Peptides and Behavior1

Peptides and Behavior1

PEPTIDES AND BEHAVIOR' By Gcorgcr Ungar Baylor College of Medicine, Houston, Texas I. Introduction . 11. Peptides and Innate Behavior A. B. C. D...

1MB Sizes 10 Downloads 65 Views

PEPTIDES AND BEHAVIOR' By Gcorgcr Ungar

Baylor College of Medicine, Houston, Texas

I. Introduction

.

11. Peptides and Innate Behavior

A. B. C. D.

.

Glutathione and the Feeding Behavior of Hydra Invertebrate Neuroseuetory Peptides Peptides in Eating and Drinking Behavior Territoriality and Aggressivity E. Peptides and Sleep F. BirdMigration III. The Role of Peptides in Learned Behavior A. Peptide Hormones and Their Derivatives B. Peptides in Learning and Memory IV. Is There a Peptide Code in the Nervous System? A. InnateMechanisms B. Formation and Labeling of Metaarcuits C. The Role of Peptides in Neural Coding References

.

.

. .

.

. .

.

.

. .

.

. .

.

37 39 39

. 4 0

. . . .

41 41 * 42 43 43 43 . 4 4 50 52 53 55 56

. . . . . .

1. Introduction

The biological revolution of the mid-twentieth century started from the idea that living organisms require for their development and survival not only energy, but also information. I n the last twenty-five years we have learned a great deal about genetic information, but we are just beginning to be interested in the elaborate biological information processing systems which integrate the life of the many cells of each individual into a harmoniously functioning whole and assure the best adaptation of the organism to the changing conditions of the environment. Some of the agents of biological coordination, the hormones, have been known for a long time. Most of these belong to two chemically distinct groups: ( a ) steroids produced by the sex glands and the adrenal cortex and ( b ) peptides or amino acid derivatives, secreted by nerve cells, the ' T h e research work done in the author's laboratory referred to in this review is supported by a grant from the U.S. National Institute for Education.

37

38

GEORGES UNGAR

pituitary and the glands of the digestive tract. Most of the latter substances can directly or indirectly be traced back to the neurosecretory process. All information-carrying molecules can be characterized by two parameters: the message they convey and the “address” to which they have to deliver it. From what we know of cells at present, the messages reaching them can elicit a limited repertoire of responses. The apparent complexity of cellular responses to hormones and other messengers depends, therefore, not so much on the variety of messages they carry as on the specificity of their destinations. This specificity varies considerably from the sharply defined selectivity of the releasing and tropic hormones to the wide range of cells acted upon by, for example, insulin, growth hormone, or thyroxine. Specificity of the address is less critical for short-distance messengers, such as the neurotransmitters. Among the many millions of cells that could respond to them, they reach only those that are across the synaptic gaps into which they are released. This may explain the comparatively small number of transmitter substances, as opposed to the number of peptide hormones. I t should be emphasized here that all of the types of substances known to carry information (steroids, prostaglandins) or potentially capable of doing it (polysaccharides, complex lipids), peptides have by far the highest information content. Transplantation data suggest that there is a sufficient number of possible peptides to mark all the individuals of all the species of higher vertebrates that have ever existed on this globe. It is easy to understand, therefore, that peptides could carry specific messages from cell to cell or, more precisely, messages between specific cells, with the highest possible accuracy of address, even if each single cell were marked with a distinctive receptor. Such luxury of specificity is probably not necessary. Even in the nervous system, which has acquired the highest degree of differentiation of all the tissues, specificity is probably limited to cell groups that form a functionally differentiated circuit. This point will be discussed below in some detail. These considerations are of importance for the role of peptides in behavior. Behavior as “the sum total of the action of the effectors” (Young, 1938) implies integration or coordination, that is, some sort of communication between several neurons and neuronal circuits. The purpose of this review is to examine the role of peptides in the integration of innate and learned behavior and to discuss the mechanisms of this integration. The term peptide is used in this review in its widest meaning. Since there is no uniformly accepted boundary bstween peptides and proteins or between oligo- and polypeptides, the term peptide will be applied to any chemical entity that contains at least one peptide bond. Although I shall deal mostly with sequences of less than one hundred amino acids, in some

PEPTIDES AND BEHAVIOR

39

cases larger molecules will be considered, since the active sites that determine their specificity may be comparatively short sequences. II. Peptider and Innate Behavior

Peptide hormones control many innate behavioral patterns : sexual behavior is dependent on gonadotropin secretion; eating can be induced by insulin ; drinking can be inhibited by vasopressin; parturition can be triggered by oxytocin; etc. I shall, however, not discuss these well known behavioral effects and mention, instead, some examples that are less known but may supply clues to some basic mechanisms. A. GLUTATHIONE AND

THE

FEEDINGBEHAVIOR OF Hydra

Probably the simplest example of behavioral modification induced by a peptide is a phenomenon discovered by Loomis (1955) and studied by Lenhoff (1968) and Rushforth (1965). They showed that the feeding behavior of several species of Hydra and other coelenterates is controlled by the presence of glutathione in their environment. Hydra littoralis, on which most of the experiments were done, is a fixed animal whose tentacles capture any small prey that happens to come near. The contact elicits the response of nematocysts lining the tentacle which puncture and poison the animals. As soon as the body fluids of the wounded prey escape, the tentacles bend toward the mouth, the mouth opens and the prey is swallowed. Whereas the stimulus for the response of the nematocysts is probably mechanical, the tentacle bending and mouth opening are elicited specifically by the presence of glutathione in the fluid escaping from the wound. The specificity of reduced glutathione is well established; none of the three amino acids that make up the tripeptide are active alone; on the contrary, glutamic acid, glutamine, and cysteinylglycine act as inhibitors. The only derivatives capable of eliciting the feeding response are ophthalmic acid (y-glutamyl-cramino-n-butyrylglycine), norophthalmic acid ( 7-glutamylalanylglycine) , and S-methylglutathione. These well documented observations are of considerable interest for several reasons. The primitive nature of the organisms may afford an insight into the very foundations of hormonal mechanisms, even if the action of glutathione does not correspond to the definition of hormonal actions in higher animals. The study of the phenomenon may also afford some insight into the agonist-receptor relationship. Lenhoff (1969) has proposed a hypothesis in which glutathione would act as a “modifier” (in the sense of the model of Koshland et al., 1966) or as an “allosteric activator” (in

40

GEORGES UNGAR

the interpretation of Monod et al., 1963) of the receptor molecule. Another important consideration is that the role of glutathione is played in other coelenterate species by single amino acids, such as proline, valine, or glutamine. The possibility that peptides take over the functions of amino acids as the complexity of the organisms increases and their information input becomes diversified will be discussed later. There is no evidence at present, one way or the other, for the intervention of neural elements in the feeding behavior of coelenterates, because of their diffuse, noncentralized nervous system. The role of amino acids and peptides as the adequate stimuli for feeding behavior have, however, been demonstrated in higher organisms (annelids, arthropods) in which the nervous system has a definite control over behavior (Lindstedt, 1971). B. INVERTEBRATE NEUROSECRETORY PEPTIDES There is a great deal of literature on neurosecretion in invertebrates (see Barrington, 1964; Fingerman, 1970), but I shall mention only two examples of behavior-controlling peptides secreted by nerve cells. The first is the crustacean color-change hormone, identified by Fernlund and Josefsson ( 1972) as pGlu-Leu-Asn-Phe-Ser-Pro-Gly-TrpNH2. I t has been known since the early studies of Parker (1948) that this important behavioral component of some crustaceans, cephalopods, fishes, and amphibians is under neurohumoral control. According to the color of the background and the illumination, these animals can change their color so as to blend in with their environment. I n shrimps, the peptide, secreted by the nerve endings in the sinus gland located in the eyestalks, represents the efferent channel of a reflex whose afferent pathway is represented by neurons going from the eye to the gland. In vertebrates, the chromatophores are under more direct neural control and the color of the skin can be influenced by the known neurotransmitters. This, however, does not exclude the possible role of peptides, and vasopressin, oxytocin, and melanocyte-stimulating hormone (MSH) are known to have an effect on skin color in amphibians and fishes. The exact relationship between these effects and direct nervous control is not known; its elucidation could be of considerable importance for understanding the role of peptides in neural function. The second example is a “polypeptide” secreted by the “bag cells” of the parietovisceral ganglion of Aplysia (Toevs and Brackenbury, 1969). It can be released by electrical stimulation of the ganglion or by K+-induced depolarization. Injection of the isolated substance into a recipient animal induces egg laying (Kupfermann, 1967). Its molecular weight is estimated to be about 6000.

PEPTIDES AND BEHAVIOR

41

Similar neurosecretory phenomena have been seen in ganglia of invertebrates from various groups (Scharrer and Weitzman, 1970). These observations, however, were based on histochemical criteria only, and no attempts have been made at isolating and characterizing the peptides. C. PEPTIDES I N EATING AND DRINKING BEHAVIOR Neural control of feeding behavior has been widely studied in the last decade. The role of neurotransmitters has received particular attention : norepinephrine in eating and acetylcholine in drinking (Grossman, 1960), but the ubiquitousness of the transmitters precludes them from being the specific agents in these behaviors. The transmitter effect varies according to species and more particularly according to the site of injection. Using the same method of intracranial injections, Epstein et a/. (1970) and Fitzsimons (1971) showed that angiotensin injected into the medial preoptic area produced drinking in nonthirsty rats. The amount necessary is just a few picomoles of the substance. It is assumed that hypovolemia induces secretion of renin in the kidney and increases production of angiotensin which would act on the thirst centers and activate drinking behavior. It is important to note that the peptide, unlike the amines, is active also by systemic administration at high doses. This fact underlines the fundamental difference in the specificity of the two types of active substances. Information supplied by the method of intracerebral injections has been supplemented by experiments measuring the release of specific substances in the brain of hungry or thirsty animals. I t has been shown (Yaksh and Myers, 1972) that the hypothalamus of the hungry rhesus monkey releases a substance that induces eating in a satiated animal. Conversely the brain of the satiated monkey releases a substance that inhibits feeding behavior. The chemical nature of the active material is unknown. The authors seem inclined to believe that the stimulating factor is a catecholamine, but some of the experimental data do not agree with this assumption. In any case, the inhibitory substance does not fit with the description of any of the neurotransmitter substances known.

D. TERRITORIALITY A N D AGGRESSIVITY Analysis of the territorial behavior of the Mongolian gerbil by Thiessen (1973) points to the possibility of the role of specific peptides or proteins. It has been established that territorial marking in this species is done principally by the secretion of a scent gland whose functioning is controlled by testosterone or progesterone. Thiessen and his colleagues have found that the effect on territorial marking of hormones implanted in the preoptic area

42

OEORGES UNGAR

of the hypothalamus can be inhibited by actinomycin D and puromycin, injected together with the hormone. This suggests that the hormone induces the synthesis of some protein necessary for the marking behavior. Yahr and Sanders (quoted by Thiessen) found a qualitative difference between proteins of the preoptic area in hormone-stimulated and control animals. There is good evidence that territoriality, as measured by making frequency, is related to social dominance and aggressivity. I t is, therefore, possible that certain portions of the brain produce substances that enter into the mechanism of aggressive behavior. There is a whole area of study here that has hardly been touched yet in spite of the availability of an appropriate methodology. Active substances could be isolated by using test systems, such as territorial marking, measure of aggressivity, or other behavioral characteristics.

E. PEPTIDESAND SLEEP A humoral mechanism of sleep was the first proposed by Legendre and PiCron in 1910 and further supported by the experiments of Schnedorf and Ivy (1939). More recently, the study of this problem has been resumed in the laboratories of Monnier in Basle and Pappenheimer at Harvard University. Using electroencephalographic criteria of sleep in their assay systems, both authors have shown that intraventricular infusion of cerebrospinal fluid from sleep-deprived goats (Pappenheimer et al., 1967) or from rabbits put to sleep by electrical stimulation of the thalamic sleep center (Monnier and Hoesli, 1965) induced sleep in recipient animals. Both laboratories found that the active substance had a low molecular weight (around 500-700 daltons) and was inactivated by heating to looo, extreme variations of pH, and repeated freezing and thawing. Further attempts at characterization indicated that it was a peptide probably containing the following amino acids: Ala, Asp, Glu, Gly, Leu, Ser and Thr, and perhaps T r p (Schoenenberger et al., 1972). The material is highly active: it induces sleep at the dose of about 6 ng/kg when infused by intraventricular route. The experimental data suggest that the peptide may be released by the thalamic sleep center or from the reticular formation (Drucker-Colin et al., 1970), which in turn are sensitive to stimuli reaching them either from the periphery or some other parts of the central nervous system. The relation of this peptide to other chemical factors known to play a role in the mechanism of the sleep cycle (norepinephrine, serotonin) is at present completely unknown.

PEPTIDES A N D BEHAVIOR

43

F. BIRDMIGRATION The annual migrations of many species of birds are controlled by a complex mechanism based on the gradual shortening of daily illumination. This stimulates food intake necessary to accumulate the energy reserves required for the long flight. This stimulation by the information on daily illumination can be effectively replaced by injections of prolactin under well defined conditions (Meier and Dusseau, 1968; Farner et al., 1967). This effect of prolactin is potentiated by administration of corticosteroids (Meier and Martin, 1971). I n Farner’s interpretation, the secretion of prolactin is part of the timemeasuring device by which birds gain information on the gradual shortening of the days. The peptide would be the trigger both for the premigratory increase in feeding and for the initiation of the migration itself. In any case, this is a clear example of a peptide hormone capable of inducing a specific behavior. 111. The Role of Peptider in learned Behavior

The few examples just discussed of peptides playing a role in innate behavior represent only an extension of the well known hormonal function of these substances. There is nothing surprising in the possibility that hormonal peptides can control such elementary behavioral patterns as feeding, sleep, reproduction, adaptation to environment, territorial defense, all intimately linked to survival. Consideration of the possible role of peptides in the acquisition and preservation of new patterns of behavior, however, implies that they may have something to do with the mechanism of learning and memory. Since there is an apparent prejudice against using the traditional methods of endocrinology for the study of higher nervous activity, I shall endeavor to show that there is no fundamental difference between innate and learned neural mechanisms and, thereby, to justify the use of the same approaches to both. AND THEIR DERIVATIVES A. PEPTIDEHORMONES

I t was first shown by Applezweig and Baudry in 1955 that hypophysectomized rats had difficulty in learning a conditioned avoidance task. Murphy and Miller (1955) found that this leaning deficiency could be corrected by administration of ACTH. This action was independent of the adrenocortical stimulating effect of the peptide since it was observed in adrenalectomized animals. ACTH had an effect on learning also in intact, non-

44

OEOROES UNOAR

hypophysectomized animals, in which it delayed significantly the extinction of learned behavior, suggesting a firmer consolidation of memory. These observations were developed by De Wied and his co-workers in several directions (De Wied, 1973). First, they found that the whole molecule of ACTH was not required for the behavioral effect: sequences 1-10 and 4-10 were equally effective, but sequences 11-24 and 25-39 were inactive. They found, furthermore, that a-MSH and P-MSH exerted similar effects. They explored the structural requirements for the behavioral action and found that substitution of D-Phe in position 7 of ACTH reversed the effect and produced a facilitation of extinction while similar substitution of D-amino acids in other positions was not critical. De Wied also found that posterior lobectomy, while not preventing the acquisition of avoidance responses, impaired consolidation and caused rapid extinction of the behavior. This deficiency could be corrected by Pitressin and later with synthetic vasopressin. The effect of this peptide, also observed in intact rats, was significantly more prolonged than that of ACTH and its derivatives. Since the behavioral effect of these peptides seemed independent of their hormonal properties, De Wied attempted to find substances that would have no hormone action, but would exhibit the behavioral effect alone. From anterior pituitary extracts, he isolated a peptide, desglycinamide lysine vasopressin, that produced a better and longer-lasting suppression of extinction than any of the hormones previously tested. Since these peptides have been tested primarily in avoidance situations, it has been proposed that they act by increasing fear, thereby maintaining and prolonging the effect of negative reinforcements. Weiss et al. (1969) proposed a hypothesis based on the antagonistic effect of corticosteroids, which decrease fear, and of ACTH, which increases it. The former would accelerate extinction of behavioral responses based on fear while the latter would delay it. The idea is far from being proved since ACTH can also inhibit extinction of appetitive responses (Leonard, 1969; Gray, 1971; Guth et al., 1971). I t seems probable that the hypothalamic and pituitary peptides have some nonspecific role in the motivational arousal which is an important factor in learning (Stratton and Kastin, 1973).

B. PEPTIDESI N LEARNING AND MEMORY This problem has been reviewed repeatedly (Ungar, 1970a,b, 1972, 1973b), and I shall give here only a brief sketch of its development and a discussion of recent data. Although there have been allusions to chemical mechanisms in the storage of acquired information in the 19th century literature, the problem could

PEPTIDES AND BEHAVIOR

45

not be adequately formulated before sufficient knowledge was accumulated on the chemistry of the nervous system and before the concept of information processing by molecular mechanisms gained acceptance in biology. The first full-fledged hypothesis, proposed by Katz and Halstead ( 1950), was inspired by the early stages of molecular biology and assumed that information was stored in nucleoprotein molecules. The hypothesis had a profound influence on those who after an interval of over 10 years resumed the problem and initiated the first experimental approaches to it. It was assumed that, if there was a chemical process involved in learning and memory, it had to be “macromolecular” (Schmitt, 1962; Gaito, 1972; Ansell and Bradley, 1973). Three main experimental strategies have been used to attack the problem: (1) search for chemical correlates of memory; ( 2 ) effect of metabolic inhibitors on memory, and ( 3 ) biological assay methods. All three methods started with the assumption that RNA molecules played the primary role in the chemical mechanism of learning and memory (HydCn and Egyhazi, 1962; Dingman and Sporn, 1961; McConnell, 1962). In later years, however, the emphasis shifted to proteins ; evidence was produced for increased turnover of brain proteins in learning animals (reviews by Booth, 1970; HydCn, 1973) and memory impairment by inhibitors of protein synthesis (review by Cohen, 1970; Squire and Barondes, 1972). Among the particular types of proteins involved, emphasis was laid on glycoproteins by Bogoch (1968) and on the brain-specific S-100 proteins by HydCn (1973). In 1963, I discussed the reasons why comparatively simple peptides could be adequate for information processing and storage in the nervous system (Ungar, 1963). Subsequently, using the bioassay approach, my laboratory demonstrated the formation of peptides in the brain of rats submitted to a series of morphine injections making them tolerant to the drug (Ungar and Cohen, 1966) and of animals habituated to a sound stimulus (Ungar and Oceguera-Navarro, 1965). Identification of the active substances as peptides was based on two criteria: dialyzability and inactivation by proteolytic enzymes. Subsequently, other laboratories working on the molecular problem of memory reached similar conclusions (Rosenblatt et al., 1966; Chapouthier and Ungerer, 1969; Giurgea et al., 1971; Zippel and Domagk, 1969). However, a doubt remained in the mind of those workers who were able to demonstrate the presence of active material in RNA extracts of brain (Fjerdingstad et al., 1965; Adim and Faiszt, 1967; McConnell et al., 1968; among others). It is noteworthy that almost all the negative results were also obtained with RNA preparations (review by Ungar, 1971; Ungar and Chapouthier, 1971) . The apparent discrepancy was explained when RNA extracts of brain from rats trained for dark avoidance were dialyzed and the active material was found in the dialyzable fraction, while the RNAcontaining nondialyzable fraction was inactive. The dialyzable material was

46

GEORGES UNOAR

identified as a peptide loosely bound to RNA from which it could be separated at low pH (Ungar and Fjerdingstad, 1971). Some of the peptides were purified, and their structure was determined. 1. Scotophobin Between 1968 and 1970, we accumulated material for the purification and structural identification of the peptide that had been detected by bioassay in the brain of dark avoidance-trained rats (Ungar et al., 1968). Out of about 5 kg of rat brain, we isolated 300 pg of a pure peptide containing the following amino acids: Ala, Asp,, Glu,, Glys, Lys, Serz, Tyr. End-group analysis of the whole peptide and its tryptic fragments gave the partial sequence of Ser (Asp, Gln, Gly) Lys-Ser (Ala, Gln, Gly) Tyr-NH,. On the basis of these data and mass spectrometric analysis, a tentative sequence was proposed in which positions 3, 5, and l l remained uncertain. By synthesizing a number of variants and testing their behavioral activity, the following most probable sequence was arrived at : Ser-Asp-Asn-Asn-GlnGln-Gly-Lys-Ser-Ala-Gln-Gln-Gly-Gly-TyrNH2 (Ungar et al., 1972a). This pentadecapeptide was given the name scotophobin. In spite of criticisms (Stewart, 1972; Goldstein, 1973), the dark avoidance-inducing effect of synthetic scotophobin was confirmed in several laboratories (Guttman et al., 1972; Bryant et al., 1972; Malin and Guttman, 1972; Thines et al., 1973). It seemed that once the structure of scotophobin was elucidated it would be possible to develop a chemical method for its detection and quantitative determination. By means of a microdansylation procedure (Neuhoff et al., 1969), the thin-layer chromatographic characteristics of the peptide were established and allowed us to make quantitative determinations of scotophobin in brain (Ungar, 1973a,b). It has never been detected in untrained rat brain, but during dark avoidance training it appears and increases gradually until the sixth day (to a maximum of about 200 ng per gram of brain). Beyond this point it decreases gradually, and at 15 days it is no longer detectable. The same method allowed us to do preliminary experiments on the regional distribution of scotophobin in the brain. About two-thirds of the peptide is present in the cortex, one-fourth in the brain stem and cerebellum, and the remaining small amount is disseminated in subcortical areas. The fate of exogenous scotophobin injected intraperitoneally into mice was investigated by the same method. I t was found in the brain 15 minutes after administration and reached its peak in about 4 hours, after which it decreased gradually and could not be detected at 48 hours. The maximum found in the brain did not exceed 1% of the injected material. All these experiments are still preliminary and will be repeated with a more sensitive method, using 'H-labeled dansyl reagent.

PEPTIDES AND BEHAVIOR

47

It is obvious that there are still some important questions to be answered regarding the specificity of the behavioral effect of scotophobin and the uniqueness of its structure. When scotophobin was synthesized by the solidphase method (Parr and Holzer, 1971), the final product had to be purified. The fraction selected by bioassay had the same dark-avoidance inducing potency and the same chromatographic characteristics as the natural material. However, when we tested a sample of scotophobin made in Dr. Weinstein’s laboratory by the classical method, we found that its activity was significantly lower than that of the substance purified from the solid-phase product. The possibility that natural scotophobin possesses some unusual linkages (involving perhaps the /3-hydroxyl of aspartic acid and the r-amine of lysine) is now being investigated. De Wied (1973) tested a number of synthetic scotophobin analogs and found that, like the ACTH fragments and vasopressin and its derivatives, they delayed the extinction of learned avoidance behavior.

2. Ameletin As mentioned above, in 1965 we published experiments suggesting the formation of a peptide in the brain of rats habituated to a sound stimulus (Ungar and Oceguera-Navarro, 1965). During a period of two and a half years, with the help of S. R. Burzynski and T . Innerarity, I have been attempting the isolation and identification of this substance. We started out with the hope that, with the experience acquired in the isolation of scotophobin, the task would be comparatively easy. In fact, the isolation proved more difficult and different schemes of extraction and purification had to be devised. Although we used a higher amount of starting material, the final yield of pure substance was considerably smaller, so that no quantitative amino acid analysis could be done. By microdansylation we detected the presence of six amino acids: Ala, Glu, Gly, Lys, Ser, Tyr. End group analysis was difficult because, as we learned later, the N-terminal was pyroglutamic acid, which is not detectable by the usual reagents unless it is converted to glutamic acid by alkali treatment. After estimating the size of the molecule by gel filtration and splitting the peptide by chymotrypsin, the following partial structure was proposed : Pglu ( Ala, Gly ) Tyr-Ser (Ala, Gly ) Lys (Ungar and Burzynski, 1973). Subsequently, by means of the enzyme dipeptidyl aminopeptidase (Callahan et al., 1972), the following dipeptides were identified : Glu-Ala; Gly-Tyr; Ser-Lys, suggesting the following tentative sequence : Pglu-Ala-Gly-Tpr-Ser-Lys. This hexapeptide was synthesized by Dr. B. Weinstein (University of Washington, Seattle) and Dr. H. Lackner (University of Giittingen) . The thin-layer chromatographic properties of the synthetic peptides are in general agreement with those of the natural material. Preliminary assays of the

48

GEORGES UNGAR

synthetic material indicate that it does reduce the startle responses of mice to the sound stimulus used in the habituation of the donor animals but the effect, although statistically significant, is less marked. Further work, including synthesis of an N-acetyllysine derivative and that of a cyclic form of the peptide, is being carried out. The natural peptide received the name ameletin (from the Greek ameletos, indifferent). 3. Other Peptides under Study Three other substances, whose behavioral effect was demonstrated by bioassay, are at present being isolated from goldfish brain. Two of them have been extracted from the brain of goldfish trained by color discrimination (Zippel and Domagk, 1969). One group of fish is trained to avoid the blue compartment of a tank by being submitted to electric shocks if they do not respond to the color cue within 20 seconds. Another group of fish is trained in the opposite direction by being shocked in the green compartment. Brain material has been accumulated from these donors for almost two years, and purification has followed the pattern established for scotophobin. The two peptides are somewhat smaller than scotophobin, around 12 amino acid residues. The blue-avoidance inducing peptide is inactivated by trypsin, and the green-avoidance inducing one by chymotrypsin (Ungar et al., 1972b). A third peptide is extracted from the brain of goldfish trained to adapt their swimming behavior to a float attached to them (Shashoua, 1968). Extract from trained fish shortens significantly the time required for adaptation (Heltzel et al., 1972). The active substance is a peptide somewhat larger than those mentioned previously (20-25 residues). It is inactivated by trypsin, and its activity is decreased by chymotrypsin. All three of these substances isolated from goldfish brain have been obtained in an almost pure state, and it is hoped that their structure will soon be elucidated. A number of other peptides have been demonstrated by bioassay to be formed in the brain of animals trained for various tasks: morphine tolerance (Ungar and Cohen, 1966; Ungar and Galvan, 1969), stepdown avoidance (Ungar, 1971), and maze learning (Radcliffe and Shelton, 1973). No attempts have yet been made at their isolation and chemical identification.

4. Emergence of a Methodology The suitability of the behavioral bioassay method for the study of chemical processes in learning and memory and the validity of its results have been the object of controversy for almost ten years. Bioassays are hardly in need of an apology; they laid the foundations of many of the most important concepts in biology; antibodies, hormones, vitamins, and neurotrans-

PEPTIDES AND BEHAVIOR

49

mitters. In spite of their disadvantages, by their sensitivity and specificity they are uniquely suitable for the detection of small amounts of biologically active but chemically unknown substances in complex mixtures. The controversy over the application of bioassays to the chemical problem of learned behavior stems largely from the reluctance to accept the idea that learning can be dependent on information-specific chemical factors (Barondes, 1972). This contrasts with the ready acceptance of the hormonal control of innate behavioral patterns. The literature of the behavioral bioassay has been reviewed repeatedly (Rosenblatt, 1970; Dyal, 1971; Ungar, 1971; Ungar and Chapouthier, 1971 ; see also volumes edited by Ad6m, 1971 ; Fjerdingstad, 1971 ; Zippel, 1973), and the controversial points have recently been critically examined (Ungar, 1973a). I shall only mention here that successful behavioral bioassays have now been published from at least 42 laboratories, not counting the many unpublished reports that came to my attention. In any case, the bioassay is only the first step in the overall strategy that has emerged from the experience of the last decade. This strategy includes the following stages: I

a. Elaboration of an Assay System. It includes the definition, by trial and error, of the optimum conditions for the training of the donors and the testing of the recipients. At this point, a first approximation can be made to the chemical nature of the active substances, which up to the present have invariably proved to be peptides. b. Isolation and Purification. Donors are trained and their brains are collected in numbers estimated adequate for obtaining the amount of material necessary for structural identification. It is probable that these numbers will be steadily decreasing with the progress of methods of separation and advances of analytical peptide chemistry. Each purification step includes the identification of the active fraction by bioassay. c. Structural Identification. This includes amino acid composition, end group analysis, and sequence determination of the whole peptide and its tryptic or chymotryptic fragments. I mentioned above the particular usefulness of dipeptidyl aminopeptidase. Most of the well established methods work well only with millimolar amounts of peptides, but new methods are available for the nanogram and even the picogram levels. I mention among these the microdansylation procedures with or without isotope labeling and mass spectrometry. I n most cases, synthesis is a necessary part of this phase of the research, to confirm the structure and, especially, to make sufficiently large amounts of material available for further research.

50

GEORGES UNGAR

d . Development of Chemical Methods of Determination. Once the structure of the active substance has been elucidated, it is usually possible to devise chemical methods for its detection and quantitative determination. This eventually allows bypassing the bioassay, which seems to be most vulnerable to criticism. The fact that the presence of scotophobin can now be detected by the dansyl method invalidates most of the criticisms of the work leading to its characterization. I n some cases, radioimmunoassays may be necessary instead of chemical techniques. If histochemical methods can be devised, these could give significant information for the next stage of the research. e. Search for the Significance of the Substance and Its Place in an Overall Scheme, As indicated above, this phase has just been started for scotophobin and involves the kinetics of its formation and distribution. Future research, using the isotope-labeled peptide, will further specify the localization of the peptide in the brain, anatomically and perhaps at the subcellular level. Another line of research concerns the chemical specificity of the peptide; the effect of its analogs and derivatives, and the determination of the active site of the molecule. A third question is the behavioral specificity: Does scotophobin, for example, cause only dark avoidance or may it elicit other avoidance behaviors? Can its formation be induced by other types of trainings besides dark avoidance? Responses to all these and other questions will contribute to the solution of the final problem, the overall significance and function of these substances. This solution will come only when a number of peptides isolated from the brain have been chemically identified. The scheme just summarized is not different, in any significant respect, from what has been accomplished for the peptide hormones. There is no particular reason why it should be different if one accepts the idea that learning and memory can be studied by essentially the same methodology as innate behavior and physiological functions. IV. Is There a Peptide Code in the Nervous System?

Before discussing this point, I shall examine briefly the presence of peptides in the brain, the evidence for their being released, and their possible role in neural function. Besides the hypothalamic peptides, present in minute amounts but detectable only by their biological activity, a large variety of brain peptides have been identified chemically (Sano, 1970). These are mostly small peptides: N-acetylaspartylglutamic acid, glutamyl di- and tripeptides (Reichelt, 1970), lysyl peptides (Gatfield and Taller, 1971 ), and tryptophanyl peptides (Edvinsson et al., 1973). There are no estimates of the total peptide content

PEPTLDES A N D BEHAVIOR

51

of brain, except for the hog hypothalamus which may contain several grams per kilogram (Shome and Saffran, 1966). Figures of this type, however, are difficult to evaluate and compare with other tissues because of the loose definition of peptides in relation to proteins. Brain peptides are known to be released from neurons as neurosecretory products. Neurosecretory cells have been divided into type A, secreting peptides, and type B, producing biogenic amines (Knowles, 1967). The distinction between the two types, based on the electron microscopic appearance and dimensions of the neurosecretory granules has been admitted to be arbitrary, and in recent years the so-called “unitary concept” of neurosecretion has prevailed (De Robertis, 1964). This concept has been formulated by Zetler (1970) as follows: “All neurons, in addition to generating and spreading electrical phenomena, have secretory functions by which active substances are synthesized and released.’’ There is now evidence that the same cell can release both peptides and amines, thus abolishing the distinction between “peptidergic” and “aminergic” neurons (Bargmann at al., 1967; Owman et al., 1973). Release of neurosecretory substances from nerve endings has been demonstrated in vitro (Musick and Hubbard, 1972 ; Edwardson et al., 1972). Release of peptides by neurons being demonstrated, it remains to be seen what the peptide does once it is released. The well known cases are the hormones released into the general circulation (posterior pituitary hormones) or the hypothalamopituitary portal system (releasing hormones) . It should be noted that evidence is accumulating to suggest that all the peptide-secreting cells derive ultimately from the neural crest by migration into the skin and the digestive tract and its glands (Weichert, 1970; Le Douarin and Teillet, 1973). Much less is known about peptides released at actual synaptic or “synaptoid” junctions. Some peptides, such as substance P, have been mentioned as possible neurotransmitters (Zetler, 1970), but the evidence is at best fragmentary. If neurons can release transmitter amines and peptides at the same time, one has to assume that the latter must have some function other than synaptic transmission. “Whether or not polypeptides act on neurons as classical synaptic transmitters, they clearly can exert powerful effects on neuronal activity” (Bloom et al., 1972). This type of effect has been designated by the general term of “modulation.” A modulator has been defined by Florey (1967) as “any compound of cellular and non-synaptic origin that affects the excitability of nerve cells and represents the normal link in the regulatory mechanisms that govern the performance of the nervous system.” Several terms of this definition require explanation and correction : “cellular” origin may be interpreted as excluding modulation by hormones released from distant cells, and the exclusion of substances of synaptic origin would reject

52

GEORGES UNGAR

the possibility of modulators being released from presynaptic endings together with transmitters. Modulators may act on the presynaptic neuron by influencing the amount of available transmitters, the amount of transmitters released and the rate and time course of release. They may also act on the postsynaptic neuron by controlling the amount of available receptors and perhaps their affinity for the transmitters. We should probably include in modulation the slower effects that may control the type of protein being synthesized by the cell (Bloom et al., 1970). Any of these effects would clearly control neuronal connectivity by deciding which synaptic inputs are accepted by the neuron and which are ignored. A. INNATE MECHANISMS During development of the embryo, the organism acquires the neural and humoral control mechanisms necessary for survival. Regulation of visceral functions, reflex responses and, in some species, elaborate instinctive behavioral patterns elicited by certain stimuli depend on prewired neural circuits. These are obviously organized according to a genetic blueprint but the mechanism of the innate organization of the nervous system is imperfectly known. Over the past twenty years the concept of “chemospecificity of pathways” has emerged to explain the complex mechanism by which the neural pathways develop (Sperry, 1958; Jacobson, 1969; Gaze, 1970). It has been supported by experimental data suggesting that neurons destined to be part of the same innate pathway find each other and make synaptic connections by means of a molecular recognition system. The nature of the labels carried by the neurons has never been investigated, but circumstantial evidence suggests that they are proteins. Only proteins have the potential information content to encode the number of specific pathways formed in the mammalian brain (estimated to be of the order of lo7 in the human brain; Ungar, 1968). I t is probable that the neuronal recognition system is but a special case of the general mechanism of histogenesis by which cells of the same type aggregate to form tissues and organs. Because of the extreme complexity of the nervous system, the labeling process reaches in it a much higher degree of differentiation than in other tissues. It seems fairly well established that homotypic cell aggregation is controlled by specific glycoproteins (Moscona and Moscona, 1963). Recent observations of Garber and Moscona (1972) suggest that regional differentiation of the brain is controlled by similar substances. I t may be significant that aggregation of embryonic brain cells is inhibited by all general anesthetic drugs (Ungar and Keats, 1973).

PEPTIDES AND BEHAVIOR

53

These recognition molecules do not fit the conventional definition of peptides, but their specificity depends in all probability on a limited peptide sequence and perhaps to some extent on carbohydrate prosthetic groups. Each innate pathway is probably marked by a peptide sequence which may be regarded as its code designation. These sequences need not be very long. There is a sufficient number of sequences in dipeptides to hexapeptides (6.7 X 10’) to label all the innate pathways in the human brain. We cannot be certain that this labeling system persists beyond embryonic life and does not disappear once the pathways are organized. It has been suggested, however, that “identity reactions could strengthen frequently used synapses if inducers were transmitted across synapses at simultaneous firings” (Roberts and Flexner, 1966). Similar opinions were expressed by Sperry ( 1963) and Jacobson ( 1969). B. FORMATION AND LABELING OF METACIRCUITS I t is widely assumed that learning, the acquisition of new behavior, involves the formation of specific neural circuits named cell assemblies by Hebb (1949) and metacircuits by Barbizet ( 1968). I t is probable that these metacircuits are formed initially by means of new connections between the innate pathways (which can be called protocircuits), and at higher levels by further connections between existing metacircuits. There are, therefore, metacircuits of first, second, third, nth order. If the labeling system of the protocircuits, mentioned in the preceding section, serves for the marking of the metacircuits, one can assume the existence of a whole hierarchy of peptide chains from the comparatively simple sequences that label the protocircuits to the increasingly complex molecules that code for the metacircuits of higher and higher order. The former are undoubtedly inscribed in the genome, since the structure of the brain is genetically determined, but the latter may be formed by a nongenetic process to deal with acquired information. RNA-directed peptide synthesis is now universally accepted, but, before the era of molecular biology, several other mechanisms were studied. They had been all but forgotten until recently, when the problem was revived by Lipmann (1971) and by Meister (1973). Lipmann described the biosynthesis of the bacterial peptides gramicidin or tyrocidine by complex enzyme systems. The amino acid is activated by ATP to form an aminoacyl adenylate-enzyme complex and is linked by the transpeptidation of a thioesterified carboxyl to the amino group of the next aminoacyl thioester. Meister showed that enzymic synthesis of peptides can take place also in eukaryotic organisms. The prototype of this process is the synthesis of glutathione, but more complex peptides could be formed by the same transpeptidation

- - -

54

GEORGES UNGAR

mechanism. T h e brain is rich in enzymes involved in this synthesis (Meister, 1973), and preliminary experiments in my laboratory suggest that the activity of one of them increases during behavioral training (Ungar, 1973b). The coding process could include the following steps (some of them illustrated in Fig. l ) : 1. Activation of the pathways involved in the behavioral training causes increased synthesis of their genetic labels. 2. Simultaneous firing of two or more protocircuits induces transsynaptic transfer of their labels ( U and C). This step, called “transprinting” in the hypothesis of Szilard ( 1964), may include the participation of interneuronal feedback systems. 3. At the same time, the peptide-synthesizing enzymes undergo activation in some of the cells involved and combine the genetic labels to form a new molecular species ( U C ). 4. The newly formed peptide is incorporated into the activated synaptic membranes and consolidates the new connection. This same process taking place a t a number of synapses creates the metacircuit corresponding to the new behavior. Some of the steps enumerated are hypothetical, but none are incompatible with the experimental data. Step 1 is supported by the observations of

ZI-.

~7 “F

FIG.1. Example of the possible formation of labeled metacircuits. An unconditioned stimulus ( U ), traveling through innately labeled pathways, elicits a preprogrammed response ( R ) . Presentation of a conditioned stimulus ( C ) activates a circuit that is not innately connected with circuit U , but if both C and U are applied in a definite temporal sequence some of the neurons involved in their respective circuits are simultaneously activated, together with a set of interneurons (which also play the role of a feedback system). This simultaneous or almost simultaneous firing, by activating the synaptic membranes and increasing their permeability, could allow penetration of the labels C and U into the interneurons. If these contain the enzyme systems necessary to combine the two peptide fragments C and U , a new molecular species CU could be synthesized and become the label or marker for the newly formed synaptic connection. Since the same process is repeated a t many other junctions, CU becomes the “codeword” for the whole metacircuit through which the newly acquired conditioned response is elicited.

PEPTIDES AND BEHAVIOR

55

increase in RNA and protein synthesis in the brain of learning animals (Booth, 1970; HydCn, 1973) and impairment of learning by inhibitors of these syntheses (Squire and Barondes, 1972). Step 2 is supported by steadily accumulating data on the possibility of transsynaptic transport (Korr et al., 1967; Globus et al., 1968; Alvarez, 1970; Grafstein, 1971). Evidence for nongenetic peptide synthesis in step 3 was mentioned above. Analysis of our data on the life cycle of scotophobin in the brain of dark avoidance-trained animals (Ungar, 1973a,b) indicates that an excess of the peptide is synthesized during the first days of training, but subsequently it disappears from the brain in spite of the persistence of the learned behavior. It is possible that retention of information depends on a small amount of the coding material bound to some structure, presumably to synaptic membranes.

C . THEROLEOF PEPTIDESIN NEURALCODING Information processing is probably a great deal more complicated in the nervous system than in the genome. There are at least two types of neural coding systems :

1 . Coding Schemes Stimuli reaching sensory receptors are transduced into bioelectric wave patterns analogous to the “coding schemes” into which data have to be translated so that the computer can handle them. A similar transduction takes place also at each neuron to convert the chemical stimulus exerted by the synaptic transmitter into the appropriate axonal output. Although this coding scheme has a basic principle of frequency modulation, there may be as many varieties of it as there are neural pathways. It is generally agreed that the neural coding scheme cannot be a long-term information store.

2. Command Codes and Program Codes They represent the repertoire of elementary operations that the brain is capable of, the order in which the operations take place, and hold the key to the manner in which these operations can be changed and their repertoire enlarged. The brain is born with a genetically determined program that allows it to respond to stimuli in a stereotyped manner. Learning consists in changing this program, i.e., reprogramming the brain. If the innate program is chemical, as suggested by data mentioned in the preceding section, it is probable that the reprogramming also takes place by a chemical process. Nobody denies that the nervous system possesses “labeled lines” (Perkel and Bullock, 1968) ; the opinions differ only over the identity of the labeling substances. Labeling by means of the transmitter

56

OEORGES UNOAR

released at the axonal ending into cholinergic, adrenergic, dopaminergic, serotonergic, etc. neurons is implicit in the neurohumoral doctrine which also postulates a complementary labeling by means of receptors present at the postsynaptic surfaces of cholinoceptive, adrenoceptive, etc., neurons. This is genuine coding since only a neuron possessing cholinergic receptors can be excited by a cholinergic axon. We do not know at present whether this type of program code can be modified by learning, but, on the face of circumstantial evidence, this seems unlikely. One would be tempted to assume that the transmitter-linked program code controls pathways that remain immutable and are not involved in learning. Many of the generally recognized neurotransmitters are amines derived from amino acids by decarboxylation with or without additional chemical modification. I n recent years, evidence has been presented suggesting that unchanged amino acids can function as transmitters, and the role of glycine and glutamic acid has been demonstrated (Curtis and Crawford, 1969). It is not clear whether amino acid transmitters operate on their own or in association with amines. One can speculate that they represent the foundations of another chemical coding system starting with amino acids and, as the number of labeled lines increases, forming amino acid chains of increasing complexity. It has been suggested that “the response to single amino acids might only be a reflection of the endogenous affinity toward a larger molecule” (Bloom et al., 1972). A peptide code based on the twenty-letter amino acid alphabet has unlimited information content, sufficient to program the protocircuits and to operate their reprogramming into the metacircuits necessary to preserve all the information acquired in a lifetime. I t is hoped that the question mark in the title of this section is sufficient to warn the reader of the speculative nature of many of its propositions. A simpler and more elegant solution of the problem may be hidden to us by our present ignorance of many relevant facts. The fact that the molecular hypothesis is at present controversial should not discourage its further study. Controversy is a normal part of the dialectical process by which science advances, and there has hardly been any new idea that did not have to contend with it. A good example, close to our topic, is the neurohumoral doctrine, for which it took over a quarter of a century to be fully recognized as one of the basic principles governing the neurosciences. Controversy may discourage the faint-hearted, but it is a challenge to those who are determined to find out the truth.

.

REFERENCES

A d h , G., ed. ( 1971 ) “Biology of Memory.” Akadhiai Kiad6, Budapest. i d h , G., and Faiszt, J. (1967). Nature (London) 216, 198.

PEPTIDES AND BEHAVIOR

57

Alvarez, J. (1970). Acta Physiol. Lat. Amer. 20, 271. Ansell, G. B., and Bradley, P. B., eds. (1973). “Macromolecules and Behaviour.” Univ. Park Press, Baltimore, Maryland. Applezweig, M. H., and Baudry, F. D. (1955). Psychol. Rev. 1, 417. Barbizet, J. (1968). Neurosci. Res. 1, 315. Bargrnann, W., Lindner, E., and Andres, K. H. (1967). Z. Zellforsch. Mikrosk. Anat. 77, 282. Barondes, S. H. (1972). Science 176, 631. Barrington, E. J. W. (1964). In “The Hormones” (G. Pincus and K. V. Thimann, eds.), Vol. 4, p. 299. Academic Press, New York. Bloom, F. E., Iversen, L. L., and Schmitt, F. 0. (1970). Neurosci. Res. Program, Bull. 8, 421. Bloom, F. E., Hoffer, B. J., and Nicoll, R. A. (1972). Neurosci. Res. Program, Bull. 10, 213. Bogoch, S . (1968). “The Biochemistry of Memory.” Oxford Univ. Press, London and New York. Booth, D. A. (1970). In “Molecular Mechanisms in Memory and Learning” (G. Ungar, ed. ), p. 1. Plenum, New York. Bryant, R. C., Santos, N. N., and Byrne, W. L. (1972). Science 177, 635. Callahan, P. X., McDonald, J. K., and Ellis, S. (1972). Fed. Proc., Fed. Amer. Soc. Exfi. Biol. 31, 1105. Chapouthier, G., and Ungerer, A. (1969), Rev. Comfiortement Anirn. 3, 64. Cohen, H. D. (1970). In “Molecular Mechanisms in Memory and Learning” (G. Ungar, ed.), p. 59. Plenum, New York. Curtis, D. R., and Crawford, J. M. (1969). Annu. Rev. Pharrnacol. 9, 209. De Robertis, E. (1964). “Histophysiology of Synapses and Neurosecretion.” Pergamon, Oxford. De Wied, D. (1973). In “Memory and the Transfer of Information” (H. P. Zippel, ed.), p. 373. Plenum, New York. Dingman, W., and Sporn, M. B. (1961). /. Psychiat. Res. 1, 1. Drucker-Colin, R. R., Rojas-Ramirez, J. H., Vera-Trueba, J., Monroy-Ayala, G., and Hernandez-Peon, R. (1970). Brain Res. 23, 269. Dyal, J. A. (1971). In “Chemical Transfer of Learned Information” (E. J. Fjerdingstad, ed.), p. 219. Amer. Elsevier, New York. Edvinsson, L., Hikanson, R., Ronnberg, A. L., and Sundler, F. (1973). /. Neurochcm. 20, 897. Edwardson, J. A., Bennett, G. W.,and Bradford, H. F. (1972). Nature (London) 240, 554. Epstein, A. N., Fitzsimons, J. T., and Rolls, B. J. (1970). 1. Physiol. (London) 210, 457. Farner, D. S., Wilson, F. E., and Oksche, A. (1967). Neuroendocrinology 2, 529. Fernlund, P., and Josefsson, L. (1972). Science 177, 173. Fingerman, M. (1970). Annu. Rev. Physiol. 32, 345. Fitzsimons, J. T. (1971). 1.Physiol. (London) 214, 295. Fjerdingstad, E. J., ed. ( 1971). “Chemical Transfer of Learned Information.” Amer. Elsevier, New York. Fjerdingstad, E. J., Nissen, T., and Rbigaard-Petersen, H. H. (1965). Scand. /. Psychol. 6, 1. Florey, E. (1967). Fed. Proc. Fed. Amer. Soc. Exfi. Biol. 26, 1164. Gaito, J., ed. (1972). “Macromolecules and Behavior,” 2nd ed. Appleton, New York.

58

GEORGES U NGAR

Garber, B. B., and Moscona, A. A. (1972). Develop. Biol. 27, 217 and 235. Gatfield, P. D., and Taller, E. ( 1971). Brain Res. 29, 170. Gaze, R. M., ed. (1970). “Formation of Nerve Connections.” Academic Press, New York. Giurgea, C., Daliers, J., and Rigaux, M. L. (1971). Arch. Znt. Pharmacodyn. Ther. 191, 292. Globus, A,, Lux, H. D., and Schubert, P. (1968). Brain Res. 11, 440. Goldstein, A. ( 1973). Nature (London) 242, 60. Grafstein, B. (1971). Science 172, 177. Gray, J. A. (1971). Nature (London) 229, 52. Grossman, S . P. (1960). Science 132, 301. Guth, S., Levine, S.,and Seward, J. P. (1971). Physiol. & Behav. 7, 195. Guttman, H. N., Matwyshyn, G., and Warriner, G. H. (1972). Nature ( L o n d o n ) , New Biol. 235, 26. Hebb, D. 0. (1949). “The Organization of Behavior.” Wiley, New York. Heltzel, J. A., King, R. A., and Ungar, G. (1972). SOC.Neurosci. Meet. p. 75. Hydtn, H. (1973). I n “Macromolecules and Behaviour” (G. B. Ansell and P. B. Bradley, eds.), p. 3. Univ. Park Press, Baltimore, Maryland. HydCn, H.,and Egyhazi, E. (1962). Proc. Nat. Acad. Sci. US.48, 1366. Jacobson, M. ( 1969). Science 163, 543. Katz, J. J., and Halstead, W. C. (1950). Comp. Psychol. Monogr. 20, 1. Knowles, F. ( 1967). I n “Neurosecretion” (F. Stutinsky, ed.), p. 8. Springer-Verlag, Berlin and New York. Korr, I. M., Wilkinson, P. N., and Chornock, F. W. (1967). Science 155, 342. Koshland, D. E., NemCthy, G., and Filmer, D. (1966). Biochemistry 5, 365. Kupfermann, I. (1967). Nature (London) 216, 814. Le Douarin, N. M., and Teillet, M. A. (1973). J . Embryol. Exp. Morphol. 30, 31. Legendre, R., and PiCron, H. (1910). C. R. SOC.Biol. 68, 1108. Lenhoff, H. M. (1968). Science 161,434. Lenhoff, H. M. (1969). Comp. Biochem. Physiol. 28,571. Leonard, B. E. (1969). Znt. J . Neuropharrnacol. 8, 427. Lindstedt, K. J. ( 197 1). Comp. Biochem. Physiol. A 39, 553. Lipmann, F. (1971). I n “Chemical Evolution and the Origin of Life” (R. Buvet and C. Ponnamperuma, eds.), p. 381. North-Holland Publ., Amsterdam. Loomis, W. F. (1955). Ann. N.Y. Acad. Sci. 62, 209. McConnell, J. V. (1962). J . Neuropsychiat. 3, Suppl. 1, 42. McConnell, J. V., Shigehisha, T., and Salive, H. (1968). J . Biol. Psychol. 10, 32. Malin, D. H., and Guttman, H. N. (1972). Science 178, 1219. Meier, A. H., and Dusseau, J. W. (1968). Physiol. Zool. 41, 95. Meier, A. H., and Martin, D. D. (1971). Cen. Comp. Endocrinol 17, 311. Meister, A. (1973). Science 180, 33. Monnier, M., and Hoesli, L. (1965). Pfiiiegers Arch. Cesamte Physiol. Menschen Tiere 282, 60. Monod, J., Changeux, J. P., and Jacob, F. (1963). 1.Mol. Biol. 6, 306. Moscona, M. H., and Moscona, A. A. (1963). Science 142, 1070. Murphy, J. V., and Miller, R. E. (1955). J. Comp. Physiol. Psychol. 48, 47. Musick, J., and Hubbard, J. I. (1972). Nature (London) 237, 279. Neuhoff, V., von der Haar, F., Schlimme, E., and Weise, M. (1969). Hopbe-Seyler’s 2.Physiol. Chcm. 350, 121.

PEPTIDES AND BEHAVIOR

59

Owman, C., HSkanson, R., and Sundler, F. (1973). Fed. Proc., Fed. Amcr. SOC. Exp. Biol. 32, 1785. Pappenheimer, J. R., Miller, T. B., and Goodrich, C. A. (1967). Proc. Nat. Acad. Sci. US.58, 513. Parker, G . H. (1948). “Animal Colour Changes and their Neurohumours.” Cambridge Univ. Press, London and New York. Parr, W., and Holzer, G. (1971). Hoppc-Scylcr’s Z . Physiol. Chcm. 352, 1043. Perkel, D. H., and Bullock, T. H. (1968). Ncurosci. Rcs. Program, Bull. 6, 221. Radcliffe, G. J., Jr., and Shelton, J. W. (1973). Fed. Proc., Fed. Amcr. SOC. Exp. Biol. 32, 818 (Abstr.). Reichelt, K. L. (1970). /. Ncurochcm. 17, 19. Roberts, R. B., and Flexner, L. B. ( 1966). Amcr. Sci. 54, 174. Rosenblatt, F. ( 1970). In “Molecular Mechanisms in Memory and Learning” (G. Ungar, ed.) , p. 103. Plenum, New York. Rosenblatt, F., Farrow, J. T., and Herblin, W. F. (1966). Nature (London) 209, 46. Rushforth, N. B. (1965). Amcr. Zool. 5, 505. Sano, I. (1970). Int. Rev. Ncurobiol. 12, 235. Scharrer, B., and Weitzman, M. (1970). In “Aspects of Neuroendocrinology” (W. Bargmann and B. Scharrer, eds.), p. 1. Springer-Verlag, Berlin and New York. Schmitt, F. 0. (1962). “Macromolecular Specificity and Biological Memory.” MIT Press, Cambridge, Massachusetts. Schnedorf, J. G., and Ivy, A. C. (1939). Amcr. /. Physiol. 125, 491. Schoenenberger, G. A., Cueni, L. B., Monnier, M., and Hatt, A. M. (1972). Pfliigcrs Arch. 338, 1. Shashoua, V . E. (1968). Nature (London) 217, 238. Shome, B., and Sathan, M. (1966). I . Ncurochcm. 13, 433. Sperry, R. W. (1958). I n “Biological and Biochemical Bases of Behavior” (H. H. Harlow and C. N. Wolsey, eds.), p. 401. Univ. of Wisconsin Press, Madison. Sperry, R. W. (1963). Proc. Nat. Acad. Sci. U.S. 50, 703. Squire, L. R., and Barondes, S. H. (1972). In “Macromolecules and Behavior” (J. Gaito, ed.), 2nd ed., p. 61. Appleton, New York. Stewart, W. W. (1972). Nature (London) 238, 202. Stratton, L. O., and Kastin, A. J. (1973). Physiol. d Behav. 10, 689. Szilard, L. (1964). Proc. Not. Acad. Sci. U.S.51, 1092. Thiessen, D. D. (1973). Amcr. Sci. 61, 346. Thines, G., Domagk, G. F., and Schonne, E. (1973). In “Memory and Transfer of Information” (H. P. Zippel, ed. ), p. 363. Plenum, New York. Toevs, L. A., and Brackenbury, R. W. (1969). Comp. Biochcm. Physiol. 29, 207. Ungar, G. (1963). Ann. N . Y . Acad. Sci. 104, 272. Ungar, G. (1968). Pcrspcct. Biol. M c d . 11, 217. Ungar, G. ( 1970a). In “Molecular Mechanisms in Memory and Learning” (G. Ungar, ed.), p. 149. Plenum, New York. Ungar, G. (1970b). Int. Rev. Ncurobiol. 13, 223. Ungar, G. (1971). In “Methods in Pharmacology” (A. Schwartz, ed.), Vol. 1, p. 479. Appleton, New York. Ungar, G. (1972). In “The Structure and Function of Nervous Tissue” (G. H. Bourne, ed.) , Vol. 4, p. 2 15. Academic Press, New York. Ungar, G. (1973a). Naturwisscnschaftcn 60, 307.

60

GEORGES UNGAR

Ungar, G. (1973b). In “Memory and Transfer of Information” (H. P. Zippel, ed. ), p. 3 17. Plenum, New York. Ungar, G., and Bunynski, S. R. (1973). Fed. Proc., Fed. Amer. Soc. Exp. Biol. 32, 367 (abstr). Ungar, G., and Chapouthier, G. (1971). Annee Psychol. 71, 153. Ungar, G., and Cohen, M. (1966). Int. J. Ncuropharmacol. 5, 183. Ungar, G., and Fjerdingstad, E. J. (1971). In “Biology of Memory” (G. AdAm, ed.), p. 137. Akadhiai Kiad6, Budapest. Ungar, G., and Galvan, L. (1969). Proc. SOC.Exp. Biol. Mcd. 130, 287. Ungar, G., and Keats, A. S. (1973). Anesthesiology 39, 362. Ungar, G., and Oceguera-Navarro, C. (1965). Nature (London) 207, 301. Ungar, G., Galvan, L., and Clark, R. H. (1968). Nature (London) 217, 1259. Ungar, G., Desiderio, D. M., and Parr, W. (1972a). Nature (London) 238, 198. Ungar, G., Galvan, L., and Chapouthier, G. (1972b). Expcrientia 28, 1026. Weichert, R. F. (1970). Amer. J . Med. 49, 232. Weiss, J. M., McEwen, B. S., Silvan, M. T. A., and Kalkut, M. F. (1969). Science 163, 197. Yaksh, T. L., and Myers, R. D. (1972). Amer. J . Physiol. 222, 503. Young, J. 2. (1938). “The Evolution of the Nervous System and of the Relationship of Organism and Environment.” Macmillan, New York. Zetler, G. ( 1970). In “Aspects of Neuroendocrinology” (W. Bargmann and B. Scharrer, eds.) , p. 287. Springer-Verlag, Berlin and New York. Zippel, H. P., ed. (1973). “Memory and Transfer of Information.” Plenum, New

York.

Zippel, H. P., and Domagk, G. F. (1969). Expcrisntia 25, 938.