa, Y., and Okuyama, R. (2020). BRAF Mutation Heterogeneity in Melanoma Lesions. Acta Derm Venereal 100, adv00045 . Sala, E., Mologni, L., Truffa, S., Gaetano, C., Bollag, G.E., and Gambacorti -Passerini, C. (2008). BRAF silencing by short hairpin RNA or chemical blockade by PLX4032 leads to different responses in melanoma and thyroid carcinoma cells. Mol Cancer Res 6, 751-759. Shinozaki, M., Fuj imoto, A., Morton, D.L., and Hoon, D.S. (2004). Incidence of BRAF oncogene mutation and clinical relevance for primary cutaneous melanomas. Clinical cancer research : an official journal of the American Association for Cancer Research 10, 1753-1757. Sirott, M.N., Bajorin, D.F., Wong, G.Y., Tao, Y., Chapman, P.B., Templeton, M.A., and Houghton, A.N. (1993). Prognostic factors in patients with metastatic malignant melanoma. A multivariate analysis. Cancer 72, 3091-3098. Smalley, KS. (2010). PLX-4032, a small-molecule B-Raf inhibitor for the potential treatment of malignant melanoma. Curr Opin lnvestig Drugs 11, 699-706. Solit, 0 ., and Sawyers, C.L. (2010). Drug discovery: How melanomas bypass new therapy. Nature 468, 902-903. Sondergaard, J.N., Nazarian, R., Wang, Q., Guo, 0., Hsueh, T., Mok, S., Sazegar, H., MacConaill, L. E., Barretina, J.G., Kehoe, S.M., et al. (2010). Differential sensitivity of melanoma cell lines with BRAFV600E mutation to the specific Raf inhibitor PLX4032. J Transl Med 8, 39. Taieb, )., Lapeyre-Prost, A., Laurent Puig, P., and Zaanan, A. (2019). Exploring the best treatment options for BRAF-mutant metastatic colon cancer. British journal of cancer. Trunzer, K, Pavlick, A.C., Schuchter, L., Gonzalez, R., McArthur, G.A., Hutson, T.E., Moschos, S.J., Flaherty, KT., Kim, KB., Weber, J.S., et al. (2013). 394 K.W.Kohn Drugs Against Cancer CHAPTER 19 Pharmacodynamic effects and mechanisms of resistance to vemurafenib in patients with metastatic melanoma. Journal of clinical oncology: official journal of the American Society of Clinical Oncology 31, 1767-1774. Tsai, J., Lee, J.T., Wang, W., Zhang, J., Cho, H., Mamo, S., Bremer, R., Gillette, S., Kong, J., Haass, N.K, et al. (2008). Discovery of a selective inhibitor of oncogenic B-Raf kinase with potent antimelanoma activity. Proceedings of the National Academy of Sciences of the United States of America 105, 3041-3046. Vogelstein, B., and Kinzler, KW. (2004). Cancer genes and the pathways they control. Nature medicine 10, 789-799. Yang, H., Higgins, B., Kolinsky, K, Packman, K, Go, Z., Iyer, R., Kolis, S., Zhao, S., Lee, R., Grippo, J.F., et al. (2010). RG7204 (PLX4032), a selective BRAFV600E inhibitor, displays potent antitumor activity in preclinical melanoma models. Cancer research 70, 5518-5527. 395 K. W. Kohn Drugs Against cancer CHAPTER 20 Qoplu20. Altd'- drltfl dl'--,-1 diMlop,o:iffltot NO Z20BJBcf Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@nih.gov CHAPTER20 Anticancer drug discovery and development at the National Cancer Institute (NCI). Attempts to find anticancer medicines date back to before the beginning of the 20 th century and were undertaken by many researchers and institutions in many countries. In this chapter, I review the anticancer drug discovery program at the National Cancer Institute (NCI), particularly the parts ofit that I have some direct knowledge of between the mid- 19SO's to the second decade of the 21 st century. I came to the NCI as a Clinical Associate in the childhood leukemia and adult cancer wards and joined research in a clinical pharmacology unit of the medicine branch (more in the Introduction). As research expanded and diversified, one of several new Laboratories established was a Laboratory of Molecular Pharmacology, which I was appointed to lead. Since the mid-1960's, I served on various committees of the Developmental Therapeutics Program (DTP), although I had direct responsibility only for my own Laboratory. My role in the broader part of the DTP was to contribute basic science input, and I was free to make suggestions or point out problems. Here, I relate some of the successes and disappointments of the Program based on articles I was able to obtain through the NIH library and on my possibly imperfect memory, as well as some old items that remained in my possession. I leave to a future historian or investigative writer to review archival material for a proper history of this complex and instructive story. "Chemotherapy" dates back to 1909, when German Nobel Prize winner physician and chemist Paul Ehrlich (Figure 20.1) developed the first effective medicine for treatment of syphilis: the arsenic-containing drug, arsphenamine. Ehrlich coined the term "chemotherapy" to denote the treatment of disease using chemical drugs. He was also the first to use an animal screen to test chemicals for their effectiveness against a disease: in 1908 he used rabbits to test chemicals for their effectiveness against syphilis. 396 K. W. Kohn Drugs Against cancer CHAPTER 20 A screen for potential anticancer activity required a suitable test system, which was first provided in the early 1910s by George Clowes at Roswell Park Memorial Institute in Buffalo, New York, who developed in mice and rats the first transplantable tumors. The first anticancer screen was set up in 1935 by Murray Shear at the newly established National Cancer Institute, which however was dropped in 1953 because of unacceptable toxicity (DeVita and Chu, 2008). Screening for potential anticancer chemical agents however received new emphasis from the effectiveness of nitrogen mustards against lymphomas (Chapter 1). The recurrence of the tumors, which then no longer responded to the drugs, caused many physicians to feel that trying to cure cancers by means of chemotherapy was hopeless. It was with this pessimistic view that the research physicians newly recruited to NCI in the 1950s to conduct cancer chemotherapy research were faced. So, where did these intrepid research physicians come from? Perhaps surprisingly, it traces back to the anti -malaria research during World War II. Soldiers fighting in malaria-infested areas of the Pacific and Asia often came down with the disease within a few weeks, and the available drugs to supplement quinine, such as quinacrine (then commonly called Atabrine), were inadequate. A program was therefore established at several hospitals and universities to find more effective drugs (Condon-Rall, 1994). One of the hospitals that was prominent in the anti-malaria drug discovery research was Goldwater Memorial Hospital located on Roosevelt Island (then called Welfare Island) in the East River between Manhattan and Queens, New York (Figure 20.2). Screening chemicals for potential drugs required an animal test system. Early in the research, they found that a suitable test species were certain birds. Using malaria-infected birds to test several thousand chemicals, they discovered what they were looking for: the suitably effective new drug was chloroquine. (Both quinacrine and chloroquine, by the way, are DNA intercalating agents, see Chapter 4.) From the success of the anti-malaria drug screen, it was thought that something akin to that success might be accomplished in the cancer field. Accordingly, several of the clinician scientists who were prominent in the anti-malaria program were recruited to head a new anticancer research program at the National Institutes of Health that was expanded inl 955 to include a research hospital, the NIH Clinical Center. Chosen to head the new cancer chemotherapy research program was Gordon Zubrod (Figure 20.1), who had a major role in the anti-malaria research at Goldwater. Much of the credit for the eventual cure of leukemias and lymphomas is attributed to Zubrod's directorship, accomplished despite strong headwinds. Zubrod's research career began in 1943 at Goldwater Hospital in New York, where he worked on the search for treatments for malaria - the first nationally organized drug discovery program - after which he had a fellowship in pharmacology and med icine at Johns Hopkins University. The NIH recruited Zubrod in 1954 to provide leadership in clinical research and chemotherapy programs at the brand new 500-bed Clinical Center. 397 K. W. Kohn Drugs Against cancer CHAPTER 20 Zubrod became clinical director of the National Cancer Institute in 1954 and became head of its Division of Cancer Treatment in 1956 and scientific director in 1961. It was with considerable apprehension that Zubrod reported to NCI on October 1, 1954. "Could I adapt to government service after 20 years of university life?" he wrote. "How would I, without experience in cancer research, provide leadership to scientists who had spent a lifetime studying cancer?" He is reported to have said, "my friends at Hopkins teased me about joining a non-clinical group to which they mockingly attached the sobriquet 'The National Mouse Cancer Institute."' (The Cancer Letter Archives, January 29, 1999). Zubrod received a medical degree from Columbia University's College of Physicians & Surgeons in 1940 and had house staff training at its Presbyterian Hospital in New York. I graduated from the same medical school in 1956 and had a 3-month elective at Columbia's medical research unit at Goldwater. That was how I found out about the newly expanded NIH and the possibility of starting on a research career there. Indeed, while at Goldwater, I heard the NIH called "the Goldwater on the Potomac." In view of my background and interest in chemistry and physics, Zubrod hired me as a Clinical Associate assigned to work with David Rall in a clinical pharmacology group in the Division of Cancer Treatment Zubrod targeted leukemia as the initial disease for intensive study, and his proposed trials ofmethotrexate became the first prospective cancer chemotherapy trial in the U.S. My first assignment when I arrived at NIH in July 1957, was to help in the care of children who had acute leukemia and were in the first trials of methotrexate and 6-mercaptopurine as single agents. Although the drugs did not cure, we could at least bri ng them out of the acute phase of blast crisis and sometimes prolong their lives for a few weeks. Fortunately, the childhood leukemia and adult solid tumor units were expertly directed by Emil "Jay" Freireich and Emil "Tom" Frei. I have recorded in the Introduction chapter my impressions of them and their clinical units of the time. Freireich gives Zubrod credit for launching him on the work that led to a first step in the control acute leukemia: the use of platelets to stop hemorrhage. According to Freireich, Zubrod would come on rounds, and in those days there would be blood splattered over over the linens and the staff. Freireich said, "Zubrod said to me, 'You're a hematologist, why don't you do something about this bleeding.' I took that as an order." Freireich found that fresh platelets could be effective, but the NIH blood bank would not give him the fresh blood needed, because at the time everyone thought that platelets wouldn't work and might even be harmful. A grand rounds meeting on blood transfusion was called. "We presented our data, but during the discussion, the director of the blood bank said platelet transfusions were not effective and the bank would not issue fresh blood," Freireich said. Recalled Frei: "Zubrod got up and said something like, 'Speaking for NCI and patients with cancer currently and in the future, we truly don't know whether we can cure cancer in the near future or if ever, but we are here to try. Progress is going to come incrementally and not all at once, and one hurdle is to control bleeding, and platelets offer the best chance to do that. I plan to support platelet research to get it done."' "It took a lot of courage to say 398 K. W. Kohn Drugs Against cancer CHAPTER 20 that" Frei said. "Within five years, we had eliminated hemorrhage as a cause of death in 90 percent of the patients," Freireich said. "I always give Zubrod credit for that." "He kept Frei and Freireich out of trouble," said DeVita, who arrived at NCI in 1963. "They were doing what was considered very wild stuff. They needed a distinguished guy like Zubrod over them." Zubrod is sometimes remembered as the organizer, enabler and pacifier who managed to shepherd an unruly bunch of NCI scientists, particularly Tom Frei and Jay Freireich, through a wild ride that demonstrated the efficacy of chemotherapy in the treatment of childhood acute leukemia, resulting in the first cures of this disease. However, the malaria program that led to the leukemia research program also had a history. Zubrod had participated in the treatment of syphilis of the central nervous system, in which the patients were given malaria to induce fever that was the only successful treatment. It was based on that experience, that Zubrod was recruited in 1943 to military service at Goldwater Hospital to work on the malaria program. As director of NCl's cancer treatment program, Zubrod then hired Frei, age 31, who had been a resident at St. Louis University, to manage the studies. A year later, he hired Freireich, a 28-year-old hematology trainee at Boston University. "Zubrod said, 'I see you have training in hematology. Do you know anything about leukemia?"' Freireich said. "I said, 'Of course,' even though I didn't know much. He said, 'I've decided we need to make progress in leukemia, and therefore, you're hired'." The headwinds Zubrod's program encountered are further shown by the following recollections. "At a conference, a pathologist once said finding a drug for cancer was like finding a drug that could dissolve off the left ear and leave the right ear intact," Frei said. "General medicine thought we were members of the Poison-of-the-Month Club," Holland said. "There was little confidence in chemotherapy." DeVita recalled attending a seminar Zubrod gave at Mt. Desert Island Biological Laboratory in Bar Harbor, Maine, in the summer of 1959. Zubrod spoke about the NCI drug development program. "I remember being stunned at how hostile the crowd was that there would be any success at random screening," he said. "He deserves a lot of credit for that program, which I would submit has been a great success." (Sources: https://www.library-archives.cumc.columbia.edu/obit/c-gordon-zubrod: The Cancer Letter Archives, January 29, 1999) In 1966, Zubrod reviewed the history of anticancer drug discovery attempts from the beginning of the 20 th century and the development by mid -century of the cancer chemotherapy program at the National Cancer Institute (Zubrod, 1966). The general view at the time was that effective cancer chemotherapy would require molecules that could reach and act on every cancer cell in the body. An important early development was the production by the R. B. Jackson Laboratories around 1946 of inbred strains of mice and transplantable tumors. These were among the first things needed for consistent and reproducible animal tests of potential anticancer drugs and chemicals. In another important early development, Lloyd Law at NCI isolated from x-rayed mice a transplantable leukemia whose characteristics were highly suited for quantitative studies of chemicals that inhibited the growth of these cells in the mice. This leukemia, which was 399 K. W. Kohn Drugs Against cancer CHAPTER 20 called L1210, was to have an important role in anticancer drug discovery for many years. It was the first of a panel of transplantable mouse tumors that came to be used to evaluate a great many chemicals as potential anticancer drugs. After extensive reviews and head-scratching, NCI officials proposed to the Congressional Appropriations Committee to build on the wartime success in producing new antimalaria drugs and to support a similarly designed research program to discover drugs against leukemia. Impressed by this proposal, Congress appropriated one million dollars for fiscal year 1954 to support the research. From that modest investment, funding in the following years grew exponentially. Paul Ehrlich Figure 20.1. Left, Paul Ehrlich (1854-1915), the founder of chemotherapy. Right, C. Gordon Zubrod (1914-1999) was clinical and research director of the National Cancer Institute (NCI) from 1956 until 1974, when he moved to direct the oncology program at the University of Miami Medical School and the Florida Comprehensive Cancer Center. He received an MD degree at Columbia College of Physicians and Surgeons in 1940. During World War II, he worked at Goldwater Memorial Hospital to find more effective drugs for prevention and treatment of malaria. That work, using birds as a test system, led to a better drug: chloroquine. Based on the success of that program, Zubrod was recruited to organize a program to discover anticancer drugs at NCI, NIH. 400 K. W. Kohn Drugs Against cancer CHAPTER 20 Figure 20.2. Goldwater Memorial Hospital in 1938, as seen from the Queens borough Bridge. This vast chronic disease hospital was located on Welfare Island (later called Roosevelt Island), a two-mile sliver ofland in the East River nestled between the Upper East Side and Astoria, Queens, New York. In addition to caring for a large number of chronic disease patients, Goldwater included clinical research departments associated with the Columbia, Cornell, and NYU medical schools. The hospital, opened in 1939, was an immense facility designed to be a new model of medical care for patients with chronic illnesses. Researchers in the Columbia unit solved the anti -malaria drug problem during World War II. Many of those researchers, including Gordon Zubrod, were recruited to lead the clinical and research programs of the newly expanded NCI at NIH. The hospital closed in December 2013, but before its destruction, a detailed photographic record was made: (http·//11rhanomnih11s net/2014/04/a11topsy-of-a-hos12ita1-a-photographic-record-of- coler-goldwater-on-roosevelt-island /). NCl's cancer chemotherapy program in 1970. A milestone in the early development of the cancer chemotherapy program was reviewed in the First Joint Working Conference on the NCI Chemotherapy Program, which convened on December 16-18, 1970, in the Hilton Inn in Annapolis, Maryland. The report summarized a milestone in the development of the program. I found a copy of this 50-year- old report among my admittedly disorderly records and will summarize it here, because the report may or may not exist in the NCI archives. A thorough historical investigation is more than I could undertake and leave it for future historians or investigative writers. In 401 K. W. Kohn Drugs Against cancer CHAPTER 20 the course of the following summary of the 1970 report, I will take the liberty of inserting a few personal impressions that may perhaps add a little to the story. The overall program was led by NCl's Scientific Director, Gordon Zubrod and was carried out in NCl's laboratories and clinics in conjunction with contracts with industry and grants to universities. Components of the program ranged from acquisition of large numbers of new compounds, to screening those compounds for anticancer action, to selection of compounds for toxicology, pharmacology, and biochemical studies, to clinical trials. Major segments of the program carried out pharmacologic and toxicologic studies in animals and patients under the Associate Scientific Director for Experimental Therapeutics, David P. Rall, MD, PhD. Gordon Zubrod was highly admired and respected for his skillful and thoughtful leadership, which was dignified yet flexible. He made courageous decisions that supported uncertain efforts, which ultimately Jed to the cures of childhood leukemia and Hodgkin's lymphoma. Dave Rall had a very relaxed and pleasant leadership style that encouraged everybody in our phrmacology group to do our best to make significant progress. He was chief of the laboratory to which I was assigned during my early years at NCI (1959-1965). The papers on his desk seemed highly disorganized, yet he would immediately find whatever he needed. Our meetings in his office were enjoyable and provoking of new ideas to investigate. One day, a wall of his office was covered with an enormous diagram of the new Linear Array, about which there will be more to say. It was already well established in 1970 for guiding the progress of new compounds through the development sequence. Clinical investigations, as well as studies of the natural history of cancer and cell population kinetics in relation to chemotherapy, were Jed by the Associate Scientific Director for Clinical Trials, Seymour M. Perry, MD. This component of the program had in it a childhood leukemia service, an adult solid tumor service, and a branch that investigated human tumor cell biology and cell control mechanisms. A Cancer Therapy Evaluation Branch, headed by Stephen K. Carter, MD, monitored and coordinated the clinical trials of investigational new drugs undertaken by the chemotherapy program. The Branch also connected with the Food and Drug Administration (FDA) for the preparation of new drug applications. In addition, NCI had a Medical Oncology Program, headed by Jerome B. Block, MD, at the VA hospital in Baltimore. This program had its own clinical and laboratory branches. We regularly went up to Baltimore for joint conferences between the Bethesda and Baltimore NCI scientists. I looked forward to these conferences, in part because they were occasions for detailed and enjoyable conversations with David Ludlum about our related research interest -- as well as other matters where our views diverged during protracted dinner conversations, since his leanings were Republican while mine were Democrat. However, that difference did not impair, and may actually have enhanced, our friendship, and we collaborated in studies, some of which we coauthored. 402 K. W. Kohn Drugs Against cancer CHAPTER 20 Another major part of the overall program was the Cancer Chemotherapy National Service Center (CCNSC), which was Jed by Saul A. Shepartz, PhD. A particularly consequential part of the CCNSC was an experimental chemotherapy unit led by Abraham Goldin, PhD, Associate Chief of Laboratory Research. Abe Goldin was highly respected and admired for his insightful innovations and personal qualities that set the tone and scientific discipline. It was carried forth, after his death, ironically of cancer, by those he had supervised, including John M. Venditti, PhD and John S. Driscoll, PhD, as well as others whom I did not know as well. Abe established the methods and protocols to evaluate the anticancer potential of new compounds, which led to a large body of reliable information about a large number of compounds that had anticancer potential. The protocol used a well-characterized strain of mouse leukemia cells (L1210) that allowed precise estimations of prolongation of life span after implantation of the L1210 cells. Three days after implantation, mice were injected with the test compound. Groups of 10 mice received a wide range of doses (according to a Fibonacci sequence). With increasing dose, an active compound would first give increased survival time relative to controls, but further increase in dose, would decrease survival due to toxicity. This gave good estimates of both the degree of activity ( maximum lifetime extension at the optimal dose) and the range of doses over which survival increased, i.e., the therapeutic ratio (Figure 20.3). Abe's successors, however, tended to be locked in with the concepts and methods used during their early research and lacked the flexibility of thinking that the originator (Goldin) had. Here I would like to insert another personal and perhaps instructive recollection. Anthony (Tony) Schrecker, PhD, whom the 1970 report listed as Associate Chief for Laboratory Research of the CCNSC had an unfortunate dislocation of his research career. Earlier, during my first few years at NCI (late 1950s and early 1960s), Tony was admired as the best organic chemist in our part of the program. I frequently visited him for advice and to borrow chemicals. He had a precise Germanic no-nonsense style that I found admirable, although sometimes overbearing. In later years, Tony responded to the perception of a demand for "relevance," which seemed like a politically inspired pressure. Trying to respond positively to this new policy, Tony changed the focus of his work from organic chemistry to biochemistry. However, the area of anticancer biochemistry in which he was engaged was not amenable to his accustomed precise discipli ne of methods. It was a case, I think, where an expert in one field became mediocre when redirected into another. As experience with various mouse tumors accrued, three were added to the standard screen: P388, a leukemia that was generally more sensitive than L1210; 816 melanoma, a relatively slow growing solid tumor, Jess sensitive than the leukemias, but nevertheless responsive to the large majority of the known clinically effective drugs, drugs that at least temporarily shrank a cancer or increased the survival time of some patients; and Lewis Lung carcinoma, a slow growing solid tumor that responded to few of the drugs, initially only cyclophosphamide, nitrosoureas, and bleomycin. A newly tested compound that was active against Lewis Lung was considered particularly noteworthy. 403 K. W. Kohn Drugs Against cancer CHAPTER 20 Table 20.1. shows an example of a significant finding in the four-mouse-tumors test system. It showed that doxorubicin (also known as Adriamycin) was better than its close chemical relative, daunorubicin (see Chapter 8). Most telling was the result in 816 melanoma in which an optimal dosage of Adriamycin cured 8 of 10 animals (Johnson and Goldin, 1975). Another example of an important finding (Table 20.2) was that drugs, such as methotrexate, that mainly killed cells that were undergoing DNA replication, were more effective when given intermittently, such as every four days, rather than daily. This was later found to be true also for leukemia and lymphoma patients (Goldin et al., 1971). lt seemed that during the few days without drug, the normal cells of the bone marrow recovered more rapidly than the leukemia or lymphoma cells. That is where the anticancer drug screen stood for many years, until an objective analysis after a few successes, such as cisplatin (which was actually not initially discovered in the screen; see Chapter 3), showed that the screen was picking up only drugs effective solely against rapidly growing cancers, such as leukemias and lymphomas. There was almost no success in finding drugs against the common cancers of lung, breast, and colon. It was therefore decided to change the drug screen to include slow growing human cancers implanted in immune-deficient mice that did not reject the foreign tissue. 160 ... ~ 2S '·: .' . .. ,. .. ., .. . ., "' 140 ..-. ~ .. .§ If ' .. 120 g' ., I'< 20 ... ' ... ' 100 "" •> > •· . 80 ""'::,~- • + "'• t" ' 60 . ''• _,,r .. . · •. i::; ' ,, ,,., C • 1S ., "' :;; ... 40 .g' m • , • ~ 10 . ' ,. . ' ~ ' ' .. ••, t 1· ' ' ' n ' • .., '·,' .,. i It· ,, " . 20 0 "::, 1 10 100 1000 Cydophosphamide (Cytoxan) (mg/ kg/day) Figure 20.3. An example of a dose-response test as designed by Abraham Go ldin for screening of chemical compounds in search of potential new anticancer drugs. Each point represents the median survival time of a group of 10 mice injected with the test compound (in this case cyclophosphamide) with the dose indicated in the horizontal axis. Eight days before the start of treatment, the mice were implanted with L1210 leukemia. Untreated mice had a median survival time of 10 days. With increasing dose, the survival time increased to a peak of 29 days, corresponding to 145% increase in survival time. Beyond that optimum survival, further increase in dose, reduced the survival time, as mice were dying due to toxicity of the test compound. A positive test in the screen was considered to require that, at the optimal dose, the test compound 404 K. W. Kohn Drugs Against cancer CHAPTER 20 increased the survival time by more than 40%. An increase beyond 100%, as in this test of cyclophosphamide, was considered to indicate a powerful drug action. (From (Goldin and Venditti, 1962).) The Linear Array and a Decision Network Committee. In 1966, the NCI chemotherapy program undertook a concerted effort to diagram the logic of the many decision steps through which a chemical compound or natural (biological) product would pass from acquisition through screening, toxicology, pharmacology, etc. to approval for clinical trial. The steps in the decision network logic were diagrammed in the form of a Linear Array based on a Convergence Technique developed by Louis M. Carrese and Carl G. Baker of the NCI staff. The Linear Array of the decision network logic was then developed over a period of several weeks by a team of NCI working scientists and planning specialists (Rothenberg and Terselic, 1970). The result was a huge logic diagram, far too large to show here. To implement the operation of the Linear Array, a Decision Network Committee of about 30 NCI scientists was appointed, whose job was to decide, based on evidence provided by relevant parts of the NCI program, whether a compound passes from one step to another through the Linear Array's logic network. The Linear Array served to focus the attention of the Committee on what decisions had to be made for each compound, thereby greatly increasing the number of compounds that could be managed at each meeting. As a member of the Decision Network Committee, however, I sometimes felt called upon to mark a decision ballot yes or no based on meager relevant data. Over perhaps about 10 years on the Committee in the late 1960s - 1970's, there were no more than one or two truly new types of clinically effective compounds detected for the first time by the screen. There were nevertheless some useful findings already mentioned. One was that doxorubicin (Adriamycin) was better than daunorubicin (daunomycin) in the test systems (Table 20.1). Another was that that intermittent treatment of leukemia in an intermittent schedule (every 4 days) was better than daily treatment (Table 20.2). 405 K. W. Kohn Drugs Against cancer CHAPTER 20 Table 20.1. Comparison of two chemically closely related drugs, daunorubicin and Adriamycin (doxorubicin), for their activity in the four-mouse-tumors test systems, showing the superiority of the latter (Johnson and Goldin, 1975). Optimal % increase in lifespan (cures) Experimental tumor system Daunorubicin Adriamycin Leukemia Ll210 4-0 (0/10) 68 (0/10) P388 leukemia 125 (0/10) >200 (5/10) Bl6 melanoma 145 (2/10) >200 (8/10) Lewis lung carcinoma 0 (0/8) 20 (0/8) Table 20.2. Intermittent treatment with methotrexate (drug given every 4 days) was better than daily treatment for mouse leukemia L1210 and for patients with acute lymphatic leukemia ((Goldin et al., 1971). Leukemia Ll210 Acute lymphocytic leukemia Median Median duration Median survival time of remission lifespan Schedule (days) Schedule (weeks) (weeb) Daily, days I-death 34 Daily 9 64 Evtl'y 4 days, day I-death 53 Twice weekly 50 91+ Control 14 M ini- and econo-scr eens. In 1970, or thereabouts, two types of smaller screens were evaluated. The first, called mini- screen, addressed the problem that many new compounds could not be tested because the amount of compound produced by synthesis was insufficient for testing in the full screen. The second, called econo-screen, aimed to reduce the cost per compound tested. Both screens reduced the number of animals used and reduced the number of injections from daily to two or three at specified times after the tumor was implanted. Eventually, a standard first screening of a new compound used three animals per dose. The cost per compound tested was thereby approximately halved (not to mention the reduced animal suffering- an ethical issue that could be debated). It was concluded that this reduced screen was almost as effective as the previous full screen (Goldin, 1973). 406 K. W. Kohn Drugs Against cancer CHAPTER 20 A new screen: human cancer xenografts in immune-deficient mice. The leaders of the NCI mouse cancer screen felt increasingly frustrated that after many years and testing of huge numbers of chemical compounds the screen had picked up hardly any truely new drugs clinically active against the major solid tumors, such as cancers of lung or colon. The previous mouse screen was therefor replaced in 1975 by a screen that included human cancers implanted in mice that were immune-deficient, so that they did not reject the human cancer tissues (De Vita and Chu, 2008). The screen included tumors derived from a human lung cancer, a human colon cancer, and a human breast cancer. These "xenograft'' tumors were slow growing, and a complete test could require 60-90 days making it necessary to reduce the number of compounds that could be tested per year. What was looked for was a reduction in the growth rate of the tumors, as measured either by reduced tumor size or weight. This drastic overhaul of the anticancer drug screening process entailed major reorientation of staff, procedures, and of the laboratories and companies that were contracted to do the work. The Associate Director of the drug screening and development program, Vincent DeVita, discussing the objectives of the new screen in a staff meeting, said that it was to be viewed as a 5-year experiment to see whether the human tumor xenograft testing would pick up new, previously unknown, drugs that would be active against some of the major human solid tumor cancers in the clinic. Because of the complexity and expense of this new screening panel, however, the number of drugs screened was reduced from about 40,000 per year to about 10,000 (De Vita and Chu, 2008). The flow through the screen was designed to start with 15,000 compounds per year selected from literature reports or voluntary submissions. The compounds would first be tested in a prescreen for activity in mouse leukemia P388, the most sensitive of the mouse tumors of the previous screens. An estimated 500-1000 compounds that passed the prescreen would then be tested in the new screen's xenograft tumors derived from human breast, colon, and Jung cancers. Natural products would flow through the same screen sequence, except that the prescreen could be in a KB -cell culture when quantity oftest material was limiting. Compounds could bypass the prescreen based on data from other anticancer systems or from biochemical or biological assays. The xenograft screen however had the downside that a complete test would take much longer than was the case for the previous mouse tumor screen: an estimated 60-90 days, which severely limited the number of compounds screened per year (Goldin et al., 1979). Another problem was that the immune-deficient mice used in the xenograft screen may become infected, which on at least one occasion decimated the mouse colony. Figure 20.4 shows an example of a response of a human breast cancer xenograft to a drug, hexamethylmelamine (HMM) that had failed to produce positive responses in the L1210, P388, B16, or Lewis lung tumor mouse cancers of the previous screen. This was perhaps 407 K. W. Kohn Drugs Against cancer CHAPTER 20 one of the few significant findings of the xenograft testing program. However, the drug had already passed a phase I clinical trial in 1965 (Wilson and De la Garza, 1965). It seems that the new screen failed to discover any truly new clinically active drugs and that the 5-year experiment of screening in human tumor xenografts announced by then Associate Director DeVita in 1975 was not a success. The failure could be attributed in part to the long time required per test and by the difficulty managing the infection-prone immune-deficient animals. HMM, by the way, is an interesting drug; it is activated by enzymes in the liver and intestines to produce an alkylating agent that can attack DNA by binding covalently to it and producing DNA inter-strand crosslinks. The drug produced tumor responses in patients with cancers of ovary, breast, lymphoma, and small cell lung cancer (Ames, 1991; Ross et al., 1981). 50 100 mg/kg HMM 10 5 "'·f ~ 200 mg/kg HMM 1.0 ,i 0.6 I- D .~ •• a; 0 .1 0.05 0.01 ...._._..__.___._..._.,_,~_..__.___._..._.,_, 0 6 10 15 20 25 30 35 Dav After Initial Treatment Figure 20.4. Response to hexamethylmelamine (HMM) by MX1 human breast cancer xenografts in immune-deficient mice (Goldin et al., 1979). HMM inhibited the net growth of the tumor and, at the highest dose (400 mg/kg), destroyed the tumor almost completely. A new screen: human tumor "stem cell"or colony -forming assays. I must now tell the story, as I remember it, of how a so-called human tumor stem cell assay temporarily became a new screen for cancer drug discovery. It was in 1980, I think, that then Associate Director Vincent DeVita, having decided that the human tumor xenograft screen was not fruitful, called a Technical Review Committee to review a proposal for a 408 K. W. Kohn Drugs Against cancer CHAPTER 20 new screen submitted by Sidney E. Salmon of the University of Arizona College of Medicine, Tucson, Arizona. In my first reading of the proposal as a member of the committee, I felt that at last we might replace the mouse tumor screens by one focused on human cancer cells in culture. On more careful reading, however, I was distressed to find what I thought to be a fatal flaw. The proposal was to screen using a "human tumor stem cell" assay developed by Sidney Salmon and Anne Hamburger (Salmon et al., 1978). The assay started with human tumor tissues derived from surgical specimens, which were then minced by a procedure that yielded a suspension of mostly single cells. A given number of cells were then deposited onto a layer of soft agar on a plate. After being incubated for 2 weeks, the number of colonies that had grown were counted. A drug, chemical compound, or natural product that reduced significantly the number of colonies grown was considered to have given a positive result. A major problem, as I saw it, was that only a limited number of assays could be done from any one tissue samples. Therefore, tissues had to come from a series of patients. Although they came from the same cancer type, say breast cancer, the drug sensitivities of the tissues likely would vary from patient to patient, making interpretation difficult I argued against approval, but the Committee voted by a narrow margin to approve the proposal. DeVita may have felt that the existing large apparatus for moving drug candidates through the Decision Network could not continue to sit idle, and no viable alternative screen proposal was at hand. A massive screening program, however, would have an inertia that would resist attempts to change its direction. As the human "stem cell" assay proceeded, some of us felt that significant problems were becoming increasingly apparent. To begin with, the ratio of colonies grown to number of cells put on the plate (the so-called plating efficiency) was extremely low (Figure 20.5). Typically, only one colony grew for every 10,000 cells put on the plate. That meant that the colonies that grew in the presence or absence oftest drug might have come from different kinds of cells, clouding interpretation. Also, unusual colony survival patterns suggested another problem - that the tissue mincing often left cell aggregates or clumps, which might be what was being counted rather than colonies growing from single cells. I had wondered why I had not seen microscopic confirmation that the tumor mincing was generally effective in producing dispersions of single cells on the agar plates. The researchers, even though highly respected, seem to have been bli nd to checking this out carefully, since Agrez and colleagues at the Mayo Clinic in Madison, Wisconsin soon reported that the various tumor disruption methods generally produced cell aggregates that could be seen within a day after plating and that grew into colonies over the next 14 days (Agrez et al., 1982) (Figure 20.6.). The presence of cell aggregates was found to affect the dose dependence of a drug effect in a way that would complicate the apparent cell kill fraction and thus whether a test was positive or negative (Rockwell, 1985), which may be one reason why the researchers judged a large fraction the drug tests to be uninterpretable. Moreover, the NCI researchers who reported those difficulties with the screening results had by 1985 stopped using the term "stem cell assay" and instead referred it as "colony-forming assay" (Shoemaker et al., 1985). 409 K. W. Kohn Drugs Against cancer CHAPTER 20 Becoming concerned about the presence of cell clumps in the disrupted tumor samples, the screen researchers tried to subtract the initially present cell clumps from the final "colony" count (Shoemaker et al., 1984). But, as can be seen in Figure 20.6. in the report by Agrez et al., colonies can grow from small cell aggregates that may not appear as obvious clumps (Agrez et al., 1982). The researchers were also concerned about the likely large and variable component of inherently non-dividing cells in the tumor tissue samples (Shoemaker et al., 1984). Another difficulty may have been that the tissue samples of a given human cancer type, derived from different patients, and then subjected to a cell dispersion process could have led to variable cell clumping and different cell types with different sensitivities to the test drug. As these problems became apparent in practice, it was eventually decided that they could be evaded by carrying out the assays instead on cultures of well -characterized cell lines that were originally derived from a single tumor of one of several cancer types. The new screening assay was to be on 60 cell lines (the "NCl-60") of several cancer types (Shoemaker et al., 1988). That story will be told shortly. The human "stem cell" or colony-forming assay, however, was dropped on or a little after 1985, a few years after a new Associate Director, Michael R. Boyd, had come to lead the program. Although the screening staff was still trying to improve the assay, Boyd had doubts. My recollection is that he called the entire staff together and began by saying that he had a problem and needed their help. There was about to be a Board meeting where he was called upon to report the progress of the screening program. But, after reviewing the data, it seemed to him that the current screen was failing. Several of the staff then agreed that the screen wasn't working adequately. It seemed to me as if it could now be admitted that the Emperor Had No Clothes. That is my recollection of how Boyd cleared the way that then led to the development by some of the same researchers of the NCl -60 assays and made possible the highly successful gene expression analysis program (Shoemaker et al., 1988), which is the subject of the next section of this chapter. 410 K. W. Kohn Drugs Against cancer CHAPTER 20 TABLE Jll. Pl,ting Efficiencies in lhe Hu m, n T umor Stem Cell System• Median number of Median% colonies/dish plating efficiency Tumor type (rnnge) (range) Neuroblnstomo 8 1 (1 2- 20,000) 0.0 I (0.002- 4.0) Ovar ian S4 (6- 650) 0.01 (0.001-0.1) Breast so (4- 60) 0.0 I (0.0008- 0.01) Melanoma 32 (10-21 0) 0.006 (0.002-0.04) Colorectal 26 (10- 118) 0.00S (0.002- 0.02) • Based on 5 00,000 nucleated cells plated per dish. Figure 20.5. Colonies formed from 500,000 cells, derived from dispersed human tumor tissue, that were deposited on a plate of soft agar. The number of colonies that grew from a given tumor type varied greatly ("range") and was tiny compared to the number of cells put on the plate (plating efficiency). A colony might have grown from a single "stem" cell or from a cluster of viable cells on the plate. Moreover, the cells plated would have included normal fibroblasts and lymphoid cells that are normally present in tumor tissues and could help neighboring tumor cells to grow. (Copied from (Von Hoff et al., 1980).) Figure 20.6. Aggregates of cells from disrupted ovarian carcinoma tissues after 1 day (left) or 10 days (right) of incubation on an agar plate (Agrez et al., 1982). The cell clumps on the plate after 10 days could have grown larger and been mistaken for colonies grown from single cells. Most of the clumps could be seen to have grown from small cell aggregates already visible after 1 day. 411 K. W. Kohn Drugs Against cancer CHAPTER 20 The NCl-60 and CellMiner stories. A new scr een: huma n NCl -60 cell lines. Having deemed the previous screen using human tumor colony-forming assays a failure, the anticancer drug screening staff directed their attention to human cell lines derived from various human tumors and designed survival assays to gauge the responses of the cells to many anticancer drugs and candidate compounds (Alley et al., 1988; Shoemaker et al., 1988). The new screen avoided the major pitfalls of the previous screen. First, cells of each line used in the survival assays came from the same passaged cultures, so the cells of a line were always of the same kind -- which is what it means to be a cell line. That helped to get reproducible results. Second, the cultures grew largely as single cells, free of clumps or aggregates. This would be less problematic also because the assay measured the growth of viable cells on a plate, rather than survival of colony-formation number. The staff, particularly Robert H. Shoemaker and his colleagues, devoted much effort developing a colorimetric viability assay that was sufficiently sensitive, reproducible, and that could be automated to measure the large numbers of compounds required by the screen. The protocol for how the original assay was conducted is summarized in Figure 20.7. Later modifications further improved the reproducibility of the assay. The new primary screen included a large number of lines derived from some of the major human tumors, including leukemia, melanoma, and cancers of breast, ovary, Jung, colon, kidney, and brain (Shoemaker et al., 1988). It was hoped thereby to find new drugs that may target one or another of those tumor types. In all, a panel of 60 cell lines was eventually selected that came to be known as the NCl -60. Add Ad d XTT Assay Plate o,ugs XTT Cells l I I Mix '24 hr 6 days 2-4 h r I Read 0 .0 . 450 nm 1ncub atP. 96 • well plates at 37° C Figure 20.7. The protocol for the colorimetric assay using a tetrazolium reagent (XTT) to measure inhibition of the growth of viable cells (Schoemaker et al., 1988). Only viable cells were able to produce the color that the assay measured. The test drug or compound was added 24 hours after seeding the cells of a given line on the plate. After allowing the cells to grow for 6 days, the XTT reagent was added and 2-4 hours later the intensity of the resulting blue-violet color (450 nm) was measured. 412 K. W. Kohn Drugs Against cancer CHAPTER 20 A drug-sensitivity analysis called COMPARE applied to the NC/-60 cell lines. In 1988, Kenneth D. Paull of the NCI staff devised a bar-graph display of the survival pattern of the cell lines in response to a drug or compound. He then developed an algorithm to quantify the difference between the survival patterns of individual compounds or group averages. This drug sensitivity display method became standard and extensively used. Paull gauged the sensitivity of a cell line to a given drug or compound as the logarithm of the IDS0 (dose producing 50% inhibition of the growth of viable cells). He then calculated a "mean graph" display centered on the mean sensitivity for the entire cell line panel (mean log(IDS0) for all the cell lines) (Paull et al., 1989). Figure 20.8 shows the first published mean graph displays using this procedure, which he dubbed COMPARE analysis, because it facilitated comparing the sensitivity patterns of different drugs or compounds (Shoemaker et al., 1988). This early example had SO cell lines; later a standard set of 60 human cancer cell lines was used: the NCl -60. In 1998, Glenda Kohlhagen, Ken Paull, Yves Pommier, and others in our Laboratory noted that a compound of distinctive chemical structure, NSC 314622, an indenoisoquinoline, produced a mean graph for growth inhibition of the NCl-60 cell lines that was highly correlated with the topoisomerase-1blocking drugs camptothecin and topotecan (Chapter 11) (Figure 20.9) (Kohlhagen et al., 1998). They showed, using our DNA filter elution methodology (Chapter 9), that the novel compound indeed produced the protein- associated DNA strand breaks we had found to be characteristic of topoisomerase inhibitors (Ross et al., 1979). Pommier and his colleagues went on to prepare several indenoisoquinoline derivatives (Figure 20.10) and found that the compounds had properties differing from camptothecin that merited further investigation (Kohlhagen et al., 1998; Marzi et al., 2018; Marzi et al., 2019). 413 K. W. Kohn Drugs Against cancer CHAPTER 20 8CNU (NSC 4099621 81eomycin tNSC 12~6) Mean IC$0 MHn ICso Fi:brOblHI [ ~ Leu.. mlo [ Leukemia [ '~"[ -~"[ ~ = SCLC [ :-:- SCLC [ ~ ~ Colon [ Colon [ 8rnst( 8reu1 [ CNS[ ( CNS[ t Melonom•[ ~ Mtlonomo [ E o,,~.., [ ~ o.."•• [ F ~ ► Ronol [ Aon•{ 1.5 w 10· 5' M 1,2 • 10· ' M Figure 20.8. The first published example of Ken Pauli's mean waph method (COMPARE analysis) for comparing the sensitivity patterns of different drugs (Schoemaker et al., 1988). This early example included SO human cancer cell lines derived from several tissues of origin. It showed that the two chemotherapy drugs, BCNU (bis(chloroethyl) nitrosourea) and bleomycin, had very different sensitivity patterns. In this display, bars to the right of the center line, indicated high sensitivity relative to the mean ICS0 (drug concentration producing 50% inhibition of viable cell growth). Thus, leukemia cell lines were sensitive to BCNU but relatively insensitive to bleomycin, whereas the converse was true particularly for the renal cell lines. Drugs of the same type, such as chloroethylnitrosoureas, on the other hand gave similar mean graphs. 414 K. W. Kohn Drugs Against cancer CHAPTER 20 t(llll(n'(C,U ~;ii,~, --- -1=--------1~ -------1==-- l(llf •t - •·'2H ........ -s-11 SIi 4'U9/AtCC ~ '°"''°' -;-:. ---- -~--- --~-E- c,n ,,.,. c_.,. - - • - - - - - - WCl•ott,6, ,..,.azs ue 1•UZ2,1 ..... l!e l • awb ---- - -- - -- ""1•11$.l2: , Colc,nc-t" IICC• ""9 lltl•116 -- --- lt: l •IS ... "" ~",,. cue...., t,,H,,I - ' -- - ...... :,t• M U,\J.9 sn•n -- -- "" ---- -• _,.... ••t- VlO( IIIYI •••• • •••• •••••••••M••••••••••••• "" __., i (•l'(l•2 $(-l• U t(◄l•'l 1.11.«•151 ....,,., -- -,---- -------1 ..... 0.-•rhn t•.,.,. - --------i---- ..... -·• -·• - -, .- -.. -=--- -=- -· .--• -- --• t(-0,,•) •- • c-•• '"' ~" CM l •I ll(f•J,9:J ~-~ 11:-10 \.O· lJ ::~4S •1- - - - 1. • - •1- - .,.Nl•l•~•" • -• • •• • •• . . . . . . . . . . . . . . . . . . . . . . . - .....,.., c.,...,., ~" ........ ~.,.,,m - JOtfl.\:ll •tfS ~ !'O&•lll--OS - -- - - --------· ------------------- ---- --- --- --- - ·• lt'l•14t • 1•U U .. , Figure 20.9. The novel compound, NSC 314622 had a mean graph for growth inhibition of the NCl -60 cell lines (middle) that correlated well with the mean graphs for the topoisomerase-1inhibitors camptothecin (left) and topotecan (right) (Kohlhagen et al., 1998). 10 •"'> A II ,<' 12 18_ 1 OH NSC 314622 Camptothectn Figure 20.10. The novel topoisomerase I inhibitor, NSC 314622, an indenoisoquinoline (left), had a multi-ring structure resembling in shape that of camptothecin (right). Several variant compounds were prepared in which the methyl group at the red arrow was replaced by other substituents. In some compounds, the two methoxy groups on the ring on the left were removed and a fluorine was added at the blue arrow (Kohlhagen et al., 1998; Marzi et al., 2018; Marzi et al., 2019). 415 K. W. Kohn Drugs Against cancer CHAPTER 20 Gene expression data for the NCl-60 human cancer cell lines. At staff meetings, I had often urged we think about acquiring molecular data for the NCl -60 cell lines and relating it to drug sensitivity data. It was some years, however, before the technology and data analysis tools, as well as staff members with the necessary skills and determination were available to us. This effort came to a head when John N. Weinstein joined our Laboratory. John had the necessary skills and determination in spades. He also had the organizational skills to take on and lead a group within the Laboratory, as well as outside collaborators to push ahead on this complex task. Their success in this effort with data analysis methods they developed were shown in two sentinel papers published in 2000 in Nature Genetics (Ross et al., 2000; Scherf et al., 2000). In the first of those papers (Ross et al., 2000), they explained how they measured gene expression by first robotically spotting 9,703 human cDNAs as microarrays on glass microscope slides, and then subjecting those cDNA microarrays to hybridization with fluorescence-labeled cDNA obtained by reverse-transcription of mRNA from each cell line. In that way, they got a measure of the complementary mRNA (reverse-transcribed cDNA) of each gene in each cell line. There were many quality-control issues to overcome, and they explained the details of how they did that; one of the consequences was that the number of well-defined human genes that they could measure reliably with the technology of the time was only 3,700. In the first analysis reported by Ross, Scherf, Weinstein, and their many coworkers and collaborators (Ross et al., 2000) (Figures 36.11 and 36.12), they applied a clustering algorithm to the average expression difference of a panel of genes between every pair of cell lines. The genes selected for the panel, 1,161 in number, were those that showed relatively large expression differences over the cell li nes, so as to be able to contribute significantly to the expression differences to be analyzed. It was satisfying to see that the expression differences of the genes sorted the cell lines, to a large degree, according to the tissue type each cell li ne came from. This was particularly striking for cell lines that came from leukemias, colon cancers, kidney cancers, melanomas, and cancers of the ovary (Figure 20.11). They also displayed the data as a two-dimensional cluster diagram. The vertical axis showed the genes, clustered according to their expression in the cell li nes. The horizontal axis showed the cell lines, clustered according to expression of the genes. Figure 20.12 shows the section of the cluster diagram where the melanoma cell lines clustered together. Several genes known to be expressed mainly in melanoma were present in the section shown in the figure (Ross et al., 2000). However, several other genes known to be expressed particularly in melanoma were absent, probably because of the number of melanoma cell lines in the NCl -60 was too small. When data for a larger number of cell li nes 416 K. W. Kohn Drugs Against cancer CHAPTER 20 became available, the missing melanoma genes showed up clearly, for example, the melanoma driver gene MITF (see Figure 20.19). In the second paper (Scherf et al., 2000), they selected 1,376 genes for analysis in the NCl - 60 cell li nes, based on showing large differenced in expression in the cell lines, and the cell lines again tended to cluster according to tissue of origin (Figure 20.13A). The cell lines also tended to cluster according to tissue of origin when clustered based in sensitivities of the cell lines to 1,400 compounds (Figure 20.138). So, tissue of origin was reflected in both gene expression pattern and drug sensitivity pattern. Moreover, the sensitivities of the cell lines to a set of 118 drugs of known likely mechanisms of action tended to cluster according to those mechanisms of action (Figure 20.13C). Clustering according to mechanism was also seen when based on the correlations between drug sensitivity and gene expression patterns (Figure 20.13D). (Those correlation values were derived by multiplying the cell line-drug sensitivity matrix by the transpose of the cell line-gene expression matrix.) The central conclusion here was that the drug sensitivity and gene expression data for the NCl -60 human cancer cell lines both contained information about tissues of origin of the cell li nes and the mechanisms of action of the drugs. It encouraged the development of similar data sets for larger numbers of cell lines that could lead to more and firmer conclusions. ·1 .00 breast -0 60 prostate non-small-lung -0 20 020 r1il CNS renal ovarian leukaemia colon melanoma Figure 20.11. Cell-line dendrogram with colored branches to reflect tissue of origin, based on hierarchical clustering of the expression of 1,161 cDNAs in the NCl -60 cell lines (Ross et al., 2000). The 1,161 cDNAs were chosen (out of a total of9,703) whose expression levels varied at least 7-fold in at least 4 of the 60 cell lines; that was done in order to select genes with the greatest differences in expression over the NCl -60 cell lines. The dendrogram shows that the cell lines tended to cluster according to tissue of origin. 417 K. W. Kohn Drugs Against cancer CHAPTER 20 60 cell lines r l ,~ --- Melaoom, bcaoch f' • . . . . . . . . .. ·•~~g:; i E " •' '. • •• '' ' ' ' •• • ' '' '' CMOAT G2 PROTEIN PTPRZ1 TYROSINE PHOSPHATASE ZETA MBP MYELIN BA SIC PROTEIN MBP MYELIN BASIC PROTEIN .... BCHE BUTYRYLCHOLINESTERASE 11' .1 ·~- " SCRAPIE RESPONSIVE PROTEIN 1 '• ........... .. ..... ARTI MONO·ADP·RIBOSYLTRANSFERASE ~ ~ SIAT6 ALPHA2 ,3-SIALYL TRANSFERASE 'I I ACP5ACID PHOSPHATASE TYPE 5 • •' RXR(; RFTINOID X RFCFPTOR C'~MMA VI Q) • • I . 'I I I I I ,I I OCT DOPA CHROME TAUTOMERASE TYR TYROSINA SE C: MUTYH MUTY HOMOLOG Q) • ,' ....... 51008 S•100 B ETA c:, I .1 '' ' • • I ' • 'I ' • ............ EDNRB ENDOTHELIN RECEPTOR TYPE B TYRP 1 TYRO SINASE.RELATED PROTEIN 1 SIAT8A ALPHA2.8.SIALYLTRANSFERASE . ' • • MLANA MELANOMA ANTIG EN RECOGNIZED BY T·CELLS G M68 • ·I ...' , " • PONP2 NUCLEO TIDE PYROPHO SPHATASE 2 ERDA3 CAGH3 .,, • • • ERDA3CAGH3 CPB2 PLASMA CARBOXYPEPTIDASE B ..................... LGAL53BP GALECTIN-6 BINDING PROTEIN A2M ALPHA-2-MACROGLOB ULIN Figure 20.12. Part of a two-dimensional hierarchical cluster diagram of genes (vertical axis) versus the NCl-60 cell lines (horizontal axis). The part of the full cluster diagram shown is the section where the melanoma cell lines clustered together, as shown by the vertical strip of red pixels on the right A red pixel indicated high expression ofa gene (cDNA) in a cell line. (Additional genes known to be highly and specifically expressed in melanoma were later revealed when a larger number of melanoma cell line became available for analysis.) 418 K. W. Kohn Drugs Against cancer CHAPTER 20 A B C D 2 1.S 2 1.5 1 ~--..-- Figure 20.13. Dendrograms for average-linkage hierarchical clustering of the NCl-60 human cancer cell li nes. For the analysis, 1376 genes were selected, based on showing large differences in expression in the cell lines (Scherf et al., 2000). A. Linkage hierarchy cluster tree of the NCl -60 cell li nes based on their gene expression patterns, showing clus tering according to tissue of origin. B. Linkage hierarchy cluster tree of the NCl -60 cell li nes based on their sensitivities to 1,400 compounds, again showing clustering according to tissue of origin. Abbreviations in A and B of the tissues of origin of the cell lines: LE, leukemia; LC, Jung cancer; CO, colon cancer; BR, breast cancer; OV, ovarian cancer; ME, melanoma; RE, renal cancer; CNS, brain cancer; PR; prostate cancer. C. Linkage hierarchy cluster tree of 118 drugs of likely mechanism of action based on the sensitivities of the NCl-60 cell lines to the drugs, showing clus tering according to mechanism of action. D. Linkage hierarchy clus ter tree of the 118 drugs of likely mechanism of action based on correlation of their drug activity patterns with their gene expression patterns in the NCl-60 cell lines, again showing clustering according to mechanism of action. Abbreviations in C and D of likely mechanisms of the drugs: A7, alkylation at guanine N7; A6, alkylation at guanine N6; AZ, alkylation at guanine NZ; Db, DNA binder (non-covalent); Df, antifol DNA synthesis inhibitor; Di, incorporation in DNA; Dr, ribonucleotide reductase inhibitor; Ds, DNA synthesis inhibitor; Pi, protein synthesis inhibitor; P90, binds HSP90; Rs, 419 K. W. Kohn Drugs Against cancer CHAPTER 20 RNA synthesis inhibitor; Tl, topoisomerase-1blocker; T2, topoisomerase-11 blocker; TU, tubulin binder; Uk, unknown. COMPARE analysis for g ene expression patterns of human cancer cell lines. With the success of COMPARE analysis for growth inhibition by drugs and compounds (Figures 36.8 and 36.9), an attractive idea was to apply the method to gene expression patterns in the cell lines. That meant acquiring a large amount of gene expression data on each of the NCl-60 cell lines, as described in the previous section and Figures 36.11-36.13. The power of the method was shown by our use of COMPARE analysis to examine gene expression expected in cells having an epithelial phenotype (Kohn et al., 2014). Among the NCl-60 human tumor cell lines there was indeed a very consistent pattern for expression of several genes characteristic for epithelial cell types (Figure 20.148). Moreover, there was a remarkably consistent inverse of expression patterns of epithelial versus mesenchymal cell types, as expected from the epithelial-mesenchymal-transition that many cancer cells undergo. The inverse patterns were, for example, clearly seen between the epithelial marker gene, E-cadherin, and the mesenchymal marker gene, vimentin (Figure 20.14A). COMPARE also revealed genes expressed in particular cell types. For example, the cadherin-17 gene tended to be expressed particularly in the colon cell lines (Figure 20.14C). The functional relationships of the epithelial genes were diagrammed in a molecular interaction map using the notation we had developed (Kohn, 1999) (Figure 20.15). The map shows the molecular interaction of the genes that hold epithelial genes together via tight junctions and adherence junctions. 420 K. W. Kohn Drugs Against cancer CHAPTER 20 Epilhtlalmarktr MeseftC.hytNI ma,1te, C COH17 A ( OHi (f.Qdhtrin) VIM lvimentin) __ ,____, . --- I r7" ·■- jiii '-iii:rn 1( .....-::; ~ • §.; 8 ' = ~1 i~ .9 't1 . .._ -::,,: ..,. !' ...... • 3 :: : I g _.,.,,,l ~ .. .. . . --- ' --- . . Average intensity (2 score) .-• ' Figure 20.14. COMPARE analysis of expression of epithelial genes in the NCl -60 cell li nes. (A) Inverse mean-graph gene expression patterns for an epithelial (E-cadherin) versus a mesenchymal (vimentin) gene. (B) Coherent of expression of several epithelial genes. (C) Gene expression pattern for a gene (cadherin-17) that was selectively expressed in the colon cancer cell lines (Kohn et al., 2014). 421 K. W. Kohn Drugs Against cancer CHAPTER 20 e ' 1'i{Jtl(JiN,cf/M$ 9 n I - ~ ' ~ 1 1 - ""' IIOTCKil 1 ..,._1Mllln l L ,~ 7 ICt.OHJ-4 I T[ ..... ,.,., J h U<>O I - ' T _L P\.$1/flmbrln Mk~ Ub I ~ ~-..... - lN:STOJ L..'. I MVOG8,C I _j_ I SERPIN8~n I I CI.DNr T I (PN31epeln3 I I»MYn02 J "'1,1 ~hO l fcrc'•n01 I -lt.t.811.A (MUii/iL~ ) lloWII. I ~ Ve,;lcle traMpolt I ~ I T I " O.lj)'!Oili,olllbO, I °"'" _L . ,21~ I~ ~ L..;i{ A,,,,.,,,..,,.,.. tl,t'~~J I ..... AA:FS I M~n-ns b,ctiom, , l COM\ll-- I H ~ ~~~ ' ) ,Ndng . . .<>I I l &w10$0t'IIU ~ °r COIKA1 I c:,11,.,,, ~ ~ I IU)(lfA7 l I C::nncin t I li$11P1, i l l ~ U IMU · 1 Meta&.tasi8 (r,-.......,./ J I ¥ T \ I CMIIAP'l / I T Figure 20.15. Part of a molecular interaction map showing functions at tight junctions and adherence junctions of epithelial cells (Kohn et al., 2014). Finding new drugs using COMPARE analysis as a scr een. The pattern recognition feature of Ken Pauli's COMPARE algorithm made it possible to search for compound that exhibited a unique growth inhibition pattern in the NCl-60 cell lines, which would suggest that the compound had a unique mechanism of action. As of March 1997, 74,196 compounds had been screened for unique growth inhibition pattern, and, after further studies, 5 went on to Phase I clinical trial as potential new anticancer drugs (Figure 20.16) (Monks et al., 1997). One of the most notable findings of the new screen was flavopiridol (also known as alvocidib), a plant product modified by organic chemists. Flavopiridol was found to block the cell's progress through the cell cycle by inhibiting cyclin-dependent kinases (CDK's), particularly CDK9, and became the first CDK inhibitor to enter clinical trial. Also contributing to the drug's anticancer action may have been its ability to suppress the expression of several proteins in the programmed cell death (apoptosis) pathway and of 422 K. W. Kohn Drugs Against cancer CHAPTER 20 the vascular endothelium growth factor (VEGF) that stimulates the growth of tumor- nourishing blood vessels (Wang and Ren, 2010). Flavopiridol was found to be effective in about half of patients with advanced acute myeloid leukemia when used as initial treatment to reduce the number of malignant cells and to accumulate the cells in a phase of the cell cycle where they would be sensitive to other cytotoxic drugs (Zeidner and Karp, 2015). Flavopiridol eventually failed in the clinic, however, because it inhibited a variety of cdk's with different functions. When inhibitors specific to cdk4 and cdk6 in the Rb pathways were developed, they became effective for treatment of common types of breast cancer (Chapter 33). Flavopiridol UCN -01 Depsipept ide NSC 649890 NSC638850 NSC630176 :__)l wi Quinocarmycin . . I Sp1camyc1n ana og • • ~-Ll_ - ~ lX) DX52-1 NSC 650426 NSC607097 Fiiure 20.16. Five compounds that passed a screen for unique growth-inhibition in the NCJ-60 cell lines, and, after passing additional tests, went on to Phase I clinical trial (Monks et al., 1997). The CellMiner story: mining molecular and pharmacological data. The NCJ-60 data were limited by the few cell lines of each tissue type in the data set. The success of the NCJ -60 analyses despite this limitation motivated the development of datasets for larger numbers of human cancer cell lines of various tissue types (Reinhold 423 K. W. Kohn Drugs Against cancer CHAPTER 20 et al., 2015). In addition to drug sensitivity and gene expression, data were assembled on gene mutations, gene copy number, DNA methylation, and protein expression. This presented a daunting complexity of analyses of various types that could be of interest. Our Laboratory met this problem by developing a set of software tools under the name Cel/Minerthat in essence facilitated the mining of the data (Reinhold et al., 2012). This effort was conducted William C. Reinhold, head of our Genomics and Pharmacology group, together with John Weinstein, Augustin Luna, Ken Paul, and Yves Pommier (Figure 20.17), their coworkers and collaborators. A recently enhanced version, Cel/MinerCDB (Luna et al., 2021; Reinhold et al., 2019), combined several databases covering larger numbers of cell lines and molecular as well as drug sensitivity data types (Figure 20.18). The new version allowed analyses between different databases. An example of a data plot for two genes expressed specifically in melanoma cell lines is shown in Figure 20.19. In addition, a database of drug sensitivity and gene expression was developed for cell lines derived from a particularly troublesome disease: small-cell lung cancers (Tlemsani et al., 2020). William Reinhold Kenneth Paull John Weinstein Yves Pommier Figure 20.17. Leaders who contributed to our Laboratory's information technology and CellMiner projects. William C. Reinhold, leader of our Genomics and Pharmacology group. Kenneth D. Paull, formerly chief of DTP's Information Technology Branch. Courtesy of The NIH Record, 7 April 1998. bttps-t(dtp cancer gov(tjmeliue/Oasb(mi!estooes(MZ COMPARE btro John N. Weinstein, formerly leader of our omics development group. Yves Pommier, Director of the Developmental Therapeutics Branch. 424 K. W. Kohn Drugs Against cancer CHAPTER 20 SCLC HCI. . Alman.ac CCLE CTRP GDSC NCI/DTP MD And.non AchlllH Cellllne 60 60 1,089 323 1,080 n es1 769 Single 24 '81 526 Combo 5 ,355 Mul•tk>n 9,307 1,667 1,667 18,099 DNA Copy• 23,232 23.3115 23.316 2•.so2 25,568 M.c.hyl• Uon 17,553 19,880 19,864 23,202 Mlcroam y (n) 25,040 RNA Eapr9'Hion M iCIWlffll)' (l092) 23,059 19.851 19,851 19,562 RNAS.q 23,826 ,,.. 52,604 - mJRNA m 800 RPPA 214 Protein Ex.prHslon I SWATH 162 3,167 452 I LC-MS - Me1-bo11te ExPtHtlon 225 Gene kllOCkoul CRISPR 18,119 Parameters Phenotyplc 10 1 3 2 Figure 20.18. Summary of data in CellMinerCDB version 1.4 (Luna et al., 2021; Reinhold et al., 2019). Drugs included activity levels, for example, for 24,047 compounds in the NCl-60 cell lines. There were 823 cell lines in the CTRP database and over 1,000 in the CCLE and GDSC databases. (RPPA, reverse-phase protein assay. SWATH, sequential window acquisition of theoretical spectra, mass spectroscopy.) 425 K. W. Kohn Drugs Against cancer CHAPTER 20 NLANA ( exp, CTRP~Broad•N IT) VS, MITF (exp, CTSlP..Broad•MIT) Pe11rson 00trelation (r)-= 0.6 , p-value=2.9e-81 12.s •• • •• •• 5,0 2.5 5 7 MITF ( exp, CTRP-Broad"""IT) • ll Figure 20.19. A CellMinerCDB data plot showing the specific expression of two genes known to function specifically in melanoma cells. Red, melanoma cell lines; blue, approximately 800 human cancer cell types of other tissue types. The plot shows that MLANA was expressed only when MITF was expressed beyond a certain level (about 7 on the horizontal scale), beyond which the expression of both genes rose. That\,vas consistent with MITF being a factor that stimulates the transcription of MLANA. The plot also shows that MLANA was expressed exclusively in melanoma cell lines. (The axes are scaled in powers of 2.) Using CellMinerCDB to explore the action offlavopiridol in acute myeloid leukemia (AML). A previous section of this chapter discussed flavopiridol ( also known as alvocidib) as one of the most notable discoveries in the NCl-60 cell li ne screen and the COMPARE algorithm. The drug went on to clinical trial and was found useful in the treatment of acute myeloid leukemia. In order to illustrate some of the capabilities of CellMinerCDB (Luna et al., 2021; Reinhold et al., 2019), I show some preliminary analyses of flavopiridol (alvocidib) in acute myeloid leukemia (AML) cell lines in the CTRP-Broad- MIT database (Figure 20.18) (Table 20.3; Figures 36.20 and 36.21). Table 20.3 illustrates how genes can be found whose expression is highly correlated with sensitivity to a given drug in cell lines of a given tissue type. By modifying the options, 426 K. W. Kohn Drugs Against cancer CHAPTER 20 you can also find genes whose expression correlates with the expression a given gene or drugs whose sensitivity correlates with the expression of a given gene or with the sensitivity to a given drug. Also, the datasets used can be selected and even mixed. The facility is completely flexible in these regards. Figure 20.20 is a plot of expression ofSl00G (the top gene in Table 20.3) versus sensitivity to flavopiridol (alvocidib) in AML cell lines. The plot suggests that sensitivity the drug would be high when expression of the gene is low, a relationship that has not previously been reported. This finding, however, is exclusive for AML cells. Figure 20.21 is a multivariate plot showing how the expression of the 8 genes at the top of Table 20.3 correlated inversely with sensitivity to flavopiridol (alvocidib) in AML cells. Explorations of these kinds could, and have indeed, translated to clinical application. Table 20.3. A set of genes showing high negative correlation of expression relative to sensitivity to flavopiridol (alvocidib) in AML cell lines. The data in the Table came from CellMinerCDB version 1.4 using the Compare Patterns option (btt12s·//discover uci nib gov/rscounect/cellrninercdh/). !iE1:iE. 1.ocaruJr31 ComlathlD E-~llu: SlO0G Xp22.2 -0.907 1.88E-05 TBC1D23 3ql2.2 -0.898 3.0lE-05 ABHD4 14qll.2 -0.872 1.02E-04 PARD3B 2q33.3 -0.869 1.15E-04 REXO2 llq23.2 -0.868 1.19E-04 MICAL2 llp15.3 -0.862 1.50E-04 UCAl 19pl3.12 -0.859 1.69E-04 TXK 4p12 -0.857 1.79E-04 LEFl 4q23-q25 -0.849 2.40E-04 RABGEFl 7qll.21 -0.845 2.81E-04 TJP2 9q13-q21 -0.844 2.88E-04 STXBP6 14q12 -0.842 3.0lE-04 AUH 9q22.31 -0.838 3.42E-04 PIP4K2C 12q13.3 -0.837 3.57E-04 DZIP3 3q13.13 -0.835 3.82E-04 PPFIAl llq13.3 -0.835 3.83E-04 GOLPH3L lq21.3 -0.835 3.86E-04 MTTP 4q24 -0.833 4.07E-04 LENGl 19q13.4 -0.832 4.19E-04 CTSL 9q21.33 -0.831 4.31E-04 427 K. W. Kohn Drugs Against cancer CHAPTER 20 HPSE2 10q23-q24 -0.830 4.49E-04 AVPll 10q24.2 -0.830 4.52E-04 DNASE2B lp22.3 -0.827 4.92E-04 TBKl 12q14.1 -0.827 4.93E-04 DSTYK lq32.l -0.826 4.96E-04 SIOOG {ex;>, CTRP.Broad•MIT} vs. alvoc::idib (act. CTIU'•8roio:Molfl) Pc11rson <orrclatlcn (r) • ·0.9 1, p•,;11h,,c • 1.9,c•OS ••• • • • •• •• • •• •• ,.• ll 22 2J 24 lllvoddlb (act, CTRP·Bn»d·MIT) Figure 20.20. This plot shows how sensitivity to flavopiridol (alvocidib; horizontal axis) correlates with the expression ofSl00G (the top gene in Table 20.3; vertical axis) in acute myeloid leukemia (AML) cell lines. The plot came from the Plot Data option in CellMinerCDB version 1.4 (httns·//discover uci nib eov/rscounect/cellrninercdh O. The facility is completely flexible in what data you chose for each axis of the plot. The result suggests that sensitivity of AML to flavopiridol (alvocidib) may be enhanced by reducing the expression of Sl00G. -- --c--. -I[- H,,_.,.,_,,.~-"""'-U..O.•• ~ ' ......,ec-,i_ _ _,_,.,ct,;, • .-ioo;a,p ,,,,-T9C1023 (W ~111>1~ ..,4>Ntol11 °"' ..~tllOGl:(17;> t:•ll""K,IIU_.(W • •P'XAJ._<11111 """'-""' 428 K. W. Kohn Drugs Against cancer CHAPTER 20 Figure 20.21. Multivariate display of a set of 8 genes whose expression is negatively correlated with sensitivity to flavopiridol (alvocidib) in acute myeloid leukemia (AML) cell lines (top 8 genes in Table 20.3). There were data for 12 AML cell lines (labeled at bottom) in the CTRP database. Sensitivity to alvocidib is in the top row. Expression of the 8 genes in the AM L cell lines is in the lower 8 rows. Red to blue: high to low sensitivity or expression. The display suggests that low expression of the 8 genes wou ld correlate with high sensitivity to alvocidib in AML cells. This display was obtained using the Multivariate Ana/ysis option in CellMinerCDB version 1.4 (httns·//discover nci nib eov/rscounect/cel1minen:dh/). References Agrez, M.V., Kovach, J.S., and Lieber, M.M. (1982). Cell aggregates in t he soft aga r "human t umour st em-cell assay" . Brit ish journal of cancer 46, 880-887. Alley, M .C., Scudiero, D.A., Monks, A., Hursey, M.L., Czerwinski, M .J., Fine, D.L., Abbott, B.J., Mayo, J.G., Shoemaker, R.H., and Boyd, M .R. (1988). Feasibility of drug screening wit h panels of human t umor cell lines using a microcult ure t etrazolium assay. Cancer research 48, 589-601. Ames, M.M. (1991). Hexamethylmelamine: pharmacology and mechanism of action. Cancer t reatment reviews 18 Suppl A, 3-14. Condon-Rall, m.e. (1994). The Army's Wa r against Malaria: Collaboration in Drug Resea rch during World War II. Armed Forces & Society 21, 129-143. DeVita, V.T., Jr., and Chu, E. (2008). A hist ory of cancer chemot herapy. Cancer research 68, 8643-8653. Goldin, A. (1973). The concept of t he mini and econo screen. In: Program Logic snd Scientific Reports (Bethesda, Maryland: Drug Research and Development, National Cancer Institute, Public Healt h Service, U. S. Department of Health, Educat ion, and Welfare), pp. 76-99. Goldin, A., Sandberg, J.S., Henderson, E.S., Newman, J.W., Frei, E., 3rd, and Holland, J.F. (1971). The chemotherapy of human and animal acute leukemia. Cancer chemot herapy reports Part 155, 309-505. Goldin, A., Schepartz, S.M., Venditti, J.M., and DeVita, V.T. (1979). Hist orical development and current st rategy of t he National Cancer Institute drug development program. Met hods in Cancer Research 16, 165-245. Goldin, A., and Venditti, J.M . (1962). A manual on quantitat ive drug evaluat ion in experimental t umor systems. Cancer chemotherapy reports Part 117, 145-178. Johnson, R.K., and Goldin, A. (1975). The clinical impact of screening and other experimental t umor studies. Cancer treatment reviews 2, 1-31. Kohlhagen, G., Paull, K.D., Cushman, M ., Nagafuji, P., and Pommier, Y. (1998). Protein-linked DNA strand breaks induced by NSC 314622, a novel noncamptothecin topoisomerase I poison. Molecular pharmacology 54, 50-58. Kohn, K.W . (1999). Molecular interaction map of t he mammalian cell cycle control and DNA repair systems. Mol Biol Cell 10, 2703-2734. 429 K. W. Kohn Drugs Against cancer CHAPTER 20 Kohn, K.W., Zeeberg, B.M., Reinhold, W.C., and Pommier, Y. (2014). Gene expression correlations in human cancer cell lines define molecular interact ion networks for epithelial phenotype. PloS one 9, e99269. Luna, A., Elloumi, F., Varma, S., Wang, Y., Rajapakse, V.N., Aladjem, M.I., Robert, J., Sander, C., Pommier, Y., and Reinhold, W .C. (2021). CellMiner Cross-Database (CellM inerCDB) version 1.2: Explorat ion of pat ient-derived cancer cell line pharrnacogeno mics. Nucleic acids research 49, D1083-D1093. Marzi, L., Agama, K., Murai, J., Difilippantonio, S., James, A., Peer, C.J., Figg, W.D., Beck, D., Elsayed, M .S.A., Cushman, M., et al. (2018). Novel Fluoroindenoisoquinoline Non- Camptot hecin Topoisomerase I Inhibitors. Molecular cancer t herapeutics 17, 1694-1704. Marzi, L., Szabova, L., Gordon, M., Weaver Ohler, Z., Sharan, S.K., Beshiri, M .L., Etemadi, M., M urai, J., Kelly, K., and Pommier, Y. (2019). The lndenoisoquinoline TOPl Inhibitors Selectively Target Homologous Recombinat ion-Deficient and Schlafen 11-Posit ive Cancer Cells and Synergize w ith Olapa rib. Oinical cancer research: an official journal of t he American Association for Cancer Research 25, 6206-6216. Monks, A., Scudiero, D.A., Johnson, G.S., Paull, K.D., and Sausville, E.A. (1997). The NCI anti- cancer drug screen: a smart screen to identify effectors of novel t argets. Anticancer Drug Des 12, 533-541. Paull, K.D., Shoemaker, R.H., Hodes, L., Monks, A., Scudiero, D.A., Rubinst ein, L., Plowman, J., and Boyd, M.R. (1989). Display and analysis of patterns of differential act ivity of drugs against human t umor cell lines: development of mean graph and COMPARE algorit hm. Journal of t he Nat ional Cancer Inst itute 81, 1088-1092. Reinhold, W.C., Sunshine, M., Liu, H., Varma, S., Kohn, K.W., Morris, J., Doroshow, J., and Pommier, Y. (2012). CellMiner: a web-based suite of genomic and pharrnacologic tools to explore tra nscript and drug patterns in the NCl-60 cell line set. Cancer research 72, 3499- 3511. Reinhold, W.C., Varma, S., Rajapakse, V.N., Luna, A., Sousa, F.G., Kohn, K.W., and Pommier, Y.G. (2015). Using drug response data to identify molecular effectors, and molecular "omic" data to identify cand idate drugs in cancer. Human genet ics 134, 3-11. Reinhold, W.C., Va rma, S., Sunshine, M., Ello umi, F., Ofori-Atta, K., Lee, S., Trepel, J.B., Meltzer, P.S., Doroshow, J.H., and Pommier, Y. (2019). RNA Sequencing of t he NCl-60: Integration into CellMiner and CellM iner COB. Cancer research 79, 3514-3524. Rockwell, S. (1985). Effects of clumps and clusters on survival measurements with clonogenic assays. Cancer research 45, 1601-1607. Ross, D.T., Scherf, U., Eisen, M .B., Perou, C.M., Rees, C., Spellman, P., Iyer, V., Jeffrey, S.S., Van de Rijn, M., Waltham, M ., et al. (2000). Systematic variation in gene expression patterns in human cancer cell lines. Nat Genet 24, 227-235. Ross, W .E., Glaubiger, D., and Kohn, K.W. (1979). Qualitative and quantitat ive aspects of intercalat or-induced DNA st rand breaks. Biochimica et biophysica act a 562, 41-50. Ross, W .E., McMillan, D.R., and Ross, C.F. (1981). Comparison of DNA damage by methylmelamines and formaldehyde. Journal of t he Nat ional Cancer Institute 67, 217-221. 430 K. W. Kohn Drugs Against cancer CHAPTER 20 Rothenberg, L., and Terselic, R.A. (1970). Commentary. Management of the Nat ional Cancer lnstitute's drug research program t hrough application of t he linear array concept . Cancer chemot herapy reports Part 1 54, 303-310. Sa lmon, S.E., Hamburger, A.W., Soehnlen, B., Durie, B.G., Al berts, D.S., and Moon, T.E. (1978). Quantit ation of differential sensitivity of human-tumor st em cells to anticancer drugs. The New England journal of medicine 298, 1321-1327. Scherf, U., Ross, O.T., Waltham, M., Smit h, L.H., lee, J.K., Tanabe, L., Ko hn, K.W ., Reinhold, W.C., Myers, T.G., Andrews, O.T., et al. (2000). A gene expression database for t he molecular pharmacology of ca ncer. Nat Genet 24, 236-244. Schoemaker, R.H., Monks, A., Alley, M.c., Scudiero, 0 .A., and al., e. (1988). Development of human t umor cell line pa nels for drug screening. Prog Clin Biol Res 276, 26S-286. Shoemaker, R.H., Monks, A., Alley, M .C., Scudiero, 0 .A., Fine, 0 .L., Mcl emore, T. L., Abbott, BJ., Paull, K.0., Mayo, J.G., and Boyd, M .R. (1988). Development of human t umor cell line panels for use in disease-oriented drug screening. Prog Clin Biol Res 276, 26S-286. Shoemaker, R.H., Wolpert-OeFilippes, M.K., Kern, O.H., Lieber, M.M., M akuch, R.W., M elnick, N.R., M iller, W.T., Salmon, S.E., Simon, R.M ., Venditti, J.M ., et al. (198S). Application of a human t umor colony-forming assay t o new drug screening. Cancer research 45, 2145-2153. Shoemaker, R.H., Wolpert-OeFilippes, M.K., and Venditti, J.M. (1984). Pot entials and drawbacks of t he human t umor stem cell assay. Behring Inst M itt, 262-272. Tlemsani, C., Pongor, L., Elloumi, F., Girard, L., Huffman, K.E., Roper, N., Va rma, S., Luna, A., Rajapakse, V.N., Sebastian, R., et al. (2020). SCLC-CellM iner: A Resource for Sma ll Cell Lung Cancer Cell Line Geno mics and Pharmacology Based on Genomic Signatures. Cell repo rts 33, 108296. Von Hoff, 0.0., Harris, GJ., Johnson, G., and Glaubiger, 0 . (1980). Initial experience wit h t he human t umor stem cell assay syst em: pot ential and problems. Prog Clin Biol Res 48, 113- 124. Wang, L.M., and Ren, O.M. (2010). Flavopiridol, t he first cycl in-dependent kinase inhibitor : recent advances in combinatio n chemot herapy. M ini reviews in medicinal chemist ry 10, 1058-1070. W ilson, W. L., and De la Garza, J.G. (1965). Phase I st udy of hexamet hylmelamine (NSC-13875). Cancer chemot herapy reports Part 148, 49-52. Zeidner, J.F., and Ka rp, J.E. (2015). Clinical activity of alvocidib (flavopiridol) in acute myeloid leukemia. leuk Res 39, 1312-1318. Zubrod, C.G. (1966). NCI chemot herapy Program, hisory, analysis, plans. Cancer Chemot herapy Reports 50, 349-540. 431 K.W.Kohn Drugs Against Cancer CHAPTER21 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@oib gov CHAPTER21 The DNA Repair Story: early discoveries. Introduction DNA damage from metabolic and environmental sources is unavoidable; cells therefore have evolved an astonishing set of mechanisms to counter the great variety of chemical damage that the cell's DNA can sustain. Defects in one or another DNA repair mechanism often lead to cancer, but cancer cells whose DNA repair mechanisms are defective or inefficient are vulnerable to DNA- damaging anti-cancer drugs. Much about DNA damage and repair first came to light from studies of how bacteria do it How our cells (or mammalian cells in general) do it, first came from studies of cells in culture, where studies could be carried out under precisely controlled conditions. Cell culture studies then unraveled much of the mystery of what anticancer drugs do to cells and how the cells respond. A specific DNA repair mechanism was already encountered and described in Chapter 2, namely the removal of drug adducts at the 06-position of guanines by methylguaninemethyltransferase (MGMT). But even that highly efficient repair process has limits - no biological process is perfect. Backup processes are therefore needed to clean up any adduct that may be left unrepaired. A normal dividing cell's DNA is constantly at risk of being damaged by environmental carcinogens or radiation, or by the rare but inevitable errors made by the machinery that normally replicates DNA. Many types of DNA damage are produced, and their frequencies of occurrence vary greatly (Table 1). It must have challenged evolution to create the collection of DNA repair 432 K.W.Kohn Drugs Against Cancer CHAPTER21 mechanisms to deal with the many different kinds of chemical damage that DNA can suffer (Figure 21.1) (Lindahl, 1982; Sancar and Sancar, 1988). Most cancer chemotherapeutic agents, including drugs and radiation, damage DNA in one way or another, and DNA repair mechanisms come into play to cope with the various kinds of damage. Cancer cells often are deficient in one or more DNA repair processes and then may become sensitive to drugs producing types of damage that the cells of the cancer are unable to repair. More generally, cancer cells often have defects in regulatory mechanisms, including those that govern responses to the genomic stress caused by DNA damage. These defects may impair the cancer cell's ability to cope with the drug-induced damage, for example by increasing the capacity of the repair machinery or by delaying DNA replication or mitosis to give more time for repair. Even though the cell has many different molecular repair machines, as well as control networks to give more time for repair, a small amount of damage inevitably gets through (Gudmundsdottir and Ashworth, 2006). The reason anticancer drugs work is often because the defects in cancer cells may allow more unrepaired damage to remain when the DNA begins to replicate or the cell begins to divide, which is when persisting DNA damage would be apt to kill the cell. Cancer cells often have DNA repair defects that make them susceptible to DNA-damaging anticancer drugs. Another kind of defect that often makes cancer cells susceptible is a defect in a "cell cycle checkpoint." These checkpoints consist of molecular circuitry that check whether it is safe for the cell to proceed from one phase of the cell cycle to the next. Unrepaired DNA damage triggers the checkpoint circuitry to delays the cell cycle in order to give more time for repair before the cell is allowed to progress to the next phase where persistent DNA damage could result in cell death. As impressive as our DNA repair armamentarium may be, we mammals are far from champions in that field. Far more impressive among animals are the tiny tardigrades and some even tinier rotifers (Figure 21.2), which can survive hundreds of times as much DNA damage than our cells could tolerate. In the course of their evolution, these creatures have acquired, by means of gene transfer, a large collection of DNA repair genes from various species of eukaryotes and bacteria. This helps them survive for years dried-out in a desiccated state in which they are subject to extensive DNA damage (Hashimoto and Kunieda, 2017). This seeming bit of trivia about DNA repair proficiency might eventually be matched by humans, if gene transfer technology became feasible and ethically acceptable: inserting extra DNA repair genes might extend lifespan and allow astronauts to survive (albeit not in a desiccated state) the radiation in space and extraterrestrial planets, moons and asteroids. The potential impact on cancer therapy is difficult to 433 K.W. Kohn Drugs Against Cancer CHAPTER21 pr edict but may be significant. Table 1Estimated frequencies of DNA lesions normally occurring in mam malian cells Damage Events per cell per day Single-strand breaks 55000 Depurinations 13000 Depyrimidinations 650 Guanine-06 methylation 3100 Cytosine deamination 200 Glucose-6-phosphate adduct 3 Thymine glycol 270 Thymidine glycol 70 Hydroxymethyluracil 620 Guanine-8 oxygenation 180 lnterstrand cross~ink 8 Double-strand break 9 DNA-protein cross-link Unknown From (Kohn and Bohr, 2001). Nuclear DNA repair pathways Mismatch Direct reversal Short patch Long patch Transcription- Global cou led enome Figure 21.1. An overview of the types of DNA repair mechanisms operating in the cell nucleus. From (Kohn and Bohr, 2001). 434 K.W.Kohn Drugs Against Cancer CHAPTER21 Figure 21.2. Scanning electron microscope image of a tardigrade: a champion DNA repairer. Discovery ofDNA repair. Discovery of DNA and its structure Before talking about the repair of DNA damage, I'd like to talk about how the genetic material was found to be made up of - well, DNA. But first I am reminded of Edwin Chargaff, discoverer of the DNA base-pairing rules (G=C and A=T), whom I encountered in 1953 during my first year at Columbia's medical school, the College of Physicians and Surgeons (P&S), when he conducted biochemistry laboratory classes. On asking him about -- admittedly wild -- ideas I had, he dismissed them out of hand; he certainly was not encouraging. My future wife, Elaine Kay Mogels, worked in a research lab at P&S and attended Chargaff's course on nucleic acids, but she didn't find it very interesting. We had his 2-volume work "The Nucleic Acids" but I did not refer to it much during the coming years, because my projects did not require the early chemistry details that it focused on. I also recall Chargaffs lecture at a molecular biology conference at Columbia in 1960 - I think it was - in which he denigrated the papers in the new molecular biology field that were coming out, even if published in journals like the Proceedings of the National Academy of Sciences, that he said were akin to what he might read in the New York Times. He thought it soft science, not up to par with the solid previous work on nucleic acids. In that regard, he extoled particularly the work of Johannes Miescher. Johannes Friedrich Miescher (Figure 21.3) wanted to find out what the cell nucleus was made of, about which next to nothing was known in 1868, when at the age of 24 he came to the University ofTubigen, Germany, to study with Ernst Hoppe-Seyler, a founder of the new field of biochemistry. In 1869, Miescher isolated from cell nuclei a strange highly viscous phosphate-rich material that he called "nuclein" --which was in fact DNA with some bound 435 K.W.Kohn Drugs Against Cancer CHAPTER21 protein. He had a hunch that the large molecules in his nuclein might be the genetic material, an idea that he expressed in a vaguely worded letter to his uncle in 1892, but this idea lay dormant for decades Oudson, 1979). Over the next three decades after Miescher's extraction of "nuclein" from cell nuclei, hard work by many researchers disclosed the chemical structures of the DNA constituents guanine, adenine, thymine and cytosine, and DNA was found to be made up of long chains of deoxyribose-phosphate bound to one or another of those 4 bases, which made up the four nucleotides: G, A, T, and C. A wrong idea emerged that DNA was a polymer of unvarying groups the 4 nucleotides. This "tetranucleotide hypothesis" implied boring tetranucleotide repeats that could not possibly be the chemical composition of genes. For a long time, many biochemists dismissed DNA as likely having a structural role and held that genes must be made up of proteins, which had complex structures that they thought to be commensurate with the complexity of genetic information. The controversy about the chemical basis of genes continued despite mounting evidence favoring DNA until 1944 with the definitive experiments by Oswald T. Avery (Figure 21.4) (Avery et al., 1944). Avery and his colleagues at the Rockefeller Institute in New York were studying peculiar phenomena that had first been observed by Frederick Griffith in 1928 in the pneumonia- inducing pneumococcus bacteria. Griffith's experiment was a bit complicated, but here goes: Each bacterial cell was able to grow into a colony on an agar surface. The colonies sometimes had a smooth appearance and sometimes a rough appearance, depending on the bacterial strain. Bacterial strains were called S or R, depending on whether they grew into smooth or rough colonies. S-strain pneumococci produced pneumonia in mice, whereas R-strains did not. There were 3 types of pneumococcus, called types I, II, and Ill; each type had Sand R strains that usually bred true to their particular strain. Injecting mice with an R strain of type II did not produce pneumonia, but - surprise! - R of type II mixed with heat killed S of type Ill did produce pneumonia! Moreover, some of the R type II apparently had transformed to behave like S type Ill!! It turned out that the transformation could be made to happen merely by incubating the two kinds of bacteria together in a nutrient-deficient broth. A genetic characteristic- indeed, a gene or genes - were being transferred from one bacterial cell to another. What Avery and his colleagues did was show that the gene transfer occurred when pure DNA from donor bacteria was mixed with recipient bacteria. They did many checking experiments to support their contention that the genetic information was in the DNA and not in a protein impurity. Some researchers nevertheless persisted for years thinking that an undetected trace of a protein 436 K.W. Kohn Drugs Against Cancer CHAPTER21 impurity, impervious to Avery's protein-digesting enzyme test, was the holder of the genes. A fond idea long held was difficult to discard. The phenomenon of genetic transformation was later found in several bacterial species. But it occurred only when the recipient bacteria were competent to take up the donor DNA What made bacteria competent became a question of intense investigation. As a post-doc in Paul Doty's laboratory at Harvard, I collaborated with Donald MacDonald Green using DNA transformation in Bacillus subtilis to show that a single nitrogen mustard- induced inter-strand crosslink abolished the DNA's transforming ability (Kohn and Green, 1966) (see Chapter 1). Successful genetic transformation of mammalian cells by DNA was first reported by Waclaw Szybalski (Szybalska and Szybalski, 1962) but was very difficult to reproduce consistently. The story of how modification of technique eventually made it an essential procedure in modern cell biology is told in Chapter 15. It was only a few years after DNA was accepted to be the bearer of genetic information that Watson and Crick presented a correct model of DNA structure (Figure 21.6). Figure 21.3. Johannes Friedrich Miescher (1844-1895) was one of the great biological chemists of his time. He received an MD degree from the medical school at Basel, Switzerland in 1868 and then joined Felix Hoppe-Seyler's laboratory at the University ofTiibingen, Germany, to investigate the constituents in cell nuclei. He succeeded in extracting from cell nuclei a highly viscous phosphate-rich material that he called nuclein, which was later found to be composed mostly of DNA Because of its high viscosity, Miescher correctly concluded that his nuclein was made up of very long molecules, and he suspected it to be the genetic material, an idea that was dismissed for decades before it was shown to be true. 437 K.W. Kohn Drugs Against Cancer CHAPTER21 Figure 21.4. Oswald Theodore Avery, Jr. (1877-1955) and his colleagues at the Rockefeller Institute in New York purified DNA and proved that it contained genetic information. His definitive experiments, published in 1944, became a famous landmark that propelled later studies of DNA and genetics. Figure 21.5. Erwin Chargaff (1905-2002) was professor of biochemistry at Columbia College of Physicians and Surgeons (P&S) from 1938 to 1970 and served as chair of the department until his retirement in 197 4. His family had moved to Vienna in 1914 from their home in Czernowitz, now Chernivtsi, Ukraine. He attended the Vienna Technische Hochschule and earned a PhD in chemistry from the University of Vienna in 1928. He became renowned for his discovery that the frequencies of the nucleotides (G, C, A, 11 in DNA, while differing among various organism, had a remarkable pattern that the frequencies of G and C tended to be equal, as was the case also for A and T. This "Chargaffrule" was later found to reflect the base-pairing in the DNA double-helix (G=C and A=T). 438 K.W. Kohn Drugs Against Cancer CHAPTER21 Figure 21.6. James D. Watson (left) and Francis Crick (right) with their DNA model circa 1952. How DNA repair was discover ed. When DNA is damaged by radiation or chemicals, the cell can usually repair the resulting mutation. I will begin by looking back at how this understanding developed. A key discovery about mutation of genes was made by Hermann J. Muller in 1927 (Figure 21.7), for which he was awarded the Nobel Prize in Physiology or Medicine in 1946. Muller had been studying hereditary changes in fruit flies and discovered that x-rays produced mutations in proportion to the x-ray dose. That finding. in a sense, initiated the field of DNA damage (even though it was not yet known that DNA was the genetic material) (Friedberg, 1997). The mutation story however dates back even further, to Charles Darwin's evidence, published in Origin ofSpecies on 24 November 1859. His evidence implied that hereditary changes in species were due to "mutation" of genes (although the term "mutation" was not introduced until the late 1880's). Long before DNA became known to be the genetic material, it was already known that genes were lined up in chromosomes and that x-rays or ultraviolet light caused chromosome breaks resulting in mutations (Goldschmidt, 1951). The ends of broken chromosomes were sticky and could join up, the end of one chromosome break becoming joined to the end of a different chromosome break (Mcclintock, 1951) (Figures 21.8 and 21.9. ). The chromosome breaks were thereby repaired, but at the cost of changes in the lineup of the genes, and, if a break occurred within a gene, the function of the gene was destroyed. 439 K.W. Kohn Drugs Against Cancer CHAPTER21 Figure 21.7. Hermann Joseph Muller (1890-1967) with his x-ray machine. Muller entered Columbia College in New York at age 16, where he developed a long-time concern for the relationship between biology and society and became a proponent of eugenics. After completing his PhD degree at Columbia, Muller joined Thomas Hunt Morgan's "Fly Room" at Columbia where fruit fly genetics was under intense investigation, for which Morgan was awarded a Nobel Prize in Physiology and Medicine in 1933. Muller was awarded a Nobel Prize in Physiology and Medicine in 1946 for his discovery that x-rays cause mutations and quantifying the effect. Figure 21.8. Barbara Mcclintock (1902-1992) discovered mobile genetic elements, a concept so revolutionary at the time that it was long before geneticists accepted or understood it. Many years later after the importance of her of her discovery was grasped, she was awarded a Nobel Prize in Physiology or Medicine in 1983. She had received a PhD in Botany at Cornell University in 1927 and used maize as her subject of investigations. 440 K.W. Kohn Drugs Against Cancer CHAPTER21 (From http·/ /siarchives si.edu /collections /siris arc 30631 O) r ... ...I / Figure 21.9. An example of how Barbara Mcclintock observed chromosome breaks and their consequences (Mcclintock, 1951). The two homologous chromosomes are joined during mitosis at their centromeres (b); one of the chromosomes of the pair is broken (c). The first evidence that the damage caused by ultraviolet light (UV) or x-ray could undergo some kind of repair, came from experiments with bacteria. When E. coli bacteria were irradiated, the individual bacteria lost their ability to grow into colonies. However, if the irradiated bacteria were held for a period of time in a medium that lacked ingredients needed by the bacteria to grow and then put back into their growth medium, they recovered some of their ability to grow into colonies (Roberts and Aldous, 1949) (Harm, 1966). That remarkable observation suggested to the researchers that some kind of repair was happening. (The interpretation of those early experiments became complicated by a later discovery that photoreactivation, a process wherein the UV light itself stimulated an enzyme that removed some of damage. However, the broth in which the irradiated bacteria were held may have absorbed enough light so that photoreactivation was insignificant Our body cells, by the way, do not have the ability to photoreactivate UV-induced DNA damage. That is one of the many metabolic abilities that microbes have that we lack (or that humans might some-day recover by gene transfer, if that were ever to become permitted.) 441 K.W. Kohn Drugs Against Cancer CHAPTER21 Another important early discovery was of a strain of bacteria, called E. coli B/r, that was much less sensitive to DNA damage by ultraviolet light or x-ray than its parent strain, E. coli B (Witkin, 1946) (Figure 21.10). Although Evelyn Witkin, who discovered the resistant strain in 1946, knew little yet about DNA, her B/ r strain became an important tool in DNA damage and repair studies. Then in 1958, R. F. Hill isolated a hypersensitive mutant called Bsl, which joined the B/ r strain to become mainstays of DNA damage and repair studies (Hill, 1958). Ultraviolet light X-Ray .. .J > j, > I.. ~ ~ "' ~ I> ... - Figure 21.10. Evelyn Witkin's discovery in 1946 of a strain of E. coli bacteria, called B/ r, that was resistant to ultraviolet light and x-rays, compared to its parental strain, E. coli B (Witkin, 1946). Firs t evidence of DNA d amage repair in mamma lian cells. A major figure in the early studies of DNA damage and repair in mammalian cells was Mortimer Elkind at the National Cancer Institute and then at the Argonne National Laboratory (Figure 21.11). Elkind demanded rigorous quantitative discipline in his research and put the relationships between DNA damage, repair, and cell survival on a sound basis. He proved that unrepaired DNA damage caused cells to die (Elkind, 1979) -- which resulted in DNA damage and repair becoming a focus of expanded research in many laboratories on how anticancer drugs kill cancer cells. Elkind's precise quantitative methods were already evident in his first major publication in this field (Figures 21.12 and 21.13) (Elkind and Sutton, 1960). Figure 21.ll shows how Elkind and Sutton quantified the killing of cells by x- rays. In these kinds of experiments, it was important to choose the best materials to work with. The cells chosen had to grow well on the surface of a glass dish, and they had to have a consistent growth curve (number of cells versus incubation time). Elkind used a cell line that fit that requirement well: a clone of fibroblasts from Chinese hamster cells, called V79. After being 442 K.W. Kohn Drugs Against Cancer CHAPTER21 incubated for a suitable number of days, each viable cell on the dish formed a colony, and the number of colonies could be counted. The survival of cells after a given dose of x-rays was gaged as the fraction of cells that retained the ability to form colonies. Elkind and Sutton used that quantitative method to study the relationship between colony survival and x-ray dose. The experiment in Figure 21.13 is marked as number 92, showing the thorough persistence in carrying out this labor-intensive work. The precision of their data, together with the mathematical theory they derived, gave the first indication that mamma)i,in cells had the ability, although limited in extent, to repair the damage caused by x-rays. Moreover, the linear part of the survival curves indicated that a single unrepaired radiation-induced damage event could kill a cell (Elkind, 1984). The P .E. (plating efficiency of individual cells) in the experiment shown in Figure 21.13 was marked 84.1%, meaning that 84.1% of the unirradiated cells put on the plate grew into colonies. In later experiments, the plating efficiency was even higher. A high plating efficiency was important for quantitative interpretation of the data. Low plating efficiencies of individual cancer cells later came to haunt us when our Developmental Therapeutics Program tried to use colony forming ability of cells extracted directly from human cancers to look for drugs that would work against the common solid tumors: the apparent plating efficiencies in those attempts was miniscule! After several years of effort and large investment of resources, the project had to be dropped - all because the tiny plating efficiencies of individual cancer cells were overwhelmed by cell clumps. This instructive fiasco was related in Chapter 20. Figure 21.11. Mortimer (Mort) Elkind (1922-2000) was one of the greats of Radiation Biology and quantitative cell culture studies, which he carried out over many years in the National Cancer Institute. He pioneered the precise quantitation of cell killing by radiation and anti-cancer drugs. Originally trained in engineering and physics, he applied the strict discipline of those fields to cell biology (Withers, 2003). 443 K.W. Kohn Drugs Against Cancer CHAPTER21 Figure 21.12. Quantitative determination of the killing of mammalian cell by x-rays. Left, no radiation; Ri9ht, 542 rad of x-rays. The cells used were from Chinese hamsters and grew on the surface of a glass dish. A single cell could divide and form a colony that was then made visible by means of a stain. The colonies were counted in order to determine the fraction of the cells that survived to form colonies after a given dose of x-rays (Elkind and Sutton, 1960). DOSE, Krad EXP. 92 0 .2 .4 .6 .8 1.0 1.2 1.0 t-o;;;;:r--;--,,--,--,--,--,--y--,-...;;,.:-,,......::; CLONE V79-I P.E.• 84.1% z 0 .:: u <I ,I a:: IL (!) z > > a:: :::, Cl) .01 Figure 21.13. An x-ray survival curve with a "shoulder" at low doses (below 0.4 krad in the Figure). This was the first indication ofrepair of x-ray-induced damage in mammalian cells (Elkind and Sutton, 1960). Cells were grown on the glass surface of a dish, as described in Figure 21.12, and the survival of colony-forming ability was determined by counting the number of colonies formed before and after various doses of x-ray. The shape of the survival curve, including the linear portion in this semi-logarithmic plot fit a 444 K.W.Kohn Drugs Against Cancer CHAPTER21 mathematical theory derived by Elkind and Sutton and presented in their 1960 paper. DNA damage and repair investigated at the molecular level. The year 1960 was a harbinger of things to come in the field of DNA damage and repair. The first clues to the chemistry of DNA damage came from two landmark discoveries, both reported in 1960 (Brookes and Lawley, 1960) (Beukers and Berends, 1960). Peter Brookes and Philip Lawley at the Chester Beatty Cancer Institute in London treated various sources of DNA and RNA, as well as a tumor in mice, with sulfur mustard having a radioactive sulfur isotope. Analyzing the treated DNA and RNA for altered nucleotides they found one dominant product: the major part of the sulfur mustard molecule became bound to guanines at the N7 position (Figure 21.14) (Brookes and Lawley, 1960). To honor their achievements, the UK in 2003 established the Brookes-Lawley Laboratory for research on the genetic nature of cancer, as part of the Institute for Cancer Research (ICR) in London. Further studies described in Chapter 1, including those I too began in 1960, showed that this reaction was the first step in the production of a DNA inter-strand crosslink (Kohn et al., 1966). Also in 1960, Dutch researchers R. Beukers and W. Berends, working in the Biochemical and Biochemical Laboratory of the Technological Institute of Delft, The Netherlands, showed that ultraviolet light caused thymine (in frozen solution) to link in pairs to form dimers (Figure 21.15) (Beukers and Berends, 1960) (Beukers et al., 2008). The significance of this discovery soon became evident Figure 21.14. The chemical structure of the adduct at the N7 position of guanine that Peter Brookes and Philip Lawley identified in DNA and RNA treated with sulfur mustard (Brookes and Lawley, 1960) (the chemistry of sulfur and nitrogen mustards was discussed in Chapter 1). 445 K.W. Kohn Drugs Against Cancer CHAPTER21 Figure 21.15. Left, the General Electric germicidal low-pressure mercury vapor ultraviolet lamp used by Beukers and Berends in 1960 to create thymine dimers by irradiating a frozen aqueous solution of thymine in the small dish shown (Beukers et al., 2008). Right, the chemical structure of the thymine dimer, which they correctly inferred from infrared spectra that suggested the presence a cyclobutene (4-membered) ring (Beukers and Berends, 1960). In 1963, Richard Setlow and his colleagues at Oak Ridge National Laboratory in Tennessee made a landmark discovery that initiated the long and tortuous history of DNA damage repair studies. Their experiments showed that E. coli bacteria were able to repair the DNA damage produced by ultraviolet light (UV). When they irradiated bacteria with ultraviolet light (UV, 265 nm), thymine dimers were produced similar to those reported in 1960 by Beukers and Berends in UV-irradiated frozen solutions of thymine. Setlow inferred that UV-induced thymine dimers in the DNA inhibited the ability of the bacteria to synthesize DNA and to grow to form colonies (Setlow et al., 1963). Investigating further, they compared the UV-sensitive strain, E. coli Bsl , with the resistant strain, E. coli B/r, for the abilities of the bacteria to recover DNA synthesis after exposure to UV light (Seti ow et al., 1963). They found that E. coli B/r was able to remove thymine dimers from DNA, whereas E. coli Bsl was defective in this ability. After removal (repair) of the thymine dimers, the bacteria recovered their ability to synthesize DNA (Setlow and Carrier, 1964) (Figure 21.16). This was the first evidence for the existence of a DNA repair mechanism that came to be called nucleotide excision repair (Chapter 23) (Setlow et al., 1963). 446 K.W. Kohn Drugs Against Cancer CHAPTER21 1 0 16 1 I 0.161w.- - -- -- • • • • ~ 012 ~ 8/r 0.12 • ; .-: 0.08 0 "',\o 0 .0 8 ~ 004 I • \; 0.0 4 :iE • > :r 000 • - - -· 0.0 0 I- 30 60 90 120 30 60 90 MINUTES AFTER EXPOSURE TO UV LIGHT Figure 21.16. The first demonstration of DNA repair in living cells ((Setlow and Carrier, 1964), modified). E. coli bacteria were exposed to ultraviolet light (UV), and the thymine dimer content of the DNA was then measured after various lengths of time. The UV-resistant variant (B/r, left) removed dimers with time, whereas the UV-sensitive variant (Bsl, right) did not. (The upper curve in the B/ r experiment was when the incubation after UV was in medium lacking an energy source (glucose); in all other experiments, the incubation was in complete medium.) Next, in 1969, Setlow, Carrier and their colleagues found that normal human cells, like the bacteria, have the ability to remove thymine dimers from their DNA. Additionally, however, another landmark discovery was that cells from patients with the genetic disease xeroderma pigmentosum (XP) were unable to do that (Chapter 23). It seemed that normal cells could cut out the thymine- dimer damage from their DNA. The inability of XP patients' cells to carry out this DNA repair caused patients with this inherited disease to be extremely sensitive to daylight. Those early discoveries have been copiously summarized by Jim Cleaver (Cleaver, 2003) and by Errol Friedberg (Friedberg, 2011). Firs t evidence of repair ofDNA of inter-strand cross/inks. My involvement in the DNA repair story began when I started as a post- doctoral fellow in Paul Doty's laboratory at Harvard in 1959. 1was looking for ways to create chemical crosslinks between the paired stands of DNA. 1 had most in mind nitrogen mustard as a possible DNA inter-strand crosslinking agent that might relate to its therapeutic capabilities. How it happened that I came to suspect that nitrogen mustard produced DNA interstrand crosslinks was told in Chapter 1. 1 discussed my idea about DNA interstrand crosslinking, 447 K.W. Kohn Drugs Against Cancer CHAPTER21 which was a new idea at the time, with Professor Doty and we considered various ways in which DNA interstrand crosslinks could be produced. Remarkably, all of the possibilities we considered were soon discovered independently by various investigators. It was evidently a case that, when concepts and methods become available, a field is ripe for rapid discoveries. During our discussions in the laboratory, a possible way to produce DNA crosslinks was brought to my attention by K. Leszek Wierzchowski, a visiting scientist from Poland who had coauthored an extensive review article with David Shugar about the effects of ultraviolet light (UV), published in an obscure Polish journal. Lech told me about the just then published reports by Beuker and Berends that UV caused thymine to link in pairs to form dimers. We thought that UV might link together thymines in opposite DNA strands to form interstrand crosslinks. This turned out to be true, although most of the thymine dimers were later found to form between neighboring thymines in the same strand. However, the path to the discovery of DNA repair progressed when Setlow and his colleagues demonstrated UV-generated thymine dimers in bacteria, described above and in the Chapter that follows. My focus, however, was on the nitrogen mustard crosslinking idea, which I was enabled to pursue based on the helix-coil transition phenomena and theory that had been developed by Paul Doty, Julius Marmur, and others in his lab. The main instrument I used in the nitrogen mustard experiments was the analytical centrifuge, an extraordinary technology which I soon managed to master with initial guidance by Carl Schildkraut, who was at that time an advanced graduate student in the lab. All of this is described in greater detail in Chapter 1, which tells how it happened that nitrogen mustard, a close relative of mustard gas, became the first chemical agent that shrank solid tumors in humans, and how my notion of crosslink production emerged. After completing the nitrogen mustard crosslinking studies in physical chemical systems (Kohn et al., 1966), I returned to the National Cancer Institute in Bethesda with the aim of extending those studies to living organisms. I was convinced that nitrogen mustard worked in cancer chemotherapy by producing DNA crosslinks in cancer cells. The methods to prove that required development of new technology, which only became feasible years later after we developed DNA filter elution methods capable to measuring several types of DNA damage in mammalian cells (Chapter 9). 448 K.W. Kohn Drugs Against Cancer CHAPTER21 In the 1960's however we were able to test our ideas in bacteria, namely the DNA repair proficient and deficient strains, E. coli Bs1 and E. coli B/r, respectively, and Setlow's findings using those strains, as described above. As expected, we found that the radiation-sensitive mutant, E. coli Bsl, was unable to repair DNA inter-strand crosslinks produced by nitrogen mustard, whereas the resistant E. coli B/r removed the crosslinks with high efficiency (Figure 21.1 7) (Kohn et al., 1965). This was the first evidence for repair of interstrand crosslink repair in living organisms. • • ~-..c.oJi B ...... ,o 0 E.coli B51 ...... No drug /\ .r I\ ,o r.. \,.___,.,. 0 b. IQ'~ .. MN2: b 1()._ M Nll2 ,o Nitrogen murtard /\ t reatment •• ' ..__/ ' 0 C. 10,. M H!l2: ,.o C. ♦•,i..to'•S1" .... - l@ltll'¥'/013f• 90 minutes A "· later ' •• 0 '-----1 0 00 20 JO Figure 21.17. The first demonstration of DNA crosslink repair (Kohn et al., 1965). In this experiment, we found that wild-type E. coli B removed nitrogen mustard crosslinks from DNA, but that the radiation-sensitive variant, E. coli Bsl , was unable to do so. The peak representing the unrepaired crosslinks in E. coli B5 1 is indicated by the arrow in the panel on the lower right. The DNA from the bacteria was radioactively labeled and is represented by the solid lines; the left and right peaks are non-crosslinked and crosslinked DNA, respectively. The DNA from the bacteria was treated with sodium hydroxide to separate the strands; after neutralizing the solution with citric acid, only the crosslinked DNA recovered its double-stranded form and banded in the peak on the right The non-crosslinked DNA remained single-stranded and banded on the left. (The dashed curves show where the single-stranded DNA (left peak) and double-stranded DNA (right peak) would band.) The samples were ultracentrifuged in a concentrated CsCI solution for 60 hours to reveal the banding patterns shown. 449 K.W. Kohn Drugs Against Cancer CHAPTER21 DNA in the cell nucleus can be damaged in many ways: carcinogens in the environment, workplace, and foods; ultraviolet light from the sun; radiation from diagnostic x-rays, power plants, and natural background; chemotherapy drugs; normal metabolism and genetic defects. These produce a wide variety of chemical damage to DNA, for which an amazing variety of repair mechanisms have evolved. It is remarkable how evolution has come up machinery to fix or cope with almost any kind of DNA damage. Even when some damage remains, there are fail-safe mechanisms to cope with it. It seems that almost nothing will permanently stop a cell, except when the insult is overwhelming. Those issues are of great importance in cancer treatment The ability of cancer patients to survive chemotherapy, depends greatly and the ability of normal tissues to repair the DNA damage caused by nearly all of those drugs. Moreover, the effectiveness of many of those drugs depends in part on the DNA damage they produce and the difficulty that cancer cells have in trying to repair them. References Avery, O.T., Macleod, C.M., and McCarty, M. (1944). Studies on the Chemical Nature of the Substance Inducing Transformation of Pneumococcal Types : Induction of Transformation by a Desoxyribonucleic Acid Fraction Isolated from Pneumococcus Type Iii. J Exp Med 79, 137-158. Beukers, R, and Berends, W. (1960). Isolation and identification of the irradiation product of thymine. Biochimica et biophysica acta 41, 550- 551. Beukers, R, Eker, A.P., and Lohman, P.H. (2008). SO years thymine dimer. DNA repair 7, 530-543. Brookes, P., and Lawley, P.D. (1960). The reaction of mustard gas with nucleic acids in vitro and in vivo. The Biochemical journal 77, 4 78-484. Cleaver, J.E. (2003). Excision repair--its bacterial beginnings. DNA repair 2, 1273-1274. Elkind, M .M. (1979). DNA repair and cell repair: are they related? International journal ofradiation oncology, biology, physics 5, 1089- 1094. Elkind, M .M. (1984). Repair processes in radiation biology. Radiat Res 100, 425-449. Elkind, M .M., and Sutton, H. (1960). Radiation response of mammalian cells grown in culture. 1. Repair of X-ray damage in surviving Chinese hamster cells. Radiat Res 13, 556-593. Friedberg. E.C. (1997). Correcting the Blueprint of Life: An historical account of DNA repair mechanisms. (Plainview, New York: Cold Spring Harbor Laboratory Press). 450 K.W.Kohn Drugs Against Cancer CHAPTER21 Friedberg, E.C. (2011). Nucleotide excision repair of DNA: The very early history. DNA repair 10, 668-672. Goldschmidt, R.B. (1951). Chromosomes and genes. Cold Spring Harbor symposia on quantitative biology 16, 1-11. Gudmundsdottir, K., and Ashworth, A. (2006). The roles of BRCAl and BRCA2 and associated proteins in the maintenance of genomic stability. Oncogene 25, 5864-5874. Harm, W. (1966). The role of host-cell repair in liquid-holding recovery of u.v.- irradiated Escherichia coli. Photochem Photobiol 5, 747-760. Hashimoto, T., and Kunieda, T. (2017). DNA Protection Protein, a Novel Mechanism of Radiation Tolerance: Lessons from Tardigrades. Life (Basel) 7. Hill, R.F. (1958). A radiation-sensitive mutant of Escherichia coli. Biochimica et biophysica acta 30, 636-637. Judson, H.F. (1979). The Eighth Day of Creation: Makers of the Revolution in Biology (New York: Simon and Schuster). Kohn, K.W., and Bohr, V.A. (2001). Genomic Instability and DNA Repair. In Cancer Handbook (Nature Publishing Group), pp. 85-104. Kohn, K.W., and Green, D.M. (1966). Transforming activity of nitrogen mustard-crosslinked DNA. Journal of molecular biology 19, 289-302. Kohn, K.W., Spears, C.L., and Doty, P. (1966). Inter-strand crosslinking of DNA by nitrogen mustard. Journal of molecular biology 19, 266-288. Kohn, K.W., Steigbigel, N.H., and Spears, C.L. (1965). Cross-linking and repair of DNA in sensitive and resistant strains of E. coli treated with nitrogen mustard. Proceedings of the National Academy of Sciences of the United States of America 53, 1154-1161. Lindahl, T. (1982). DNA repair enzymes. Annual review of biochemistry 51, 61-87. Mcclintock, B. (1951). Chromosome organization and genie expression. Cold Spring Harbor symposia on quantitative biology 16, 13-47. Roberts, R.B., and Aldous, E. (1949). Recovery from Ultraviolet Irradiation in Escherichia Coli. Journal of bacteriology 57, 363-375. Sancar, A., and Sancar, G.B. (1988). DNA repair enzymes. Annual review of biochemistry 57, 29-67. Setlow, RB., and Carrier, W.L (1964). The Disappearance of Thymine Dimers from DNA: An Error-Correcting Mechanism. Proceedings of the National Academy of Sciences of the United States of America 51, 226-231. Setlow, RB., Swenson, P.A., and Carrier, W.L. (1963). Thymine Dimers and Inhibition of DNA Synthesis by Ultraviolet Irradiation of Cells. Science 142, 1464-1466. Szybalska, E.H., and Szybalski, W. (1962). Genetics of human cess line. IV. DNA-mediated heritable transformation of a biochemical trait. Proceedings of the National Academy of Sciences of the United States of America 48, 2026-2034. Withers, H.R. (2003). In Memoriam. Radiat Res 159, 137. 451 K.W. Kohn Drugs Against Cancer CHAPTER21 Witkin, E.M. (1946). Inherited Differences in Sensitivity to Radiation in Escherichia Coli. Proceedings of the National Academy of Sciences of the United States of America 32, 59-68. 452 K. W. Kohn Drugs Against cancer CHAPTER 22 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@oib gov CHAPTER22 Genetic diseases reveal DNA nucleotide excision repair. Rare inherited diseases reveal DNA repair genes and how cells cope. It is remarkable how studies of certain rare inherited diseases gave clues that helped unravel the complexities of DNA damage repair, the factors that predispose to cancer, and the molecular targets for drug treatments tailored for the molecular defects in individual cancers. Table 22.1 lists genetic diseases that led to the discovery of genes for the repair of certain types of DNA damage or that help cells to survive the damage. The current chapter relates the stories of the first two on the list (xeroderma pigmentosum and Cockayne syndrome). Table 22.1. Genetic disease Defective nrocess Defective 2enes Chanter Xeroderma oi1m1entosum (XP1 Nucleotide excision reoair XPA-XPG; XPV 22, 23 Cockavne svndrome Transcrintion-counled renair CSA,CSB 22 Fanconi anemia Inter-strand crosslink renair FANC 2enes 31 Ataxia telangjectasia Cell rvcle checknoint activation ATM, ATR 29 Lvnch svndrome DNA mismatch renair MLH, MSH, PMS 25 Prone to breast cancer Homolo2ous recombination BRCAl, BRCA2 26 Li-Fraumeni svndrome "Guardian of the 2enome" TP53 32 Bloom's syndrome Premature a2in2 in childhood BLM Werner's svndrome Premature a2in2 in adult WRN The Xeroderma pigme ntosum (XP) story: a curse and a clue . 453 K. W. Kohn Drugs Against cancer CHAPTER 22 The esoteric name, xeroderma pigmentosum (or simply XP), became common parlance among researchers and clinicians dealing with skin diseases and cancer. Likewise, the name will appear so many times in the chapter that the reader may find it enter his/her familiar vocabulary. Literally, xeroderma pigmentosum means "pigmented dry skin," but that benign-sounding name does not match the severity of the disease. Children affected with this fortunately very rare inherited disease came to be called children of the dark and even vampire children, because for them daylight could be deadly. If they failed to adhere to a discipline of the dark, many of them had little chance of surviving beyond the age of 20. However, some who practiced sun protection led nearly normal lives. Their quandary of how to meld their extreme light sensitivity with their social lives became the subject of several films ("Children of the Dark" - TV movie 1994). It may not have consoled these affected people very much to know that their suffering would lead to new knowledge that would help others, including many suffering with cancer. Children with xeroderma pigmentosum (XP) tended to get freckle-like pigmentation in sun- exposed skin, severe burns after minimal sun exposure, and damage to sun-exposed parts of the eyes with loss of vision and ocular cancer. At least 8 molecular defects were discovered. About 25% of the patients had progressive neurological degeneration. Their greatest risk of sun exposure, however, was potentially deadly skin cancers. The older medical literature often showed patients with advanced stages of the disease. Since it affects sun-exposed skin, its effects are exposed to view. The worst of it, however, was hidden: some of the skin nodules become malignant cancers that spread to organs within the body. Xeroderma pigmentosum was one of the most cancer-prone conditions with a 10,000-fold increase in skin cancer (Bradford et al., 2011). It became one of the most extensively studied diseases in all of cancer research. The advanced cases commonly shown in the medical literature look so bad, that I have selected an early and relatively mild case to show in Figure 22.1 (Anderson, 1889). The disease was first described by Ferdjnand Rjtter yon Hehra and Morris Kaposi in 187 4, who coined the name "xeroderma pigmentosum" (Ueber Xeroderma pi9mentosum. Medizinische Jahrbiicher, Wien, 1882: 619- 633). The case in Figures 22.1 and 22.2 dates back to 1889, when its inherited nature was known, but its cause was still clouded in mystery (Anderson, 1889). Up to that time, only 44 cases of the disease were known. This case was reported by T. McCall Anderson of the University of Glasgow in the British Medical Journal. It was of a 9-year old boy, who, despite his skin lesions and the loss of his left eye -- which was removed because it had a tumor growing on it -- still looked to be otherwise in generally good health. You can see from Figure 22.1 that his skin lesions were mainly on the face, neck, upper shoulders, and lower arms: the parts of his body that would have been most exposed to the sun. Microscopic examination of his tumors was noted to be typical of epithelioma (epithelial cancer) (Figure 22.2). His parents were both in good health, but his only sister was similarly affected and had died at the age of 9. This was consistent with the known recessive inheritance of the disease. The exact genetics however were confusing; we now know the reason for the confusion: it was because the disease, 454 K. W. Kohn Drugs Against cancer CHAPTER 22 with clinical variation, can be caused by mutation of any one of several genes, located on different chromosomes (Table 22.2). Figure 22.1. Woodcut of a 9-year old boy with xeroderma pigmentosum (Anderson, 1889). The child had multiple cancers on his face, and his left eye was removed because there was a cancer in it. Unless protected against sunlight his condition could have become much worse. At the time this picture was made (circa 1889), little was known about the disease. Pigmentation (darkening) is also shown in the sun-exposed areas around his neck and the lower parts of his arms. Figure 22.2. A micrograph (x250) of a skin tumor from the patient in Figure 22.1 (Anderson, 1889). The central portion (labeled A in the micrograph) was typical for a squamous carcinoma where the cancer retained a bizarre memory of the arrangement of the multi layered cells of the outer skin. The cancer cells showed the usual large irregular nuclei, compared with the fibroblasts in the stroma (labeled B) within which the tumors are growing. A blood vessel is seen on the left. 455 K. W. Kohn Drugs Against cancer CHAPTER 22 Another case ofxeroderma pigmentosum, also published in 1889 in the British Medical Journal, was of a 10-year-old boy. The report noted the role of sun exposure by the unusual involvement of this child's feet and lower legs (Hunter, 1889): It may be remarked here that these children went to school during the summer months without shoes or stockings and the trousers often rolled up. The latter is done so as not to interfere with running or jumping, and in wet weather to keep them (the trousers) clean. This is a common habit in the country districts, and thus, in these cases, the feet and legs which are affected were also exposed (Hunter, 1889). These two reports (Anderson, 1889; Hunter, 1889) gave a detailed description of the skin lesions and the course of the disease in all-together 4 patients. These early reports, while little was as yet known about the disease, show what physicians of ~130 years ago saw and experienced. Next, we move ahead 78 years. Several different forms of the disease were by then known with different clinical manifestation, although all of them entailed skin lesion caused by sunlight The cells from patients having mutations in a particular gene were referred to as a "complementation group" because the normal version of the respective gene could cure the defect in the corresponding mutant cells. The defect measured was "unscheduled DNA synthesis" which will be explained in a moment A complementation group, in common parlance, became nearly the same as referring to a particular mutant gene (Table 22.2). It was 1967, I think, in a small conference in the NIH Clinical Center; we had a guest speaker. It was James E. Cleaver from the University of California. He claimed to have detected a defect in unscheduled DNA synthesis in cells from patients with xeroderma pigmentosum. Some of us were skeptical, because we had expected that phenomenon, but no one had yet been able to demonstrate it Moreover, his data seemed to be barely above the noise level. Within a year or so, however, he published more convincing data (Cleaver, 1968) and the skeptics were soon on the bandwagon as their investigations corroborated and extended Cleaver's claims (Kraemer and DiGiovanna, 2015). So now, what is "unscheduled DNA synthesis" and how is it related to DNA repair? Well, our concept was that cells whose DNA was damaged by ultraviolet light (UV) would be undergoing a small amount of DNA synthesis for repair after the damage was cut out. This would be occurring even in cells that were resting, i.e., not cycling in cell division and hence not undergoing the normal DNA synthesis phase. Normally, those cells would not be synthesizing DNA But to repair DNA damage, even those cells would be undergoing a small amount of unscheduled DNA synthesis: it was "unscheduled" because it would be occurring even if the cells are not in the cell division cycle. We had reason to believe that cells from xeroderma pigmentosum (XP) cells were defective in their ability to repair DNA damage caused by ultraviolet light. Therefore, they might not exhibit unscheduled DNA synthesis, as Cleaver had shown to be the case, namely the absence of unscheduled DNA synthesis in UV-irradiated XP cells. 456 K. W. Kohn Drugs Against cancer CHAPTER 22 To put it in another way: exposing normal skin to ultraviolet light induced DNA damage, which was promptly repaired by a process that involved a small amount of DNA synthesis to replace the damaged DNA regions that were cut out by the repair mechanism. This DNA synthesis would be occurring even in cells that were not in division cycle and hence would not ordinarily undergo any DNA synthesis. Since that DNA repair synthesis would be occurring in quiescent (i.e. non-dividing, non-cycling) cells, it was termed "unscheduled." Xeroderma pigmentosum patients were presumed to be defective in such DNA "excision" repair, and their cells therefore would be unable to carry out unscheduled DNA synthesis. The inability of skin cells from xeroderma pigmentosum (XP) patients to exhibit unscheduled DNA synthesis in response to ultraviolet light (UV) was later confirmed directly in patients (Figure 22.4) (Epstein et al., 1970). A small region of the patient's skin was irradiated with ultraviolet light; then, tritium-labeled thymidine was injected under the irradiated portion of skin, which was then biopsied and radiographed to show the cells that had incorporated the radioactive thymidine into DNA (Figure 22.4). I do not know whether the patients had given informed consent for this procedure of injecting radioactive material into the skin - it would probably not be permitted now with or without informed consent As thymine dimers were being cut out from DNA of UV-irradiated cells, transient DNA strand breaks occurred. Detection and study of that step became possible using the DNA filter elution technique described in Chapter 9. Before that, the available methods were not sensitive enough to detect those breaks. Application of the filter technique allowed Al Fornace, who had recently joined my lab as a post-doctoral fellow, to fill in that gap in the evidence (Fornace et al., 1976). We showed, as expected, that DNA strand breaks appeared after UV exposure of normal cells, but not in XP cells (Figure 22.5). Using the filter methods, Fornace and I also showed that UV produces DNA-protein crosslinks and that XP cells were unable to repair them (Fornace and Kohn, 1976) and later confirmed the inability ofXP cells to repair DNA-protein crosslinks by using transplatin to produce them (Chapters 3 and 9) (Fornace and Seres, 1982). Moreover, the XP cells had increased vulnerability to being killed by transplatin. Those results showed that the nucleotide excision repair (NER) pathway was able to recognize and repair, not only thymine dimers, but also DNA-protein crosslinks. 457 K. W. Kohn Drugs Against cancer CHAPTER 22 Figure 22.3. James Edward Cleaver (1938- ), discoverer of the molecular process that is defective in the genetic disease, xerodenna pigmentosum. This was the seminal discovery of a human DNA repair deficient disease. 458 K. W. Kohn Drugs Against cancer CHAPTER 22 Figure 22.4. Skin from a xeroderma pigmentosum patient was unable to repair DNA damage caused by ultraviolet light The repair entailed a small amount of DNA synthesis to replace the damaged DNA regions. These studies demonstrated defective DNA repair synthesis ("unscheduled DNA synthesis") in skin of xeroderma pigmentosum patients. Left, normal skin: no radioactivity grains. Middle, irradiated normal skin: many cells show radioactivity grains, indicative of unscheduled DNA synthesis. Right, irradiated xeroderma skin: no unscheduled DNA synthesis (the heavily labelled cell is undergoing replicative DNA synthesis) (Epstein et al., 1970). . :r.96) NORMAL CEUS XP:CEUS BAq ' - S 10 15 5 10 15 HOURS OF ELUTION Figure 22.5. The cutting out of UV-induced thymine dimers in the course of nucleotide excision repair (NER) was expected to produce transient DNA strand breaks, but it had been difficult to detect these breaks. The DNA filter elution methods described in Chapter 9, however, were sensitive enough to show up those transient breaks - shown by a high DNA elution rate (solid circles in the left panel). Cells from a xeroderma pigmentosum (XP) patient showed almost no DNA elution (sold circles in the right panel), as expected when thymine dimer excision is defective (Fornace et al., 1976). (The open circles were for internal standards used to increase the precision of the measurements.) 459 K. W. Kohn Drugs Against cancer CHAPTER 22 An unusual variant ofXP is discovered. As often happens in science, a new discovery, when further investigated, becomes challenged by findings that don't fit the original concept And so it was that the fourth NIH XP patient had the high light sensitivity of the skin, but showed no defect in unscheduled DNA synthesis (Robbins et al., 1974). About the same time, Jim Cleaver reported 3 more XP patients whose cells had normal unscheduled DNA synthesis after being exposed to ultraviolet light (Cleaver, 1972). Those patients defined a variant ofXP that had normal nucleotide excision repair (NER); this variant of XP became known as XPV. The defect in XPV turned out to be a mutation in a special DNA polymerase, known as Pol-eta or PolH, which is needed to synthesize DNA across a defect in the template strand, such as may remain after removal of pyrimidine dimers (Di Giovanna and Kraemer, 2012). Cleaver later remarked that, if his first patient had been of the XPV type, he might have erroneously concluded that XP had normal DNA repair (Kraemer and DiGiovanna, 2015). Subtypes ofxeroderma pig mentosum: complementation groups Even among XP patients whose cells were defective in unscheduled DNA synthesis, there was considerable variability in how severe the disease was, how sensitive the cells were to be killed by ultraviolet light, and the clinical picture in general. Particularly puzzling was that the disease of some patients had neurological symptoms, sometimes quite severe. An extreme case of the latter was a syndrome that deserved its own name: De Sanctis- Cacchione syndrome. It was one of the rarest, most severe forms of xeroderma pigmentosum (XP). In addition to being highly sensitive to sunlight, these XP patients were of short stature and developed progressive neurologic degeneration. The syndrome was first recognized by de Sanctis and Cacchione in 1932, who described three brothers with XP who had microcephaly, mental deficiency, dwarfism, gonadal hypoplasia, progressive neurologic deterioration, deafness, and ataxia beginning at the age of 2 years. This new syndrome was described in 1932 in an obscure Italian journal: de Sanctis C., Cacchione A. L'idiozia xerodermica. Riv Sper Freniatr Med Leg Alien Ment. 1932;56:269-292. The cause of these different types ofXP could only be investigated after new methods and concepts were developed, which took several decades. The story began to unfold in 1971 when Dirk Bootsma (Figure 22.6), a Professor of Genetics at Erasmus University in Rotterdam, thought that the large differences in the clinical picture ofXP patients might be due to different genes, each of which, when mutated, would cause a particular clinical form of the disease. He reasoned that, if two of those defective genes, each on a different chromosome, were put into the same cell, they might complement each other and restore normal unscheduled DNA synthesis. He wanted to test that idea using cells from two very different forms of XP: the classic variety and the De Sanctis-Cacchione type. 460 K. W. Kohn Drugs Against cancer CHAPTER 22 But how could one put the presumed different defective genes into the same cell? Bootsma and his colleagues developed a method that used the ability of certain viruses (inactivated Sendai virus) to fuse cells together to produce cells that sometimes had two nuclei. However, although some of the binucleate cells had a nucleus from each of the XP types, oftentimes they had nuclei from the same type. In order to distinguish whose different cases, they used a clever trick: they fused cells from a male patient who had one XP type with cells from a female patient who had the other XP type of the disease. They then exposed the cells to UV to produce DNA damage. By carefully examining the chromatin in each nucleus of a binucleate cell, they could tell whether the nucleus came from a male or female cell. They indeed found that those and only those binucleate cells that had both a male and a female nucleus - one nucleus from each XP type -- complemented each other to restore unscheduled DNA synthesis (Figure 22.7) (De Weerd-Kastelein et al., 1972). Other investigators, particularly Ken Kraemer at NIH (Figure 22.8), then jumped in and, using a modified method, found that there were in fact several different complementation groups among the XP patients, their cells, and their mutated genes (Table 22.2). Decades later, the genes that were mutated in the various complementation groups were cloned and their functions determined. Quite remarkably, the proteins produced by all of these XP genes were found to work together to repair DNA damage by a very important mechanism: DNA nucleotide excision repair (NER). How this mechanism works is the subject of the next chapter. How complementation groups were determined is explained in Figure 22.9. Figure 22.6. Dirk Bootsma (1936- ), a Professor of Genetics at Erasmus University Rotterdam, Netherlands, developed a cell-fusion method by which he discovered xeroderma pigmentosum complementation groups. (From Ned Tijdschr Geneeskd 2002 12 oktober;146(41).) 461 K. W. Kohn Drugs Against cancer CHAPTER 22 Figure 22.7. How Bootsma and his colleagues showed that a gene responsible for what was then called the classic type of XP and a gene responsible for the De Sanctis- Cacchione type of XP complemented each other to restore the unscheduled DNA synthesis process of DNA repair (De Weerd-Kastelein et al., 1972). They fused cells from a male child who had one type of XP with cells from a female child who had the other type ofXP. Each panel shows a cell with two nuclei. The cells had been exposed to UV to produce DNA damage and then incubated with radioactive thymidine. The cell on the right shows radioactive spots scattered in both nuclei where unscheduled DNA synthesis was occurring. The cell on the left shows no unscheduled DNA synthesis. The nuclei in the cell on the left both came from male donors. The same was true if both nuclei came from females. Only cells that had one male and on female nucleus - therefore, a mixture of the XP types - showed unscheduled DNA synthesis: the cell had a good copy of both genes, one from each nucleus. Figure 22.8. Kenneth Kraemer (left) and Vilhelm Bohr (right) in 2005, commemorating four decades of research on DNA repair at NIH. Kraemer received an MD degree at Tufts Medical School and became board certified in Dermatology and Internal Medicine. He came to NIH in 1971 as a clinical associate in the Dermatology Branch and has been leading ground-breaking research on xeroderma pigmentosum and related diseases. In 1980, he established and since co-chaired with Vilhelm Bohr an NIH Special Interest Group on DNA Repair in which he pioneered internet 462 K. W. Kohn Drugs Against cancer CHAPTER 22 conferencing to bring together researchers from several institutions in different cities. Bohr, a descendent of Niels Bohr, received an MD degree in 1978, followed by PhD and D.Sc. degrees in 1987 from the University of Copenhagen in Denmark. Together with Philip Hanawalt at Stanford University, he pioneered investigations of transcription-coupled DNA repair, which he continued in my Laboratory. In 1992, he became Chief of the Laboratory of Molecular Genetics in the National Institute of Aging where he has been leading studies of DNA repair and cancer. $USPE.fl$0H Of f lMOk.ASTS ....... Vl l!\lllACt(l,RPTOr, ClU. AOGlUTiflllATICIN , .. B CI U. fUSIOH @0 ..,.......@_ '0 ♦ 0, Hl»IOltAR'f'ONS Figure 22.9. Complementation was determined by fusing together a mixture of cells from two different xeroderma pigmentosum (XP) patients. Complementation occurred when the patients had defects in different XP genes: combining their cells would give at least some fused cells that did not show the defect - because some of the fused cells would have a good copy of both genes. The patients therefore were of two different complementation groups. On the other hand, if the cells from the two patients had defects in the same in the same gene, then all of their fused cells continued to show the defect: the two patients were of the same complementation group. The experiments utilized killed Sendai virus, which caused the cells to stick to each other. When the cells were then warmed up, the cells fused together and many of them had two nuclei, showing that they were a combination of two different cells. If some of the binucleate cells restored the defect present in both of the original cell strains, then complementation was said have occurred and that the cells came from different complementation groups. If none of the binucleate cells came from cells that all had defects in the same gene, then they were of the same complementation group. (From (Robbins eta!., 1974).) 463 K. W. Kohn Drugs Against cancer CHAPTER 22 Table 22.2. Xeroderma pi2mentosum 2enes (complementation 2roups) Complementation Genes/ proteins/ definitions Chromosome Group XP-A XPl 9q22.33 XPB ERCC3 (excision repair 3, helicase 2q14.3 subunit) XPC RAD4 3015.1 XPD ERCC2 (excision repair 2, helicase 19q13.32 subunit) XPE DDB2 (damae:e-specific DNA bindine: 2) llnll.2 XPF FANCQ, RADl, ERCC4 (excision repair 16p13.12 4, endonuclease) XPG ERCC5 (excision repair 5, 13q33.1 endonuclease) XPV XP variant, DNA polymerase eta 6q21.1 (Information from the human gene nomenclature committee (HGNC) website.) The Cockayne syndrome story But the XP story had yet another twist in a different, but closely related, genetic disease. In 1936, Edward Alfred Cockayne, a pediatrician at th~ Great Ormond Street Hospjta) for Sjck Children (which still stands (Figure 22.10)) in Bloomsbury, London, described two children with a previously unknown syndrome that was to bear his name (Figure 22.11) (Cockayne, 1936). He characterized the syndrome as "dwarfism with retinal atrophy and deafness." Although Cockayne's syndrome became closely linked with xeroderma pigmentosum, and these children had sun sensitivity, their skin was clear (Figure 22.11). That 85-year-old paper was not easy to find and may become increasingly difficult to find as time goes by; therefore, I am reproducing here Cockayne's original description of the syndrome, as well as images he included in his paper (Figures 22.12-14): The two children with this dystrophy, a girl aged seven years and eleven months and a boy aged sixyears and three months were admitted to the Hospital for Sick Children, Great Ormond Street in June, 1935. The parents, who are natives of north Hampshire, are of English race, normal and not blood-relations, and they have been unable to trace the occurrence of the condition in their ascendants or co/laterals .. The dwarfs are so much alike In facial appearance, build and disposition, that the same general description will suffice. Both have small heads, that of the girl being the smaller, bu~ although the vault of the skull Is flattened and the circumference small, the general shape Is normal, and neither child has the receding forehead characteristic of mlcrocephaly. Their faces are small with sunken eyes and prominent superior maxillae. They are slightly built with sho~ slender trunks and unduly long legs, and their feet and hands are too large in proportion. The third and fourth fingers oftheir hands are deviated a little towards the mes/a/ line. Both are active and their movements are quick and bird-like. They are friendly and playful, Invariably good tempered, and laugh with obvious enjoyment at the slightest provocation. Although they are Imitative, they have a certain amount of Initiative and in playing with toys are no more destructive than most children of their age and class. They frequently make noises 464 K. W. Kohn Drugs Against cancer CHAPTER 22 which at first sound like speech, but actual words can seldom be recognized, although the girl has been heard to say' mother' and' do it again' and the boy has said 'doctor ' several times. They do not answer to their names or obey spoken words, nor do they take any notice of a sound made behind their heads, but they are quick to obey signs. Mr.James Crooks, F.R.CS., who saw them, says that although not totally deaf their hearing is greatly impaired. It is difficult to tell how much of their backwardness is due to deafness and how much to mental deficiency. Their behaviour is not the usual behaviour ofdeafchildren. They appear to be a little below the average in intelligence and are far more excitable and laugh much more readily than children ofnormal mentality whether deaf or not. Children with Cockayne syndrome rarely survived to adulthood. Figure 22.10. The Great Ormond Street Hospital for Sick Children in Bloomsbury, London, where Edward Alfred Cockayne saw two children in 1935 who had a new syndrome, which came to bear his name (Cockayne, 1936). (Creative Commons, Wikipedia) 465 K. W. Kohn Drugs Against cancer CHAPTER 22 Figure 22.11. A 7-year old girl with the syndrome described by Cockayne in 1936 (right) standing next to a normal girl of the same age (left) (Cockayne, 1936). The affected child, although much shorter than normal, had relatively long legs and large hands. Her head, however, was relatively small. Her skin was clear and had no sign of the sun-induced damage that is characteristic in xeroderma pigmentosum. Figure 22.12. Cockayne said the affected child had an abnormally small head with sunken eyes and prominent front upper jaw (Cockayne, 1936). 466 K. W. Kohn Drugs Against cancer CHAPTER 22 Figure 22.13. Cockayne described her skull as being small of small circumference with thickened bones and prominent upper jaw (Cockayne, 1936). Figure 22.14. The retinal atrophy Cockayne noted in the affected children (Cockayne, 1936). He noted markedly narrowed retinal arteries and atrophic changes, particularly in the central region of the retina. 467 K. W. Kohn Drugs Against cancer CHAPTER 22 "' Normal and CS cells ~ . I I OJ I I a •• '--..,.,---'-,--',,--,.._..,.o--':•--'~:--,'7',~ 0·01 '---'-000,----:-200 '-:--,.,::'co : ---:- _,'-:--ooo,.._, U\! OOSE ( J/M t ) X RAY (xPOSUM: IA) Figure 22.15. Fibroblast cells from the skin of Cockayne syndrome (CS) children were unusually sensitive to being killed by ultraviolet light (UV) (left) but had normal sensitivity to x-rays (right) (Schmickel et al., 1977). Xeroderma pigmentosum cells had the same pattern: high sensitivity to UV but not to x-rays. It was four decades after Cockayne's description before the first clue to the cause of the syndrome arrived. It came from the laboratory of Schmickel and coworkers at the University of Michigan. They showed that fibroblast cells derived from the skin of affected children were unusually sensitive to ultraviolet light (UV), whereas their sensitivity to x- rays was normal (Schmickel et al., 1977) (Figure 22.15). This curious difference of being highly sensitive to DNA damage caused by UV but not to DNA damage caused by x-ray was exactly the same as in xeroderma pigmentosum (XP). However, despite the cells of most Cockayne's syndrome (CS) children being highly sensitive to UV, the children were not highly sensitive to sunlight and their cells removed thymine dimers from their DNA at a near normal rate, contrary to the inability of xeroderma pigmentosum cells to remove those dimers (Schmickel et al., 1977). lt took almost another decade to resolve the puzzle of why CS cells resembled XP cells in being unusually sensitive to UV, even though they (the CS cells) removed thymine dimers from their DNA normally, as opposed to the inability of XP cells to do so. The confusing findings about the relationship between Cockayne syndrome and xeroderma pigmentosum began at last to be clarified by Vilhelm Bohr (Figure 22.8), working at Stanford with Philip Hanawalt and later at NCI in my Laboratory. He pinned down a special nucleotide excision repair (NER) mechanism designed specifically and exclusively to repair DNA damage in regions of the genome that were being actively transcribed at the time that the damage was present. It was a distinct type of repair and was named "transcription- coupled nucleotide excision repair" (TCNER) (Bohr et al., 1985). Researchers in The Netherlands and the UK then showed that Cockayne's syndrome had a defect in TCNER (Venema et al., 1990). But why did a defect in TCNER (due to a defective CS gene) make cells sensitive to being killed by UV (Figure 22.15)? The reason was that transcription (RNA synthesis), as it 468 K. W. Kohn Drugs Against cancer CHAPTER 22 progressed along the DNA, occasionally collided with UV-induced thymine dimer (or other pyrimidine dimer). The collision produced a peculiar DNA damage configuration that was apt to lead to death of the cell. This disaster was avoided by a special NER machinery (TCNER) that was attached to transcription machinery. When transcription encountered a UV-induced dimer, it was promptly excised by TCNER, allowing the transcription machinery to continue merely on its way. The TCNER machinery in Cockayne syndrome (CS) cells however was defective, which put the cells at risk whenever transcription collided with a UV-induced dimer. Thus, while most of the dimers scattered in the genome were efficiently removed by NER, the small fraction of dimers involved in transcription- collision needed the special TCNER to be removed. Consequently, CS cells were killed by UV even though the large majority of dimers were removed from their DNA. However, it remained puzzling why different cases of Cockayne's syndrome sometimes had different patterns and severities of the symptoms, and there were cases that had both Cockayne syndrome and xeroderma pigmentosum symptoms (Nance and Berry, 1992). As so often happens in research, the real world, as opposed to simpler worlds of theory, hides complications that challenge researchers, as in a treasure hunt. More recent findings, it turns out, indicate that the molecular defects in Cockayne's syndrome have more widespread aspects, including defects in base excision DNA repair and mitochondrial functions (Karikkineth et al., 2017). References Anderson, T.M. (1889). Note of a Rare Form of Skin Disease: Xeroderma Pigmentosum. British medical journal 1, 1284-1285. Bohr, V.A., Smith, C.A., Okumoto, D.S., and Hanawalt, P.C. (1985). DNA repair in an active gene: removal of pyrimidine dimers from the DHFRgene of CHO cells is much more efficient than in the genome overall. Cell 40, 359-369. Bradford, P.T., Goldstein, A.M., Tamura, D., Khan, S.G., Ueda, T., Boyle, J., Oh, K.S., Imoto, K., lnui, H., Moriwaki, S., et al. (2011). Cancer and neurologic degeneration in xeroderma pigmentosum: long term follow-up characterises the role of DNA repair. J Med Genet 48, 168-176. Cleaver, J.E. (1968). Defective repair replication of DNA in xeroderma pigmentosum. Nature 218, 652-656. Cleaver, J.E. (1972). Xeroderma pigmentosum: variants with normal DNA repair and normal sensitivity to ultraviolet light. J Invest Dermatol 58, 124-128. Cockayne, E.A. (1936). Dwarfism with retinal atrophy and deafness. Arch Dis Child 11, 1-8. De Weerd-Kastelein, E.A., Keijzer, W., and Bootsma, D. (1972). Genetic heterogeneity of xeroderma pigmentosum demonstrated by somatic cell hybridization. Nature: New biology 238, 80-83. DiGiovanna, J.J., and Kraemer, K.H. (2012). Shining a light on xeroderma pigmentosum. J Invest Dermatol 132, 785-796. 469 K. W. Kohn Drugs Against cancer CHAPTER 22 Epstein, J.H., Fukuyama, K., Reed, W.B., and Epstein, W.L. (1970). Defect in DNA synthesis in skin of patients with xeroderma pigmentosum demonstrated in vivo. Science 168, 1477-1478. Fornace, A.J., Jr., and Kohn, K.W. (1976). DNA-protein cross-linking by ultraviolet radiation in normal human and xeroderma pigmentosum fibroblasts. Biochimica et biophysica acta 435, 95-103. Fornace, A.J., Jr., Kohn, K.W., and Kann, H.E., Jr. (1976). DNA single-strand breaks during repair of UV damage in human fibroblasts and abnormalities of repair in xeroderma pigmentosum. Proceedings of the National Academy of Sciences of the United States of America 73, 39-43. Fornace, A.J., Jr., and Seres, D.S. (1982). Repair oftrans-Pt(II) diamminedichloride DNA- protein crosslinks in normal and excision-deficient human cells. Mutation research 94, 277-284. Hunter, W.B. (1889). Notes of Three Cases ofXeroderma Pigmentosum, or Dermatosis Kaposi. British medical journal 2, 69-71. Karikkineth, A.C., Scheibye-Knudsen, M., Fivenson, E., Croteau, D.L., and Bohr, V.A. (2017). Cockayne syndrome: Clinical features, model systems and pathways. Ageing Res Rev 33, 3-17. Kraemer, K.H., and DiGiovanna, J.J. (2015). Forty years of research on xeroderma pigmentosum at the US National Institutes of Health. Photochem Photobiol 91, 452- 459. Nance, M.A., and Berry, S.A. (1992). Cockayne syndrome: review of 140 cases. Am J Med Genet 42, 68-84. Robbins, J.H., Kraemer, K.H., Lutzner, M.A., Festoff, B.W., and Coon, H.G. (1974). Xeroderma pigmentosum. An inherited diseases with sun sensitivity, multiple cutaneous neoplasms, and abnormal DNA repair. Annals of internal medicine 80, 221-248. Schmickel, R.D., Chu, E.H., Trosko, J.E., and Chang, C.C. (1977). Cockayne syndrome: a cellular sensitivity to ultraviolet light. Pediatrics 60, 135-139. Venema, J., Mullenders, L.H., Natarajan, A.T., van Zeeland, A.A., and Mayne, L.V. (1990). The genetic defect in Cockayne syndrome is associated with a defect in repair ofUV- induced DNA damage in transcriptionally active DNA. Proceedings of the National Academy of Sciences of the United States of America 87, 4 707-4 711. 470 K. W. Kohn Drugs Against cancer CHAPTER 23 Chapur-ZJ. Th~ DNA nuckotitk ad$1/Jn ~pair story ZZ072Say 3 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@nih.gov CHAPTER23 The DNA Nucleotide Excision Repair Story: cutting out the damage. Evolution has come up with an amazing set of tools to repair the great variety of damage that DNA can accrue. The repair tools can sustain life despite damage from radiation, environmental chemicals, chemotherapy, metabolic errors, reactive oxygen species from aerobic metabolism, cosmic rays, etc. The major sites of chemical damage to DNA, as understood in 1993, are shown in Figure 23.1. In view of these hazards, Tom Lindahl surmised that specific mechanisms must exist to repair those various kinds of damage to DNA (Lindahl, 1982). Lindahl shared the 2016 Nobel Prize in Chemistry with Aziz Sancar and Paul Modrich for their discoveries of DNA repair mechanisms (Figure 23.2) (Cleaver, 2016; Orren, 2016; Van Houten, 2016; Zagorski, 2005) . The various DNA repair mechanisms are outlined in Figure 23.3, which also indicates in the legend the chapters that discuss each of them. The current chapter delves into the tools that repair the greatest variety of chemical damage to DNA: nucleotide excision repair (NER), which cuts out a damaged piece of DNA strand and replaces it with a good piece. There are two types of NER: global-NER and transcription coupled NER (TCNER). Global-NER repairs DNA damage that may be present anywhere in the genome, whereas TCNER repairs DNA damage at sites where RNA polymerase is transcribing the genome. Genetic defects in NER cause xeroderma pigmentosum whereas defects in TCNER are associated with Cockayne's syndrome (Chapter 22). 471 K. W. Kohn Drugs Against cancer CHAPTER 23 Sites susceptible to: --+ Hydrolytic attaclc. ~ Oxidative damage -Z-. Non tie methvtation bysTnosy,methronif'le o· Figure 23.1. Sites where DNA is vulnerable to chemical damage, as summarized by Tomas Lindahl is 1993. The large red arrows pointto the bonds in guanine or adenine (the purines) that are vulnerable to cleavage, thereby dislodging the guanine or adenine base from the DNA The smaller red arrow pointing to the bond at cytosine's amino (HN2) group indicated where hydrolysis can replace the amino group by an oxygen atom, thereby converting cytosine to uracil. The open blue arrows show where oxidative damage can occur; those that point to the double-bond of thymine or cytosine show where UV light pairs them together as thymine dimers and thymine-cytosine dimers, collectively known as pyrimidine dimers (Lindahl, 1993). 472 K. W. Kohn Drugs Against cancer CHAPTER 23 Figure 23.2. The three scientists who shared the 2016 Nobel Prize in Chemistry for their discovery of mechanisms by which human cells repair DNA damage. Aziz Sancar, born on 8 September 1946 in the Mardin Province of southeastern Turkey, received an MD degree at Istanbul University in 1969 and a PhD in molecular biology at the University of Texas at Dallas in 1977. He was elected to the National Academy of Sciences in 2005 as the first Turkish-American member. He discovered several DNA repair enzymes, elucidated how they work, and made notable discoveries also in other areas of cell biology. Thomas Robert Lindahl, born on 28 January 1938 in Stockholm, Sweden, received a PhD in 1967 and an MD in 1970, both from the Karolinska Institute in Stockholm. He isolated a mammalian DNA ligase that operates in NER and discovered a the totally unanticipated DNA glycosylases as mediators of DNA base excision repair. Paul Law rence Modrich, born on 13 June 1946 in Raton, New Mexico, received a Ph.D. degree from Stanford University in 1973 He has been working at Duke University, Durham, North Carolina since 1976 and is also affiliated with Howard Hughes Medical Institute, Chevy Chase, Maryland. He discovered DNA mismatch repair and was elected to the National Academy of Medicine and the National Academy of Sciences. Nuclear ONA repair pathways Base excision Recombination Mismatch Direct reversal ShOrt patch Long patch Topic of this chapter. Figure 23.3. An overview of the types of DNA repair mechanisms operating in the cell nucleus. The topic of this chapter is outlined in red. From (Kohn and Bohr, 2001). Base excision repair will be the topic of Chapter 24; recombinational repair, Chapter 26; mismatch repair, Chapter 25; nucleotide excision repair (NER), this chapter; direct reversal was the topic of Chapter 2. 473 K. W. Kohn Drugs Against cancer CHAPTER 23 Nucleotide excision repair (NER): cutting out and mending damaged DNA. An evolutionary solution to the wide variety DNA damage might seem easy: simply cut out the damaged section of the DNA strand and patch the resulting gap with new DNA. Easily said -- which is not to imply that the "blind watchmaker" said or thought anything at all -- but it must have taken a great deal of more-or-less random evolution to design (or, rather, to select) the chemical and enzymatic processes to carry it out. Nucleotide excision repair (NER) was perfected eons ago in the simplest microorganisms, possibly even before or during the time when the accumulation of oxygen from photosynthesis became the most toxic environmental mutagen in the history of our planet. Some DNA repair genes and mechanisms in animals might have arisen by gene transfers or incorporation of whole organisms (as in the case of mitochondria) from the microbes that originally evolved the repair machinery. The major NER genes were discovered in the course of investigations of the various complementation groups of xeroderma pigmentosum (XP), as I explained in Chapter 22. They became known as XPA through XPG, plus a variant called XPV. These genes were listed in Table 22.2 in the previous chapter, but is reproduced here for convenience: Table 23.1. Xeroderma oil!mentosum l!enes (comolementation l!rou is) Gene Synonyms Chromosome XPA XPl 9q22.33 XPB ERCC3 (excision reoair 3, helicase subunit) 2a14.3 XPC RAD4 3p15.l XPD ERCC2 ( excision repair 2, helicase subunit) 19q13.32 XPE DDB2 (damal!e-soecific DNA bindinl! 2) lloll.2 XPF FAN CO RADl, ERCC4 (excision reoair 4, endonuclease1 16013.12 XPG ERCCS (excision reoair 5, endonuclease1 13a33.l XPV XP variant, DNA polymerase eta 6q21.1 (Information from the human gene nomenclature committee (HGNC) website.) Repair by cutting out damaged sections ofDNA: a ne w idea. In his review of the "very early history of nucleotide excision repair," Errol Friedberg says that this new idea "was much in the air in the 1950s and early 1960s and was circulating freely in the informal grapevine of seminars and meetings" (Friedberg, 2011). Those airy ideas also circulated freely among us in 1960 in Paul Doty's laboratory at Harvard, as well as among our colleagues at nearby Brandeis University. Friedberg cites a "prophetic notion" by Evelyn Witkin in the early 1960s, based on observations of the induction of certain mutations by UV light in bacteria, that "some type of enzyme-catalyzed dark repair" was involved. 474 K. W. Kohn Drugs Against cancer CHAPTER 23 Friedberg goes on to cite a report in 1963 by Robert Haynes that, if yeast irradiated with x- rays or UV were left for a few hours of delay before being allowed to grow on an agar surface, they were able to form a greatly increased number of colonies; it seemed that the yeast were repairing the damage during the delay period. A related, at first puzzling, observation was that, if the same dose of x-rays was delivered to bacteria, but spread out in time, the bacteria survived much better; it was soon grasped that repair was taking place during the protracted x-ray treatment. But the first experimental evidence for DNA excision repair was obtained by Richard Setiow. Friedberg cites Haynes recalling that in the Fall of 1963, Setlow visited him at the University of Chicago and triumphantly asked "Do you know how E. coli repair [UV-damaged] DNA?" When Haynes replied "No, how?", Setlow said "They cut out the [thymine] dimers and throw them away!" A lucky break, Setlow said, was his early introduction to thymine dimers, and he recalled that"One of the people on the [Oak Ridge] staff approached me one day and asked what I thought about the experiments carried out by 'those crazy Dutchmen'?" referring to the experiments of Beukers and Berends showing the production of dimers by UV-irradiating frozen solutions of thymine (see Figure 21.15 and associated text in Chapter 21). On the basis of experiments showing that, after being irradiated with UV, bacteria gradually released small pieces of DNA containing the thymine dimers, "throwing them away," Setiow surmised correctly that there must be two cutting events -- it turned out later that the cuts are produced by two of the XP enzymes: XPF and XPG: a left-side cutter and a right-side cutter, respectively (Friedberg, 2011). I found a picture of Richard Setiow that shows his bright interest in all there is in the nature world (Figure 23.4). I don't know what was swimming in the small fish tank he was holding: maybe it was a species whose DNA repair peculiarities he was studying. When I visited him at his home on Long Island in the 1970s or 1980s I felt his warm friendly informal manner - how he prepared an excellent dinner ad hoc without fuss and invited my collaboration in that enterprise; the relaxed intense way he wanted to talk about all kinds of things; it was hard to keep up with his thinking. It was only on reading his obituary that I learned that he stemmed from The Bronx, New York, where I too grew up a decade later. It is remarkable how many well-known people I met later in life or read about were from The Bronx of that era. A next step in the saga was made by a former graduate student of Setlow's: Philip Hanawalt, who was by then on the faculty at Stanford University. His experiments are a bit complicated, and we don't need the details here; they are summarized by Friedberg (Friedberg, 2011). The story was essentially this: some time in 1963, Hanawalt called Setlow to tell him about some puzzling findings he had made. Setlow responded in a letter, dated August 23, 1963, telling him that his (Hanawalt's) findings were in accord with new results in his own laboratory and showed that UV-resistant E. coli strains had enzymes to cut the damage out of their DNA and that a UV-sensitive mutant E. coli could not do so, perhaps because the hypothetical "repair enzymes" whose job it was to cut out the damaged pieces of DNA were defective in those bacteria. Hanawalt said it was the first time he heard the word "repair" used in this context (Friedberg, 2011). The following year, Hanawalt and his student David 475 K. W. Kohn Drugs Against cancer CHAPTER 23 Pettijohn, published a paper with the title: "Evidence for the repair-replication ofultraviolet- damaged DNA in bacteria" (Pettijohn and Hanawalt, 1964). A third laboratory that was simultaneously homing in on DNA repair in bacteria was that of Paul Howard-Flanders, a noted radiobiologist at Yale. The interactions between the two laboratories were complex and sometimes competitive, as Friedberg described them, (Friedberg. 2011). Finally, we should note that almost all of those advances were facilitated by making use of the radiation-sensitive mutant E. coli Bs1strain that had been isolated by Ruth Hill (see Chapter 21). Studies in mammalian cells began later, after cell culture techniques were refined, particularly by the work of Mortimer Elkind described in Chapter 21. Figure 23.4. Richard ("Dick") Setlow (1921-2015) was a biophysicist at Oak Ridge National Laboratory in Tennessee, as well as an adjunct Professor at Stony Brook University on Long Island, New York, and a member of the National Academy of Science. He is best known for advancing the frontiers of what we know about how DNA damage is repaired. (Photo published in Long Island/Obituaries, provided by Setlow family.) Nucleotide excision r epair (NER): discovering how it works. As Bernard Strauss noted not so long ago, NER works only because DNA is double-stranded, which allows a damaged region of one strand to be cut out and replaced by copying undamaged information from the undamaged complementary strand. He suggested that this duplication of information may be one reason why DNA is double-stranded (Strauss, 2018). Other types of life may be discovered thriving in other planets of our solar system under what would for us be noxious conditions; those alternative living systems would do well also to have chemical duplicates of their genetic information. 476 K. W. Kohn Drugs Against cancer CHAPTER 23 An early clue to the workings of NER, came from tracking down some peculiar differences among xeroderma pigmentosum (XP) cells of different complementation groups (see Chapter 22 for explanation ofXP complementation groups and Table 23.1 or 22.2 for a list). In 1975, Ken Kraemer and his colleagues reported that cells defective in XPC retained the ability to respond to UV light albeit with a reduced rate of unscheduled DNA synthesis (see Chapter 22) (Kraemer et al., 1975a; Kraemer et al., 1975b). It seemed that there was a type of NER that did not require a normal XPC gene. Subsequent research disclosed two modes of NER: global-NER and transcription-coupled NER (TCNER), which came to light from studies of cells from the XP-related genetic disease, Cockayne's syndrome. A peculiar characteristic of those TCNER-deficient cells was that that they did not require the function of a normal XPC gene to carry out unscheduled DNA synthesis. Global-NER repaired damage anywhere in the genome, while TCNER specialized in repairing damage in regions of the genome that were in the process of being transcribed. The proteins responsible for TCNER were found to be bound to the transcription complex as it moved along to transcribe the DNA. The molecular interactions involved in both types of repair are diagrammed in Figure 23.5 in the form of a molecular interaction map (Kohn, 1999). I ,' 2 ---- - -Q9--DNA ', . r+-....----, . R23B ·..... ...............................,1 TFIIH XPB XPD ~-------·-·-----t ......_:_~....'...\~-~~: RPA i 7 t6 1 5~ ' --0-+" - : 1 3' ! _ _ 1_3___,__:·_4...., -, , <" · 12 ERCCl 1:: 15 iliiiAJ • !: B 10 ' XPA 477 K. W. Kohn Drugs Against cancer CHAPTER 23 Figure 23.5. Molecular interaction map (Kohn (1999) of nucleotide excision repair (NER) in humans showing the roles of the XP proteins, as understood in 2001 (Kohn and Bohr, 2001). The meaning of the symbol for each step is explained in Box 1. Essentially: first, XPC (in complex with HR23B) detects and binds to the damaged site (shown as an x in a small circle). XPC then brings in XPB and XPD, which unwind the DNA helix on both sides of the damage. XPA then replaces XPC at the damage site. The XP's, together with RPA and some other proteins all bind together to form a big complex, which allow XPG to cut the strand at one side of the damage and XPF to cut it on the other side. That is how the damaged section of DNA is cut out. We see that the XP's, together with some other proteins, work together to cut out a section of damaged DNA strand. A DNA repair polymerase then comes along to extend the cut strand from its 3' cut end and copies the information in the complementary strand. Finally, a DNA ligase seals the newly replicated DNA to the 5' cut end of the strand. (The repair polymerase and ligase steps are straight-forward and are not included in the diagram.) Figure 23.5 shows how the XP proteins work together in nucleotide excision repair. Global- NER must first detect and then cut the damaged region out of the damaged DNA strand. It must then replace the gap in the damaged strand with DNA copied from the undamaged strand. The mechanism whereby NER detects and cuts out the damage, as understood in 2001, is diagrammed in Figure 23.5. What happens at each step is explained in Boxl. The first challenge for global-NER is to find where the damage sites are located within the enormous length of the genome. The main actor to accomplishing this feat is the protein encoded by the XPC gene. The upper dashed box in Figure 23.5 depicts a normal double- stranded DNA helix bearing a lesion that is recognized and bound by a combination of XPC and HR23B. A DNA segment surrounding the lesion is then unwound by the XPB and XPD helicases, and XPA replaces XPC at the damage site (indicated by an x within a small circle). RPA, a single-strand binding protein, then binds to the undamaged DNA strand in the unwound region of the DNA helix. The lower dashed box depicts the region of unwound helix and the cutting of the DNA single- strand segment on either side of the lesion by the endonucleases XPF-ERCCl and XPG. The transition from the closed to the unwound state of the DNA (with associated proteins) is indicated by the long vertical dotted line, labelled interaction 4. The transition replaces the XPC:HR23B complex by XPA XPA then assembles proteins that participate in the excision of the lesion. In the case of transcription-coupled repair (TCNER), the repair proteins don't have to search for damage sites, because the repair machinery is bound to the transcription machinery and comes into play when the transcription encounters a damaged site. Moreover, the DNA helix around the lesion is already unwound due to the transcription process, and therefore repair can begin with XPA (lower box in Figure 23.5) and therefore does not need XPC to initiate the unwinding. That explains why normal XPC function is not required for TCNER. 478 K. W. Kohn Drugs Against cancer CHAPTER 23 Box 1 . Explanation of the reactions/interactions of the steps numbered in Figure 23.5: (1) A lesion in one strand of an intact DNA helix binds a dimer consisting of the XPC and HR23B proteins. (2) XPC:HR23B binds the TFIIH transcription complex, which contains the DNA helicases XPB and XPD (Yokoi et al., 2000). All of these proteins are needed for the initial opening of the DNA helix at the site of the lesion. XPB and XPD on opposite sides of the lesion to unwind the DNA for a short distance on both sides of the lesion. (3) XPA can then bind to the lesion (however, in the case of TCNER, the helix is already opened by the transcription process, and XPA can bind to the lesion without the aid of XPC:HR23B). (4) The vertical hatched line with solid triangle arrowhead indicates that the DNA helix opens and XPC:HR23B is replaced by XPA. The TFIIH complex now is bound to XPA instead ofto XPC:HR23B. (5) XPA binds the DNA single-strand-binding protein RPA. (6) RPA binds the undamaged strand where the helix has been opened. Thus, RPA helps to stabilize the XPA complex at the site of the lesion. (7) RPA recruits endonuclease XPG. (8) XPG binds XPA, which is bound to the lesion. (9) XPG cuts the lesion-containing strand on the 3' side of the lesion. (10) XPA recruits the XPF:ERCCl heterodimer to the lesion site. (11) RPA binds XPF and directs it to cut on the 5' side of the lesion. (12) XPF cuts the lesion-containing strand on the 5' side, thereby releasing a segment of DNA single-strand containing the lesion. (13) TFIIH binds to XPG. (14) TFIIH positions XPG on the 3' side of the lesion. (15) XPG recruits PCNA, which is required for the subsequent DNA repair synthesis that fills the gap left by the excised single-strand segment. (16) RPA binds p53 and thereby signal the presence of DNA damage. XPF-ERCC1 as a chemotherapy target. In the late 1970's and early 1980's, many genes were identified that corrected DNA-repair defects in human and rodent cells. In addition to the XP genes, several other DNA repair- correcting genes were identified, among which there was a gene that came to be called ERCCl. This ERCCl gene was cloned in 1984 (Westerveld and Naylor, 1984), and in 1993 it was found to correct the defect in XPF cell lines (Biggerstaff et al., 1993). The XPF and ERCCl proteins were then found to bind tightly to each other. The XPF-ERCCl complex constituted the enzyme that cuts the DNA strand on the 5' side of the lesion. Looking at Figure 23.5, we see that XPF-ERCCl functions in NER to cleave the damaged strand on the 5' site of the damage site (reaction 12 in Figure 23.5). Moreover, the XPF-ERCCl complex 479 K. W. Kohn Drugs Against cancer CHAPTER 23 also has a similar role in other DNA repair pathways, such as in the repair of DNA interstrand crosslinks. In general, it cleaves a DNA strand where a region of unwound DNA joins normal duplex DNA, as we see in Figure 23.5. ERCCl attracted attention when high ERCCl expression was found to favor the long-term survival of lung cancer patients whose tumors were resected by surgery without chemotherapy (Simon et al., 2005) (Figure 23.6). On the other hand, high ERCCl expression reduced the survival in advanced non-small cell lung cancer (NSCLC) treated with chemotherapy (Lord et al., 2002) (Postel-Vinay et al., 2012) (Figure 23.7). Although the circumstances of the two studies were quite different, they suggested that ERCCl had two- faces when it came to whether it benefits or harms. It was presumed that high ERCCl level would increase DNA repair activity. Lung cancer patients whose tumors were caught early enough to be amenable to surgery without chemotherapy benefited if their tumors expressed ERCCl at a high level. The benefit would have been due to enhanced DNA repair when there was lots of ERCCl, which would reduce the DNA damage that happens because of the unrestrained division of cancer cells. On the other hand, the detrimental effect of high ERCCl would come from the consequent high DNA repair activity countering the DNA damage whereby chemotherapy kills cancer cells. Hence, high DNA repair capacity could counter the therapeutic action agents, such as cisplatin, whose therapeutic action depends on producing DNA damage. Therefore, inhibitors ofXPF-ERCCl functions were recently developed and were found to synergize with the DNA damaging agents, cisplatin and mitomycin, in killing cancer cells culture (Ciniero et al., 2021). • 1.00 • .• u "- 0 .7S High ERCCl L-- • 1 ERCC I > 50 -•. !' o.so ·-- ---- ..•------- · . 0 mctl \ 50 0 0 .2:s > . > ~ 0 . f O ' - . - - - - -- ~ - - ~- - ~ ~ - -, - - - , - - - - - r ' 0 to ◄O •• •• Su,-vl w.l l_,.,t.hsl 100 120 110 Figure 23.6. Lung cancer (NSCLC) patients whose tumors were localized enough to be resected by surgery, had a better chance of long-term survival if their cancer expressed ERCCl at a high level (Simon et al., 2005). 480 K. W. Kohn Drugs Against cancer CHAPTER 23 1.0 .8 ERCC1 mRNA < median value (6.7) .2 ERCC1 mRNA > median value (6,7) 0 .0 + - - - - - - - ~ . . . . . . - - - - ~ 0 20 40 60 80 100 120 Overall Survival (weeks) P"(),009 I.ti& nint ICM Figure 23.7. Advanced lung cancer (NSCLC) patients who were treated with a combination of cisplatin and gemcitabine survived longer if their cancer's ERCCl expression was low (Lord et al., 2002). References Biggerstaff, M., Szymkowski, D.E., and Wood, R.D. (1993). Co-correction of the ERCCl, ERCC4 and xerodenna pigmentosum group F DNA repair defects in vitro. The EMBO journal 12, 3685-3692. Ciniero, G., Elmenoufy, A.H., Gentile, F., Weinfeld, M., Deriu, M.A., West, F.G., Tuszynski, J.A., Dumontet, C., Cros-Perrial, E., and Jordheim, L.P. (2021). Enhancing the activity of platinum-based drugs by improved inhibitors of ERCCl-XPF-mediated DNA repair. Cancer chemotherapy and pharmacology 87, 259-267. Cleaver, J.E. (2016). Profile of Tomas Lindahl, Paul Modrich, and Aziz Sancar, 2015 Nobel Laureates in Chemistry. Proceedings of the National Academy of Sciences of the United States of America 113, 242-245. Friedberg, E.C. (2011). Nucleotide excision repair of DNA: The very early history. DNA repair 10, 668-672. Kohn, K.W. (1999). Molecular interaction map of the mammalian cell cycle control and DNA repair systems. Mol Biol Cell 10, 2703-2734. Kohn, K.W., and Bohr, V.A. (2001). Genomic Instability and DNA Repair. In Cancer Handbook (Nature Publishing Group), pp. 85-104. Kraemer, K.H., Coon, H.G., Petinga, R.A., Barrett, S.F., Rahe, A.E., and Robbins, J.H. (1975a). Genetic heterogeneity in xerodenna pigmentosum: complementation groups and their relationship to DNA repair rates. Proceedings of the National Academy of Sciences of the United States of America 72, 59-63. 481 K. W. Kohn Drugs Against cancer CHAPTER 23 Kraemer, K.H., De Weerd-Kastelein, E.A., Robbins, J.H., Keijzer, W., Barrett, S.F., Petinga, R.A., and Bootsma, D. (1975b). Five complementation groups in xeroderma pigmentosum. Mutation research 33, 327-340. Lindahl, T. (1982). DNA repair enzymes. Annual review of biochemistry 51, 61-87. Lindahl, T. (1993). Instability and decay of the p rimary structure of DNA. Nature 362, 709- 715. Lord, R.V., Brabender, J., Gandara, D., Alberola, V., Camps, C., Domine, M., Cardenal, F., Sanchez, J.M., Gumerlock, P.H., Taron, M., et al. (2002). Low ERCCl expression correlates with prolonged survival after cisplatin plus gemcitabine chemotherapy in non-small cell lung cancer. Clinical cancer research : an official journal of the American Association for Cancer Research 8, 2286-2291. Orren, D.K. (2016). The Nobel Prize in Chemistry 2015: Exciting discoveries in DNA repair by Aziz Sancar. Sci China Life Sci 59, 97-102. Pettijohn, D., and Hanawalt, P. (1964). Evidence for Repair-Replication of Ultraviolet Damaged DNA in Bacteria. Journal of molecular biology 9, 395-410. Postel-Vinay, S., Vanhecke, E., Olaussen, K.A., Lord, C.J., Ashworth, A., and Soria, J.C. (2012). The potential of exploiting DNA-repair defects for optimizing lung cancer treatment Nat Rev Clin Oncol 9, 144-155. Simon, G.R., Sharma, S., Cantor, A., Smith, P., and Bepler, G. (2005). ERCCl expression is a p redictor of survival in resected patients with non-small cell lung cancer. Chest 127, 978-983. Strauss, B.S. (2018). Why Is DNA Double Stranded? The Discovery of DNA Excision Repair Mechanisms. Genetics 209, 357-366. Van Houten, B. (2016). A tale of two cities: A tribute to Aziz Sancar's Nobel Prize in Chemistry for his molecular characterization of NER. DNA repair 37, A3-A13. Westerveld, A., and Naylor, S. (1984). Report of the Committee on the Genetic Constitution o f Chromosomes 18, 19, 20, 21, and 22. Cytogenet Cell Genet 37, 155-175. Zagorski, N. (2005). Profile of Aziz Sancar. Proceedings of the National Academy of Sciences o f the United States of America 102, 16125-16127. 482 K. W. Kohn Drugs Against cancer CHAPTER 24 Chapur24. 1',~ DN A Bow &dsion RqH,Ir Sro,y: removing bod bost.t 220 7Z6an3 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@nih.gov CHAPTER24 The DNA Base Excision Repair Story: removing bad bases. From nucleotide excision repair (NER), which was the topic of the previous chapter, we now move on to a related DNA repair process, namely base excision repair (BER) (Figure 24.1) . I Nuclear DNA repair pathways I I I I I I Base excision Recombination I Mismatch Nucleotide excision Direct reversal I I I I I ShOrt patch I I Long patch I I Transcription- couoled I I (lenome Global I Topic of this chapter. Figure 24.1. The topic of this chapter, base excis ion repair (BER), in relation to the other DNA repair pathways in the cell nucleus. From (Kohn and Bohr, 2001). 483 K. W. Kohn Drugs Against cancer CHAPTER 24 The uracil-DNA glycosylase story In 1974, Thomas Lindahl reported his investigations of why DNA slowly loses its biological activity when it is left in a warm solution for a long time (Lindahl, 1974, 1976). DNA had been reported to suffer several types of chemical changes spontaneously at a rate that increased with increasing temperature. Lindahl focused on one of those chemical reactions that he thought likely to be biologically significant even in unstressed organisms. He focused on cytosine's amino group that very slowly comes off and is replaced by a hydroxyl group. This is a typical hydrolysis reaction in water solutions. The biological problem was that the slow reaction changed the cytosine to uracil, thereby changing the base-pairing: cytosine pairs with guanine while uracil pairs with adenine. Lindahl reasoned that evolution must have found a way to overcome that problem. Therefore, he searched for an enzyme that might be able to remove the offending uracil whenever it reared its ugly head in DNA. What he discovered was the first of a large group of enzymes, each of which can pluck off a particular abnormal base from DNA. The enzyme he discovered specifically removed uracil by means of a "glycosylase" reaction, which breaks the bond between one of uracil's nitrogen atoms and deoxyribose. The uracil -DNA glycosylase, as it came to be known, only removed uracil from DNA, but did not remove uracil from RNA, where it belongs. Nor did it remove thymine (which is like uracil, but with an extra methyl group) from DNA. Thus, the DNA-uracil glycosylase enzyme was carefully designed to act only where it is needed. Moreover, DNA evolved to use thymine instead of uracil to pair with adenine in DNA, because uracil (generated by the inevitable hydrolysis of cytosine) would be removed by the glycosylase. Evolution is a remarkable designer! After an offending uracil is removed, it leaves behind a deoxyribose lacking any base. This is known as a base-free site in DNA. It is like a nucleotide with its head chopped off (left side of Figure 24.2). The base-free site that is left behind still needs to be repaired -- which is accomplished by previously discovered enzymes, as I will explain later in this chapter. The early work on DNA glycosylases was done in bacteria. But Lindahl later remarked that the major DNA repair pathways are surprisingly similar between E.coli bacteria and mammals and that cells of higher organisms have not evolved any novel DNA repair pathways; moreover, that DNA repair mechanisms probably evolved very early, because they appear to be present in all living organisms (Lindahl, 1982). He noted that the most important of the DNA repair pathways, in E coli as well as in human cells, repair damaged bases, which are among the most frequent type of DNA lesion. The first step in this pathway would be to locate the abnormal base and then to cut it out of the DNA, leaving behind a base-free site that must then be repaired. That entire process has come to be known as base-excision repair (BER) to distinguish it from nucleotide-excision repair (NER), which was the subject of the preceding chapter. Errol Friedberg too had a major role in the discovery of BER and gives an interesting personal account of its history (Friedberg, 2016). 484 K. W. Kohn Drugs Against cancer CHAPTER 24 Figure 24 .2. DNA base-excision repair (BER) as depicted by Thomas Lindahl in 1976 (Lindahl, 1976), with additions shown in red. The left side shows the removal of an offending uracil (U), leaving a base-free site. The right side then shows how the base-free site is repaired. In reaction a, hydrolysis of a cytosine's amino group changes the cytosine to uracil (U), which does not belong in the DNA of eukaryotes. In reaction b, DNA-uracil glycosylase cuts off the U and leaves behind a base-free site (empty red circle). In reaction c, an AP-lyase cuts the bond on the 3' side of the base-free unit (enclosed in a red box). In reaction d, an AP-endonuclease cuts the bond on the S' side of the base-free unit, leaving a gap in the DNA strand (shown by the empty red box). In reaction e, a DNA-repair polymerase adds the correct nucleotide (C) to the 3 ' end of the DNA strand. In reaction f, a DNA ligase finally seals the C in place, thereby completing the repair. (The diagram omits the complementary DNA strand.) Discovery of other DNA glycosylas es. Aside from uracil-DNA glycosylases, ma ny enzymes were discovered t hat cut off oth er abn ormal bases from DNA. By 1982, 15 DNA glycosylases had been d iscovered, each design ed to remove a particular abn ormal base fro m DNA, analogous to the removal of uracil sh own in Figur e 24.3 (Lindahl, 1982). The job of th ese enzymes was to cut th e bond between the abnor mal base a nd th e d eoxyr ibose sugar in DNA; t he enzymes simply catalyzed a hydrolytic cleavage of the bond a nd did not requ ire ener gy or any cofactor. (Hydrolytic cleavage is a reaction th at ind uces a water molecule to split a bond; an OH 485 485 K. W. Kohn Drugs Against cancer CHAPTER 24 becomes bound to one end of the split bond and an H to the other end.) Each glycosylase was found to act only on the abnormal base that evolution designed it to remove. ("glyc" refers to a sugar, in this case deoxyribose; thus, glycosylase means enzyme that cleaves a bond to a sugar.) One of the early reports that BER can remove and repair chemically modified bases from DNA came from Jacques Laval of the lnstitut Gustav-Roussy in Villejuif, France, in 1977 (Laval, 1977). He found that BER can remove and repair 3-methyl -adenine (3-meA) from DNA that had been treated with an alkylating agent (methyl-methanesulfonate). Here I cannot resist relating an anecdote about Jacques Laval. Sometime in the 1970s, my wife and I met Jacques at a DNA repair conference in Lyon, France. Jacques was as always ebullient, friendly and a great pleasure to be with and talk about DNA damage chemistry. At a meeting session one morning, he came over to chat and asked where we had dinner last night With a tiny bit of reluctance, I said we had found this fine Chinese restaurant... He said, "What! Here you are in the culinary capital of the world with so many 5-star French restaurants, and you go to a Chinese?" Well, a day or two later, he came over and asked: "Where did you say that Chinese restaurant was located?" It seems that, when one is constantly immersed in fine things of a particular kind, one could eventually tire of them. But I think that none ofus tired of the emergent story of DNA repair. As of 2009, eleven of the DNA glycosylases had been isolated (Robertson et al., 2009). Figure 24.3 shows the chemical structures of several abnormal bases that could erroneously be incorporated into DNA by a DNA polymerase and then removed by one or another of the glycosylases. Since specific enzymes have evolved to remove these abnormal structures from DNA, one may suspect that organisms have frequently formed or encountered such structures. After the uracil-DNA glycosylase, the next to be discovered and extensively studied was the glycosylase that removes an oxidation product of guanine (8-oxoG, also known as oxo8 G, Figure 24.3) (Klungland and Bjelland, 2007). 8-oxoG is an unavoidable by-product of normal oxidative metabolism in mitochondria. DNA polymerase can erroneously and easily incorporate 8-oxoG into DNA. But instead of pairing with C of the template strand as it should, the oxo8 G sometimes wiggles around in a manner that its alternative hydrogen bonding capability allows it to mis-pair with A or G (Figure 24.4). Such mis -pairings are a major cause of mutation. 486 K. W. Kohn Drugs Against cancer CHAPTER 24 fNt NH2 N.,,-CH, ,x5.· fN J I HC,N+:j 3 NH2 N ) ~ OA N I dR N I dR CH 3 I dR 1-meA 3-meA 3-meC ~i:oNANHNH o=<N I o}-~~ H H- N I ~ NH NH I 2 I 2 dR dR 8-oxoG FaPyG --~ 0 {Nl 'NH OA N I NJl..J I N dR dR Uracil Hypoxanthine Figure 24.3. Some of the chemically altered bases that BER is capable of removing from damaged DNA (Robertson et al., 2009). 487 K. W. Kohn Drugs Against cancer CHAPTER 24 A B C Figure 24.4. Oxidation processes of cell metabolism causes an oxygen atom addition to position 8 of guanine, which then allows the resulting 8-oxo-guanine to mis-pair with adenine or guanine. A, pairing of8-oxoG with C (like the normal G:C pair). B, mis-pairing of 8-oxoG with A. C, mis -pairing of 8-oxoG with G. (From (Klungland and Bjelland, 2007)). Repair of base-free sites The APE1 story After base excision repair (BER) removes a bad base, a base-free site remains that has to be repaired. The first step in that repair is carried out by an enzyme that is so remarkable and so important that it became one of the most intensively studied of all enzymes. It came to be called APEl (AP-endonuclease 1), although it had other confusing names as well. The enzyme was found to cleave the DNA strand in step c of Lindahl's 1976 diagram (Figure 24.2) and also contributed to step d. These are the two steps that cut the base-free unit out of the DNA. 488 K. W. Kohn Drugs Against cancer CHAPTER 24 Investigators soon realized that base-free sites actually were produced quite frequently. They were produced, not only during DNA repair, but also appeared spontaneously due to a slight instability of the bond that binds a purine or pyrimidine bases to the DNA. Lindahl estimated that a cell's DNA genome spontaneously loses about 10,000 bases per day. Since base-free sites occur so frequently, investigators searched for enzymes that might be implicated in their repair. In 1970, W. G. Yerly and his coworkers at the University of Montreal, Canada, found an enzyme activity that cleaved DNA at or near a base-free site, and by 1973, they had purified the enzyme from both bacteria and mammalian cells (Yerly et al., 1973). That is how APE1 was discovered. However, it soon emerged that APE1 is more clever than just to cleave a base-free site: it also detects and removes mismatched nucleotides if any such mismatches exist at the cleavage site. Thus, it is not only an endonuclease (which cleaves an intact DNA strand) but also an exonuclease that chews away mismatched nucleotides from the end of a cleaved strand. Moreover, APE1 turned out to have even more functions than that: it was found to be at the nexus and central regulator of DNA repair in all its complexity. Furthermore, APE1 was found to be overexpressed in cells of a variety of cancers, which made it a potential chemotherapy target (Fishel and Kelley, 2007). How base-excision repair (BER) works. A diagram of BER, as it was understood in 2009 (Robertson et al., 2009), is shown in Figure 24.5. It gives a more complete picture than was available in the 1976 diagram (Figure 24.2). In the first step, a glycosylase removes the damaged base, leaving behind a base-free site. APE1 then cleaves the DNA strand on one side of the site, creating a 3'-OH end that is suitable for the DNA repair polymerase, POLB, to hook onto. POLB (DNA polymerase beta) specializes in DNA repair. It extends the broken DNA chain from the 3'-OH end of the break, adding one or more nucleotides. Now, POLB is faced with a choice: whether to add only one or whether to continue adding several nucleotides. As of 2009, it was still unknown how the choice was made. In either case, however, the added nucleotides were chosen to match the complementary DNA stand, which is almost always what the major DNA polymerases do. The diagram suggests that the nucleotide addition continues when POLB hands off its work to POLD (Figure 24.5). The two branches of the BER mechanism each required a different set of enzymes and proteins; hence, the distinct mechanisms were awarded different names: short-patch versus long- patch repair. The short-patch mechanism was obviously the simpler process. It involved relatively few enzymes and proteins. In addition to the enzymes already mentioned, a protein called XRCC1, which lacked enzyme activity, came into play. It was thought to function as a kind of scaffold that held together the needed enzymes, presumably helping to make the process 489 K. W. Kohn Drugs Against cancer CHAPTER 24 fast and efficient- which made sense, in view of the frequent occurrence of base-free sites reported by Lindahl. But why was the more complicated long-patch process needed? In 2009 that was still a mystery (Robertson et al., 2009). One might speculate, however. A simple idea would be that POLB was occasionally too exuberant and continued adding nucleotides when it should have stopped after adding just one. Maybe that happens when the DNA strand it is facing happens to breathe. Breathe? Yes, the DNA double-helix would be expected occasionally and very transiently to come apart in regions where it is not locked in place by histones or other nuclear proteins. After having displaced and copied one more nucleotide than it should have, momentum might carry the polymerase forward until it runs out of steam and stops. Then a large collection of other proteins come in to fix this awkward situation where a displaced DNA single-strand is hanging out like a flap. I will mention but a few of the many proteins that were implicated in the long-patch repair process diagrammed in Figure 24.5 (Robertson et al., 2009). The ring protein complex, PCNA, clamps like a donut around replicating DNA, carrying a DNA polymerase as it slides along. synthesizing a replicated double helix. RFC is a clamp loader that assembles the PCNA clamp around the DNA. The way this happens is of course complicated, but by 2012 it had all been worked out (Kelch et al., 2012). RPA binds and stabilizes the DNA single- strand segment that is displaced by the long-patch repair process, and FENl is a special nuclease that cuts off that flapped segment. Finally, LIGl is a DNA ligase that seals the end of the newly synthesized DNA segment to form an intact strand. The whole process, including the replication of the two DNA strands, was found to be similar in life forms from bacteria to humans; it is an astonishing accomplishment of evolution. Although long-patch repair has to replicate a segment of only one of the DNA strands, it seems to need much of the same machinery as replication that duplicates the cell's DNA during the cell cycle. Long-patch repair, however, needs additional factors to take care of the displaced DNA single-strand segment, such as BLM and WRN, which are implicated in homologous recombination, and the MSH proteins that are implicated in DNA mismatch repair (Robertson et al., 2009). 490 K. W. Kohn Drugs Against cancer CHAPTER 24 Damaged DNA base l glycosylase Base-free site J,. TGGTACC 0 c A+GG I I I I I I I ♦ - APEl 3'- 0 H '\_ ; --,...,--.-,--.-, ---,-, --,-, A CATGG TGGTACC Short patch XRCCl ! POLB ! POLB POLO -<> 0 Long patch PCNA RFC RPA BLMWRN MSH 6 yb ~ ~~-.-~OH ,-, 7 I I I I A C C A T G G A C C A T G G T G G T A C C TGGTACC i POLB ,: OH - -,--1 p ~ ~--.---,-- p ~ A C C A T G G A C C A T GG T G G T A C C TGGTACC i A C C A T G G A C C A T GG T G G T A C C TGGTACC Figure 24.5. Steps in the repair of base-free sites after removal of an altered or damaged base from DNA where there is a damaged base. ((Robertson et al., 2009) with labels in red added.) BER helps repair a topoisomerase-1 (TOP1) blockage at a DNA lesion. Base-excision repair (BER) can help repair a variety of other DNA damage problems. A recent example comes from a collaboration that included several members of Yves Pommier's laboratory at NCI (Saha et al., 2020). The problem arises when TO Pl in its strand opening and closing reaction (see Chapter 11) encounter a DNA lesio n. The problem and its solution are depicted in Figure 24.6 and explained in its legend. 491 K. W. Kohn Drugs Against cancer CHAPTER 24 5' ... 6-4 pp Pol ~ XRCC I J © I (D ... ~ ='- ..... I ••• ® I ® r ... ~:;Nl Proteasome ® I TDP l/2 Ligase I © OH ... PNKP VIII Figure 24.6. A DNA repair problem arising when TOP1 encounters a DNA lesion, in this case a 6-4 pyrimidine dimer (Saha et al., 2020). In step 1, TOP1 has cleaved the DNA strand adjacent to a 6-4 PP and has bound to one end of the break by way of a tyrosine (Y) (see Chapter 11). The TOP1 is now trapped and cannot proceed. In step 2, a proteasome cuts away most of the trapped TOP1 protein. Step 3 removes the remaining peptide, including the Y, leaving a 3'OH. In steps 4 -6, long-patch BER comes into play and completes the repair, as in Figure 24.5. The b rief life of "DNA ins ertase". Because glycosylase enzymes were capable of removing improper bases from DNA, it seemed plausible that there might be an enzyme able to carry out the reverse reaction. The enzyme would directly insert the proper base into a base-free site. The existence of such a "DNA insertase" was reported in 1979. The following year, however, Errol Friedberg, working in Tom Lindahl's laboratory was unable to confirm the existence of such an enzyme. Subsequent reports suggested that the apparent "insertase" activity resulted from a combination of an enzyme that cleaved the base-free site and a polymerase that inserted the proper base - essentially like the short-patch repair on the left side of Figure 24.5. Doubts about insertase had already been raised, because the direct insertion of a base would require energy, whereas the reported reaction did not seem to need any. No further reports of DNA insertase activity appeared after the early 1980's, and so the albeit attractive insertase idea was quietly laid to rest (Friedberg, 2016). 492 K. W. Kohn Drugs Against cancer CHAPTER 24 References Fishel, M.L., and Kelley, M.R. (2007). The DNA base excision repair protein Apel/Ref-1 as a therapeutic and chemopreventive target. Mol Aspects Med 28, 375-395. Friedberg, E.C. (2016). A history of the DNA repair and mutagenesis field: The discovery of base excision repair. DNA repair 37, A35-39. Kelch, B.A., Makino, D.L., O'Donnell, M., and Kuriyan, J. (2012). Clamp loader ATPases and the evolution of DNA replication machinery. BMC Biol 10, 34. Klungland, A., and Bjelland, S. (2007). Oxidative damage to purines in DNA: role of mammalian Oggl. DNA repair 6, 481-488. Kohn, K.W., and Bohr, V.A. (2001). Genomic Instability and DNA Repair. In Cancer Handbook (Nature Publishing Group), pp. 85-104. Laval, J. (1977). Two enzymes are required from strand incision in repair of alkylated DNA. Nature 269, 829-832. Lindahl, T. (197 4 ). An N-glycosidase from Escherichia coli that releases free uracil from DNA containing deaminated cytosine residues. Proceedings of the National Academy of Sciences of the United States of America 71, 3649-3653. Lindahl, T. (1976). New class of enzymes acting on damaged DNA. Nature 259, 64-66. Lindahl, T. (1982). DNA repair enzymes. Annual review of biochemistry 51, 61-87. Robertson, A.B., Klungland, A., Rognes, T., and Leiros, I. (2009). DNA repair in mammalian cells: Base excision repair: the long and short of it. Cell Mol Life Sci 66, 981-993. Saha, L. K., Wakasugi, M., Akter, S., Prasad, R., Wilson, S.H., Shimizu, N., Sasanuma, H., Huang, S.N., Agama, K., Pommier, Y., et al. (2020). Topoisomerase I-driven repair of UV- induced damage in NER-deficient cells. Proceedings of the National Academy of Sciences of the United States of America 117, 14412-14420. Yerly, W.G., Paquette, Y., and Thibodeau, L. (1973). Nuclease for DNA apurinic sites may be involved in the maintenance of DNA in normal cells. Nature: New biology 244, 67-69. 493 K. W. Kohn Drugs Against cancer CHAPTER 25 Chapur-25. Th~ DNA ml:tmotd! ~pair story 220911.bgJ Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@nih.gov CHAPTER25 The DNA Mismatch Repair Story: fixing base-pairs that don 't match. Replicating a cell's genome is a challenge: all 6.4 billion base-pairs of the human genome must be copied correctly. The DNA polymerases that carry out this function are highly accurate: they make only one copying mistake every 10,000 to 100,000 nucleotides. But that leaves nearly 100,000 errors each time a cell divides, and each uncorrected error is apt to result in a mutation. These polymerases however have evolved a proofreading capability that allows them to detect and correct about 99% of the errors they have made. But that still leaves about 1,000 errors uncorrected. Those remaining errors still would have to be corrected to avoid accumulating mutations. Quite remarkably, almost all organisms from bacteria to mammals have evolved a backup system. This "DNA mismatch repair" system works similarly in all organisms and is carried out by variations of some of the same genes. It is the topic of this chapter, as shown in Figure 25.1 in relation to other DNA repair mechanisms. 494 K. W. Kohn Drugs Against cancer CHAPTER 25 I Nuclear DNArepair pathways I I ~ I I I I Base excision Recombination MiSmatch Nucleotide Direct excision reversal I I IShort patch I ILong patch I Transcription- coupled Global genome Figure 25.1. The topic of this chapter, DNA mismatch repair (MMR) outlined in red, in relation to the other DNA repair pathways in the cell nucleus. From (Kohn and Bohr, 2001). Cancer-prone families that have DNA mismatch repair defects. The story begins in 1895 when a seamstress who worked for Aldred Scott Warthin (Figure 25.2 left), Chairman of the Department of Pathology at the University of Michigan, said she was distressed that 5 of her 9 siblings had died of cancer, and she feared the same would happen to her, which unfortunately it did (Boland and Lynch, 2013). A family tree over 3 generations showed 33 of 70 family members having died of cancer of the uterus, stomach or colon. The first cancer-prone family tree, published by Warthin in 1913, is reproduced in Figure 25.3. Warthin traced the cancer tendency back to a German family who immigrated before the Civil War (Figure 25.2 right). Warthin's report indicated that cancer could have a familial origin, which was a new idea not readily accepted at the time and remained dormant until Henry T. Lynch (Figure 25.2 center) revived it many years later. The case for inherited factors disposing to cancer was eventually revived in 1970, when Henry T. Lynch of Creighton School of Medicine in Omaha, Nebraska, started compiling evidence from family histories. But, even then, his research grants were rejected, because of bias against the idea, and because his reported cancer frequencies were thought statistically coincidental. It was another 20 years of persistence that finally yielded evidence that could not be denied (Kunkel and Erie, 2005, 2015). With some justification Lynch became noted as "the father of cancer genetics," although he himself said that the designation rightfully belonged to Alfred Warthin. The most decisive case was "non- polyposis colon cancer," also known as Lynch syndrome, which was later found to be caused by an inherited mutation in a DNA mismatch repair gene. 495 K. W. Kohn Drugs Against cancer CHAPTER 25 Figure 25.2. (Left) Aldred Scott Warthin, MD, PhD, (1866-1931) Chairman of the Department of Pathology at the University of Michigan in Ann Arbor was first to report a cancer-prone family. (Center) Henry T. Lynch (1928-2019) compiled family trees to show that susceptibility to cancer is sometimes inherited. He defined hereditary non-polyposis colon cancer (HNPCC), also known as Lynch syndrome. Warthin and Lynch could share the designation "fathers of cancer genetics." (Ri9ht) The German family, who immigrated before the Civil War, to whom Aldred Warthin traced the first familial cancer disease (Boland and Lynch, 2013). 496 K. W. Kohn Drugs Against cancer CHAPTER 25 A f"AMJLY C. ,c;_ .__ S't• ... ••"- " t -.t , ~ u..• ~-~~----l ._.. u,- - .. c--.,- ,, .. , ....1 C-..• •t-.... •t--"- ""' c.-, 11,... Cl...,__ u .u 4l . ~ l. T-- \U•-. C•- $t_......,. -'-•.,.·· Figure 25.3. The first reported cancer-prone family, published in 1913 by Aldred Scott Warthin. The history traced back to a male founder (top of diagram) who died in 1856 at age 60 of cancer of stomach or intestine. (Squares, male; circles, female; hatched, died of cancer.) (Boland and Lynch, 2013). Discovery ofDNA mis match repair. Perhaps the most telling lesson ofthe past forty years has been the recognition that very different forms oflife are built around essentially similar mechanisms. All species are discovered to have more in common with each other than their differences would suggest (John Maddox in 'News and Views' Nature, Lond. 1993 363, 13.) DNA mismatch repair is a prime example of a system that exists in nearly all organisms from bacteria to mammals. Indeed, some of the genes of the bacterial system resembled the corresponding mammalian genes sufficiently to help find the mammalian genes once the bacterial ones were known (Radman et al., 1995). DNA base-pair mismatches (other than the normal A-TofG-C matches) can happen due to replication errors or to chemical DNA damage. An early question was whether such mismatches could be repaired. Studies of DNA damage in bacterial viruses in the early 497 K. W. Kohn Drugs Against cancer CHAPTER 25 1970's suggested that bacteria could do that. This was confirmed in 1975 by Wildenberg and Meselson, who prepared a bacterial virus bearing mismatched base-pairs. Upon infecting bacteria with this defective virus, the bacterial host was able to repair the mismatch and allowed the virus to multiply normally (Wildenberg and Meselson, 1975). Analogous experiments in mammalian cells infected with viruses inactivated by a base- base mismatch showed that the cells were able to reactivate these viruses. All of the possible mismatches were found to be repairable (Brown and Jiricny, 1988). This method: ability of a cell to rescue a virus having DNA damage, became a common way to detect the repair of various kinds of DNA damage, both in bacteria and in mammalian cells. Although known to exist in bacteria, much about mismatch repair remained a mystery in 1982, when Tom Lindahl reviewed what was then known about DNA repair (Lindahl, 1982). Mismatch repair would have to distinguish and repair the newly synthesized DNA strand, which is the strand that would incur errors during replication, but how the repair system did that was not clear. When Paul Mod rich reviewed what was known about mismatch repair in 1991, the enzymes and pathways were already well worked out in bacteria, but the details of how it worked in higher cells remained fuzzy (Modrich, 1991). Remarkably, however, a defect in mismatch repair resulted in high mutation rates in bacteria (Mod rich and Lahue, 1996). This was likely true also in mammalian cells because the repair systems functioned similarly. The genes and mechanis ms of DNA mismatch repair (MMRJ. DNA damage sometimes alters or deletes bases in DNA, so that they cannot associate with their complementary bases. This causes problems when the cell tries to replicate through such mismatches in its DNA The replicative DNA polymerases rarely make mistakes, and can self-correct most of them, but rare errors inevitably get through. When that happens, the base added to the end of the growing DNA chain does not match (A-Tor G-C) the corresponding base on the opposite DNA strand. Virtually all organisms have molecular machinery to repair such mismatches. When mismatch repair was defective, cancer was likely to ensue. Anticancer drugs also can result in base-pair mismatch, which impacts on the clinical outcome. In particular, and contrary to what one might have expected, the mismatch machinery sometimes made a drug more instead of less toxic to the cell. This chapter aims to clarify these mechanisms, and to review how all of this knowledge came to be uncovered. The human versions of the MMR genes were found by the albeit modest DNA sequence similarity with the bacterial versions (Radman et al., 1995). Six human MMR genes were discovered, whose proteins products interacted as shown in Figure 25.4. The six proteins were found to come together to recognize and repair two types of defects in DNA: mismatch of a single base-pair and loops formed by inserted or deleted base-pairs, which I will describe in tum. Defects in any of these genes (or of a gene called EPCAM that is located near the MSH2 gene) caused high mutation rates all over the genome, making people vulnerable to developing cancer (Baretti and Le, 2018). 498 K. W. Kohn Drugs Against cancer CHAPTER 25 ( hMSH2 ) ( hMSH6 ) < hMutSr.c e 1 hMutS p e •( hMSH3 ) (hPMS2)~•-◄•-~r e • ( hMLH3 ) ( hMLHl ) Single-bp Insertion/deletion MMR loop repair Figure 25.4. Molecular interactions and functional capabilities of the six human proteins implicated in DNA mismatch repair (MMR). hMSH2 combines with hMSH6 or with hMSH3, and hMLHl combines with hPMS2 or hMLH3. These pairs then come together to form the functional units (Kohn and Bohr, 2001). (The solid circles on the lines represent the species formed by the bindings - for example, the circle labeled hMUTSa., represents the hMUTS2- hMUTS6 dimer.) How MMR r epairs mismatch of a sing le base-pair. How the MMR repair process works in mammalian cell was still not completely understood at the time of this writing. A detailed model of how a mismatch is repaired was proposed by (Hsieh a nd Zha ng, 2017). A simplified version of their model is presented in Figure 25.5. The repair begins with a MSH2-MSH6 (or a MSH2-MSH3) dimer forming a clamp around the DNA. With the aid of PCNA, the clamp diffuses along the DNA in search of a base-base mismatch (C in Figure 25.5) (Pluciennik et al., 2010). It is somehow able to distinguish the newly replicated strand, which has the misincorporated base, but how it did that remained unknown at the time of the review by (Hsieh and Zhang, 2017). Then, with the aid of MLH1-PMS2, a break is created in the newly synthesized stra nd on one side or other of the mismatch (D in Figure 25.5). The segment of newly synthesized DNA that has the mismatch in it is then digested away by Exol, leaving a single-strand segment of template strand that becomes coated with the single-strand binding protein, RPA (E in Figure 25.5) (Kadyrov et al., 2009; Kunkel and Erie, 2015). Finally, a DNA polymerase that 499 K. W. Kohn Drugs Against cancer CHAPTER 25 specializes in DNA repair fills in the gap in the newly synthesized strand, inserting the correct base (Fin Figure 25.5). A ! L••ding Str>nd Synth•sis I Polymerase o/i; B ------J:::::::::~ 0 PCNA ! Mismatch recognition • RPA C ! '""~ Incision \ - -,y D --.--"=u,---i-v· ! R•mov•I E ·····••u••·· (,ol .141 RuynlhO>ls •nd Uc•lion F -----,G----- Figure 25.5. Simplified scheme of how MMR was thought to repair single mismatches in DNA (Brandon D'Arcy, ada pted and from (Hsieh a nd Zha ng, 2017).) Starting from the top: A shows the replicative synthesis of a DNA strand by a DNA polymerase (assisted by donut- shaped PCNA) p rogressing from left to right B shows a G:T mismatch in the DNA. In C, a MSH2:MSH6 (or MSH2:MSH3) dimer recognizes and binds to the mismatche d base- pair. In D, a MSH1:PMS2 dimer is recruited and diffuses along the DNA; it induces a break on either side of the newly replicated DNA. In E, an exonuclease (Exol ) has digested away t he segment between t he breaks, and the DNA single-stra nd segment left behind becomes sta bilized by binding an a rray of RPA molecu les. In F, the repair is completed by a DNA-repair polymerase. 500 K. W. Kohn Drugs Against cancer CHAPTER 25 Microsatellite ins tability indicates defective mismatch repair (MMR). Microsatellite instability is caused by and is a common feature in mismatch repair deficient cancers, especially cancers of colon, stomach, and ovary. To begin with, however, what is microsatellite DNA? Before answering that, however, I would like to digress briefly by mentioning how the jargon "satellite DNA" arose. Colleagues in Paul Doty's laboratory, when I was there inl 960, were banding mammalian DNA by CsCI equilibrium ultracentrifugation when they noticed a bump in the otherwise symmetrical peak, which indicated that a small fraction of the DNA had an unusually low GC/AT ratio. They called it satellite DNA because it showed itself as a small component adjacent to the bulk of the DNA (Figure 25.6). lt turned out to be the largest repetitive DNA component in mammalian cells, alphoid DNA, which is associated with the centromeres of all chromosomes. "Satellite DNA" eventually came to mean any set of repeated sequences in the genome. "Microsatellite DNA" came to refer to any relatively short DNA segment consisting ofrepeats of one or two (or rarely up to six) nucleotides. A major discovery about microsatellite DNA came in 1993 from researchers in Southern California (lonov et al., 1993). They discovered that many colon cancer patients had reduced numbers base-pair repeats in the microsatellite DNA in their tumors (Figure 25.7). They thought that the changes in the microsatellite DNA might be the cause of the cancer. Later investigations however revealed that the microsatellite changes were not the cause of the cancers, but rather were caused by a defect in a DNA repair mechanism: DNA mismatch repair, that was in fact a main cause of the cancers (Kunkel and Erie, 2015). Main peak ! Satellite 1 - DNA base ratio (GC/AT} Figure 25.6. Origin of the term "satellite DNA". It was first noted as a small sideband in the DNA of mammalian cells that was banded on the basis GC/AT ratio by ultracentrifugation to equilibrium in a concentrated CsCI gradient. The AT-rich satellite was later found to have a monomer length of a few hundred base-pairs in arrays of up to 100 million bases in the centromeres of chromosomes. 501 K. W. Kohn Drugs Against cancer CHAPTER 25 Normal Tumor ACGT ACGT Figure 25.7. Reduced number of A:T base-pair repeats in a microsatellite DNA in the tumors of a patient with colon cancer (Ionov et al., 1993). How micros atellite ins tability arises. The replicative DNA polymerases, in addition to mispairing errors, sometimes produced a different type of error. While copying a string of repeated nucleotides (or nucleotide pairs), the polymerase could slip forward or backward, causing deletion or insertion of one or more nucleotides in the newly replicated strand (Figure 25.8). Sequence repeats of say 6 to 20 nucleotides were found all over the genome. If a sequence repeat was in or near a gene, an insertion or deletion of one or two nucleotides in the repeated sequence was apt to cause a frame-shift mutation - that is when the triplet code that defines the amino acid sequence of the gene's protein product gets to be out of synch so that a subsequent triplet may code for nonsense or STOP (Bhattacharyya et al., 1995; Eshleman and Markowitz, 1996). The mismatch repair (MMR) system was found to recognize and repair insertions and deletions, thereby preventing those kinds of mutations. People who had frequent changes in the lengths of their sequence repeats - a condition called microsatellite instability - were found to have defects in their mismatch MMR genes. If a MMR defect was in one of their chromosomes, it predisposed them to developing cancer when a MMR mutation occurred in the sister chromosome (Baretti and Le, 2018). As long as the MMR genes in one chromosome were ok, adequate MMR function remained. But if MMR mutations existed in both sister chromosomes, then MMR was defective. 502 K. W. Kohn Drugs Against cancer CHAPTER 25 A (a) A-A-A-A-A-AA-A-A-A-A T-T-T- T-T -T- T-T-T- - A-A-A-A-A-A-A-A-A-A-A (b) T-T-T- T-T -1\ /J -T-T- - T Figure 25.8. Deletion (a) or insertion (b) of a nucleotide - such as A or Tin a string of repeats of the same nucleotide. This happens when the DNA polymerase slips forward or backward as it copies. In this example, the polymerase is copying a string of A's to produce a sting of the complementary nucleotide T (red). Her editary non-polyposis colon cancer (HNPCC). Did colon cancer sometimes run in families? This was clearly the case for a type of cancer where family members developed hundreds of pre-cancerous polyps in the colon. The gene responsible for this type of cancer -APC, familial adenopolypsis coli - was identified in 1991. However, in 1993, another type of colon cancer that did not arise in polyps was shown to have a genetic origin. This type of familial disposition for colon cancer (Lynch syndrome) was several times less frequent than the APC type and was found to be caused by a defect in DNA mismatch repair (Nicolaides et al., 1994; Papadopoulos et al., 1994; Peltomaki et al., 1993). Although the most common colon cancer arises in polyps in the descending (left side) colon, HNPCC is quite different. It does not arise in polyps and it occurs in the ascending colon which is on the right side of the body. The cancers arising in that part of the intestine, the HNPCC cancers, were usually caused by a defect in one of the genes that carry out DNA mismatch repair (MMR) - mismatch is when there are base-pairs in DNA that don't have the usual A-Tor G-C match. MMR gene defects are often inherited, as in Lynch syndrome, but also can sometimes happen when a DNA polymerase inserts the wrong base or when there is chemical damage to a base. If a mismatch is not repaired, then replication of the mismatch will cause mutation and may eventually lead to cancer (Kunkel and Erie, 2005, 2015). Expression of MMR genes is coupled to DNA replication - the genes are most active during S and G2 phases of the cell cycle, which is when they would be most needed (Kunkel and Erie, 2015). The MMR gene story had already begun in 1993, when Bert Vogelstein, Richard Kolodner and their colleagues isolated a gene in chromosome 2 that was mutated in patients who 503 K. W. Kohn Drugs Against cancer CHAPTER 25 had HNPCC or microsatellite instability. They found that the nucleotide sequence of the gene resembled the sequence of a MMR gene, namely MSH2, of microorganisms. They inferred that the mutation and the microsatellite instability were causally related to the disease (Fishel et al., 1993; Leach et al., 1993). A defective MMR gene in a single chromosome would not by itself cause trouble, provided that the corresponding gene in the other chromosome was normal. However, the defect in one chromosome made people vulnerable to develop cancer later if DNA damage or a replication error resulted in the MMR defect in both chromosomes. That is why people with Lynch syndrome did not develop cancer until later in life. The MMR defect in colon cancers of the HNPCC type were most often caused by mutation of MLHl or MSH2 - which conferred a lifetime risk of colon cancer by as much as 80%. Germline mutations in MSH2 and MLH l accounted for approximately 60% of HNPCC, although nearly one-third of HNPCC patients had a MMR gene that was silenced epigenetically, for example by DNA methylation of the promoter region of the gene. Non- inherited inactivation of the mismatch-repair gene MLH l happened in approximately 15% of patients by suppression of the gene by epigenetic methylation of DNA. Mutations of microsatellite repeat length were extraordinarily common in MMR-deficient cancers. Such cancer cells had thousands of microsatellite mutations, and the presence of this microsatellite instability strongly indicated that the cells were cancerous (Modrich and Lahue, 1996). Treatment of metastatic colon cancer Once colon cancer has spread to distant metastases, surgery was no longer an option. Chemotherapy was then able to extend the survival of many of the patients. Particularly effective were 5-tluorouracil combined with oxaliplatin, irinotecan and/or capecitabine. Further progress then used monoclonal antibodies to target epidermal growth factors in patients whose cancers were driven by overexpression of these receptors. The first effective monoclonal antibody was cetuximab (see Chapter 17), and the responses of many of the patients was enhanced by adding irinotecan to the treatment (Figure 25.9.). Cetuximab, however, combined a human antibody chain with a mouse-derived antigen- recognition part. The mouse part sometimes produced toxicity by causing an immune response. That problem was solved by creating a fully humanized monoclonal antibody, panitumumab, which replaced cetuximab (Xie et al., 2020). 504 K. W. Kohn Drugs Against cancer CHAPTER 25 100 ., * .,- ~ C 75 u.. .2 -.,"' .,"'"' C ·,p 0t>0 ~ 50 Cetuximab plus irinoteca n "' ....~ 25 0.. 0.. Cetuximab'• ···a .. ··-~ 0 ----------·-.. 0 0 2 4 6 8 10 12 Time to Progression {months) Figure 25.9. A monoclonal antibody, cetuximab, produced responses in about half of colon cancer patients who had previously failed to respond to chemotherapy that included irinotecan, a camptothecin-related drug. Adding irinotecan to cetuximab produced even better responses (Cunningham et al., 2004). lmmunotherapy of mismatch repair-deficient cancers. Mismatch repair-deficient cancers, such as many cancers of the ascending colon, acquired exceptionally large numbers of mutations due to microsatellite instability that often caused frameshift mutations. These cancers therefore produced many structurally abnormal proteins, which the protein-degrading machinery in the cell would break down into fragments that the cell displayed on the cell surface. Cells of the immune system recognized the abnormal protein fragments and acted to kill the cells that produce them. Mismatch repair-deficient cancer cells were particularly vulnerable to attack by the immune system because of the large number of abnormal protein fragments they displayed. Cells however have a protective system that limits the immune process so that it does not kill too many normal cells. This is a delicate balance between a cell killing system and a protective system. lmmunotherapy was developed to push the balance in favor of killing cancer cells. For that purpose, monoclonal antibodies were developed that blocked the molecules that inhibited the cell killing action of the immune system's killer T-cells. The first discovered natural inhibitor of the cell-killing action of the immune system's killer T-cells was PD-1, although its action was at first misunderstood. PD-1 was initially isolated as an immunoglobulin-related gene whose production was increased in cells undergoing programmed cell death, which was how it derived its name (Ishida et al., 1992). Before the action of PD-1 was understood, a monoclonal antibody targeting a related T-cell inhibitor, CTLA-4, was found to produce a few good responses in metastatic melanoma (Phan et al., 2003). Monoclonal antibodies targeting PD· 1 were then tested in patients who had several types of metastatic cancer, and there were some good responses (Brahmer et al., 2010; Brahmer et al., 2012). 505 K. W. Kohn Drugs Against cancer CHAPTER 25 Since mismatch repair deficiency produced large amounts of abnormal protein fragments that the immune system would recognize, a large group of researchers from several cancer centers conducted a phase-2 study to evaluate the clinical activity ofpembrolizumab, a monoclonal antibody targeting PD-1 (Le et al., 2015). The idea was that mismatch repair deficient cancers would potentially engender a strong immune response that would kill the cancer cells, but that PD-1 would block this potentially therapeutic action. PD-1 was known to bind receptor molecules, PD-Ll, on the cancer cell surface and thereby to inhibit the immune response. Moreover, the cancer cells produce lots of PD-Ll to evade the immune system. The monoclonal antibody would bind PD-1 so that it could not bind PD-Ll. The immune system would then be more free to act against the cancer. The researchers focused their attention on colon cancer, because cancers of the ascending colon often were notably deficient in mismatch repair. They cited a study in which 1 of 33 colon cancer patients had a good response to the PD-1 monoclonal antibody and asked what was different about that one responding patient. They thought that patient might have been the only one in the group that had a mismatch repair deficient cancer. The 33 patients would have included the most common colon cancers that arise in polyps in the descending colon and rarely have mismatch repair deficiency. That would explain why there was only one responding patient out of 33: the responding patient might have had a mismatch repair deficient cancer in the ascendin9 colon. They therefore investigated whether mismatch repair deficiency would make cancers more responsive to a monoclonal antibody, pembrolizumab, that binds PD-1 on T-cells and blocks its binding to its ligands, PD-Ll or PD-L2, on the cancer cells (Le et al., 2015). The expected response of mismatch repair-deficient colorectal cancers was indeed clearly seen (Figure 25.10). Good responses were also seen in other mismatch repair-deficient cancers. 1.0 P<0.001 by log,rank test P""0,03 by log-rank test -;; -~~ 0.8 ~ VI Mismatch repair-deOcient Mismatch repair-deficient 1! 0.6 " > 0 'o 0.4 & :;; M ismatch repair-pro ficient ~ 0.2 ..0 0.2 0.0•+ -- '-----~ ~ - -~ - Mismatch repair-proficient .........- -~ - ~ l 0.0 0 3 6 9 12 0 ) 6 9 12 IS Months Months Figure 25.10. Patients with progressive metastatic colorectal cancer were treated with pembrolizumab, a monoclonal antibody that binds PD-1 and prevents its binding to PD-Ll or PD-L2. Deficiency in mismatch repair was associated with much longer progression-free survival (left) and better overall survival (ri9ht) (Le et al., 2015). 506 K. W. Kohn Drugs Against cancer CHAPTER 25 In addition to PD-1, another natural inhibitor, CTLA-4, was discovered that T-cells display on their surface to limit their own cell-killing function. CTLA-4 interacts with and inhibits cells of the immune system that activate the T-cell's cell-killing function. To enhance the T- cell's capacity to kill cancer cells, CTLA-4 inhibiting monoclonal antibodies were developed as potential therapeutics in addition to the monoclonal antibodies directed against PD-1 and PD-Ll/2. A next step was to test whether adding a CTLA-4 directed monoclonal antibody, ipilimumab, to a PD-1 directed monoclonal antibody, nivolumab, would enhance the therapeutic effect of the latter against mismatch repair-deficient metastatic colon cancer (Overman et al., 2018; Overman et al., 2017). The combination of these monoclonal antibodies had already been approved for treatment of metastatic melanoma. The results showed that the combination indeed gave better clinical results than nivolumab alone against mismatch repair-deficient metastatic colon cancers (Figure 25.11). The monoclonal antibodies offered promising new treatments for patients with DNA mismatch repair-deficient cancers. However, the cancers sometimes developed resistance to the treatments, and this problem remained to be solved (Thomas et al., 2020). 100 90 80 70 - ;::;,:: 00 en LL. 40 60 50 a. 30 20 -<>- Nivolumab 10 - - Nivolumab + ipilimumab 0 3 6 9 12 15 18 21 24 27 Time (months) Figure 25.11. Durable responses of metastatic colon cancers to treatment with monoclonal antibodies. The cancers were mismatch repair-deficient and had high microsatellite instability. Adding ipilimumab, a CTLA-4 monoclonal antibody, to nivolumab, a PD-1 monoclonal antibody, improved the progression-free survival (PFS) of the patients (Overman et al., 2018; Overman et al., 2017). 507 K. W. Kohn Drugs Against cancer CHAPTER 25 From the European Society for Medical Oncology in Medscape Oncology of a remarkable success in a clinical trial, reported in 2022, of ipimumab plus nivolumab before surgery in patients with DNA -mismatch-repair-deficient colon cancer: PARIS - "Unprecedented" pathologic responses were seen after a neoadjuvant 4- week course of ipi/imumab (Yervoy) plus nivolumab (Opdivo) was given before surgery to patients with DNA mismatch repair deficient (dMMR) colon cancer. say researchers reporting new results from the NICHE-2 trial. The trial involved 112 patients with dMMR colon cancer who were given one cycle of low-dose ipilimumab and two cycles of nivolumab followed by surgery. The results show that 95% of patients had a major pathologic response (MPR) and 67% had a pathologic complete response (pCR) to immunotherapy. To date. none of these patients have had disease recurrence after a median follow-up of 13. 1 months. Study presenter Myriam Chalabi. MD. an oncologist at the Netherlands Cancer Institute, Amsterdam, described the findings as •unprecedented," especially as many of the patients had stage 3 and high-risk disease, and the expected disease recurrence rate with standard-of-care adjuvant chemotherapy in these patients would usually have been around 15%. "Importantly, this treatment was very well-tolerated,• she added. Chalabi presented the new results here during a presidential session at the European Society for Medical Oncology (ESMO) Annual Meeting 2022, held in Paris, France. Neoadjuvant immunotherapy "has the potential to become standard of care• in these patients, she said, adding that the "future has never been brighter" for dMMR colon cancer. Around 10%-15% of colon cancers are dMMR, and around 33% of these are associated with Lynch syndrome. she noted. She also urged pharmaceutical companies to seek approval for immunotherapy in this patient population, to warm applause from the audience. Commenting on the results, Andres Cervantes, MD, PhD, professor of medicine at the University of Valencia, Spain, said in an ESMO press release that the "innovative" study "questions the need for surgery and postoperative chemotherapy in all patients in whom the primary tumor has disappeared." He observed that adjuvant chemotherapy has remained standard of care, "despite the fact that chemotherapy is not so active and a complete disappearance of the tumor in the surgical specimen is not observed. Overall, Cervantes said that dMMR status is a "strong predictor of the positive effect observed with this short-course immunotherapy," adding that "determining dMMR can be easily done by immunohistochemistry in the conventional pathology lab, without the need for complex molecular testing.• 508 K. W. Kohn Drugs Against cancer CHAPTER 25 The "minimal toxicity" seen in the study "may also facilitate the implementation of this strate9y, potentially sparin9 patients from sur9ery. • •· Reported by Liam Davenport on September 11, 2022. References Baretti, M ., and Le, D.T. (2018). DNA mismatch repair in cancer. Pharmacology & therapeutics 189, 45-62. Bhattacharyya, N.P., Ganesh, A., Phear, G., Richards, B., Skandalis, A., and Meuth, M. (1995). Molecular analysis of mutations in mutator colorectal carcinoma cell lines. Hum Mo) Genet 4, 2057-2064. Boland, C.R., and Lynch, H.T. (2013). The history of Lynch syndrome. Fam Cancer 12, 145· 157. Brahmer, J.R., Drake, C.G., Wollner, I., Powderly, J.D., Picus, J., Sharfman, W.H., Stankevich, E., Pons, A., Salay, T.M., McMiller, T.L., et al. (2010). Phase I study of single-agent anti· programmed death•l (MDX-1106) in refractory solid tumors: safety, clinical activity, pharmacodynamics, and immunologic correlates. Journal of clinical oncology: official journal of the American Society of Clinical Oncology 28, 3167-3175. Brahmer, J.R., Tykodi, S.S., Chow, L.Q., Hwu, W.J., Topalian, S.L., Hwu, P., Drake, C.G., Camacho, L.H., Kauh, J., Odunsi, K., et al. (2012). Safety and activity of anti·PD·Ll antibody in patients with advanced cancer. The New England journal of medicine 366, 2455-2465. Brown, T.C., and Jiricny, J. (1988). Different base/base mispairs are corrected with different efficiencies and specificities in monkey kidney cells. Cell 54, 705-711. Chan, E.M., Shibue, T., McFarland, J.M., Gaeta, B., Ghandi, M., Dumont, N., Gonzalez, A., McPartlan, J.S., Li, T., Zhang, Y., et al. (2019). WRN helicase is a synthetic lethal target in microsatellite unstable cancers. Nature 568, 551-556. Cunningham, D., Humblet, Y., Siena, S., Khayat, D., Bleiberg, H., Santoro, A., Bets, D., Mueser, M., Harstrick, A., Verslype, C., et al. (2004). Cetuximab monotherapy and cetuximab plus irinotecan in irinotecan-refractory metastatic colorectal cancer. The New England journal of medicine 351, 337-345. Eshleman, J.R., and Markowitz, S.D. (1996). Mismatch repair defects in human carcinogenesis. Hum Mo) Genet 5 Spec No, 1489-1494. Fishel, R., Lescoe, M.K., Rao, M.R., Copeland, N.G., Jenkins, N.A., Garber, J., Kane, M., and Kolodner, R. (1993). The human mutator gene homolog MSH2 and its association with hereditary nonpolyposis colon cancer. Cell 75, 1027-1038. Hsieh, P., and Zhang, Y. (2017). The Devil is in the details for DNA mismatch repair. Proceedings of the National Academy of Sciences of the United States of America 114, 3552-3554. Ionov, Y., Peinado, M.A., Malkhosyan, S., Shibata, D., and Perucho, M. (1993). Ubiquitous somatic mutations in simple repeated sequences reveal a new mechanism for colonic carcinogenesis. Nature 363, 558-561. 509 K. W. Kohn Drugs Against cancer CHAPTER 25 Ish ida, Y., Agata, Y., Shibahara, K., and Honjo, T. (1992). Induced expression of PD-1, a novel member of the immunoglobulin gene s uperfamily, upon programmed cell death. The EMBO journal 11 , 3887-3895. Kadyrov, F.A., Genschel, J., Fang, Y., Penland, E., Edelmann, W., and Modrich, P. (2009). A p ossible mechanism for exonuclease 1-independent eukaryotic mismatch repair. Proceedings o f the National Aca demy of Sciences of the United States of America 106, 8495-8500. Kohn, K.W., and Bohr, V.A. (2001). Genomic Instability and DNA Repair. In Cancer Handbook (Nature Publishing Group), pp. 85-104. Kunkel, T.A., and Erie, D.A. (2005). DNA mismatch repair. Annual review of biochemistry 74, 681-710. Kunkel, T.A., and Erie, D.A. (2015). Eukaryotic Mismatch Repair in Relation to DNA Replication. Annu Rev Genet 49, 291-313. Le, D.T., Durham, J.N., Smith, K.N., Wang, H., Bartlett, B.R., Aulakh, L.K., Lu, S., Kemberling, H., Wilt, C., Luber, B.S., eta/. (2017). Mismatch repair deficiency predicts response o f solid tumors to PD-1 blockade. Science 3 57, 409-413. Le, D.T., Uram, J.N., Wang, H., Bartlett, B.R., Kemberling, H., Eyring, A.D., Skora, A.D., Luber, B.S., Azad, N.S., Laheru, D., et al. (2015). PD-1 Blockade in Tumors with Mismatch- Repair Deficiency. The New England journal o f medicine 372, 2509-2520. Leach, F.S., Nicolaides, N.C., Papadopoulos, N., Liu, B., Jen, J., Parsons, R., Peltomaki, P., Sistonen, P., Aaltonen, L.A., Nystrom-Lahti, M., et al. (1993). Mutations o f a mutS homolog in hereditary nonp olyposis colorectal cancer. Cell 75, 1215-1225. Lindahl, T. (1982). DNA repair enzymes. Annual review of biochemistry 51, 61-87. Modrich, P. (1991). Mechanisms and biologica l e ffects of mismatch repair. Annu Rev Genet 25, 229-253. Modrich, P., and Lahue, R. (1996). Mismatch repair in replication fidelity, genetic recombination, and ca ncer biology. Annual review of biochemistry 65, 101-133. Nicolaides, N.C., Papa dopoulos, N., Liu, B., Wei, Y.F., Carter, K.C., Ruben, S.M., Rosen, C.A., Haseltine, W.A., Fleischmann, R.D., Fraser, C.M., et al. (1994). Mutations of two PMS homologues in hereditary nonpolyposis colon ca ncer. Nature 371 , 75-80. Overman, M.J., Lonardi, S., Wong, K.Y.M., Lenz, H.J., Gelsomino, F., Aglietta, M., Morse, M.A., Van Cutsem, E., McDermott, R., Hill, A., et al. (2018). Durable Clinical Benefit With Nivolumab Plus lp ilimumab in DNA Mismatch Repair-Deficient/ Microsatellite Instability-High Metastatic Colorectal Cancer. Journal of clinical oncology: official journal o f the America n Society of Clinical Oncology 36, 773-779. Overman, M.J., McDermott, R., Lea ch, J.L., Lonardi, S., Lenz, H.J., Morse, M.A., Desai, J., Hill, A., Axelson, M., Moss, R.A., eta/. (2017). Nivolumab in patients with metastatic DNA mismatch repair-deficient or microsatellite instability-high colorectal cancer (CheckMate 142): an open-label, multicentre, phase 2 study. The Lancet Oncology 18, 1182-1 191. Papadopoulos, N., Nicolaides, N.C., Wei, Y.F., Ruben, S.M., Carter, K.C., Rosen, C.A., Haseltine, W.A., Fleischmann, R.D., Fraser, C.M., Adams, M.D., et al. (1994). Mutation of a mutL homolog in hereditary colon cancer. Science 263, 1625-1629. Peltomaki, P., Aaltonen, L.A., Sistonen, P., Pylkkanen, L., Mecklin, J.P., Jarvinen, H., Green, J.S., Jass, J.R., Weber, J.L., Leach, F.S. et al. (1993). Genetic mapp ing o f a locus p redisposing to h uman colorectal cancer. Science 260, 810-812. 510 K. W. Kohn Drugs Against cancer CHAPTER 25 Phan, G.Q., Yang. J.C., Sherry, R.M., Hwu, P., Topalian, S.L., Schwartzentruber, D.J., Restifo, N.P., Haworth, L.R., Seipp, C.A., Freezer, L.J., et al. (2003). Cancer regression and autoimmunity induced by cytotoxic T lymphocyte-associated antigen 4 blockade in patients with metastatic melanoma. Proceedings of the National Academy of Sciences of the United States of America 100, 8372-8377. Picco, G., Cattaneo, C.M., van Vliet, E.J., Crisafulli, G., Rospo, G., Consonni, S., Vieira, S.F., Rodriguez, LS., Cancelliere, C., Banerjee, R., et al. (2021). Werner Helicase Is a Synthetic-Lethal Vulnerability in Mismatch Repair-Deficient Colorectal Cancer Refractory to Targeted Therapies, Chemotherapy, and lmmunotherapy. Cancer Discov 11, 1923-1937. Pluciennik, A., Dzantiev, L., Iyer, R.R., Constantin, N., Kadyrov, F.A., and Modrich, P. (2010). PCNA function in the activation and strand direction of MutLalpha endonuclease in mismatch repair. Proceedings of the National Academy of Sciences of the United States of America 107, 16066-16071. Radman, M ., Matic, I., Halliday, J.A., and Taddei, F. (1995). Editing DNA replication and recombination by mismatch repair: from bacterial genetics to mechanisms of predisposition to cancer in humans. Philos Trans R Soc Lond B Biol Sci 347, 97-103. Thomas, J., Leal, A., and Overman, M.J. (2020). Clinical Development of lmmunotherapy for Deficient Mismatch Repair Colorectal Cancer. Clin Colorectal Cancer 19, 73-81. Wildenberg. J., and Meselson, M. (1975). Mismatch repair in heteroduplex DNA. Proceedings of the National Academy of Sciences of the United States of America 72, 2202-2206. Xie, Y.H., Chen, Y.X., and Fang. J.Y. (2020). Comprehensive review of targeted therapy for colorectal cancer. Signal Transduct Target Ther 5, 22. 511 K. W. Kohn Drugs Against cancer CHAPTER 26 Chapt,er-Z6. Th~ BRCA and bomok>gousrttDfflbinorkm ttory Z2 0ll61Jp3 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@nih.gov CHAPTER26 The BRCA and homologous recombination story. An old suspicion had it that genes or their mutations may somehow cause cancer, in particular breast cancer. In the 1990's, new methods made it possible to tackle the question. There was reason to focus on breast cancer. This tragically common cancer occasionally occurred in families, suggesting that some family members might have inherited a gene that was causing the cancer, and it might be possible to identify the gene. The story actually goes back to 1866, when the distinguished French surgeon, Pierre Paul Broca (Figure 26.1), noted that a surprisingly large number of members of his extended family had died of breast cancer [Broca, Traite des tumeurs (1866)] (Krush, 1979). He compiled the cause of death of all 38 members of five generations of his family between 1768 and 1856. Ten of the 24 women died ofbreast cancer. Since then, many family histories were reported in which breast cancer was abnormally frequent (Papadrianos et al., 1967). An example of a breast cancer family tree is shown in Figure 26.2. Interestingly, the causative genes were given the name BRCA as contraction for ''hI:east,!;a,ncer," but could equally well refer to the discoverer of breast cancer families, Broca. Researchers suspected that, when breast cancer occurred in several members of a family, a rogue gene may be lurking among the family members that made them prone to developing breast cancer. Also remarkable and significant was that familial breast cancer tended to develop in unusually young women (Figure 26.2). There was strong incentive to find and identify the gene, because drugs might then be developed to block the cancer-causing action of the gene. 512 K. W. Kohn Drugs Against cancer CHAPTER 26 Figure 26.1. Pierre Paul Broca (1824-1880), a French surgeon and scientist who is best known for his research on an area of the brain involved with language: Broca's area. Less known is that he accumulated evidence for a hereditary factor in cancers. His wife had a family history of breast cancer, which piqued his interest in exploring possible hereditary causes of cancer. The relevant breast cancer genes were to become known as BRCA - for 'breast cancer' and/or 'Broca' (Krush, 1979). (Source: Wikipedia; public domain.) Fam ily S 8olol. t 8ilot. Figure 26.2. Example of a breast cancer-prone family among several reported before 1967 (Papadrianos et al., 1967). Black symbols are individuals who had breast cancer. The numbers are the age of onset of the cancer or age of death. Circles were women; squares were men. The family tree shows that the grandfather died at 63, and the grandmother died at 69. Neither of them had cancer, but at least one of them, probably the grandfather, must have carried of the breast cancer causative gene. Of their 5 daughters, 2 had breast cancer, one of them at the early age of 47. Of their 11 granddaughters, 6 had breast cancer, which began between 31 and 47 years of age; 2 of them developed cancers in both breasts. 513 K. W. Kohn Drugs Against cancer CHAPTER 26 Years later, new methods were developed that unveiled the mutated BRCA genes that increased breast cancer susceptibility in a given family. This was an example of how focus on the right question yielded important answers. The question was whether a particular gene was associated with breast cancer in families that were prone to the disease. The target of the investigation, familial breast cancer, was rare, but the discovered causative genes were later found to be mutated in many cases of the common non-familial breast cancer, as well as some other cancers, and therapy was developed to block the biochemical effects of a mutated or otherwise overactive BRCA gene. Moreover, breast cancer-prone families often had family members with cancer of the uterus. Were there genes that made women prone to both types of cancer? If so, what was wrong with the causative genes? Probably, it was abnormal in some way. Maybe it was a mutated gene that was passed down, inherited, in the family. That may have been how researchers were thinking. Identifying the mutated gene was important, because a cell in a normal breast may sometime during the life of the woman acquire a mutation in that gene. The mutation could be produced by background radiation or cancer-causing chemicals, or even by errors during the replication of the part of the DNA that had the gene in it The cell with the mutated gene could be a first step that, together with other gene changes accumulating in the cell over the years, would eventually cause the cell to multiply uncontrollably and form a malignant cancer (Hall et al., 1990). The mutated gene, once identified, could be a target for attack at the molecular level as therapy for the cancer. Those speculations turned out be correct. There were mutated cancer-causing genes in the breast cancer-prone families. And the same genes were found to be associated with cancer of the uterus. Of even more importance, the same mutated genes were often the culprits in the common non-inherited cases (Easton et al., 1993a; Easton et al., 1993b; Hall et al., 1990). Thus, investigating the rare familial cases led to basic knowledge that became relevant to therapy for many patients with the common cancers. How the BRCA genes wer e discover ed. In their investigations of breast cancer-prone families, researchers started looking at the chromosomes of family members who developed breast cancer. They tried to pin down where in the chromosomes a gene associated with the cancer might be located. Long and painstaking effort was made to locate a difference in the chromosomes of family members who did or did not develop the cancer. You might ask why that chromosome search could not be done just as well in breast cancers occurring in the general population. Why focus on the rare cancer-prone families? The answer is that there were several different genes that led to the cancer. But in the cancer-prone families there likely would be a single particular gene that was associated with the cause in that family. Indeed, there were different breast cancer-prone families that had different cancer-associated genes. That made it possible to track down the particular gene that caused the cancer in a particular family. Once a single causative gene was identified and its DNA sequence determined, it became possible to search for related genes by sequence similarity. 514 K. W. Kohn Drugs Against cancer CHAPTER 26 The first success in the search was in a breast cancer-prone family where the genome difference pointed to a narrow region of chromosome 17. Perhaps there was a breast cancer-associated gene located in that region? Indeed, this presumption was correct, and the offending gene was named 'BRCAl ' (Easton et al., 1993a; Easton et al., 1993b; Hall et al., 1990). The BRCAl gene itself was soon identified, its DNA nucleotide sequence was determined, and the cancer-causing changes in the mutated genes were revealed (Miki et al., 1994). Not all familial breast cancers, however, were associated with BRCAl mutations. Another mutated breast cancer-causing gene, found in a different family, was located in chromosome 13 and was named 'BRCA2'. As in the case of BRCAl, the DNA nucleotide sequence changes in the mutated genes were soon determined (Wooster et al., 1995). Although BRCAl and BRCAZ were occasionally mutated in other cancers, I don't know why they were most often found in cancers of breast and ovary. What do BRCA1 and BRCA2 do? So, what is it about those BRCA gene mutations that incites cancer to erupt? Somehow, the normal versions of those genes protect against the development of cancer. Accordingly, the BRCA genes were considered to be "tumor suppressors." But what do the normal BRCA genes do that prevents cancer? The first clue to the function of BRCAl came inl 996 from Ralph Scully and coworkers in David Livingston's laboratory at the Dana Farber Cancer Institute of Harvard Medical School in Boston. They found that the BRCAl protein (the product of the BRCAl gene) binds to a protein, RadSl, that was known to be part of the machinery that repairs radiation-damaged DNA (Scully et al., 1997). Moreover, they made the remarkable observation that BRCAl is seen in discrete spots ("foci") in the cell nucleus - but only in cells that were replicating their DNA (in "S-phase" of the cell division cycle) (Figures 26.3 and 26.4). The RadSl protein became localized in the same spots as the BRCAl protein. The BRCAl-RadSl protein pair evidently did something important during S-phase at those particular spots in the nucleus, perhaps at the very places in the chromosomes where damaged DNA was being repaired. A few months later, similar observations were reported for BRCA2 (Kinzler and Vogelstein, 1997; Sharan et al., 1997). Moreover, BRCAl, BRCA2, and RadSl were bound all together in the same spots in the S-phase cell nucleus. In addition to those 3 proteins, several others, including proteins associated with Fanconi's anemia, were found in the same complex, and most of those proteins were known to function in the repair of damaged DNA (Chapter 31 will tell about the Fanconi anemia DNA repair system). All of those proteins (gene products) seemed to work together in a critically important DNA repair process. The spots where they were co-localized within the cell nucleus were perhaps the locations in the genome where DNA damage was being repaired. If the BRCAl or BRCAZ gene were inactivated by mutation, DNA repair might be impaired and DNA damage would accumulate, increasing the chance that a cell would become cancerous. 515 K. W. Kohn Drugs Against cancer CHAPTER 26 Figure 26.3. The BRCAl protein localized in spots ("foci") in the nuclei, but only in cells that were in the process of replicating their DNA (S-phase of the cell division cycle). BRCAl lights up in bright fluorescent spots. Breast cancer cells (MCF7) were synchronized, so that few cells were in S-phase (left), and later when most cells were in S-phase. Only the cell nuclei are visible in these images. From (Scully et al., 1997). Figure 26.4. An S phase cell nucleus co-stained for the BRCAl protein (green fluorescence, left) and for the Rad51 protein (red fluorescence, center). We see the BRCAl and Rad51 proteins localized in the same spots in the nucleus. When BRCAl's green and RadSl's red fluorescence were in the same spot in the nucleus, the color was white (right). From (Scully et al., 1997). The BRCA proteins in DNA r epair. It seemed that BRCAl and BRCA2 may function in one or more steps in the cell's DNA repair pathways. But, which steps? A clue came from Rad51, whose function in DNA repair had already been partially elucidated: it functioned in a remarkable DNA repair process based on homologous recombination. Since BRCAl and BRCA2 were bound together with Rad51, it seemed plausible that all three might function together in a homologous recombination process in DNA repair. So ... what is "homologous recombination"? 516 K. W. Kohn Drugs Against cancer CHAPTER 26 Homologous recombination Homologous recombination is like a conjurer's trick. It functions both in chromosome crossover during meiosis and in DNA repair. I will focus on DNA repair and the role of BRCAl and BRCA2 in homologous recombination. But to understand it and its history, I will recount the story of its discovery in studies of meiosis and "jumping genes". But before relating those stories of discovery, here is an introductory summary of how homologous recombination works and how the BRCA's became implicated: When the growing point of a replicating DNA strand encounters damage in the template strand that it is copying, it can -- quite remarkably -- switch to copying a different DNA strand that is "homologous" (having the same or a very similar DNA sequence) to the damaged strand. "Homologous" means that the target DNA strand has nearly the same nucleotide sequence as the damaged strand had before it was damaged. In effect, the replicating strand that encounters a blockage of the template strand it's trying to copy says, "Ok, forget it, I'll switch over to copying this other guy who's just as good as you were before you got damaged!". But how could a blocked growing strand find and switch over to copying a homologous strand in another DNA duplex? For that caper to work, first of all there must be a homologous strand nearby for the replicating strand to find and switch to. That becomes feasible when the DNA duplexes in a chromosome are being replicated, because the two DNA duplex pairs (the two daughters produced by Watson and Crick's "semi-conservative" replication) remain near each other: they remain connected at their "centromeres" until the chromosomes separate in the latter part of mitosis. The strands of a damaged duplex therefore can find the needed undamaged section in the homologous newly replicated DNA duplex, because it lies nearby. That is why homologous combination is only effective during replication (S-phase) or between the end ofS-phase and mitosis (G2 phase). (A "newly replicated DNA duplex" consists of a preexisting ("conserved") strand and a newly replicated strand, i.e., the replication is "semi-conservative", which is the essence of Watson and Crick's discovery in 1953, proven experimentally by Meselson and Stahl in 1958.) The role of the BRCAs in DNA repair was further highlighted by the remarkable observation that a complex of BRCAl, BRCA2, RAD51, as well as some other proteins, was seen at sites in the nucleus where DNA damage was being repaired; the sites ("foci") were seen as spots in the nucleus made visible by fluorescent tags on one or another of the proteins in the complex (Garcia-Higuera et al., 2001), as seen in Figures 26.3 and 26.4. This complex of proteins was thought likely to act in DNA repair steps in which homologous recombination steps comes into play and where BRACAl and BRCA2 are needed. By the way, BRCAl and BRCA2 also function the in the repair of DNA crosslinks by the Fanconi repair system (Chapter 31). But now let's look back to the story of how homologous recombination was discovered. 517 K. W. Kohn Drugs Against cancer CHAPTER 26 How homologous recombination came to be discovered. Homologous recombination is a remarkable process whereby genetic traits are assorted during meiosis among offspring and produces the variety of progeny for selection that enables evolution. We are interested in homologous recombination, because of its importance in repairing DNA damage. Moreover, cancer cells often are defective in their ability to repair DNA damage, which is one reason that anticancer drugs work The concept ofrecombination between two homologous chromosomes goes back to Thomas Hunt Morgan's famous work on the genetics of fruit flies. Morgan (Figure 26.5) proposed that genes are arranged on chromosomes like beads on a string, and that sometimes there is "crossing over" between chromosomes giving rise to recombination of the genes, as shown in Morgan's drawing of the concept in 1916 (Figures 26.6). Then in 1930, Barbara Mcclintock (Figure 26.7) obtained a direct image of chromosomes in the process of exchanging their associated pairs, a prelude to homologous recombination (Mcclintock, 1930) (Figures 26.8). Figure 26.5. Drawing of Thomas Hunt Morgan (1866-1945), discoverer of the phenomena of chromosome crossing over and genetic linkage in 1913. (Created 31 December 1930; Wikipedia, public domain.) 518 K. W. Kohn Drugs Against cancer CHAPTER 26 Figure 26.6. Thomas Hunt Morgan's illustration of his concept of crossing over between homologous chromosomes and their rows of genes from his pioneering studies of fruit fly genetics (1916) (from Wikipedia). Figure 26.7. Barbara Mcclintock (1902-1992) discovered mobile genetic elements, a concept so revolutionary at the time that it was long before geneticists accepted or understood it. Many years later after the importance of her of her discovery was grasped, she was awarded a Nobel Prize in Physiology or Medicine in 1983. She had received a PhD in Botany at Cornell University in 1927 and used maize as her subject of investigations. (From http://siarchives.si.edu/collections/siris arc 306310) 519 K. W. Kohn Drugs Against cancer CHAPTER 26 ( Figure 26.8. Barbara Mcclintock published this drawing in 1930 of chromosomes interchanging their associated partners. Switching DNA partners is an essential part of homologous recombination. The cell from which these chromosomes came was in the midst of preparing for mitosis (mid-prophase): the chromosomes were already condensed but had no spindle microtubules attached yet. The clear regions are the centromeres (one for each chromosome), where spindle microtubules will attach during mitosis. She used a "camera lucida" to project the image (magnification x1875} on a page, so as to make an accurate drawing - as was commonly done before photography through a high-power optical microscope became available (Mcclintock, 1930). Homologous recombination was first discovered in genetics, where sections of a chromosome often moves from one chromosome to another during sexual reproduction. It happens in meiosis, during the production of sex cells. During meiosis, homologous sections of sister chromatids often became interchanged . When a chromosome is replicated, each daughter chromosome is called a chromatid, and the two daughter chromosomes are called sisters. Meiosis is a bit complicated, because it entails two chromosome doublings and "reduction division." Wait! Before your eyes glaze over: You may have learned about all that in high school biology, and maybe forgotten the details. Shortly, we will review the essentials of meiosis, because it is where DNA-damaging anticancer drugs often cause trouble: it is where reproductive cells (egg and sperm) are produced, a process that is especially vulnerable to those drugs. Homologous recombination comes into play as our story unfolds, because of the role of the BRCA genes and the proteins they code for. The first clear observation of recombination by exchange of segments between different chromosomes ("recombination") came from Barbara McClintock's studies in 1930 (Mcclintock, 1930). That classic paper of 1930 demonstrated a genetic exchange between different chromosomes in the same cell (Figure 26.8), which visualized how this happens during meiosis, and foreshadowed the concept of "Holliday Junctions" whose importance in DNA repair we will see later in this chapter. Homologous recombination was also found in microorganisms and described by physicist- turned-biologist Max Delbruck in 1946 . He had observed exchange of genome sections between different strains of bacteriophage that had infected the same E. coli bacterial cell. 520 K. W. Kohn Drugs Against cancer CHAPTER 26 Delbruck wrote about the views commonly held at the time of his conversion from physics to biology in his "retrospective of 20 years as a biologist" (Delbruck, 1970). He begins by thinking back to how Aristotle viewed the world. Aristotle viewed cycles of origin, development, and demise as characteristic of living organisms, where demise at the end of one cycle leads to the origin of another. Only in the astronomy of the time did there appear to be cycles without end, with neither generation nor decay. Delbruck viewed his conversion from physics to biology as a break to a different conceptual mode: from the inanimate to the animate. His conversion occurred at a time when "life seemed to have unique properties quite irreducible to the world of physics and chemistry: motion generated from within ... aspects that were foreign to the physicist" (Delbruck, 1970). Nevertheless, the "physicist" in Delbruck showed itself in mathematical theory he developed to account for genetic recombination findings in his bacteriophage crossings (Visconti and Delbruck, 1953). This conformed with the tone of the time, (expressed by Francis Crick, as I recall) that "the objective of molecular biology is to destroy the last vestiges of vitalism." The vita list notions that dominated genetics had however begun to be overturned by Gregor Mendel in his cross-fertilization studies of different kinds of peas and other plants. He concluded that inheritance was mathematical - thus an inanimate, rather than a vitalist process. Both Mendel and Delbruck showed that mathematical ideas could apply to living organisms. Johann Mendel (1822 -1884) lived in a German-speaking part of Moravia in the Austrian Empire. As a young man, he studied theoretical philosophy and physics. But then, in order to pursue his studies free of "perpetual anxiety about a means of livelihood," he joined the Augustinian Friars in Brno, and was given the name "Gregor"; eventually, becoming Abbot. Much has been written about his plant hybridizing studies that were not comprehended by scientists of the time, and which lay obscure for several decades, before a more enlightened age rediscovered them. Barbara McClintock's work near the middle of the 20 th century then overturned some of the tenants of genetics, although her major findings took a long time to be appreciated. After much success in a remarkable variety of genetic research, mostly on plants, and recipient of multiple awards, including election to the U.S. National Academy of Science, Mcclintock embarked on investigation of the genetics of maize, which led to observations for which the scientific community was not yet ready to comprehend. She was first to figure out how to see the maize chromosomes under the microscope, which allowed her to relate changes in genetic traits with changes in the physical chromosomes. In the 1940's, however, she detected maize genes (or "genetic elements"), whose behavior made no sense from what was then known and understood about genetics. One of those genes was jumping from one place on a chromosome to another place. Genes were not supposed to be able to do that It defied the accepted Mendelian inheritance patterns that had become ingrained in scientific lore. Although many geneticists refused to accept her findings, she persisted in accumulating and assembling data. The strange genetic behavior she observed produced complicated data that were challenging to make sense of. It is said that, when she presented her findings at conferences, the audience listened politely in deference to her previous achievements, but remained silent, because they could make neither head nor tail of her complicated new data. Her meticulous studies of genetic recombination in maize 521 K. W. Kohn Drugs Against cancer CHAPTER 26 demonstrated radically non-Mendelian patterns of inheritance that were understood only after genetically mobile genes became understood, which was not until the 1960's. When her work was finally understood, the scientific world was so astounded by her "out-of-the- box" achievement that in 1983 she was awarded, unshared, the Nobel Prize in Physiology or Medicine. Earlier, after visiting her, Joshua Lederberg. himself a Nobel Prize winner in Physiology or Medicine (1958), is said to have remarked, "that woman is either crazy or a genius!" That statement was logically true; moreover, she was NOT crazy! Homologous recombination is essential to how McClintock's "jumping genes" move from one chromosome to another. It happens by way of homologous recombination and is related to DNA repair, as well as to how new therapies work that use CRSPR-CAS9 systems to engineer CAR-T cells by inserting or replacing DNA sections at key locations in the chromosomes. Roles ofBRCA1 and BRCA2 in homologous recombination. As already described earlier in this chapter, the first indication of a connection between the BRCA genes and homologous recombination came in 1997 from studies led by David M. Livingston of the Dana-Farber Cancer Institute at Harvard Medical School in Boston (Scully et al., 1997). They found BRCAl colocalized with RADSl in the same discrete foci in the nuclei of cells preparing for mitosis or undergoing meiosis (Figure 26.4). Then, in 1999 and 2001, a pair of papers in Molecular Cell in 1999 and 2001 by Mary Ellen Moynahan, and her coworkers in Maria Jasin's laboratory at the Memorial Sloan-Kettering Cancer Center in New York reported that both BRCAl and BRCA2 not only colocalized with RADSl in the same nuclear foci, but actually were physically bound in a trimer consisting of the three proteins all bound together (Moynahan et al., 1999; Moynahan et al., 2001). Moreover, cells deficient in any one of the three had increased sensitivity to agents that produced DNA double-strand breaks. Furthermore, RADSl was similar to the bacterial RECA gene that was known to be required for homologous recombination in those organisms. Similarly, RADSl was shown to be required for efficient homologous recombination in mammalian cells. Those findings set a firm basis for the conclusion that the BRCA gene played a role in homologous recombination (Chen et al., 2018). How DNA repair by homologous recombination works. Homologous recombination (HR) works by way of a complicated choreography that repairs the DNA without any errors in its nucleotide sequence. However, it is blocked in a manner that allows it to take place only during or after DNA replication. It is blocked during the Gl phase of the cell division cycle, during which the DNA remains unduplicated and there is no sister chromosome nearby. There are many different ways in which homologous recombination can occur, some of them rather complicated, and many more have been proposed over the years. The main point is that, when a DNA replication process encounters a blockage, such as by a damaged template strand, replication can proceed by 522 K. W. Kohn Drugs Against cancer CHAPTER 26 switching to copying a strand in a homologous sequence in the newly replicated sister chromosome. As often happens as a science develops, a general rule turns out to have exceptions. Thus, it turned that homologous recombination may occasionally occur even outside of S or G2 and even in non-dividing cells (GO). In 2013, researchers at the University of Pittsburg reported that homologous chromosomes in GO cells can find each other within minutes after the cells were exposed to x-ray (Gandhi et al., 2013). Moreover, the DNA-damage-induced pairing between homologous chromosomes occurred without pairing of their centromeres. Therefore, it seemed to be happening by a different mechanism than the pairing of newly replicated sister chromosomes. Also, it seemed that the genome regions that paired up were only in the RNA-transcribing genes. So, how did homologous chromosomes find each other in DNA-damaged GO cells where no newly replicated sister chromosome was available? The researchers thought it likely that the transcribed RNA was in fact what was recognizing a homologous sequence in the homologous chromosome. The idea that DNA repair by homologous recombination may occur in actively transcribing homologous regions was supported by clever experiments by investigators in other departments at the University of Pittsburg (Wei et al., 2015). (The US National Center for Biotechnology Information (NC BI), lists 20,203 protein-coding genes and 17,871 non-coding genes (genes that are transcribed into RNA that does not code for protein but has other functions). Thus, all together there may be about 38,000 RNA-transcribing genes in the human genome. That, however, comprises only a small fraction of the DNA in the chromosomes.) Getting back to how homologous recombination works, the essentials of an early step in the process was depicted in a simple diagram by Charles Radding in 1979 (Figure 26.9). The dark curve in the diagram shows an invading strand base-pairing with a homologous region of a recipient DNA. The process displaces the other strand of the recipient, forming a "D-loop" pattern. Later work disclosed that the invading strand had to be coated with RECA, the homolog of mammalian RADSl that had the same function, and both BRCAl and BRCA2 were required for that to happen. Homologous recombination became important in how cells deal with circumstances when the template DNA strand is broken. When a moving replication fork encounters a DNA single-strand break or gap, the replicating strand was found to switch to a homologous region of an undamaged chromosome, where it can base-pair and copy an undamaged strand (Figure 26.10). Further details about how homologous recombination works will be told in the next chapter that deals with its critical role in the repair of the potentially lethal DNA double-strand breaks. 523 K. W. Kohn Drugs Against cancer CHAPTER 26 Dloop Figure 26.9. The beginning of homologous recombination (HR) from part of a diagram by Charles Radding and his colleagues in 1979, which showed that the RecA protein and ATP drives the process (Shibata et al., 1979). They had worked this out by experiments using pure components. The diagram shows a single-stranded DNA segment (heavy line) invading a DNA double helix. The invading DNA single-strand base-pairs with one strand of the DNA duplex while displacing the other strand. C 3' I I I I I I I I I I I I I I I I I II II IIII II II II I A s·,,, '.,.'",'.,.'",',..'",',..'",, -+ ,,.,... ~.,.,. ,. E F -: :::::::-::::: S',...,. i i i i i i i i i i i i ~ • 3 111111II1:tlS:ii 1111!11111111111 iiiiiiiiiiiiiiii 3' I I I I I I I I I I I I I I I I ....... """"' ~ ~ ·· 1111111, :JV'\;;:::::: I' ii I I I I I I I I ii I I I I B s·.w. 111::::: II II PP II II II II PP II II II II II II PP It -+ l S' I I I I I I I I I I I I I I I I 3 r:111111111::rr: II 11I11111111111 D iliiiiiiiiliiiii 11 11 11111 11 11111 ii ii ii ii ii ii ii ii ::::::::::r:r::: G Figure 26.10. A scheme for homologous recombination triggered by a single-strand break (A) or a single-strand gap (B) in the DNA of one of the sister chromatids, as proposed by Charles Radding in 1983. Homologous recombination yields C or D, respectively, which both lead to E, then F, then finally to products such as G. (Simplified from (Vriend and Krawczyk, 2017)). References Chen, C.C., Feng, W., Lim, P.X., Kass, E.M., and Jasin, M. (2018). Homology-Directed Repair and the Role of BRCAl, BRCA2, and Related Proteins in Genome Integrity and Cancer. Annu Rev Cancer Biol 2, 313-336. Delbruck, M. (1970). A physicist's renewed look at biology: twenty years later. Science 168, 1312-1315. Easton, D., Ford, D., and Peto, J. (1993a). Inherited susceptibility to breast cancer. Cancer Surv 18, 95-113. Easton, D.F., Bishop, D.T., Ford, D., and Crockford, G.P. (1993b). Genetic linkage analysis in familial breast and ovarian cancer: results from 214 families. The Breast Cancer Linkage Consortium. American journal of human genetics 52, 678-701. 524 K. W. Kohn Drugs Against cancer CHAPTER 26 Gandhi, M., Evdokimova, V.N., Cuenco, K.T., Bakkenist, C.J., and Nikiforov, Y.E. (2013). Homologous chromosomes move and rapidly initiate contact at the sites of double- strand breaks in genes in G(O)-phase human cells. Cell Cycle 12, 547-552. Garcia-Higuera, I., Taniguchi, T., Ganesan, S., Meyn, M.S., Timmers, C., Hejna, J., Grompe, M., and D'Andrea, A.O. (2001). Interaction of the Fanconi anemia proteins and BRCAl in a common pathway. Mol Cell 7, 249-262. Hall, J.M., Lee, M.K., Newman, B., Morrow, J.E., Anderson, L.A., Huey, B., and King, M.C. (1990). Linkage of early-onset familial breast cancer to chromosome 17q21. Science 250, 1684-1689. Kinzler, K.W., and Vogelstein, B. (1997). Cancer-susceptibility genes. Gatekeepers and caretakers. Nature 386, 761, 763. Krush, A.j. (1979). Contributions of Pierre Paul Brocato Cancer Genetics. Trans Nebraska Acad Sci VII, 125-129. Mcclintock, B. (1930). A Cytological Demonstration of the Location ofan Interchange between Two Non-Homologous Chromosomes of Zea Mays. Proceedings of the National Academy of Sciences of the United States of America 16, 791-796. Miki, Y., Swensen, J., Shattuck-Eidens, D., Futreal, P.A., Harshman, K., Tavtigian, S., Liu, Q., Cochran, C., Bennett, L.M., Ding, W., et al. (1994). A strong candidate for the breast and ovarian cancer susceptibility gene BRCAl. Science 266, 66-71. Moynahan, M.E., Chiu, J.W., Koller, B.H., and Jasin, M. (1999). Brcal controls homology- directed DNA repair. Mol Cell 4, 511-518. Moynahan, M.E., Pierce, A.J., and Jasin, M. (2001). BRCA2 is required for homology-directed repair of chromosomal breaks. Mol Cell 7, 263-272. Papadrianos, E., Haagensen, C.D., and Cooley, E. (196 7). Cancer of the breast as a familial disease. Ann Surg 165, 10-19. Scully, R., Chen, J., Plug, A., Xiao, Y., Weaver, D., Feunteun, J., Ashley, T., and Livingston, D.M. (1997). Association ofBRCAl with Rad51 in mitotic and meiotic cells. Cell 88, 265-275. Sharan, S.K., Morimatsu, M., Albrecht, U., Lim, D.S., Regel, E., Dinh, C., Sands, A., Eichele, G., Hasty, P., and Bradley, A. (1997). Embryonic lethality and radiation hypersensitivity mediated by Rad51 in mice lacking Brca2. Nature 386, 804-810. Shibata, T., Cunningham, R.P., DasGupta, C., and Radding, C.M. (1979). Homologous pairing in genetic recombination: complexes of recA protein and DNA. Proceedings of the National Academy of Sciences of the United States of America 76, 5100-5104. Visconti, N., and Delbruck, M. (1953). The Mechanism of Genetic Recombination in Phage. Genetics 38, 5-33. Vriend, L.E., and Krawczyk, P.M. (2017). Nick-initiated homologous recombination: Protecting the genome, one strand at a time. DNA repair 50, 1-13. Wei, L., Nakajima, S., Bohm, S., Bernstein, K.A., Shen, Z., Tsang, M., Levine, A.S., and Lan, L. (2015). DNA damage during the GO /Gl phase triggers RNA-templated, Cockayne syndrome B-dependent homologous recombination. Proceedings of the National Academy of Sciences of the United States of America 112, E3495-3504. Wooster, R., Bignell, G., Lancaster, J., Swift, S., Seal, S., Mangion, J., Collins, N., Gregory, S., Gumbs, C., and Micki em, G. (1995). Identification of the breast cancer susceptibility gene BRCA2. Nature 378, 789-792. 525 K. W. Kohn Drugs Against cancer CHAPTER 27A Chapr,er-2 7A DNA doubk-strartd b~ k repair by homok>gous r«Dmbinorkm zz on7cv3 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@nih.gov CHAPTER27A DNA double-strand break repair by homologous recombination Wha t is a DNA double-strand break (DSB)? Although DNA single-strand breaks (SSB) production and repair were actively studied in the 1970's, studies of double-strand breaks (DSB) lagged. One reason was confusion about what qualified as a DSB as opposed to two SSB's near each other, one on each DNA stand. If both DNA strands were cleaved at precisely the same place along the double helix, that would clearly be a DSB. But how far apart on opposite strands could SSB's be to qualify as a DSB? If there were a sufficient number of base-pairs between SSB's, the DNA would only need the simpler SSB repair mechanism to restore continuity. If, on the other hand, there were insufficient base-pairs between the breaks, then the DNA would separate at the breaks (Figure 27 A.1), and the more complicated DSB-repair would be needed to restore continuity. How far apart the SSB's had to be to maintain DNA continuity depended also on the conditions used in the DSB measurement. A 8 ======== -- C - Figure 27 A.1. What is a double-strand break (DSB)? A and B qualify as DSB's, but C does not In B, the single-strand breaks (SSB's) are so close that the number of base-pairs between them is insufficient to hold the DNA together. In C, there are sufficient base-pairs between the SSB's to hold the DNA together. 526 K. W. Kohn Drugs Against cancer CHAPTER 27A What causes DNA double-s trand breaks? We are all subject to background radiation and cosmic rays, which damage DNA by producing both single- and double-strand breaks (SSB and DSB), as well as DNA base damage. In cells that are ongoing DNA replication, SSB are sometimes converted to DSB. That occurs when a replication fork, as it moves along the DNA, happens to collide with an SSB -- as shown and explained in Figure 2 7A2. The ratio of DSB relative to SSB produced by radiation depends on the type ofradiation. The most effective DSB-producing radiations are those that consist of atomic nuclei, such as a-particles. Although they have a short range of travel and do not penetrate the skin, a- particles are apt to cause DSB if they pass through or very close to DNA, as shown in Figure 27 A3. Ordinary x-rays may produce some DSB, but an equal dose of a-particle radiation would produce a much higher frequency of DSB, which are harder than SSB to repair and are likely to produce chromosome breaks. Therefore, a-particle radiation is very dangerous if it is produced within the body. From the 1930s and into the early 1950s, radiologists commonly administered a suspension of thorium dioxide, known as Thorotrast, to produce good contrast in x-ray images. Several million people received Thorotrast. Although it produced no immediate ill-effects, the thorium was slightly radioactive, emitting a-particles within the body. Years later, many who had received Thorotrast eventually developed cancers. Occasional DSB, as well as SSB, also result from reactive oxygen molecules escaping from the energy-producing electron- transfer chain in mitochondria. DNA double-strand breaks produced by chemicals and anticancer drugs. While studying DNA damage produced by various alkylating agents, Bradley and Dysart (Bradley and Dysart, 1985) obtained some surprising results. They used the alkaline and neutral filter elution methods to quantify the production of both single- and double-strand breaks (SSB and DSB) in the DNA of cells and animals (Chapter 9). The surprise was that alkylating agents that simply added methyl or ethyl groups to guanines in DNA produced DSB as well as SSB. Those methylated or ethylated guanines would have been repaired by base excision repair (BER; Chapter 24), which could transiently produce SSB at an intermediate step in the repair, but not DSB. Here is how DSB may result during the repair of methylated or ethylated guanines or adenines. In the first step, a DNA glycosylase removes the alkylated base and leaves behind a base-free sites in the DNA (Chapter 24). A base-free site in DNA is a nucleotide that has no base attached to it During the repair of a base-free site, a transient single-strand break (SSB) appears and is soon resealed. However, if a SSB is encountered by a progressing DNA 527 K. W. Kohn Drugs Against cancer CHAPTER 27A replication machine, the molecular collision is apt to produce a DSB, as suggested by (Vriend and Krawczyk, 2017) (Figure 27A.2). Other sources of DSB are DNA-binding drugs, such as bleomycin, that produce reactive molecular species close to where the drug is bound (see Chapter 13). In addition, topoisomerase II inhibitors, such as doxorubicin and etoposide, produce DSBs by blocking or perturbing the normal function of the enzyme (see Chapter 10). Moreover, topoisomerase I inhibitors, such as camptothecin, produce DNA double-strand ends when a DNA replication fork collides with a blockage where the drug is bound to the enzyme (see Chapter 11). , \/, ,,,,,,,,,1,lf!lf 'l'llll"ll'l'llll"PI "'•'l"I,..111"11'1"19'1111"11 ! ~ \ e I v·· .. Replication fork (REO) __,,. colliding with an SS8 results in a 0S8. ~""' "" II..,..... F~ collapse Figure 27 A.2. Disaster when a replication fork collides with a single-strand break (SSB). The newly replicated strands are shown in red. The newly replicated double-helix falls off and the replication of that part of the chromosome has to start over again - which prolongs the duration of S-phase. (Modified from (Vriend and Krawczyk, 2017)). 528 K. W. Kohn Drugs Against cancer CHAPTER 27A eledron;---._,__....-'I (IOW•LET) Figure 27A.3. The ratio of DSB relative to SSB produced by ionizing radiation depends on the kind of radiation used. Low energy x-rays produce slow electrons that scatter widely, because they produce low linear energy transfer (LET) in the medium through which they travel. Therefore, they produce mainly SSB, as shown in the upper part of the diagram. a- particles, on the other hand, are heavy and don't scatter as much. When they pass through or very close to DNA, they mostly damage both strands in a small neighborhood, resulting in DSB whose strand ends are staggered as shown in Figure 27 A.1, and there may also be base damage at the DSB sites (from (Iliakis et al., 2019)). Overview of how DNA double-s trand breaks (DSB) are repaired. Two remarkable mechanisms for DSB repair were discovered: homologous recombination repair (HRR) and non-homologous end joining (NHEJ) (Figure 27 A.4). The story of how homologous recombination came to be discovered was told in the previous chapter (Chapter 26), including the discovery of RADSl, BRCAl, and BRCA2. Repair by NHEJ almost always loses or occasionally gains sequences when it rejoins the ends of the DNA breaks, whereas repair by HR is usually error-free; it is nevertheless NHEJ that cells most frequently to use to repair DSB (Iliakis et al., 2019). The likely reason is that HR requires a nearby undamaged sister DNA helix from which to copy normal sequence information to replace damaged regions. A nearby sister DNA is available only late in DNA replication (S-phase ), when there are DNA sister chromosomes held together at a centromere. NHEJ, on the other hand, removes the damaged region without replacing it with normal sequence and therefore does not need a sister chromosome. DSB repair by NHEJ therefore can occur during almost any phase of the cell cycle - and can occur even in quiescent cells that are not multiplying (Critchlow and Jackson, 1998). Indeed, DNA repair by NHEJ may have evolved very early in the history of life in haploid organisms that never passed through a stage where sister chromosomes were present to permit repair by HR. Figure 27 A.4 presents an overview of how the two major DSB repair processes work The diagram and legend of the Figure is a prelude for going on to the history and mechanisms of those processes. 529 K. W. Kohn Drugs Against cancer CHAPTER 27A Non-homologous Homology-directed end Joining (NHEJ) repair (HOR) I MREII • KU70/$0 = ==81111== • + ONA·PKc,' ' 8RCA1 I ~AROI♦ = ~ -= + Figure 27 A.4. Outline of the two major double-strand break repair alternatives (from (Kaplan and Glazer, 2020)). The left side of the diagram shows the essentials of the non-homologous end joining (NHEJ) pathway, in which the first step involves binding of the Ku70-Ku80 dimer to each end of the break. In the next step, DNAPKcs is recruited to each end, which binds the two ends together. Finally, XLF and LIG4 ligate the ends together. The right side of the diagram shows the essentials of the homologous recombination repair (HRR) pathway, which was called "homology-directed repair (HDR)" in the diagram. In the first step, MREl 1 chews away part of one strand at each end, so as to produce a single-strand region long enough to invade and base-pair with a region of a sister DNA. The projecting single-strand at each end of the break becomes coated with RADSl molecules with the aid of BRCA2. The RADSl-coated single-strand at one or both ends of the break then invade a homologous region of a nearby newly replicated sister DNA. Further actions by a polymerase and ligase then restore an accurately repaired DNA. The homologous r ecombination DNA r epair (HRR) story. In 1979, Charles Radding (Figures 27A.SA and B) wrote: "Nothing is more intriguing about homologous recombination than its beginning. How, for example, do homologous double-stranded molecules recognize each other, and what enzymic events, overcoming the energetic barrier posed by the stability of duplex DNA, begin an exchange of parts?" (Shibata et al., 1979). 530 K. W. Kohn Drugs Against cancer CHAPTER 27A A diagram of homologous recombination that Radding and his colleagues came up with in 1979 is depicted in Figure 27 A.SC. It was based on experiments they had done using purified RecA, the bacterial version of mammalian RadSl. They showed that a DNA single- strand, with RecA bound to it, was able to unwind a region of DNA helix and associate with one of its strands to form the D-shaped loop in their diagram. They also showed how ATP provided the energy to drive the reaction. The RadSl gene was first isolated from yeast mutants that were unable to carry out recombination between sister chromatids. Since recombination in bacteria required RecA, it seemed that the two genes might be related to each other. Also, the RadS l and RecA proteins had similar DNA binding properties. Indeed, the two proteins had similar amino acid sequences (Figure 27A.6) (Shinohara et al., 1992). Further research confirmed that the two proteins carried out similar homologous recombination functions in their respective organisms. But what exactly did these proteins do to help the homologous recombination process? That will be part of the story of homologous recombination DNA repair that will be related in the next section. Figure 27 A.SA. Charles M. Radding (1930-2020) at a laboratory celebration. Image credit: David Keith Gonda (West and Kowalczykowski, 2021). (Radding, by the way, was a friend and classmate of mine at Harvard College and then at NIH.) 531 K. W. Kohn Drugs Against cancer CHAPTER 27A Figure 27 A.SB. The blackboard in the Radding laboratory with new ideas about recombination models and mechanisms for how RecA might drive recombination. Ima9e credit: David Keith Gonda (West and Kowalczykowski, 2021). Figure 27A.SC. A model of homologous recombination, proposed by Charles Radding and colleagues in 1979 (Shibata et al., 1979). The model was based on experiments with purified components using the bacterial RecA, which corresponds to the 532 K. W. Kohn Drugs Against cancer CHAPTER 27A mammalian RADSl. In step 1, RecA binds ATP. In step 2, RecA-ATP coats a segment of single-strand DNA (ssDNA), which starts as random coil but becomes linear when coated with RecA. In steps 3 and 4, the RecA-coated DNA single-strand binds to the recipient double-strand DNA, unwinding a section of helix as it invades and base-pairs with one of the strands, thereby forming a D-loop. This process is driven by ATP release as ADP. The process can continue with the next section of the ssDNA. Rad St 154 RS-E LICL7!GSKNLOTLLG-GGvtTGS!'l'ELFGEFRTGKSQLCHTl,AVTCQ I PLO IGGGEGK-CLYI DTEGTFRPVRLVS IAQR I i :: : : I ll I I : II I I : I I 1:: 1 : 11 : I ::: I 111 f:: I I I : ! : :I : RecA 33 RSMDVl:TJSTCSLSLDIALGAGGLPMGRIVEIYGPESSGKTTLTLQV!AAAQR-- --- --6GKTCI\F[DA£HALOP l --- -YARK RadS I 236 FGLDPDDALNNVAYARA~NAOHQLRLL-DMAQMMSESRFSL!VVDSVMALY-RTOFSGRGELSARQMHLA-KFMRALQR-oADQ :' : I I :I I: : I I:I : : I 111 I I : I I : :: 11: I II : I : I II : Rec A 101 LGVD !ONLLCSQPDTGE-- ---QALE ICOALA- --RSGAVDVIVVOSVAAL TPKAE IE· - GEIGDSHMGLMRMMSQAMRKLAGII Rad ~: 317 f G- --VAVVVTNQWAQVDGGMAF· NPDPKKPlGGHIM.l\HSSTTRLGFKKGKGCQRLC- KV\,'DS :: : : II : : : 1:: 1 II : I l l ::: : 11 ::: I : 11 I ~eel\ 182 1KQS!l'f£.LlF INQIRMK! - ·GVMFGNP&'T'.l'T--GGNALKrYASVRLDIRRl - GAVKEGENVVGS Figure 27 A.6. Amino acid sequence homology between bacterial RecA and yeast RadSl (Shinohara et al., 1992). How homologous r ecombination repairs DNA double-strand breaks (DSB). As already told in the previous chapter, the story DNA repair by homologous recombination (HRR) can be traced back to 1866, when Paul Broca noted a high incidence of breast cancer among his relatives. Other families were discovered that eventually led to the discovery the BRCAl and BRCA2 genes; mutation of either of those genes was associated with the high breast cancer-incidence families. As explained in the previous chapter, the two BRCA genes were found to bind to each other and to another protein, the above-mentioned RadSl that was implicated in HRR For HRR to work, a single-strand extension from at least one end of the break must become coated with RadSl protein, which enables the strand to find and invade another double-stranded DNA with the same nucleotide sequence (Figure 27 A. 7 A). But how could the broken strand find another DNA of the same sequence? The answer is that, when DNA replicates, it produces two identical DNA molecules that are held together at the centromere, as explained in Figure 26.8 of the previous chapter. The resulting two identical DNA molecules become "sister chromatids" of nearly identical DNA sequence, and recombination can occur between them. Unless efficiently repaired, double-strand breaks (DSB) are among the most lethal kinds of DNA damage. The repair based on Radding's model uses homologous recombination between the DNA of sister chromatids. First, one strand from each end of the DNA break is resected to produce a single-strand extension from each end of the break (top of Figure 27 A.7 A). The single-strand extension then becomes coated with RadSl proteins, which 533 K. W. Kohn Drugs Against cancer CHAPTER 27A occurs with the aid of BRCAl and BRCA2 (the three proteins bind together, as explained in the previous chapter). We get a picture of what a RadSl-coated DNA single-strand looks like from the molecular structure shown in Figure 27 A.7B: the relatively much greater size of the protein overwhelms the thin DNA strand. The invaded structures at the bottom of Figure 27A.7A can then resolve in several ways, some of which result in recombination between the two sister-chromatid DNA's (Figure 27 A.8). ,; 3' l' S' l Strand resection - ,; 3' - from s·~nd - 3' s· l Rad51-0>ated - strand DNA •- ,; - -r ·3•- s 1 -\ f.. , . _ . - D·Loop s3' -- l' ,; Figure 27A.7A. Two simplified versions of the early stages of homologous recombination in the repair of a DNA double-strand break In the first step, one of the strands on each side of the break are resected to form single-strand extensions that then become coated with RadSl protein. The RadSl-coated DNA strand can then invade an undamaged DNA and set the stage for homologous recombination. The diagram on the left shows a single-strand from the damaged DNA invading an undamaged DNA, forming a D-loop (Greene, 2016). The diagram on the right shows another possibility, where both single-strand segments of the damaged DNA invade the undamaged DNA. 534 K. W. Kohn Drugs Against cancer CHAPTER 27A 4 RecA Figure 27A. 7B. Part of a molecular structure of a RecA-coated DNA single-strand. The structure shows a section of a DNA single-strand with four RecA protein molecules wrapped around it. RecA is the bacterial version of mammalian Rad51. The RecA molecules (alternating blue and green) fit into each other and wrap around the relatively much thinner DNA strand. ((Greene, 2016) with label in red added) . ...... _., 0 =---..IX. C - I ....... ::::r.==:x= I - ---------~ d ::::>c==x e A non-crossover crossover Figure 27A.8. How the final repair products form, as portrayed by (Szostak et al., 1983). The process shown begins after one strand at each end of the double-strand break (DSB) has been resected, leaving a single-strand extension at each end (a). The arrowheads show the 535 K. W. Kohn Drugs Against cancer CHAPTER 27A 5' ➔3' direction of each DNA strand. In (b ), the single-strand extension of the right end has invaded an undamaged sister DNA and paired with one of the strands of the sister. In (c), repair synthesis, shown by the dashed line, has extended the invaded strand. In (d), the single-strand extension of the left end has paired with the other strand of the sister DNA and has been extended by repair synthesis (dashed line). (e) shows two possible outcomes, depending on whether the junction on the left has crossed over. Repair of double-strand breaks (DSB) was intensively studied over the years and new information gradually accumulated. By 2007, new information allowed Thorslund and West of the London Cancer Research Institute to diagram DSB repair by homologous recombination, as shown in Figure 27A.9 (Thorslund and West, 2007). They portrayed repair of DSB by homologous recombination (HR) beginning with step a, which determines whether a DSB will be repaired by HR or by non-homologous end joining (NHEJ). HR repair occurs almost only when DNA replication has produced a sister chromatid that is nearby and from which homologous sequences could be copied. HR can occur during Sor G2 of the cell cycle but rarely happen during GO or Gl , because it would then be less likely for a homologous target DNA to be located nearby. Accordingly, DSB repair during GO and Gl happens predominantly via NHEJ. The 2007 diagram shows repair via HR beginning in step a, where the MRN complex resects one of the strands at each end of the DSB -- which allows RPA molecules to associate with the exposed single-strand segments. Steps b and c in Figure 27A.9 bring in Rad52 and BRCA2, leading to RadSl-coatingof the single-strand segments. In 2016, more details about this complicated process were clarified in the later diagram by George Illiakis and colleagues (Figure 27 A.10). Step d in Figure 27A.9 shows a RadSl -coated strand (blue) invading the DNA ofa sister chromatid (red). In step e, a DNA-repair polymerase copies sequences in the sister chromatid, leading in step f to two crossovers called Holliday junctions, which can be resolved in several ways, one of which is shown in step g. The final product is an error-free reconstruction of the DNA. 536 K. W. Kohn Drugs Against cancer CHAPTER 27A NHEJ HR Predominant in G 1 Predominant in SIG2 h a jjjjjjjfjj,j jjjjjjjjjjjj j jjj j j jjjj j j j j j jj ■ j j jj j j Resection RPA # ~ MRN complex s· b ~ # • # er "• , nu ~ .,..,rn Filament RAD51 e RAD52/BRCA2 Formation ~ C Invasion Into sister chromatid RAD51 ~ RAD52/RAD54 I SOSA d m l IIDllll~~,,:-u--i: : ~ :: 11 I 11 11 I 11 11 - 11111 ~ ~ II ii 1111 ii II ~~ IlII JUI ____,_...,_._ I DNA synthesis & RAD51 I RADS2/RAD54 T Second-end capture Polymerase T iii iii iii 11111 I iii I iii ii iii ii iii iii ii I 1lll tii jjjfj i ,, ii ii iii iii e lllll ~ ~ •IIIIII k Ii ii i ii i ii ii ii ii i i ii i ii Ii i ii ii i ii i ii ii ii I Ii iii I Iii I Ii I iii ii I ii I ii ii iii Ii iii I DNA synthesis & Polymerase I Ligation Ligase T f :::::::::x::::::::::::::):::::::::: ~ Holliday Junction Resolution II i II Ii ii q P i O i i Ii II i II I i ii i I ■ ii i i II 9 i ii ii i ii ii ii i ii i i ii ii i i jj i ii i ii ii ii ii i Figure 27A.9. Steps in the repair of double-strand breaks (DSB) by homologous recombination (HR) as diagrammed in 2007 by Thorslund and West (Thorslund and West, 2007). The steps are explained in the text above the Figure. Then, in 2016, George Iliakis and his colleagues at the University of Duisburg-Essen Medical School, Germany, summarized further details of the steps from strand resection to the coating of the strand with RadSl (Mladenov et al., 2016) (Figure 27 A.10). They began by pointing out that repair of DSB by homologous recombination may have to overcome complications such as damaged nucleotides and staggering of the breaks on the two DNA strands. Such complicated DSB are produced particularly by the high-energy radiation used in modern radiotherapy (see Figure 27 A.3). 537 K. W. Kohn Drugs Against cancer CHAPTER 27A The first step in repair of DSB by homologous recombination repair (HRR) therefore is to cut away any nucleotide damage and then to resect one of the strands from each end, as shown in Figure 27A.7A. This is accomplished by a combination of three proteins: Mrell, RadSO and Nbsl (step A in Figure 27A.10). The complex of those three proteins is often abbreviated MRN. The action of MRN in resecting the strands from the S' -ends is crucial for the initiation of repair by HRR, and it is what determines that repair will go by HRR, rather than by non-homologous end joining (NHEI). MRN (with the aid of another protein, Ct!P) however only begins the resection. The story of how the MRN genes were discovered and what happens when one of those genes is defective will be told later in this chapter. Step B in Figure 27A.10 unwinds the DNA to allow the Exol exonuclease to nibble away more of the other strand, so as to produce the long single-strand extension needed to probe, reach and invade the sister DNA. The helix unwinding is carried out by the helicase BLM together with another protein, Dna2, to which it is bound. A genetic defect in BLM causes Bloom syndrome, which will be discussed later in this chapter. In step C, the long single-strand extensions become bound by a line-up of RPA protein molecules that have a high affinity for such binding. That stabilizes the strand and prevents untoward binding events. Next, the RPA has to be removed before the required coating by RadSl can take place (see Figure 27 A.7). The problem here is that RPA binds rather tightly to single-stranded DNA, so some special effort is required, which is carried out by a triplex consisting of BRCAl , BRCA2, and BALB2 (step D). The stories of BRCAl and BRCA2 were told in Chapter 26. Then, in step E, the BRCA1-BALB2-BRCA2 trimer, with the additional aid of Rad52, manage to remove the RPA molecules and replace them with RadSl - at last! Step F then allows strand invasion to take place as shown in Figures 27A.7. 538 K. W. Kohn Drugs Against cancer CHAPTER 27A 0S8 Induction ONA ecnd nt&Ktion Rad51 fllame-nt formatiOfl Strand invasion F Figure 27 A.10. The steps in homologous recombination repair (HRR) from DNA double- strand breaks (DSB) to stand invasion, according to George Illiakis and colleagues in 2016 (Mladenov et al., 2016). The steps are explained in the text. Some of the essentials of DSB repair by homologous recombination, as understood at the time of this writing. were presented by Eli Rothenberg. Michael R. Lieber and colleagues (Zhao et al., 2019), as shown in Figures 27 A.1 lA and B. The process is explained in the captions in the two parts of the Figure. After this survey of how our understanding of the events in the repair of DNA damage by homologous recombination developed, we turn next to some of the relevant stories of discovery and clinical implications. 539 K. W. Kohn Drugs Against cancer CHAPTER 27A ONA double-strand break y ~~ l 538PI-RIFI- Shieldin-CST . . /X\/X\/X\~ ~ "'"'"' W'\JVVJ ~ ~ \XI\X/\)V j End resection MRN,C~P BLM-TOPOllla- RMl/2 DNA2,EX01, APA Nuclea.ses RPA ~ ~ Presynaptlcfilament fo,matlon j DSSI RADS! ~ Figure 27 A.1 lA. The first part of the scheme for repair of a DNA double-strand break by homologous recombination. This first part of the process cuts back one strand of DNA double-helix to reveal a single-strand section projecting from the end of the break The cutting back is done by the nuclease, EXOl. BRCAl, together with its partner, BARDl, as well as several other proteins, then coats the projecting single-strand with RPA, which consists of 3 parts (RPAl, RPA2, and RPA3) . The BRCAl-BARDl pair then bring in BRCA2 together with two other proteins (PALB2 and DSSl). The complex of those 5 proteins, all bound together, then manage to replace the RPA molecules arrayed along the single-strand with an array RADSl molecules. RADSl confers a helical structure to the strand filament. The DNA single-strand coated with RADSl, aided by its helical structure, has the capability to invade another DNA double-helix and to initiate the recombination process, as shown in the next part of the scheme in Figure 27 A.11B. (From (Zhao et al., 2019).) 540 K. W. Kohn Drugs Against cancer CHAPTER 27A RAOS1 ~ ~ ONA strnnd invasion 0-loop formation Repa.ir ONA synthesis SOSA DSBR dHJ dissolution (noncrossovers) (croS5oOvers o r (noncrossovers) noncrossovers) Figure 27 A.11B. The second part of the scheme for repair of a DNA double-strand break by homologous recombination. The RADSl-coated DNA single-strand filament is now ready to invade an undamaged homologous DNA helix located nearby. The BRCAl-BARDl pair also functions here to bring about the invasion. From here, several pathways can complete the error-free repair of the damaged DNA. (From (Zhao et al., 2019).) The MRN story. The MREll-RADS0-NBSl (MRN) complex is one of the first sensors and responders to DNA damage and initiates the repair of DNA double-strand breaks (DSB). It has a key role in deciding between the two major DSB repair pathways: whether the repair will go by way of homologous recombination (HRR) or by way of non-homologous end joining (NHEJ). The current chapter focusses on the former, the next chapter on the latter. This section of the chapter will take a closer look at how the remarkable MRN molecular machine does its job. But first a brief review of how the three parts of the MRN complex were discovered. Of the three genes whose protein products make up the Mrell-RadS0-Nbsl (MRN) complex, the first to be discovered were found in yeast strains that needed those genes to initiate homologous recombination during meiosis. Genetic recombination during meiosis begins by creating a double-strand break (DSB) and resecting one of the strands so as to produce a single-strand extension. Looking for yeast mutants that were unable to carry out this first step of meiosis, Hideuki Ogawa and colleagues at Osaka University isolated two 541 K. W. Kohn Drugs Against cancer CHAPTER 27A genes whose mutation destroyed the ability to produce the DSB (Ogawa et al., 1995). One of the genes was new and was dubbed Mrell (for meiotic E,combination); the other was the previously known yeast gene Rad50. They then went on to show that the two genes, Mrell and Rad50, bind to each other in a complex required for DSB repair Oohzuka and Ogawa, 1995). However, where did the Nbsl part of the MRN complex come from? The Nijmeg en Breakage Syndrome. The third gene of the MRN complex came from a clinical and cytological description of a 10- year-old boy in 1981 by clinicians at the University of Nijmegen, The Netherlands. An older brother with similar clinical features had died at age 6 of severe infections. The main clinical features were reduced head size (microcephaly), stunted growth, mental retardation, and respiratory tract infections. The overall picture suggested a relationship with the previously known familial chromosome abnormality diseases: xeroderma pigmentosum, ataxia telangiectasia, Fanconi's anemia, and Bloom's syndrome, but there were distinguishing features that suggested a new genetic disease. As in those other DNA repair defect syndromes, there were immunoglobulin abnormalities and frequent chromosome breaks and translocations. The new syndrome came to be known as Nijmegen Breakage Syndrome (NBS) and the responsible gene Nbsl . In addition to random chromosome breaks, NBS patients often had specific translocations between the immunoglobulin and T-cell receptor genes on chromosomes 7 and 14 (Figure 27A.12.) (Digweed and Sperling. 2004; Weemaes et al., 1981). Head cirOJmference .. •"• b • ., •• ' t • • ";>,-,-+++1+++1+++1+++H Proband at age. 9. Proband's brother I 2 3 4 :, G TraMloc:ation of chromosome:s 7 at ag•6. Age (yea,s) and 14 in proband's lymphocyte:s. Figure 27 A.12. The first described cases of Nijmegen breakage syndrome. The proband (defining case), born in 1969, had repeated upper respiratory infections; in addition to the freckles, he had several cafe-au-lait spots on the trunk. His brother, born in 1964, had multiple severe infections and died at age 6 of respiratory failure. Both had reduced head circumference and mental retardation. They had chromosome abnormalities, such as translocation between chromosomes 7 and 14 to produce abnormal recombined products, such as m2 and m3 shown on the right (Weemaes et al., 1981). 542 K. W. Kohn Drugs Against cancer CHAPTER 27A The Mre11-Rad50-Nbs1 (MRN) molecular machine. A DNA double-strand break (DSB) presents a challenging problem for any repair machinery, because the break completely separates the DNA into two parts. A general repair device would have to grab onto both ends of the separated DNA ends, bring them back into alignment, and process the ends in a manner that allows enzymes to reconnect the two strand breaks. To make this possible, the process has to digest away a few bases, which entails loss of some information and makes the repair error-prone. This repair process is known as non-homologous end joining (NHEJ) and is the subject of the next chapter. The current chapter focusses on repair by homologous recombination (HRR), which is an even more daunting process and more restricted in when it can occur. But it has the merit of being error-free. The MRN machinery is needed for both of those repair pathways. MRN was thought of as having three actions: (1) binding to the ends of DSBs and processing the ends chemically to allow further repair steps, (2) bridging between the ends of a DSB in repair by NHEJ or between homologous DNA regions of two sister chromatids in repair by HRR, and (3) signaling control networks to pause cell cycle events in order to give more time for DNA repair (Bian et al., 2019). The importance of MRN function for the cell was indicated by its widespread conservation among species in evolution; Mrell and RadS0 are conserved from bacteria to mammals, and Nbsl is conserved among eukaryotes. Figure 27 A.13 shows a structure of the MRN complex as deduced by (Hopfner et al., 2002) and depicted by (Bian et al., 2019). Figure 27 A.14 shows how a pair of MRN complexes may link a DSB with a homologous sequence in a sister chromatid in HRR (left) or link together the ends of a DSB in NHEJ (ri,qht) . Zinc hooks connecting .,.,-- two RADSO coiled coils l-• Coiled coil of RAD50 ./ RADSO RADSO A,8, Globular DNA domams of RADSO Mrell I NBS1 NBS1 Figure 27A.13. The MRN complex consists of two molecules, each of the Mrell, RadS0, and Nbsl triplet and could form a structure such as shown in the figure. RadS0 is made up of a long coiled-coil, each end of which has a globular domain (A and B) that fits into a pocket in the Mrell protein. The midpoint of the coiled coil, where it loops back, has a pair of 543 K. W. Kohn Drugs Against cancer CHAPTER 27A cysteines that bind a Zn atom. The Zn regions of the coiled coils of the two Rad50 molecules form hooks that link them together. DNA can fit between the two Mrell molecules. Nbsl can bind ATM (ataxia telangiectasia mutated protein) to signal cell cycle delay while DNA repair is in progress and could link to other control molecules. ((Bian et al., 2019) with added labels.) DNA repair by homologous DNA repair by non-homologous end joining (NHEJ). recombination (HRR). '' I I I I ' ' • ONA tfokcn sites Figure 27A.14. How a pair of NRM complexes may function in DNA repair by homologous recombination (left) or by non-homologous end joining (right) (Modified from (Bian et al., 2019).) Homologous recombination repair (left): The DNA with the double-strand break is at the bottom. The breaks in its two strands are indicated by red stars. A pair of MRN complexes (see Figure 27 A.13) holds together the broken DNA and the DNA of a homologous chromatid. Non-homologous end joining (right): In this case, the DSB separates the DNA into two parts, the ends of which are held together by a pair of MRN complexes. A structural change in the MRN complexes may bring the broken ends into alignment. References Bian, L., Meng. Y., Zhang. M., and Li, D. (2019). MRE11-RAD50-NBS1 complex alterations and DNA damage response: implications for cancer treatment. Molecular cancer 18, 169. Bradley, M .O., and Dysart, G. (1985). DNA single-strand breaks, double-strand breaks, and crosslinks in rat testicular germ cells: measurements of their formation and repair by alkaline and neutral filter elution. Cell biology and toxicology 1, 181-195. Critchlow, S.E., and Jackson, S.P. (1998). DNA end-joining: from yeast to man. Trends Biochem Sci 23, 394-398. Digweed, M., and Sperling. K. (2004). Nijmegen breakage syndrome: clinical manifestation of defective response to DNA double-strand breaks. DNA repair 3, 1207-1217. 544 K. W. Kohn Drugs Against cancer CHAPTER 27A Greene, E.C. (2016). DNA Sequence Alignment during Homologous Recombination. The Journal of biological chemistry 291, 11572-11580. Hopfner, K.P., Craig, L., Moncalian, G., Zinke!, RA., Usui, T., Owen, B.A., Karcher, A., Henderson, B., Bodmer, J.L, McMurray, C.T., et al. (2002). The Rad50 zinc-hook is a structure joining Mrell complexes in DNA recombination and repair. Nature 418, 562- 566. Iliakis, G., Mladenov, E., and Mladenova, V. (2019). Necessities in the Processing of DNA Double Strand Breaks and Their Effects on Genomic Instability and Cancer. Cancers (Basel) 11. Johzuka, K., and Ogawa, H. (1995). Interaction of Mrell and Rad50: two proteins required for DNA repair and meiosis-specific double-strand break formation in Saccharomyces cerevisiae. Genetics 139, 1521-1532. Kaplan, A.R., and Glazer, P.M. (2020). Pharmacological methods to transcriptionally modulate double-strand break DNA repair. Int Rev Cell Mo) Biol 354, 187-213. Mladenov, E., Magin, S., Soni, A., and Iliakis, G. (2016). DNA double-strand-break repair in higher eukaryotes and its role in genomic instability and cancer: Cell cycle and proliferation-dependent regulation. Semin Cancer Biol 37-38, 51-64. Ogawa, H., Johzuka, K., Nakagawa, T., Leem, S.H., and Hagihara, A.H. (1995). Functions of the yeast meiotic recombination genes, MREll and MRE2. Adv Biophys 31, 67-76. Shibata, T., Cunningham, R.P., DasGupta, C., and Radding, C. M. (1979). Homologous pairing in genetic recombination: complexes of recA protein and DNA. Proceedings of the National Academy of Sciences of the United States of America 76, 5100-5104. Shinohara, A., Ogawa, H., and Ogawa, T. (1992). Rad51 protein involved in repair and recombination in S. cerevisiae is a RecA-like protein. Cell 69, 457-470. Szostak, J.W., Orr-Weaver, T.L., Rothstein, R.J., and Stahl, F.W. (1983). The double-strand- break repair model for recombination. Cell 33, 25-35. Thorslund, T., and West, S.C. (2007). BRCA2: a universal recombinase regulator. Oncogene 26, 7720-7730. Vriend, L.E., and Krawczyk, P.M. (2017). Nick-initiated homologous recombination: Protecting the genome, one strand at a time. DNA repair 50, 1-13. Weemaes, C.M., Hustinx, T.W., Scheres, J.M., van Munster, P.J., Bakkeren, J.A., and Taalman, R.D. (1981). A new chromosomal instability disorder: the Nijmegen breakage syndrome. Acta Paediatr Scand 70, 557-564. West, S.C., and Kowalczykowski, S.C. (2021). Charles M. Radding: A love of science and art. Proceedings of the National Academy of Sciences of the United States of America 118. Zhao, W., Wiese, C., Kwon, Y., Hromas, R., and Sung, P. (2019). The BRCA Tumor Suppressor Network in Chromosome Damage Repair by Homologous Recombination. Annual review of biochemistry 88, 221-245. 545 K. W. Kohn Drugs Against cancer CHAPTER 27B Chapr,er-278. DNA douhle-strand b~lc repair by nonhomologou:$ ffld joining zzon 7at3 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@nih.gov CHAPTER27B DNA double-strand break repair by nonhomologous end joining. DNA double-strand breaks (DSB) are notoriously toxic and difficult to repair. Evolution has managed to develop several DSB-repair pathways that operate in a wide range of organisms. Of the two major pathways, the previous chapter deltwith repair by homologous recombination, while the current chapter will focus on repair by non- homologous end joining (NHEJ). NHEJ has the advantage of being able to work at any time during the cell cycle but at the cost of being error-prone by deleting a few base-pairs of DNA sequence. How non-homologous end joining (NHEJ) and its components were discovered. In 1983, P. A. )eggo and L. M. Kemp at the National Institute for Medical Research in London, England isolated seven x-ray-sensitive mutants of a Chinese-hamster cell line (Figure 27B.1A.) (Jeggo and Kemp, 1983). Six of those mutant cell lines were also sensitive to several DNA-damaging drugs, including bleomycin. The mutations were all of the same complementation group, indicating that they were all mutations of the same gene. )eggo and his colleagues went on to use the DNA filter elution methods (see Chapter 9) to show that the mutant cell lines were indeed defective in their ability to repair DNA double-strand breaks (DSB), but not single-strand breaks (SSB) (Kemp et al., 1984) (Figure 27B.1B). It took several more years to isolate that mutated gene, but by 1992, they had identified it as XRCCS (Jeggo et al., 1992), whose protein product came to be called Ku80 and was found to be an essential part of the NHEJ machinery. 546 K. W. Kohn Drugs Against cancer CHAPTER 27B But where did this 'Ku' come from? Surprisingly, it came from clinical studies totally unrelated to cancer or DNA repair. In 1981, Tsuneyo Mimori and colleagues at the Keio School of Medicine in Tokyo had discovered that some patients with connective tissue diseases related to systemic lupus erythematosus and scleroderma produced antibodies against a previously unknown antigen that they called Ku after the first two letters in the name of the first patient in whom they found the antigen (Mimori et al., 1981). Mimori and Hardin, then at the Yale University School of Medicine in New Haven, Connecticut, went on to investigate Ku's structure and function (Mimori and Hardin, 1986; Mimori et al., 1986). They found that Ku actually consisted of two proteins that function together. They called the two proteins Ku80 and Ku 70, after their approximate molecular weights, 80 and 70 kilodaltons. As the first clue to their role in non -homologous end joining (NHEJ), they discovered that the two proteins bind tightly to the ends of double-stranded DNA, such as would occur in DNA double-strand breaks (DSBs). Ku80 and Ku70 were found to bind to each other to form a kind of donut shape that assembled around the DNA helix near its broken end. Once bound to DNA at a DSB end, Ku seemed to be able to slide along the DNA like a bead on a string, and would not come off until it reached a DSB or to some other kind of DNA terminus (Paillard and Strauss, 1991). Ku protein was found to be abundant in cell nuclei (Figure 27B.2) and was well placed there to search for and find DSBs and then to initiate their repair. Next came a discovery by Carl W. Anderson at Brookhaven National Laboratory in Upton, New York, of a protein kinase in cells of several species that was strongly activated by double-stranded DNA (Chen et al., 2021b; Lees-Miller and Anderson, 1989). (A protein kinase is an enzyme that phosphorylates other proteins, sometimes also itself.) This DNA- dependent protein kinase (DNAPK) was found to bind Ku plus a 135 kilodalton protein that was the catalytic subunit that could phosphorylate selected serine or threonine amino acids in a variety of proteins. Its activity required both Ku and DNA (Lees-Miller et al., 1990; Peterson et al., 1995; Suwa et al., 1994). A DNA-dependent protein kinase activity actually was discovered earlier, in 1985, as a kinase activity that was stimulated by double-stranded DNA to phosphorylate a variety of proteins (Walker et al., 1985). The discovery was accidental. Anthony Walker and his colleagues at Cambridge University, England, noted a kinase activity while studying the stimulation of protein synthesis by RNA. The kinase activity was not stimulated by RNA, but, surprisingly, by DNA that contaminated their RNA. DNAPK was found to be a trimer consisting of a 135 kilodalton protein that was the catalytic subunit and the two subunits of Ku (Ku80 plus Ku70) . The catalytic subunit (DNAPKcs) was active only when it was bound to DNA (Gottlieb and Jackson, 1993). Relevance to the repair of DNA double-strand breaks (DSB) came from evidence that Ku80 was the product of the XRCCS gene that had been found to function in DSB repair (Rathmell and Chu, 1994). These findings about Ku and DNAPK were intriguing. but much about their functions in DNA repair remained unknown or uncertain. It may well have been suspected 547 K. W. Kohn Drugs Against cancer CHAPTER 27B that DNAPK phosphorylates and thereby brings into play other components of the DSB repair machinery. Other early clues were that Ku binds to DNA double-strand ends and recruits DNAPKcs to those ends. Moreover, Ku stimulated DNAPKcs to phosphorylate various DNA-bound proteins, such as transcription factors (Dvir et al., 1992; Gottlieb and Jackson, 1993). DOSE (AD) Figure 27B.lA. The increased x ray-sensitivity of mutant cells lines, relative to a non- mutated cell line (filled circles), isolated by Jeggo and Kemp in 1983 Qeggo and Kemp, 1983). Six of the mutant cell lines were also hypersensitive to bleomycin. The mutations in the six cell lines were all of the same gene, XRCCS, whose protein product, Ku 80. was later found to have an essential role in DNA repair by the NHEJ mechanism. 1 eo eo Figure 27B.1B. The panel on the right showed by neutral DNA filter elution (see Chapter 9) that an XRCC5-mutated cell li ne was defective in repair ofx-ray-induced DNA double- strand breaks (DSB). (The x-ray-sensitivity of this mutant was shown by the filled squares in the right panel of Figure 27B.lA.) The panel on the left shows the normal repair in a non - mutated cell line. Open symbols: triangles, 0 min; squares, 20 min; diamonds, 60 min; circles, 120 min; inverted triangles, 240 min repair times after irradiation with 100 Gy ofy- rays. Filled circles, unirradiated controls. (From (Kemp et al., 1984).) 548 K. W. Kohn Drugs Against cancer CHAPTER 27B Figure 27B.2. HeLa cell nuclei were loaded with many copies of the Ku proteins, which showed up as fluorescent speckles (Mimori et al., 1986). (The cell line used was reported as Hep-2 but was later found to have been contaminated by HeLa cells - a common occurrence in those years due to the vigorous growth of HeLa cells that tended to take over cell cultures after the slightest contamination.) How non-homologous end joining (NHEJ) repairs DNA damage. In view of NHE)'s prominence in DNA repair - particularly the repair of the potentially disastrous DSB's - much research effort went into working out the details of that remarkable repair process. By 1998, the binding of the Ku80-Ku70 dimer to the ends of DSBs had been firmly established (Critchlow and Jackson, 1998). Also known was that the Ku dimer can bind and activate DNAPKcs, as the catalytic subunit was called. The activation of DNAPKcs, however, also required DNA, which is why the enzyme was designated 'DNAPK' for 'DNA-dependent protein kinase.' In other words, 'DNAPK' referred to the Ku80-Ku70 -DNAPKcs trimer, a catalytically active protein kinase. It was correctly supposed that the first step in DSB repair by NHEJ was the binding of the Ku80-Ku70 to the ends of the breaks and then bring in the DNAPKcs. The resulting DNA-bound trimer would then be an active protein kinase that could phosphorylate other proteins that might assemble around the DSB site in the course of the repair. The li mited knowledge, as of 1998, about DSB repair by NHEJ was shown in a model by Critchlow and Jackson (Figure 27B.3.). If we move ahead a decade to 2008, we find that researchers had come to understand the basic steps in the repair of DSBs, although the order in which some of the steps occurred was uncertain. They were amazed by the versatility of NHEJ to reconnect the ends of DSBs when the ends had a variety of abnormal structures, even when the ends seemed to be incompatible with each other - as reviewed by Michael Lieber in 2010 (Lieber, 2010). New 549 K. W. Kohn Drugs Against cancer CHAPTER 27B information had accrued by 2008 to allow Katheryn Meek and her colleagues to suggest a model of how it all happens (Figure 27B.4.) (Meek et al., 2008). It was understood that something first had to recognize the DSB. That something was the Ku dimer consisting of two similar protein molecules, Ku80 and Ku 70, which were named according to their respective approximate molecular weights, with Ku80 having some additional structure compared to Ku 70. Ku was found to bind tightly to the end of a DSB. Its molecular structure had a loop into which the DNA at the end of a DSB would insert. Attached by its loop encircling the DNA, Ku could then slide along the DNA, safely attached by way of the loop around the double-helix (B in Figure 27B.4). Next, enzymes to carry out the repair were inferred to assemble to form a molecular complex at the DSB. Which enzymes came into the complex would depend on the type of molecular abnormalities existing at the DSB - which is what conferred the flexibility of the NHEJ mechanism to repair a variety of problematic DSB ends. The first step in the assembly, however, was the binding ofDNAPKcs to Ku and the DNA (C in Figure 27B.4). The structure of Ku was revealed to have a ring that encircles the DNA near the end of the break that can glide along the DNA a short distance from the end of the DNA break; DNAPKcs bound firmly to Ku and had contact also with the DNA (C in Figure 27B.4). Next, the two ends of the DSB had to find and bind to each other, a process called synapsis. One way that was found to happen was through the DNAPKcs molecules at the two ends of the DSB binding to each other. According to the model, the DNAPKcs at each DNA end changes shape in a manner that allows two DNAPKcs molecules at two ends of the break to join and phosphorylate each other (D in Figure 27B.4). Seven proteins were then known to take part in NHEJ: Ku70, Ku80, DNAPKcs, XRCC4, DNA ligase IV (Lig4), the nuclease Artemis, and XLF. DNAPKcs would carries out many phosphorylations to facilitate the NHEJ process. The model suggested that the two DNAPKcs molecules phosphorylate each other. The Artemis nuclease would then chew away a few nucleotides from the ends of the DNA strands until the strands from the two ends of the break found short regions of sequence homology them to join (E-G in Figure 27B.4). The final ligation of the DNA strand ends was found to be carried out by a complex of Lig4-XRCC4-XLF (Meek et al., 2008). Moving ahead another decade to 2018, we find that NHEJ has much flexibility for carrying out its task under a variety of circumstances, with different sets of proteins coming in to join the ends of DNA breaks that have different molecular configurations. To begin with, there is great variability in the length of single-strand overhang (see Figure 27 A.1 in Chapter 27 A). Then there is the question of how far have the broken ends drifted away from each other. What happens if the ends of a break do not find each other? It was proposed that the separated ends could eventually engage in homologous recombination; however, if that occurred when the DNA had not replicated, so that there was no sister DNA strand nearby for accurate recombination, then the recombination would be apt to occur by joining with inappropriate homologous regions of the genome, thereby causing 550 K. W. Kohn Drugs Against cancer CHAPTER 27B chromosome anomalies. A particularly problematic situation would be when a replication fork encounters a DSB, or when the replication fork encounters a single-strand break (SSB) thereby generating a DSB. There appeared to be NHEJ processes configured to meet the many variation of DSB structure. Especially noteworthy during that decade were the new molecular structures that were determined for many of the NHEJ participants. Much, although incomplete, information accrued about how those participant proteins combined in different way to carry out functions appropriate for various configurations that DSBs may have (Hnizda and Blundell, 2019; Pannunzio et al., 2018) (Figures 27B.6-7). Exactly how those proteins assembled and functioned together in the various circumstances was only partially elucidated. As already said, the Ku70 -Ku80 dimer was found to form a loop through which a DNA broken end could be threaded (Figure 27B.5). After the dimer has assembled on the DNA and moved a short distance from the end, a DNAPKcs molecule bound by associating with the Ku80 unit of the dimer. The two broken ends of the DSB, each with a Ku70-Ku80 dimer and DNAPKcs must somehow find each other and align correctly (D of Figure 27B.4). This process, called synapsis, was difficult to study, and some experiments even suggested that synapsis may occur by way of other proteins without implicating DNAPKcs at this step in the repair (Zhao et al., 2020). Another difficulty was that the DSB end could have many different structures, depending on what caused the break. DSB repair appeared to be able to handle different structural problems through several possible end-rejoining pathways involving different multi-protein assemblies. A common feature however was an exonuclease called Artemis that cut away abnormal DNA structures from the end-regions of the DSB. The action of Artemis was already portrayed in 2008 in the green units in E of Figure 27B.4 (Meek et al., 2008). By the end of 2020, many additional proteins had been found to interact with components of NHEJ, and their molecular structures were determined, but exactly how those structures assemble and function together was still not clear. It appeared that there were several ways those structures could assemble and function in NHEJ under different circumstances. By 2022, structural studies by electron microscopy gave further detail on how DSB repair by NHEJ works, although exactly how it brings the two broken DNA ends together remained uncertain (Menolfi and Zha, 2022). At the core of the NHEJ mechanism were the Ku70-Ku80 dimer, XRCC4, Lig4, XLF, and PAXX (Figure 27B.8). Versions of these components were found in all eukaryotes (with some exceptions) (Chen et al., 2021b), pointing to an early evolution of NHEJ and the critical need of most eukaryotic organisms to be able to repair DSBs. The picture as of 2021 was that Ku binds DNAPKcs to initiate NHEJ, and that Ku would then bind and recruit to the complex Lig4-XRCC4 and XLF, which would bring together the DNAPKcs units from the two broken ends of the DSB and induce them to phosphorylate each other, thereby allowing them to dissociate from the complex. There were contrasting reports, however, as to whether DNAPKcs was required for DNA end-bridging during NHEJ, and exactly how these factors coordinated to bring the broken DNA ends together remained unclear (Chen et al., 2021a). 551 K. W. Kohn Drugs Against cancer CHAPTER 27B Figure 278.3. An early model by Critchlow and Jackson in 1998 of the repair DNA double- strand breaks (DSB) by non -homologous end joining (NHEJ) (Critchlow and Jackson, 1998). They concluded correctly that the Ku80-Ku70 dimer would bind to the ends of the breaks and then recruit the catalytic subunit, DNAPKcs, to generate an active DNA-dependent protein kinase. But much about the configuration at the DSB site and how the DNA ends were brought together remained to be clarified. They also knew that DNAPKcs would recruit to the DSB site a complex consisting of Lig4 and a protein called XRCC4; they suppose correctly that the complex would come into play in final steps of the repair process to seal (ligate) the DNA strand ends to form two continuous DNA strands. artemis Q ~ / Ku DNAPKcs D ~ A ~~ ! l l B ~! '-'" E C Figure 278.4. The non -homologous end-joining (NHEJ) mechanism of DNA double-strand break (DSB) repair, as understood in 2008 by Katheryn Meek, Van Dang, and Susan Lees-Miller. (Modified and simplified from (Meek et al., 2008)). A, a double-strand break produced in DNA by ionizing radiation or oxidative molecules. B, the Ku80-Ku70 dimer bound and folded around the end of a DNA break. Ku glides a short distance (about 552 K. W. Kohn Drugs Against cancer CHAPTER 27B two turns of the DNA helix) along the DNA away from the end of the break. C, DNAPKcs binds to the Ku dimer at the end of the DNA break. The different colored patches represent the different functional regions (doma ins) of the DNAPKcs protein; the red patch is the region that has kinase activity. D, the two ends of the DNA break brought together by the purple domains of the two DNAPKcs molecules; the DNA helix at each end is unwound so as to facilitate the subsequent search for short complimentary sequences between the end- regions of the two ends of the break. E, the nuclease Artemis (green) chews away bits of nucleotide sequence from the DNA single-strand ends until a short region of sequence homology (2-4 base pairs) is found. This process is facilitated by the kinase region of each DNAPKcs molecule phosphorylating sites on the other DNAPKcs (red stars). F, a short region of base-pairs forms between strands from the two ends of the break. G, the ends of the strand breaks are sealed (ligated) by the complex of Lig4 and XRCC4 (not shown). A - B - C Figure 278.5. Molecular structures of Ku and DNAPKcs showing how they assemble at the broken end of DNA, as depicted in 2018 by Nicholas Pannunzio and Go Watanabe in Michael Lieber's laboratory at the Norris Comprehensive Cancer Center in Los Angeles, California (modified and simplified from (Hnizda and Blundell, 2019; Pannunzio et al., 2018). A, The Ku80 -Ku70 dimer binds to the broken end of a DNA double-helix. the colors red and yellow distinguish the two parts of the dimer. B, DNAPKcs binds to the Ku80 -Ku70 dimer at the end of the DNA. C, the complete structure: DNAPKcs-Ku80-Ku70 -DNA. The 553 K. W. Kohn Drugs Against cancer CHAPTER 27B different functional regions (domains) of DNAPKcs are shown in different colors. The Artemis nuclease ( whose molecular structure had not yet been elucidated) can bind to the DNAPKcs-Ku complex at the DNA break. D XRCC4 - Ligase IV Figure 27B.6. Molecular structures of central components in non-homologous end joining (NHEJ). A, Ku70 and Ku80 in the structure of the Ku dimer, showing the opening through which the DNA will thread. B, DNA within the opening in structure A. C, DNAPKcs added to structure B. D, a XRCC4-Lig4 dimer, showing the two similar parts of the Lig4 structure (BRCTl and 2). Modified and simplified from (Hnizda and Blundell, 2019). 554 K. W. Kohn Drugs Against cancer CHAPTER 27B I PAXX I XRCC4:Lig IV Lig IV Figure 27B.7. Structures of some of the components that can assemble with Ku at the end of a DNA double-strand break (DSB). Exactly how they fit together in assemblies to handle different DSB configurations had not yet been determined. Artemis connects with Lig4 by a disordered peptide chain, but the molecular structure of Artemis was not yet known (Pannunzio et al., 2018). + XLF -XLF Ku he terodimer .,Ku heterodimer gc_~:·· Ku70 homodlmer -'L-' 11.:,... 'l.- A~ I' I ·-·;"3, XRCC4 ~-~ ~ LigUe 4 Ligase4 KBM (Ku Binding Motifs) XLF (Ku80) PAXX (Ku70) - = APLF (Ku81>) ca 555 K. W. Kohn Drugs Against cancer CHAPTER 27B Figure 278.8. Further details about the interactions of some of the factors implicated in NHEJ (Tadi et al., 2016). The general shape of the XRCC4-Lig4 dimer is shown in blue and purple near the bottom of the figure. The newly discovered PAXX (red) binds to the Ku70 subunit of Ku, while XLF (green) binds to the Ku80 subunit. The Ku dimer (yellow) with its loop around the DNA is nicely portrayed. In the absence ofXLF, an abnormal version of Ku forms that consists of two Ku70 subunits, and PAXX can then bind to both of them (right side of the figure). DNAPK inhibitors for cancer therapy. Anticancer agents, such as ionizing radiation and topoisomerase inhibitors produce much of their therapeutic action by inducing DNA double strand breaks (DSB). In view of the major role of DNAPK in repairing DSBs, it was thought that DNAPK inhibitors might enhance those therapeutic actions by inhibiting DNA repair steps in NHEJ or homologous recombination, or by other actions (Mohiuddin and Kang, 2019). Supporting that idea was an early report that glioblastoma brain tumor patients treated with radiation survived longer (13 versus 9 months, p=0.02) if their tumor's DNAPK expression level was low rather than high (Kase et al., 2011). Therefore, inhibiting DNAPK expression could perhaps enhance the therapeutic action of radiation against those cancers. As hoped, a selective inhibitor of DNAPK's kinase activity, AZD7648, inhibited DNA repair in irradiated or doxorubicin -treated cells and, in combination with a PARP inhibitor, suppressed the growth of human xenograft tumors in mice (Fok et al., 2019). By 2020, several selective inhibitors of DNAPK's kinase activity were in early clinical trial as single agent or in combination with radiotherapy, a PARP inhibitor, or a topoisomerase inhibitor (Medova et al., 2020). The antitumor actions of DNAPK inhibitor, however appeared due in part to actions on immune components in the tumor cells' environment (Nakamura et al., 2021). Structure studies of the DNAPK protein with bound inhibitor set the stage for the development of new inhibitors (Liang et al., 2022). There was a bright horizon for DNAPK inhibitors in the armamentarium for cancer therapy. References Chen, S., Lee, L., Naila, T., Fishbain, S., Wang, A., Tomkinson, A.E., Lees-Miller, S.P., and He, Y. (2021a). Structural basis of long-range to short-range synaptic transition in NHEJ. Nature 593, 294-298. Chen, S., Lees-Miller, J.P., He, Y., and Lees-Miller, S.P. (2021b). Structural insights into the role of DNA-PK as a master regulator in NHEJ. Genome In stab Dis Z, 195-210. Critchlow, S.E., and Jackson, S.P. (1998). DNA end-joining: from yeast to man. Trends Biochem Sci 23, 394-398. Dvir, A., Peterson, S.R., Knuth, M.W., Lu, H., and Dynan, W.S. (1992). Ku autoantigen is the regulatory component of a template-associated protein kinase that phosphorylates 556 K. W. Kohn Drugs Against cancer CHAPTER 27B RNA polymerase II. Proceedings of the National Academy of Sciences of the United States of America 89, 11920-11924. Fok, J.H.L., Ramos-Montoya, A., Vazquez-Chantada, M., Wijnhoven, P.W.G., Follia, V., James, N., Farrington, P.M., Karmokar, A., Willis, S.E., Cairns, J., et al. (2019). AZD7648 is a potent and selective DNA-PK inhibitor that enhances radiation, chemotherapy and olaparib activity. Nat Commun 10, 5065. Gottlieb, T.M., and Jackson, S.P. (1993). The DNA-dependent protein kinase: requirement for DNA ends and association with Ku antigen. Cell 72, 131-142. Hnizda, A., and Blundell, T.L. (2019). Multicomponent assemblies in DNA-double-strand break repair by NHEJ. Curr Opi n Struct Biol 55, 154-160. Jeggo, P.A., Hafezparast, M., Thompson, A.F., Broughton, B.C., Kaur, G.P., Zdzienicka, M.Z., and Athwal, R.S. (1992). Localization of a DNA repair gene (XRCC5) involved in double-strand-break rejoining to human chromosome 2. Proceedings of the National Academy of Sciences of the United States of America 89, 6423-6427. Jeggo, P.A., and Kemp, L.M. (1983). X-ray-sensitive mutants of Chinese hams ter ovary cell line. Isolation and cross-sensitivity to other DNA-damaging agents. Mutation research 112, 313-327. Kase, M., Vardja, M., Lipping, A., Asser, T., and Jaal, J. (2011). Impact of PARP-1 and DNA-PK expression on survival in patients with glioblastoma multiforme. Radiother Oncol 101, 127-131. Kemp, L. M., Sedgwick, S.G., and Jeggo, P.A. (1984). X-ray sensitive mutants of Chinese hamster ovary cells defective in double-strand break rejoining. Mutation research 132, 189-196. Lees-Miller, S.P., and Anderson, C.W. (1989). The human double-stranded DNA-activated protein kinase phosphorylates the 90-kDa heat-shock protein, hsp90 alpha at two NH2 -terminal threonine residues. The Journal of biological chemistry 264, 17275- 17280. Lees-Miller, S.P., Chen, Y.R., and Anderson, C.W. (1990). Human cells contain a DNA- activated protein kinase that phosphorylates simian virus 40 T antigen, mouse p53, and the human Ku autoantigen. Molecular and cellular biology 10, 64 72-6481. Liang, S., Thomas, S.E., Chaplin, A.K, Hardwick, S.W., Chirgadze, D.Y., and Blundell, T.L. (2022). Structural insights into inhibitor regulation of the DNA repair protein DNA- PKcs. Nature. Lieber, M.R. (2010). The mechanism of double-strand DNA break repair by the nonhomologous DNA end-joining pathway. Annual review of biochemistry 79, 181- 211. Medova, M., Medo, M., Hovhannisyan, L., Munoz-Maldonado, C., Aebersold, D.M., and Zimmer, Y. (2020). DNA-PK in human malignant disorders: Mechanisms and implications for pharmacological interventions. Pharmacology & therapeutics 215, 107617. Meek, K., Dang, V., and Lees-Miller, S.P. (2008). DNA-PK: the means to justify the ends? Adv lmmunol 99, 33-58. Menol fi, D., and Zha, S. (2022). DNA-PKcs kinase activity orchestrates both end-processing and end -ligation. Trends Cell Biol 32, 91-93. Mimori, T., Akizuki, M., Yamagata, H., lnada, S., Yoshida, S., and Homma, M. (1981). Characterization of a high molecular weight acidic nuclear protein recognized by 557 K. W. Kohn Drugs Against cancer CHAPTER 27B autoantibodies in sera from patients with polymyositis-scleroderma overlap. J Clin Invest 68, 611-620. Mimori, T., and Hardin, J.A. (1986). Mechanism of interaction between Ku protein and DNA. The Journal of biological chemistry 261, 10375-10379. Mimori, T., Hardin, J.A., and Steitz, J.A. (1986). Characterization of the DNA-binding protein antigen Ku recognized by autoantibodies from patients with rheumatic disorders. The Journal of biological chemistry 261, 2274-2278. Mohiuddin, I.S., and Kang, M.H. (2019). DNA-PK as an Emerging Therapeutic Target in Cancer. Front Oneel 9, 635. Nakamura, K., Karmokar, A., Farrington, P.M., James, N.H., Ramos-Montoya, A., Bickerton, S.j., Hughes, G.D., lllidge, T.M., Cadogan, E.B., Davies, 8.R., eta/. (2021). Inhibition of DNA-PK with AZD7648 Sensitizes Tumor Cells to Radiotherapy and Induces Type I IFN-Dependent Durable Tumor Control. Clinical cancer research: an official journal of the American Association for Cancer Research 27, 4353-4366. Paillard, S., and Strauss, F. (1991). Analysis of the mechanism of interaction of simian Ku protein with DNA. Nucleic acids research 19, 5619-5624. Pannunzio, N.R., Watanabe, G., and Lieber, M.R. (2018). Nonhomologous DNA end-joining for repair of DNA double-strand breaks. The Journal of biological chemistry 293, 10512-10523. Peterson, S.R., Kurimasa, A., Oshimura, M., Dynan, W.S., Bradbury, E.M., and Chen, D.J. (1995). Loss of the catalytic subunit of the DNA-dependent protein kinase in DNA double-strand-break-repair mutant mammalian cells. Proceedings of the National Academy of Sciences of the United States of America 92, 3171-3174. Rathmell, W.K., and Chu, G. (1994). Involvement of the Ku autoantigen in the cellular response to DNA double-strand breaks. Proceedings of the National Academy of Sciences of the United States of America 91, 7623-7627. Suwa, A., Hirakata, M., Takeda, Y., Jesch, S.A., Mimori, T., and Hardin, J.A. (1994). DNA- dependent protein kinase (Ku protein-p350 complex) assembles on double-stranded DNA. Proceedings of the National Academy of Sciences of the United States of America 91 , 6904-6908. Tadi, S.K., Tellier-Lebegue, C., Nemoz, C., Drevet, P., Audebert, S., Roy, S., Meek, K., Charbonnier, J.B., and Modesti, M. (2016). PAXX Is an Accessory c-NHEJ Factor that Associates with Ku70 and Has Overlapping Functions with XLF. Cell reports 17, 541- 555. Walker, A.I., Hunt, T., Jackson, R.J., and Anderson, C.W. (1985). Double-stranded DNA induces the phosphorylation of several proteins including the 90 000 mol. wt heat- shock protein in animal cell extracts. The EMBO journal 4, 139-145. Zhao, 8., Rothenberg, E., Ramsden, D.A., and Lieber, M.R. (2020). The molecular basis and disease relevance of non -homologous DNA end joining. Nature reviews Molecular cell biology 21, 765-781. 558 K. W. Kohn Drugs Against cancer CHAPTER 28 Chapt,er-28. 11,~gammaHZAX story Z21009dh3.docx Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Developmental Therapeutic Branch National Cancer Institute Bethesda, Maryland kohnk@oib gov CHAPTER 28 The gamma-H2AX story: DNA double-strand breaks revealed in cell nuclei. In 1976, my Laboratory was fortunate to be joined by William M. Bonner, who had become interested in histones during his post-doctoral work at Oxford and Cambridge in England (Figure 28.1). Bill was one of those senior researchers who did much of the lab work himself. Thus, he had a hands-on feel for the experiments that, perhaps much to his own surprise, yielded ground-breaking advances for investigations of DNA damage repair and cancer. Our Laboratory had begun to study histones in the late 1960's, because these proteins were intimately associated with the chromatin where DNA damage and repair were taking place. But in those early days, we were still abysmally ignorant about histones and chromatin: the identity of the histone proteins was still hazy, and we did not yet know about nucleosomes! When Bill joined our laboratory, the nucleosome structure of chromatin had already come to light and the histone proteins had been well defined. Bill had a new and better understanding of chromatin structure and began in his modest lab to study the individual histone proteins carefully and systematically. He proceeded by carefully pinning down one step after another, without speculating too far into the future. I think he had no idea that his work would lead to something so important. His initial objective was to improve the gel electrophoresis method to increase its ability to separate the different histones and their sub-types. That would enable him to investigate how each of those sub-types would respond after cells were treated in different ways. He devised a new 2-dimensional gel system in which the histones could be separated, first by electrophoresis in one direction, followed by electrophoresis in the perpendicular direction in a different medium (e.g., acetic acid-urea). (gel electrophoresis separates histones, largely based on their positive charge, which causes them to move in an electric field.) 559 K. W. Kohn Drugs Against cancer CHAPTER 28 His new electrophoresis method revealed histone variant proteins that had never been seen before. He found the variants ofhistone H2A particularly amenable for study, because they were seen as distinctly separated spots in his gels. He had no idea what proteins were in those spots, so he called them H2AX and H2AZ. I don't know what happened to H2AY, but the other two had very bright futures as major players in the cell. Bill investigated both of them but focused mostly on H2AX- and this led to one of the most far-reaching discoveries in the field of DNA damage repair. Ironically, his early work on histone proteins had no obvious direct bearing on cancer research, but I supported it, because the histones surely had a role in most DNA functions, including what happens when anticancer drugs attack DNA. The irony was that, although we had no idea how his work would be relevant to that specific focus of our Laboratory, his contribution was one of the most important advances to come from our Laboratory - and it turned out to be directly relevant to our focus on anticancer drugs that attack DNA! When Bill Bonner modified the electrophoresis conditions to further increase its resolution, three modified forms of H2AX were revealed, which he marked alpha, beta, and gamma. A remarkable observation was that the amount of gamma-H2AX increased when the cells were exposed to radiation that damaged their DNA: gamma-H2AX became an exquisitely sensitive measure of the amount of DNA damage in the cells (Rogakou et al., 1998) (Figure 28.2 and 28.3). Importantly, they found that the gamma-H2AX that appeared after radiation was phosphorylated at a single position, namely at amino acid number 139, which is a serine near the C-terminal end of the protein's amino acid chain. A question that then arose immediately, of course, was, what is the enzyme that brings about that phosphorylation and what might be its role in the response of cells to DNA damage. It took much research to answer these questions, and the answer had important bearing on the cell's response to anticancer drugs, as we shall see. What they did soon find out, however, was that the kind of DNA damage to which gamma- H2AX responded was double-strand breaks (DSB). The response developed within a mere few minutes, during which a large number of gamma-H2AX molecules clustered around each double-strand break. Each of those clusters became visible as "foci": fluorescent spots in the cell nucleus, made visible by means of a fluorescent antibody (Figure 28.4) (Bonner et al., 2008). The amount of radiation damage existing in a cell could be gaged by counting the number of foci in its nucleus, which was equivalent to counting the number of DNA double-strand breaks! More practical was to measure the intensity of the fluorescence produced by the gamma-H2AX antibody in a cell or tissue sample. The repair of the DNA double-strand breaks (DSB) could be seen as the gamma-H2AX foci disappeared. The appearance and disappearance of these foci became a powerful tool for studying the effects of anticancer drugs and the cell's response. Also, it turned out that other proteins clustered together, some within the gamma-H2AX foci and some in foci of other kinds. Later, we will discuss some of those proteins and their bearing on how proteins in the cell signal the presence of DNA damage and protect the cell by inhibiting the cell division cycle or kill the cell by inducing programmed cell death (apoptosis). 560 K. W. Kohn Drugs Against cancer CHAPTER 28 Bonner's group also found that, co-localized with those same gamma-H2AX foci, were the DNA repair proteins Rad50, BRCAl, and Rad51 (Rogakou et al., 1998). (Figure 28.5). The DNA repair functions of these proteins was already discussed in the precious two chapters. Clearly and remarkably, therefore, a set of DNA repair proteins clustered around each DNA double-strand break, consistent with the DSB repair models discussed in those chapters. The different repair proteins, however, did not appear in the foci all at the same time. The gamma-H2AX foci appeared first, shortly after the cells were irradiated. Brcal then gradually appeared in the foci, followed by Rad50 (Paull et al., 2000), which exists in a complex of three proteins: Mrell:Rad50:Nbsl (MRN complex), whose function was discuss later in the previous chapter. So, what did this method reveal when applied to anticancer drugs? In cells treated with camptothecin, DNA double-strand breaks appeared, but only in cells that were in S phase (Bonner et al., 2008) (Figure 28.4). This was consistent with how camptothecin perturbs DNA replication (see Chapter 11). Figure 28.1. William M. Bonner in his laboratory in 1995. Bonner is noted for his landmark discovery of the histone gamma-H2Ax protein and its role in DNA damage repair. After receiving a PhD from Harvard University in Biochemistry and Molecular Biology, working with Nobel Prize winner Konrad Bloch, he carried out postdoctoral studies at Oxford University and the MRC Laboratories in Cambridge, England, where he became interested in histones. He continued this work at NIH as a Staff Fellow in the National Institute of Child Health and Human Development Two years later, in 1976, he moved to NCl's Laboratory of Molecular Pharmacology. In 1980, he identified two important variants of histone H2A, which he named H2AX and H2AZ. (Source: National Cancer Institute; photo by Bill Branson; public domain.) 561 K. W. Kohn Drugs Against cancer CHAPTER 28 \ ,.u·i , / ulA.2121 u2A.U2) - 2. . . ... WW / Xt31 O '7\ \ S.llll 2A. 1111 2".2131 ' zcs, Figure 28.2. Two-dimensional gel electrophoresis, developed by William M. Bonner in my laboratory, showing the separation ofhistone variant proteins (spots below the diagonal). The protein in the spot labeled X(3) was to become known as H2AX (Bonner et al., 1980). A H2AX B H2AX y C H2AX D Figure 28.3. Higher resolution gels disclosed three variants in addition to the main H2AX spot The variant labelled gamma (arrow) increased when the cells had been subjected to radiation: A, unirradiated cells; B· D, cells exposed to increasing doses of radiation. Gamma- H2AX became famous as a measure of DNA double-strand breaks in the cell. (From Ro9akou, Pilch, Orr, Ivanova, and Bonner, 1998,Journal of Bio/09ica/ Chemistry (Rogakou et al., 1998)) 562 K. W. Kohn Drugs Against cancer CHAPTER 28 ., • ,. • ,. • ~ . •"' • I' ~ -.. • . . _..• I • .. " ' fu'"'~ ' . • - \i:~ Figure 28.4. Gamma-H2AX foci in the n uclei of cells exposed to radiation (30 minutes after 1 Gy) (left) or to camptothecin (right). Radiation produced foci in all of the cells, whereas camptothecin produced foci only in cells that were in S phase. (Bonner et al., 2008). Figure 28.5. Gamma-H2AX foci in the nuclei of irradiated cells (left, green); each focus is a cluster of gamma-H2AX proteins at a DNA double-strand break. Another DNA repair protein, Rad50, also clusters in foci (right, red), and co-localizes with the gamma-H2AX foci, as can be seen in the merged image (center), where the coinciding green and red are seen as white. Therefore, gamma-H2AX and Rad50 clustered together around each DNA double- strand break. Nine double-strand breaks are evident in this cell nucleus (white spots in the merged image). Similar experiments showed that Brcal and Rad51 also cluster together in those same foci. (Gamma-H2AX was labeled with a green fluorescent antibody, and Rad50 was labeled with a red fluorescent antibody.) (Rogakou et al., 1998).) References Bonner, W .M., Redon, C.E., Dickey, J.S., Nakamura, A.J., Sedelnikova, O.A., Solier, S., and Pommier, Y. (2008). GammaH2AX and cancer. Nature reviews Cancer 8, 957-967. Bonner, W .M., West, M.H., and Stedman, J.D. (1980). Two-dimensional gel analysis of histones in acid extracts of nuclei, cells, and tissues. Eur J Biochem 109, 17-23. Paull, T.T., Rogakou, E.P., Yamazaki, V., Kirchgessner, C.U., Gellert, M., and Bonner, W.M. (2000). A critical role for histone H2AX in recruitment of repair factors to nuclear foci after DNA damage. Curr Biol 10, 886-895. 563 K. W. Kohn Drugs Against cancer CHAPTER 28 Rogakou, E.P., Pilch, D.R., Orr, A.H., lvanova, V.S., and Bonner, W.M. (1998). DNA double- stranded breaks induce histone H2AX phosphorylation on serine 139. The Journal of biological chemistry 273, 5858-5868. West, M.H., and Bonner, W.M. (1980). Histone 2A, a heteromorphous family of eight protein species. Biochemistry 19, 3238-3245. 564 K. W. Kohn Drugs Against Cancer CHAPTER 29 Qoplu29. fllcoto:ic.b~#OtyOl/ldtlw ATN(JlfN220729#t}3 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@oib gov CHAPTER29 The ataxia telangiectasia story and the ATM gene. In 1941 Madame Denise Louis-Bar, at the time a resident in the neurological clinic of the Bunge Institute at Antwerp, Belgium, saw an unusual case of a 9-year-old boy, who had a combination of neurological and blood vessel symptoms. On the neurological side, the child had difficulty walking, discoordination of movements and slurred speech ("ataxia"), which she attributed to a problem in the cerebellum. There were several known diseases with ataxia. However, the child also had clusters of enlarged small blood vessels ("telangiectasia") in several places, such as the eyes, face, and ears. She thought this combination of symptoms was unusual and wrote up the case in great detail in an article entitled "Sur un syndrome pro9ressif comprenant des telan9iectasies capillaires cutanees et conjonctivales symmetriques a disposition naevoide et des troubles cerebelleux," which was published in Confinia neural. 4:32, 1941 (Boder and Sedgwick, 1958). I could not find the original 1941 article, nor a picture of Madame Louis-Bar. However, her paper became a landmark, and for a time the disease was called "Louis-Bar syndrome." However, there was an earlier report, in 1926, by two Czech neurologists of 3 siblings with this syndrome (Henner, 1968; Syllaba and Henner, 1926) (Figure 29.1). These children's ataxia was in the form of"chorea-athetosis" (involuntary and writhing movements) and they had telangiectasias, but no pulmonary infections. As their family's economic situation deteriorated and the siblings' disability became unmanageable, all 3 of them eventually resorted to suicide (Etzioni et al., 2014). Between 1950 and 1957, eight children were seen with this combination of progressive ataxia of the cerebellar type beginning in early childhood together with striking telangiectasia in the eyes and in areas of the upper face ("butterfly area"). In addition, many of them had recurrent infection of the lungs, sinuses, and middle ear. Often, there were other family members with similar symptoms. This pattern of symptoms was noted by Elena Boder and Robert P. Sedgwick of the University of Southern California, Los Angeles 565 K.W.Kohn Drugs Against Cancer CHAPTER29 Children's Hospital, and Cedars of Lebanon Hospital, who concluded that it was a syndrome, which they named "ataxia-telangiectasia." In 1958, these authors published an extensively detailed report of these cases (Boder and Sedgwick, 1958) (Figure 29.2). But they were still unaware of the 1941 paper by Madame Louis-Bar, until it was brought to their attention by Professor Ludo van Bogaert of the Bunge Institute, who had in fact cared for the patient described by Madame Louis-Bar. However, none of the aforementioned were aware of the even earlier report by Syllaba and Henner (Syllaba and Henner, 1926). Typically, the disorder was first noted when the 2-year-old began to walk and seemed unsteady and clumsy, and the unsteady gait became worse with time. Dilated blood vessels (telangiectasia) in the eyes and over the bridge of the nose typically appeared at the age of 4 or 5. One of the 8 children in the original report was noted to have some grey hair at the age of 2. The children began to speak at the usual time and their intelligence was near normal, but their speech became slurred and eventually it became hard to understand what they were saying. In addition, they had tremor and increasingly discoordinated movements typical for pathology of the cerebellum. As they grew older, the children were much below average height and weight for their age. I was especially moved by the picture of one of their patients, a 9-year-old girl (Figure 29.3). She was a typical case of this disease, which I will summarize below. She had an older sister with this same rare disease; it is hard to imagine the anguish of a family with even one child with this slowly fatal disease. Figure 29.4 shows a close-up of her eye that shows the dilated blood vessels in the corner of the eye and bridge of her nose (telangiectasias). These areas are typically affected in ataxia- telangiectasia. They are areas that could be exposed to sunlight and cause DNA damage that could not be repaired because of a gene mutation in a DNA repair gene characteristic of the disease (to be explained later in this chapter). Occasionally, an older child with this syndrome was found to have a malignant lymphoma or sarcoma. There was sometimes autopsy evidence of premature aging, such as early atherosclerosis and arthritis. Since the ataxia symptoms were of the kind produced by pathology of the cerebellum, special attention was given to that part of the brain in patients who came to autopsy. Although the outward appearance of the cerebellum seemed normal, microscopic examination showed marked degeneration of certain cell types, especially the large Purkinje cells (Figures 29.6 and 7). There will be much to say about how this connects up with the other symptoms of ataxia telangiectasia so as to make a coherent story. But first, what are Purkinje cells? They are large cells arranged side-by-side in a layer in the cerebellum. Coming out of each of them towards the surface of the layer is a large bush of dendrites. Johanne Evangelista Purkinje (Figure 29.8) discovered these cells in 1839, and Santiago Ramon y Cajal developed a stain that clearly showed their amazing axon bushes (Figure 29.6). The Purkinje cell layer is the output funnel for all of the input from the rest of the cerebellum. Their extensive bush of dendrite branches allows them to assemble inputs from a great many other neurons. There is a big burden on the Purkinje cells, however, because they 566 K. W. Kohn Drugs Against Cancer CHAPTER29 have to fire rapidly to keep up with rapid changes in balance and coordinated movements, including the remarkable feats of athletes and circus performers. To carry out their tasks during rapid movements, the Purkinje cells need a large continual supply of energy. The energy is generated in the cells' mitochondria, which produce oxidation products as a side-reaction that can damage DNA. The cells have powerful antioxidants that neutralize most of these molecules, but inevitably some escape and attack the DNA. The cells have mechanisms to repair the DNA damage, and these IDf>Cbaoisms are critically important to prevent the accumulation of DNA damage. We will see that ataxia teiangiectasia is due to an inherited mutation in a DNA repair gene calledA1M; the ataxia may be due to this DNA repair defect having a particularly strong impact on Purkinje cells (Baltanas et al., 201 la; Baltanas et al., 2011b). By 1964, a research team led by Robert A . Good at the University of Minnesota (" the Good guys") showed that ataxia telangiectasia is a primary immunodeficiency disease, which accounts for the frequent infections that were the most frequent cause of death (Peterson et al., 1964) (Etzioni et al., 2014). This however did not explain the ataxia or the telangiectasia. Figure 29.1. Czech Professors Ladislav Syllaba and Kami! Henner, the first to describe patients with the inherited disease that came to be called ataxia telangiectasia (picture from (Etzioni et al., 2014). 567 K. W. Kohn Drugs Against Cancer CHAPTER29 Figure 29.2. Elena Boder and Robert P. Sedgwick definitively described 8 cases of ataxia- telangiectasia in 19S8 (picture from (Etzioni et al., 2014)). Figure 29.3. A 9-year old girl with ataxia-telangiectasia (Boder and Sedgwick's case 1) (Boder and Sedgwick, 19S8). She was unable to walk and hardly able to speak or eat, but her mind was clear; she died 2 years after this picture was taken. I like to think she may sometimes have escaped her distress by living in an imaginative world of adventure, maybe like that depicted by Maurice Sendak in "Where the Wild Things Are" 1963, Harper Collins Publishers. 568 K. W. Kohn Drugs Against Cancer CHAPTER29 Figure 29.4. Closeup of her right eye, showing enlarged blood vessels in the comer of the eye (which might be where exposure to sunlight could occur). Telangiectasia is also seen along the bridge of the nose, which is also an area subject to sunlight exposure (right side of the picture). !-eio=:=J- Middl• ,,.dur.d, I n/crior r>rou11~l$ Figure 29.5. The cerebellum in relation to other parts of the brain. Although smaller than the cerebrum, it contains many more neurons. In contrast to the convoluted gyri of the cerebral cortex, the cerebellum is formed of closely spaced deep grooves that provide a large surface area for a layer of neurons giving an enormous computation power to coordinate body movements. (From Gray's Anatomy, 1918.) 569 K.W.Kohn Drugs Against Cancer CHAPTER29 Figure 29.6. Purkinje cells of the cerebellum. These large cells are lined up in a layer; they have the most extensive interlocking system of branches in the nervous system. This drawing was made by the famous cytologist Santiago Ramon y Cajal in 1899; Instituto Cajal, Madrid, Spain. It shows two Purkinje cells (A) and connections from the smaller granular cells (B). ln ataxia-telangiectasia, both of these cell types are decimated. (From Wikipedia.) '. . ;;,; . .' • \ Figure 29.7. Section through the patient's cerebellum showing Purkinje cells (arrows): they are few and far between, whereas normally there would be many more lined up side-by- side in a layer (from case 1 of Bader et al. 1958 (Bader and Sedgwick, 1958), arrows added). The paucity of these cells explains her inability to carry out controlled movements, but the loss of these and other cells of the cerebellum did not impair her mind. 570 K. W. Kohn Drugs Against Cancer CHAPTER29 Jan Evangelista Purkyne Figure 29.8. Johanne Evangelista Purkinje (1787-1869), Czech anatomist who discovered the Purkinje cells of the cerebellum. He had graduated from University of Prague and worked at University of Breslau. (From Wikipedia). The g ene responsible for ataxia telangiectasia identified at last The next step to locate and clone the gene responsible for the ataxia telangiectasia disease was not undertaken until the 1980's, after the required technology was developed. It was nevertheless a complicated task consuming the efforts of many researchers in several laboratories (Etzioni et al., 2014; Savitsky et al., 1995). The first breakthrough along the way to this accomplishment, reported in 1988 by researchers at UCLA School of Medicine Los Angeles and their collaborators, applied a genetic method called linkage analysis that localized the genome region responsible for the disease to the q22 -23 region of chromosome 11 (Gatti et al., 1988). Mutations in this region seemed to be responsible for most cases of the disease. With the new methods of molecular genetics, a large group of researchers, led by Yosef Shiloh, were then able to isolate and sequence the DNA of a large part of the responsible gene, which they dubbed ATM for ataxia telangiectasia mutated (Savitsky et al., 1995). In this way, the name of the disease evolved from Louis-Barr syndrome to ataxia telangiectasia to ATM disease. By 1963, more than 100 cases of ataxia-telangiectasia had been identified and shown that the disease was inherited as a single mutated gene with an autosomal recessive trait, meaning that the disease occurred only when the person had a mutated ATM gene in both chromosomes 11, having inherited one from the father and one from the mother. Each parent would have had one chromosome 11 with a mutated gene and one with the normal gene, and the unfortunate child would have inherited both defective chromosomes. Those "haploid" individuals (about 1% of the population) who had one chromosome 11 with and one without the mutation were carriers of the ATM disease. They were disease- free but had increased sensitivity to radiation and an increased cancer risk, although their 571 K. W. Kohn Drugs Against Cancer CHAPTER29 risk was much less than in ataxia telangiectasia patients, for whom a cancer of the lymphatic system was the second most frequent cause of death (Etzioni et al., 2014). People who carried one copy of the A TM gene were estimated to have a three or fourfold increased risk of developing cancer, although the increased risk of breast cancer in women was about fivefold and tended to occur at an earlier age (Etzioni et al., 2014; Savitsky et al., 1995). How ATM is activated. In 2003, Christopher Bakkenist and Michael Kastan at St. Jude Children's Hospital in Memphis, Tennessee investigated the phosphorylation of ATM and its consequences in response to DNA damage (Bakkenist and Kastan, 2003). They found that ATM exists as an inactive homodimer that responds to x-ray-induced DNA damage with each ATM unit phosphorylating its partner at serine-1981. Over 50% of the cell's ATM became autophosphorylated within a few minutes after the irradiation. The authors felt that there were too few DNA damage sites for direct action at those sites solely to account for the rapid phosphorylation of such a high fraction of the ATM. The resulting ATM monomers were active protein kinases. The activated ATM then phosphorylated several proteins that had been found to block or delay cell cycle at a checkpoint. These included, among others, p53 and Chk2 that blocked the cells at the Gl/S checkpoint; Nbsl and Brcal that delayed S phase; Brcal also blocked the cells at the G2/M checkpoint Cells from ataxia-telangiectasia patients, both of whose ATM genes were mutated, repaired DNA strand breaks normally; the defect in the cell's DNA damage response was instead thought to be in the cell cycle checkpoints. The many actions ofATM in response to DNA damage. Ataxia telangiectasia (where both chromosomes 11 had a mutated ATM gene) was found to have chromosome instability, meaning that the patients' cells often had an abnormally high frequency of chromosome breakage both with and without DNA damage, induced for example by x-rays (Lehmann and Carr, 1995). The patients were often extraordinarily sensitive to x-rays -- which was first noted when some died even after having an ordinary diagnostic x-ray. The high x-ray sensitivity was noted also in their cells, which were killed by unusually low doses of x-rays or by DNA damaging anticancer drugs. However, the ataxia telangiectasia cells seemed to be able to repair DNA strand breaks normally (Taylor et al., 1975), in contrast to cells from xeroderma pigmentosum patients, which had greatly reduced ability to repair DNA strand breaks (Chapter 22). Evidently, the ataxia telangiectasia genetic defect was not simply in the DNA repair machinery itself. In 1980, Bob Painter at the University of California, San Francisco discovered the main defect in ataxia telangiectasia cells: the cells were defective in responding to damaged DNA by slowing their rate of DNA replication (Painter and Young, 1980) (Figure 29.9). The decreased DNA synthesis rate helped normal cells survive DNA damage by giving the cells 572 K. W. Kohn Drugs Against Cancer CHAPTER29 more time to repair their damaged DNA before initiating mitosis, when persistent DNA damage would produce unrepairable chromosome damage. It was as if the ataxia telangiectasia cells were unable to detect or respond properly to DNA damage. This was the first evidence for a "checkpoint" at which cells checked whether it was safe to proceed along the cell division cycle (Lehmann and Carr, 1995). This kind of"checkpoint" response to DNA damage was found in a wide variety of organisms. Two major checkpoints were at two points in the cell division cycle. The first was at the point in time when the cell begins to replicate its DNA; this checkpoint response (known as the Gl-S checkpoint) would delay the onset of DNA replication. The second (known as the G2-M checkpoint) was at the point in time when the cell begins to condense its chromosomes in preparation for mitosis; this checkpoint would delay the preparation for mitosis. Both checkpoints served to give more time for DNA repair before the cell begins a process during which the presence of DNA damage would likely damage or possibly kill the cell. 100 - 'uii so 90 °'ii-........... 2;; 70 2"-J~ 0 u .9 60 !l f ~ 50 .i ~ 5 ~ 40 < z Q ~ 0 ~ so 1 2 s 4 5 no.., Jerad Figure 29.9. Normal cells (lower curve) responded to x-rays by decreasing their rate of DNA synthesis. Ataxia telangiectasia cells (upper curve) were largely defective in their ability to do that (Painter and Young. 1980). Horizontal axis: radiation dose in thousands of rad. Vertical axis: % of DNA synthesis compared to unirradiated cells. How the ATM gene and protein control the cell cycle checkpoint began to be elucidated in 1992 by researchers led by Michael Kastan at Johns Hopkins Oncology Center in Baltimore, Maryland and by a group led by Albert Fornace in my Laboratory at NIH (Kastan et al., 1992). They found that inducing DNA damage by x-rays in cells from ATM patients did not cause the usual increase the level of p53 protein seen in normal cells. The function of a normal p53 protein was required for the normal cell cycle checkpoint response to DNA damage. (p53 will be the subject of a later Chapter.) This indicated that the checkpoint response required an action of a normal ATM protein on a normal p53 protein: there was a 573 K. W. Kohn Drugs Against Cancer CHAPTER29 pathway from DNA damage to ATM to p53 to checkpoint response. Knocking out p53 prevented the response; there was no response if ATM was mutated (the cells were from an ATM patient) or if the cells' p53 was inactivated by mutations. They also found another factor in the pathway, a protein called GADD45 whose gene was discovered and characterized by Al Fornace in my Lab. DNA damage caused GADD45 to bind tightly to p53, and this did not happen in ATM cells. In the absence of GADD45, there was no checkpoint response to DNA damage. Thus, there was a pathway: from DNA damage to ATM to p53 to GADD45 to checkpoint response. ATM's molecular functions were further investigated in Michael Kastan's laboratory at St Jude Children's Research Hospital in Memphis, Tennessee (Bakkenist and Kastan, 2003; Berkovich et al., 2007). They found that ATM normally exists as an inactive dimer. In response to DNA damage, the two ATM molecules of the dimer phosphorylate each other, thereby causing the dimer to separate to form active monomer units able to phosphorylate the hydroxyls of serine or threonine in several proteins. This dimer separation and activation was somehow brought about through the action of a remarkable trimer that we encountered in Chapter 27B as an agent that detects DNA damage and initiates its repair. The three parts of the trimer were Mrell , RadS0, and Nbsl. The trimer, abbreviated MRN, here again detects the DNA damage and induces the inactive ATM dimer to separate and use its kinase (i.e., phosphorylating) activity The relationship between ATM and MRN turned out to be quite deep. To begin with, ataxia telangiectasia (ATM disease) was found to be closely related to Nijmegen breakage syndrome (NBS) (described in Chapter 27A), both in the symptoms of the diseases and in the functional relationships of their respective genes (Shiloh, 1997). The two diseases shared high sensitivity to radiation, high incidence of cancer, chromosome instability and deficiency of the immune system, and they both lacked the checkpoint responses to DNA damage that delayed the progress of normal cells through the cell cycle. The only major difference was that NBS patients lacked the cerebellar deterioration that caused ataxia in ATM disease. The most frequent cancers of patients with ATM disease were lymphomas and lymphoid leukemias of both the B-cell and T-cell type, which were hundreds of times more frequent than in the general population (Shiloh, 1997). ATM was found to have amazingly many actions in response to DNA damage and coordinating the repair. The gene is large and located in chromosome 11. The domain structure of the protein, summarized in Figure 29.l0A, closely resembled the ATM-related ATR protein as well as the DNAPKcs that was a major player in the repair of DNA double- stand breaks (DSB) (Chapter 27B). A more detailed function diagram of ATM's domains showed the many interactions of this remarkable protein (Figure 29.l 0B) (Phan and Rezaeian, 2021). Particularly notable was the interaction with Nbsl , a component of the Mrell -RadS0-Nbsl (MRN) complex that signals the presence of DNA damage response elements such as delayed mitosis to allow time for DNA repair (Figure 29.11 and Chapter 27 A, Figure 27 A.13). Figure 29.11 summarizes the actions of ATM: By way of MRN and BRCAl, ATM stimulates repair of DNA double-strand breaks. By phosphorylating the checkpoint kinase Chk2, 574 K. W. Kohn Drugs Against Cancer CHAPTER29 which then phosphorylates and inactivates the phosphatase CDC25A, ATM inhibits progress of the cell cycle. In response to DNA damage, ATM phosphorylates and activates p53, which then activates the transcription ofp21/CDKN1A that inhibits the cyclin- dependent-kinase Cdk2 -- which blocks the cell cycle in Gl. P53 also stimulates the transcription of the apoptosis initiators Bax, Bid, and PUMA, causing cells to die by suicide. This is but an outline of the central role of ATM in the cell's complex responses to DNA damage, as currently understood (Phan and Rezaeian, 2021). A S1981t ATM 1 I HEAT repeats FAT I Kinase l 3056 FATC ATR 1 HEAT repeats FATC S2056~ T26~ T395~ ONA-PKcs 1 I HEAT repeats FAT Kinase 14128 ' FATC Figure 29.l0A. Similarity of ATM's domain structure to that of the functionally related ATR and DNAPKcs proteins (Blackford and Jackson, 2017). All three proteins have a serine/threonine kinase domain in the C-terminal regions. They have a so-called FAT domain that serves to bind proteins that the kinase domain phosphorylates. The large HEAT repeat region has multiple functions. It contains repeats, consisting of two helices separated by a short loop, that form a kind of solenoid structure. MM: 3056 aa, 370 kOa In Its monomeric form (active form) 49 alpha helical HEAT motifs ! • Ph0<phorytation sites 1. 3 189. .981 2996 3016 N I 1■•1-....,.._ ATM ktnJSC domJm 'fc Nuc'ear localization sequence NBSl interactkm I 607 aa, autophosphorylation site and interacting with S\lbstrates Ovomatin binding and 10 aa: For Interaction with c•Abl Highty homologous to ttie kinase substrate interaction domain of Pl3K Leocine zipper motif (22 aa): for ATM FATC, For lnteracdon with partners and full dimerization and interaction with partners ATM activation and substrates Figure 29.l0B. A more detailed diagram of the many functions of the various parts of the human ATM protein (Phan and Rezaeian, 2021). By way of its N-terminal region, the 575 K. W. Kohn Drugs Against Cancer CHAPTER29 protein binds to chromatin and to DNA damage response proteins, such a p53. Nearby, there is a nuclear localization sequence that enables the protein to enter the nucleus. The "leucine zipper" motif keeps the protein in an inactive homodimer state until a phosphorylation event serves to activate it Then there are sites for interaction with ABL (see Chapter 14) and NBS (see Chapter 27A). Several of the phosphorylations, as well as the acetylation site at position 3016 at the C-terminus, serve to activate ATM's kinase function. , ONA damage (double strand breaks) 4 -··™•~:;:~ MRN complex-mediated { i MRN complex Mre11/RadS0/N851 .. ~i ~ -- MRN complex Mrcll/RadSO/NBSl ■n:ts.;; - I &.PW ■Ulb- ■- -■-- -■II--■■■-- ■L1UIOtiP w;u~w 11111 ' a.aw - _.J ■-- li&M ! 4+¥M,~ F+&HihllilL_L 1 ■@A I F+IM I I I I l v · FWTIP WM:N '4 iijifofi V I V {, ++iii MHIM ! ONA damaet repair Cell cycle arrest E¥NM#I (ell cydc ched<.polnt Figure 29.11. The multiplicity of the actions of ATM in response to DNA double-strand breaks (DSB). ATM helps DNA repair and signals cell cycle delay and, as a last resort, cell suicide by apoptosis (Phan and Rezaeian, 2021). References Bakkenist, C.J., and Kastan, M.B. (2003). DNA damage activates A1M through intermolecular autophosphorylation and dimer dissociation. Nature 421, 499-506. Baltanas, F.C., Casafont, I., Lafarga, V., Weruaga, E., Alonso, J.R., Berciano, M.T., and Lafarga, M. (201 la). Purkinje cell degeneration in pcd mice reveals large scale chromatin reorganization and gene silencing linked to defective DNA repair. The Journal of biological cliemistry 286, 28287-28302 . Baltanas, F.C., Casafont, I., Weruaga, E ., Alonso, J.R., Berciano, MT., and Lafarga, M. (201 l b). Nucleolar disruption and cajal body disassembly are nuclear hallmarks of DNA damage- induced neurodegeneration in purkinje cells. Brain Pathol 21 , 374-388. Berkovich, E., Monnat, R.J., Jr., and Kastan, M.B . (2007). Roles of A1M and NBSl in cliromatin structure modulation and DNA double-strand break repair. Nat Cell Biol 9, 683- 690. 576 K. W. Kohn Drugs Against Cancer CHAPTER29 Blackford, A .N ., and Jackson, S.P. (2017). ATM, ATR, and DNA-PK The Trinity at the Heart of the DNA Damage Response. Mol Cell 66, 801-817. Boder, E., and Sedgwick, RP. (1958). Ataxia-telangiectasia; a familial syndrome of progressive cerebellar ataxia, oculocutaneous telangiectasia and frequent pulmonary infection. Pediatrics 21 , 526-554. Etzioni, A., Ochs, H.D., McCurdy, D., and Gatti, RA. (2014). Finally Found: The Ataxia- Telangiectasia Gene and its Function. In Primary Immunodeficiency Disorders: A Historic and Scientific Perspective (Elsevier Inc.), pp. 83-86. Gatti, R.A., Berkel, I., Boder, E., Braedt, G., Charmley, P., Concannon, P., Ersoy, F., Foroud, T., Jaspers, N.G., Lange, K., et al. (1988). Localization of an ataxia-telangiectasia gene to chromosome 1 lq22-23. Nature 336, 577-580. Henner, K. (1968). Apropos de la description par MMe Louis-Bar de l'Ataxia teleangiectasia. Priorite de la description par L . Syllaba et K. Henner en 1926, de reseau vasculaire conjonctival. Soc Franc Neurol 118, 60-63. Kastan, M.B., Zhan, Q., el-Deiry, W .S., Carrier, F., Jacks, T., W alsh, W .V., Plunkett, BS., Vogelstein, B., and Fomace, A.J., Jr. (1992). A mammalian cell cycle checkpoint pathway utilizing p53 and GADD45 is defective in ataxia-telangiectasia. Cell 71, 587-597. LPbruann, A..R , and Carr, A.M. (1995). The ataxia-telangiectasia gene: a link between checkpoint controls, neurodegeneration and cancer. Trends Genet 11 , 375-377. Painter, RB., and Young, B.R. (1980). Radiosensitivity in ataxia-telangiectasia: a new explanation. Proceedings of the National Academy of Sciences of the United States of America 77, 7315-7317. Peterson, RD., Kelly, W .D., and Good, RA. (1964). Ataxia-Telangiectasia. Its Association with a Defective Thymus, Immunological-Deficiency Disease, and Malignancy. Lancet 1, 1189- 1193. Phan, L.M., and Rezaeian, A .H . (2021). ATM: Main Features, Signaling Pathways, and Its Diverse Roles in DNA Damage Response, Tumor Suppression, and Cancer Development. Genes (Basel) 12. Savitsky, K., Bar-Shira, A., Gilad, S., Rotman, G., Ziv, Y., Vanagaite, L., Tagle, D.A., Smith, S., Uziel, T., Sfez, S., et al. (1995). A single ataxia telangiectasia gene with a product similar to PI-3 kinase. Science 268, 1749-1753. Shiloh, Y. (1997). Ataxia-telangiectasia and the Nijmegen breakage syndrome: related disorders but genes apart. Annu Rev Genet 31, 635-662. Syllaba, L., and Henner, K. (1926). Contribution in !'independence de l'atheltose double idiopathique et congenital. Rev Neurol l , 541-562. Taylor, A.M., Harnden, D.G., Arlett, C.F., Harcourt, S.A., Lehmann, A.R, Stevens, S., and Bridges, B.A. (1975). Ataxia telangiectasia: a human mutation with abnormal radiation sensitivity. Nature 258, 427-429. 577 K. W. Kohn Drugs Against cancer CHAPTER30 ~t«30. fl111tPA/I.Pstoiyaftdo IWWSO'Otql)'fo,rlOnart/w:nJpyZ2.'111'9eo3 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Laboratory of Molecular Pharmacology Developmental Therapeutics Branch National Cancer Institute Bethesda, Maryland kohnk@oib gov CHAPTER3 0 The PARP story and a new strategy for cancer therapy. Since most chemotherapy drugs damage DNA, the ability of cancer cells, as well as a patient's normal cells, to repair that damage had long been high in the minds of cancer researchers. Therefore, when it was discovered in the 1980's that poly(ADP- ribose)polymerase (PARP) had a role in DNA repair, cancer researchers began intensive studies of the effects of PARP inhibitors, which after many years led to useful new cancer therapy. But what is PARP, and how was it discovered? This chapter is about the PARP enzyme, and the polymer it produces, poly(ADPR). Both seemed unusual and strange at the time of their discovery, yet both turned out to be important players in many DNA repair processes, and PARP became an important target for novel cancer therapies. One of the first clues pointing to PARP inhibitors as potential anti-cancer drugs was that they increased the cell killing potency of alkylating agents (Durkacz et al., 1980) (Durkacz et al., 1981). But a major clinical success eventually came from the remarkable success of PARP inhibitors in patients whose cancers had inactivating mutations in a BRCA gene. Discovery of a strange nucleic acid-like polymer, poly(ADPR). The first time I heard about the newly discovered poly(ADPR) polymer and saw its chemical structure displayed on a poster at a conference in the 1960's, it seemed so bizarre that I had doubts about its existence. I was still imbued with the Watson-Crick lore of nucleic acid structures and could not believe that a chemical structure like poly(ADPR) had any right to exist in biology. 578 K. W. Kohn Drugs Against cancer CHAPTER30 The discovery of this strange-seeming polymer traces back to 1963 to the Institut de Chimie Biologique in Strasbourg, France, where P. Chambon, J.D. Weill, P. Mandel and their coworkers noticed that there was an enzyme in cell nuclei that incorporated NAO (nicotinamide-adenine-dinucleotide, a prominent molecule in much of biochemistry) into an insoluble product that did not dissolve even in acid solvents (Chambon et al., 1966). I don't know whether the experiment that led to this observation was designed or whether it was an incidental observation. When they examined this phenomenon closely, they found that the enzyme's activity was greatly stimulated by DNA, but not by RNA. Chambon and colleagues might not have imagined that this observation of theirs was to develop into a body of knowledge of enormous importance in cell biology and cancer therapy. Figure 30.1 depicts the structure of the new polymer, as originally conceived by Chambon and coworkers in 1966. It was noteworthy that the polymer was made up of NAO units, but without the nicotinamide part. A more up-to-date structure of poly(ADPR) and of its NAO building block is shown in Figures 30.2 and 30.3. After so many years, culminating in so many important findings about its functions, this polymer no longer seems strange at all; in fact, it seems quite natural for it to exist and notable for all the things it does. It's remarkable how new knowledge can convert mystery into mundane experience; which is not to say that there is anything dull about poly(ADPR) or the enzyme that produces it, now known as PARP for.J!oly(ADP-ribose) J!Olymerase. What at first seemed so bizarre about poly(ADPR) was how it violated the rules of ordinary DNA or RNA structures, as many of us conceived them in the 1960's. Although poly(ADPR) was composed of ordinary adenosine-ribose units, these units were linked to each other in a very non-nucleic acid-like manner (Figure 30.3). Instead of the regular head to tail order of nucleotide chains in DNA and RNA, the structural units of poly(ADPR) were linked via their di phosphates and, even stranger, they were also linked directly between their respective ribose units. Stranger yet, and very non-DNA or RNA-like, was that poly(ADPR) was a branched polymer. The branches come out of the second hydroxyl (OH) group of the ribose in the polymer chain. Ribose has two hydroxyl groups, one of which connects in the primary polymer chain, and the branches grow out of the other hydroxyl group on the riboses (Figure 30.3). (As I write this, I am reminded ofa conference I attended in 1951 or 1952 at Harvard during my senior year at the college. The speakers were excited by some of the first information about nucleic acid structure, which they obtained by using the then brand-new technique of ion-exchange chromatography. I recall their exciting conclusion that RNA was a branched molecule! I don't remember the evidence that led to that erroneous conclusion, but it was at least plausible in view of the apparently available extra hydroxyl group on the ribose parts of RNA. One might imagine that branched RNA molecules might once have existed during the early development of life on Earth, during what is called the RNA World, or that polymers of that kind will be found in life that remains to be discovered elsewhere in our Solar System. Linear RNA might have been required for replication. But primitive 579 K. W. Kohn Drugs Against cancer CHAPTER30 RNA is thought also to have functioned as enzymes, which might be helped by a branched structure. The branches might have been added after the RNA chain was replicated.) We can begin to understand the curious structure ofpoly(ADPR) from the fact that the building blocks from which the polymer was made differed from the ones from which DNA and RNA are made. DNA and RNA are composed of nucleotide units, assembled from nucleotide triphosphates, whereas poly(ADPR) is made from nicotine adenine di nucleotides (NAO). NAO had long been known to have essential roles in much of the cell's biochemistry. But to be used to make a biologically important polymer seemed bizarre. Moreover, the nicotinamide part of the NAO molecule was not even in the final polymer. Now, to understand all that, we have to look more closely at the poly(ADPR) structure. We see (reading from right to left in Figure 30.3) that the polymer consists of chains of adenosine-phosphate-phosphate-ribose-adenosine-phosphate-phosphate-ribose- etc. (Ueda and Hayaishi, 1985). So, what happened to the nicotinamide part of the NAO molecule that went into making poly(ADPR)? Why was it not in the polymer? Not only that, but it seemed strange to have one ribose bound to another ribose; I had never seen that before in any nucleic acid-like structure. Those questions whirled in my mind when I first saw the structure of poly(ADPR) displayed on the poster in the 1960's. Soon the fog began to lift however when Hayaishi and coworkers noted that the bond holding the nicotinamide to the ribose is a high-energy bond (Nishizuka et al., 1969). That meant that the bond could easily break. Moreover, the energy released upon breakage of the bond drove the creation of the bond between the two ribose units that connected one unit of the polymer to the next. Thus, it was realized that the departure of the nicotinamide in fact drove the polymerization -- which was why there were no nicotinamide units in the polymer. As already mentioned, the polymer had branches coming out of the ribose's second hydroxyl (OH) group. The polymer could grow a branch out of the second of the two ribose hydroxyls. The branched structure of poly(ADPR) was actually visualized in an electron microscope image in 1991 (Figure 30.4). Next, we should talk about negative charge: The poly(ADPR) polymer has loads ofit, because each phosphate bears a negative charge. Furthermore, the polymer's branched structure concentrates the negative charges into an even smaller volume of space. The poly(ADPR) molecule can be composed of hundreds of units and therefore can be quite large (Figure 30.4). When that happens, an extensive region of concentrated negative charge surrounds the chromatin region where poly(ADPR) polymers were later found to become attached. The concentrated negative charge would loosen the bonds between the DNA strands and between the DNA and its associated proteins, making it easier for the DNA repair machinery to access the sites of damage. That is getting ahead of the story, but it may be useful to have in mind where the story is heading -- particularly about the reason for the concentrated negative charge. The early picture was soon strengthened, when it was 580 K. W. Kohn Drugs Against cancer CHAPTER30 discovered that poly(ADPR) polymers become attached to proteins, particularly the positively charged histones that are associated with DNA in chromatin (Hayaishi and Ueda, 1977). The story, as it unfolded, was that, when DNA is damaged, PARP arrives at the scene, binds to the DNA at the damage site, and causes poly(ADPR) chains to grow from lysine amino acids of nearby histones. The large, branched polymer that hovers over the region of damage tends to make the region negatively charged, which would weaken the bonds holding together the DNA chains and their associated histones. Consequently, it would become easier for the DNA repair machinery to come in and do its job. Pyrophosphatase-;:= =1-- - -- -- - - -- - - - - , +ADPR Figure 30.1. The structure ofpoly(ADPR) as originally conceived by Chambon and coworkers in 1966 (Chambon et al., 1966). The ribose parts are represented by carbon atoms C1through Cs. The repeated unit of the polymer is in the box. 0 NAO NH, H-N.C "¾, 0 0 /~N N"" 1'_ fH)J P O f, 0 Ct-,o.__j N?' -...__ _,.. ''-- o-r" o o )-{ HO OH HO OH Figure 30.2. Chemical structure of NAD (nicotine-adenine-dinucleotide), the building block from which the Poly(ADPR) polymer is formed. The nicotinamide moiety within the red circle becomes cleaved away and does not become part of the polymer. The bond between the nicotinamide and the ribose is a high energy bond; breakage of that bond provides the energy for the polymerization of the poly(ADPR) polymer. 581 K. W. Kohn Drugs Against cancer CHAPTER30 ribose diphospho-i Adenosine- Figure 30.3. Chemical structure of adenosine-diphospho-ribose (ADPR) polymer, showing a chain of ADPR units (poly-AD PR) attached to a protein (left end in the diagram). Additional ADPR units may be connected via the dashed bond at the lower right. The enzyme that assembles the chain is poly(ADPR) polymerase (PARP). The negatively charged phosphates in the structure help to open the chromatin locally to allow repair enzymes to access the DNA at DNA damage sites. The remaining OH groups on the ribose units could attach to another adenosine-diphospho-ribose, leading to a branched polymer structure, which would concentrate the negative charge even more. The polymer can be broken apart by an enzyme (glycohydrolase), which breaks the bond between the ribose units (red arrow). Notice that there is no nicotinamide in the poly(ADPR) polymer. The cleavage of the bond to nicotinamide in NAD provided the energy for the assembly of the poly(ADPR) polymer. Figure 30.4. An electron microscope image ofa large branched molecule ofpoly(ADPR) (de Murcia et al., 1991). The polymer spans a little over 0.1 microns, which could encompass about a hundred base-pairs or 10 turns of DNA 582 K. W. Kohn Drugs Against cancer CHAPTER30 Discovery of the poly(ADPR) polymer and the enzy me that produces it, PARP. As mentioned above, the poly(ADPR) polymer was first observed in the 1960's by P. Chambon and coworkers, who found it as a product made from NAO in a reaction catalyzed by an enzyme in cell nuclei of chicken liver and beef spleen (Chambon et al., 1966). They remarked that, despite being made from NAO, the nicotinamide part of the NAO molecule was absent from the polymer. More information about the new poly(ADPR) polymer and how it is formed soon followed, largely from the laboratory of Osamu Hayaishi in Kyoto, Japan (Nakazawa et al., 1968; Nishizuka et al., 1968; Nishizuka et al., 1967; Nishizuka et al., 1969), who prepared it from rat liver. They noted that free nicotinamide was released from NAD coincident with the incorporation of the remainder of the NAO molecule into the polymer -- which was consistent with polymer assembly being driven by scission of the high energy bond to nicotinamide. The enzyme that catalyzed the polymerization, which became known as PARP, was located exclusively in the cell nuclei, where it grew poly(ADPR) chains onto histones in chromatin, as well as onto the PARP molecule itself. The role of PARP and poly(ADPR) in DNA repair, however, was not to emerge for several years. But first, we should talk about the PARP protein and its DNA-binding and enzyme activities. The PARP1 protein and its domains offunction. Several proteins with PARP-like structure were discovered that eventually were considered to make up a family of at least 18 members. However, only PARPl and PARP2 bound DNA, and PARPl accounted for the great majority of the poly(ADPR) produced in the cell. Mouse embryos survived if either PARPl or PARP2 w as missing but died if both were missing (Ferraris, 2010; Schreiber et al., 2006). An overview of the PARP protein, as diagrammed by (Schreiber et al., 2006) (Figure 30.5) shows the major regions ("domains") with the functions of each domain. The N-terminus of the amino acid sequence on the left and the C-terminus on the right. The amino acids are numbered from 1 to 1,014, starting from the N-terminus. A domain in the N-terminal region was found to be capable of binding DNA, and a domain near the C-terminus w as where the PARP enzyme activity was located. In the central region, there was a region (an "automodification" domain) where the PARPl protein's enzyme activity could grow poly(ADPR) polymer chains onto itself. PARPl was found to have a way of controlling its own enzyme activity. A region at the C- terminus was able to bend back on itsel f to bind and inhibit the adjacent enzyme activity region. When PARPl's N-terminal region binds to a DNA damage site, the bend is relieved, and the enzyme becomes highly active. In that way, PARPl limits itself to generating poly(ADPR) chains only near sites of DNA damage. 583 K. W. Kohn Drugs Against cancer CHAPTER30 DNA-binding Automod ifica tion doma in domain PARPdomain F PARP-1 795 859 Figure 30.5. Diagram of the PARPl amino acid chain, showing the locations of its functional parts (domains), as described by Schreiber et al in 2006. The N-tenninal end of the molecule is on the left; and the C-terminus is on the right. The numbers are the amino acids, counting from the N-tenninus. The N-tenninal region was found capable of binding DNA, and the C-terminal region had the enzyme activity. The automodification domain near the center was where the PARPl molecule attached poly(ADPR) polymer and onto itself (from (Schreiber et al., 2006)). Discovery ofPARP's role in DNA repair. The discoveries that were to implicate PARP and poly(ADPR) in DNA repair, however, began long before anything was known about the polymer or the enzyme that makes it As often happens in break-through research, it all began with a puzzling observation. The first clue in the story goes back to 1956, with a curious observation by I. M. Roitt at the Courtauld Institute for Biochemistry in Middlesex Hospital, London (Roitt, 1956). When Roitt treated cells with an alkylating agent (triethyleneiminotriazine), he found that NAD (nicotinamide-adenine-dinucleotide), a major component in the cell's metabolic network, nearly disappeared. That was the first observation linking a DNA damaging agent with NAD, the building block that was later found to be used by PARP to make poly(ADPR). However, it took an additional 20 years before the fall in NAD levels in DNA damaged cells was shown to be due to consumption of NAD during production of poly(ADPR) for linkage to histones in chromatin (Davies et al., 1977; Whish et al., 1975). ln the meantime, there were over 100 reports about NAD depletion and the function ofpoly(ADPR), but all of them seem to have missed or ignored the relationship to DNA damage repair (Hayaishi and Ueda, 1977). There were just so many possibilities to consider among NAD's many functions in the cell that the DNA repair aspect apparently was not seriously considered. The significance of Roitt's early observation in 1956 remained clouded until 1979, when Sidney Shall and his coworkers at the University of Sussex in England studied the effects of the DNA alkylating agent, dimethylsulfate, on mouse leukemia cells. They made a similar observation to that of Roitt in 1956: the alkylating agent caused a severe fall in the level of NAD in the cell. And as the NAD level fell, the PARP activity rose (Durkacz et al., 1980) (Figure 30.6.); PARP and its enzyme action were by then known, but their role in DNA 584 K. W. Kohn Drugs Against cancer CHAPTER30 repair was just beginning to be revealed. It all suggested that perhaps the NAO was being used up to make a huge amount ofpoly(ADPR). They soon found out that PARP was, in fact, required to repair DNA strand breaks. The Sidney Shall group had used dimethyl sulfate, which was known to methylate the N7- position of guanines in DNA, resulting in dissociation of the methylguanine followed by breakage of the DNA strand at the base-free site. They found that dimethyl sulfate caused NAO levels to fall and PARP activity to rise (Figure 30.6). In addition, they found that, when PARP was inhibited, the survival of the dimethylsulfate-treated cells was reduced, which indicated that PARP helped the cells survive DNA damage caused by the alkylating agent. Then in 1982, Leonard Zwelling and Yves Pommier in my laboratory showed that PARP inhibitors impaired the ability of cells to repair x-ray induced DNA strand breaks (Zwelling et al., 1982) (Figure 30.7). They measured the DNA strand breaks using the alkaline elution technique that we had developed (see Chapter 9). In 1984, Mortimer Elkind and his coworkers then showed that a PARP inhibitor reduced the ability of cells to recover from DNA damage caused by x-rays (Ben-Hur et al., 1984) (Figure 30.8). Taken together, those three studies supported the idea that PARP helped to repair DNA damage produced by alkylating agents and x-rays. Sixteen years later, in 2000, a research group in Newcastle upon Tyne led by Barbara Durkacz and David Newell reported that PARP inhibitors increased the killing of cultured human cancer cells when added to treatment with a topoisomerase I inhibitor (Figure 30.9) (Delaney et al., 2000) (see Chapter 11). The finding of synergy between PARP and a topoisomerase expanded the types of DNA damage whose cell killing was enhanced by PARP inhibitors and suggested that combining a PARP inhibitor with a topoisomerase inhibitor might be clinically beneficial. However, the PARP inhibitors available at that time all had low potency and also inhibited enzymes other than PARP. Those early PARP inhibitors were used in many attempts to pin down an increased anti-cancer cell activity when combined with DNA damaging drugs. But the inhibitors were not good enough to create enthusiasm among clinical researchers. Further research into clinical applications therefore had to await the development of better PARP inhibitors. To recapitulate to this point: After cells were exposed to radiation or alkylating agents, there was a marked reduction in the cell's content of a key molecule of metabolism, nicotinamide adenine dinucleotide (NAO). That was tied to another observation: the DNA damaging agents cause histones (the proteins around which DNA is wrapped in chromatin) to be modified by chains of adenosine-diphospho-ribose (ADPR) becoming stuck to them. The enzyme that catalyzed that reaction was poly(ADPR) polymerase (PARP), whose activity increased when there was DNA damage. It turned out, as inferred, that NAO became depleted because the molecule was used to make the poly(ADP-ribose) chains that were added onto the histone proteins, a process that was a required step in most DNA repair pathways. It seemed surprising that so much poly(ADPR) was made that it actually depleted the amount of NAO in the cell. 585 K. W. Kohn Drugs Against cancer CHAPTER30 B C -e c0 ... I) 0 t0 <( z 10 II hours hours Dimethyl sulfate (µM) Figure 30.6. Treatment of mouse leukemia L1210 cells with dimethyl sulfate, which damages DNA by adding methyl groups at guanines-N7 positions, suppressed NAO levels (A), stimulated PARP enzyme activity (B), and impaired the viability of the cells (C) (Durkacz et al., 1980). The curves from top bottom in A are for increasing concentrations of dimethyl sulfate. The lower curve in C shows the increased cell killing when a PARP inhibitor (3-aminobenzamide) was added to the dimethyl sulfate treatment. The NAO suppression and PARP activity enhancement reverted after several hours, except at the highest dimethyl sulfate concentration. Control,300R JOOR JOOR 300 .- IOmM S..Methvlnl(Ollt\flTUde +SmM J,Aminobenzaml(Je A B I \ • C \ • \ \ • • \ .!2 100 \ \ \ ill \ ~ \ • \ .0 ~ • \ \ \ • • • \ \ "' ~ ~ <( 50 • \ \ ' ,, \ • \ \ '\ z l) • \ '-,..._____ "" 0 ,,_ •A B C 0 5 10 15 0 5 10 15 0 5 10 15 M inutes after x-ray Figure 30.7. PARP inhibitors retarded the ability of cells to repair DNA single-strand breaks (Zwelling et al., 1982). Mouse leukemia L1210 cells growing in suspension culture were exposed to x-rays, which produced DNA strand breaks, mainly single-strand breaks. Panel A shows the rate at which the DNA strand breaks disappeared (were repaired) after exposure of cells to x-rays. Panels Band C showed that PARP inhibitors (5- methylnicotinamide and 3-aminobenzamide) reduced the rate of repair of x-ray-induced 586 K. W. Kohn Drugs Against cancer CHAPTER30 strand break, compared with their rate of repair after x-ray alone (dashed curves). The DNA strand breaks were measured using the alkaline filter elution method we had developed (see Chapter 9). The PARP inhibitors used were among those available at the time, which had low potency and low specificity. They nevertheless indicated that PARP function was required for full effectiveness of the cell's ability to repair DNA strand breaks. $ X 10 l V 79 - 8310H PE=99" (plat. phase) 12 Gy X only X + 50mMNA 3 X 10 · 3 ~~-~-~~~~-~ 0 2 3 4 5 6 Time (hi after X-ray Figure 30.8. An experiment from Mortimer Elkind's laboratory that showed that a PARP inhibitor (nicotinamide, NA) suppressed the ability of cells to recover after x-radiation (Ben-Hur et al., 1984). The vertical axis is the fraction of the cells that were able to grow into colonies on an agar plate. '" Topotecan only · · Topotecan + PARP inhibitor ..- - •.+-- - ,00--,,.-~.., TOf"O'reCANOW) Figure 30.9. Synergy between a PARP inhibitor (NU1025) and a topoisomerase I inhibitor (topotecan) in killing a human colon cancer cell line grown in culture (Delaney et al., 2000). Cells were exposed to topotecan with or without a PARP inhibitor and then tested for survival of their ability to grow into colonies on an agar plate. The graph shows that the PARP inhibitor increased the killing of topotecan-treated cells, as measured by% survival of cells able to form colonies. 587 K. W. Kohn Drugs Against cancer CHAPTER30 How did PARP assist in the repair ofDNA damage, and what would happen if PARP's activity were blocked? By the 1980's there was good reason to think that PARP's synthesis ofpoly(ADPR) helped to repair DNA strand breaks and to repair DNA damage whose repair path included DNA strand breaks along the way. An important discovery supporting that reasoning was that single-strand breaks in DNA induced PARP to synthesize poly(ADPR). Efforts were then made to find out how that happens and what its effects may be. To begin with, PARP was found to bind to DNA single-strand break sites. PARP binding was found to be the earliest response to at least some types of DNA damage. The binding of PARP to a DNA break site was actually caught in electron microscope images in 1994 (de Murcia and Menissier de Murcia, 1994) (Figure 30.10). The DNA binding activated PARP's enzyme domain, which then grew large branched poly(ADPR) chains (up to 200 ADPR units per branched chain) onto several chromatin proteins, including histones. It made sense that PARP would add these ADPR units to chromatin proteins near the DNA break site, where PARP became bound, and it seemed likely that these modifications of chromatin proteins in the vicinity of the DNA damage site would in some way prepare the damage for repair. In accord with that idea, several DNA-repair proteins were found to bind to the DNA-bound PARP. It was suspected also that the high concentration of negative charge conveyed by poly(ADPR) might be important. It all pointed to PARP being the first player to enter the DNA damage repair scene. Figure 30.10. Electron microscope image of a segment of DNA that has a single-strand break (white patches between the dashed red lines) and a molecule of PARP bound to the break (white arrowhead). The break caused a bend in the DNA, which was accentuated by the bound PARP. (From (de Murcia and Menissierde Murcia, 1994), modified by addition of the dashed red lines.) However, the cell had to limit the amount ofpoly(ADPR) allowed to accumulate, as well as the extent of the concurrent drop in NAO level that the cell could tolerate. In 1992, an enzyme (a glycohydrolase) was found that addressed that problem. The glycohydrolase 588 588 K. W. Kohn Drugs Against cancer CHAPTER30 broke down the poly(ADPR) chains within minutes of their production (Althaus, 1992; Pieper et al., 1999), and was surmised to be important for balancing the production and removal of poly(ADPR) at the DNA damage sites. Also, the PARP molecule inactivated its own enzyme activity by growing poly(ADPR) onto itself, which was thought to cause PARP to dissociate from the DNA. A sufficiently large amount of DNA damage, together with the large amount of PARP that existed in the cell, could however overwhelm the glycohydrolase's capacity to eliminate the huge amount of poly(ADPR) that could be produced; the cell could die for lack of NAD. The earliest model of how PARP works may have been that proposed by Tom Linda! and Masahiko Satoh in 1992 (Satoh and Lindahl, 1992) (Figure 30.llA). In 1995, Lindahl presented a somewhat more developed model (Lindahl et al., 1995) (Figure 30.11B). PARP was known to be divided into two separable parts, both of which were required for its ability to produce poly(ADPR) polymers. In addition, DNA was required for this activity (Nakazawa et al., 1968; Nishizuka et al., 1968; Nishizuka et al., 1967; Nishizuka et al., 1969). Lindahl's diagrams included much of what was then known. He depicted the PARP molecule as made up of a DNA-binding domain, connected by way of a middle segment to an enzyme domain that synthesized poly(ADPR) when the DNA-binding domain was bound to DNA. His diagrams showed poly(ADPR) chains becoming attached onto the PARP molecule itself. However, his diagrams did not show poly(ADPR) chains becoming attached to histones, evidence for which had already been reported by Nishizuka and coworkers in 1968 (Nishizuka et al., 1968). Lindahl may have been unaware of that older evidence, or perhaps thought it was not securely enough established to include in his model - or perhaps he felt that the reason for the histone binding was unknown and not relevant to what he wanted to show. The steps of PARP's actions, as understood by Lindahl in 1992 and 1995, were: (a) PARP's DNA-binding domain binds to the site of damage, while DNA repair proteins wait in the wings; (b) PARP's enzyme domain adds poly(ADPR) chains onto itself, particularly onto the region that connects the two domains of the PARP molecule; (c) the PARP molecule is then repelled away from its binding site on the DNA, and repair proteins come in and bind to the vacated site. PARP adds poly(ADPR) chains to other molecules, such as histones, but only while it is attached to the DNA break Two decades later, the picture had filled out, as shown in Figure 30.12. In 2017, Lord and Ashworth (Lord and Ashworth, 2017) showed the elegant way that PARPl efficiently administers the early steps in DNA repair. Here is their concept of how it works (the Roman numerals refer to the designations in Figure 30.12): (i) The domain structure of PARPl is shown from the N-terminus on the left to the C- terminus on the right. The diagram shows the Zn-fingers (ZnF) (the DNA-binding elements) at the N-terminus and the catalytic domain near the C-terminus. (ii-iii) PARPl recognizes and binds to a DNA strand break (accomplished by 3 Zn-finger structures at the N-terminus of the PARPl molecule). (iv) PARPl's catalytic domain near the C-terminus then binds a molecule ofNAD (diamond shape in the Figure). The NAD concentration in the cell must be high enough for an NAD molecule to bind and allow poly(ADPR) polymer to assemble. This normally limits the rate 589 K. W. Kohn Drugs Against cancer CHAPTER30 at which NAO is consumed, thereby tending to avoid reducing the NAO store to dangerously low levels. (v) The catalytic domain, now activated by NAO, assembles poly(ADPR) chains onto histones in the vicinity of the break, as well as to proteins of the repair machineries and to PARP itself. The poly(ADPR) additions help to loosen DNA-bound proteins and thereby allow access of repair proteins to the DNA damage site. (vi) The DNA-bound PARPl adds poly(ADPR) chains onto itself1 Neat! Because that causes the PARPl molecule to release from the DNA and complete the cycle. (vii) An inhibitor of PARP's enzyme activity would prevent the addition of poly(ADPR) chains to chromatin proteins as well as onto itself, which would prevent the PARP molecule from dissociating from the DNA. The inhibitor would thereby keep the PARP molecule trapped on the DNA, which would block DNA replication or RNA transcription attempting to pass through that point on the DNA, potentially killing the cell. This is where there was an opportunity to make potentially therapeutic inhibitors that would mimic NAO and bind PARPl the way NAO does. An NAO mimic, ifit binds PARP stably, would prevent future reaction steps, thereby locking PARPl onto the DNA- which could be lethal to the cell, unless fixed by a complicated repair machinery. It was estimated that there are typically about one million molecules of PARP in a mammalian cell, tightly bound to chromatin in the nucleus. This large number of PARP molecules scattered about in the chromatin was thought an efficient way for them to find DNA damage sites quickly, wherever they may be, as the first step in damage detection and setting in motion the DNA repair process. Binding to a DNA damage site would activate the poly(ADPR) production by PARP. But, if unchecked, this action might consume so much NAO that the cell could die for lack of that essential metabolite. The addition of poly(ADPR) polymer onto itself was a self-inhibitory action that reversed the PARP-DNA binding and stopped further PARP activity. The rapid breakdown of poly(ADPR) by glycohydrolase was another essential part of the PARP control mechanism. The PARP trapping scenario was investigated further by Junko Murai, Yves Pommier and coworkers in our laboratory. They showed that, when bound to an inhibitor (niraparib or olaparib), PARP becomes trapped exactly at sites of DNA single-strand breaks (Murai et al., 2012). They found that, when PARP has bound to damaged DNA, it prevented the DNA strands at a single-strand break from swiveling around each other. They diagrammed their concept of how PARP becomes trapped at DNA single-strand breaks using the notation we had devised to make unambiguous molecular interaction diagrams (Figure 30.13). 590 K. W. Kohn Drugs Against cancer CHAPTER30 A ) Figure 30.llA. The concept of PARP trapping, according to a PARP function scheme proposed by Satoh and Lindahl in 1992. The PARP molecule was considered to be made up of a DNA-binding domain, shown on the left in A, and a separate enzyme domain on the right. The region of the molecule between those two domains was where they thought PARP added poly(ADPR) onto itself. A shows the PARP molecule and a segment of DNA that has a strand break. B shows the PARP molecule bound to the DNA strand break by way of its DNA-binding domain. In C, the enzyme domain would normally grow a poly(ADPR) chain onto the intermediate region of the molecule, which would release the PARP molecule, shown in D, and allow repair of the strand break (not shown). A PARP inhibitor would block the enzyme domain's ability to make the polymer chain. Therefore, the PARP molecule would remain trapped at the DNA damage. (From (Satoh and Lindahl, 1992), modified and simplified.) 591 591 K. W. Kohn Drugs Against cancer CHAPTER30 (a) ,v~,v '\V.nVI PARP's enzyme domain Repair·~· enzyme (b) ./ '\~ --PARP :'/0\Yd fnhibilOr& ----l t PARP's ONA-binding domal. \ , (c) (d) Figure 30.11 B. A more developed view by Tom Lindahl in 1995 of the role of PARP and poly(ADPR) in DNA repair (From (Lindahl et al., 1995) with text in red added). ARTcatalytJC clorroin PARP trapped on DNA. bbeks ' .E . ~~ "Trapping·· Auto•Parylabon C, Bnnd>ed ~ADPR) re~icati~n fcrk progression, (e(J.rtesHRR PARPI stbstrate~ II for repair. protein _..., T ...) _ w. 111 Figure 30.12. The beautiful way that PARPl efficiently administers DNA repair, as depicted in 2017 by Lord and Ashworth. (From (Lord and Ashworth, 2017) with added label.) See text for explanation. 592 K. W. Kohn Drugs Against cancer CHAPTER30 e ■kJ§#ffiffiJ\HH SSBs iiilii 11111 .LLLLI .uJJ.L a Figure 30.13. How PARP interacts at DNA breakage sites, as conceived by Murai, Pommier and coworkers (Murai et al., 2012). The interactions are diagrammed using the notation for molecular interaction maps (Kohn, 1999). The N-terminal end of the PARP molecule is at the left; the C-terminal end is at the right The DNA-binding domain is indicated in blue; the enzyme domain is in red; the region where PARP can bind poly(ADPR) chains onto itself is in yellow. The diagram shows the main features of how PARP binds to DNA strand breaks and how PARP inhibitors can trap PARP on the DNA damage site. The main interactions are: (a) the DNA-binding domain of PARP binds to single-strand breaks in DNA; (b) when that binding has occurred, it stimulates the catalytic activity of the enzyme domain; (c) the enzyme domain then adds poly(ADPR) chains to the central region of the molecule; (d) those poly(ADPR) chains then release PARP from the DNA strand break. PARP inhibitors block the production ofpoly(ADPR) polymers, thereby preventing PARP from adding poly(ADPR) onto itself and preventing PARP from dissociating from the DNA PARP is able to add poly(ADPR) to other molecules (not shown), but only while attached to the DNA break Finding better PARP inhibitors. Much of the desire for PARP inhibitors was fueled by its role in DNA repair, and by early signs that combining PARP with a DNA-damaging drug enhanced the killing of treated cells (Smulson et al., 1977). As the PARP story developed, the number of research programs aiming to discover PARP inhibitors that could be used for treatment of patients mounted 4- fold during the 1990's (Ferraris, 2010). Although at least 18 members of the PARP family of proteins were eventually discovered, only PARPl and PARP2 bound DNA, and PARPl accounted for the great majority of the poly(ADPR) produced in the cell (Ferraris, 2010). Therefore, inhibitors were sought primarily against PARPl. Then, in 2005, the stature of the search for better PARP inhibitors for cancer treatment increased enormously, because of an amazing discovery that wove PARP together with defects in certain genes. ?ARP-inhibitor drugs were found to be highly effective in patients with breast or ovarian cancers of a certain type, namely, cancers whose BRCAl or BRCA2 genes were defective. The PARP-BRCA story is told later in this chapter. But first we focus on discoveries of PARP's role in DNA repair and therapeutics 593 K. W. Kohn Drugs Against cancer CHAPTER30 independent of the BRCA status of the patient However, it was first necessary to develop better PARP inhibitors. The easiest route to an inhibitor was to target the enzyme site that binds NAO for use in making the polymer. An NAO-like molecule could bind to the enzyme site on the PARP molecule and prevent poly(ADPR) production. As already explained, Poly(ADPR) production would cease, and PARP would remain trapped on the DNA. In other words, if the drug only inhibited PARP's enzyme site, its DNA-binding site could still bind to the DNA damage. Then, the PARP molecule would remain bound, because the drug would prevent PARP from growing poly(ADPR) chains onto itself to release the PARP molecule from the DNA (Figures 30.llA, 12, and 13). Furthermore, the trapped PARP would block normal events seeking to access or pass through that location on the DNA. The result would be an anticancer action by the inhibitor-bound PARP that would be trapped bound to its DNA-binding site, and the cell would be left with a difficult repair problem. Taking advantage of the synergy between PARP inhibitors and DNA damaging drugs. To review and expand on how the role of PARP in DNA repair was discovered: A major part of the story, as already mentioned, began in 1980 in Sidney Shall's laboratory at the University of Sussex, England, when they noticed that adding a PARP inhibitor increased the lethal effects of a DNA alkylating agent (Durkacz et al., 1980) (Nduka et al., 1980). It seemed that PARP helped cells withstand the lethal effects of an alkylating agent. Research following up on the 1980's findings about combining PARP inhibitors with DNA damaging agents were at first inconclusive, because of the low potency of the PARP inhibitors that existed at the time. At last, nearly 15 years later, organic chemists succeeded in synthesizing more potent inhibitors. Many studies then combined the new PARP inhibitors with temozolomide, an alkylating agent that adds methyl groups to the 06 position of guanine in DNA. Focus on this drug combination was driven by the use oftemozolomide in the treatment of brain cancers, because temozolomide was one of the few anticancer drugs able to penetrate the blood-brain barrier (see Chapter 2). It was assumed that both drugs in the combination, both being uncharged, would be able to pass through the blood-brain barrier into the brain (de la Lastra et al., 2007). Meanwhile, more evidence was obtained that the synergy between PARP inhibitors and DNA damaging agents was due to PARP being required for repair of the DNA damage. PARP was found to bind some of the proteins (XRCCl, DNA polymerase~. and DNA ligase 111) that were part of the DNA base excision repair (BER) mechanism (Dantzer et al., 1999). A knotty problem in chemotherapy was that cancer cells often migrated into the brain and caused brain metastases. To be fully effective when cancer had gone into the brain, the drugs must cross the blood-brain barrier. Many drugs bore a positive or negative charge, which usually blocked their ability to enter the brain. Temozolomide was one of the few 594 K. W. Kohn Drugs Against cancer CHAPTER30 anticancer drugs that could pass into the brain, and it was therefore studied especially for its effect on brain cancers (see Chapter 2). Many PARP inhibitors also were devoid of charge and could pass into the brain. Hence, both drugs in the combination could enter the brain and potentially act on brain cancers or brain metastases. When a PARP inhibitor was combined with temozolomide, the 2-drug combination had a tremendous synergistic effect in prolonging the lives of mice bearing a lymphoma in the brain (Ten tori et al., 2005) (Figure 30.14). Then, in 2008, researchers reported that the most effective among many potential PARP inhibitors that they had up to that time synthetized was olaparib (originally called AZD2281) (Menear et al., 2008). A big advantage of olaparib, as well as PARP inhibitors discovered subsequently was the lack of a positive or negative charge, thereby allowing them to move into the brain and act on cancers there. The ?ARP-inhibitor plus temozolomide combination also had a striking effect on human colon cancers grown in immune deficient mice (xenografts) (Figure 30.15) (Menear et al., 2008). PARP inhibitors approved for cancer treatment come onto the scene. Those findings spurred the search for even better PARP inhibitors that would be effective in cancer treatment The first PARP inhibitor to be approved for treatment of cancer patients was olaparib. It was the first of several structurally related drugs to become approved. Figure 30.16 shows the PARP inhibitors that were in clinical use by 2017, listed according to their potencies for trapping PARP at DNA stand breaks. The most potent was talazoparib, the least potent was veliparib, and olaparib had intermediate potency. Figure 30.17 shows the inhibitors according to their structural relatedness. They all had a structural feature (shown in red) resembling the nicotinamide part of NAO, which indicated that they all acted by binding and inhibiting the NAO site (the catalytic site) of PARP. But they differed in their strength of binding to DNA break sites as shown in Figure 30.16. As of 2019, we still did not know which of the PARP inhibitors was best, because there had not yet been any clinical trials that compared them head-to-head (Mateo et al., 2019). This question however met a complication, when in 2014 Junko Murai and Yves Pommier in our laboratory and their colleagues noted that PARP inhibitors had two separable actions (Murai et al., 2014a). They compared olaparib, rucaparib, and talazoparib (BMN 673) for inhibition of enzyme activity, ability to trap PARP at DNA strand breaks, and killing of cancer cells. They found, remarkably, that, although all three drugs inhibited PARP enzyme activity with comparable potency, talazoparib was about 100 times as potent as the other two drugs when it came to potency of trapping on DNA. Also, talazoparib in combination with temozolomide was more potent in killing cancer cells compared to other PARP inhibitor-temozolomide combinations. It seemed that a major cell toxic effect of the inhibitors came from trapping PARP to DNA, not merely from inhibiting PARP enzyme activity. 595 K. W. Kohn Drugs Against cancer CHAPTER30 A notable fact about talazoparib was that its stereoisomer, where the configurations of its sites labeled Rand Sin Figures 30.16 and 30.17 were reversed, was nearly inactive (Murai et al., 2014a). That meant that the 3-dimentional geometry oftalazoparib was critical in allowing the drug to bind to its site on the PARP molecule; the mirror image of the inhibitor would not fit at the binding site. Figure 30.18 shows an experiment that demonstrated the strong synergy between talazoparib and temozolomide (Shen et al., 2013). 100 ;;; 80 -~ t': :, 60 -.,.., (/) C: 40 No drug Temozolomide + PARP inhibitor ., ~ CL 20 0 10 15 20 25 30 35 Day Figure 30.14. A PARP inhibitor (GPI 15427) together with temozolomide greatly increased the survival of mice that had a lymphoma in the brain. The PARP inhibitor by itself or temozolomide by itself had much less effect (Tentori et al., 2005). Both drugs were able to pass through the blood-brain barrier. 596 K. W. Kohn Drugs Against cancer CHAPTER30 ♦ Vehicle 8 • 10mg,1(g Cpd •7 (olaparib) ½ 50mg,1(g TMZ (temozolo mide) Q) E 7 + 10mg,1(g Cpd 47 + SOmg/kg TMZ :, 0 no drug > 6 :i olaparib ~ 5 :, f- 4 Q) > ~ 3 ai a:: 2 temozolomide C ~+ o laparib l.'l 1 ::;; 0 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 Time (days} Figure 30.15. The PARP inhibitor, olaparib, together with temozolomide had an impressive synergistic effect in suppressing the growth of human colon cancer cells (SW620) in mice (Menear et al., 2008). The size of the tumor (vertical axis) is plotted against time (horizontal axis). PARP trapping potency (high to low) O~N H, ~ N ,H 'N~\_/ F 1 Talazoparib 2 Niraparib H 0 -~~ Hf O~,N F ("'•,,.J...__ '- I I I N__)' V N 0 0 3 Rucaparib H 4 01aparib OXN_H, lX17Q H H 5 Veliparib Figure 30.16. PARP inhibitors that were used in cancer treatment as of 2017, showing their relative potencies from 1 (most potent) to 5 (least potent) for trapping PARP on DNA, which correlated with potency for killing cells in culture (Lord and Ashworth, 2017). 597 K. W. Kohn Drugs Against cancer CHAPTER30 0 il N N ,/ C- ,CHD-P 0-P-0-C> ~- N NAO ,..,,,,,, ' ' o ... , - , , -UOOOH O O Y. / HO Ot 0 SH Rucaparib Veliparib CH, ~H I~ NH NH f CH Talazoparib 0 NH, NH 1, N ~ 'i" t .() Niraparib --0~-·Q NH f f O cN'b !_i" Y£7016 I I f"'-i 0 .# 0 ~OH Figure 30.17. PARP inhibitor drugs showing their chemical structure relationships. A feature they had in common with NAO is shown in red (Pommier et al., 2016). This feature is like the nicotinamide part of the NAO molecule. It is the part of the molecule that binds to the active site of the PARP enzyme, which breaks its bond to the rest of the molecule. The cleavage of that bond drives the poly(AOPR) polymerization reaction (blue arrow in NAO structure). The nicotinamide-like part of the inhibitor structure is attached by multiple bonds to the rest of the structure and cannot be released. Therefore, when inhibitor binds to the enzyme at the nicotinamide-binding site, the inhibitor remains stuck there, because the reaction to release the nicotinamide-like part cannot occur. 598 K. W. Kohn Drugs Against cancer CHAPTER30 • TMZ only • BMN 673 (40 nmol/L) ~ 100 t,. BMN 673 (10 nmol/L) ! - e C 80 • 0 (.) ~ 0 60 .c j 0~ 40 Ol Q) 20 () 0 0 100 200 300 400 Temozolomide [µmol/L] Figure 30.18. Synergy between the DNA alkylating agent, temozolomide, and the potent PARP inhibitor, talazoparib (originally called BMN 673) (Shen et al., 2013), in inhibiting the growth of a human colon cancer cell line, called LoVo. Temozolomide by itself had little effect (upper curve). Talazoparib by itself inhibited cell growth by about 40% (lower curves at zero temozolomide); when temozolomide was added at increasing concentrations, cell growth was inhibited up to 90%. That was a remarkable degree of synergy between the two drugs. PARP has a major role in DNA r epair. The poly(ADPR) chains that PARP fixes onto histones in chromatin serve to recruit DNA repair enzymes to a variety of sites of DNA damage (Murai et al., 2012; van Wietmarschen and Nussenzweig, 2018). Evidence explained earlier in this chapter pointed to single- strand breaks and base-free sites as the DNA lesions where PARP comes into play during DNA repair. It therefore became important to know which drugs produce such DNA lesions. For temozolomide, the DNA damage and repair scenario was described in Chapter 2. The most important damage it caused was methylation of guanine-06 sites on DNA, in other words, the addition of a methyl groups to the 06 position of guanines, yielding a potentially toxic DNA product A wonderful enzyme was discovered that efficiently plucked off the methyl group off and restored the normal DNA structure. The enzyme that did that was 06- methylguanine-methyl-transferase (MGMT) (see Chapter 2). Although that enzyme was highly efficient, it was not foolproof (no biological process is). Moreover, some cancer cells made little or no MGMT. As discussed in Chapter 2, that made those kinds of cancers vulnerable to treatment with drugs like temozolomide, because the cells did not have 599 K. W. Kohn Drugs Against cancer CHAPTER30 enough MGMT to remove the methyl groups from all the methylated guanines. The remaining methylguanines then had to be repaired by a different mechanism: DNA base excision repair (BER), whose discovery was described in Chapter 24. BER plucks the bad base off the DNA, leaving a base-free site in the DNA chain: a deoxyribose unit without a base (adenine, guanine, thymine or cytosine) attached to it In the case of temozolomide, the BER mechanism removed the persistent O6- methylguanines in toto and left behind a base-free deoxyribose in the DNA. To PARP, the base-free site may look sufficiently like a single-strand break for it to bind there and bring in some of the needed DNA repair proteins. Those repair proteins, it seemed, removed the base-free deoxyribose and filled the resulting gap in the DNA strand, but left behind a single-strand break. It seemed that the PARP molecule would remain attached and continue to help in the repair. Although the essentials may be correct, the mechanism may be more complicated than the relatively simple picture had suggested. A collaboration of several laboratories in 2017 enabled the use of new physics techniques to investigate in amazing detail some of the events during the repair of base-free sites at the level of individual molecules (Liu et al., 2017). Their essential findings were: (1) PARPl and the endonuclease (APEl) that cleaves away the base-free deoxyribose both move around freely in search of base-free sites on a DNA molecule. (2) A PARPl molecule can bind to a single-strand break, base-free site, or DNA end. (3) When PARPl and APEl are bound to the same base-free site, the PARPl molecule is enabled to slide along the DNA while remaining close to the damage site; perhaps that opened the damage site for access to other repair proteins. (4) Addition ofpoly(ADPR) chains onto itself allowed the PARPl molecule to slide for greater distances along the DNA in the vicinity of the damage site. (5) The PARP inhibitor, olaparib, did not affect the dissociation of PARPl from the DNA, but increased the ability of PARPl to slide along the DNA for greater distances away from the damage site. A caveat of this investigation, however, was that the DNA they were able to study was bare, without any histones or other chromatin proteins. Nevertheless, they demonstrated a new technology for detailed investigation of how DNA repair works. In general, drugs that caused DNA damage requiring base-excisions repair (BER) for good cell survival should be sensitive to PARP inhibitors, because BER entails the production of DNA single-strand breaks during the repair process. Aside from drugs like temozolomide that damage DNA bases directly, base analog drugs, such as thioguanine and 5-fluorouracil, become incorporated into DNA by the DNA replication machinery that is unable to distinguish the base analog nucleotide triphosphate from the normal nucleotide triphosphate. The DNA polymerase thus adds the bad base into the DNA chain. The bad base could then be removed by BER, leaving a base-free site and leading to a single-strand break where PARP would bind. 600 K. W. Kohn Drugs Against cancer CHAPTER30 How do PARP inhibitors exert toxic effects on cells? In 2014, as already mentioned, my colleagues Junko Murai and Yves Pommier, together with other researchers in our laboratory, gave us more insight into the workings of the PARP drug combinations. They measured the toxic effects on cells when a PARP inhibitor was combined with a DNA damaging drug (Murai et al., 2014b). Insight into the effects of PARP came from measuring what happens in cells whose PARPl gene was deleted. Cells that had no functional PARPl had increased sensitivity to temozolomide, as expected, since PARP was necessary for efficient DNA damage repair. Also as expected, PARP inhibitors had no effect on cells that had no functional PARPl gene. But quite remarkably, cells having normal PARP genes were extremely sensitive to the combination oftemozolomide and a PARP inhibitor. The inference was that, even though PARP itself helped to repair the DNA damage, the ?ARP-with-bound-inhibitor unit became trapped at temozolomide-induced DNA damage sites, resulting in extreme toxicity. It was all in accord with the idea that PARP inhibitors cause PARP to become trapped at sites of DNA damage, thereby producing severe toxicity to cells, possibly more severe against cells in a cancer than against critical normal cells in the body. Hence, it seemed that PARP inhibitors had two different kinds of toxic effects on cells: (A) inhibition of poly(ADPR) production and (B) trapping of PARP at sites of DNA damage, particularly at strand breaks. The different ?ARP-inhibitor drugs had similar potencies for mechanism A, but different potencies for mechanism B. Moreover, the major toxic mechanism appeared to be mechanism B (Murai et al., 2014b). In the same paper, Murai et al. asked what kinds of DNA damage would produce the extreme toxicity of PARP trapping. In other words, in what kinds of DNA damage did or did not PARP come in to help repair and at the same time become trapped on the DNA? They were able to infer answers from experiments such as those summarized in the previous paragraph. Further studies of PARP looked to see when and where poly(ADPR) was produced (van Wietmarschen and Nussenzweig, 2018). Since poly(ADPR) is rapidly degraded by glycohydrolase, the researchers inhibited the enzyme, which allowed the polymer to accumulate at its sites of production where it could then be measured. They found that poly(ADPR) was produced only during DNA replication and at or near the replication sites. When they inhibited PARP, the gaps in the replication lagging strand (producers of Okazaki fragments) were not ligated. The gaps persisted into the next replication cycle, and then caused the replication fork to collapse. Repair of that problem required homologous recombination for which BRCAl and BRCA2 were both needed (Gudmundsdottir and Ashworth, 2006). This gave some insight into the BRCA-PARP synthetic lethality strategy for cancer therapy described in a section at the end of this Chapter. 601 K. W. Kohn Drugs Against cancer CHAPTER30 Topoisomerase 1 inhibitors produce another DNA damage scenario requiring PARP for repair. A drug that produced DNA damage of a kind different from that of DNA base modifiers, such as temozolomide, was the topoisomerase inhibitor camptothecin (Chapter 11). ln 2011, my colleagues led by Yves Pommier had, in the vein of our long interest in topoisomerase-targeted drugs, found that PARP inhibitors become trapped at the DNA sites of camptothecin action (Zhang et al., 2011). As explained in Chapter 11, camptothecin was known to produce transient single-breaks that allow swiveling of the DNA strands to relieve the torsional stress that accumulates as DNA is transcribed or replicated. Those single-strand breaks however open and close quickly in a controlled manner -- and do not qualify as DNA damage. Camptothecin however did induce DNA damage, because it bound and stabilized the open state during the topoisomerase I swivel cycle long enough to cause problems. Camptothecin nevertheless dissociated easily from this bound state. As described in Chapter 11, camptothecin induced DNA damage mainly while DNA was being replicated during S-phase of the cell cycle, which suggested that the damage happened when the DNA replication machinery collided with a camptothecin-blocked topoisomerase- 1 site. That idea held up and was extended by evidence that damage was induced, albeit to a lesser degree, by collisions involving RNA transcription. The essentials of an experiment showing these DNA damaging events are shown in Figure 30.19 (Zhang et al., 2011). In general, DNA damaging drugs that required repair by BER, as well as drugs like camptothecin that impaired topoisomerase I, were synergized by PARP inhibitors. On the other hand, drugs like cisplatin that crosslink DNA or drugs that impair topoisomerase II, were repaired by mechanisms that did not require PARP, and these drugs were not synergized by PARP inhibitors (Zhang et al., 2011). A dual role of PARPl and TD Pl in removing trapped Topl-DNA complexes on DNA is diagrammed in Figure 30.20. EdU yH2Ax EdU yH2Ax - E C 0 l) co co Q) ~§ ~CG Figure 30.19. The PARP inhibitor, veliparib (ABT-888), combined with the topoisomerase 1 inhibitor, camptothecin, produced DNA damage, but mainly in cells that were undergoing 602 K. W. Kohn Drugs Against cancer CHAPTER30 DNA synthesis (S-phase) (Zhang et al., 2011). In this experiment, human cancer cells (osteosarcoma U2OS) were treated with camptothecin (CPT) with or without veliparib (ABT-888). Cells undergoing DNA synthesis were labeled by adding a thymidine analog (EdU), which became incorporated into the newly synthesized DNA. The cells were then put on a slide and stained with two fluorescent antibodies that make cells that are undergoing DNA synthesis glow pink and cells that have DNA damage (where yH2Ax is bound) glow green. The left panel shows the results for cells that were treated with veliparib alone. Treatment with camptothecin alone showed some DNA damage (green in upper right of the right-hand panel). But there was much more green when camptothecin was combined with veliparib (lower right in the right-hand panel). Thus, camptothecin produced DNA damage mainly in S-phase cells (green), but the combination of camptothecin and veliparib produced a great deal more DNA damage (more green), mainly in S-phase cells, but to a smaller degree also in non-S-phase cells. (Modified from (Zhang et al., 2011).) Phosphodiesterase pathway --- PARP1 1 -- ~ Topi Figure 30.20. PARPl works together with TD Pl and DNA repair enzymes to cleave away Topl trapped as cleavage complexes on DNA) (Das et al., 2014). (See Figure 10.9 in Chapter 10. TDPl acts similarly to remove trapped Topl and Top2.) The diagram uses the molecular interaction map notation (Kohn, 1999). 1. PARPl binds TDPl. 2, 3 . The PARPl- TDPl dimer stimulates TD Pl to cleave any remaining Topl fragment away from the DNA 4 . DNA enzymes complete the repair. PARPl may also serve to recruit the DNA repair enzymes to the site where they are needed (not shown in the diagram). 603 K. W. Kohn Drugs Against cancer CHAPTER30 PARP inhibitors begin to be used to treat cancer pa tients. The approved PARP inhibitor drugs, however, did not come easily. Far from it! A huge number of compounds were prepared and studied by several pharmaceutical companies to unravel the chemistry of exactly how the inhibitors worked and to prepare some of them for clinical trial. Going back to the early days of development of clinically approved PARP inhibitors, Figure 30.21 lists the PARP inhibitors that had been approved as of 2010 for testing in cancer patients, and the steps in development that each drug had passed (Ferraris, 2010). The first step was to obtain approval of the drug as an Investigational New Drug (IND) by the U.S. Food and Drug Administration (FDA). Approval of an IND depended on animal studies indicating that the drug's toxicities and therapeutic actions are well enough understood to allow preliminary testing in a small numbers of advanced cancer patients for whom there was no longer any approved therapy available. Next came Phase I, which aimed to determine the toxicity and safe dosage limits in a limited number of cancer patients, most of whom had relapsed after chemotherapy and for whom there was no approved therapy available. About 70% of the drugs went on from Phase I to Phase II, where several hundred patients with advanced cancers were treated with the aim of finding evidence that the drug was active against cancer, while continuing to monitor for untoward actions. If Phase II studies provided sufficient evidence of action against some types of cancer, the drug went on to Phase III, which aimed to determine whether the drug was better than the best previous treatment for some types of cancer. Several thousand patients were typically recruited for Phase III studies that were usually carried out in double-blind fashion. Two drugs, at the top of the list in Figure 30.21, reached Phase III. The first was BSl-201, a small molecule whose structure is shown below the table. This structure was simpler than the structures of the drugs (Figures 30.17 and 30.18) that were later approved as drugs in the clinical armamentarium. BSI-201 did not pass beyond Phase III however and was dropped. The next Phase III drug in Figure 30.21, became "olaparib" and was the first PARP inhibitor to become an official clinical drug. The next in the list became "veliparib," and was in Phase II at the time that the list was compiled in 2010. 604 K. W. Kohn Drugs Against cancer CHAPTER30 Table I. Current Clinical Status of PARP-1 Inhibitors' company compd IND phase I phase II phaseIll therapeutic indications 8iPar/Saoofi BSl-201 • • • • lriple negatil~ breast cancer KuDOS/AstraZencca KU59436(AZD8821 . Olaparib) • • • • metastatic breast cancer. advanced ovarian cancer Abboll ABT-888 (Veliparib) • • • metastatic breast cancer, metastatic mekrnoma, braincancer Pfizer AG 14699 (PF-01367338) • • • metastatic breast cancer, ad,~nred ovarian cancer lnotck lNO-1001 • • • malignant melanoma. heartJlung bypass surgery Cephalon CEP-9122 • • advanced sotid tumors Merck MK4827 • • advanced sotid tumors Mitsubishi MP-124 • cerebral ischemia Guilford/MGI/Eisai GPI 21016 (E7016) • glioblastoma 'The asterisk (•l indicates participationin the indicated stage. ,J/""' N02 64 (BSl-201) Figure 30.21. PARP inhibitor drugs approved for clinical test as an lnvestigational New Drug (IND) as of 2010 (see text) (Ferraris, 2010). The combination of a PARP inhibitor and the alkylating agent, temozolomide, began to be used to treat brain cancer patients, because both drugs were able to pass through the blood-brain barrier. Temozolomide, however, was expected to be most effective in cancers that had low levels of the DNA repair enzyme that removes methyl groups from the 06 position of guanine in DNA (DNA-06-methylguanine-methyltransferase, MGMT) (see Chapter 2). Clinical trials of PARP-inhibitor plus temozolomide combination were under way at the time of this writing (Gupta et al., 2018). A response was already seen in a phase 1/11 trial (Figure 31.22). 605 K. W. Kohn Drugs Against cancer CHAPTER30 Baseline Nadir Progression Figure 30.22. Response of a small-cell lung cancer (SCLC) tumor in the lung of a patient treated with olaparib plus temozolomide. The patient had relapsed after previous chemotherapy and was treated with the olaparib-temozolomide drug combination in a phase 1/11 trial. The green arrowheads point to the tumor. The middle panel showed that the tumor had shrunk almost to the point of invisibility. The tumor unfortunately grew again, as shown in the right panel (Farago et al., 2019). A new paradigm for cancer treatment: sy nthetic lethality. An unexpected finding by Alan Ashworth and his colleagues at Guy's Hospital, London, and Cambridge Science Park in 2005 led to a new strategy for cancer therapy, based on a concept called "synthetic lethality" (Farmer et al., 2005). Here is how their unexpected discovery came about: since BRCA enzymes were required for repair of DNA double-strand breaks, and PARP was required for repair of DNA single-strand breaks, they thought that a BRCA mutation together with a PARP inhibitor might make cells particularly sensitive to DNA-damaging agents. Unexpectedly, however, they found that cells whose BRCA function was inactive due to mutation were killed by PARP inhibitors, even without introducing a DNA-damaging drug! It was not clear exactly how PARP produced that effect, but it was nevertheless a clear case, they soon realized, of synthetic lethality that might have therapeutic potential (Figure 30.23.). In the same issue of Nature, Thomas Helleday and his colleagues in the UK and Stockholm confirmed and gave more substance to the BRCA-PARP synthetic-lethality concept (Bryant et al., 2005). Both research teams found that PARP inhibition by itself generated DNA damage that required BRCA for repair and that, if the functions of both BRCA and PARP were lacking. cells died. Lack of BRCAl or 2 function was not by itselflethal to cancer cells, nor did inhibition of PARP by itself kill the cells. However, the combination of a non- functioning BRCA and an inhibitor of PARP killed them. That was the essence of synthetic lethality (Figure 30.23). Synthetic lethality, as first defined in 1946 by Dobazhansky (Dobzhansky, 1946) was a situation where the combination of 2 genetic changes was lethal, whereas either genetic change by itself was not Figure 30.23 shows how synthetic lethality produced by combining a BRCA defect with a PARP inhibitor was a strategy for cancer therapy. 606 K. W. Kohn Drugs Against cancer CHAPTER30 Body cells usually have two copies of each BRCA gene, one on each of two homologous chromosomes. In some families, one of the two copies (most commonly of the BRCAl of BRCA2 gene) has an inherited mutation that destroys its function. As long as the second copy is ok, the cell has normal BRCA functions. However, a random mutation can inactivate the second copy and frequently occurs as an initiating event in cancer particularly of breast cancer. Family members who had inherited an inactivating mutation in one copy of a BRCA genes tended to have an unusually high incidence of breast cancer (see Chapter 26). They were vulnerable to developing cancer when a random mutation inactivated the second copy of the same BRCA gene. With BRCAl or BRCA2 totally inactive, the cancer cells could not repair DNA by homologous recombination. For repair of DNA damage, such as double- strand breaks, the homologous recombination repair option would not be available to them. Such cancer cells could then be killed by PARP inhibitors that blocked the other major DNA repair pathways. People whose BRCA genes were all normal had a much lower risk of breast cancer. However, a first inactivation of a BRCA gene could occur as a rare random mutation, for example in a normal breast cell from which a cancer could eventually develop. That was thought to be a route by which cancer could arise even when there was no family history of cancer. In both the familial and the random circumstance, however - and this is the critical point - BRCA function would be defective in the cancer, but not in the normal cells. That difference was the basis for therapy: the cancer cells were vulnerable to PARP inhibitor, while the normal cells was not. It was a case of synthetic lethality with selectivity against cancer cells (Kaelin, 2005). Synthetic lethality in cancer therapy: Cells die if both Gene A and Gene Bare defective or inhibited by drug. Normal cell Cancer cell with defective gene A Gene A - nor mal Gene A - normal Gene A - defective Gene A - defective Gene 8 • normal Gene 8 - inhibited Gene 8 • nor mal Gene B - inhibited by drug by drug l Viable Viable l Viable Lethal Gene A = BRCA Gene B = PARP Figure 30.23. BRCA-defective cancer cells are killed by PARP inhibitor. This was the first therapy based on synthetic lethality, where cells die if an only if their Gene A and Gene B are both defective or drug-inhibited. 607 K. W. Kohn Drugs Against cancer CHAPTER30 The BRCA-PARP synthetic-lethality strategy showed its promise against cancers in 2009 in a phase I study of ovarian cancer patients who had an inherited BRCAl or BRCA2 mutation in one allele. Such patients were prone to develop cancer from cells that spontaneously acquired an inactivating mutation in the second allele. These cells lacked BRCAl or BRCA2 function. Since both BRCA genes were needed for homologous recombination, the cells were unable to carry out this DNA repair function. Consequently, more mutations accumulated, leading to a high probability of cancer. Those cancers had a non-functional BRCA, making them vulnerable to PARP inhibitors, based on synthetic lethality when both of those genes were non-functional. Figure 30.24. shows the response to olaparib (an inhibitor of PARP) of an ovarian cancer in a patient who had a BRCAl gene mutation (Fong et al., 2009). Olaparib was approved in 2014 by the US Food and Drug Administration (FDA) for treatment of ovarian cancer in patients with a BRCAl mutation. Another PARP inhibitor, talazoparib, was approved in 2018 for treatment of breast cancer patients who had a BRCA mutation (Turk and Wisinski, 2018; Zimmer et al., 2018). Before treatment 4 months after t reatment with olaparib. with ola parib. Figure 30.24. Regression of an ovarian cancer in response to treatment with the PARP inhibitor, olaparib; the patient had an inherited mutation of the BRCAl gene. The tumor disappeared completely in these computerized tomography (CT) scans of the abdomen (red circles) (Fong et al., 2009). Unfortunately, however, only about half of the BRCA mutant ovarian cancer patients responded to olaparib, the response was rarely complete, and the duration of response was 8 months at best (Fong et al., 2010). To begin to understand how the remarkable cooperation between the BRCA and PARP genes led to new cancer drug therapy, we must go back to the story of PARP (see above). PARP is abundant in the cell nucleus, where it catalyzes the production ofpoly(ADP-ribose) polymers and attaches them to certain essential cell proteins, particularly histones, in chromatin. It is an essential part of the molecular machinery that repairs DNA single-strand breaks and other types of DNA damage. However, there is an effective backup mechanism to deal with DNA breaks that may remain unrepaired: homologous recombination repair of DNA double-strand breaks, which can form from unrepaired single-strand breaks, and requires both BRCAl and BRCA2 (see Chapter 27 A) (Gudmundsdottir and Ashworth, 608 K. W. Kohn Drugs Against cancer CHAPTER30 2006). Therefore, if both PARP and either BRCAl or BRCA2 are defective, then the cells die. That, in short, was the basis for the excitement engendered by prospects for therapy specific for cancers that have inactivating mutations of BRCAl or BRCA2. Treatment of such cases with PARP-inhibiting drugs would cause the cancer cells to die. More recently, Niek van Wietmarschen and Andre Nussenzweig at the National Cancer Institute reviewed evidence, some of it from our Laboratory (Murai et al., 2012), that suggested that the normal production of Okazaki fragments in the replication of the lagging DNA strand could lend itself to PARP-BRCA synthetic lethality (van Wietmarschen and Nussenzweig, 2018). The sealing of the fragments would be impaired by inhibition of PARP and would then rely on BRCA-dependent homologous recombination to deal with the difficulty. Consequently, this could be a of synthetic lethality mechanism. The success of the PARP-BRCA therapy recently stimulated much thought and investigation of other potential therapy based on a synthetic lethality strategy (Ashworth and Lord, 2018; Setton et al., 2021). Summary The story developed in this chapter grew out of three roots: First, the discovery of a strange new polymer, poly(ADPR). Second, the unexpected observation that alkylating agents caused a fall in nicotinamide adenosine dinucleotide (NAD) level in the cell. Third, the discovery that inhibiting the enzyme activity of PARP (the enzyme that produces poly(ADPR)) impaired the ability of cells to repair DNA strand breaks and to survive DNA damage produced by alkylating agents and x-rays. The inference that PARP had a role in DNA repair then spurred investigation of how PARP worked and how to develop better PARP inhibitors that could become useful anti-cancer drugs. Later findings indicated that it was PARP's DNA binding, rather than inhibition of its enzyme activity per se, that was the main factor in killing cancer cells, at least at moderate dosage suited for treatment of patients. However, inhibitors of PARP's enzyme activity were effective, because the enzyme activity was required for addition ofpoly(ADPR) chains to the PARP molecule, which allowed PARP to dissociate from the DNA The inhibition of PARP's enzyme activity thus tended to keep the PARP molecule attached to the DNA, thereby blocking DNA functions from passing through that point Along with the development of clinically approved PARP inhibitors, focus was on combination of PARP inhibitors with DNA-damaging drugs, particularly the alkylating agent temozolomide. The drug combinations were highly effective against cancer cells in culture and against human tumors grown as xenografts in immune-deficient mice. Clinical trials of the drug combinations were then begun. Then, came the unexpected observation that PARP inhibitors were remarkably effective in cancers that had defects in BRCA gene function in DNA repair by homologous recombination. Cancer cells died if and only if both their BRCA and PARP functions were inactivated. This was the first case of synthetic lethality in mammalian cells and gave promise as a new strategy for cancer therapy. 609 K. W. Kohn Drugs Against cancer CHAPTER30 References Althaus, F.R. (1992). Poly ADP-ribosylation: a histone shuttle mechanism in DNA excision repair. Journal of cell science 102 ( Pt 4), 663-670. Ashworth, A., and Lord, C.J. (2018). Synthetic lethal therapies for cancer: what's next after PARP inhibitors? Nat Rev Clin Oncol. Ben-Hur, E., Utsumi, H., and Elkind, M.M. (1984). Inhibitors of poly (ADP-ribose) synthesis enhance radiation response by differentially affecting repair of potentially lethal versus sublethal damage. Br J Cancer Suppl 6, 39-42. Bryant, H.E., Schultz, N., Thomas, H.D., Parker, K.M., Flower, D., Lopez, E., Kyle, S., Meuth, M., Curtin, N.J., and Helleday, T. (2005). Specific killing of BRCA2-deficient tumours with inhibitors ofpoly(ADP-ribose) polymerase. Nature 434, 913-917. Chambon, P., Weill, J.D., Doly, J., Strasser, M.T., and Mandel, P. (1966). On the formation of novel adenylic compounds by enzymatic extracts of liver nuclei. Biochemical and biophysical research communications ZS, 638-643. Dantzer, F., Schreiber, V., Niedergang, C., Trucco, C., Flatter, E., De La Rubia, G., Oliver, J., Rolli, V., Menissier-de Murcia, J., and de Murcia, G. (1999). Involvement ofpoly(ADP- ribose) polymerase in base excision repair. Biochimie 81, 69-75. Das, B.B., Huang, S.Y., Murai, J., Rehman, I., Ame, J.C., Sengupta, S., Das, S.K., Majumdar, P., Zhang, H., Biard, D., et al. (2014). PARPl-TDPl coupling for the repair of topoisomerase I-induced DNA damage. Nucleic acids research 42, 4435-4449. Davies, M.I., Shall, S., and Skidmore, C.J. (1977). Poly(adenosine diphosphate ribose) polymerase and deoxyribonucleic acid damage [proceedings]. Biochem Soc Trans 5, 949-950. de la Lastra, C.A., Villegas, I., and Sanchez-Fidalgo, S. (2007). Poly(ADP-ribose) polymerase inhibitors: new pharmacological functions and potential clinical implications. Curr Pharm Des 13, 933-962. de Murcia, G., and Menissier de Murcia, J. (1994). Poly(ADP-ribose) polymerase: a molecular nick-sensor. Trends Biochem Sci 19, 172-176. de Murcia, G., Menissier-de Murcia, J., and Schreiber, V. (1991). Poly(ADP-ribose) polymerase: molecular biological aspects. BioEssays : news and reviews in molecular, cellular and developmental biology 13, 455-462. Delaney, C.A., Wang, L.Z., Kyle, S., White, A.W., Calvert, A.H., Curtin, N.J., Durkacz, B.W., Hostomsky, Z., and Newell, D.R. (2000). Potentiation oftemozolomide and topotecan growth inhibition and cytotoxicity by novel poly(adenosine diphosphoribose) polymerase inhibitors in a panel of human tumor cell lines. Clinical cancer research : an official journal of the American Association for Cancer Research 6, 2860-2867. Dobzhansky, T. (1946). Genetics of natural populations; recombination and variability in populations of Drosophila pseudoobscura. Genetics 31, 269-290. Durkacz, B.W., Irwin, J., and Shall, S. (1981). Inhibition of(ADP-ribose)n biosynthesis retards DNA repair but does not inhibit DNA repair synthesis. Biochemical and biophysical research communications 101, 1433-1441. Durkacz, B.W., Omidiji, 0 ., Gray, D.A., and Shall, S. (1980). (ADP-ribose )n participates in DNA excision repair. Nature 283, 593-596. 610 K. W. Kohn Drugs Against cancer CHAPTER30 Farago, A.F., Yeap, B.Y., Stanzione, M., Hung, Y.P., Heist, R.S., Marcoux, J.P., Zhong, J., Rangachari, 0 ., Barbie, D.A., Phat, S., et al. (2019). Combination Olaparib and Temozolomide in Relapsed Small-Cell Lung Cancer. Cancer Discov 9, 1372-1387. Farmer, H., McCabe, N., Lord, C.J., Tutt, A.N., Johnson, D.A., Richardson, T.B., Santarosa, M., Dillon, K.J., Hickson, I., Knights, C., et al. (2005). Targeting the DNA repair defect in BRCA mutant cells as a therapeutic strategy. Nature 434, 917-921. Ferraris, D.V. (2010). Evolution ofpoly(ADP-ribose) polymerase-1 (PARP-1) inhibitors. From concept to clinic. Journal of medicinal chemistry 53, 4561-4584. Fong, P.C., Boss, D.S., Yap, T.A., Tutt, A., Wu, P., Mergui-Roelvink, M ., Mortimer, P., Swaisland, H., Lau, A., O'Connor, M.J. et al. (2009). Inhibition of poly(ADP-ribose) polymerase in tumors from BRCA mutation carriers. The New England journal of medicine 361, 123- 134. Fong, P.C., Yap, TA, Boss, D.S., Carden, C.P., Mergui-Roelvink, M ., Gourley, C., De Greve, J., Lubinski, J., Shanley, S., Messiou, C., et al. (2010). Poly(ADP) -ribose polymerase inhibition: frequent durable responses in BRCA carrier ovarian cancer correlating with platinum-free interval. Journal of clinical oncology : official journal of the American Society of Clinical Oncology ZB, 2512-2519. Gudmundsdottir, K., and Ashworth, A. (2006). The roles of BRCAl and BRCA2 and associated proteins in the maintenance of genomic stability. Oncogene ZS, 5864-5874. Gupta, S.K., Smith, E.J., Mladek, A.C., Tian, S., Decker, P.A., Kizilbash, S.H., Kitange, G.J., and Sarkaria, J.N. (2018). PARP Inhibitors for Sensitization of Alkylation Chemotherapy in Glioblastoma: Impact of Blood-Brain Barrier and Molecular Heterogeneity. Front Oncol 8,670. Hayaishi, 0 ., and Ueda, K. (1977). Poly(ADP-ribose) and ADP-ribosylation of proteins. Annual review of biochemistry 46, 95-116. Kaelin, W.G., Jr. (2005). The concept of synthetic lethality in the context of anticancer therapy. Nature reviews Cancer 5, 689-698. Kohn, K.W. (1999). Molecular interaction map of the mammalian cell cycle control and DNA repair systems. Mol Biol Cell 10, 2703-2734. Lindahl, T., Satoh, M.S., Poirier, G.G., and Klungland, A. (1995). Post-translational modification ofpoly(ADP-ribose) polymerase induced by DNA strand breaks. Trends Biochem Sci ZO, 405-411. Liu, L., Kong, M., Gassman, N.R., Freudenthal, B.D., Prasad, R., Zhen, S., Watkins, S.C., Wilson, S.H., and Van Houten, B. (2017). PARPl changes from three-dimensional DNA damage searching to one-dimensional diffusion after auto-PARylation or in the presence of APEL Nucleic acids research 45, 12834-12847. Lord, C.J., and Ashworth, A. (2017). PARP inhibitors: Synthetic lethality in the clinic. Science 355, 1152-1158. Mateo, J., Lord, C.J., Serra, V., Tutt, A., Balmana, J., Castroviejo-Bermejo, M., Cruz, C., Oaknin, A., Kaye, S.B., and de Bono, J.S. (2019). A decade of clinical development of PARP inhibitors in perspective. Annals of oncology : official journal of the European Society for Medical Oncology/ ESMO. Menear, K.A., Adcock, C., Boulter, R., Cockcroft, X.L., Copsey, L., Cranston, A., Dillon, K.J., Drzewiecki, J., Garman, S., Gomez, S., et al. (2008). 4 -(3-(4- cyclopropanecarbonylpiperazine-1-carbonyl)-4-fluorobenzyl)-2H-phthalazin- 1-one: a 611 K. W. Kohn Drugs Against cancer CHAPTER30 novel bioavailable inhibitor ofpoly(ADP-ribose) polymerase-1. Journal of medicinal chemistry 51, 6581-6591. Murai, J., Huang, S.Y., Das, B.B., Renaud, A., Zhang. Y., Doroshow, J.H., Ji, J., Takeda, S., and Pommier, Y. (2012). Trapping of PARPl and PARP2 by Clinical PARP Inhibitors. Cancer research 72, 5588-5599. Murai, J., Huang, S.Y., Renaud, A., Zhang, Y., Ji, J., Takeda, S., Morris, J., Teicher, B., Doroshow, J.H., and Pommier, Y. (2014a). Stereospecific PARP trapping by BMN 673 and comparison with olaparib and rucaparib. Molecular cancer therapeutics 13, 433-443. Murai, J., Zhang. Y., Morris, J., Ji, J., Takeda, S., Doroshow, J.H., and Pommier, Y. (2014b). Rationale for poly(ADP-ribose) polymerase (PARP) inhibitors in combination therapy with camptothecins or temozolomide based on PARP trapping versus catalytic inhibition. The Journal of pharmacology and experimental therapeutics 349, 408-416. Nakazawa, K., Ueda, K., Honjo, T., Yoshihara, K., Nishizuka, Y., and Hayaishi, 0. (1968). Nicotinamide adenine dinucleotide glycohydrolases and poly adenosine diphosphate ribose synthesis in rat liver. Biochemical and biophysical research communications 32, 143-149. Nduka, N., Skidmore, C.J., and Shall, S. (1980). The enhancement of cytotoxicity of N-methyl- N-nitrosourea and of gamma-radiation by inhibitors of poly(ADP-ribose) polymerase. Eur J Biochem 105, 525-530. Nishizuka, Y., Ueda, K., Honjo, T., and Hayaishi, 0. (1968). Enzymic adenosine di phosphate ribosylation ofhistone and poly adenosine diphosphate ribose synthesis in rat liver nuclei. The Journal of biological chemistry 243, 3765-3767. Nishizuka, Y., Ueda, K., Nakazawa, K., and Hayaishi, 0. (1967). Studies on the polymer of adenosine diphosphate ribose. I. Enzymic formation from nicotinamide adenine dinuclotide in mammalian nuclei. The Journal of biological chemistry 242, 3164-3171. Nishizuka, Y., Ueda, K., Yoshihara, K., Yamamura, H., Takeda, M., and Hayaishi, 0 . (1969). Enzymic adenosine diphosphoribosylation of nuclear proteins. Cold Spring Harbor symposia on quantitative biology 34, 781-786. Pieper, A.A., Verma, A., Zhang. J., and Snyder, S.H. (1999). Poly (ADP-ribose) polymerase, nitric oxide and cell death. Trends Pharmacol Sci 20, 171-181. Pommier, Y., O'Connor, M.J., and de Bono, J. (2016). Laying a trap to kill cancer cells: PARP inhibitors and their mechanisms of action. Science translational medicine 8, 362ps317. Roitt, J.M. (1956). The Inhition of Carbohydrate Meatolism in Ascites Tumor Cells by Ethyleneimines. Biochem J 63, 300-397. Satoh, M.S., and Lindahl, T. (1992). Role ofpoly(ADP-ribose) formation in DNA repair. Nature 356, 356-358. Schreiber, V., Dantzer, F., Ame, J.C., and de Murcia, G. (2006). Poly(ADP-ribose): novel functions for an old molecule. Nature reviews Molecular cell biology 7, 517-528. Setton, J., Zinda, M., Riaz, N., Durocher, D., Zimmermann, M., Koehler, M ., Reis-Filho, J.S., and Powell, S.N. (2021). Synthetic Lethality in Cancer Therapeutics: The Next Generation. Cancer Discov 11, 1626-1635. Shen, Y., Rehman, F.L., Feng, Y., Boshuizen, J., Bajrami, I., Elliott, R., Wang. B., Lord, C.J., Post, LE., and Ashworth, A. (2013). BMN 673, a novel and highly potent PARPl/2 inhibitor for the treatment of human cancers with DNA repair deficiency. Clinical cancer research : an official journal of the American Association for Cancer Research 19, 5003- 5015. 612 K. W. Kohn Drugs Against cancer CHAPTER30 Smulson, M.E., Schein, P., Mullins, D.W., Jr., and Sudhakar, S. (1977). A putative role for nicotinamide adenine di nucleotide-promoted nuclear protein modification in the antitumor activity of N-methyl-N-nitrosourea. Cancer research 37, 3006-3012. Tentori, L., Leonetti, C., Scarsella, M., Vergati, M., Xu, W., Calvin, D., Morgan, L, Tang. Z., Woznizk, K., Alemu, C., et al. (2005). Brain distribution and efficacy as chemosensitizer ofan oral formulation of PARP-1 inhibitor GPI 15427 in experimental models of CNS tumors. Int J Oncol 26, 415-422. Turk, A.A., and Wisinski, K.B. (2018). PARP inhibitors in breast cancer: Bringing synthetic lethality to the bedside. Cancer 124, 2498-2506. Ueda, K., and Hayaishi, 0 . (1985). ADP-ribosylation. Annual review of biochemistry 54, 73- 100. van Wietmarschen, N., and Nussenzweig, A. (2018). Mechanism for Synthetic Lethality in BRCA-Deficient Cancers: No Longer Lagging Behind. Mol Cell 71 , 877-878. Whish, W.J., Davies, M.I., and Shall, S. (1975). Stimulation ofpoly(ADP-ribose) polymerase activity by the anti-tumour antibiotic, streptozotocin. Biochemical and biophysical research communications 65, 722-730. Zhang. Y.W., Regairaz, M., Seiler, J.A., Agama, K.K., Doroshow, J.H., and Pommier, Y. (2011). Poly(ADP-ribose) polymerase and XPF-ERCCl participate in distinct pathways for the repair of topoisomerase I-induced DNA damage in mammalian cells. Nucleic acids research 39, 3607-3620. Zimmer, A.S., Gillard, M., Lipkowitz, S., and Lee, J.M. (2018). Update on PARP Inhibitors in Breast Cancer. Current treatment options in oncology 19, 21. Zwelling. LA, Kerrigan, D., and Pommier, Y. (1982). Inhibitors ofpoly-(adenosine diphosphoribose) synthesis slow the resealing rate ofx-ray-induced DNA strand breaks. Biochemical and biophysical research communica tions 104, 897-902. 613 K. W. Kohn Drugs Against cancer CHAPTER31 Qoplu3J. fllcFolicM(ON!llto~r:tqondthcnpof'r o/ONAO'OSSl'tllh120731ffll1 Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus National Cancer Institute Bethesda, Maryland kohnk@nih.gov CHAPTER31 The Fanconi anemia story and the repair of DNA crosslinks. Fanconi anemia: a clue to how DNA cross/inks are repaired. In 1927, the Swiss pediatrician Guido Fanconi (Figure 31.1), reported an unusual inherited anemia affecting three brothers. The red blood cells of these 3 anemic children were enlarged, leading Fanconi to describe the disease as a form of pernicious anemia. However, the clinical pattern, family inheritance and macrocytic anemia of the children were unusual, and indeed, Fanconi had discovered a new disease, which came to bear his name: Fanconi anemia (Walden and Deans, 2014). This rare genetic disease was to provide a key to unraveling how DNA crosslinks and some other DNA derangements are repaired (Alter et al., 2003; Boisvert and Howlett, 2014; Howlett et al., 2009; Rego et al., 2009; Rosenberg et al., 2003; Vuono et al., 2016). The first step to this unraveling was made in the early 1960's by several research groups who obtained cells from Fanconi anemia patients and grew them in culture and looked at cells undergoing mitosis, where chromosome structure was clearly seen. They saw chromosome abnormalities in an unusually high frequency (Digweed and Sperling, 1996) (Figure 31.2). The next step came when researchers tried to find out what might be causing this high frequency of chromosome breaks and abnormalities. A clue had already come from testing a variety of drugs for their ability to cause chromosome damage. Remarkably, chromosome damage in cells from Fanconi anemia patients was most effected by drugs or chemicals that were known to produce DNA inter-strand crosslinks (Weksberg et al., 1979) (Chapters 1 and 3). The sensitivity of Fanconi anemia cells specifically to DNA crosslinkers was observed for cisplatin, carboplatin, nitrogen mustard, cyclophosphamide, and diepoxybutane (Niraj et al., 2019). Apparently, cells had a mechanism to repair DNA cross li nks that was defective in Fanconi anemia cells. 614 K. W. Kohn Drugs Against cancer CHAPTER31 Chromosome abnormalities, seen in the patient's lymphocytes during mitosis, became a criterion for the diagnosis of Fanconi anemia. The chromosome breaks and the underlying DNA damage made the patients prone to developing cancer, with acute myeloid leukemia occurring in about 10%. In addition, newborns bearing mutated Fanconi anemia genes often had congenital abnormalities, presumably because the embryos were defective in their ability to repair certain types of DNA damage that occurs occasionally in all cells. In addition to the chromosome abnormalities, Fanconi anemia cells were killed by unusually low concentration of DNA crosslinking drugs (Digweed and Sperling, 1996) (Figure 31.3). Researchers surmised that Fanconi anemia cells were defective in ability to repair DNA crosslinks. Indeed, the high sensitivity to treatment with DNA crosslinkers became a diagnostic test for Fanconi anemia. The most reliable DNA crosslinker for the test was diepoxybutane, a bifunctional alkylating agent that has a chemically simple and direct crosslinking mechanism (Auerbach, 1988). One may wonder why such a complicated DNA repair as the Fanconi system would have evolved specifically to deal with inter-strand crosslink producers rarely found in nature. The answer to this conundrum may be normal metabolic processes that produce rare, but in aggregate many, inter-strand crosslinks. Processes such as lipid peroxidation, histone demethylation, and alcohol metabolism can generate formaldehyde and acetaldehyde byproducts that can react to form inter-strand crosslinks (Niraj et al., 2019). Figure 31.1. Guido Fanconi (1892-1979) was a Swiss pediatrician, regarded as a founder of modern pediatrics. In 1927, he described the hereditary anemia that bears his name. His name in fact became associated with two different diseases: 'Fanconi syndrome' is a disorder of kidney function that must not be confused with 'Fanconi anemia'. (Photo in 1959 by Israeli photographer Ze'ey AJeksaodmwjcz.) 615 K. W. Kohn Drugs Against cancer CHAPTER31 ~\' .J 'vt ' ~ Ir,.l -~ , -'! ''" ' .~• • / Figure 31.2. Chromosomes of a Fanconi anemia patient's lymphocyte. The arrows point to chromosome breaks and to an abnormal joining of2 chromosomes (center). (Photograph by Dr Rolf Wegner, Berlin (Digweed and Sperling. 1996). 120 100 ,F 80 ~ ~ .. :6 5 60 40 20 10 100 1000 Mitonycin C (nM) Figure 31.3. Cell lines from two Fanconi anemia (complementation group A) patients (open symbols) were killed by 100-fold lower concentrations of mitomycin than a cell line from a normal person (solid symbols) (Digweed and Sperling.1996). The clinical diagnosis of Fanconi anemia was often difficult because symptoms among patients was variable. This problem was solved by testing for abnormally high sensitivity to DNA crosslinkers, such as diepoxybutane or mitomycin. But why was there such variability 616 K. W. Kohn Drugs Against cancer CHAPTER31 in the clinical p icture among d ifferent patients? Several research ers suspected that each type of clinical pattern was caused by a d iffere nt gen e. In other words, Fanconi an em ia might be caused by a defect in any on e of two or more gen es. This possibility was th oroughly investigated in 1985 by G. Duckwor th-Rys iecki, M. Buchwald, and th eir coworkers usi ng a cell fusion tech nique (Duckwor th-Rys iecki et al., 1985). When they fused togeth er pairs of lymphocytes, each from a d iffere nt patient, sometimes th e response to crossli nking d rug became n ormal a nd sometimes th e high sensitivity remai ned. They surmised th at, when fused cell pairs had normal drug resp on ses, then the disease of the two patients fro m whom the cells were taken was caused by defects in different genes. That meant that th ere were at least two d ifferent types or "complementation groups" of Fanconi an em ia. When cells of different complementation groups were fused together, the resulting cell duos were able to repair the crosslinks and swvive normally, but the combination of cells from patients who had the same complementation group did not do so: that was how "complementation group" was defined. In other words, cells of the same complementation group had the same underlying defect, whereas cells of different complementation groups had defects caused by different, independent factors. By 1992, four different complementation groups ofFanconi anemia had been defined (Strathdee et al., 1992a). The story was becoming complicated. And it didn't stop there. By 1999, eight complementation groups were defined. Thus, there were at least 8 different types ofFanconi anemia, each caused by a mutation in a different gene. Moreover, the proteins encoded by at least three of these genes were found to bind together to form a functional complex (Garcia- Higuera et al., 1999). It began to look like some of the proteins encoded by the Fanconi anemia genes bound to each other and worlced together as a multi-protein complex. By 2006, at least 12 Fanconi anemia genes had been discovered, and information began to accrue about how those genes assemble into a complex and how the system functions (Medhurst et al., 2006). As often happens is a developi ng field of research, the story gradually b eca me increasingly complicated. As of 201 7, 2 2 distinct Fanconi a nemia complementatio n groups had been found, each of which defined a different gene mutation (Che et al., 2018; Nepal et al., 2017). As many as 13 of these Fanconi genes had rare mutated forms on various chromosomes th at were inherited. Family members who car ried on e of th ese mutated genes had a n increased chance of developi ng cancer. Repair of crosslinks that covalently linked the two strands of a DNA double helix must be a lot more complicated than repair of damage to just one strand, and the repair had to be error-free, because DNA crosslinks occurred occasionally even in normal cells. That may be why so many different gene products, i.e., proteins, were required to repair DNA inter-strand crosslinks. Fanconi anemia resulted when both copies of any one of a person' s 22 Fanconi anemia genes were defective, usually because of mutations. Because of the defect, the patient' s cells were unable to repair DNA crosslinks. The proteins encoded by the 22 Fanconi anemia genes, together with several other genes, were found to cooperate in the crosslink repair (Nepal et al., 2017). How that repair worlcs is described below. 617 K. W. Kohn Drugs Against cancer CHAPTER31 Inactivating mutations of both copies of one of the Fanconi anemia genes often pushes cells on road to malignancy (Alter et al., 2003). (Those mutations occur fresh in body cells and are relatively common, in contrast to Fanconi anemia inherited from genn cells, which is rare) . Although disordered blood cell production and anemia was common in Fanconi anemia in children, cancers of various types usually appeared years later when Fanconi anemia patients had reached adulthood. Cancer treatment in Fanconi anemia patients was difficult, because the patient's normal tissues are sensitive to some of the best anti-cancer drugs. Fanconi anemia was found to be genetically recessive and occured when both parents has one normal and one mutated gene of a given complementation group. Each child then has a l -in-4 chance of inheriting the disease. Each child had a l -in-2 chance of becoming a carrier, like both parents, with a mutated gene paired with a normal gene. Individuals who were carriers of a Fanconi anemia gene mutation had an increased risk of developing cancer sometime during their lives, and the risk varied, depending on which Fanconi gene he/she was canying (Alter et al., 2003). Unraveling the roles of the Fa11coni a11emia ('Fane') protei11s in DNA repair. Unravelling the molecular details of how the Fane genes and their protein products functioned was important because of their role in repair of DNA damage, particularly the kind of DNA damage caused by some of the most useful cancer chemotherapy drugs: the DNA inter-strand crosslinkers, including nitrogen mustards, cyclophosphamide, platinum complexes, and mitomycin (d iscussed in Chapters 1 and 3). Moreover, one or another Fane gene was found defective in 40% of cancer cases and put the Fanconi problem high on the list of questions about both cancer cause and opportunities for therapy. The unravelling problem was hard, because at least 22 Fane anemia genes plus a number of functionally related genes all seemed to work together. How did they all work together to repa ir DN A crosslinks? The first objective was to clone and determine the nucleotide sequence of each gene and the structure of the protein s encoded by them. But that was only the start of the difficulty. Finding out how all of those protein s worked together to repair DNA crosslinks might have seemed a nearly impossible task. It might have surprised the early researcher who started on this effort that a large part of the story would be revealed with in less than 3 decades. The road to unravelling the complicated story of how the Fanconi system repaired DNA crosslinks began in 1989, when Martin Digweed and Karl Sperling identified an mRNA fraction from non-Fanconi cells that could correct the crosslink repair defect in Fanconi anemia cells (Digweed and Sperling, 1989). The cDNA then served to identify and clone the Fane genes. The molecular part of the story began with the cloning of the first Fane gene, which was accomplished in 1992 by Craig Strathdee, Manuel Buchwald and their coworkers at the University of Toronto, Canada (Strathdee et al., 1992b). At that time, 4 Fanconi anemia complementation groups had been defined, and it was known that fusing together cells of different complementation groups corrected the DNA repair defect. But nothing was 618 K. W. Kohn Drugs Against cancer CHAPTER31 known about the genes suspected of being mutated in each of the complementation groups. The Toronto researchers reasoned that the repair defect in cells of a given complementation group might be correctable by a normal version of the defective gene. Although they could not introduce the intact genes, they could get viruses to carry a library of cDNA molecules from normal cells into the recipient cells (each virus particle would carry one of the myriads of normal cDNA molecules). As recipients, they chose cells of complementation group C, because these were highly sensitive to mitomycin, and would give high sensitivity for correction of a DNA repair defect. Some of the cells that survived a good slug of mitomycin might then have harbored a virus that conveyed the correcting DNA sequence. Then it was merely a matter of transferring the viruses that did the correcting back into bacteria for cloning, and voila! the cDNA sequence of the complementation group C gene was at hand. It was not quite as simple as that, but that was the essence. (What is "cDNA"? Each cDNA molecule has a sequence that is complementary to a protein- coding RNA molecule in the cytoplasm. It is produced artificially by means of a reverse transcriptase enzyme that copies protein-coding mRNA sequences found in the cytoplasm into a complementary DNA version.) The next Fane gene cDNA to be cloned and sequenced was FancA, whose gene mutations accounted for 65% of the cases of the disease. This was accomplished in 1996 by an international group led by Hans Joenje and Manuel Buchwald using a method similar to how the Toronto group had cloned and sequences the cDNA of FaneC (Lo Ten Foe et al., 1996) The first information about how the Fane genes function came in 1997 from Paul D'Andrea's laboratory at the Dana-Farber Cancer Institute and Harvard Medical School. They found that the FancA and FancC proteins bound to each other and then were able to enter the cell nucleus (Kupfer et al., 1997). This binding was likely important, because a mutant FancC protein from complementation group C patients failed to bind to the FancA protein. A drug that prevented this binding might therefore increase the sensitivity of cancer cells to DNA crosslinking agents. The researchers went on to show that mutation of any one of several other Fane genes prevented FancA-FancC proteins binding each other and moving into the nucleus (Garcia- Higuera et al., 1999). It seemed that several Fanconi proteins were needed to form a functional multiprotein complex. The next player to enter the Fanconi dance was FancG. In 1999, Quinten Waisfisz and Hans Joenje at the Free University in Amsterdam, and their colleagues, discovered that FancA, in addition the binding FancC, also bound FancG (Waisfisz et al., 1999). Therefore, FancG seemed to be part of the functional multi protein complex. In 2000, the Amsterdam group added another piece to the story by cloning FaneF and proposing that the Fanconi proteins combine to form a complex that maintains the integrity of the chromosomes in the nucleus (de Winter et al., 2000). By 2001, altogether 619 K. W. Kohn Drugs Against cancer CHAPTER31 six Fane genes had been cloned and several of the proteins were found to bind to each other (Joenje and Patel, 2001). It seemed that the binding together of several Fane proteins formed a multiprotein complex that entered the nucleus to exert its functions, and the idea emerged that disrupting the complex might become a new cancer therapy. The next major addition to the story was made in 2002 by K. J. Patel and colleagues at the Universities of Cambridge and Amsterdam, who showed that the multi -protein Fanconi complex activated another Fanconi protein, FancD2, by adding a molecule of the small protein ubiquitin to it (Pace et al., 2002). This was a key discovery, as we shall see. Another key discovery came from an international group of scientists in 2005 (Meetei et al., 2005), who discovered FancM, the gene and protein defective in Fanconi anemia complementation group M. They showed that FancM bound to the other Fane proteins in the core multiprotein complex and was required for ubiquitylation of FanD2. But more importantly, they found that FancM protein interacted with some abnormal DNA structures. It seemed that FancM's role might recognize and bind DNA crosslink sites and bring other Fane proteins to the damage site. By 2006, it became possible to suggest a model of how various Fanconi proteins assemble into complexes and subcomplexes (Medhurst et al., 2006) (Figure 31.4). The model proposed that most of the then-known 11 Fanconi proteins bound to each other to form a multiprotein complex in the cell nucleus and that this complex ubiquitylated FancD2: it stuck a molecule of the small protein, ubiquitin, onto the FancD2 protein. See legend of Figure 31.4 for details. 620 K. W. Kohn Drugs Against cancer CHAPTER31 Figure 31.4. A model proposed in 2006 by Medhurst and colleagues of how Fanconi anemia proteins assemble in the cell nucleus at a DNA crosslink site and form a multi-protein complex that ubiquitylates FancD2 (Medhurst et al., 2006). The right side of the diagram proposed that a complex containing Fanc's A, B, C, E, F, G, L, and M ubiquitylates FancD2. The left side of the diagram suggested how the Fane multi protein complex forms. The model proposed that FancM, which was thought to recognize and bind to the crosslink site on the DNA, recruits Fanc's A, B, G, and Lat the DNA damage site in the nucleus. Fancs A and B had nuclear localization signals in their protein structures, which would carry them, as well as Fancs G and L into the nucleus. Fancs C, E, and F would then somehow become recruited to the complex that then would become capable of ubiquitylating FancD2. The next important discovery was the gene and protein defective in Fanconi anemia complementation group I, Fancl (Smogorzewska et al., 2007). This eye-opening discovery came in 2007 from a research group led by Steve Elledge at Harvard Medical School, that found that Fancl protein resembled FancD2 in amino acid sequence. Moreover, the two proteins bound tightly to each other when both of them became ubiqutylated by the core Fane multiprotein complex. Also, unlike other Fane proteins, the FancD2-Fancl pair was not required for the assembly of the core Fane complex. The FancD2-Fancl pair seemed special, because the other known Fane proteins had individually unique amino acid sequences, and, unlike FancD2 and Fancl, each of them was required for assembly of the core Fane multiprotein complex. 621 K. W. Kohn Drugs Against cancer CHAPTER31 Another remarkable aspect of the FancD2-Fancl dimer was its structure. The structure was determined by crystallography in 2011 by a research group led by Nikola Pavletich at Memorial Sloan-Kettering Cancer Center in New York, together with Steve Elledge's group at Harvard (loo et al., 2011). The structure of the duo resembled two saxophone-like shapes fitted together, and the ubiquitins were at the junction between the two proteins, evidently serving to lock the protein pair together (Figure 31.5). -~-----------. -,;.,,•11: I .FANCI ' /"\. ~. '. , •'<t ' ' I L----------- Figure 31.5. The saxophone-like shapes of FancD2 and Fane!, showing how the couple nestles together. Each has a ubiquitin attached, and the two ubiquitins are located at the junction between the two Fancs, where they lock the couple together (loo et al., 2011). The alpha helices of the proteins are shown in yellow. On the right is a section through the dimer, showing the locations of the ubiquitins (red with arrows pointing to them). From (Swuec et al., 2017) with arrows and labels added. Yet another remarkable finding came in 2007. When cells were treated with a DNA crosslinking drug, FancD2 and Fane! entered the cell nucleus, but were not spread all over the nucleus. Instead, the ubiquitylated FancD2 and Fane! became concentrated in spots or 'foci' where a DNA crosslink was located (Smogorzewska et al., 2007) (Figure 31.6). It made sense that these and perhaps other Fancs should become localized to places where they were needed to repair the crosslinks. But the fact that the foci were so clearly visible meant that a huge excess of these molecules somehow collected at the site of the crosslink - many more than were needed for the repair. How and why did that happen? As far as I know, this remains an open question. (There is more discussion about nuclear foci in the story of histone gamma-H2Ax in Chapter 28.) 622 K. W. Kohn Drugs Against cancer CHAPTER31 Antibody to Fancl Antibody to FancD2 Merge Figure 31.6. Fancl and FancD2 both became localized at foci (spots) where there were DNA cross li nks in the nucleus of a cell. The images show a single nucleus of a mitomycin-treated cell. Left: stained with a blue antibody to Fancl. Middle: stained with a red antibody to FancD2. Right: merge of the two images. In the merged image, the blue and red spots combine to form white spots, showing that Fancl and FancD2 were both present in each spot From (Smogorzewska et al., 2007). The Fane DNA damage response and repair network. By 2010, it was surmised that most of the known Fanconi anemia proteins (13 of the 22 Fanconi anemia genes were then known) bind together in a large multi-protein core that has essential functions in the DNA crosslink repair pathway (D 'Andrea, 2010). Furthermore, FancD2 and Fancl were special: they were found to bind to each other and to become activated by becoming ubiquitylated (i.e., by having the small protein, ubiquitin, bound to each of them) (Figure 31.5). The Fanconi core multi-protein complex (or some part ofit) was found to carry out this ubiquitylation reaction. The general picture was that FancM detects and binds DNA at the site of a crosslink and then binds and signals the Fanconi core complex to ubiquitylate the FancD2-Fancl dimer. The ubiquitylated dimer would then activate downstream proteins that carry out the initial DNA repair steps, but exactly how the repair itself worked was not yet known. By 2017, 22 Fane genes had been identified and cloned, and a general picture of the roles of the 22 Fane proteins emerged {Nepal, 2017 #1010) (Figure 31.7). The diagram depicts the multitude of protein species whose functions connect in one way or another to the Fane DNA repair pathway. The center of the diagram shows the Fane core multiprotein complex ubiquitylating the FancD2 -Fancl dimer. FancM (upper left in the diagram) is shown liked to a DNA damage site. The diagram proposed a sequence of steps leading to phosphorylation of the FancD2-Fancl dimer by ATM (whose gene is mutated in ataxia telangiectasia; see Chapter 29). The Fane core multiprotein complex then somehow converted the FancD2- Fancl dimer to a ubiquitylated form. The ubiquitylated FancD2-Fancl dimer would then be the executor of several outputs of the Fane pathway (shown along the bottom part of the diagram). 623 K. W. Kohn Drugs Against cancer CHAPTER31 Replication stress DNA damage __ , Phosphorylated FancD2-Fa ncl Oownweam function ~ GD Tra ns-lesio n -----~ --- \♦-~ DNA polymerases - i ..... TLS DNA repair Geno me sta bility Figure 31.7. Overview of the Fane DNA repair pathway. (From (Nepal et al., 2017) with labels and ovals added.) Near the upper left of the diagram, FancM (together with two associated protein) recognizes and binds to DNA at a site of damage, such as an inter- strand cross li nk. Through a sequence of steps, shown from left to right in the upper part of the diagram, this results in phosphorylation of the FancD2-Fancl dimer by ATM (mutated in ataxia telangiectasia; see Chapter 29). The Fane core multi protein complex (shown in the middle of the diagram) then causes the FancD2-Fancl dimer to become ubiquitylated, which then executes several functions, including stimulation of trans-lesion DNA synthesis. Thus, ubiquitylated FancD2 -Fancl dimer is special: it is not part of the core multiprotein complex and is the output or executor of the Fane pathway. Another Fane that has special functions in the pathway is FancM, the recognizer of DNA damage sites (top left in Figure 31.7). All about FancM. FancM has an unusual story of discovery. Rather than being found as a gene in Fanconi anemia patients, if was found as a previously unknown protein that bound to Fane proteins, such as FancA (Meetei et al., 2005). FancM was a part of the Fane pathway, because, in the 624 K. W. Kohn Drugs Against cancer CHAPTER31 absence of FancM, cells did not ubiquitylate FancD2 and had increased sensitivity to being killed by the DNA crosslinker, mitomycin. Mutation of FancM was noted to be especially associated with uterine and breast cancers (Basbous and Constantinou, 2019). Remarkably, FancM resembled a gene called Hefin ancient microorganisms, the archaea. FancM seemed to have a critical function dating back to the early history of life. What was that critical function? The greatest similarity between FancM and Hefwas in a region of amino acid sequence that conferred helicase function (Figure 31.8), which is the ability to unwind the DNA helix. Helicase becomes important during DNA replication and repair, when the growing end of the new strand encounters a break in the strand it is trying to copy. The problem is particularly severe when both template strands have breaks in nearly the same place. This may well have been problematic for organisms since early in evolution. How was the growing strand to find a complementary sequence to copy beyond the break in its template strand? The best solution in diploid organisms (which have a pair of each chromosome) was for the blocked strand to copy the relevant sequences from DNA in the other chromosome of the pair by homologous recombination (Chapter 27A). (The alternative method of double stand break repair, non -homologous end joining Chapter 278, did not involve FancM.) We see in Figure 31.8 that, although the FancM helicase and endonuclease functions reside in the same gene product protein in archaea, they are separated in humans and other vertebrates, the helicase function being in FancM and the endonuclease function in ERCC4. What does FancM actually do in the Fane DNA repair pathway; how do its DNA damage recognition and DNA unwinding (helicase) abilities come into play? The key may be that FancM can translocate along the DNA duplex; it may be looking for a damage site to lock on to and bring the repair proteins of the FancM pathway to the site (Meetei et al., 2005). Helicase Endonuclease domain domain Human FANCM -------- Chimpanzee FANCM Vertebrates Mouse FANCM Fish FANCM Fruit fly CG7922 ~ YeastMPH1 ~ Archaeal Hefs Human ERCC4 625 K. W. Kohn Drugs Against cancer CHAPTER31 Figure 31.8 The FancM protein in various organisms has a DNA helicase domain (red) and a DNA endonuclease domain (green) in its amino acid sequence. Both domains indicate a role of the FancM protein in DNA repair. However, in some organisms, a domain may be functionally defective (marked by a X), in which case the amino acid sequence is not quite correct for function. Even a defective domain however indicates an evolutionary relationship of the proteins: a common ancestor. The only organisms in which both domains are functional are the ancient Archaea microorganism, whose FancM-Jike protein is called Hef. Human ERCC4 (green arrow) is a FancM relative that functions as an endonuclease (DNA-cutting) enzyme in DNA repair. FancM in vertebrates from Human to fish has a functional helicase (DNA unwinding) domain, which is important in DNA repair, but its endonuclease function is defective (and provided by ERCC4). (From (Meetei et al., 2005).) An important finding about inter-strand crosslink repair was that most of the repair occurred at sites where DNA replication was blocked. In other words, sites along the DNA where progressing replication forks have encountered a crosslink and cannot progress further. The repair mechanism is in fact linked to the replication machinery. An experiment demonstrating this, reported in 2013 from Michael Seidman's laboratory at NIH in Baltimore, is shown in Figure 31.9. Under the conditions of the experiment, blocked replication forks coming from one direction (B) or from both directions (C) occurred with about equal frequencies (the crosslink is indicated by a vertical red bar). The method that revealed growing DNA strands blocked at crosslinks was somewhat complicated but illustrated the remarkable power of the new technology. 626 K. W. Kohn Drugs Against cancer CHAPTER31 RED GREEN A DIG-TMP/ UVA CldU ldU Replication 20min 20min patterns B ~~..._, ____ single fork / C Figure 31.9. An experiment reported in 2013 by Ji ng Hang and colleagues in Michael Seidman's laboratory at NIH in Baltimore, showing how DNA replication blocks at inter- strand cross link sites were visualized (Huang et al., 2013). An example of a replication fork coming from one direction or from both directions are shown in B and C, respectively. In both cases, the replication forks were blocked at the crosslink (vertical red bar in the diagrams). An example of how the replication blocks were visualized are shown above each diagram. The protocol of the experiment is shown in A, Inter-strand crosslinks were produced by treating cells with a psoralen compound followed by ultraviolet light (UV) (see section about psoralen in Chapter 1). The psoralen part of the compound had linked to it a part that recognized an antibody that would fluoresce red. The cells were then treated for 20 minutes with chlorodeoxyuridine (CldU), which became incorporated into growing DNA stands; the CldU would be recognized by an antibody that fluoresced red. The CldU was then washed away and replaced by iododeoxyuridine (IdU), which would be recognized by an antibody that fluoresced green. DNA fibers from the cells were then stretched out and treated with the fluorescent antibodies. The more recently replicated DNA glowed green, while the older part of the newly replicated DNA glowed red. We see that the most recently replicated part of the DNA stand (green) stops at the crosslink, where replicated had stopped after proceeding only a short distance (red spot). In 2013, Kottemann and Smogorzewska (Kottemann and Smogorzewska, 2013) had outined DNA interstrand crosskink repair (Figure 31.10) as going by the following, perhaps 627 K. W. Kohn Drugs Against cancer CHAPTER31 necessarily compicated, steps: First, the core Fane complex (Figure 31.7), together with accessory proteins, detects the crosslink and becomes activated by phosphorylation mediated by the ATM-related kinase, ATR. The the core complex (perhaps mediated by its FancL component (Garcia et al., 2009)) then adds ubiquitins to the Fancl-FancD2 dimer (see Figure 31.7), which induces SLX4, functioning as a scaffold for three nucleases (XPF, MUSBl and SLXl ), to bring those nucleases to the crosslink site. The nucleases then cut a DNA strand on either side of the crosslink (red arrows in Figure 31.10). The final steps of the repair were not well understood at the time, although it was known that DNA polymerases that carry out trans-lesion synthesis (TLS) were required (see Figure 31.7). Evidence also indicated that the final process involved homologous recombination with the participation of BRCA2 (Figure 31.11). In 2015, Xue et al (Xue et al., 2015) depicted the role of FancM in the early steps of inter- strand crosslink repair as shown in (Figure 31.12) and explained in the legend. 5'~ ~ 3' ~ 3' ---r--- :::-- 5' Figure 31.10. First excision step in removing a DNA inter-strand crosslink (!SC) (red bar) (Kottemann and Smogorzewska, 2013). In this representation, DNA replication processes have approached from both directions and have become blocked at the !SC. The Fane proteins bring a host of DNA repair proteins to the site of the crosslink (Figure 31.7). Additional repair proteins are brought in by FancD2-Fancl, after it has become ubiquitylated by the Fane core complex at the crosslink site. Among the brought-in repair proteins are enzymes that cut the DNA on both sides of the crosslink (red arrows). 35''- - - -,,,..1..,...,,......___ 3'5, 628 K. W. Kohn Drugs Against cancer CHAPTER31 Figure 31.11. The homologous recombination step in DNA inter-strand crosslink repair. The green segment in the middle of the upper stand represents the part of the DNA strand that was healed by a trans-lesion DNA polymerase, which copy a damaged nucleotide in the template by inserting an arbitrary base, usually an A. I Recruitment of .. FANCM to ICL HCU<2 D interaction B ATR-regulation ~ ofFANCM J Recruitment of FA core Cheekpolnl • and BTR to chromatin activation C Figure 31.12. The early steps in repair of DNA inter-strand crosslinks, according to a model proposed in 2015 by Xue et al (Xue et al., 2015). First, FancM recognizes and binds to the inter-strand crosslink (with the help of a complex of 2 other proteins, labelled MHF in the diagram) (A and B). FancM then recruits other proteins required for the repair: the Fane core complex and 3 other proteins, including the DNA-unwinder BLM (mutated in Bloom's disease) and topoisomerase Topllla, which relieves tortional stress in the DNA helix (C). ln an alternative path, FancM recruits proteins that signal the cell to delay DNA replication and cell division (D). The Fanconi anemia core complex in C(FancT and FancL (Garcia et al., 2009), within the complex) goes on to ubiquitylate FancD2-Fancl (Figure 31. 7), which leads to the next phase of the repair. (From (Xue et al., 2015), simplified.) The final steps ofDNA cross/ink repair: homologous recombination. A clue for a role of homologous recombination in the Fane cross link repair pathway came as early as 2002 with the discovery by Niall Howlett, Alan D ' Andrea and their coworkers at the Dana-Farber Cancer Institute and Harvard Medical School that, surprisingly, the breast cancer-associated genes BRCAJ and BRCA2 were an intimate and essential part of the network of Fane genes that repaired DNA crosslinks (Howlett et al., 2002). It was later found that, in fact, the BRCAJ gene was the same as FancS, and the BRCA2 was the same as FancD l (Nepal et al., 2017). BRCAl and BRCA2 were known to be part of the homologous recombination system that 629 K. W. Kohn Drugs Against cancer CHAPTER31 is required by several DNA repair processes (Chapter 27A). Apparently, some of the FANC genes were at a convergence of several DNA repair pathways (Niraj et al., 2019). A key step then came in 2010, in a paper by Fiona Vaz, Helmut Hanenberg, Detlev Schindler, Christopher Mathew, and their colleagues in London and Germany (Vaz et al., 2010). At the time, there were 13 known Fanconi anemia genes. The authors investigated a Fanconi anemia family where there was no mutation in any of known Fanconi anemia genes. One of the affected family members was a 10-year-old boy whose cultured fibroblast cells had many chromosome rearrangements and whose blood lymphocytes were killed by low concentrations of DNA crosslinking agents (mitomycin and diepoxybutane). The boy, as well as other affected family members, had multiple developmental abnormalities; some died in infancy. The boy had a mutated gene, but, surprisingly, the mutation was in Rad51C, a gene that was involved, together with its close relative Rad51, in homologous recombination(Figure 27 A). It was later confirmed that mutation of Rad51C caused clinical Fanconi anemia. Therefore, Rad51C was given the alternative name, FancO; the homologous recombination gene, Rad51C, was the same as the newly discovered Fanconi anemia gene, FancO. But FancO was not part of the Fane core complex. Therefore, it was thought to function downstream of the FancD2-Fancl ubiquitylation step, perhaps in a recombination process that repairs an intermediate DNA structure. The bottom-line message was that, since mutation of Rad51C caused Fanconi anemia, the repair of DNA inter-strand crosslinks may involve DNA recombination. That was consistent with the already mentioned finding that BRCAl, BRCA2, and Rad51C, which were known to function in genetic recombination, were in fact the same as FancS, FancDl , and FancO, respectively. Fane DNA repair components function also in other DNA repair pathways. In 2002, the D'Andrea Jab at Harvard had discovered a connection between the Fanconi DNA repair pathway and the repair pathways that are controlled by ATM gene that is mutated in ataxia telangiectasia (see Chapter 29) (Taniguchi et al., 2002). They discovered that FancD2 phosphorylates and thereby activates the ATM kinase enzyme, which is the product of the ATM gene. The remarkable finding, reported from Alan D' Andrea's laboratory linked the Fanco ni anemia ('Fane') DNA repair pathway with DNA repair genes associated with ataxia tela ngiectasia (Andreassen et al., 2004) (see Chapter 29). They and later researchers found that phosphorylation of FancD2, which switches on the Fane DNA repair pathway, was carried out by ATR, a kinase related to the ATM gene. What's more, phosphorylations by ATM were also implicated in the activation switch. It seemed that phosphorylations by ATM and ATR, as well as ubiquitylations by the Fane core multiprotein complex, were all required to activate the FancD2-Fancl dimer and for the consequent activation of the Fane 630 K. W. Kohn Drugs Against cancer CHAPTER31 DNA repair pathway. ATM and ATR also signaled to the cell cycle checkpoints to delay the cell cycle in order to give more time for DNA repair. A connection of Fanconi anemia genes with another DNA repair process was discovered by Amom Meetei, Weidong Wang and their coworkers at the NIH National Institute of Aging and colleagues in the Netherlands (Meetei et al., 2003). They found that FancD2 binds to BLM, the product of the gene, which, when mutated, caused the rare premature aging genetic disorder known as Bloom's syndrome. FancD2 and BLM were in a multiprotein complex different from the Fanconi anemia core complex. The BLM complex functioned to unwind the DNA helix, a step required for many DNA repair pathways, including the Fanconi anemia repair pathway. It was the first time that a biochemical step, namely DNA helix unwinding, was identified as a step in the Fanconi anemia DNA repair pathway; this was an early step in which FancD2 functions as part of the Fanconi core complex. Yet another connection was that the Fane complementation group gene, FancQ, was the same as the gene for the xeroderma pigmentosum XPF enzyme that function to cleave a DNA strand during repair of UV-induced pyrimidine dimers (see Chapter 22) (Figure 31.7). Fancgene mutations in cancers. Figures 31.13 and 31.14 show how often the Fanconi anemia genes were found to be altered in cancers (Niraj et al., 2019). The alterations were classified as mutations (excluding changes in gene copy numbers), amplifications (increased number of gene copies), or deletions of parts of the gene. We see that the most frequent types of alterations were mutations in some genes and amplifications in some other genes, while deletions were most common in only a few genes. As many as 40% of the cancers had Fanconi gene alteration. The gene alterations almost always were in the cancers, not in the genome of the patients. Evidently, a Fanconi anemia gene alteration is frequently acquired by the cancer as it develops and may contribute to the development of the malignancy. DNA damage, such as crosslinks, may occur naturally in normal metabolism that can produce small amounts of DNA damaging compounds, such as formaldehyde. The ability of a cell to repair those DNA lesions may therefore be critical. A cell's ability to repair those DNA lesions would be defective if the cell's genome has a mutation or deletion in a Fanconi anemia gene; the consequence would be a step towards malignancy. Why amplification of certain of those genes also may lead to cancer is less clear to me. Amplifications were particularly common in cancers of ovary and breast, while mutations were the dominant alterations in cancers of the uterus (Figure 31.13). lt remains to be found out why there is such dependence on cancer type. 631 K. W. Kohn Drugs Against cancer CHAPl ■ Mutation m 31S ■ Amplification !>- so 963 287 ■ Deep deletion ■ Multiple alte,atlor .. V C 4-0 242 .. :, 504 - <T 492 ~ 30 149 22 C .5! 20 .. :;; ~ < 10 0 t .,.;; C ~ •E ., .,& .,. ~ :t C ~ ~ ~ •~ t . .,".,." f! 0 > • 0 C "• C C e ~ ~ V C • c e .s. :l: ":x::3 Figure 31.13. The frequency at which Fanconi anemia genes were altered in a total of 3,406 cancers of various types. Mutations, amplifications, or deletions in Fanconi anemia related genes were found in 40% of the cancers. (From Niraj et al (Niraj et al., 2019), who state that the data were generated by The Cancer Genome Atlas and were downloaded from cBioPortal.) C Pro 100 90 I Mutations 80 I Delet.ions 0c- 1 70 I Amplifications 70 67 63 0 ~ 63 62 61 62 61 6<I 6() S< 57 56 51 -.. 0 C so 53 4949 S2 ~ 0" ·- 5-0 '445 45 46 ... -.. a. f 4-0 JS JJ 35 0 .. JO 31 19 30 ~ J-0 C 20 II 10 8 s 0 ,., ... e t; t e; § ~ :;: 3" 3 ~ (:j ~ ;;; 3 !::l tl "11: "11: "11: "11: 11:" 11:" 8 ~ 11: ~ 1l:" ~ 11: o' .,"' °'., <!: ~ lS~ ~ C ~ ~ ~ ~ "' r:A rnro ln?rnmnlov ~A/MR f:A ror•nt Figure 31.14. Mutations, deletions, and amplifications in Fanconi anemia genes (and several related genes) found in a total of l,363. For some of the genes the most frequent alterations were mutations; for some genes, amplifications were the most common; for a few genes, including FancA, deletions were common. The frequencies of alterations of gene in the tumors were FancA 64%; FancC 12%; FancG 8%; FancD2 4%; Fancs B, F, J, and D1 2% each; Fancs E and I 1% each; all others <1%. FA, Fanconi anemia; ID2, FancD2-Fancl; HR, homologous recombination. (From Niraj et al (Niraj et al., 2019), who state that the data were generated by The Cancer Genome Atlas and were downloaded from cBioPortal on May3, 2018.) 632 K. W. Kohn Drugs Against cancer CHAPTER31 Summary of the Fanconi anemia (Fane) genes and proteins in DNA repair. The state of knowledge as of 2018 was reviewed by Niraj et al (Ni raj et al., 2019). DNA repair by the Fane proteins primarily targeted inter-strand crosslinks. The repair occurred mainly during DNA replication, when the replication machinery becomes stuck at the crossli nk sites on the DNA. Crossli nk repair at sites where a crosslink blocks the replication machinery must begin with recognition of the crossli nk blockage sites on the DNA. However, proteins of the replication machinery that remain stuck to the crosslink site must be removed in order to make the site accessible to the repair machinery. According to the current picture (Ni raj et al., 2019), the cleaning away of replication proteins from the cross li nk site is a special job of Fan cl, which also has other critical functions in the repair process. The cleaned crosslink sites are then recognized by the Fanconi anemia protein, FancM, which binds to the DNA at those places and initiates the repair process by bringing to the site a multiprotein complex consisting of most of the Fane proteins (the Fane core complex). The next step, which actually begins the crosslink repair, was found to be carried out by the combination of two Fane proteins: FancD2 and Fane! (Swuec et al., 2017). (FancD2 was found to be the same as BRCA2). In order to function, however, the FancD2 -Fancl pair had to be activated by having a molecule ofubiquitin linked to each of them (Figure 31.5). The ubiquitylation that activates the two proteins of the FancD2-Fancl dimer was found to be carried out by one of the proteins in the Fane core complex, namely FancT. The remarkable way that FancD2 and Fan cl nestle together and bind to each other by way of their ubiquitins is shown in Figure 31.5. The shapes of the pair was described as saxophone-like, but their shape changes somewhat when they become associated with a DNA damage site; the shape change made them fully active (Niraj et al., 2017). The picture that emerged was that the DNA at the replication-blocked crossli nk site was first cleaned of replication proteins with help of Fane!. The crosslinked site then brings in FancM, which then brings the other Fane core complex of proteins to the crosslink site on the DNA. FancL or FancT within the core complex may then ubiquitylate the two proteins in the FancD2-Fancl dimer (which is not itself part of the core complex) (Niraj et al., 2019) (Figure 31.7). The ubiquitylated dimer then proceeds to activate a variety of DNA damage- response processes (Figure 31.7) (Nepal et al., 2017). The FancD2-Fancl dimer was found to receive inputs from several DNA-damage signaling pathways and to respond by transmitting signals to several systems that repair or counteract the effects of the DNA damage (Nepal et al., 2017). The key event, initiated by the DNA damage, was to add a molecule of ubiquitin to both FancD2 and Fane! in the dimer. This ubiquitylation activated the dimer to allow it to signal to the down-stream DNA damage-response processes. In brief, the ubiquitylation was attributed to FancT as part of a complex consisting of as many as 12 other Fane proteins. DNA-binding ability in this big complex was found to reside in FancM, together with 2 other proteins. FancM was found to bind and detect places on DNA that have structural damage, such as crosslinks. 633 K. W. Kohn Drugs Against cancer CHAPTER31 Ubiquitylation of the FancDZ-Fancl dimer required that both proteins of the dimer be phosphorylated, an action that was carried out by ATM, the kinase that is mutated in ataxia telangiectasia, and/or by the ATM -related kinase, ATR (Chapter 29). After FancM and its associated proteins have recognized and bound to the DNA crosslink site, the FancDZ-Fancl duo comes into play. Together with some other proteins, it cuts out a segment of DNA that has the crosslink in it and prepares the cut ends of the DNA for rejoining. Rejoining of the resulting DNA double-strand ends involves homologous recombination with participation of BRCAl, which is the same as FancS (Fane complementation group S protein) (Niraj et al., 2019). Thus, BRCAl had a role in both breast cancer and Fanconi anemia. References Alter, B.P., Greene, M.H., Velazquez, I., and Rosenberg, P.S. (2003). Cancer in Fanconi anemia. Blood 101, 2072. Andreassen, P.R., D'Andrea, A.D., and Taniguchi, T. (2004). ATR couples FANCDZ monoubiquitination to the DNA-damage response. Genes Dev 18, 1958-1963. Auerbach, A.D. (1988). A test for Fanconi 's anemia. Blood 72, 366-367. Basbous, )., and Constantinou, A. (2019). A tumor suppressive DNA translocase named FANCM. Crit Rev Biochem Mo! Biol 54, 27-40. Boisvert, R.A., and Howlett, N.G. (2014). The Fanconi anemia !DZ complex: dueling saxes at the crossroads. Cell Cycle 13, 2999-3015. Che, R., Zhang,)., Nepal, M., Han, B., and Fei, P. (2018). Multifaceted Fanconi Anemia Signaling. Trends Genet 34, 171-183. D'Andrea, A.D. (2010). Susceptibility pathways in Fanconi's anemia and breast cancer. The New England journal of medicine 362, 1909-1919. de Winter, J.P., van der Wee!, L., de Groot, )., Stone, S., Waisfisz, Q., Arwert, F., Scheper, R.J., Kruyt, F.A., Hoatlin, M.E., and Joenje, H. (2000). The Fanconi anemia protein FANCF forms a nuclear complex with FANCA, FAN CC and FANCG. Hum Mo! Genet 9, 2665- 2674. Digweed, M., and Sperling, K. (1989). Identification of a He La mRNA fraction which can correct the DNA-repair defect in Fanconi anaemia fibroblasts. Mutation research 218, 171-177. Digweed, M., and Sperling, K. (1996). Molecular Analysis of Fanconi Anemia. BioEssays: news and reviews in molecular, cellular and developmental biology 18. Duckworth-Rysiecki, G., Cornish, K., Clarke, C.A., and Buchwald, M. (1985). Identification of two complementation groups in Fanconi anemia. Somat Cell Mo! Genet 11, 35-41. Garcia, M.J., Fernandez, V., Osorio, A., Barroso, A., Fernandez, F., Urioste, M., and Benitez, J. (2009). Mutational analysis of FAN CL, FAN CM and the recently identified FANCI suggests that among the 13 known Fanconi Anemia genes, only FANCDl/BRCAZ plays a major role in high-risk breast cancer predisposition. Carcinogenesis 30, 1898-1902. 634 K. W. Kohn Drugs Against cancer CHAPTER31 Garcia-Higuera, I., Kuang, Y., Naf, D., Wasik, J., and D'Andrea, A.D. (1999). Fanconi anemia proteins FAN CA, FAN CC, and FANCG/XRCC9 interact in a functional nuclear complex. Molecular and cellular biology 19, 4866-4873. Howlett, N.G., Harney, J.A., Rego, M.A., Kolling, F.W.t., and Glover, T.W. (2009). Functional interaction between the Fanconi Anemia D2 protein and proliferating cell nuclear antigen (PCNA) via a conserved putative PCNA interaction motif. The Journal of biological chemistry 284, 28935-28942. Howlett, N.G., Taniguchi, T., Olson, S., Cox, B., Waisfisz, Q., De Die-Smulders, C., Persky, N., Grompe, M., Joenje, H., Pals, G., et al. (2002). Biallelic inactivation of BRCA2 in Fanconi anemia. Science 297, 606-609. Huang, J., Liu, S., Bellani, M.A., Thazhathveetil, A.K., Ling, C., de Winter, J.P., Wang, Y., Wang, W., and Seidman, M.M. (2013). The DNA translocase FANCM/MHF promotes replication traverse of DNA interstrand crosslinks. Mo! Cell 52, 434-446. Joenje, H., and Patel, K.J. (2001). The emerging genetic and molecular basis of Fanconi anaemia. Nat Rev Genet 2 , 446-457. Joo, W., Xu, G., Persky, N.S., Smogorzewska, A., Rudge, D.G., Buzovetsky, 0 ., Elledge, S.J., and Pavletich, N.P. (2011). Structure of the FANCI-FANCD2 complex: insights into the Fanconi anemia DNA repair pathway. Science 333, 312-316. Kottemann, M.C., and Smogorzewska, A. (2013). Fanconi anaemia and the repair of Watson and Crick DNA crossli nks. Nature 493, 356-363. Kupfer, G.M., Naf, D., Suli man, A., Pulsipher, M., and D'Andrea, A.D. (1997). The Fanconi anaemia proteins, FAA and FAC, interact to form a nuclear complex. Nat Genet 17, 487-490. Lo Ten Foe, J.R., Rooimans, M.A., Bosnoyan-Collins, L., Alon, N., Wijker, M., Parker, L., Lightfoot, J., Carreau, M., Callen, D.F., Savoia, A., et al. (1996). Expression cloning of a cDNA for the major Fanconi anaemia gene, FAA. Nat Genet 14, 320-323. Medhurst, A.L., Laghmani el, H., Steltenpool, J., Ferrer, M., Fontaine, C., de Groot, J., Rooimans, M.A., Scheper, R.J., Meetei, A.R., Wang, W., et al. (2006). Evidence for subcomplexes in the Fanconi anemia pathway. Blood 108, 2072-2080. Meetei, A.R., Medhurst, A.L., Ling, C., Xue, Y., Singh, T.R., Bier, P., Steltenpool, J., Stone, S., Dokal, I., Mathew, C.G., et al. (2005). A human ortholog of archaeal DNA repair protein Hef is defective in Fanconi anemia complementation group M. Nat Genet 37, 958-963. Meetei, A.R., Sechi, S., Wallisch, M., Yang, D., Young, M.K., Joenje, H., Hoatlin, M.E., and Wang, W. (2003). A multiprotein nuclear complex connects Fanconi anemia and Bloom syndrome. Molecular and cellular biology 23, 3417-3426. Nepal, M., Che, R., Zhang, J., Ma, C., and Fei, P. (2017). Fanconi Anemia Signaling and Cancer. Trends in cancer 3, 840-856. Niraj, J., Caron, M.C., Drapeau, K., Berube, S., Guitton -Sert, L., Coulombe, Y., Couturier, A.M., and Masson, J.Y. (2017). The identification of FANCD2 DNA binding domains reveals nuclear localization sequences. Nucleic acids research 45, 8341-8357. Niraj, J., Farkkila, A., and D'Andrea, A.D. (2019). The Fanconi Anemia Pathway in Cancer. Annu Rev Cancer Biol 3, 457-478. Pace, P., Johnson, M., Tan, W.M., Mosedale, G., Sng, C., Hoatlin, M., de Winter, J., Joenje, H., Gergely, F., and Patel, K.J. (2002). FANCE: the link between Fanconi anaemia complex assembly and activity. The EMBO journal 21 , 3414-3423. 635 K. W. Kohn Drugs Against cancer CHAPTER31 Rego, M.A., Kolling, F.W.t , and Howlett, N.G. (2009). The Fanconi anemia protein interaction network: casting a wide net Mutation research 668, 27-41. Rosenberg, P.S., Greene, M.H., and Alter, B.P. (2003). Cancer incidence in persons with Fanconi anemia. Blood 101, 822-826. Smogorzewska, A., Matsuoka, S., Vinciguerra, P., McDonald, E.R., 3rd, Hurov, K.E., Luo,)., Ballif, B.A., Gygi, S.P., Hofmann, K., D'Andrea, A.D., eta/. (2007). Identification of the FANCI protein, a monoubiquitinated FANCD2 paralog required for DNA repair. Cell 129, 289-301. Strathdee, C.A., Duncan, A.M., and Buchwald, M. (1992a). Evidence for at least four Fanconi anaemia genes including FACC on chromosome 9 . Nat Genet 1, 196-198. Strathdee, C.A., Gavish, H., Shannon, W.R., and Buchwald, M. (1992b ). Cloning of cDNAs for Fanconi's anaemia by functional complementation. Nature 356, 763-767. Swuec, P., Renault, L., Borg, A., Shah, F., Murphy, V.j., van Twest, S., Snijders, A.P., Deans, A.J., and Costa, A. (2017). The FA Core Complex Contains a Homo-dimeric Catalytic Module for the Symmetric Mono-ubiquitination of FANCI-FANCD2. Cell reports 18, 611-623. Taniguchi, T., Garcia -Higuera, I., Xu, B., Andreassen, P.R., Gregory, R.C., Kim, S.T., Lane, W.S., Kastan, M.B., and D'Andrea, A.D. (2002). Convergence of the fanconi anemia and ataxia telangiectasia signaling pathways. Cell 109, 459-4 72. Vaz, F., Hanenberg, H., Schuster, B., Barker, K., Wiek, C., Erven, V., Neveling, K., Endt, D., Kesterton, I., Autore, F., eta/. (2010). Mutation of the RAD51C gene in a Fanconi anemia-like disorder. Nat Genet 42, 406-409. Vuono, E.A., Mukherjee, A., Vierra, D.A., Adroved, M.M., Hodson, C., Deans, A.J., and Howlett, N.G. (2016). The PTEN phosphatase functions cooperatively with the Fanconi anemia proteins in DNA crosslink repair. Sci Rep 6, 36439. Waisfisz, Q., de Winter, J.P., Kruyt, F.A., de Groot, )., van der Weel, L., Dijkmans, L.M., Zhi, Y., Arwert, F., Scheper, R.J., Youssoufian, H., eta/. (1999). A physical complex of the Fanconi anemia proteins FANCG/XRCC9 and FANCA. Proceedings of the National Academy of Sciences of the United States of America 96, 10320-10325. Walden, H., and Deans, A.J. (2014). The Fanconi anemia DNA repair pathway: structural and functional insights into a complex disorder. Annu Rev Biophys 43, 257-278. Weksberg, R., Buchwald, M., Sargent, P., Thompson, M.W., and Siminovitch, L. (1979). Specific cellular defects in patients with Fanconi anemia. J Cell Physiol 101 , 311-323. Xue, X., Sung, P., and Zhao, X. (2015). Functions and regulation of the multitasking FANCM family of DNA motor proteins. Genes Dev 29, 1777-1788. 636 K. W. Kohn Drugs Against cancer CHAPTER32 Qoplu32. fllcpS3st1117 • ,-dk»o/tNa-Z21.01SblJ Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Developmental Therapeutic Branch National Cancer Institute Bethesda, Maryland kohnk@nih.gov CHAPTER32 The pS3 story - guardian of the genome and the Li-Fraumeni Syndrome. The most famous and perhaps most important discovery of a familial cancer syndrome was made in 1969 by Frederick Li and Joseph Fraumeni at NIH in Bethesda, Maryland (Li and Fraumeni, 1969) (Figures 32.1 and 32.2). The cause of the Li-Fraumeni syndrome was eventually traced to mutations in what came to be the most famous and most important of all cancer-causing genes, not only in its rare inherited form, but in at least half of all cancers. Mutation of the gene, TP53 (T for transcription factor), was found to be a very frequent early step as normal cells in various tissues begin on their path to malignancy. In their famous 1969 paper in the Annals of Internal Medicine, Li and Fraumeni reported on four families in which cancers were unusually common (Figure 32.1 ). What drew their attention to these families was that the cancers occurred at unusually young age and that each family had a pair of children who had rare soft-tissue sarcomas. By 1988, they had assembled data from 24 families, which confirmed their conclusions (Li et al., 1988). A mutated TP53 gene evidently was highly potent, since several cancers of different kinds, including sarcomas, often occurred during an affected young person's life. 637 K. W. Kohn Drugs Against cancer CHAPTER32 • II • Ill IV V C. $0n.nuu1 $AltC0"4A DKIAKO C) •U:....ST CANCU / ,-aO aAMO ~ OTHI• MAUGHANT NlOf'lASM © ,PIQOHS <I> CAHCt• a y "1$f0tt'f (NO C:ONll'taMATIO... Figure 32.1. One of the cancer-prone families reported by Frederick Li and Joseph Fraumeni in 1969 (Li and Fraumeni, 1969). Both the frequency and the variety of cancers, including sarcomas, in members of this and their other cancer-prone families were remarkable. Frederick Li and Joseph Fraumeni, 1991, Figure 32.2. Frederick Li and Joseph Fraumeni, who in 1969 described the famelial cancer- prone syndrome that bears they name, and who later helped discover the mutations of the responsible gene, TP53. The molecular unraveling of the Li-Fraumeni familial cancer disposition began in 1979, when Albert DeLeo, Wolfgang Dippold and Lloyd Old at Memorial Sloan-Kettering Cancer Center in New York City, together with Gilbert Jay, Ettore Appella, George Khoury, Garrett Dubois, and Lloyd Law at the National Cancer Institute in Bethesda, Maryland, found a protein, which they estimated to be 53,000-daltons in size, in various cancer tissues of the 638 K. W. Kohn Drugs Against cancer CHAPTER32 mouse (DeLeo et al., 1979; Dippold et al., 1981; Jay et al., 1981) (Figure 32.3). They accordingly dubbed the protein p53 (although it later was found to be closer to 44,000 in size) - and more than 38,000 papers with "p53" in the title have since then been published. They also found a serologically similar protein in human cells, both normal and cancerous (Dippold et al., 1981). In 1990, researchers at the Uniform Services University of Health Sciences and the National Cancer Institute in Bethesda, Maryland, analyzed the TP53 gene of members of a Li- Fraumeni family (Srivastava et al., 1990). They found a single-base mutation, a G replaced by an A, in the gene in several family members but not in others. The mutation was in codon 245, which normally codes for glycine, but in the mutant coded for aspartate. This region of the TP53 sequence, which codes for DNA binding. turned out to be a hot spot for mutations. The affected individuals had (in their non-cancer cells) the mutation in one chromosome but not in the other chromosome of the pair (Figure 32.4). Cancer would arise when the normal copy of TP53 became deleted, leaving only the mutated copy. Thus, the cell would be left without a functional p53 and would lack the protein's tumor-suppressor function. Codon 245 is in the DNA-binding region of the p53 protein (Figure 32.5), where most p53- inactivating mutations occur. Inactivating mutations in this region prevent p53 from binding to DNA, thereby abolishing p53's major functions. Having a mutated TP53 (as well as a normal copy of the gene) made the person a carrier of the Li-Fraumeni cancer-prone condition. Carriers developed cancer when one or more of their cells lost the normal copy of the gene. Characteristic of the Li-Fraumeni syndrome was that a great variety of cancers could develop and that they often developed at an early age (Malkin, 1993). Among the malignancies noted were childhood sarcomas (ordinarily a very rare disease), breast cancers, brain tumors, leukemias, and adrenal carcinomas. Individuals with an inherited loss-of-function-mutations in p53, as in Li-Fraumeni syndrome, had a 50% chance of developing cancer by age 30, when only 1 % of the general population did so, and a 90% chance of developing cancer by 70 (Mullard, 2020). TP53 was at first thought to be an oncogene, whose presence promoted cancer; only later was it realized that TP53 was actually a tumor suppressor gene: it inhibited cancer formation (Levine, 1992). Another characteristic, however, was that the inactive mutated p53 protein was more stable than the normal p53; consequently, p53 concentration was usually higher in cancer than in normal tissues - but the high concentration did not signify cancer causation, because the p53 was mutated and inactive; its cancer suppression function was inactivated. 639 K. W. Kohn Drugs Against cancer CHAPTER32 M, M AFibro- x ,o-3 el h blasts 120 76 --- pS 3 48.5 62 - • 42 27 12 - Figure 32.3. This may be the first published display of the p53 protein (DeLeo et al., 1979). It was a gel-electrophoresis experiment that separated proteins according to molecular weight The p53 band showed up only in the lane labeled Meth A on top and a.Meth A at bottom. This lane was loaded with radioactively (3 5S-methionine)-labeled proteins extracted from a mouse sarcoma (Meth A); the extract was then incubated with Meth A antibodies, after which the resulting immune complexes were isolated and loaded onto the gel. The other 3 lanes in the group were controls showing that p53 did not show up in extracts from normal fibroblasts or when the antibodies came from normal mouse serum. (The lane at the left had molecular weight markers.) (From (DeLeo et al., 1979) with text in red added.) 640 K. W. Kohn Drugs Against cancer CHAPTER32 Family member mutant mutant mutant normal 1 2 3 4 TGC AT GC ATG - -- C AT GCA mutant codon 2 45 - --- -· --- norma l codon 2 45 - - G G Gf G A 1G G ]~ - - C C C - -- - Figure 32.4. Mutations in the TP53 gene in three members of a Li-Fraumeni cancer-prone family. The mutation was in one copy of the gene; the other copy - on the other chromosome of the pair - was normal. The affected individuals had both a normal and a mutant copy of TP53 codon 245 (showing both a G and an A), one in each of the two chromosomes of the pair. The mutation changed a G to an A - which changed the codon from glycine GGC to aspartate GAC (red boxes). The fourth family member was normal and had a GGC sequence in both copies of codon 245. These DNA sequences were from non- cancer cells of the patients; their cancer cells would usually have one chromosome with the inherited mutation and the other chromosome deleted by a random cancer-causing event; thus, the cell had no functional TP53 to suppress cancer formation. (From (Srivastava et al., 1990) with markings in red added.) Another major finding came in 1988 from David Meek of the Salk Institute in San Diego, California; he found that cells often phosphorylate their p53 and identified several amino acids in the protein that were phosphorylated (Meek and Eckhart, 1988). This was the beginning of countless and seemingly endless studies in many laboratories disclosing the enormous complexity of the interactions of the p53 gene. Reviewing the intricacies of all these interactions in 2009, Meek, then at University of Dundee, UK, and Carl Anderson at Brookhaven National Laboratory in Upton, New York, displayed all known at the time in a diagram so complicated that I reproduce it here only to indicate the complexity of p53's functions (Meek and Anderson, 2009) (Figure 32.5.). We had represented some of these interactions in another way in a comparably complicated molecular interaction map (Kohn and Pommier, 2005) (Figure 32.6.). Another view of the central role ofp53 in DNA repair according to Lindahl in 1995 is reproduces in Figure 32. 7 (Lindahl et al., 1995) . 641 K. W. Kohn Drugs Against cancer CHAPTER32 0.~hylllH K370 LS01 (IW01) K313 ""' ... K>20 ""'''" """" .. HO.tC1 s» 546 sss S3tS SIRT1 .,,,, PP> PPI PP'M PP2A C0Ct4A PPM10(Vf191J PP2A 8S6y COl<S RSK2 AIJPI( ""'' I I Po1end11I moclil',w!9 Cd<S - ""° 11n:tyf!Wls CAl(/Cdlk7 p300 mro~ TTK = PKC, """" l)CAlSJ (KAT3B} '''" CK II.~ 1,1,,2 Kl Pl~ c111c2i [El - ERK2 1 Po1ittr1J'151~10fllll modific-..iions 8l ! ... P,'f:8 pp pp =ll•=· p53 A TA01 Pollklnllft110fllll modirie~ion5 UOl',Q Ub N6(F8XO II ) Protein >30 l)IOWinlir'd.d"9' >SO pn:iwlne inel.dng T8P, lnl•l'M1iOIU BACA IIBAAO I, !138P I, TAFI, YY I, PARC, o,••m. A.V>P' AA.I( H lF1..., P'TFU HfM'll Figure 32.5. A comprehensive diagram by David Meek and Carl Anderson in 2009 of the then-known interactions ofp53 (Figure from (Meek and Anderson, 2009) with markings in red added.) 642 K. W. Kohn Drugs Against cancer CHAPTER32 A • C 0 [ • 2 • .. s i:--'-+---,--f-~~~--------~,;-----+--_--_-_-_-_-_-_-_-~-,- --~-~-~-~-$_[!_)_,: ~ --1 '' " l. AIQt' ,.........................• ft ,.__ _ _ _ _,. _ _ _ _ _-!-~ ) 'r-H•~ ,~,-<!i;!e~••!l: • '' ' '' . " '' ' B Th~ doubll:•am:11'~d line indk atn lhat protdl\'I A and U C/11\ bitid 1ocacb 01her. The "ti.,'l®" placed 01, lhe l ine rc-prcsc:n1s lhe A:8 ~ex. Rcrrrsen1:uml afmulti mak-c:ulucomplc:u:!!:.T ii; A: D: y is ( A:S);C'. 'rbis no1a1ioo is c1t1~sibk 10 aBy nun1bc-r of compooen1s in a complex. Co~lcn1 modiflClllion of proccin A . The singlwrrowc<I line indic:1IC'.i th:11 A can l."Xi.st in :i ph(11¢phoryl:lted sta te. 'The nooo n.-,,resems the phos.-phoc')'t:1100 species. Clctwigc of Bocwa1¢n1 bond: dephospboryla1fon of A b}' Dphc)'Sph:11n$C, Stu,khiomctrk CCIIIVl..'f'llicln o(A 10 IJ (Clf' ffll)Vl..'fflenl (ro11) 0 1.c COll'lplltl rne,u 10 MIOlhtr). @--o@ - --0 ---<> - ---i Enzymotic i,1im11lu.tioo oh rcaccioo, 0¢n¢tal $}'mbol for ~ im11lotion of n process. A ~ r bthind lhc arrowhead llignilk!! nccci~ity Gc-11=1 .Jyn1bul fo, inhibi tWfl. Dcgr~llllion prod.ui:ts Figure 32.6. (A) Molecular interaction map focusing on p53 and Mdm2 (Kohn and Pommier, 2005). (Figure from (Kohn and Pommier, 2005) with correction made at bottom: reaction 73 stimulates the cell cycle, reaction 35 inhibits it.) (B) The symbols used in the map (Kohn, 1999, 2001). 643 K. W. Kohn Drugs Against cancer CHAPTER32 111< Kil l!I! Sc ...itr 8tft~ Figure 32.7. Part of a ONA damage repair cartoon showing the central role of p53 (Lindahl et al., 1995). (Drawn for Trends in Biochemical Sciences, Octobe r 1995, based on information from To m Lindahl.) The Mdm2 s tory complicates and clarifies the p53 story. In 1991, Donna George and her colleagues at the University of Pennsylvania identified a gene, Mdm2, that became known to be a central controller of p53 function and cancer initiation. They isolated the gene from a mouse cell line that contained amplified genes present in numerous extrachromosomal nuclear bodies, called double minutes (Figure 32.8.). Working with that cell line for some time, George had noted that the presence of the double minutes seemed to be giving the cells a growth advantage, and she imagined that maybe the amplified DNA in those double minutes might contain a previously unknown oncogene. It may have seemed a far-out possibility but turned out to be spectacularly correct (Fakharzadeh et al., 1991; George and Powers, 1982). Focusing on the DNA in the double minutes in their cell line, they isolated three genes that they called Mdml, Mdm2, and Mdm3 (mouse .double minute). They cloned each gene and transfected it into mouse cells that they then implanted into mice to see if it would cause tumors in the mice. Of the three genes, only Mdm2 produced tumors (Figure 32.9.) 644 K. W. Kohn Drugs Against cancer CHAPTER32 (Fakharzadeh et al., 1991). Moreover, they found that humans had a gene that was homologous to the mouse Mdm2. So, whatever Mdm2 did in mice, the human homolog likely did the same thing in humans. Figure 32.8. Chromosome spread from a mouse cell line containing extrachromosomal nuclear bodies (arrows) from which George and her colleagues later isolated the Mdm2 gene (George and Powers, 1982). The double minutes lacked centromeres and therefore did not connect to the spindle during mitosis, enabling them to multiply and causing the genes they contain to become amplified. (From (George and Powers, 1982).) Table I. Tumorigenicity testing of transfectants Cell line Tumorigenicity• Cell line Tumorigenicity• N/mdm2 12/12** R/mdm2 4/4 N/mdml 1/ 8 R/mdm l 014 N/ mdm3 0/8 R/ mdm3 0/2 N/pCVOOI 018 R/ pCVOO I 2/8 Figure 32.9. Donna George and her colleagues transfected each of the three genes they had isolated from double minutes into cell lines that they implanted into animals to see if tumors would be produced. Only mdm2 consistently produced tumors. The experiments were done in mice (N) and in rats (R). pCV00l was the empty vector that served as control. (From (Fakharzadeh et al., 1991 ).) 645 K. W. Kohn Drugs Against cancer CHAPTER32 Well, what was it that made Mdm2 an oncogene? It did not take long to find out. In a collaboration between Donna George at University of Pennsylvania and Arnold Levine at Princeton, they found that Mdm2 binds and inhibits p53 (Momand et al., 1992). They found that Mdm2 worked by inhibiting the tumor suppressor functions of p53, thus behaving as an oncogene. The role ofp53 in DNA damage r epair -- synopsis A key component of the DNA damage response network was found to be TP53, a gene that is mutated in more than half of all cancer cases, whose transcription product was the p53 protein. When DNA damage signals arrive, TP53 was found to respond in two ways: its first action was to delay the onset of DNA replication, which gave more time for the DNA repair machinery to do its job. If the damage signal persisted, TP53's second action was to activate genes that caused the cell to commit suicide in an orderly fashion (by apoptosis, which is Greek for the falling of Autumn leaves). When TP53 was inactivated by mutation, as it was in most cancers, the cancer cells did not wait long enough for repair to be completed, and the cells entered mitosis while DNA damage was still in place; consequently, chromosomes would become scrambled. In some of the cells - perhaps only a small minority -- the resulting change in chromosome complement would give those cells a greater growth and metastasis potential. Thus, TP53 mutation was found often to be an early step in the route to a malignant tumor. Another, albeit less common, way that TP53 function was impaired was by overexpression of Mdm2, for example by amplification of the Mdm2 gene. Initiation of apoptosis by TPS3 can play a major part in the anti-tumor effect of chemotherapy; thus, cancers whose TPS3 is inactivated by mutation do not respond well to cisplatin. The situation is not so simple, however, because sometimes the opposite is observed (O'Grady et al., 2014). In those cases, it may be that lack ofTPS3 functions in tumor cells allows them to start DNA replication before the DNA damage has been fully repaired; then most of the tumor cells undergo abnormal mitoses and die. If the cancer's TP53 is defective, DNA damage is less likely to kill cancer cells by apoptosis, but more likely to kill the cells because they don't wait for DNA damage to be repaired before they start DNA replication. Thus, the dependence of the sensitivity of a tumor to a DNA-damaging drug. such as cisplatin, on TPS3 function may depend on the balance of those two TPS3- dependent actions -- cell cycle arrest or apoptosis -- in a particular tumor (Vogelstein et al., 2000). Another p53 response to DNA damage was stimulation of DNA repair (Figure 32.7; {Lindahl et al., 1995)). 646 K. W. Kohn Drugs Against cancer CHAPTER32 Molecular interaction map notation The sections below describe some of the details ofp53 function, as understood in 2005 (Kohn and Pommier, 2005), and refer to the molecular interaction map in Figure 32.6 that uses graph symbols defined in the lower part of the figure. The molecular species appear only once on a map and are connected by various types of lines to show their interactions, such as binding. phosphorylation, enzyme action, stimulation, and inhibition. The product of an interaction is represented by a small filled circle placed on a connecting line. For example, a node on a binding interaction line indicates the product of the binding (e.g., a dimer); a node on a phosphorylation line indicates the modified (e.g., phosphorylated) protein. When a line branches, it indicates alternative possibilities, such as competitive binding. The number on a line is used in the text (italicized in brackets) when referring to the function indicated by the line. How p53 and Mdm2 respond in a controlled manner to DNA damage. Cells that are not under stress, normally keep p53 function at a very low level. That is because p53 arrests the cell cycle and, if the stress is high or continuous with DNA damage, causes cells to die by apoptosis. Hence, p53 helps tissues - both normal and cancerous - survive DNA damage, but at a cost. Among the several molecular mechanisms that keep p53 function low, the two most import were binding of Mdm2 and phosphorylation of p53 (Koo et al., 2022). How Mdm2 binds and inhibits p53, as we understood it in 2005, is shown by interaction {15) in Figure 32.6 (Kohn and Pommier, 2005). The N-terminal region of Mdm2 binds to the N-terminal region ofp53, where p53 has transcription-activation domains (TAD). Mdm2 blocks those domains and prevents p53 from activating genes {16), such as apoptosis gene Bax {36}, cell cycle arrest gene p21cipl {82), and Mdm2 itself {55). Thus, there was a negative feedback loop wherein p53 stimulated the production of Mdm2, and Mdm2 inhibited that action by binding p53 (McCoy et al., 2003; Weinberg et al., 2004). Another likely control ofp53 and Mdm2 was the degradation of both proteins in a ubiquitin (Uh) dependent manner (reactions {30} and {32)), which appeared to depend in a complicated way on several components of the network (Kohn and Pommier, 2005). Further intricacies in the controls of p53 and Mdm2 were the effects of phosphorylations of several sites on both proteins. These actions implemented the responses to DNA damage via the activation ATM (the product of the ataxia telangiectasia gene discussed Chapter 29), as shown in the molecular interaction map (Figure 32.6). ATM responded to DNA double- strand breaks and acted in part through the cell-cycle checkpoint regulator, Chk2, {14). A group of mostly parallel interactions (not shown in Figure 32.6) were carried out by the ATM-related protein, ATR, which responded to DNA single-strand breaks and acted via Chkl in parallel to Chk2. 647 K. W. Kohn Drugs Against cancer CHAPTER32 ATM, ATR, Chk2 and Chkl, as well as a few other protein kinases, were able to phosphorylate selected sites on p53 and Mdm2 with a variety of effects as shown in Figure 32.6. ATM phosphorylated serine-15 in the N-terminal region of p53, thereby blocking the binding of Mdm2 to that site (reactions {12) and {17!). ATM also phosphorylated Chk2 {14), which in turn phosphorylated two additional p53 sites {13} that cooperated to block the binding and inhibition by Mdm2 ofp53's transcription-activation domain (TAD) {17). Consequently, ATM efficiently relieved p53 of Mdm2 binding and inhibition. But that's not all. ATM also phosphorylated Mdm2 {18} and did so in Mdm2's N-terminal region where Mdm2 binds p53, thereby blocking that binding. The net effect of all those phosphorylations was the robust activation ofp53 by ATM in response to DNA damage. Activated by DNA damage via ATM and/or ATR, p53 then acted as a tumor-suppressor by stimulating the transcription of genes exerting two major actions: (1) arrest of the cell cycle by the cyclin-dependent-kinase inhibitor gene p21cip1 and (2) initiating cell demise by the apoptosis-activating genes BAX and PUMA (Tyteca et al., 2006). Cell cycle arrest helped normal cells survive DNA damage unscathed by allowing more time for repair; the cells were then less likely to sustain cancer-promoting mutations. The stimulation of apoptosis served as a backup to kill badly DNA-damaged cells that had a high likelihood of becoming cancerous. The p21cip1fWAF1 story - cell cycle arrest in response to DNA damage. In 1993, Wafiq EI-Deiry, working in Bert Vogelstein's laboratory at Johns Hopkins School of Medicine in Baltimore, Maryland, was investigating how p53 suppresses cell division. They knew that p53 was often inactivated in human cancers, perhaps by mutation, and that normal p53 suppressed tumor growth. But how p53 suppressed cell division and tumor growth in response to DNA damage was yet unknown. There were some clues however: p53 could bind to DNA at certain sequences, and it could enhance transcription, but it was not known of which genes. That was what they sought to find out. They thought that the genes induced by p53 may mediate its biological role as a tumor suppressor (Vogelstein and Kinzler, 1992). They found a gene whose expression was stimulated by normal but not by mutated p53 (El-Deiry, 2016; el-Deiry et al., 1993). Introducing the gene into human cancer cells, in the form of a cDNA, suppressed the growth of the cells. Also, they found that p53 in fact binds to the promoter region of the gene. They concluded that the gene's expression was induced by p53 and could be an important mediator of p53-dependent tumor growth suppression (el-Deiry et al., 1993) (Figure 32.10.). They initially called the gene WAFl after the researcher's first name, Wafiq (el-Deiry et al., 1993) - an unusual practice. The protein product of the gene had molecular weight 21,000 and came to be called p21cipl/WAF1; "cipl" for "cdk-interacting protein" was added when the protein was found to inhibit cyclin-dependent kinases (Harper et al., 1993) (reaction 648 K. W. Kohn Drugs Against cancer CHAPTER32 {79) in the lower right of Figure 32.6). The use of the researcher's name as an eponym for the gene was criticized but still often continued, perhaps mostly by authors unaware of its eponymic origin. It nevertheless stimulated the whimsy of my lab colleague and friend, Al Fornace. When EI-Deiry's first paper about WAFl came out in 1993, Al looked over the collection of genes he had isolated and called "gadd" for "growth arrest and DNA damage"; he had isolated those genes on the basis of their being associated with DNA damage and arrest of cell growth (Fornace et al., 1992); he had investigated some of them, like gadd45, which became famous. And indeed, there it was, still waiting to be investigated, Wafiq's new gene. Whereupon Al quipped with some regret that, had he chosen to investigate that gene earlier, he could have called it ALFl! In 1994, Patrick M. O'Connor, then a post-doctoral fellow in our Laboratory, and Wafik EI- Deiry collaborated in studies of p21 WAFl that included members of Vogelstein's lab and my lab (el-Deiry et al., 1994). We already knew that p53 was a transcription factor and tumor suppressor that responded to DNA damage by arresting the cell cycle before onset of DNA synthesis and by initiating apoptosis. We also knew that p53 induced p21 WAFl, and we wanted to find out whether p21WAF1 acted to arrest the cell cycle or to initiate apoptosis. At about that time, Wade Harper and Steve Elledge at Baylor College of Medicine in Houston, Texas, found a 21,000 molecular weight protein (Harper et al., 1993)in bound in protein complexes of cyclin-dependent kinases (cdk's) (Harper et al., 1993). The cdk's, consisting of a kinase whose activity depended on it being bound to a cyclin protein, moved the cell cycle through the Gl/S transition where DNA synthesis begins. The 21,000 dalton protein bound cdk's, and they called the protein CI Pl for cdk-interacting protein. It was the same as p21WAF1 and thus came to be called p21cipl. We found that DNA damage triggered p53 to induce p21cipl, which then bound to cdk's and blocked the cell cycle at Gl/S. The path from p21cipl to effect on cell cycle includes a sequence of three inhibitory steps ([79, 77, 75], bottom right in Figure 32.6A) the net effect of which is inhibition. (Step (35] is a redundant indicator of this.) -- - -~ 0 4 6 8 16 hr p53 ra WAF1 --· WAF1 ◄30 (PROTEIN) -◄21 ◄14 - + Figure 32.10. Identification ofp21cipl (also called WAFl) as a 21,000 molecular weight protein whose expression was stimulated by a normal (W) but not by mutated (M) p53 649 K. W. Kohn Drugs Against cancer CHAPTER32 gene (el-Deiry et al., 1993). The cells were treated with dexamethasone to stimulate gene expression. The cell's p53 gene was inactive because of a mutation. However, a normal p53 was induced in the cells by way of a cDNA. The upper part of the figure shows that the inactive mutant gene (M) was already expressed at zero hours, while the normal "wild type" gene (W) began to be by or before 4 hours. The p21cipl gene (WAFl) began to be expressed increasingly after 6 hours; hence its expression became induced by normal p53 but not by the mutant p53. The lower part of the figure shows that the p21cipl (WAF) protein was relatively pure and had a molecular weight of 21,000. (From (el-Deiry et al., 1993).) Activation of apoptosis by p53 in response to DNA damage. In addition to cell cycle arrest, the second major tumor-suppressor action of p53 was to activate the transcription of genes, such as BAX and PUMA, that stimulated apoptosis of cells that had persistent DNA damage. The molecular interaction map in Figure 32.6 showed phosphorylations that controlled gene expression by p53 generally. Since then, however, acetylation sites were discovered that controlled apoptosis-inducing genes specifically. But the control and effects of an increased number of phosphorylation and acetylation sites on p53 has made the full story too complicated for me to try to unravel here (Sabapathy and Lane, 2019). We could note, by the way, that phosphorylation and acetylation have similar effects on the electrostatic environment of the proteins they bind: phosphorylation adds a negative charge to the serine or threonine they bind: acetylation removes a positive charge from the lysine it binds. An important aspect of the control of apoptosis by p53 was uncovered soon after the map in Figure 32.6 was made (Sykes et al., 2006; Tang et al., 2006; Tyteca et al., 2006). In its response to DNA damage, p53 was shown to become acetylated at a lysine located in p53's DNA-binding domain (K120) by an enzyme known as Tip60 (as well as by a closely related enzyme). If the lysine was mutated to arginine (K120R), the mutated p53 could still bind to DNA and activate the transcription ofp21cipl and arrest the cell cycle, but it could no longer activate the transcription of BAX or PUMA and cause apoptosis. Lysine (K) and arginine (R) both bear a positive charge and, when lysine becomes acetylated, it loses its positive charge. Moreover, the K120R (lysine to arginine) mutation still allowed the p53 to bind to the BAX and PUMA promoters but could not induce them to make the proteins. It is as if the positive charge was required for p53 to bind to the BAX and PUMA promoters (as well as to the p21cipl 2awpromoter) but had to be removed to activate specifically the former two promoters. Thus, K120 acetylation of p53 seemed to tip the balance of the DNA damage response toward apoptosis as opposed to cell cycle arrest (Tyteca et al., 2006). More recently, many additional acetylation sites on p53 have been found and another class of acetylating enzymes has been characterized, exemplifies by an enzyme called p300 (Xia et al., 2022). The molecular interaction map in Figure 32.6 already showed p300 and some of its actions on p53. Reaction [5) indicated its acetylation of several lysine sites on p53. This acetylation was favored by p300 binding to the N-terminal region ofp53 at a site 650 K. W. Kohn Drugs Against cancer CHAPTER32 where Mdm2 also binds (reaction [Bl); the two proteins may compete for binding in this region. p300 was found to bind directly to p53 and to acetylate several lysines in the p53 C- terminal region (Xia et al., 2022) as shown in the map. However, it did not acetylate K120, which was an unusual site that was acetylated only by the Tip60 family. The role of this increasingly complex set of acetylations on the tumor suppressor function of p53 however still remained to be determined. Therapy by targeting the p53-Mdm2 interaction. In 1999, Arnold J. Levine, who had led a team 20 years earlier to discover p53, proposed a novel strategy for cancer therapy based on inhibiting the binding between p53 and Mdm2. The TP53 gene was known to be mutated in about half of cancers that arose in part due to lack of the mutated p53 protein's functions: primarily the DNA damage responses of cell cycle arrest and apoptosis, which normally prevented tissue cells from becoming cancerous. Most of the other cancers, whose TP53 was normal, had overactive Mdm2, often due to amplification of the Mdm2 gene, which suppressed p53 excessively. The new strategy was to inhibit p53-Mdm2 binding with the idea to relieve the excessive suppression of p53 and allow its cell cycle and apoptosis effects to act against those cancers. Based on that idea, researchers at Hoffmann-LaRoche in Nutley, New Jersey, embarked on efforts to find such inhibitors (Vassilev et al., 2004). They expected that a small molecule inhibitor could be effective because of the way p53 bound to the Mdm2 structure (Figure 32.11): the Mdm2 protein had a deep hydrophobic groove into which part of a hydrophobic amino acid chain of p53 could bind securely. They hoped to find a molecule that would bind in that groove and prevent p53's peptide chain from binding there. They began by screening a large number of divers synthetic compounds and found a few that exhibited some degree of Mdm2 binding. Modifications of those compounds to optimize their binding strength and specificity led to potentially useful inhibitors that they called nutlins (for Nutley inhibitor) (Figure 32.12). The nutlins prevented the inhibition of p53 by Mdm2 and allowed the p53, as expected, to inhibit the cell cycle and to kill cancer cells that had normal p53, but not those that had mutated p53 (Vassilev et al., 2004). The discovery of p53-Mdm2 inhibitors after much hard work "was spectacular stuff," recalls Michael Andreeff, an oncologist at MD Anderson Cancer Center. "I was super impressed, and I jumped up and down when I saw it in Science" (Mullard, 2020). Pharmaceutical companies worked hard to develop p53-Mdm2 binding inhibitors good enough for clinical use, but, although better and more promising compounds were made, clinical trials were disappointing and discouraging. Many chemical variants were tried and the outcomes with the resulting new compounds were thoroughly reviewed in 2022, but none of potential drugs had so far passed beyond phase-3 clinical trial (Koo et al., 2022). 651 K. W. Kohn Drugs Against cancer CHAPTER32 Figure 32.11. Structure of a deep hydrophobic groove in the Mdm2 protein where an amino acid chain of p53 binds (Kussie et al., 1996). Three hydrophobic amino acids of the p53 chain (yellow) bind in the groove; they are phenylalanine-19, tryptophane-23, and leucine- 26. The aromatic rings of Fl 9 and W23 lie face-to-face in the groove; the consequent interaction between the rings helps stabilize the binding structure. Cl Ct Figure 32.12. The chemical structure ofnutlin-3, the strongest inhibitor ofp53-Mdm2 binding among the three chemical relatives, called nutlins, made by Hoffmann-LaRoche scientists in 2004 (Vassilev et al., 2004). The compound had two enantiomer (mirror- image) forms, of which only one, called nutlin-3a, was active -- because only that one had the proper 3-dimentional structure to fit in Mdm2's p53-binding groove. (The mirror- image symmetry arises from the two asymmetric carbon atoms in the 5-membered ring.) Additional control ofp53: Mdmx/ Mdm4. As guardian of the genome and suppressor of cancer, p53 helps repair damaged DNA in cells that have sustained such damage or to kill cells with excessive DNA damage. When a cell detects DNA damage, its p53 becomes activated. However, p53 activation becomes harmful to cells undergoing normal cell division in tissues, especially during the development of the embryo. To keep p53 inactive, as we have seen, is the job of Mdm2. Evidently, this job was so critical that evolution added another actor, called Mdmx or Mdm4, that refines and further complicates the control network 652 K. W. Kohn Drugs Against cancer CHAPTER32 In 1996, researchers in the laboratory of A. G. Jochemsen in Leiden, the Netherlands, discovered a protein, similar to Mdm2, that also bound and inhibited p53. They called the new protein Mdmx (Shvarts et al., 1997). The p53-binding domains of the two proteins had 53.6% amino acid identity (Figure 32.13), and the same amino acids in p53 mediated p53- Mdm2 and p53-Mdmx binding. In critical cell types, both Mdm2 and Mdmx were required to fully inhibit p53 activity. There were however significant differences between the actions of the two p53 inhibitors (Marine et al., 2007). p53 stimulated the transcription of Mdm2 but not Mdmx. Absence of Mdmx induced the consequent increase in p53 to transcribe more Mdm2, thereby partially making up for the lack of Mdmx. That was also why absence of Mdm2 produced greater increase in p53 and more severe consequences than absence of Mdmx. Also contributing to that difference was that, in contrast to Mdm2, Mdmx did not induce ubiquitylation and degradation ofp53. Mdm2 and Mdmxwere able to bind to each other via their RING domains (Figure 32.13). Mdm2 could ubiquitylate Mdmx, as well as itself by way of its RING domain, leading to degradation of both proteins (Marine et al., 2007). It is remarkable that the TP53 gene is inactivated by mutation in most cancers, and, moreover, that in most of the cancers whose TP53 gene is normal, p53 (the product of the TP53 genes) is nevertheless inactivated by Mdm2 and/or Mdmx; full inactivation ofp53 required both Mdm2 and Mdmx. p53-binding Acidic domain domain Zn RING NL$ NE$ Nol.S hMOM2 ·~ [ 18 I LIi 101 II r .11 178 192 237 \ 260 289 331 436 I I~... )\ 466 473 4132 53.2% ,. 53.6% 41 .9% ,02 215 255 290 332 hMOMX ' 400 Figure 32.13. Comparison of the domain structures of the human Mdm2 and Mdmx proteins. The greatest similarities were in the p53-binding domain (53.6% amino acid identity) and the RING domain (53.2% amino acid identity). From (Marine et al., 2007)). The p53-Mdm2-Mdmx interaction network a nd f eedback regulation loop. 653 K. W. Kohn Drugs Against cancer CHAPTER32 Mdmx added complexity to the p53 control network, but also added new opportunities for therapy. In response to DNA damage, ATM can phosphorylate Mdm2 (reaction {18} in Figure 32.6) and inhibit the p53-Mdm2 binding (reaction {19n. This would release the inhibition of p53, which could then activate the transcription of Mdm2 (reactions {54) and {ssn and to a lesser extent Mdmx. The abundance of Mdm2 and Mdmx however initially decreases, due to their ATM-induced ubiquitin ligase activity, which leads to degradation of both components of Mdm2-Mdmx dimers (Phillips et al., 2010). p53 can then resume production of Mdm2 and Mdmx, which then limits the duration of p53 activity. The intricacies of the p53-Mdm2-Mdmx interaction network are explained in Figure 32.14. We simulated versions of the network and found domains of parameter space where p53 may have switch-like behavior or oscillations (Kim et al., 2010). It is perhaps not surprising that the network can have multiple complex behaviors. Particularly interesting and possibly important was our finding that p53 activity could switch on relatively suddenly in response to small changes in the reaction rate parameters, as might occur in response to DNA damage. ~~ 1 Mdm2 I I I p53 ' J I ' 1 ' Mdmx -I> ' I ¢ - Ub ' Ub Ub ~ ,, t1 ~ ' ' _- " ' ~I G2/ M arrest _ ..... PUMAet c I Apo ptosis I Figure 32.14. The p53-Mdm2-Mdmx interaction network displayed using the molecular interaction map notation. (Please see Figure 32.6 for symbol definitions.) The map is based on information in (Yu et al., 2020). The map shows p53 activating the transcription of p21cipl, leading to cell cycle arrest at G2/M, and activation of PUMA and other genes, leading to cell death by apoptosis. The transcription activity of p53 is inhibited by Mdm2 and further suppressed by Mdmx, which can bind Mdm2. Mdm2, p53, and Mdmx are degraded after being ubiquitinated by Mdm2. The auto-ubiquitination of Mdm2, however is inhibited by Mdmx, while the ubiquitination ofp53 is enhanced by Mdmx. 654 K. W. Kohn Drugs Against cancer CHAPTER32 Mdmx as a therapy target As already described above, much effort to target the p53-Mdm2 interaction with Mdm2- targeted inhibitors had met with limited therapeutic success. When researchers became aware of the role of Mdmx in p53 regulation, their interest shifted to developing Mdmx- targeted inhibitors. Mdmx was frequently amplified and overexpressed in various cancers and seemed to contribute to the malignancy: it was reported to be overexpressed in a remarkably wide range of cancers, including some cases of breast (19%), colon (19%), lung (18%), stomach (36%), bone osteosarcoma (71%), brain, and thyroid cancers, as well as melanomas and some leukemias and lymphomas (Yu et al., 2020). Many Mdmx-targeted inhibitors of different steps of the interaction network were obtained and some of some of them had promising therapeutic potential. Inhibition of Mdmx would restore the tumor-suppressor actions ofp53 and suppress the tumor-promoting actions of Mdm2. The Mdmx-targeted strategies to meet those goals aimed to (1) block the p53- MDMX interaction, (2) inhibit MDMX expression, and (3) induce MDMX degradation (Yu et al., 2020). These strategies led to the development of Mdmx-targeted drugs that inhibited human xenograft tumors in mice, but clinical trials against human cancer had not yet been reported. Addendum: Nutlin -3 in CellMiner . CellMiner is a set of database analysis tools that can relate (among other things) the activity of a drug with expression or mutation of genes in cell lines (see Chapter 20). I applied these tools to elicit relationships between nutlin-3 activity and TP53 expression or mutation (Figure 32.15). Although previous studies of p53-Mdm2 binding inhibitors focused on acute myeloid leukemia (AML), Figure 32.15 suggests that the nutlin-3 activity relationships may apply as well to acute lymphocytic leukemia (ALL) and non-Hodgkins lymphoma (NHL) cell lines and that there is a subgroup of cell lines with high nutlin-3 response and high expression of non-mutated p53; those may be the lines driven by high Mdm2 expression. (See legend of Figure 32.15.) Therefore, patients potentially responsive to nutlin-3 or other p53-Mdm2-binding inhibitors might be identified as having high Mdm2 expression but no inactivating TP53 mutation. 655 K. W. Kohn Drugs Against cancer CHAPTER32 nllll....~ (4'C"f, C:l"ltl'.01V~•f'•UT) ~• • TP,» ( U 'J>. < ; f ' t ~ N . l l ) nutfln•) (,a, CTRP•lllQ.lO•M IT}'4, Tl'Sl (CIIP, CTI\f>,8n)b(l,,MIT) "'-"O'"IC<l'T'dotlP"I (r)-0,2), P- " '• l ,"4e· l0 ·- ~r,,o,'1«MTlll!%tt"<On (r)• O.•, p-"'1~• 7,h-OS ... " -· . . ., • •• • • • • • . ............ .-.,,..> ·- AWl!l l~ \ A I O~ ~ . ........,.. ,I.ti.I(> AMe H,~ )<l:l ~ t < U ' \ ) " • ••• • •• • --~ ♦ '---- ♦ • CltU() A(.U..,.,_(oof .....~-•!UM t) ~ •b'!"ll l"U> ... ~..-,..~ A<-NpUUl,._,... ~ l) " I) . ... ,. ·- ..• • • ♦ ::-;:i::..(1,,1~'$ ~tf(fo\t ( lllt.l [ '" • \ • ,:. • •• • • i -\ . .. • •, • • t .. • •••• • -~ ! •; t•..,, . " .. .. •• .-··· • • •• • • •.••• ' j • • •• •••• • • " • " • •• • •• •<. • •.: " • • ••• • • • ,. . ' Tll$1 ( lt,P, (;f'tP.e,,o.-.Mll) . .. • TPS) (tl<P, (:U;P..6t04td~ll) • " • Nutlin3 acti\lity 11 TP53 expression ~.,"fr,,,. . ~,,,ii# . , . . . •.~s.,•<.•Y -,, ••••• •,),)>; !J/,,Y,.-✓~W,!§,Z•YN-''>',:,• >•,••1,•, n ~.,.,...,.r'❖t$/ q~o>~V.,, ,;❖-,v~.,..~.-~"fiW 111111 -r,-;,~~, IH • "-"" •• •'-' >/,>~,>N;;,)il '••-> ',>-.,,,,,,,.P. ~•'s,"<'~Y.•-<N "' -.,. ~· :;-~-,.._.,, 7; 'l'o/~•-• "if Nutlin3 • ~ activity TP53 mutation Figure 32.15. Activity of Mdm2-inhibitor nutlin3 correlated positively with TP53 expression and negatively with TP53 mutation in acute lymphocytic leukemia (ALL), acute myeloid leukemia (AML), and non-Hodgkins lymphoma (NHL). From CellMinerCDB version 1.5 (April 2022 release); cell line dataset CTRP-Broad-MlT. Upper left:: nutlin3 activity versus TP53 expression for all cell lines. Upper right: nutlin3 activity versus TP53 expression for ALL, AML, and NHL cell lines. A subset of the cell lines (within the oval) showed high nutlin-3 sensitivity and high TP53 expression. The TP53 in most of these lines was non-mutated (bottom, lower). Bottom: (upper) nutlin3 activity and TP53 expression; red-to-blue is high-to-low. (lower) nutlin3 activity and TP53 mutation (red, mutated; blue, not mutated). High nutlin-3 sensitivity (red) correlated with absence of TP53 mutation (blue). 656 K. W. Kohn Drugs Against cancer CHAPTER32 References Deleo, A.B., Jay, G., Appella, E., Dubois, G.C., Law, LW., and Old, U. {1979). Det ection of a transformation-related antigen in chemically induced sarcomas and ot her transformed cells of the mouse. Proceedings of the Nat ional Academy of Sciences of the United St ates of America 76, 2420-2424. Dippold, W.G., Jay, G., Deleo, A.B., Khoury, G., and Old, L.J. {1981). p53 transformat ion-related protein: det ect ion by monoclonal antibody in mouse and human cells. Proceedings of the National Academy of Sciences of t he United States of America 78, 1695-1699. EI-Deiry, W .S. {2016). p21{WAF1) Mediat es Cell-Cycle Inhibition, Relevant to Cancer Suppression and Therapy. Cancer resea rch 76, 5189-5191. el-Deiry, W.S., Harper, J.W., O'Connor, P.M., Velculescu, V.E., Canman, C.E., Jackman, J., Pietenpol, J.A., Burrell, M., Hill, D.E., Wang, Y., et al. {1994). WAF1/CIP1 is induced in p53- mediat ed Gl arrest and apoptosis. Cancer research 54, 1169-1174. el-Deiry, W.S., Tokino, T., Velculescu, V.E., Levy, D.B., Parsons, R., Trent, J.M., Lin, D., Mercer, W.E., Kinzler, K.W., and Vogelstein, B. {1993). WAFl, a potential mediator of p53 t umor suppression. Cell 75, 817-825. Fakharzadeh, S.S., Trusko, S.P., and George, D.L {1991). Tumorigenic pot ential associated with enhanced expression of a gene that is ampl ified in a mouse t umor cell line. The EMBO journal 10, 1565-1569. Fornace, A.J ., Jr., Jackman, J., Hollander, M.C., Hoffman-Liebermann, B., and Liebermann, D.A. {1992). Genotoxic-stress-response genes and growth-arrest genes. gadd, MyD, and other genes induced by t reat ments eliciting growth arrest. Annals of the New York Academy of Sciences 663, 139-153. George, D.L, and Powers, V.E. {1982). Amplified DNA sequences in Yl mouse adrenal t umor cells: association with double minutes and localization to a homogeneously staining chromosomal region. Proceedings of the Nat ional Academy of Sciences of the United States of America 79, 1597-1601. Harper, J.W., Adami, G.R., Wei, N., Keyomarsi, K., and Elledge, S.J. {1993). The p21 Cdk- interacting protein Cipl is a potent inhibitor of Gl cyclin-dependent kinases. Cell 75, 805- 816. Jay, G., Khoury, G., Deleo, A.B., Dippold, W .G., and Old, L.J. {1981). p53 transformat ion-related protein: det ect ion of an associated phosphotransferase activity. Proceed ings of the National Academy of Sciences of t he United States of America 78, 2932-2936. Kim, S., Aladjem, M .I., McFadden, G.B., and Kohn, K.W. {2010). Predicted f unctions of MdmX in fine-t uning the response of p53 to DNA damage. PloS computat ional bio logy 6, e1000665. Kohn, K.W. {1999). Molecular interaction map of t he mammalian cell cycle control and DNA repair syst ems. Mol Biol Cell 10, 2703-2734. Kohn, K.W. {2001). Molecular interaction maps as information o rganizers and simulation guides. Chaos 11, 84-97. Kohn, K.W., and Pommier, Y. {2005). Molecular interact ion map of t he p53 and Mdm2 logic elements, which control the Off-On switch of p53 in response to DNA damage. Biochemica l and biophysical resea rch communications 331, 816-827. 657 K. W. Kohn Drugs Against cancer CHAPTER32 Koo, N., Sharma, A.K., and Narayan, S. (2022). Therapeutics Targeting p53-MDM2 Interaction to Induce Cancer Cell Death. Int J Mol Sci 23. Kussie, P.H., Gorina, S., Marechal, V., Elenbaas, B., Moreau, J., Levine, AJ., and Pavlet ich, N.P. (1996). Structure of the MDM2 oncoprotein bound to the p53 tumor suppressor transactivation domain. Science 274, 948-953. Levine, A.J. (1992). The p53 tumor-suppressor gene. The New England journal of medicine 326, 1350-1352. Li, F.P., and Fraumeni, J.F., Jr. (1969). Soft-tissue sarcomas, breast cancer, and other neoplasms. A familial syndrome? Annals of internal medicine 71, 747-752. Li, F.P., Fraumeni, J.F., Jr., Mulvihill, JJ ., Blattner, W.A., Dreyfus, M .G., Tucker, M.A., and Miller, R.W . (1988). A cancer family syndrome in twenty-four kindreds. Cancer resea rch 48, 5358- 5362. Lindahl, T., Sat oh, M.S., Poirier, G.G., and Klungland, A. (1995). Post-t ranslational modification of poly(ADP-ribose) polymerase induced by DNA strand breaks. Trends Biochem Sci 20, 405-411. Malkin, D. (1993). p53 and the Li-Fraumeni syndrome. Cancer Genet Cytogenet 66, 83-92. Marine, J.C., Dyer, M.A., and Jochemsen, A.G. (2007). MDMX: from bench t o bedside. Journal of cell science 120, 371-378. McCoy, M.A., Gesell, J.J., Senior, M.M., and Wyss, D.F. (2003). Flexible lid to the p53-binding domain of human Mdm2: implications for p53 regulation. Proceedings of t he National Academy of Sciences of t he United Stat es of America 100, 1645-1648. Meek, D.W ., and Anderson, C.W. (2009). Posttranslational modification of p53: cooperative integrat ors of f unction. Cold Spring Harb Perspect Biol 1, a000950. Meek, D.W ., and Eckhart, W. (1988). Phosphorylat ion of p53 in norma l and simian virus 4Q- t ransformed NIH 3T3 cells. Molecular and cellular bio logy 8, 461-465. Momand, J., Zambetti, G.P., Olson, D.C., George, D., and Levine, AJ. (1992). The mdm-2 oncogene product forms a complex with the p53 protein and inhibits p53-mediated transactivation. Cell 69, 1237-1245. Mullard, A. (2020). p53 programmes plough on. Nature reviews Drug discovery 19, 497-500. O'Grady, S., Finn, S.P., Cuffe, S., Richard, DJ., O'Byrne, KJ ., and Barr, M.P. (2014). The role of DNA repair pathways in cisplatin resistant lung cancer. Cancer t reatment reviews 40, 1161-1170. Phillips, A., Teunisse, A., Lam, S., Lodder, K., Darley, M., Emaduddin, M ., Wolf, A., Richter, J., de Lange, J., Verlaan-de Vries, M., et al. (2010). HDMX-L is expressed from a functional p53- responsive promoter in t he first intron of t he HDMX gene and participates in an autoregulatory f eedback loop to control p53 activity. The Journal of biological chemist ry 285, 29111-29127. Sabapathy, K., and Lane, D.P. (2019). Understanding p53 functions t hrough p53 antibodies. J Mol Cell Bio l 11, 317-329. Shvarts, A., Bazuine, M., Dekker, P., Ramos, Y.F., Steegenga, W .T., Merckx, G., van Ham, R.C., van der Houven van Oordt, W., van der Eb, AJ., and Jochemsen, A.G. (1997). Isolation and identificat ion of the human homolog of a new p53-binding protein, Mdmx. Genomics 43, 34-42. 658 K. W. Kohn Drugs Against cancer CHAPTER32 Srivastava, S., Zou, Z.Q., Pirollo, K., Blattner, W ., and Chang, E.H. (1990). Germ-line transmission of a mutat ed pS3 gene in a ca ncer-prone family w ith Li-Fraumeni syndrome. Nature 348, 747-749. Sykes, S.M., Mellert, H.S., Holbert, M .A., Li, K., Marmorst ein, R., Lane, W.S., and McMahon, S.B. (2006). Acetylation of t he p53 DNA-binding domain regulates apopt osis induction. Mol Cell 24, 841-851. Tang, Y., Luo, J., Zhang, W ., and Gu, W. (2006). Tip60-dependent acetylation of p53 modulates t he decision between cell-cycle arrest and apoptosis. Mol Cell 24, 827-839. Tyteca, S., Legube, G., and Trouche, D. (2006). To die or not to die: a HAT trick. Mol Cell 24, 807- 808. Vassilev, LT., Vu, B.T., Graves, B., Carvajal, 0., Podlaski, F., Filipovic, Z., Kong, N., Kammlott, U., Lukacs, C., Klein, C., et al. (2004). In vivo activation of t he p53 pathway by small-molecule antagonist s of MDM2. Science 303, 844-848. Vogelstein, B., and Kinzler, K.W. (1992). p53 function and dysfunction. Cell 70, 523-526. Vogelstein, B., Lane, 0 ., and Levine, A.J . (2000). Surfing the p53 network. Nature 408, 307-310. Weinberg, R.L., Freund, S.M., Veprintsev, D.B., Bycroft, M., and Fersht, A.R. (2004). Regulation of DNA binding of p53 by its C-t erminal domain. Journal of molecular biology 342, 801- 811. Xia, Z., Kon, N., Gu, A.P., Tavana, 0 ., and Gu, W. (2022). Deciphering the acetylat ion code of p53 in transcription regulation and tumor su ppression. Oncogene 41, 3039-3050. Yu, D.H., Xu, Z.Y., Mo, S., Yuan, L., Cheng, X.D., and Qin, JJ. (2020). Targeti ng MDMX for Cancer Therapy: Rationale, St rategies, and Challenges. Front Oncol 10, 1389. 659 K. W. Kohn Drugs Against Cancer CHAPTER33 Chapt,er-3.1 '"'~ n-t.inoblastoma sro,y - ctNICl'OI ofal.I divltlM cycle 221113a} Drugs Against Cancer: Stories of Discovery and the Quest for a Cure Kurt W. Kohn, MD, PhD Scientist Emeritus Developmental Therapeutic Branch National Cancer Institute Bethesda, Maryland kohnk@oib gov CHAPTER33 The retinoblastoma story - control of cell division. The story dates back to 1657 with the description of a large tumor in the left eye of a 3-year old child by Petrius Pawius, a Professor of Anatomy in Leiden, the Netherlands (Albert, 1987). His description of the tumor as "filled with a substance similar to brain tissue mixed with thick blood and like crushed stone" likely was what is now called a retinoblastoma. The next mention of an eye tumor of this kind did not appear until 176 7 in an article published in Medical Observations and Inquires entitled "The Case of a Diseased Eye Communicated to Mr. William Hunter by Mr. Hayes, Surgeon," which described a tumor in both eyes of a 3-year-old girl (Albert, 1987). We will see why sometimes there is a tumor in only one eye and sometimes in both eyes. Retinoblastoma as a disease entity was described at last in 1809 by the colorful Scottish surgeon and ophthalmologist James Wardrop (Figure 33.1) "to bring under one general view a considerable number of facts, the greater part of which are to be found insulated and not arranged in the works of different authors-and also to describe the disease in particular organs where it has not been hitherto known to exist" Wardrop's meticulous dissections led him to conclude that the tumor in the eyes of children usually arose from the retina. His drawings and clinical descriptions accurately reflected the clinical course of the disease, including the extension of the tumor to the optic nerve and brain, as well as to metastases to different parts of the body (Albert, 1987). The next major event in the story was the invention of the ophthalmoscope in 1851 by the famous German physicist Hermann von Helmholtz ((Helmholtz, 1951); a centennial article published in the AMA Archives of Ophthalmology). This allowed the doctor to see the retina in the living patient and thus to diagnose tumors early enough to save the patient's life by removing the eye (Figure 33.2). Before the advent of anesthesia, this was of course a horrendously painful procedure that sometimes made doctors resort to drastic measures, 660 K. W. Kohn Drugs Against Cancer CHAPTER33 such as bleeding the patient to the point of unconsciousness (from which the patient eventually recovered). Wardrop was already convinced in 1809 that earlier removal of the eye could be life-saving (even though all his patients died from recurrence of the tumor and the procedure was for many years controversial): " ... were I in any case to be assured of the existence of the disease in the early stage, I would have no hesitation in urging the performance of the operation." After chloroform became available for general anesthesia and the ophthalmoscope for early diagnosis, removal of the eye led to reports in the German literature of survival rates of 5% in 1869, 1 7% in 1897, and 57% in 1911 (Albert, 1987). It seems that it took much time and experience for ophthalmologists to become assured when removal of the eye was needed to save life. When there were tumors in both eyes, one of the eyes was sometimes preserved by destroying the tumor with radiation or implanted radon seeds despite the risk of cancer developing later in life (Moore et al., 1931). Chemotherapy with intravenous nitrogen mustard or other chemotherapy was also attempted (Kupfer, 1953) but eventually abandoned. Studying the occurrence rates of various common cancers over time, geneticists came to suspects that various cancers developed over a period of years as consequence of a sequence of several, usually between 3 and 7, mutations (Ashley, 1969). Retinoblastoma was unusual in that the cancer developed as a consequence of just 2 mutations, one in each of a diploid pair in the genome; thus only a single chromosome function needed to be defective (Knudson, 1971). The first mutation was often inherited and made the child susceptible to developing the cancer when a cell in the developing retina acquired the second mutation. It turned out that the two mutations were in the same gene, later called RB in the two homologous chromosomes bearing that gene (Fung et al., 1987). Thus, retinoblastoma was unique in resulting from loss of function solely of the RB gene. Moreover, the mutations would occur early, before the age of five, while the retina was being formed by dividing cells (retinoblasts). Retinoblastomas were familial in about 40% of the cases, while the remainder derived susceptibility from a new mutation in a parent's germ cell or occurred during development. The mutation was in a region ofa chromosome 13 (13q14), where the RB gene was later found to be located (Yandell et al., 1989). A mutation could then occur in the RB gene of the second chromosome 13 of a retinoblast cell that already had the RB mutation in one chromosome 13. The probability of that happening when an RB mutation was inherited and already present in all cells of the embryo was high enough to produce tumors in both eyes. Most familial cases, where the mutation was in the germline, had tumors in both eyes (Benedict et al., 1983). Those cases - bilateral tumors - would occur as a function of age by so-called one-hit kinetics, because there already was one RB mutation in every cell and only one more would be needed to abolish RB function. These kinetics would be linear on a semi-logarithmic plot, as actually seen in the lower plot in Figure 33.3 (Knudson, 1971 ). Patients who survived inherited retinoblastoma were at risk of later developing osteosarcoma, as well as occasional other tumors, initiated by the RB mutations (Hansen et al., 1985). 661 K. W. Kohn Drugs Against Cancer CHAPTER 33 In non-inherited cases, however, tumor development would require two mutations - one in the RB gene in each chromosome 13 of the same cell. This would require a sequence of two low-probability events and would show a delay, as shown in the upper curve in Figure 33.3 - which fit two-hit kinetics (Knudson, 1971). The long-term survival recently of children with the non-heritable form of the disease was 96%, whereas for the inherited form it was 90%; these high survival rates, however, were only in countries that had adequate resources and routine vision testing of children (Manrique et al., 2021). Before the 1990's, an eye with a localized retinoblastoma tumor was sometimes saved by external beam radiation. However, patients with the inheritable form of the disease -- where there usually were tumors in both eyes, one of which was saved by radiation treatment -- would often develop new cancers later in life, because cells in every tissue already had one mutation and only a single new RB mutation was needed to send a cell on its way to malignancy. Radiation treatment of course increased the risk of such mutation. Patients with the non-inherited disease, on the other hand (who had a tumor in only one eye) were much less likely to develop new cancers, because their cells would need two mutations to become malignant (Figure 33.4A and 33.4B) (Eng et al., 1993). The first sign of retinoblastoma most commonly was a white pupil noticed by a parent or pediatrician. Less commonly the first sign was strabismus (crossed eyes) or reduced vision due to the tumor obscuring the macula (central vision part of the retina). More serious eye problems in more advanced cases occurred mainly in so-called developing countries. If the tumor had already extended beyond the eyeball or had metastasized, the survival outlook was for less than one year (Manrique et al., 2021). Chemotherapy of retinoblastoma began in the 1990's, most commonly with melphalan, etoposide, and vincristine administered intravenously. A major advance however was to administer the drugs into an artery that led into the eye. This was done by threading a catheter into the ophthalmic artery under fluoroscopic guidance and permitted many-fold higher local drug concentrations delivered directly to the eye with little or no systemic toxicity. But, if the disease had spread beyond the eye, systemic chemotherapy was needed. In early cases, where the tumor was still localized to the eye, intraarterial chemotherapy cured most patients without a great deal of toxicity and often saved the eye (Manrique et al., 2021). Since cancers generally require mutations in several different genes before they become malignant, why was loss function of the RB gene sufficient to produce retinoblastomas and later sometimes osteosarcomas (or more rarely other sarcomas), but specifically those tumors? The inherited RB mutation would be present in other developing tissues of the embryo and infant, why don't cancers appear in those tissue as often as in the retina or bone? I don't know and it may not be known. 662 K. W. Kohn Drugs Against Cancer CHAPTER33 Figure 33.1. James Wardrop (1782-1869), the colorful Scottish surgeon and ophthalmologist who described the disease entity that came to be known as retinoblastoma. (Portrait by Andrew Geddes; from Wikipedia.) According to Wikipedia, "Wardrop was associated with Thomas Wakley in the founding of The Lancet in 1823 in London, for which he first wrote savage articles and, later, witty and scurrilous lampoons in his column 'Intercepted Letters'. The letters, under the pseudonym "Brutus", were thinly disguised as by leading London surgeons, revealing their nepotism, venality and incompetence. There was enough truth in them to make the parodies sting." Figure 33.2. Ophthalmoscopic view of the retina with an early retinoblastoma in the eye of a child. In early cases like this, where vision was likely to be preserved, radioactive radon seeds were sometimes implanted to destroy the tumor, despite the risk of a radiation- induced cancer developing later in life (from (Moore et al., 1931) with label and arrow added). More recently, treatment of such early cases with intraarterial chemotherapy destroyed the tumor without risk of radiation-induced toxicity. 663 K. W. Kohn Drugs Against Cancer CHAPTER 33 10 0 .8 0 ... ,60 0 ~ 5 ~ ,40 .... "' ~ ,q 0 .20 l:i 0 z .., "' 0 ~ ,o w >- .08 ~ 0 z .06 = w '{/ ,04 ... '-' 0 z . 0 ;::: ~ ... "' ,02 o o lkli!att,ol Costs 12S) S~ tttot Coses (23> - - Two • flit O,n-e - Ofit. llil C11n• .O f 0 10 20 30 40 50 60 AGE IN MONTHS Figure 33.3. Kinetics of the development of retinoblastoma as a function of age of the child in the bilateral cases of retinoblastoma (inherited ; lower curve; one-hit kinetics) or in the unilateral cases (non-inherited; upper curve; two-hit kinetics). See text. 30 i 26.0-i.3.9% I 25 ! > ,-. :; 20 ;5 a: 0 ::E 15 w > ~ ... :, 10 ::E :, " 5 UN:t..ATERA!. i 0 10 TIME AFTER OIA.GNOSIS (YJI) Figure 33.4A. Children with retinoblastomas in both eyes - in whom there was an RB mutation in the germline -- had a high risk of dying from new cancers later in life, in part secondary to eye-saving radiotherapy. Those with a tumor in only one eye had much less risk of dying later in life (Eng et al., 1993). (This was before intraarterial chemotherapy was introduced to treat early cases.) 664 K. W. Kohn Drugs Against Cancer CHAPTER33 35,---- - - - - - - - - - - - - -- - , 30 30,3:. • ., ..i > :; 25 ~ a; 20 0 2 RA.OKITHERAPY • ~ 15 ~ 5 10 HO RA.DCOTMERAPY 2 :, ,- ♦---- 6.4t3h 0 5 ,---- - I AACXOT>,ERAPV 83~ ,..,. 3S9 ., 2S NO RADIOTHERAPY - --- -------- ----- .. ~ " Of Q!llOAEN WITH 811.ATEAA\. RETIHOl!LJ.Si Ot.JA Figure 33.4B. Eye-saving radiotherapy ofretinoblastoma patients increased their risk of dying later in life, usually because they developed new cancers (Eng et al., 1993). Retinoblastoma at the pineal gland: trilateral cases. Curiously, some inherited retinoblastoma cases are called "trilateral"! It is because these cases have, in addition to tumors in both eyes, also a tumor in the pineal gland located near the center of the brain (Figure 33.5). Among the children who have a germline mutation in the RB gene and retinoblastomas in both eyes, about 5% also develop a similar tumor in or near the pineal gland. Before 1995, few of these children survived, but, since then, early detection and intensive chemotherapy with stem cell rescue of the bone marrow has led to long-term survival of more than half of the children (de Jong et al., 2014). Trilateral retinoblastoma is rare but connects with the ancient idea that the pineal was a kind of vestigial third eye. How did that idea arise, what do we know now, and why is the pineal called a gland? The name 'pineal', by the way, comes from the shape of the gland, which resembles a pinecone. Descartes (1594-1650) thought it to be a connection between the soul and consciousness, but this idea was soon dismissed when it was noted that many animals had a pineal yet lacked those special qualities. The notion of the pineal as a kind of third eye traces to ancient Egypt, where the pineal was considered to be the eye of Horus and to Hindu spiritual enlightenment that imagined it as an atrophied eyeball (Shoja et al., 2016). The idea of an atrophied eye may have derived from it seeming to be attached to the brain by way of a stalk. The pineal is the source of melatonin, which it secrets during the night; thus it is indeed a gland. Moreover, this light-dark dependence is functionally like what the eye does, although the gland does not respond directly to light and there seems to be no confirmed neuronal connection between the pineal and the retina. Nonetheless, the 665 K. W. Kohn Drugs Against Cancer CHAPTER33 fact that the same rare tumor type in infants occurs in both retina and pineal gland suggest a relationship in the origin of the two tissues. Figure 33.5. Left, location of the pineal gland in the center of the brain (from Bock's Handbuch der Anatomie des Menschen (1841) Leipzig. Germany, with red oval added). Right, nuclear magnetic resonance image of a retinoblastoma tumor of the pineal gland (from (De loris et al., 2014) with arrow added). Cell cycle control by the Rb pathway. Retinoblastoma tumors arise from uncontrolled growth ofretinoblasts, the cells that form the retina. The uncontrolled growth is initiated by loss of RB gene function. RB is said to be a tumor suppressor gene because tumors develop when both copies of the gene, and hence hence of its protein product, pRb, are inactivated by mutation (or other process). Thus, if even one copy of a cell's pRb protein is in working condition, it suppresses the cell's likelihood of growing into a tumor. RB was one of the first tumor suppressor genes to be described. It was found to be a key factor (although not the sole cause, as it was in retinoblastoma) in the initiation of some of the most common cancers including those of lung. breast, and bladder. Sometimes osteosarcoma appeared without retinoblastoma; in these cases, there was also loss of RB gene functions, showing that both malignancies occurred by the same mechanism. After the RB gene was cloned, definitive evidence was obtained that the gene was a tumor suppressor. The inactivating mutation in the gene was often a point mutation - a change in only one base pair of its DNA - which did not cause a change in the chromosome that could be seen under the microscope (Benedict et al., 1990). Aside from pRb's well-known suppression of the cell division cycle, it was also found to suppress the transcription of many genes of DNA repair and genome maintenance (Lee and Kim, 2022). Would that be useful? Speculatively, high expression of those genes in some normal cell types might entail a risk of genome derangement High expression of those genes in pRb-deficient tumors may help tumors derived from those normal cell types - namely retinoblastomas and osteosarcomas -- to thrive (thus, possibly as an unfortunate 666 K.W.Kohn Drugs Against Cancer CHAPTER33 side effect). However, it also presented possibilities for therapy of those tumors by inhibition of the functions of one or another or several of the DNA repair or genome maintenance genes, as recently proposed by (Lee and Kim, 2022). A critical point in the life of both normal and cancer cells is a commitment to begin DNA replication on the way to mitosis. The process begins with growth factor stimuli that in normal cells come primarily in response replication stress via RAS (Chapter 18) or via estrogen receptor, but in cancer cells come mainly in response to DNA damage via ATM or ATR (Chapter 29) (Matthews et al., 2022). In the previous chapter, we saw that, when p53 is activated in response to ATM or ATR, it stimulates the transcription ofp21cipl, which arrests the cell cycle, thereby allowing more time for DNA repair before cells start mitosis. The steps whereby p21cipl arrests the cell cycle show the central role of the retinoblastoma gene and protein. Our understanding of those steps as of 2005 are shown in Figure 33.6, which is part of the molecular interaction map shown in Figure 32.6A of the previous chapter and in (Kohn and Pommier, 2005). The steps go from p53 to an effect on the cell cycle. But it is easier to understand it in the opposite direction. We begin with the final effect: the induced expression of genes that activate the cell cycle (action 73 in the map), which is stimulated by E2F bound to the E2 promoter of the genes (action 74). The pRb protein binds and inhibits E2F (actions 76 and 75). A Cyclin-Cdk dimer phosphorylates pRb (action 78), thereby inhibiting the binding of pRb to E2F (action 77). Finally, the activity of Cyclin-Cdk is inhibited by p21cipl (action79), the production of the latter being enhanced by p53 (action 82). Now, tracing the steps from p21cipl to the stimulation of the cell cycle, we see a sequence of3 inhibitory actions (actions 79, 77, 75) - the net effect is inhibition, because a sequence of an odd number of inhibition steps yields inhibition. That is essentially how p21cipl could inhibit the cell cycle when p53 is activated in response to DNA damage. Mdm2 was found to counter the actions of p53, consistent with its functions described in the previous chapter. Mdm2 induced the degradation ofp21cipl (actions 80 and 81) and inhibited the binding ofpRb to E2F (actions 83 and 84). Both of these actions of Mdm2 would counter the p53-induced cell cycle inhibition. All of this was a plausible but simplistic yet instructive narrative -- but as often happens as knowledge progresses, further investigation disclosed a more complicated picture, as we shall see. 667 K. W. Kohn Drugs Against Cancer CHAPTER 33 ------~'--- -_-_-_-_-_-_-_-_-80_-_-_ -_-_-_-~--------, Mdm2 83 81 p53- - - - - - - -82 --l-e>@~ I I I I I I I I ]4 .--- f-75 Cell :73 cycle '-------------------------~ Figure 33.6. A molecular interaction map showing how p53, by producing p21cipl, inhibits the cell cycle. This happens when p53 is activated in response to DNA damage. Inhibiting the cell cycle gives more time for DNA repair before the cell begins steps to mitosis. This molecular interaction map is part of a larger map shown in Chapter 32 (Figure 32.6A) and in (Kohn and Pommier, 2005). The steps in the map are explained in the text The symbols are defined in Figure 33.7. T'hc double-arrowed line indic:u es 1h:1.1p-to1ei.ns A and 8 c.1n bind 10 each Qlhct. The "nQdc" l)lacccl <,n the line repte$a'nlli 1he A:8 complex. Represenuiion of multimolcculnr complexes: :r is A:B: y is (A:8) :C. n1is nOiation is cxtensibk to any numb,:r of componcn1s in a compkx. Cov.1le111modilic<1tioo of p,ot-.:in 1\ . The $lngle....1rrow«I li.ne indicates 1ba1 A c..-m ell.isl in a phospho,,ylo•ed st,ue. Tiw: node re1m~seni.s die pbOSphorylated s.pecies. p ~ Cl~vage of a C(>Vl'I ICl'lt bood: de1>hosphorylatioo ofA by a phospha1:1s,e. 0 S1oichio1ne1ric eon\'ersio,1 of A 10 8 (<>r m<>\·eme111 @- @ from one Comp.lrlment to :mo1her). A induce!- ti confonnationul dwngc in 8. -0 Enzymatic stimulation of a reaction. ----e, C.cncral symbol for s1imulation of a process. --tC> A bar behind the anowhc3d signifies ncccssccity -I General symbol for inhibition. ~ Deg.r:1d.'1 1ion productS Figure 33.7. Symbol definitions for molecular interaction maps (Kohn, 1999, 2001). 668 K. W. Kohn Drugs Against Cancer CHAPTER 33 I I ao P I I ''I I I , ,f..i) EtL~ I 3 ~ Ry_:__'cC'1I ...~ i ' ppm,, ~ _,( A-1-- ( .,. E< I ,~ ~°' I ""::::::l I~ <§> · s• C30 cs ~ •"" <€-~•D @ £13 .._c;,~ 5 ~ ?SC) I ' r---t- LI+ -~+- ~-=,-, ---.-1-',;,rl'IT~-, plUYll <§B> f:1,L @+,t C<t ow 1 CIT "-= ""' (1i;i) I - E24 Ji ~ ~ Eaf l ,. C41 c» - _r ~~ ~ "¥ •' I lkl ~ Ct C8 ~ .. U3 l •11 • '•1-------------~~:~~-i ; ~----------' 142 - L .. ~ EIS ' ____ ----------- t-________ ' ······---------------------- , '-------------..,,...,...-1-_J....__.___. __, Figure 33.8. Part of a molecular interaction map of the control of the cell division cycle as understood in 1999 (Kohn, 1999). See explanations in the text. The individual steps in the map are annotated in the paper (Kohn, 1999). (Symbol definitions are listed in Figure 33.7.) To begin with, there is more to be said about Cyclin-Cdk dimers, how they are affected by p21cipl and how they affect pRb. Moreover, each of the aforementioned molecules have several relatives, some of which we will distinguish in our description of the control of the cell cycle via pRb and its relatives. The actions of some of these were already shown in an earlier molecular interaction map (Figure 33.8) (Kohn, 1999). 669 K. W. Kohn Drugs Against Cancer CHAPTER33 There are several types of cyclins, of which we will be concerned mainly with cyclin D and cyclin E. (Cyclin D has three types which function similarly and need not be distinguished here.) Of the several types of Cdk's, the ones that come into play are cdk4 and cdk6, which combine similarly with cyclin D, function similarly, and need not be distinguished here -- and cdk2, which combines with cyclin E. p21cipl has an important relative, p27kipl . There are several E2F species, which we need not distinguish. Finally, pRb has important relatives, p107 and p130. You can spot all of them in Figure 33.8. We move next to the more current picture of how the cell division control system works - the system that controls a cell's commitment to start DNA replication. This is a key decision point, fraught with danger if the genome that is to be replicated is damaged, as it often is in cancer cells, especially after chemotherapy. The decision is made at a so-called cell cycle checkpoint - specifically, the Gl/S checkpoint when cells that are in cycle are confronted with the critical decision of whether to start DNA synthesis. The retinoblastoma protein, pRb, plays a central role in this decision. Most cells in the common slow-growing tumors, by the way, are not in cycle: they are dormant in a so-called GO state. Such cells can be activated by growth factors to move into a Gl state in the cell cycle and, after passing the Gl /S checkpoint, to move toward cell division. To begin on the path to cell division, cancer cells must traverse this checkpoint, also called the restriction point, that holds up the cells' ability to begin DNA replication. The decision to traverse the Gl/S checkpoint is governed by a complicated pRb-dependent molecular interaction network that is still not completely understood (Baker et al., 2022; Matthews et al., 2022). An early version of the details was assembled in the molecular interaction map with annotations for all the steps in that complicated map (Kohn, 1999). But I now show the essentials in a more up-to-date map in Figure 33.9. that should be easier to follow than a full map would be. I'll describe the essentials starting from the bottom of the map, i.e., from the final outcome: the beginning of DNA synthesis with transcription of the necessary genes, one of which is dihydrofolate reductase (DHFR) (Chapter 5). Transcription of these genes is initiated by transcription factors E2F, of which there are several and which operate as dimer complexes with DPl or 2, as shown in the earlier more detailed map in Figure 33.8. E2Fl, 2, and 3 can bind pRb and are thereby dislodged from the E2 promoter sequences; thus pRb (as well as its relatives p107 and p130) inhibit the onset of DNA replication. (There is some uncertainty about the functions of E2F4 and 5. (Baker et al., 2022; Matthews et al., 2022).) For DNA synthesis to begin, pRb (as well as its relatives) have to be inhibited. This happens by a sequence of two regulated steps, the first regulated by cyclin D and the second by cyclin E. The steps are successive phosphorylations of pRb, which is inactivated only when fully phosphorylated through the successive actions by cyclin D-cdk4 and cyclin E-cdk2 (or their respective relatives). 670 K. W. Kohn Drugs Against Cancer CHAPTER33 (To present the effects of the sequence of pRb phosphorylations on the map, I took some liberties with the notation and recognize that a strict interpretation of the inhibition step marked in red would be ambiguous. But hopefully the description in the text will be clear.) As pRb begins to be inhibited, transcription can begin. Interestingly, positive feedback loops come into play - in fact, two of them - because, among the transcription products from the E2 promoters, there are two critical components of the system itself: E2F and cyclin E (Figure 33.9). The effect of these positive feedbacks would be to make pRb inhibition and transcription activation of DNA synthesis genes relatively sudden, as if turning on a switch. Next, I'll describe the two pRb phosphorylation processes, the first governed by cyclin D and coming from replication signals, the second governed by cyclin E and inhibited by DNA damage. Each process is governed by a series of molecular events that I will continue to describe in turn from the bottom towards the top (Figure 33.9). The first set of pRb phosphorylations (P0 ) is carried out by cdk4 when it is activated by binding to cyclin D. The cdk4-cyclin D dimer is stabilized by binding p27kipl but is separated when cdk4 binds p16ink4, which prevents cdk4 from binding cyclin D. Thus, p16ink4 is an effective cdk4 inhibitor. Cyclin Dis produced by transcription in response to replication signals by way of RAS, MYC, or estrogen receptor pathways. In response to DNA damage, p16ink4 is transcribed from a gene that also produces a protein called ARF (for alternative reading frame protein) that in turn inhibits Mdm2, which inhibits p53 and blocks cell cycle entry (Chapter 32) (Matthews et al., 2022). These steps were omitted for clarity in Figure 33.9. p27kipl binds cyclin D-cdk4 (or cyclin D-cdk6 in different cell types) and increases cdk4 or cdk6 activity, apparently by stabilizing the complexes (Baker et al., 2022). p27kipl binds and stabilizes these complexes thereby increasing their cyclin-dependent kinase (cdk) activities. These cdk complexes then phosphorylate the pRb protein (as well as the pRb- relatives p107 and p130) as a first step in the inactivation of these proteins. The second set of pRb phosphorylations (PE) is carried out by cdk2 when it is activated by binding cyclin E. The pathway is in large part similar to the cdk4-cyclin D path. The cdk2- cyclin E dimer too may be stabilized by binding p27kipl and is inhibited when it binds p21cipl. The latter is produced when its transcription is activated by p53, which in turn is activated by being phosphorylated by ATM, which in turn is activated in response to DNA damage. The story of how ATM is activated is told in Chapter 29, including the role of the MRN complex (not shown in Figure 33.9). The fully phosphorylated pRb can no longer bind E2F (red inhibitory step in Figure 33.9). E2F then is free to activate transcription of genes required for DNA synthesis and to initiate progress toward cell division. 671 K. W. Kohn Drugs Against Cancer CHAPTER33 The above description of the molecular interaction map in Figure 33.9 started from the final outcome, the synthesis of genes that propel the cell into S-phase and proceeded upward according to successive regulatory steps. The advantage of this type of display, which we called hierarchical, was presented in (Kohn et al., 2009), which also discussed some additional aspects of the network that were omitted for clarity in Figure 33.9. We see that, up to the level of cdk4 or cdk2, the net action is to stimulate DNA synthesis. The steps above this level show the response to DNA damage that inhibit the events below them. Replication signal DNA damage p ~ ~ Estrogen Receptor p53 p16ink4a ~ ~ Cdk4 @v<H----, ~ ~~ I I l----+-➔ @)<1-------- ~b-----.---< DNA- -t....._!E ~2~ - - IDNA synthesis I Figure 33.9. A molecular interaction map showing the hierarchy of steps leading to onset of DNA synthesis (S-phase) from responses to replication signals or DNA damage (discussed in the text above.). 672 K. W. Kohn Drugs Against Cancer CHAPTER33 The cdk4 inhibitor story. As we saw in Figure 33.9, entry into S-phase of the cell division cycle is tightly controlled by the pRb protein, whose block of cell cycle progression can be relived by the action, first by cdk4 or cdk6 and then by cdk2. Only then can a normal cell set off on the way to dividing. In order to progress, the cell also needs cyclin D, which cdk4 needs to function. But cyclin D will only appear if its production is stimulated by replication signals coming by way of pathways from growth factors, such as RAS, MYC, or estrogen receptor. Cancer cells often have weakened pRb control and tend to enter S-phase inappropriately. Thinking that the weakness may reside at the level of cdk4 or cdk6, medicinal chemists began an intensive search for specific inhibitors of those kinases (Goel et al., 2022). The first cdk inhibitor had actually been discovered earlier in the NC l's cancer drug screen. The drug, flavopiridol (also known as alvocidib), was active against acute myeloid leukemia in some patients (Chapter 20). But toxicity was deemed high relative to benefit and the drug was dropped from clinical trial. It turned out that the drug acted largely on cdk9, which acts on RNA during transcription - which overwhelmed its actions on cdk4/6. The search was on, therefore, for drugs that inhibit only cdk4 and cdk6. The problem was difficult and success was far from assured. Nevertheless, and quite remarkably, by 2004, medicinal chemists had succeeded in synthesizing a highly specific cdk4/6 inhibitor that arrested tumor cells prior to S-phase accompanied by reduced pRb phosphorylation (Fry et al., 2004). By 2015, the drug, palbociclib, received accelerated FDA approval for treatment of hormone receptor positive, HERZ-negative breast cancer. (HERZ is a human epidermal growth factor receptor; see Chapter 1 7.) Palbociclib, as well as two related drugs, abemacicilib and ribociclib, entered the mainstream of clinical practice and were considered one of the most significant advances in breast cancer treatment over the past two decades (Goel et al., 2022) (Figure 33.10). Recent findings however suggest that exactly how the cdk4 inhibitors work may be more complicated than might have been supposed (Baker et al., 2022). 673 K. W. Kohn Drugs Against Cancer CHAPTER33 Palbociclib Abemaciclib Ribociclib N H N'Y'~ N N0 9 r HN.__} O N ii NA " h I N- Figure 33.10. Cdk4/Cdk6 inhibitors approved for treatment of HR•, HERZ" advanced breast cancer (from (Goel et al., 2022)). Summary The retinoblastoma story dates back to pre-anesthesia times when the eyes having malignant tumors in them were removed from very young children but nevertheless failed to save their lives. The outlook improved after the invention of the ophthalmoscope, which allowed removal of the eye before the tumor spread. Therapy improved gradually to a current 90% cure rate of early cases by administration of chemotherapy drugs through an artery to localize chemotherapy drug to the eye. The tumor, retinoblastoma, developed from the retina and had a genetic origin, inherited from the genome of a parent or from a mutation in a stem cell in the early embryo. The former origin usually produced tumors in both eyes, while the latter resulted in a tumor in only one eye. Retinoblastoma was one of many inherited malignancies where a mutated gene was found to have an important role in many cancers. Most cancers were found to have a defect in the molecular pathway dominated by the protein product of the retinoblastoma gene, pRb. The pRb pathway controls the progress of the cell through the cell division cycle. A defect in that control allows cancer cells to move toward inappropriate cell division. Drugs were 674 K.W.Kohn Drugs Against Cancer CHAPTER 33 developed to counter this process and were found to be effective treatment of common types of breast cancer. References Albert, D.M. (1987). Historic review of retinoblastoma. Ophthalmology 94, 654-662. Ashley, D.J. (1969). The two "hit" and multiple "hit" theories of carcinogenesis. British journal of cancer 23, 313-328. Baker, S.J., Poulikakos, P.I., Irie, H.Y., Parekh, S., and Reddy, E.P. (2022). CDK4: a master regulator of the cell cycle and its role in cancer. Genes Cancer 13, 21-45. Benedict, W.F., Banerjee, A., Mark, C., and Murphree, A.L. (1983). Nonrandom chromosomal changes in untreated retinoblastomas. Cancer Genet Cytogenet 10, 311-333. Benedict, W.F., Xu, H.J., and Takahashi, R. (1990). The retinoblastoma gene: its role in human malignancies. Cancer investigation 8, 535-540. De loris, M.A., Valente, P., Randisi, F., Buzzonetti, L., Carai, A., Cozza, R., Del Bufalo, F., Romanzo, A., Angioni, A., Cacchione, A., et al. (2014 ). Baseline central nervous system magnetic resonance imaging in early detection of trilateral retinoblastoma: pitfalls in the diagnosis of pineal gland lesions. Anticancer Res 34, 7 449-7454. de Jong, M.C., Kors, W.A., de Graaf, P., Castelijns, J.A., Kivela, T., and Moll, A.C. (2014). Trilateral retinoblastoma: a systematic review and meta-analysis. The Lancet Oncology 15, 1 157-1167. Eng. C., Li, F.P., Abramson, D.H., Ellsworth, R.M., Wong. F.L., Goldman, M.B., Seddon, J., Tarbell, N., and Boice, J.D., Jr. (1993). Mortality from second tumors among long-term survivors ofretinoblastoma. Journal of the National Cancer Institute 85, 1121-1128. Fry, D.W., Harvey, P.J., Keller, P.R., Elliott, W.L., Meade, M., Trachet, E., Albassam, M., Zheng. X., Leopold, W.R., Pryer, N.K., et al. (2004). Specific inhibition of cyclin-dependent kinase 4/6 by PD 0332991 and associated antitumor activity in human tumor xenografts. Molecular cancer therapeutics 3, 1427-1438. Fung, Y.K., Murphree, A.L., T'Ang. A., Qian, J., Hinrichs, S.H., and Benedict, W.F. (1987). Structural evidence for the authenticity of the human retinoblastoma gene. Science 236, 1657-1661. Goel, S., Bergholz, J.S., and Zhao, J.J. (2022). Targeting CDK4 and CDK6 in cancer. Nature reviews Cancer 22, 356-372. Hansen, M.F., Koufos, A., Gallie, B.L., Phillips, R.A., Fodstad, 0 ., Brogger, A., Gedde-Dahl, T., and Cavenee, W.K. (1985). Osteosarcoma and retinoblastoma: a shared chromosomal mechanism revealing recessive predisposition. Proceedings of the National Academy of Sciences of the United States of America 82, 6216-6220. Helmholtz, H.V. (1951). Description of an ophthalmoscope for examining the retina in the living eye. (Centenial after the original Beschreibung eines Augen-Spiegels zur Untersuchung der Netzhaut im lebenden Auge. Berlin: A Forstner'scnne Verlaqsbuchhandlung, 1851). AMA Arch Ophthalmol 46, 565-583. Knudson, A.G., Jr. (1971 ). Mutation and cancer: statistical study of retinoblastoma. Proceedings of the National Academy of Sciences of the United States of America 68, 820-823. 675 K. W. Kohn Drugs Against Cancer CHAPTER33 Kohn, K.W. (1999). Molecular interaction map of the mammalian cell cycle control and DNA repair systems. Mol Biol Cell 10, 2703-2734. Kohn, K.W. (2001). Molecular interaction maps as information organizers and simulation guides. Chaos 11, 84-97. Kohn, K.W., Aladjem, M.I., Weinstein, J.N., and Pommier, Y. (2009). Network architecture of signaling from uncoupled helicase-polymerase to cell cycle checkpoints and trans- lesion DNA synthesis. Cell Cycle 8, 2281-2299. Kohn, K.W., and Pommier, Y. (2005). Molecular interaction map of the p53 and Mdm2 logic elements, which control the Off-On switch of p53 in response to DNA damage. Biochemical and biophysical research communications 331, 816-827. Kupfer, C. (1953). Retinoblastoma treated with intravenous nitrogen mustard. Am J Ophthalmol36, 1721-1723. Lee, C., and Kim, J.K. (2022). Genome maintenance in retinoblastoma: Implications for therapeutic vulnerabilities. Oncology letters 23, 192. Manrique, M., Akinbolue, D., Madigan, W.P., and Bregman, J. (2021). Update on the Treatment of Retinoblastoma. Neoreviews 22, e423-e437. Matthews, H.K., Bertoli, C., and de Bruin, R.A.M. (2022). Cell cycle control in cancer. Nature reviews Molecular cell biology 23, 74-88. Moore, RF., Stallard, H.B., and Milner, J.G. (1931). Retinal Gliomata Treated by Radon Seeds. Br J Ophthalmol 15, 6 73-696. Shoja, M.M., Hoepfner, L.D., Agutter, P.S., Singh, R., and Tubbs, RS. (2016). History of the pineal gland. Childs Nerv Syst 32, 583-586. Yandell, D.W., Campbell, T.A., Dayton, S.H., Petersen, R., Walton, D., Little, J.B., McConkie- Rosell, A., Buckley, E.G., and Dryja, T.P. (1989). Oncogenic point mutations in the human retinoblastoma gene: their application to genetic counseling. The New England journal of medicine 321, 1689-1695. 676