a. Ske ley

PROGRAM IN MOLECULAR NEUROBIOLOGY

(Submission to National Institute for
Neurological Diseases and Blindness, 8/30/62, NB04270-01)

Joshua Lederberg

Department of Genetics
Stanford University School of Medicine
PROGRAM IN MOLECULAR NEUROBIOLOGY

1. General theme: the neural engram as a chemical entity.

The underlying hypothesis of our program is that the engram has a material
basis which it is profitable to pursue by the methods of macromolecular
chemistry (=molecular biology).

This hypothesis may be incorrect, and one can hardly pretend that current
knowledge permits a reliable judgement one way or the other ---many neuro-
physiologists have despaired of finding a chemical engram and would focus

their efforts on dynamic signal analysis rather than material circuit net-

works, If correct, the hypothesis may still take many forms. Thus, the

engram may be more a problem of morphological rather than chemical analysis.
Nevertheless, even if the results of such investigations were merely to define
the chemical composition of synaptic terminals whose distribution rather than
whose structurews the engram, this would still be Important for further progress
in neurobiology, .

.
One approach to bhe hypothesis is to inquire whether any macromolecular constit-
uents of the brain have some metabolic properties which parallel the learning
process. For example, are any proteins synthesized during early life which have
the same life-long persistence as memory, in contrast to the rapid turnover of
most of the functional elements of the organism? Because proteins are the
specific effecters for many cellular processes, we focus our attention on these

in preference to other elements, which, however, cannot be ignored. in addition
to the metabolic, cytochemical and developmental studies that are implied by

these precepts, any genetic variations that could be discovered in brain proteins
would furnish tools of immense value for further work, [It should hardly be
necessary to point out that a thorough knowledge of normal structure is absolutely
essential for the understanding and thence prevention and mitigation of disease.

The hypothesis can also be stated in developmental terms: Learning is the
development of the structure and chemistry of the brain under the influence of
external stimuli. Thus, human intelligence is a corollary of the relative
immaturity of the human brain a*birth which allows for the extensive participa-
tion of environmental Information In the actual development of the brain during
its maturation. Another corollary of this conceptual approach is the importance
of studying the metabolism of the brain during the early postnatal period when
the most rapid development of brain structures and the most intense learning"
are fundamentally occurring. The question must also be posed, though it cannot
now be answered, whether animals having more primitive intellectual functions
can be suitable experimental material, in terms of the relative allocation of
biosynthetic activity within their brains to learning processes, as compared

to maintenance and consequent functions. .
2.

If some developmental control of protein structure is part of learning we may
speculate on some specific mechanism that may be consistent with our general
biochemical insight. We might just ask how a bit of information may be stored
at a given switch point:

A. Choice among proteins, i.e., specific activation of one among a series of
RNA messages from the DNA.

B. Choice among protein subunits to form diverse "isozymes" ~ analagous
to the differentiation of adult hemoglobin from fetal hemoglobin, or cardiac
from skeletal lactic dehydrogenase.

C. Episequential modification of the primary amino acid sequence - e.g.,
the selective removal of glutamine amide groups to give a series of variant
proteins differing in ionic charge distribution from the basic fully amidated

pattern.

It may be foolhardy to discuss these alternatives more deeply until we have
accumulated much more information about the proteins of the brain, but these
speculations may give some insight into the long range goals of neurochemical
studies. Choice A may be substantiated through careful analytical work to
separate the range of brain proteins into fractions which cou!d then be localized
by refined ultrastructural techniques. Choice B has had great encouragement
from the realization that many mammalian proteins occur as series of "isozymes",
different combinations of subunits offering another dimension in the sirategy
of developmental variation, Choice C has had new impetus since it was proposed
in the original application: Zubay (Proc. Natl. Acad. Sci, 48: 894, May 1962)
has found that glutamic acid is not bound directly to transfer RNA but only as
glutamine, so that some secondary mechanism must be general lyinvoked to account
for the choice between glutamic acid and glutamine in polypeptide sequences.

We can scarcely overestimate the technical problems that may hinder the resolution
of these questions. Fortunately, Dr. Shooter's extensive experience with the
analysis of hemoglobin is an ideal basis for an effort. Our genera! strategy Is to
enter the problem in a modest way to find the elements that give the most amanable
opportunities to apply current technique, and to use this experience to make whatever
technical Innovations are within our capacity. We also plan to exploit, as far as
possible, the opportunity for development of new analytical Instrumentation which,
fertunately, becomes available to us through the contiguity of the NASA supported
programs in this department.

Having stated our general strategy 1 would then turn to the tactics, the specific
problems chosen for study In the immediate future.

1. The genetic dimorphism of cerebral proteins in the mouse.

2. Identification of long lasting proteins in the brain.

3. Microscan mass spectrometer - new [Instrumentation for fine structure
chemical analysis at the ultramicroscopic level.
}. The genetic dimorphism of cerebral proteins In the mouse

Bailey and Heatd (J. Neurochem. 6: 342, 1951) have fractionated soluble brain
proteins of the mouse by starch gel electrophoresis. They mention, in passing,

the distinction of one component In two mouse strains. This suggests a genetic
dimorphism, in a brain protein, analogous to the variation of hemoglobins in man,
from the study of which we have learned so much of the molecular medicine of the
blood. This study should be repeated, and, if confirmed, extended to other mouse
strains. Then a choice should be made of any technically favorable opportunity

to establish the genetic basis of the dimorphism and to characterize the protein

as a basis for other work on its functional significance. To do this, it will be
necessary to Isolate the dimorphic protein and establish a convenient assay for it,
perphaps an immunological one. This would then be the basis of a breeding program
to establish co-isogenic strains differing only in the gene controlling this
protein , and identical in every other respect. This breeding work could take
several years unless we were fortunate to find the dimorphism in co-isogenic lines
for other tissue antigens already established in this department or procurable from

other centers.

The study would thus start out as a verification of existing claims to which the
following elements would then be added: i) the examination of a wider range of
Inbred strains already avaflable in the Stanford mouse colony, and being studied
immunogenetically in respect to their tissus transplantation antigens by

Dr. L.A. Herzenberg (Associate Professor of Genetics), ii) the comparison of the
brain antigens with these differential tissue antigens, iii) the screening of the
protein bands of the electrophoretograms for enzyme activity (esterase, phospha-
tase, glycosidase, nucleotidase, depolymerase) with sensitive fluorescent reagents
being developed in che exobiology project in this deparmment; other well establish~
ed histochemical techniques for identifying enzyme activity are also in mind, iv)
the use of further methods of protein fractionation, e.g., column and thin-layer
electrophoresis and chromatography on a wide range of substrates, and pretreatment
of the mixed extracts with selective reagents, v) the preliminary fractionation
of the mlxed extracts by sedimentation, thermal denaturation, disulfide reducing
agents, and similar procedures in searching for improved methods of isolating the
differential fractions, vi) isotopic labelling of the total cerebral protein
followed by measures of the turnover rate of the respective fractions, vii)
developmental studies of the fractions, viii) the development of immunological
reagents for the characterization of the fractions, particularly important in
searching for the distribution of these components in tissues other than the
brain, ix) finally, the genetic program to obtain co-iscgenic lines of mice to
allow for testing the further impact of the protein dimorphism on the develcpment
of the animal's behavior.

Patently, it will not be possible to explore all of these lines of investigation
within the scope of the present program. It is our Intention to explore which
aspects of the problem are more nearly within a realistic assessment of our reach.
2, Identification of long lasting proteins in the brain.

On one speculative hypothesis of the mechanism of memory, specific structures are
laid down In learning and persist as material substances for the duration of mem-
ory, i.e., for a large fraction of the life span. As most proteins and other
substances undergo rapid turnover, especially during the growth period, we can
approach a test of the hypothesis by long term labelling experiments, For example,
a group of very young mice (or perhaps mopkeys , if preliminary experiments
encourage it) would be injected with a C labelled protein hydrolysate for an
Interval of several days, to allow the equilibration of the amino acid pool through-
out the body. These animals would then be reared to maturity on normal, non-
Tsotopic diggs for long periods, with intermittent monitoring of serum and
catabolic C. During this interval, the dynamic elements of the brain (and other
organs) should have turned over extensively and the label should have disappeared,
but the label should be retained in any site which fulfills the conditions of

the hypothesis. Whatever residual label can be found in this way by counting of
bulk samples should then also be pursued: for its location by autoradiography,

and its identity by chemical fractionation. Whatever the outcome of this experi-
ment, it should be an important guide to our Jong-range program,

A.N. Davison (biochemical journal 78: 272, 1961) has now reported an experiment
analogous to the one suggested here with findings quite encouraging to the
hypothesis of the chemical engram. While most of the brain protein has a half
life of only 22 days, a small fraction, characterized as lipoprotein and presu-
mably originating from cell membranes, persists, having once been formed, without
appreciable loss throughout the lifetime of the animal. The localization of these
long lasting components, and their chemical characterization, are thus the next
chal lenge in, this analysis. We propose to repeat these experiments using labels
other than C " glycine, verifying the persistence of the C " in the form of
polypeptide, and attempting to localize these fraction by cytofractionation and
microscopic studies (cf. Gray and Whittaker, The isolation of Nerve Endings from
Brain, J. Anat. 96: 79, 1962; Cummins and Hyden, Adenosine triphosphate ievels
and adenosine triphosphatases in neurons, glia and neuronal membranes of the
vestibular nucleus, Biophys. Acta 60: 271, 1962).
De

3. Microscan mass spectrometer - new instrumentation for fine structure
chemical analysis at the ultramicroscopic level.

In many fields of biology we urgently need new methods of chemical analysis of
Structures that can now be seen under the light microscope at a resolution of 0.2
micron or under the electron microscope at .001 micron. For example, in planning
instrumentation for exobiology, we found that many microscopic particles can be
collected from the atmosphere or from soil that we could see but could not
identify. A conclusion whether these were microorganisms or not would be very
much helped if we could analyze such particles, e.g., for protein and nucleic acid,
Further consideration of the instrumentation called for by this problem has led to
the present project, the development of a microscan mass spectrometer, which may
have the most exciting Implications for neurobiology, as well as many other fields
of biology and medicine.

Our point of departure is the electron probe microanalyzer (Castaing, Adv.
Electronics and Electron Physics 13: 317, 1960) which in a few years has become

a widely used analytical instrument, especially In industrial analysis of metals

in alloys, An electron beam is focused on the specimen, and the X-ravs emitted
‘are then collected; their energy spectrum is characteristic of the excited atom and
can be used to identify it. Cosslett (Ann. N.Y. Acad. Sci. 97: 329, 1962) had
elaborated this instrument further by scanning the specimen with the electron beam,
allowing an image of it to be created which shows the distribution of a given
species of atom. For technical reason, the method is mainly effective for heavier
atoms - such elements as fron or calcium can be readily mapped to a resolution of
better than Iu. Useful,ebut much poorer performance is indicated for carbon,
oxygen and nitrogen on account of their characteristics as X-ray em|tters under
electron excitation. It is characteristic of the method that it will give no
information on the molecular, only on the atomic composition of the target.

We propose to construct a similar device for electron beam scanning, but to
analyze the microtarget by the mass spectrum of the gas burned out of the specimen
by the beam. High frequency mass spectrometers are commercially available that
can anlyze the molecular masses In a gas in an interval of 100 microseconds. Thus
a 100 »” specimen could be fully scanned in a period of one second at a resolution
of | micron, Under ideal conditions the mass spectrum can give a wealth of infor-
mation on the chemical composition of the target -- the structure of many new
products and other organic compounds have been fully elucidated from their mass
spectra alone (e.g. at Stanford university, Chemistry Department, Bjerassi et al.,
Proc, Nat. Acad, Sci, 48: 113, 1962 - Mass spectrometry in structural and stereo
chemical problems). Whether so much detail can be usefully obtained In a micro-
scanning instrument remains to be seen. The large yield of information that can be
generated (in principal, a million mass spectra per microgram of target materlal )
will require careful attention to the simplification of the data by computer -
techniques. This problem is also shared by the exobiology program, and it must be
thought of especially forthe later stages of the instrument development.

The most Immediate application of the microscan, requiring the least optimism as
to its capabilities, is for the localization of isotope Jabels in the metabolic
experiments. Although less suitable for some other purposes, tritium, the mass-3
isotope of hydrogen would be an ideal label for microscan localization, having
such a large proportional mass shift compared to hydrogen. With such a label we
could stress the spatial rather than mass spectral resolution, and hope to map
the distribution of tritium to a fraction of a micron. After feeding tritium-
6.

labelled amino acids we could then inquire whether long-lasting molecules
synthesized from these amino acids are localized on the neuron membranes, or

in synaptic knobs, or whether they are also represented in intracellular struc-
tures as well. With further ¢laboration, the instrument could also localize
other Isotopes, especially C " and N~ with the special advantages of a) higher
resolution, higher sensitiyity and faster readout than autoradiography; b) avail-
ability for labels, like N° that are not radioactive; c) the abundance of ad-
ditional chemical information from the complete mass spectrum.

Even a crude microscan mass spectrometer could have a rewarding application in the
direct determination of molecular weights of macromolecules. A solution of protein
or DNA heavily labelled with tritium would be spread and dried on the target. The
scan would give a number of bursts of tritium, each one representing a cluster
source, a labelled macromolecule. Even without counting how many tritons occured
per burst, the number of bursts is the number of macromolecules which, divided

into the total mass deposited, gives the average molecular weight of the particles,
(These calculations are analogous to those in Rotman's work In this deparmment on
Microfluorometry for the detection of single enzyme molecules -- Proc. Nat. Acad.
Sci. 47: 1981, 1962). ,

The chief technical problems we expect to encounter jn this development have to
-do with the efficient gathering of the signal gas from the specimen, its
ionization, and transport in the analyzer beam. Within the last few days, at a
symposium on X-ray microscopy at Stanford University, Castaing has reported some
new work of his own along related lines that gives the strongest encouragement to
the practicability of our approach.