I. Introduction.

Cancer is a common, complex, and frequently fatal disorder, in which the
mechanisms that normally govern the growth and functions of our cells go awry.
How can this disease be mastered when our knowledge of the behavior of normal
cells remains rudimentary? We contend that the best hope lies in a newly flour-
ishing enterprise, one significantly nourished by our own work, that applies the
techniques of molecular biology to a relatively small set of genes recently im-
plicated in the development of cancer ("oncogenes"). ,

The conviction that cancer is a disease involving structural alterations
(mutations) of genes has its roots in a variety of clinical and experimental ob-
servations. Carcinogens are frequently mutagens; certain neoplastic diseases
display patterns of inheritance resembling genetic disorders; cancer cells often
manifest gross distortions of their chromosomes; and the phenotype of a cancer
cell has a stability evocative of genetic change. However, to the many experi-
mentalists entering cancer research as we did 10 to 20 years ago, the most
striking fact was that the permanent addition of one or a few viral genes to a
normal cell could convert it to a cancer cell. Compared with the daunting com-~
plexity of the vertebrate genome, the simplicity of certain tumor viruses (the
polyomaviruses and retroviruses) seemed refreshingly approachable. Although the
study of such viruses and affected cells from chickens and rodents seemed at
first to represent an expedient compromise with Nature, there have been rich and
unexpected rewards, particularly for those of us working with retroviruses. The
many different oncogenes found to be carried by retroviruses (v-onc 's) have pro-
ven to be derived from normal cellular genes (known as cellular oncogenes [c-
one's} or proto-oncogenes); moreover, the cellular oncogenes show many signs of
being the targets for various mutational events that lead to eancer in animals
and man.

Our laboratory first unveiled cellular oncogenes by finding that the DNA of
birds and mammals contains homologues of v-sre, the oncogene of Rous sarcoma
virus. This and subsequent findings prompted us to propose that v-sre arose by
the capture of a cellular gene (c-sre) by a pre-existing retrovirus without an
oncogene; that c-sre is a gene highly conserved in Nature and vital to normal
cells; and that c-sre (or genes like it) might figure in the genesis of many
cancers, regardless of precipitating cause.

- In the decade that followed, several developments enlarged the repertoire
of cellular oncogenes and strengthened the argument that they are involved in
many forms of cancer. (i) Definition of the composition and origin of retro-
viral oncogenes other than v-sre, here and elsewhere, uncovered many new cellu-
lar oncogenes, nearly twenty at last count; as the number grew, so did the
variety of biochemical mechanisms for inciting neoplastic growth and the pros-
pects for perceiving mechanisms relevant to human cancer. (ii) The study of re-
troviruses lacking their own oncogenes provided the first direct evidence that
known cellular oncogenes could participate in carcinogenesis, as the targets for
activation by viral insertional mutations. This mechanism also affords a novel
means to search for new oncogenes. (iii) DNA-mediated gene transfer into rodent
cells has uncovered active oncogenes in human and other tumors. Again cellular
oncogenes previously identified by their homology with retroviral oncogenes have
frequently been implicated, and an explicit definition of carcinogenic change at
the nucleotide level has been possible for the first time. (iv) Cellular on-
cogenes or their close relatives have been encountered amidst chromosomal
- 2a

anomalies with surprising frequency, the genes sometimes amplified in numbers
and extent of expression and sometimes translocated from one chromosome to
another.

Such findings not only validate the study of retroviral oncogenes as models
for the biochemical basis of human cancer, they also invite a direct assault on
an apparently limited set of cellular genes important in the creation of a canc-
er cell, irrespective of external cause. The questions we address in our work
and this proposal spring directly from this perspective. What is the full rost-
er of cellular genes instrumental in carcinogenesis? What are the normal func-
tions of these genes? How do their oncogenic homologues---either mutant alleles
in tumors or transduced derivatives in viral genomes--~-differ structurally and
functionally from their normal progenitors? What are the biochemical conse-
quences of oncogene activation and how do those consequences lead to the loss of
growth control and the other changes that typify a cancer cell? Might we be
able to use the emerging knowledge of oncogenes to begin the rational design of
strategies for the control of cancer?

II. Expanding the roster of oncogenes.

The size of the complete repertoire of cellular oncogenes is unknown. Re-
troviruses have brought to light at least twenty such genes, and more appear to
be in the offing. For example, two leukemia viruses (E26 and MH-2) under study
in our laboratory harbor as yet uncharacterized genetic loci that may be on-
cogenes. Several oncogenes have also been added to the list by the study of
insertion mutations, oncogenic DNA, and chromosomal rearrangements.

We are placing special emphasis upon the use of insertion mutations to
identify new oncogenes. The value of this approach has been substantiated---and
given rise to yet more experimental opportunities described below-~-in studies
of avian B cell lymphoma and mouse mammary carcinoma. In the lymphomas, a cel-
lular oncogene (c-myc), previously discovered here by its homology with the on-
cogene of MC-29 virus, was found to be activated in virtually all tumors by ad-
jacent insertions of avian leukosis virus DNA. This finding presaged other
kinds of evidence for the involvement of c-mye in human and murine tumors: am-
plifications of the c-myc gene and chromosomal translocations that join c-myc to
immunoglobulin loci. In mouse mammary tumor virus (MMTV)-induced carcinomas, we
have traced proviral DNA to the site of insertion mutations and thereby identi-
fied a cellular gene (called int-1) that is activated by the insertions. This
gene, like other putative oncogenes, has been highly conserved during evolution,
but expression of it has been observed to date only in mouse mammary tumors
bearing nearby proviral DNA. We now seek to know the function of this gene (see
below), and whether it figures in non-viral carcinogenesis in human beings and
other animals. We are also attempting to discover new oncogenes in two other
contexts in which insertion mutations may be operative: in nephroblastomas in-
duced in chickens by myeloblastosis-associated virus, and in primary hepatic
carcinomas (PHCs) associated with infection by hepatitis B virus (HBV). The
study of nephroblastomas may have special rewards: no gene has yet been impli-
cated in human renal cancer, though several chromosomal abnormalities have been
described; the precedent of c-myc sugggests that targets for insertion mutations
may also be involved in chromosomal rearrangements. PHC has a significance that
is self-evident: this disease is among the most common fatal cancers of man
worldwide. In both contexts we seek chromosomal domains that are physically and
-3-

transcriptionally altered by viral DNA. Previously known cellular oncogenes
seem thus far not to serve as targets for insertion mutation in these tumors,
but in one hepatoma we have an interesting lead: an integrated unit of HBV DNA
and flanking cellular DNA is many-fold amplified.

Similar considerations influence the search for oncogenes in tumors without
apparent viral cause. For example, the thought that gene amplification might
bring cellular oncogenes into the tumorigenic scheme led to our recent discovery
of a gene we call N-myc - a distant kin to c-mye that is amplified in human neu-
roblastomas. Our study of neuroblastoma was motivated by karyological evidence
that the tumor frequently, if not inevitably, harbors a domain of amplified DNA.
Searching within this domain for representatives of known oncogenes, we encoun-
tered the previously unrecognized N-myc. We now know that N-mye is present
throughout the vertebrate phyla (as expected for a candidate cellular oncogene);
that the haploid genome of human cells contains a single copy of N-myc, situated
on the short arm of chromosome 2; and that amplification of N-myc may be a
molecular marker for neuroblastoma, perhaps even an etiological factor in the
genesis of the tumor. The rationale that engendered the discovery of N-myc
seems worthy of pursuit in the numerous other examples of human tumors present-
ing with karyological evidence of gene amplification prior to therapy.

We wish to note here an aspect of our work that may appear not to be
directly related to oncogenes and their functions: the study of mechanisms by
which the retroviruses and hepatitis B viruses replicate. In reality, this work
has provided an important theoretical base for many of our excursions among on-
cogenes; in particular, it inspired the search for retroviral insertion muta-
tions that activate cellular oncogenes. We have a continuing commitment to
study retroviral and host factors important for integrative recombination; the
transposition of genetic elements structurally related to retroviral proviruses
in Drosophila; and the genetic program and replicative strategy of hepatitis B
viruses in lower mammals.

III. The functions of retroviral oncogenes.

 

The oncogenes carried by highly tumorigenic retroviruses are among the most
potent and experimentally malleable carcinogenic reagents known. Within hours
after infection---or within minutes after temperature shift of cells infected
with thermosensitive mutants---normal cells can be converted to tumorigenic
cells through the action of modest amounts of a single protein. Furthermore,
work here and elsewhere has sketched some preliminary versions of how such pro-
teins act; for example, several of the v-one proteins exhibit protein kinase ac-
tivity specific for tyrosine and reside in the plasma membrane. Future chal-
lenges include the identification of (1) the functionally-significant cellular
targets for tyrosine kinases; (ii) the structural features of the transforming
proteins important to their localization and activity; and (iii) the biochemical
properties of those oncogenic proteins that are not tyrosine kinases. The ulti-
mate goal is to understand in detail the apparently varied oncogenic mechanisms
used by viral genes whose relevance to human cancer can now hardly be ques-
tioned.

We continue to place special emphasis upon the src gene of Rous sarcoma
virus and its product, a 60,000 dalton phosphoprotein (pp60.~S£°) that phospho-
rylates tyrosine residues in many proteins. The power of a joint genetic and
a

biochemical approach is manifest here: we are isolating and engineering new sre
mutants to probe (i) the basis and significance of the tyrosine kinase activity,
(ii) the interaction of pp60’ Src Src with various cellular proteins, (iii) the im-
portance of serine and tyrosine. phosphorylations within pp60"-S=<, (iv) the ter-
tiary structure and membrane attachment of the protein, and (v) our recent find-
ing that sre may encode a second protein in an alternative reading frame. In
addition, we we are attempting to isolate host mutants that fail to respond , to v-
sre in hopes of identifying: host factors required to mobilize pp60° ——, im-
mediate targets for phosphorylation, and other components in the transformation
pathway. We have mutants in hand that display only a portion of the transformed
phenotype and some that are competent to affect some host cells but not others;
these too will be useful in a dissection of transformation by sre. Recently we
have placed the v-sre gene under hormonal regulation (by linking _ it to the
glucocorticoid-sensitive promoter from the mouse mammary tumor virus); this
maneuver has produced another system ripe for biochemical exploitation, since we
can now reproducibly modulate the phenotype by making relatively small adjust-
ments in the dose of pp60’ == and its kinase activity.

If we are to realize the full benefits of genetic analysis, we must learn
much more of the biochemistry of pp60” “Ste. Important objectives include: (i)
more effective purification of the protein in bulk, perhaps by exploiting pro-
duction in bacterial hosts; (ii) detailed characterization of the enzymatic
reaction catalyzed by pp60 “sre and comparisons with better known protein ki-
nases; (i114) identification of the amino acid sequences that define preferred
sites for phosphorylation by pp60” “Src. (iv) the design and synthesis of pep-
tides that represent model substrates and potential inhibitors of the kinase ac-
tivity; (v) the use of inhibitors to obtain more decisive evidence that tyrosine
phosphorylation is responsible for the tumorigenic capacity of pp60” “sre, and
(vi) the development of more effective strategies for the biochemical recogni-
tion of cellular proteins phosphorylated ye¥ré the enzyme (approaches could include
affinity chromatography on purified pp60” a and antisera to both phosphotyro-
sine and peptide substrates). As the mechanistic analyses of sre advance, prin-
ciples by which the effects of the gene on cells might be reversed should em-
erge, presenting us with new opportunities to interdict malignant growth.

There is little in the design of our studies on v-sre that cannot also be
applied to the other viral oncogenes that have caught our ur fancy (v-fps, v-myc,
v-erb-A, v-erb-B, and v-myb). Although only one of these (v-fps) is a protein
kinase, genetic analysis of both viral gene and cellular response, purification
of the protein products of the genes, and biochemical quests for the cellular
macromolecules with which the transforming proteins may interact will neverthe-
less be central to our explorations of how these varied genes transform cells to
neoplastic growth. We believe that it is essential to study a multiplicity of
oncogenes in order to achieve a fair sampling of the biochemical mechanisms that
can underlie tumorigenesis, and to learn how these mechanisms affect cells of
different embryological lineages.

IV. The normal and carcinogenic functions of cellular oncogenes;

 

The findings that implicate specific cellular genes in tumorigenesis---
whether the genes were encountered first in retroviral genomes (e.g., sre), at
sites of insertion mutations (e.g., int-1), or in amplified DNA (e.g., N-myc)
---raise two obvious but unanswered questions: what are the normal functions of
-5-

those genes? and how do alterations of them contribute to disease? The clues
at hand are inspirational but not decisive: many of these genes are strikingly
conserved throughout evolution, arguing that they are required for some funda-
mental life process; variations in expression of these genes among tissues or
during embryogenesis and the affiliation of certain oncogenes with certain types
of tumors suggest roles in development or differentiation; and comparisons of
viral oncogenes (or mutant cellular oncogenes) with their progenitors reveal
differences in both the structure and abundance of products, implying that both
Qualitative and quantitative factors may operate.

The specific objects of our attention are the cellular genes homologous to
the viral oncogenes discussed in the preceding section (sre, myb, myc, fps,
erb-A, erb-B); genes uncovered as the targets of insertion mutation (int-1, as
well as myc and erb-B, and perhaps others yet to come); and genes involved in
chromosomal rearrangements (N-myc and c-myc, principally). The level at which
we can approach the overriding questions must vary in each case. For genes of
which we are largely ignorant (e.g., int-1 and N-myc), the primary issues are a
more complete structural definition of the genes and the identification of the
protein products. In cases. for which proteins have been identified (e.g., c-sre
and c-myc), the immediate issues are more sophisticated: discovering the
relevant biochemical properties of the gene products and defining the genetic
and biochemical distinctions between normal and altered proteins.

We and others have wagered that cellular oncogenes may play important roles
in the growth and development of normal organisms. This is an exciting pros-
pect, because biological scientists have previously had little genetic purchase
on.cell division and differentiation in higher eukaryotes. But how are the
roles of cellular oncogenes in normal cells to be sought? In the belief that
genetic strategies are likely to be most telling, we have turned to the two ex-
perimental systems that offer the most facile access to the genetics of eu-
karyotes: yeast (Saccharomyces cerevesiae) and the fruit fly (Drosophila melano-

gaster).

Genes related to many (perhaps all) of the cellular oncogenes of birds and
mammals can be found in Drosophila. The great store of information accumulated
from classical genetic analysis, as well as recent remarkable progress with
techniques to modify the genome of Drosophila, should make it possible to ask
directly how cellular oncogenes might contribute to the growth and development
of the fly. The prospects for this work can be demonstrated by our experience
to date with Drosophila sre: the identity of the gene has been documented by nu-
cleotide sequencing; the gene has been mapped to position 64B on chromosome 3; a
tyrosine-specific protein kinase apparently encoded by the sre locus has been
identified in tissues of Drosophila; and expression of the locus has been shown
to fluctuate dramatically during the course of Drosophila embryogenesis. These
findings prepare the way for a mutational analysis of Drosophila sre and have
emboldened us to proceed with isolates of other cellular oncogenes, including
fps, ras, myc, N-mye and int-1.

Although it is conceivable that cellular oncogenes would serve different
physiological purposes in yeast and in metazoan organisms, the ease and preci-
sion with which the genome of yeast can be manipulated call for exploitation
whenever possible. Cautious excitement has therefore greeted the discovery here
and elsewhere that yeast may harbor recognizable members of the cellular on~
cogene family. Candidates to date include sre, fps, myc, mos and ras, and the
- 6-

search is on for others. Once the identity of any of these genes has been vali-
dated, targeted mutagenesis and other genetic manipulations should quickly re-
veal whether the gene is essential to yeast and what its physiological purpose
might be.

Genes can now be introduced into the genome of mice by microinjection of
single-cell embryos. We hope eventually to use this remarkable technology in
our search for the normal and pathogenic functions of cellular oncogenes, and we
have begun to develop the necessary reagents. But there are presently serious
constraints on the utility of the procedure, since it has not yet proven possi-
ble to dictate whether, when or where the implanted gene will be expressed. We
will wait for relief of these limitations before pursuing the creation of
transgenote mice for our purposes.

Views of how the growth of normal and neoplastic cells is governed have
merged dramatically in the study of polypeptide growth factors. At least one of
these, platelet-derived growth factor (PDGF), is a close genetic relative of the
protein encoded by the viral oncogene v-sis; several (PDGF, epidermial growth
factor [EGF], and perhaps insulin) elicit pl phosphorylation of tyrosine in cellu-
lar proteins, possibly by activating tyrosine-specific kinase activity residing
on the receptors to which the growth factors bind; and illicit production of
growth factors may be a central anomaly in many forms of cancer cells. Antici-
pating (and provoked by) these associations, we have initiated two lines of at-
tack designed to explicate the cellular response to growth factors. First, we
have sought and found modifications in the structure and function of the protein
encoded by the cellular sre gene during the early cellular response to PDGF.

The nature of the modifications and their role in the response to PDGF are under
study. Second, we have begun the molecular cloning of genes that encode the re-
ceptors for EGF and PDGF. The cloned genes will reveal valuable details of re-
ceptor structure. More importantly, the clones can be manipulated in vitro and
then transplanted back into living cells in order to dissect the role of the re-
ceptors in the response to their ligands.

 

How large is the role of cellular oncogenes in carcinogenesis? What
changes must be wrought in these genes to make them pathogenic? How might these
changes affect the functions of the genes? We seek the answers to these ques-
tions in a multifaceted approach that lies at the core of our research program,
and that employs all of the oncogenes now under our scrutiny.

(1) The protein products of these genes must be identified, produced in
quantity, characterised in detail and compared to their pathogenic kin. In
these efforts we are assisted by the remarkable progress that has occurred in
the cloning and expression of eukaryotic genes in bacteria and in the production
of poly- and monoclonal antisera with antigens synthesized chemically or in bac-
teria.

(11) The structural details of cellular and viral oncogenes must be ela-
borated and compared in a search for features that can modify function.

(iii) The carcinogenic capacity of cellular oncogenes needs to be tested
persuasively in cell culture and in animals, using viral vectors and other means
of gene-transfer. Here the dividends of pursuing retroviral replication are
again apparent: various versions of cellular oncogenes - including those such as
int-1 and N-myc, not previously found in retroviral genomes - can be harnessed
-7-

to retroviral elements and delivered throughout a host animal or to all cells in
a culture.

(iv) The genes can be manipulated in vitro to describe the variety of
structural alterations that can produce or enhance pathogenicity. The starting
point in such efforts is the genesis of hybrid genes that combine domains from
viral oncogenes and their cellular progenitors, but manipulation of the cellular
genes alone should also be revealing, especially when retrieved from normal and
cancerous cells. The functional consequences of the manipulations can be tested
by gene-transfer in culture and in animals.

(v) Further evidence to implicate cellular oncogenes in the genesis of hu-
man tumors must be sought. All of the strategies we have learned before, and
others yet to come, can be employed in this search: pursue new oncogenes identi-
fied by insertional mutagenesis in the appropriate human tumors (for example,
int-1 in carcinoma of the breast, and the presently unknown genes we hope to
find in hepatic and renal carcinomas); exploit the amplification of DNA in human
tumors to implicate known cellular oncogenes in tumorigenesis and to find poten-
tially novel oncogenes; continue our search for inordinate expression of known
cellular oncogenes in diverse samplings of human tumors; use established pro-
cedures for gene-transfer to seek active oncogenes in the tumors we study; use
viral vectors to develop more efficient and more comprehensive strategies for
the rescue of tumorigenic genes from human neoplasia; and explore the use of
gene-transfer and somatic cell genetics for the detection of faulty regulatory
devices that might contribute to neoplastic growth but that would not be counted
as members of the oncogenic family.

Vv. Epilogue.

We make this proposal with two firm convictions: that cancer is at its
heart a genetic malady, and that retroviruses have provided us with our best
grip on the malady. If there is a common genetic substrate on which various
carcinogens act, cellular oncogenes are likely to be part of that substrate. If
we can unravel the means by which these genes act, we should be able to perceive
at least the outlines of a rational design for the prevention and cure of canc-
er.