Technical Report No. IRL 1128

CYTOCHEMICAL STUDIES OF PLANETARY MICROORGANISMS
EXPLORATIONS IN EXOBIOLOGY

Status Report Covering Period January 1, 1971 to December 31, 1971
For

National Aeronautics and Space Administration

Grant NGR-05-020-004.

 

Instrumentation Research Laboratory, Department of Genetics
Stanford University School of Medicine
Stanford, California 94305
Report to the National Aeronautics and Space Administration

“Cytochemical Studies of Planetary Microorganisms - Explorations in Exobiology”

NGR 05-020-004

Summary Report Covering Period January 1, 1971 to December 31, 1971

Instrumentation Research Laboratory, Department of Genetics
Stanford University School of Medicine

Stanford, California

elocan cect a

[pect Lederberg
[princival Investigator

Copp

lliott C. Levinthal, Director
Instrumentation Research Laboratory
TABLE OF CONTENTS

A. INTRODUCTION

B. PROGRAM RESUME

I.

II.

IIl.

IV.

VI.

VII.

VIII.

IX.

Dendral
a. Dendral Chemistry
b. Computer Science

1. Applications of Artificial Intelligence to
Mass Spectrometry

2. Extending the Theory of Mass Spectrometry by
a Computer (Meta-Dendral)

c. Computer Aided Research Instrumentation
1. Remote Mini-Computer Programming
2. Computer Assisted Experiment Set-up
3. High Resolution Mass Spectrum Acquisition and Analysis
4, Instrument Control Interface
5. Function Allocation Between Computers
Constituents of Urine
Applications of Gas Chromatography
Chlorination of DNA Bases
Mass Spectrometry of Organic Compounds
Analysis of Natural Products by Mass Spectrometry
Mariner Mars 1971 Orbiter Photography
Viking Lander Imagery Investigation
Cell Separation
a. High Speed Fluorescent Cell Sorter

b. High Speed Volumetric Cell Sorter

C. REPORTS AND PAPERS
A. INTRODUCTION

This report covers the activities of the Instrumentation Research
Laboratory during the calendar year January 1, 1971 to December 31,

1971.

While the main support of the IRL activities during this period
continued to be the NASA grant NGR-05-020-004, some of the funds have
come from other grants, other agencies, and in some cases private
institutions. This report includes all the activities of the laboratory
which relate to or have benefited from NASA support regardless of

whether or not they were primarily supported by this NASA grant.

The Dendral project receives direct support from NIH grant GM 00612-01
and indirectly from the ACME grant NIH RR 00311-06 and Professor

Feigenbaum'’s ARPA grant SD-183.

Our work in the area of cell separation was supported in this period

by NIH grant GM 17367 and NIH contract 69-2064. We received some
support for Dr. Duffield's work on the interaction of chlorine compounds
with DNA under NIH grant AM 12797 and are seeking additional support

for this effort.
The efforts on image processing for the 1971 Mars Mariner are a

joint effort with the Artificial Intelligence Laboratory of the Computer
Science Department under Professor John McCarthy. It receives support
from NASA grant NGR-05-020-508 and JPL Contract 952489. Our work on

lander camera imagery receives support under Langley contract NAS 1-9682.

These interrelationships benefit our NASA program in two ways. They

aid the rapid utilization for medicine, and biology in general, of the
ideas and skills developed because of our interests in space missions.

They not only provide us with the opportunity of testing instrumentation
methods, applicable to future NASA programs, in circumstances of

solving current scientific problems but also provides much needed
additional support. This support hopefully will permit the Instrumentation
Research Laboratory to continue to function with sufficient technological

depth and breadth to allow responses to future NASA needs.

For these reasons we have always felt it desirable to carry out some
laboratory work that could lead to new and fruitful relationships. For
example, we have collaborated with Professor Marvin Chodorow of the
Applied Physics Department on a preliminary evaluation of the application

of microwave acoustic scattering to problems of cell discrimination.
The general areas of the Program Resume, Part B of the report, are:

Il.

IIl.

IV.

VI.

VII.

VIII.

IX.

Dendral

Constituents of Urine

Applications of Gas Chromatography

Chlorination of DNA Bases

Mass Spectrometry of Organic Compounds

Analysis of Natural Products by Mass Spectrometry
Mariner Mars 1971 Orbiter Photography

Viking Lander Imagery Investigation

Cell Separation
B. PROGRAM RESUME
I. DENDRAL

This section is a general review of what is known as the Dendral
research project. It is presented in three parts; a) Dendral
Chemistry, b) Computer Science, and c) Computer Aided Research

Instrumentation.

a. Dendral Chemistry

Historically and in summary the Dendral algorithmwas developed by
Lederberg)“ for the enumeration of all the possible isomers of a

given empirical composition. The atomic connectivity is represented

by a linear notation and the format itself contains the information
required te rank the linear formulas for a set of isomers in a canonical
dictionary order. Provision exists for the exclusion from the output
of chemically unstable (or otherwise undesirable) structures by placing
that subgroup of atoms on a Badlist. This list is referenced during
structure generation and all structures containing the Badlist subgroup
are not produced. Conversely, if only one functional group (or a
specified number of such groups) is required then this subgroup is
placed on Goodlist and only those structures containing this entity are

constructed by the program.
A computer program, Heuristic Dendral, was developed for the computer
interpretation of aliphatic ketone mass spectra.” This program consisted
of several distinct sections. First, the Preliminary Inference Maker,
whose function was to examine the low resolution mass spectrum of an
unknown aliphatic ketone to deduce whatever portions of the structure of
the unknown were possible. This was achieved by presenting this section
of the program with a theory of the fragmentation processes of aliphatic
ketones in a mass spectrometer. The results of this survey of the data
are then presented to a structure generator (the Dendral Algorithm
mentioned above) where all possible chemical structures which are
consistent with the mass spectrum are enumerated. Those structures which
are inconsistent with a second order theory of mass spectrometry are then
removed. This was achieved using a predicted mass spectrum for every
candidate and comparing this predicted spectrum with the original unknown.
Any significant differences between the two mass spectra result in the
removal of that candidate structure from further consideration. Finally
each of the remaining structures were ranked by. a scoring function on the
basis of the closeness of fit between the predicted and unknown mass spectra.
This resulted in an ordered list of plausible molecules as a solution

for the unknown mass spectrum.

Our results, gathered from a series of aliphatic ketone mass spectra,
indicated that the correct solution was invariably ranked either first,
or in some rare instances, second. These results provided the impetus
for further development of the Heuristic Dendral program, and for the
next task, domain aliphatic ethers® were chosen. The program worked
satisfactorily with this class of compound as shown from the output

5
lists of structures. The correct answer to a particular problem
always appeared in the output, generally as the first or second

choice of the program.

The program was then expanded to accomplish the interpretation of

alphatic amine mass spectra.’ This represented an added complexity
insofar as the number of amines of a given carbon content is considerably
larger than either the corresponding ethers or ketones. In other

words the total search space the program had to encounter is significantly

enlarged.

The next logical extension of the Heuristic Dendral project was in the

interpretation of mass spectra generated by the general class of

compounds represented by 1.°
R-X-R'
I, X = 0 (ethers and alcohols)
X = N (amines)
X= S (thioethers and thiols)

Confronted with an unknown mass spectrum, the program was capable of
deciding to which of these class of compounds the unknown belonged.
The required heuristics were then employed to identify the precise
compound which had produced the mass spectrum. Using nuclear magnetic
resonance data,the program invariably arrived at only one solution per
problem, viz the correct answer. Without this spectral evidence (i.e.
using mass spectrometry alone) the number of possible answers was
reduced from many thousands (often millions) to a few hundred. The

speed with which the program operated (of the order of 20 spectra per
minute) made it superior to the human chemist in the interpretation
of mass spectra in the limited area of aliphatic ethers, thioethers,

amines, alcohols and thiols.

A review of the development of the Dendral Algorithm and Heuristic

Dendral to this point has been published.”

b. Computer Science

The close interaction between computer scientists, chemists, and
biomedical scientists and engineers has always been a unique feature of
the Dendral project. In the last year cooperating scientists have

pushed development of the computer programs in two directions of interest
to the computer science community: (1) Application of Artificial

Intelligence and (II) Theory Formation by Computer.

As an application, the artificial intelligence work is invested with a
richness of concept and organization that is .a direct consequence of the
requirement to attend to the detail of a real physical problem - and that
is lacking in most other A.I. studies that use "toy" problem environments.
Because of this richness of the problem environment, and the success of
Dendral in solving real problems, the effort is the focus of much attention

and comment in the A. I. community.
In theory formation, the Meta-Dendral program has the potential for
greatly enhancing the power of mass spectrometry (and perhaps other
instrumental methods) as a research resource. A successful Meta-Dendral
program will be a "resource amplifier" in the link between mass spectral

data and useable scientific knowledge.

1. Applications of Artificial Intelligence to Mass Spectrometry
Objectives:

The overall objective of this part of the research program is to
extend the Heuristic Dendral program to analysis of the mass spectra of
complex organic molecules. The following specific objectives, which now
rest on a firm foundation of experience through our efforts on the
steroidal hormones in the estrogen family, can be identified:

(A) Generalization of the programming techniques to make them

compound class independent as far as possible.

(B) Incorporation of the cyclic generator into existing
algorithms to expand the capabilities of structure and/or
substructure generation.

(C) Refinement of planning rules to minimize the number of
molecular structures considered by the Dendral algorithm.

(D) Exploitation of the information contained in ancillary
mass spectrometric techniques, such as metastable ion
spectra, low ionizing voltage spectra and mass spectral

pattern shifts in isotopically labeled molecules, for
inclusion in Dendral.
(E) Structuring of the programs to allow use of information
obtained from other spectroscopic or chemical techniques.
(F) Use of ACME facilities for the complete integration of the
presently separated tasks of data acquisition and data

analysis.
Progress:

Considerable progress has been made in the structural analysis of
estrogenic steroids by Heuristic Dendral. The results obtained to this
date are important from two standpoints. Firstly, this class of
molecules represents the first instance of application of the problem
solving capabilities of artificial intelligence to biologically
important molecules. Secondly, the initial success of this approach for
estrogens is encouraging as it has important implications for extension
of the method to other complex molecular systems. A paper describing

this program has been submitted for publication.*

*wapplications of Artificial Intelligence for Chemical Inference. VII.

An Approach to the Computer Interpretation of the High Resolution Mass
Spectra of Complex Molecules. Structure Elucidation of Estrogenic
Steroids." D. H. Smith, B. G. Buchanan, R. S. Engelmore, A. M. Duffield,
A. Yeo, E. A. Feigenbaum, J. Lederberg and C. Djerassi. Submitted for
publication in J. Am. Chem. Soc. (1972).
The operation of the program described in the paper may be briefly
summarized. Given the basic estrogen skeleton (without substituents),
information on fragmentation processes relevant to estrogens, metastable
information and a complete high resolution mass spectrum, Part I of the
program, termed analyzer, identifies molecular ions, determines the
numbers of various substituents that must be placed on the skeleton,
performs the fragmentations for a given placement and searches the mass
spectral data for peaks supporting this placement. Part II of the program,
termed synthesizer, attempts to reconstruct plausible molecular
structures to fit the conclusions of Part I. Part III of the program
allows input of rules or chemical knowledge derived from other techniques
to search through the Part II results for a most plausible molecular

structure.

As the performance of the existing estrogen analysis program has
improved through a continuing feedback mechanism whereby results are
examined and programs revised as required, certain. facts have become
apparent that bear directly on the specific objectives (A-F) mentioned
above. For example, it is now apparent that both the analyzer and the
synthesizer can be structured to be independent of the specific compound
class treated by the program, with changes necessary only in the input
data to the analyzer mentioned above. Restructuring of these routines,
now largely accomplished, will result in a much more general program

that can easily be utilized for other classes of organic compounds.

10
Parts II and III of the program are formally equivalent to a
cyclic structure generator, although on a much more simplistic level.
The present cyclic generator is nearing completion, and efforts have
been made to ensure a relatively easy incorporation of this more

rigorous approach into the existing program.

As other classes of complex molecules are examined by these
techniques, planning rules will be necessary to provide information
on which compound class may be represented by a given mass spectrum.

This topic has received some attention, but much more needs to be done.

Metastable ion spectra are discussed in detail in the subsequent
section, so that it is sufficient to say that some of this information
can presently be utilized by the program to rationalize the spectra of
mixtures of compounds to determine the structure of each component.
Incorporation of isotopic labeling information is now underway in a
preliminary form, using deuterium labeling to identify the number of
labile hydrogen atoms (e.g., O-H functionalities) associated with each

molecular ion.

Part III is presently being written to provide the capability of
facile input of other chemical or spectroscopic information with which
to test plausible structures. This includes, for example, the above
deuterium labeling information. It also includes 'natural' rules such
as the expectation of an oxygen functionality at C-3 and C-17 in the

case of estrogens.

11
The LISP programming language has been successfully implemented
on the IBM 360/50 at the ACME Computing Facility by R. I. Berns, who
co-authored and maintains Stanford 360-LISP. The full capabilities of
LISP are now available to all users of the ACME Facility. Existing
ACME programs for acquisition and reduction of the high resolution mass
spectral data have been improved over the previous period to ensure
that high quality data are available for use by the Dendral (LISP)

algorithm.
Plans:

Many of the plans are implicit in the above discussion of
objectives and progress. To summarize, it is planned to continue work
on the generalization of the Dendral algorithm, used successfully for
estrogens, and begin the study of other biologically and pharmacologically
important molecules, e.g., progesterone and perhaps cortico-steroids.
The cyclic generator will be implemented on its completion. The
inclusion of information from other techniques, e.g, isotopic labeling
metastable spectra, 13, NMR, will be refined considerably from its
present, preliminary state. It is hoped to implement the Dendral
algorithm on ACME soon, so that the complete system is resident on one

computer, and to begin evaluating the performance of the system in semi-

real-time as opposed to present batch processing of the data.

12
2, Extending the Theory of Mass Spectrometry by a Computer (Meta-DENDRAL)
Objectives:

The broad objective of Meta-Dendral is to study theory formation
in science. Previous work on Heuristic Dendral has given us enough
experience with writing computer programs which work within the theory
of mass spectroscopists who must formulate new sets of mass spectrometry
rules for each new class of molecules under consideration. There are
many gaps in current textbooks because of the relatively small number of
classes of molecules which have been studied systematically by mass
spectroscopists. If the computer system can aid in the process of
filling those gaps, then it will have advanced the goals of this project
by extending the reasoning power of the Heuristic Dendral System. In
addition, it is possible that the program could advance the state of

knowledge of mass spectrometry.
Progress:

Substantial progress has been made toward Meta-Dendral goals. A
detailed research plan has been formulated which makes explicit the
nature of the computer programs that need to be written, and the ways
this work will build on previous Heuristic Dendral work. The research
plan was elaborated in a paper presented to the second International

Joint Conference on Artificial Intelligence (London, September 1970).*

*"A Heuristic Programming Study of Theory Formation in Science" by

B. G. Buchanan, E. A. Feigenbaum, J. Lederberg. Stanford Artificial
Intelligence Project, Memo AIM-145, Computer Science Department Report
No. CS 221, June 1971. Available from the Clearinghouse for Federal
Scientific and Technical Information, Springfield, Virginia 22151.

13
Plans:

We expect to finish soon the basic theory formation program. Most
parts are completed now but have not been interfaced with each other.
Running the program with small, well-structured sets of data will provide
a good test for the soundness of this approach. We are confident that
the program will perform satisfactorily, but we are also ready to program
an attractive alternative strategy in case that is necessary. For the
purpose of testing the theory formation program, the mass spectra
already collected for testing the Heuristic Dendral System will be

sufficient.

In the near future we expect to add more capabilities to the computer
program and test it on large classes of data. For example, generalization
will be programmed initially to proceed with extreme caution. In
subsequent versions of the system, less cautious forms of generalization
will be tried and capabilities for specification will be added. Also,
the first version will be able to formulate only rules about combinations
of molecular cleavages, while it is desirable to add the capability for
formulating rules encompassing more complex mass spectrometric processes
such as group migration. It will also be necessary to expand the
library of mass spectra in order to provide adequate tests of this

system. So,these data will have to be collected during this time period.

The development of data acquisition systems for mass spectrometric
data is not new. The system under development at Stanford will add

to the data acquisition phase an intelligent interpretation system. With

14
routine use of gas chromatography-mass spectrometry systems a vast

avalanche of data threatens to inundate the analytical chemist. For

instance during the analysis of the constituents of urine several

hundred mass spectra would be recorded in the space of one hour. Clearly

the sheer volume of material is beyond the interpretive power of
chemists within any reasonably short period of time. One solution to
this problem is to develop computer programs capable of reasoning their

way through a large compilation of complex data. Heuristic Dendral is

the beginning of one such program.

References:

1. J. Lederberg, "DENDRAL~64, A System for Computer Construction,
Enumeration and Notation of Organic Molecules as Tree Structures
and Cyclic Graphs. Part I, Notational Algorithm for Tree Structures,"

NASA Report CR-27029, December 15, 1964. Copies are available from
the author.

2. J. Lederberg, "Topological Mapping of Organic Molecules," Proc. Nat.
Acad. Sci. U. S., 53, 134 (1965).

3. J. Lederberg, "Topology of Molecules" in the National Research

Council's Committee on Support of Research in the Mathematical Sciences,
The Mathematical Sciences, M.I.T.Press, Boston, Mass., 1969, p. 37.

4. J. Lederberg, '"DENDRAL, A System for Computer Construction, Enumeration
and Notation of Organic Molecules as Tree Structures and Cyclic Graphs.
Part II, Topology of Cyclic Graphs" (NASA Report CR-68898, December 15,
1965); Part III, Complete Chemical Graphs: Embedding Rings in Trees"
(March 13, 1968). Copies are available from the author.

5. A. M. Duffield, A. V. Robertson, C. Djerassi, B. G. Buchanan, G. L.
Sutherland, E. A. Feigenbaum, and J. Lederberg, 'Applications of
Artificial Intelligence for Chemical Inference I. The Number of
Possible Organic Compounds. Acyclic Structures Containing C.H.O, and
N. J. Amer. Chem. Soc. 91, 2973 (1969).

6. G. Schroll, A. M. Duffield, C. Djerassi, B. G. Buchanan, G. L.
Sutherland, E. A. Feigenbaum, and J. Lederberg, "Applications of
Artificial Intelligence for Chemical Inference. III. Aliphatic
Ethers Diagnosed by Their Low-Resolution Mass Spectra and Nuclear
Magnetic Resonance Data. J. Am. Chem. Soc. 91, 7440 (1969).

15
7. A. Buchs, G. Schroll, A. M. Duffield, C. Djerassi, A. B. Delfino,
B. G. Buchanan, G. L. Sutherland, E. A. Feigenbaum, and J. Lederberg,
"Applications of Artificial Intelligence for Chemical Inference. IV.
Saturated Amines Diagnosed by Their Low Resolution Spectra and
Nuclear Magnetic Resonance Spectra." J. Amer. Chem. Soc. 92, 6831 (1970).

8. A. Buchs, A. B. Delfino, A. M. Duffield, C. Djerassi, B. G. Buchanan,
E. A. Feigenbaum, and J. Lederberg, "Applications of Artificial
Intelligence for Chemical Inference, VI. Approach to a General
Method of Interpreting Low Resolution Mass Spectra with a Computer."
Helv. Chim. Acta 53 (6) 1394-1417 (1970).

9. 8B. G. Buchanan, A. M. Duffield and A. V. Robertson. "An Application
of Artificial Intelligence to the Interpretation of Mass Spectra,"
in Mass Spectrometry: Techniques and Applications, Ed. G. W. A.
Milne, John Wiley & Sons, New York, 1971, pp. 121-178.

c. Computer Aided Research Instrumentation

 

Since the last report, the major research objective has been the
continued development of the concept of using dispersed computer elements
to interface research instrumentation to large time-shared computer
systems. The goal of this approach is to utilize dedicated, inexpensive
mini-computer capabilities to satisfy real time, reflexive instrument
interface and data acquisition requirements, and to buffer extracted
information from a raw data stream into sophisticated analysis procedures
available, without real time turn-around commitments, on time-shared
large machines. To act as a test bed for these ideas and equally
importantly to provide a necessary information source for advanced
Dendral project and other chemical structure analysis work, the ability
to acquire and reduce high resolution mass spectrometer data is being
developed. A PDP-11l mini-computer is used to interface existing mass
spectrometers to the ACME IBM 360/50 time-shared computer. Progress

has been made in five major aspects of the problem.

16
Remote Mini-Computer Programming:

A capability has been developed to program and control the mini-
computer using the sophisticated processor, filing, and editing
abilities of the time-shared machine. A basic assembler for the
PDP-11 was written on the ACME machine. Programs can be written,
edited, assembled, and filed on the 360/50 and then retrieved and
transferred to the PDP-1l as required for execution via an interface
and a minimal bootstrap loader resident in the PDP-1l. Required
supervisor and high resolution mass spectrum acquisition software

has been written for the PDP-1l in this way. At this time only the
basic assembler is available making programming for the PDP-11 fairly
time consuming and requiring expert programmer support. Given the
large machine capabilities of the 360/50, higher level language
processors could be written for the PDP-11 thereby making programming

the PDP-1l considerably easier.

Computer Assisted Experiment Set-up:

Programs have been written to assist the mass spectrometer operator
in setting up the instrument and computer interface to perform a mass
spectrometer run. This assistance takes two forms: Prompting of
operator actions to assure completion of necessary set-up procedures
and interaction with data interface adjustments to minimize data loss
and/or distortion, It is desirable to automate these operations as
much as possible to assure consistent set-up and to make data

acquisition procedures adaptive to instrument variations.

1?
High Resolution Mass Spectrum Acquisition and Analysis:

Because of hardware changes within the ACME computer system a new
set of interface hardware between the PDP-1l and the 360/50 had to
be built. This hardware (270-Z) allows the transmission of data
either way between the machines. Using primarily previous interface
hardware developed for the MS-9 mass spectrometer, the new MAT 711
spectrometer was brought on-line to the PDP-11. With this hardware
configuration, programs were written for the PDP-1l and the 360/50
to acquire and reduce high resolution (M/AM 2 20,000) mass spectral
data. This presents a good test of mini-computer capacity in that
in excess of 5 x 10° spectrum samples must be processed in 60 seconds
at present and in approximately 20 seconds in the future. The
initial version of the operational system performs fairly well,
although given the high data rate, the need for adaptation to
instrument fluctuations becomes a necessity because computer data
buffer overflows can result from poor peak detection threshold
settings. The PDP-11 transmits only the significant mass peak
information to the 360/50 thereby reducing the data volume by
approximately 95%. The 360/50 analyzes the remaining peak information
to apply instrument calibration data and to derive elemental compo-
sitions for mass peaks. These are then displayed for the chemist,
filed for future retrieval, or transmitted to the Dendral structure

interpretation programs via a telephone line interface.

18
Instrument Control Interface:

Because of the high volume of time varying diverse-data available
from a gas chromatograph/mass spectrometer (e.g. normal versus
metastable spectra, variable ionization energy, etc.), not all data
can be analyzed or even collected in a run. It is thus necessary

to select that subset of most significance based on real time
interpretive analyses and to command the instrument to operate with
the proper mode and parameters to optimize data collection. Initial
design work has begun for a general interface between the PDP-1l and
the mass spectrometer to allow the control of selected instrument
parameters such as scan initiation, sample probe temperature, and

metastable spectrum acquisition.

Function Allocation Between Computers:

As the sophistication of analysis procedures which must be

performed in real time increases along with data rates and volumes,
the requirements on real time computer support increase. An initial
study was made, in the context of the Dendral high resolution spectrum
acquisition and analysis problem, of the desirability of using an
increasing number of coordinated mini-computers versus increasing
the capacity of a single processor to achieve the required computing
support. The study was constrained to consider the ACME 360/50 as
the larger processor alternative to the multi-mini approach. Based
on initial Dendral requirements and this constraint it appears that
the coordinated multi-mini approach supplies the needed computing

support most economically. Given recent experience with high

19
resolution mass spectrum analysis, this study will be redone
with better estimates of long term Dendral requirements and with no
constraint on large machine selection to finalize the selection of

long term approach.

An additional problem arises in considering the automated closed
loop computer analysis of instrument data. One of the primary
functions of the front-end processor is to extract "significant"
information from the raw stream and to pass it on to subsequent
analyses. It is clear that if either the definition of "significant"
changes due to later analysis or if data which otherwise would be
insignificant shows an instrument or experiment peculiarity which
should be brought to the attention of the downstream analyzer, then
valuable information may be lost unless the front end processor is
"intelligent" enough to adapt to these situations. Furthermore, the
quality of extracted information varies both as instrument
performance varies and as signal-to-noise properties, An objective
quality measure could qualify extracted data in the sense that the
amount of detailed analysis time spent downstream could be adjusted
based on inherent data quality hereby minimizing processor loading.

This problem will be better defined and investigated in future work.

During the next report period work will continue to integrate the
low and high resolution mass spectrometer data acquisition and
analysis procedures. Capabilities to control instrument operation

will be developed further with the goal of closed loop operation in

20
conjunction with the Dendral programs. A broader study of the
optimum computer system configuration and function allocation

design for long term Dendral needs will be undertaken.

21
II. Constituents of Urine

The separation of the components present in urine has been approached

using an initial resolution into acidic and basic fractions. Following
chemical derivatization of the acidic fraction we have been able to

observe over 80 individual peaks on a gas chromatogram. This work is
connected with our interest in developing computer programs for the

computer interpretation of the mass spectra of biologically significant
molecules and is discussed more fully in the preceding section of this report.
Work with the basic fraction present in urine has concentrated on the
isolation of catecholamines because of their importance in neurological
systems. To date we have not been successful in the isolation of
catecholamines from urine and we attribute this failure to their facile
oxidation. We have, however, been successful in developing a derivatization
procedure for the gas chromatographic-mass spectrometric identification

of catecholamines.

An all glass silicone membrane separator has been developed for use in
the direct coupling of a gas chromatograph with a mass spectrometer.
Preliminary studies indicate that this apparatus provides a high
transfer of material from the gas stream of the chromatograph into the
mass spectrometer and no thermal fragmentation of compounds in the

separator has been observed.

22
III. Applications of Gas Chromatography

During the past year we have constructed an automatic instrument for the
analysis of phenylalanine in serum. The basic approach involves the
chemical derivatization of phenylalanine in the modified injector port

of a gas chromatograph which then provides the separation necessary for
the identification of the derivatized phenylalanine. This system has

now been completely automated and requires no technician in attendance
during routine operation. For instance, sample injection, injector
pyrolysis, oven temperature control and carrier gas flow are all
automatically operated according to a pre-scheduled mode of operation.
This technique would allow over one hundred plasma samples to be screened

for phenylalanine content per day using a four-column gas chromatograph.

As a method of determining amino acid constituents of soil it is planned
to obtain these compounds by solvent extraction and then derivatize their
amino group. A known amount of deuterated amino acid will be added (to
act as an internal standard) and the quadrupole mass spectrometer will

be assigned to scan selected significant ions in the mass spectra of the
amino acid derivatives. By also scanning the mass positions of the
deuterated amino acid derivatives quantitation will be introduced into

the analysis.

Gas chromatography was also employed as the analytical tool in the
construction of a technique for the determination of the amount of
cyclohexyl amine in aqueous solutions of sodium cyclamate. Detection
limits in the 191+ gram range were achieved using this method. The
rate of hydrolysis of sodium cyclamate to cyclohexylamine in acid

solution was determined using this procedure.

23
IV. Clorination of DNA Bases

Chlorination studies using aquoeus hypochlorous acid and the bases
present in DNA has continued. Work has centered around a clarification
of the chemistry involved in the reaction of hypochlorite solution

with cytosine. We have been able to identify pure crystalline compounds
from this reaction depending upon the amount of hypochlorous acid used.
These compounds are N-chloroamines and further work is in progress to
elucidate the chemistry behind the reaction of hypochlorite solution
with the other bases found in DNA. It is hoped that this study will
provide some clues to the manner in which chlorine kills bacteria and

whether this involves attack on the DNA of the bacteria by chlorine.

The isolation of 4-N-chlorocytosine and its reaction with one equivalent
of hydrobromic or hydroiodic acids results in a new, simple and direct
route to the chemical preparation of 5-bromo- and 5-iodocytosine,

respectively.

Another possible cause of the lethal power of hypochlorous acid on
bacteria could be due to the cessation of amino acid synthesis in the
cell. In order to examine the chemical implications of this concept we
have examined the reaction of hypochlorous acid with amino acids. To
date phenylalanine, tyrosine,and glutamic acid have been examined. We
observed that the product from these reactions were good yields of the
corresponding nitriles formed by the following reaction sequence,
Previous workers using a much higher pH (8.0 compared to pH 3 in our
studies) report the isolation of aldehydes (B) in good yield. Under the

24
the conditions used we find the aldehydes to be only a minor component
of the reaction. In addition chlorination of the aromatic ring of

tyrosine occurs when it is treated with hypochlorous acid. This had not

previously been reported.

R-CH-CO,H R-CH-CO,H R-CH R-CH
; + HOCL ——> ; —> " Hocly, N
NH, NHC1 NH N-Cl
Y u,0 Vener
2
(A) R-CHO (B) R-C = N (C)

The chemical reaction of di- and larger peptides with hypochlorous acid
does not seem to have been studied previously in any detail. We have
observed the formation of the N,N-dichloro compound (D) from several
dipeptides examined. The chemistry of this new class of compound is
being examined with reference to its possible hydrolysis under mild

conditions to the amino acid of the carboxyl end of the original peptide.

0 0
" " pol
HO, C-CH-N-C-—CH-NH HOCL HO.,C-CH-N-C-CH-N
2 i H t 2 —_—_—_— 2 i | { \
R R R H R Cl
dipeptide (D)

25
V. Mass Spectrometry of Organic Compounds

We became interested in the mass spectra of promazine sulfoxide (a
member of the phenothiazine group of tranquilizers) during an investi-~
gation of its metabolism in sheep. The mass spectra of the sulfoxide of
promazine contained a distinct feature not observed in the mass spectrum
of promazine itself. This difference involved the transfer of two
hydrogen atoms and we have commenced the labeling of the side chain of
promazine sulfoxide with deuterium in order to better understand the

precise mechanism involved in this mass spectrometric fragmentation,

Another study in the field of mass spectrometry has been concerned with
the fragmentation behavior of a series of pimelic acid derivatives.
From extensive deuterium labeling we hope to be able to discover to
which functional group the hydrogen atom preferentially moves in the

McLafferty rearrangement process.
New Publications:

L. G. Brookes, M. A. Holmes, I. S. Forrest, V. A. Bacon,

A. M. Duffield, and M. D. Solomon. "Chlorpromazine Metabolism

in Sheep II. In Vitro metabolism and preparation of 3H-7-Hy droxychlor-
promazine." (Submitted to Agressologie 1971.)

M. D. Solomon, W. E. Pereira and A. M. Duffield. "The Determination
of Cyclohexylamine in Aqueous Solutions of Sodium Cyclamate by

Electron-Capture Gas Chromatography." Analytical Letters 4(5), 301-304
(1971).

W. E. Pereira, B. Halpern, M. D. Solomon and A. M. Duffield.

"Electron-Impact Promoted Fragmentation of Alkyl-N-(1-Phenylethyl) -Carbamates
of Primary, Secondary and Tertiary Alcohols." Organic Mass

Spectrometry 5, 157-169 (1971).

26
B. Halpern, W. E. Pereira, Jr., M. D. Solomon and E. Steed. "A
Rapid and Quantitative Gas Chromatographic Analysis for
Phenylalanine in Serum." Anal. Biochem. 39, 156 (1971).

A. M. Duffield, W. E. Reynolds, D. A. Anderson, R. A. Stillman, Jr.
and C. E. Carroll. "Computer Recognition of Metastable Ions."
Presented at the Nineteenth Annual Conference of Mass Spectrometry
and Allied Topics, Atlanta, Georgia, May 1971.

27
VI. Analysis of Natural Products by Mass Spectrometry

During the past year research has continued on the structural
analysis by mass spectrometry of natural products isolated from plant,
animal and marine sources. A list of publications resulting from

this experimentation follows:

G. Eadon, J. Diekman and C. Djerassi. "Application of Ion
Cyclotron Resonance to the Structure Elucidation of the C H.0.
Ion Formed in the Double McLafferty Rearrangement". J. Am. Chem.
Soc. 92, 6205 (1970).

R. T. Gray, J. Diekman, G. L. Larson, W. K. Musker and C. Djerassi.
"Mass Spectrometry in Structural and Stereochemical Problems -
CLXXXIX: Electron-Impact Induced Fragmentations and Rearrangements
of Alkoxycyclohexanol Trimethylsilyl Ethers and Alkoxycyclohexyl
Trimethylsilanes." Org. Mass Spectrometry 3, 973 (1970).

G. G. Smith and C. Djerassi. "Mass Spectrometry in Structural and
Stereochemical Problems - CXCIX. Contrasting Fragmentations of
Singly- and Doubly-Charged o-, m- and p-Hydroxyalkylphenones and
their Trimethylsilyl Ethers. Org. Mass Spectrometry 5, 505-523 (1971)

R. J. Liedtke, A. M. Duffield and C. Djerassi. ''Mass Spectrometry in
Structural and Stereochemical Problems - CXCIII: On the Combined
Hydroxyl and Water Loss in Nitrophenylhydrazones." Org. Mass

Spectrometry 3, 1089 (1970).

28
VII. Mariner Mars 1971 Orbiter Photography

The picture processing capacity described in Technical Report IRL 1123
has been implemented in support of the Mariner spacecraft presently
orbiting Mars. The system was developed and tested utilizing Mariner
Mars '69 data and simulation data designed to determine its capacity to
isolate image brightness changes of as little as 5%. The "W" cloud
formation thought to have appeared in some of the Mariner Mars '69 Far
Encounter images was quite clearly revealed by image differencing

utilizing the system.

The above mentioned report describes, in detail, the geometric and
photometric normalization and alignment procedure necessary for

automated picture differencing. Various digital filtering and enhancement
techniques have also been developed in support of the picture differencing
work. These techniques are also helpful in aiding with the visual
inspection of the images. Data tapes containing the raw TV images
(PTV-EDA) and the geometrically photometrically corrected TV images

(RDR) are bing sent to us as they become available at the Jet Propulsion
Laboratory. Also received is supplementary data (TQL-SEDR) describing

the spacecraft conditions under which the images were recorded.

Figure 1 is an example of a picture differencing operation. The area
in question is along the edge of the South Polar Cap at about -87°
latitude, 20° longitude (West). The upper left image was acquired on

orbit 28, the upper right three days later on orbit 34. Both images

29
warn

i

ne
os

  
 

rr rr ere

 

  
 

2202 £2028 Oo

Tai |e

~
es
a

 

FIGURE 1.

30
were transformed to the same orthographic projection and then the
upper right image was registered to the other using the correlation

techniques described in the above mentioned report.

The lower left image is the difference image gotten by subtracting the
upper right from the upper left. The lower right image is the negative
of that difference (upper left from upper right). The bright and dark
region on the difference images represent changes in the original
images. However, some apparent differences appearing along the border

are anomalies and are due to poor image registration along the edges.

Special enhancement operations are also being performed. Figure 2
shows two views of Phobos (innermost moon) with each view subject to
two enhancements. The upper image was taken on orbit 43,the lower on

orbit 48.

Work of the type described above is intended to continue through the

primary mission, the extended mission, and the post-mission analysis.

The work is being carried out jointly with the Stanford Artificial
Intelligence Laboratory of the Computer Science Department. The
Stanford Linear Accelerator Center (SLAC) has provided hardware support
in the form of two 70 mm film viewers; one installed at Jet Propulsion

Laboratory, the other at the Artificial Intelligence Laboratory.

31
 

FIGURE 2.

32
VIII. Viking Lander Imagery Investigation

Activity during the past year has fallen into two general areas:
l. Participation in the development of camera and mission
operation specifications; and, recently,
2. Commencement of computer analysis of prototype and projected

lander camera imagery data.

The specification work has been concerned with camera details and with

requirements for image reconstruction hardware and data manipulation

software.

The computer effort has been devoted to problems associated with the
extraction of near-field ranging information from lander camera stereo
image pairs. This latter work, which has used the facilities of the

Stanford Artificial Intelligence Laboratory, is described below.

Figure 3 illustrates a stereo pair of images recorded of a portable
terrain model by a lander camera prototype. The figures displayed in
this and the following illustrations are reproductions of Polaroid
pictures taken of the output of a computer driven video-synthesizer.

The upper and lower images represent left and right views, respectively,
of the model as recorded by a single lander camera prototype located at
two different positions 100 cm apart, approximately 135 cm above the
model and at roughly 100-200 cm horizontal range. The model consists of
an undulating surface of dark sand upon which have been placed four

conspicuous rocks and some smaller pebbles. Since the horizontal range

33
Ssetqperaepeaas

teataeas

tlecerpessebensrdsersbrege

2D oeee ee Pees eee ae

7

ores CT eee cee!

 

Figure 3. A stereo pair of images recorded of a portable terrain

model by a single lander camera prototype located at

two different positions. Upper image left view;

lower image right view.

34
is comparable to the inter-camera displacement, henceforth called the
"baseline", the two views of the scene appear appreciably different.
It has been determined experimentally that it is impossible to visually

fuse such disparate images, a prerequisite to visual depth perception.

Figure 4 indicates the same pair of images upon which have been overlayed
two smooth curves. These curves have been constructed as follows. A
plane, henceforth referred to as the "camera~-centers plane", is defined to
contain the two effective camera-center positions. This plane has been
rotated about the baseline until it passes through the selected point of
interest in the left view at the center of the prepositioned box located
directly beneath the central rock. This plane projects into the left

and right camera views as the curves shown. The fact that the plane
projects as a curve rather than a straight line is a consequence of the
scanning geometry of the facsimile camera and of the associated display
format. Specifically, the camera scans the scene, point by point, in
uniform polar and azimuthal steps. The display format is linear in

polar and azimuthal angles, and hence represents a linear mapping of the
original recording. The lowest point in each of these curves corresponds
to an azimuthal viewing direction perpendicular to the baseline at the
camera location in question. Scene points lying on the camera-centers
plane and common to both views must necessarily lie on the projections

of this plane in each image.

Figure 5 illustrates the same image pair overlayed with image point
location boxes, used in the scene ranging mode. The box in each view
has been visually positioned to a scene point that is common to both

35
i ails Sothtnad este geae tesa yraas stesPrevepecueguasegeeripeccegs

See Tsecetscsebccsshessebessslsessbssastasesdaccsts

Lae oe

eee bee SERS MORES CRESS CELLS CEES Seespessngesespssss
Bi: i

FeRa BREE

ee % . : : sige .
seeadvsiebeecsfoss teaal eevee sserbossabaateboaes PT eT eee

 

Figure 4. Overlay of a projection of the camera-centers plane onto

the stereo images.

36
 

toosfsrecbaeeeQecerdsses ts eettons

 

Figure 5. Stereo images overlayed with image point location boxes

in the scene point ranging mode.

37
views and is to be ranged. The positioning procedure is as follows.

The box is first interactively centered about a point of interest in the
left camera image. A camera-centers plane is then tilted to contain
this pointing direction. The physical point of interest in the right
view now must lie on the projection of the plane in that image. Thus,
only a single degree of freedom remains for the definition of the
corresponding box location in the righthand view. This latter parameter
has been arbitrarily chosen to be the x-coordinate of the point, in
image coordinates. The location of the matching point in the right view
is determined visually. The box is then brought to position by the
adjustment of the x-coordinate. The associated value of the y-coordinate

of the point is then immediately evaluated.

Once the corresponding pointing directions in the two scenes have been
established, we have sufficient information to evaluate the spatial
location of the selected point relative to the cameras. The ranging

information is promptly computed and recorded and/or displayed.

Following development of the above programs we moved on to a
familiarization study of some of the characteristics of the unique
data format, preparatory to investigating automation of ranging and

contour map production.

A first step in this process is indicated in Fig. 6, which contains
additional overlays to those discussed above. The oscillatory curves
appearing in these views represent plots of the scene intensity scanned

along the camera-centers plane and parameterized linearly in the

38
&

~ e 3 : - . ;
PAAR. eee eer PERE EERE eee Cee

 

PAE hE Lh Le ae RO Ae Ao bee Ge Re Lae Re

 

 

Figure 6. Overlays of the scene point intensity along the camera-—
centers plane as a function of the x-coordinate of the

image point.

39
x-coordinate. It will be noted that the lines of constant phase for

the sand ripples running across the lower lefthand portion of the left
view are roughly orthogonal to the baseline. These same waves, when
viewed from the righthand camera position are observed obliquely and
consequently exhibit foreshortened projected wavelength. The relative
distortions associated with the grossly different perspectives of the
two views pose special picture point correlation problems for automation

of ranging.

Figure 7 illustrates a remapping of the intensity plots of Figure 6

into a format that would be more amenable to a one-dimensional correlation
of data from the two scenes. The intensity curves in Figure 7 have been
obtained by transforming the curves in Figure 6 to the appearance they
would have if recorded by a camera located at the mid-point between the
left and right camera positions, under the assumption that the scene is
perfectly flat, horizontal, and at the nominal value of observed relative
elevation. The horizontai scale has been changed to utilize the full
width of the screen. The mapping between the scene and the image now is
nonlinear and the mutual interrelation no longer is as readily
discernable as in the previous illustrations. It is evident, however,
that one-dimensional picture point correlation would be more readily
conducted in this transformed image intensity space than in the initial

space.

40
Ce ae

Pttesteseataves Lak C2 nant Lee eee rer

i

 

Figure 7. Overlays of transformed scene point intensities.

41
We do not plan to proceed further along the above lines. Rather, we

are considering a more general and powerful approach toward the develop-
ment of near-field automated ranging. The proposal is to explore the
portion of the 3~space mutually accessible through the left and right
windows, by correlation of left and right imagery data over a variously
tilted and positioned probing planar patch. Once it has been determined,
by prescribed correlation criteria, that we have succeeded in "landing"
on a surface, we can "crawl" over the surface in any manner desired,
accumulating ranging information as we go. Elevation contour lines
“could for example be generated by instructing the probe to explore at

fixed elevation. An automated contour map could thus be constructed.

We also are commencing various image transformations directed both
toward compensating for inherent projective distortions and toward
facilitating the comparison both of images taken from the same camera
under different recording conditions and of images taken from the two

different lander camera positions.

42
IX. Cell Separation

While this work was initiated by the subject grant most of the support
is now coming from NIH Grant GM 17367. The work has obvious applications

in the medical field as well as for exobiology.

A. High Speed Fluorescent Cell Sorter

This unit is designed to measure the fluorescence of cells in a jet of
liquid, break up the jet into uniform drops and collect the drops in a
series of containers, with all drops containing cells with similar

fluorescent characteristics collected in the same container.

As previously reported, the jet has been modified by incorporation of a
modified Crossland-Taylor sheath flow system which confines the cells to
the central portion of the stream and thus minimizes variation in cell

output as a function of cell position. The system sensitivity has been
increased greatly by substitution of a medium power argon ion laser for

the mercury arc.

Improvements made during the past year have made it possible to increase
efficiency of separation by a factor of about two by reducing the
uncertainty regarding the number of drops which must be deflected to be
certain of deflecting the desired cell. A number of other modifications
have been made resulting in a system of improved speed, reliability,

reproducibility and ease of operation.

43
Much of the recent instrumental effort has been devoted to design and
construction of a multichannel cell separator, able to separate on

the basis of signals from fluorescence in two different spectral ranges.
This unit is essentially completed and is undergoing preliminary testing.
In order to take full advantage of the capabilities of this unit a two

parameter pulse height analyzer has been incorporated into the system.

Also incorporated into the system is a third photoelectric channel which
uses low angle forward scattering to provide signals from all cells in

the stream. This signal is roughly proportional to cell volume, and

will be useful in determining flow rates and relative number of fluorescent
cells. It is also needed to permit optimum use of an anti-coincidence
circuit we have designed. This unit, which substantially increases the
purity of separated cell fractions, deflects groups of drops which

contain both a desired cell and one or more undesired cells into a

separate container, and thus prevents contamination of the desired cell

fraction with undesired cells.

The scatter unit can be adapted for absorption measurements, if desired,
by simple mask changes, allowing us to experiment with a wide assortment

of absorption stains as separation criteria.

In order to optimize system performance we have undertaken studies

aimed at determining the relationships between operating conditions and
variations of signals from cells with equal fluorescence output.

Minimum coefficient of variation so far has been observed with a suspension

of very uniformly fluorescent beads kindly supplied to us by Los Alamos

44
Scientific Laboratory. In our system these beads had a coefficient

of variance (the ratio of standard deviation to the mean) of less than
9%. This appears to be low enough for any separations we presently
envision. Our standard test objects, chicken erythrocytes stained with
quinacrine mustard, have a coefficient of variation of about 152,

indicating remarkable uniformity for a biological specimen.

As a result of the system improvements the value of the instrument for
biological experiments has been greatly increased. Papers by M. Julius,
Masuda and Herzenberg , and by Herzenberg, describe some of the
experiments involving use of immunofluorescent techniques to separate
cells with particular antigen binding characteristics. In one series of
of experiments we were able to separate virtually all cells capable of
giving rise to antibody production against human serum albumin (HSA or
Keyhole Limpet Hemocyanin (KLH). Suspensions of spleen cells from a
mouse primed with HSA were incubated with HSA and then with fluorescein
labelled goat anti-HSA, and the fluorescent cells were separated. These
cells were unable to give rise to anti-HSA production when injected

into an irradiated host alone, but when supplemented with thymus derived
cells also unable to give antibody respmse alone they proved to contain
nearly all of the precursors to antibody producing cells. Similarly

the fraction of spleen cells stained with a fluorescent anti-immunoglobulin

reagent proved to contain the great majority of precursors of cells

M. Julius, T. Masuda and L. A. Herzenberg. "Isolation of Functional
Antibody Forming Cell Precursors Using a Fluorescence Activated Cell
Sorter." (In preparation)

** LL. A, Herzenberg in '"Immunogical Intervention", Jonathan Uhr and
Maurice Landy, eds. Academic Press. 1971.

45
capable of producing antibodies to sheep red cells. These findings are

in accordance with the hypothesis that the precursors to antibody

producing cells carry immunoglobulin and specific antigen binding

markers, and require thymus derived (T) cells to enable them to differentiate
to the actual antibody producing cells. It appears that the separator

may be able to play a major role in investigating the immune system and

its role in such clinical problems as transplant rejection.

These results constitute the first successful enrichment of specific

functionally active antibody forming cell precursors.

A recent most exciting result using the machine has been the direct
demonstration that cooperator cells may have IgM on their surfaces.

Such cells were obtained by indirect fluorescent staining for IgM of
spleen cells from animals injected previously with sheep red blood cells
as antigen. The stained live cells were separated. These cells added
to another portion of the same spleen cell suspension which had been
depleted for cooperator thymus-derived (T) cells fully restored the
cooperator activity. That is, the T cell depleted population did not
give rise to antibody forming cells while this population plus added
IgM stained and separated cells did. This then may be the first

direct demonstration that cooperator T cells have surface IgM.

Another area of current interest is research into the relationships
between uptake of conconavalin A by cell surfaces and cell characteristics.
This protein, derived from jack bean meal, has a special affinity for

certain carbohydrate groups on cell surfaces, particularly for a D

46
glucopyranosyl residues. For similar cells the amount of Con A taken
up is probably proportional to cell surface area. However there are
reports that it shows special affinity for virus transformed cells
and also for cells in the process of mitosis. It may also show
different affinities for different functional types of cells, so that

relative Con A binding might be used to separate such types.

We have been able to incorporate fluorescein in the Con A molecule and
reacted this fluorescent protein with cell populations from mouse bone
marrow, thymus and spleen. When kept cold the cells do not clump. There
are measurable differences between the various populations but we have
not determined yet how much of the differences are the result of size

as compared to qualitative differences in cell surface carbohydrates.

We have, however, measured mouse thymocyte binding at various ages and
found significant differences, with maximum binding apparently occurring
in the juvenile (4 wks) compared to newborn or adult (78 wks). This
indicates that functional differences (or position in the mitotic cycle)
do affect Con A binding. Quantitative binding curves have been obtained
over a 2i8 fold range of Con A concentration. We have treated human
leucocyte suspensions with Con A and found two distinct peaks in the
pulse height distribution. The treated cells were separated. The
deflected fraction (consisting mainly of cells giving higher amplitude
signals) was about 90% granulocytes, the undeflected fraction about

90% lymphocytes. The original suspension was about 75% granulocytes and

25% lymphocytes.

47
A series of similar proteins react with different carbohydrate moieties,

so if the Con A proves able to distinguish between functional types a
completely new parameter for cell separation will be available. In
addition the fact that Con A binding appears to be affected by transforming
virus and by mitosis strongly suggests that it or some of the similar
proteins may be of use in identifying and separating neoplastic (tumor)

cells from the general cell population.

Studies using the binding of Bh? antibodies in a sandwich with
fluoresceinated goat anti-human gamma plobulin to detect Rht blood cells
are currently underway. This reaction appears to work well, producing
tagged cells with relatively uniform signals and good signal to noise
ratio. The technique holds promise in detecting and quantitating leakage
of Bht cells to an Rh™ mother's blood during pregnancy and delivery since
we have been able to detect Rht cells in a frequency of 1074 to 107° the

number of Rh” cells, i.e. 1 Rht cell per 10*-10° Rh~ cells,

Other work has shown that the instrument can detect reticulocytes
stained with acridine orange and thus may be useful for providing
automated reticulocyte counts. It is faster than the presently used
manual methods, and would have the advantage that many more cells could
be processed, reducing the relative statistical variation. The counts
recorded so far on a variety of samples correspond roughly to those
obtained manually on the same samples in the clinical laboratory. Work

to determine the advantages and disadvantages of the method is continuing.

48
Acridine orange can also be used to stain human leucocytes. The ratio
between the red and green fluorescence of these cells after staining
appears to vary depending on cell type. We have conducted some
preliminary experiments in this region with our single channel instruments,
and Melamed and Bio Physics Inc. have demonstrated the differences
between three groups of such cells, tentatively identified as lymphocytes,
monocytes and granulocytes on a two channel analytical instrument
incapable of separation. They also have found that the apparent
lymphocyte fraction seems to vary with the state of health of an
individual. Such variations are difficult to interpret with only an
analytical instrument. Separating cells giving specific intensity

signals would allow identification. We intend to explore this area

intensively when the dual channel separator is in operation.

B. High Speed Volumetric Cell Sorter

Work on this unit has been given a relatively low priority, primarily
because of the exciting results obtained with the fluorescent separator.
A report describing the instrument and its operations has been published

in The Review of Scientific Instruments.*

x
, J. T. Merrill, N. Veizades, H. R. Hulett, P. L. Wolf, and L. A.

Herzenberg. “An Improved Cell-Volume Analyzer.'' Rev. Sci. Instr.
42, 8, pp. 1157-1163 (Aug. 1971).

49
Publications

1. W. A. Bonner, H. R. Hulett, and L. A. Herzenberg. "Highspeed
Sorting of Fluorescence Labeled Cells." Fed. Proc. 30, 699 abs. 1971

2. L. A. Herzenberg. Chairman, Conference Session, "Fluid Transport
Methods", Engineering Foundation Research Conference on Automatic
Cytology, New England College, Henniker, New Hampshire, July 26-30,
1971.

3. L. A. Herzenberg and R. G. Sweet. "Fluorescence Activated Cell
Sorting", presented at Engineering Foundation Research Conference
on Automatic Cytology, New England College, Henniker, New Hampshire,
July 26-30, 1971.

4, L. A. Herzenberg, T. Masuda, and M. Julius. Invited paper on
Symposium on Thymus and Bone Marrow Cells in the Immune Response,
Annual Meeting of the American Society + Hematology, San Francisco,
Dec. 4, 1971.

5. L. A. Herzenberg. Invited participant in symposium on Cell Purification
by use of Surface Antigens and Receptors, Midwinter Conference of
Immunologists, Asilomar, California, Jan. 22, 1972.

6. L. A. Herzenberg, R. G. Sweet, M. Julius, T. Masuda, and R. A. Merker.
Invited paper, "Fluorescent Activated Electronic Cell Sorting in
Immunology", to be presented at Biophysical Society Annual Meeting,
Toronto, Canada, Feb. 19, 1972.

7. W. A. Bonner, H. R. Hulett, R. G. Sweet and L. A. Herzenberg.

"Fluorescence Activated Cell Sorting." Rev. Sci. Inst. (In press,
scheduled for publication March 1972)

50
C. REPORTS AND PAPERS

This section lists papers and reports not referred to in preceding

sections of the Program Resume.

REPORTS

l. Cytotoxicity Assay Automation Final Report. Contract 69-2064
NIH Transplantation Immunology Branch. Technical Report No.
IRL 1129 (1971).

PUBLICATIONS

1. W. E. Reynolds, V. A. Bacon, J. C. Bridges, T. C. Coburn, B. Halpern,
J. Lederberg, E. C. Levinthal, E. Steed and R. B. Tucker.

"A Computer Operated Mass Spectrometer System." Analytical Chem.
42, 1120 (1970).

2. 8B. Halpern, W. E. Pereira, Jr., M. D. Solomon and E. Steed.
"A Rapid and Quantitative Gas Chromatographic Analysis for
Phenylalanine in Serum." Anal. Biochem. 39, 156 (1971).

3. YY. M. Sheikh, A. Buchs, A. B. Delfino, G. Schroll, A. M. Duffield
C. Djerassi, B. G. Buchanan, G. L. Sutherland, E. A. Feigenbaum
and J. Lederberg. "Applications of Artificial Intelligence for
Chemical Inference-V. An Approach to the Computer Generation of
Cyclic Structures. Differentiation Between All the Possible Isometric
Ketones of Composition C¢Hj)0. Org. Mass Spectrometry 4, 493 (1970).

 

4. E. A. Feigenbaum, B. G. Buchanan, and J. Lederberg. "On Generality
and Problem Solving: A Case Study Using the Dendral Program," in
Machine Intelligence 6, B. Meltzer and D. Michie (eds). Edinburgh
University Press, 1971.

5. G. Buchanan, J. Lederberg. "The Heuristic Dendral Program for
Explaining Empirical Data." To be presented at the 1971 Congress of
the International Federation of Information Processing Society
(August, 1971) and published by North Holland Publishing Co. (1971)
in press.

51
6.

J. Cymerman Craig, W. E. Pereira, Jr., B. Halpern, J. W. Westley.
"Optical Rotatory Dispersion and Absolute Configuration -XVII
a-Alkylphenylacetic Acids."" Tetrahedron 27, 1173 (1971).

W. E. Pereira, Jr. and B. Halpern. "The Steric Analysis of
Aliphatic Amines with Two Asymmetric Centers Via Gas Liquid
Chromatography of Diastereoisomeric Amides." Submitted to
Aus ralian Jour. of Chem. (1971).

 

A. M. Duffield and 0. Buchardt. "Thermal Fragmentation of
Quinoline and Isoquinoline N-Oxides in the Ion Source of a Mass
Spectrometer." Submitted to Acta Chemical Scandanavica (1971).

52