International Review of Neurobiology Volume 1

INTERNATIONAL REVIEW OF Neurobiology VOLUME 1 This Page Intentionally Left Blank INTERNATIONAL REVIEW OF Neurobio...

Author: Carl C Pfeiffer

26 downloads 898 Views 22MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

INTERNATIONAL REVIEW OF

Neurobiology VOLUME 1

This Page Intentionally Left Blank

INTERNATIONAL REVIEW OF

Neurobioloav Edited by CARL C. PFEIFFER Emory University, Atlanfa, Georgia

JOHN R. SMYTHIES The Worcester Foundation for Experimental Biology, Shrewsbury, Massachusetts

Associate Editors V. Amassian J. A. Bain D. Bovet Sir Russell Brain Sir John Eccles

VOLUME

E. V. Evarts H. J. Eysenck F. Georgi G. W. Harris R. G. Heath

C. Hebb A. Hoffer K. Killam S. M6rtens

1

@

1959

ACADEMIC PRESS, New York and London

Copyright

0, 1959, by Academic Press Inc. ALL RIGHT8 REBERVED

NO PART OF THIB BOOK MAY BE REPRODUCED I N ANY FORM, BY PHOTOBTAT, MICROFILM, OR ANY OTHER MEANB, WITHOUT WRITTEN PEBMIBBION FROM THlD PUBLIBHERB.

ACADEMIC PRESS INC. 111 FIFTHAVENUE NEWYORK3, N . Y.

United Kingdom Edition Published by ACADEMIC PRESS INC. (LONDON)LTD. 40 PALLMALL,LONDON S.W. 1

Library of Congrea Catalog Card Number 69-15822

PRINTED I N THE UNITED STATES OF AMERICA

CONTRIBUTORS W. R. ADEY,Departments of Anatomy and Physiology, School of Medicine, University of California, Los Angeles, California, and the Veterans Administration Hospital, Long Beach, California F. GEORGI, The Research Laboratory of the University Psychiatric Hospital and the University Neurological Policlinic, Basel, Switzerland G. W. GRANGER, Psychology Department, Institute of Psychiatry, Maudsley Hospital, London, England ROBERTG. HEATH,Department of Psychiatry and Neurology, Tulane University School of Medicine, New Orleans, Louisiana CATHERINE 0. HEBB,The A.R.C. Institute of Animal Physiology, Babraham, Cambridge, England C. G . HONEGGER, The Research Laboratory of the University Psychiatric Hospital and the University Neurological Policlinic, Basel, Switzerland D. JORDAN, The Research Laboratory of the University Psychiatric Hospital and the University Neurological Policlinic, Basel, Switzerland S. MRRTENS,The Beckomberga Hospital and the Research Department, A / B Kabi, Stockholm, Sweden B. MELANDER, The Beckomberga Hospital and the Research Department, A / B Kabi, Stockholm, Sweden CARLC. PFEIFFER, Division of Basic Health Sciences, Emory University, Atlanta, Georgia DOMINICK P. PURPURA,Paul Moore Research Laboratory, School of Neurological Surgery, College of Physicians and Surgeons, Columbia University, New York, New York H. P. RIEDER,The Research Laboratory of the University Psychiatric Hospital and the University Neurological Policlinic, Basel, Switzerland M. ROTTENBERG, The Research Laboratory of the University Psychiatric Hospital and the University Neurological Policlinic, Basel, Switzerland S. VALLBO,The Beckomberga Hospital and the Research Department, A / B Kabi, Stockholm, Sweden

V


PREFACE Progress in neurobiological research must maintain a delicate balance between the fascination of basic explanation of clinical and physiological phenomena by means of chemical and physical concepts on the one hand and the pressing needs for the development of new and effective treatments of disease on the other. Advances in basic biochemistry and biophysics often give rise to developments in the clinical field, but mature judgment is required to select from the vast detail of biochemistry and biophysics, those parts which are likely to apply to human disease. The aim of this Review is to enable active workers in these fields of neurobiology, neurochemistry, neuroanatomy, neuropharmacology, neurophysiology, psychopharmacology, psychology, etc. as well as those in biological psychiatry and neurology to give an account of recent progress in their fields, The Review covers the whole field of neurobiology and includes work within a particular basic science as well as in neurology and psychiatry. Particular emphasis has been laid on the recent development of ideas of fundamental importance and general interest and also of ideas from these basic fields likely to further our understanding of nervous and mental disease. In the past the basic neurobiological sciences have played no little part in progress toward these ends. They are most active a t present and they hold great promise for the future. The purpose of this Review is to contribute something to this objective.

CARLC. PFEIFFER JOHNR. SMYTHIES July 10, 1969

Vii


CONTENTS CONTRIBUTORS . . P R E F A C E .. . .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

v

vii

Recent Studies of the Rhinencephalon in Relation to Temporal Lobe Epilepsy and Behavior Disorders W. R. ADEY

I. Introduction . . . . . . . . . . . . . . 11. Anatomical and Physiological Studies of Neuronal Systems . 111. Epileptic Phenomena in the Temporal Lobe . . . . . IV. Behavior Studies Following Rhinencephalic Stimulation and lation . . . . . . . . . . . . . . . . V. Neuropharmacological Studies of Rhinencephalic Functions VI. Summary and Conclusions . . . . . . . . . . References . . . . . . . . . . . . . . .

.

.

.

.

.

1 2 20

. Ab. . . . . .

40

.

42

.

27 33

Nature of Electrocortical Potentials and Synaptic Organizations in Cerebral and Cerebellar Cortex DOMINICKP. PURPURA

I. Brain Waves as Axon-Spikes, Cell, “Dendritic,” and Postsynaptic Potentials I. General Introduction . . . . . . . . . . . . . . 11. Electrical Responses of Excitable Cells . . . . . . . . 111. Applied Axonology . . . . . . . . . . . . . . IV. Cortical Potentials as Synchronized, Spontaneous Fluctuations in Cell Membrane Polarization; the “Somatic Potential” Theory . . V. Brain Waves as “Graded Responses”; Current Theories of Cortical Electrogenesis . . . . . . . . . . . . . . . .

PART

11. The Synaptic Origin of Cortical Surface Potentials I. Introduction . . . . . . . . . . . . . . . . 11. Nature of Cortical Synapses . . . . . . . . . . . . 111. Applied Synaptology . . . . . . . . . . . . . . IV. Functional Differences in Cortical Axosomatic and Axodendritic . . . . . . . . . . . . . . . . . Synapses

48 50 53 58 61

PART

PART

111. Transmissional and Conductile Activity in Cortical Neuronal Organizations . . . I. Relationship of Spontaneous and Evoked Neuronal Discharges . . . . . . .

ix

79 80 82 97

Different

. . . . . . . 101 Potentials to Single . . . . . . . 101

X

CONTENTS

I1. Synaptic Organizations in Cerebral Cortex . . . . . . . . I11. Analysis of Different Synaptic Organizations in Cerebellar Cortex IV. General Considerations. Conclusions. and Summary . . . . . References . . . . . . . . . . . . . . . . . Chemical Agents of the Nervous System CATHERINE 0. HEBB I. Introduction . . . . . . . . . . . . . . . . I1 Chemical Morphology of Nerve Cells . . . . . . . . . I11. Intracellular Storage and Transport of Neurohumoral Agents : Neurosecretion . . . . . . . . . . . . . . . . IV Chemical Transmission in the Central Nervous System of Mammals V. Conclusions . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . .

.

.

Parasympathetic Neurohumors; Possible Precursors and Effect on Behavior CARLC . PFEIFFER I Introduction . . . . . . . . . . . . . . . . I1. The Role of Acetylcholine in Brain Function . . . . . . . I11. Effect of Autonomic Agents on the Conditioned Avoidance Response (CAR) . . . . . . . . . . . . . . . IV Effect of Tranquilizing Drugs on the Conditioned Avoidance Response . . . . . . . . . . . . . . . . . . V. Effect of Tertiary Amine Parasympathetic Stimulants on the CAR VI Effect of Arecoline in Schizophrenic Patients . . . . . . . VII Failure of Acetylcholine or Methacholine to Inhibit the CAR . . VIII. Limited Passage of Quaternary Amines Across the Blood-Brain Barrier . . . . . . . . . . . . . . . . . . I X Literature Survey on Deanol (2-Dimethylaminoethanol) . . . X Pharmacological Studies on Deanol . . . . . . . . . . XI Trial of Deanol in Human Subjects . . . . . . . . . . XI1. Choice of a Deanol Salt for Clinical Trial . . . . . . . . XI11 Possible Modes of Action of Deanol . . . . . . . . . . XIV . Deanol in Clinical Disorders . . . . . . . . . . . . XV Biochemical Adjuvants to Deanol Therapy . . . . . . . XVI . Hallucinatory Effect in Man of Acetylcholine Inhibitors . . . XVII Critique of Acetylcholine as the Parasympathetic Neurohumor . . XVIII . The SAR of Muscarinic and Nicotinic Ends of Acetylcholine Congeners . . . . . . . . . . . . . . . . . . XIX . Congeners of Deanol Designed to Increase Muscarinic Effect . . XX . A Theory for the Mode of Action of Benaotyzine . . . . . . XXI Summary . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . .

.

.

112 142 152 155

165 166 176 184 189 191

196 197 199

. .

200 201 202 202

. . . . . .

202 203 214 220 222 223 223 226 229 231

.

233 238 239 240 241

xi

CONTENTS

Psychophysiology of Vision G. W. GRANGER I. Introduction . . . . . . . . . . . . . . 11. Visual Effects of Physiological Stresses and Drugs . . . . 111. Summary and Conclusions . . . . . . . . . . . References . . . . . . . . . . . . . . . . ,

. 245 . 247

. 292 . 294

Physiological and Biochemical Studies in Schizophrenia with Particular Emphasis on Mind-Brain Relationships ROBERT G. HEATH I. Introduction . . . . . . . . . . . . . . . . 11. Serum Oxidative Studies in Schizophrenia: Ceruloplasmin and Inhibitors . . . . . . . . . . . . . . . . . . 111. Taraxein: Effects on the Brain and Behavior . . . . . . . IV. Amine Oxidase Inhibition . . . . . . . . . . . . V. Brain Tissue Extracts and Schizophrenia . . . . . . . . VI. Relationship of Psychological Factors to Biological Findings in Schizophrenia . . . . . . . . . . . . . . . . VII. Studies of Biological Changes Resulting from Stress . . . . . VIII. A Working Hypothesis in Schizophrenia . . . . . . . . IX. Summary . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . .

300 30 1 309 315 316 317 323 325 328 328

Studies on the Role of Ceruloplasmin in Schizophrenia S. MARTENS,S. VALLBO,and B. MELANDER I. Review of Previous Research on Serum Copper and Ceruloplasmin 11. Effects of Ceruloplasmin Administration to Schizophrenics . . . 111. Experimental Findings Related to the Administration of Ceruloplaamin to Schizophrenics . . . . . . . . . . . . 1V.Summary . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . ,

333 335 338 340 341

Investigations in Protein Metabolism in Nervous and Mental Diseases with Special Reference to the Metabolism of Amines F. GEORGI,C . G. HONEQGER, D. JORDAN, H. P. RIEDER,and M. ROTTENBERG I. Introduction . . . . . . . . . 11. Abnormal Findings in Protein Metabolism; Proteins . . . . . . . . . . . 111. Liver Function and Hippuric Acid Test .

.

. . . . . . 343

Stabbility of Plasma .

.

. . .

.

. .

.

I

. .

. 344 . 346

xii

CONTENTS

IV . Therapeutic Experiments . . V . Research in Amine Metabolism VI . Further Outlook . . . . VII . Summary . . . . . . References . . . . . .

AUTHOR INDEX. . . SUBJECT INDEX. . .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. . . . . . 348 . . . . . . 349 . . . . . . . 362 . . . . . . . 362

.

.

.

.

.

.

.

.

.

.

. 364

.

.

.

.

.

.

.

.

. 367 381

. . . . . . . . .

RECENT STUDIES OF THE RHINENCEPHALON IN RELATION TO TEMPORAL LOBE EPILEPSY AND BEHAVIOR DISORDERS By W. R. Adey Departments of Anatomy and Physiology, School of Medicine, University of California at Angeles, and the Veterans Administration Hospital, Long Beach, California

I. Introduction

. .. .. .....

....................

. . . . . . . . . . .. . . ............................ Area . . . . . . . . . . . . . terns

A. The Hippocampal-Fornix System . ,

C. Interrelations between the Entorhinal Area and the Brain Stem . . . . . . D. Interrelations between Rhinencephalic and Subcortical Structures; General Considerations . . . . . . , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E. Interrelations between the Rhinencephalon and the Corpus Striaturn . . 111. Epileptic Phenomena in the Temporal Lobe . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Animal Studies . . . . . . . . . . . . . . . . . . , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Human Studies in Temporal Lobe Epile .... ........ IV. Behavior Studies Following Rhinencephalic A. Cortical Ablation Studies B. Cortical Stimulation Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Behavior Changes Following Subcortical Lesions . . . . . . . . . . . . . . . . . . . . V. Neuropharmacological Studies of Rhinencephalic Functions A. The Possible Role of Serotonin aa a Central Transmitter B. Evidence concerning the Effect of Striatal Lesions on the Action of Reserpine . . . . . . . . . . C. The Mode of Action o phalic Mechanisms .......... VI. Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References ...............................

Lor 1 2 3 8 10 12 16 20 20 26

27 27 30 30 33

36 37

39 40 42

1. Introduction

To tread once more the well-trodden paths of the reviewer of rhinencephalic functions, to seek amid the arches of the allocortex at least some aspect of its activities thrown into bold relief by recent research, may seem, indeed, a dubious field of endeavor, since in recent times many aspects of its functions have been comprehensively reviewed (Brodal, 1947; Pribram and Kruger, 1953; Maclean, 1955; Thomalske et al., 1957; Adey, 1956, 1958). Nevertheless, within the limited scope of a review 1

2

W. B. ADEY

directed mainly to some of the major contributions in the last three years, there would seem to be significant pointers which may yet lead to the unraveling of the tangled skein that binds the rhinencephalon to the diencephalon, the basal ganglia, and the rostra1 midbrain. It is proposed, therefore, to deal with this topic from four different aspects, including anatomical and physiological arrangements of neuronal systems, investigations of seizure discharges in the limbic system, behavior studies concerned with stimulation and ablation in these areas, and with metabolic and neuropharmacological investigations. The classic contributions of Elliot Smith (1910), Herrick (1933), and Papez (1937) are too well known to need elaboration here, but they have indeed formed a valuable point of departure for many studies in the last twenty years. Papez’ contention that the pathways between the hippocampus and the diencephalon provided an anatomical substrate for neuronal activity concerned with emotion has been shown to be substmtially correct. The Papez “circuit” proposed that activity initiated in the hippocampus would pass through the fornix to the mammillary bodies, thence via the mammillothalamic tracts to the anterior thalamic nuclei. Thalamic radiations carry the activity to the cingulate cortex, and, thence, further relays, proceeding partly through the cingulum bundle, would ultimately reach the presubiculum and the entorhinal area around the caudal border of the corpus callosum. The temporoammonic tracts of Cajal would provide the final pathway for this activity to re-enter the hippocampus from the entorhinal area (Fig. 1). Although the role of hippocampus and adjacent allocortical structures iQ emotional functions is generally conceded, their mode of interconnection is more complex than Papez envisaged, with important byways and even converse paths revealed by anatomical and physiological studies in recent years. We may now turn to these in more detail.

II. Anatomical and Physiological Studies of Neuronal Systems

Attention has frequently been directed by Maclean (1955) to the participation of a ring of cortical structures a t the medial border of the cerebral hemisphere forming the limbic lobe, as suggested by Broca (1878), and to which Maclean has applied the term “visceral brain.” Evidence that the hippocampus does indeed receive visceral aff erents has been supplied by Dunlop (1958). Stimulation of the stomach electrically, mechanically, and chemically modifies the hippocampal activity, with electrical stimulation eliciting the most striking results. A slow-wave response is evoked in the hippocampus 500 msec after stimulation. Where

THE RHINENCEPHALON AND BEHAVIOR DISORDERS

3

a primary response with short latency occurred, no slow-wave ensued, and evidence is presented that the primary response may have resulted from stimulation of adjacent somatic structures. By contrast, no viscerosensory representation was seen in the amygdala, which is of interest in view of Dell and Olsen’s findings (1951) of amygdaloid responses to stimulation of the cervical vagus.

A. THEHIPPOCAMPAL-FORNIX SYSTEM The neuronal circuit proposed by Papez envisaged the fornix as carrying activity from the hippocampus t o the mammillary bodies. It is only recently that the complexities of the fornix bundles have been reexamined. Cajal (1911) had described it as the only efferent pathway from the hippocampus. There is evidence that many of its fibers terminate before reaching the hypothalamus, and that others enter the precommissural fornix to end in the septum. T. P. S. Powell and coworkers (1957) have counted the fibers in the postcommissural fornix and the number of cells in the medial mammillary and anterior thalamic nuclei in rabbit, cat, monkey, and man. The absolute number of fibers in rabbit, cat, and monkey is of the same order (200,000). In the rat they number 50-60,OOO and in man, 1-2 million. In primates, the number of fibers in the precommissural fornix approximately equals that in the postcommissural fornix. One-half to one-third of the postcommissural fibers fail to reach the mammillary body in all species. The majority are lost in the rostral third of the hypothalamus and probably form a direct hippocampothalamic pathway. The anterior thalamic nuclei appear t o receive the same number of afferent fibers directly from the fornix as by way of the mammillary nuclei. Extension of the connections of the postcommissural fornix beyond the immediate confines of the mammillary bodies has also been seen in the rat by Guillery (1956) and Nauta (1956). They have detected in experimental anatomical material terminations of fornix fibers in the anterior thalamus, periventricular system and rostral midbrain. Many fibers leave the dorsal aspect of the postcommissural fornix and enter the anteromedial and anteroventral thalamic nuclei. Few, if any, enter the anterodorsal nuclei. In the hypothalamus, fibers terminate in the dorsal hypothalamic area, preoptic region, and nucleus of the diagonal band. Nauta describes a separate projection from the caudal third of the hippocampus to the periventricular zone of the hypothalamus, and particularly to the arcuate nuclei of the tuberal region. It is of interest that some of the fornix fibers terminate in the periaqueductal gray matter of the midbrain, since major pathways to this region have been found from the entorhinal area (Adey et al., 1956), following lesions which spared

4

W. R. ADEY

FIO.1. A. A mechanism of emotion (after Papea). Activity from the entorhinal area pames via the hippocampus to the fornix bundles and thence to the mammillary body. The mammillothalamic tract then conveys activity to the anterior thalamic nuclei which project to the cingulate cortex. Activity returns to the hippocampal formation once again via the cingulum bundles running posteriorly in the cingulate gyrus. B. A scheme suggested by recent investigations, with activity in the fornix running in the converse direction to the paths in A. Here, hippocampal activation is induced from the reticular thalamic nuclei and passes in turn to the entorhinal area. Pathways exist between the entorhinal area and the reticular formation (see Fig. 2). Abbreviations : A.C., anterior commissure; AM., amygdaloid complex : ANT. CING., anterior cingulate area; CORP. CALL., corpus callosum ; HIPP, hippocampal fornation; INS, insular cortex; M, mammillary body; MED. FRONT., medial frontal cortex; POST. CING., posterior cingulate area; RET. FORM., reticular formation; SEPT., septa1 area; TEMP. POLE, cortex of temporal pole; TH., thalamus (from Adey, 1956).


5

the hippocampus. The size of the fornix contribution to these midbrain zones appears much less than from the entorhinal area, a t least in our electrophysiological studies described below. There remains the much more intriguing aspect of fornix functions concerned with its septal connections. It has become clear that fornix fibers form, in part, an afferent pathway to the septum, but much less is known of possible septofugal pathways coursing caudally in the fornix and perhaps terminating in the hippocampus. Such pathways would offer a channel of entry to the hippocampus for activity ascending from the brain stem and thalamus to the septum. Behavioral and electrophysiological evidence supports the notion that the septum may be a waystation on diencephalic paths from the brain stem. Green and Arduini (1954) found that septal lesions block the slow-wave response in the hippocampus to an alerting stimulus, and, in a systematic survey of stimulus thresholds in diencephalic areas yielding hippocampal responses, Green and Adey (1956) found the most sensitive zones in the septum, fornix, and anterior thalamus. Precautions taken in these experiments could not entirely preclude the possibility of either antidromic activation of pyramidal cells, or orthodromic activation of some cells through axon collaterals running a recurrent course in the alveus. These problems have to some extent yielded to histological and further electrophysiological investigations. A septofugal component ascending in the fornix of the guinea pig was described by Morin (1950), but he was unable t o detect its termination. McLardy (1955a, b) found, after fornix section in the monkey, finely myelinated fibers running to or through the opposite lateral septal nucleus, but considered that there might also be some decussating septofugal fibers coursing caudally from the medial septal nucleus. Fibers of the medial fifth of the body of the fornix, when followed caudally, curved dorsally and posteriorly through the corpus callosum to emerge into the cingulum, and thence fanned deeply into the parietotemporal white matter. Such an arrangement would provide an additional path for septal activity to reach the hippocampus through relays in the cingulate gyrus. White et al. (1958) have recently re-examined caudally directed activity in the cingulate gyrus, reaching the presubiculum and entorhinal area along pathways as proposed by Papee (1937). Certain difficulties are attached to histological investigations following fornix section, due to the well-known resistance of hippocampal cells to retrograde degeneration in these circumstances. McLardy could detect no cellular changes in hippocampus, dentate gyrus, subiculum, or entorhinal cortex. He ascribes the absence of change to persistence of collateral axons which pyramidal hippocampal cells are known to distribute within the ipsilateral hippo-

6

W. R. ADEY

campus. By contrast, in 8 of 25 hemispheres following fornix section or temporopolar lesions, nerve cells in the medial septal nucleus were degenerated widely or focally, suggesting that there may be a septofugal system coursing between the septal region and isocortex of the temporal lobe, although it would seem that septo-hippocampal connections are not thereby excluded. Since section of the fimbria rostra1 to the hippocampal recording site does not abolish the large response to fornix stimulation (Green and Adey, 1956)) the question arises as to the course which such afferent fibers might follow within the hippocampus. It has been noted that during exploration in depth radially through the hippocampus, “turnovers” occur in the responses to fornix stimulation (Green and Adey, 1956), and that various components of the response are differentially susceptible to anesthesia and asphyxia. There is an early response which may be construed as “presynaptic” t o the later “cellular” potential. This presynaptic component displays a greater resistance to asphyxia and anesthesia than the later cellular component. Exploration of the hippocampus with a bipolar electrode arranged t o pass radially through the hippocampus (and, thus, to be oriented at first in the long axis of the hippocampal pyramidal cells in region CA2 of Lorente de N6 (1934), and a t greater depths to pass through region CA4, between dentate and hippocampal pyramidal cell layers, before finally penetrating the dentate pyramidal cell layer) , indicates a “turnover” in the region CA4, which may be interpreted as a synaptic activation at this level (Adey et al., 1957b). On the basis of this evidence, it may be suggested that this hippocampal afferent pathway runs deeply within the hippocampus to synapse on the dentate granule cells. Here activity is relayed t o the arc of hippocampal pyramidal cells situated more peripherally in the Ammon’s horn. Convergence of different sensory modalities on the cells of the dentate gyrus has been described (Green and Machne, 1955). Further anatomical and physiological evidence supports this hypothesis (Fig. 2). In histological investigations of hippocampal commissural and septal connections in the rabbit, Cragg and Hamlyn (1956) sectioned the fimbria on one side and found degenerating fibers in the posterior half of the fimbria of the opposite side, extending into fields CA3 and CA4 of the hippocampus. In the ipsilateral hemisphere additional degeneration was seen in the fimbria extending back over the dorsal alveus to all layers of the presubiculum. The authors consider that these fibers could have come from the septal nuclei. Lesions of the angular tract of Cajal produced degeneration sharply delimited to the upper part of the entorhinal cortex bilaterally and present in all layers of the cortex. Black-

THE) RHINENCEPHALON AND BEHAVIOR DISORDERS

7

stad (1956), however, has undertaken a more extensive survey of the hippocampal commissural connections in the rat, in the belief that the terminal ramifications from interhemispheric fibers end a t definite cortical levels, exclusively different for each area studied. Further electrophysiological studies have revealed that the earliest component of the hippocampal response to fornix stimulation, a sharp diphasic potential with a latency of less than 1 msec, probably represents a fiber response in the fimbria and alveus, and that if it is partly antidromic, it nevertheless fails to invade the hippocampal pyramidal cell somata (von Euler et al., 1958). A second small deflection with a latency of 1.5 msec is seen in area CA4 a t the level of the dentate granule cells. The main deflection has a latency of 4 4 msec, and is a negative potential only a t the level of the hippocampal pyramidal cell bodies. Above and below the layer of hippocampal pyramidal cells, the deflection normally has a positive sign, suggesting that these cells are synaptically activated close to the cells and, also, that propagation is blocked in the dendritic layer. This apparent block of somatofugal conduction along the pyramidal cell dendrites would seem a point of major interest in relation to current theories of cellular excitation concerned with the role of dendritic excitability. Green and his colleagues have further elaborated their data to display latency shifts and current flows in progressive penetration of the hippocampus. At 1.8 msec a negative peak occurs at the level of the granule cells, and disappears at 2.7 msec. At 3.6 msec there is very little sign of any gradient along the tract, but at 5.4 msec a negative peak appears a t the pyramidal cell level and grows progressively. A marked voltage gradient between the cell layer and the dendrites persists up to 21 msec. In the period 9-20 msec, the synaptic zone between dentate axonal terminals and pyramidal cell apical dendrites becomes isopotential and may represent the point where outward current of the pyramids ceases and outward current of the granule cells begins. Other aspects of dendritic functions in the cerebral cortex have been discussed by von Euler and Ricci (1958). 1. Summary of Hippocampal-Fornix System

Recent investigations have shown that a large part of the postcommissural fornix terminates in the anterior thalamus, dorsal hypothalamic, and preoptic areas. A few descend in the periventricular system as far as the central gray of the midbrain. Only about one-half of the postcommissural fibers reach the mammillary body in all mammals. The precommissural fornix in man is about the same size as the postcommissural.

8

W. B. ADEY

Evidence is presented concerning a septofugal pathway running in the fornix to enter the hippocampus. Its suggested intrahippocampal course is from the dentate granule cells to hippocampal pyramids.

B. RELATIONS OF

THE

HIPPOCAMPUS WITH

THE

ENTORHINAL AREA

Classic accounts of the hippocampus have emphasized the afferent paths reaching it from the entorhinal area. Cajal (1911) and Lorente de N6 (1934) have described in elegant detail the course of the temporoammonic tracts arising in the entorhinal area and passing, as the perforant and alvear paths, to terminate in relation to the arch of the hippocampal pyramidal cells. These pathways have formed the basis of Papez’ proposal that activation of the hippocampus occurred largely from the entorhinal area and this activity in turn passed to the hypothalamus through the fornix. The existence of the temporoammonic pathways has been confirmed histologically (Adey and Meyer, 1952a) , and potentials can be evoked in the hippocampus from entorhinal stimulation (Adey et al., 1957b). The functional continuity of the series of pathways from the entorhinal area to the fornix appears less certain. I n the marsupial (Adey et al., 1957b) and the cat (Adey, Killam, and Killam, unpublished observations) , despite the appearance of evoked potentials in the hippocampus from the entorhinal stimulation, only small and irregular responses occur in the fornix and septum. This confirms the earlier observations of Renshaw et al. (1940) that, following entorhinal stimulation, they were unable to discern a response with microelectrodes in the fimbria and fornix which could be satisfactorily separated from the cellular response in the hippocampus itself. These effects of entorhinal stimulation contrast sharply with the results of fornix stimulation, where impulses clearly pass freely in the opposite direction through the hippocampus to reach the entorhinal area. Indeed, unit activity in the entorhinal cortex, concurrently with a slow-wave train, is readily evoked from fornix stimulation (Adey, 1958). Some importance may be attached to these connections passing caudally from the hippocampus to the entorhinal area in a direction opposite to the course of the temporoammonic tracts. Lorente de N6 (1934) has described these fibers as a major afferent pathway to the entorhinal area. In his Golgi preparations, he observed bifurcation of axons of pyramidal cells as they entered the alveus, and noted that the collateral branches ran a recurrent course to enter the subiculum. Some of these branches could be traced still further into the entorhinal cortex.


9

Cajal (1911) had earlier pointed out that at least some hippocampal neurons pass to the subiculum, and that a further projection tract might arise in the subiculum, running to an undetermined destination. The studies of Cragg and Hamlyn (1956, 1957), cited above, suggest that a septofugal component may also run caudally over the hippocampus in the alveus to reach the presubiculum. Thus, in summary, there can be little doubt that the hippocampus receives from the entorhinal area a significant input from the temporoammonic tracts. Conversely, there is also strong evidence that axons of SUB.

ENT. AREA

F I ~ 2.. A diagrammatic representation of suggested caudally directed pathways from the fornix running through the hippocampus to the entorhinal area, thence bilaterally through a series of pathways running in part through the stria medullaris to the midbrain tegmentum. Abbreviations: DENT. PYR., cells of fascia dentate : DORSAL HIPP., dorsal hippocampus; ENT. AREA, entorhinal area; HIPP. PTR., hippocampal pyramidal cell layer; M. B., transverse section through midbrain ; 5. M., stria medullaris; SUB., subiculum (from Adey et a2, 195ib).

hippocampal pyramids may bifurcate, one branch passing forward in the fornix and the other turning caudally into the subiculum and the entorhinal area. The ease with which this caudally directed path from the septa1 areas and the fornix can be activated, in contrast to the small responses seen in the septum from entorhinal stimulation, has raised the question of the extent of projections passing still further downstream from entorhinal cortex to brain stem and, indeed, of the extent of the interrelations of rhinencephalic areas, such as entarhinal area, hippocampus, and amygdala, with the diencephalon and midbrain,

10

W. R. ADEY

C. INTERRELATIONS BETWEEN

THE ENTORHINAL ABEA BRAINSTEM Removal of the entorhinal area is possible with minimal damage to the hippocampus, and our experiments in the marsupial and baboon have indicated altered behavioral responses following ventral temporal resections which will be described below. We may draw attention here to certain electrophysiological effects which follow entorhinal ablation in the marsupial phalanger (Adey et al., 1956). Since the available data indicates that the hippocampus may be activated both anteriorly from septum and fornix and posteriorly from the entorhinal area, removal of one or another of these inflows might be expected to modify the hippocampal response to an alerting stimulus. Removal of the entorhinal area in the phalanger does not abolish the hippocampal primary response or ensuing slow-wave train to a click (Adey et al., 1956). On the other hand, damage to the septum and fornix in the rabbit abolishes this slowwave response (Green and Arduini, 1954), suggesting that activation of the hippocampus through its anterior influx may be responsible for its characteristic response to alerting stimuli, It should be mentioned that Carreras et al. (1955) were unable to elicit slow-wave responses in the hippocampus of the cat following entorhinal ablation, possibly due to the acute conditions of their experiments. Evoked “spindle” trains are commonly seen in the auditory cortex in response to a click stimulus following removal of the entorhinal area. These spindle trains occurred with a latency of about 300 maec following the primary response, with 5 to 10 spindles in each burst (Adey et al., 1956). These spindle trains may indicate a subtle change in the normal cortico-diencephalic relationships. Bremer (1954) has noted similar trains of slow waves following the primary auditory response in encdphale isole’ preparations and has correlated the phenomenon with the onset of sleep in these animals. The appearance of these evoked spindle trains may indicate a waning consciousness or, perhaps more precisely, a diminution of alerted or directed attention, for we have noted them in normal animals after many hours of intermittent auditory stimulation in the stereotaxic instrument. Gerebbaoff (1941-1942) noted spontaneous spindling a t 6 to 9 per second in cats and rabbits under hypnosis induced by exposure to bright lights or by various postural maneuvers. He also noted spindle trains in the auditory cortex in these animals.following the primary response to click stimulation. Similar spindles have been reported after midbrain tegmental lesions and, in normal sleep (Bremer, 1935; Whitlock 8t al., 1953). If the occurrence of these spindle trains, after entorhinal resection, is

AND THE


11

indeed a measure of altered interrelations between the cerebral cortex on the one hand and the rostral midbrain or diencephalon on the other, it may be suggested that the entorhinal area is capable of exercising an especially powerful influence on the central tegmental areas of the rostral midbrain, since removal of other cortical areas has not been described as producing such an effect on evoked cortical responses. It may also be relevant here that only surgical interference in the frontal and temporal lobes has been shown to produce alterations in emotional aspects of behavior, despite extensive resections in other parts of the cortical mantle of man and animals. Our attention has been directed to the possibility of assessing the effects of corticifugal volleys from various parts of the cortical mantle on ascending conduction in the central core of the midbrain tegmentum. This central zone contains both rapidly and slowly conducting systems which jointly constitute the “reticular formation,” with neurons accessible to collaterals of many of the more laterally situated lemniscal axons. This central core is known to mediate both ascending and descending influences, and contains an ascending system capable of greatly affecting activity in the cerebral cortex (Moruzzi and Magoun, 1949) , and known to account for the more obvious electrical and behavioral phenomena associated with arousal (Magoun, 1952). In studies of the effects of corticifugal volleys in cat and monkey, we have examined a slowly conducting pathway from the pontine tegmentum to the centrum medianum of the thalamus (Adey et al., 1957a). Conditioning stimuli (trains of impulses or single shocks) delivered to certain cortical loci prior to the test stimulation of the caudal reticular formation were observed to modify the response in the centrum medianum, with either blocking interaction, augmentation, or each of these in a complex sequence. Active cortical loci producing these effects include the sensorimotor cortex, frontal oculomotor area, paraoccipital region, superior temporal gyrus and temporal pole, medial frontal cortex, anterior cingulate gyrus, and the basal occipitotemporal cortex, including the entorhinal area. Other cortical loci had no apparent effect on ascending conduction in this brain stem system. The most pronounced and enduring effects followed stimulation of the basal occipitotemporal cortex. The active cortical areas defined in this study resemble in arrangement those having major projections to the brain stem (French et al., 1955). Stimulation of these areas is also capable of arousing the monkey from normal sleep, whereas stimulation of other cortical zones is ineffective (Segundo et al., 1955). In the light of these findings, we have first sought evidence of anatomical and physiological connections between the entorhinal area and

12

W. R. ADEY

the rostral midbrain. In degenerating fiber studies of silver impregnated material in the marsupial phalanger, following entorhinal ablation, a pathway has been found to run initially via the external capsule into the anterior commissure and then to be distributed bilaterally through the stria medullaris, and in subjacent areas of the dorsal thalamus, and to terminate in the dorsal tegmentum of the midbrain adjoining the periaqueductal gray matter (Adey et al., 1956). Responses from entorhinal stimulation in the periaqueductal region and dorsal tegmentum of the rostral midbrain have many of the characteristics of a monosynaptic pathway, and exhibit a shorter latency than those from hippocampal stimulation (Adey et al., 195713). Unit records from the periaqueductal gray matter a t the rostral midbrain have indicated a profound inhibition of spontaneous firing after entorhinal stimulation, whereas more ventrally placed tegmental units, which respond to sensorimotor cortical stimulation, have not beeli influenced by entorhinal stimulation. Gloor (1955a, b) observed short latency responses from amygdaloid stimulation in the same regions of the midbrain tegmentum as those in which fibers of the entorhinal area appear t o terminate in our experiments, and he emphasized that these responses have a much shorter latency than those recorded in the immediately rostral regions of the posterior hypothalamus. Whitlock and Nauta ( 1956) have described additional direct paths to the midbrain tegmentum, arising from the temporal cortex outside the allocortical areas on the basal aspect.

D. INTERRELATIONS BETWEEN RHINENCEPHALIC AND SUBCORTICAL STRUCTURES ; GENERAL CONSIDERATIONS We have presented evidence in support of the hypothesis that a series of pathways arising in the septum may convey activity through the hippocampus to the entorhinal area, and thence by further relays to the midbrain tegmentum. It is easy in these circumstances t o focus attention on a few confirmed anatomical and physiological pathways, and to endow them with a significance quite disproportionate to what may ultimately be discovered about their functional roles. It is obviously easier to ignore the multitude of unmyelinated or polysynaptic connections involved in any general scheme than to attempt their incorporation into such a scheme, despite the probable importance of their contribution to the totality. It has, nevertheless, become apparent that the pursuit of these less tangible elements in the frame of neural organization may be well worthwhile. Thus, the effects of frontal lobotomy, initially explained on the basis of interruption of connections between the dorsomedial nucleus of


13

the thalamus and the frontal lobe, have more recently been attributed to a wide variety of subcortical connections, arranged in series and parallel, and terminating at all levels of the neuraxis (Meyer and Beck, 1954; Yakovlev et al., 1950). Widespread subcortical projections, partly polysynaptic, have been described in the amygdala of the cat (Gloor, 1955a, b) . This reappraisal of comparatively simple early concepts is attributable to evidence of trans-synaptic activation in electrophysiological studies, and to careful histological search for degeneration appearing slowly in the wake of surgical resections, and which may be in part trans-synaptic. Bucy and Kluver (1955) have reviewed the degeneration in one of their monkeys, subjected to bilateral temporal lobectomy in 1936 and 1937 and have described a wide series of subcortical connections, including corticotectal fibers passing through the pulvinar to the superior colliculus, corticonigral, and temporopulvinar. The stria terminalis was largely degenerated and could be traced along the tail of the caudate nucleus as far as the head, where it dispersed fanwise toward the anterior commissure. In seeking a clearer view of the interrelations of hippocampal and entorhinal regions of the rhinencephalon with the brain stem, our attention has been directed to the relative extent of the descending projection fields of the hippocampus and of the entorhinal area (Adey et al., 195813). A systematic examination of subcortical areas a t all levels from anterior septum to inferior colliculus has been made in a series of marsupial phalangers. Descending influences from the entorhinal area appear to pass caudally through the thalamus in two streams. A dorsal series of paramedian pathways extends posteriorly through the stria medullaris and subjacent paratenial and dorsomedial thalamic nuclei as far caudally as the nucleus parafascicularis. At this posterior thalamic level, the dorsal stream fused with one coursing further ventrally through the nucleus reuniens and dorsal hypothalamus. From these posterior thalamic and periventricular regions further projections occur into the periaqueductal gray matter and adjacent dorsal tegmentum of the midbrain. These projections of the entorhinal area are more extensive at lower levels than descending hippocampal influences. Descending hippocampal influences mainly follow classic pathways of the fornix system, with caudal extension of responses into anterior thalamic nuclei and dorsomedial thalamic nucleus. In the rostra1 midbrain the responsive area tapers sharply, and only a few responses are seen in the periaqueductal gray matter. By contrast with the comparative extent of the descending fields, ascending influences reaching the hippocampus arise from much wider

14

W. R. ADEY

regions of the brain stem than do those forming the descending projection fields of either hippocampus or entorhinal area. Hippocampal responses to brain stem stimulation frequently occur in the absence of a discernible response in the entorhinal area. Entorhinal responses to brain stem stimulation usually occur only where a large hippocampal response is evoked, with the entorhinal response displaying a longer latency and time-course than the hippocampal response. It appears that the hippocampus receives afferent5 from a very wide area of the rostral midbrain and thalamus, and that a smaller portion of the hippocampal activity so engendered passes ultimately to the entorhinal area. Examination of the reciprocity of paths between the entorhinal area and brain stem supports the view that the field of descending entorhinal influences is significantly extended by multisynaptic pathways. Certain aspects of these rhinencephalic interrelations with thalamic and hypothalamic structures merit further attention. Bodian (1939, 1940) has drawn attention to the possible role of the dorsal hypothalamus in a pathway from the rhinencephalon to the periaqueductal gray matter of the midbrain in the Virginian opossum. He describes an extensive tractus amygdalohypothalamic dorsalis arising mainly in the medial amygdaloid nucleus and running with the stria medullaris component of the stria terminalis. These fibers terminate in the dorsal hypothalamic nucleus, the periventricular system, the posterior hypothalamic nucleus, and the periaqueductal gray matter. His studies suggest that these pathways form both ascending and descending connections. Our findings support this concept, and suggest that the longitudinal column of the nucleus reuniens, lying ventrally in the thalamus adjacent to the dorsal hypothalamic nuclei, sends major projections to the hippocampus. The proximity of the septa1 nuclei and fornix bundles to the nucleus reuniens anterior may provide a pathway through which this activity could reach the hippocampus. In a series of studies in the cat, rabbit, and guinea pig, Angeleri and co-workers (Angeleri and Carreras, 1956; Carreras et al., 1954, 1955; Macchi et al., 1955, 1956) have discussed the relationship of the midline and intralaminar thalamic nuclei to the rhinencephalic cortex. They found no monosynaptic connections from these nuclei t o the rhinencephalon. The extrathalamic projections of these nuclei run toward the rostral end of the thalamus, passing close to the corpus striatum, and they may terminate in the basal ganglia and mediobasal cortex of the cerebral hemisphere. These authors suggest that the centromedian, parafascicular, and suprageniculate nuclei may project to the basal ganglia, and that the paracentral, centrolateral, and mid-line nuclei may project both to the basal ganglia and mediobasal cortex of the cerebral hemisphere.

T H I RHINENCEPHALON AND BEHAVIOR DISORDERS

15

E. INTERRELATIONS BETWEEN THE RHINENCEPHALON AND THE CORPUS STRIATUM Although we have not detected responses in the basal ganglia from stimulation of either hippocampus or entorhinal area, we have found evidence of connections between the amygdala and the basal ganglia in the monkey. After amygdaloid ablation, degenerating fibers can be traced as obliquely directed fascicles into the ipsilateral globus pallidus and putamen, with numerous swollen terminals in the globus pallidus. Less extensive degeneration occurs in the contralateral globus pallidus (Adey et al., 1958a). We have sought evidence of converse connections between the globus pallidus and the amygdala, but only scattered degenerating fibers have been detected on the lateral aspect of the amygdala, with minimal degeneration within the nucleus (Adey and Rudolph, unpublished observation). These findings of extensive interconnections between the amygdala and the globus pallidus are in conformity with their development from adjacent pallial areas (Elliot Smith, 1918; Johnston, 1923). In the light of these histological findings, Dr. Dunlop and I have examined in current research the extent of evoked responses in pathways between the amygdala and the basal ganglia in the marsupial phalanger and the monkey. In the phalanger, stimulation of the basal part of the lateral amygdaloid nucleus evokes widespread responses in the ipsilateral globus pallidus and putamen, in a responsive zone which passes without interruption into the subjacent amygdala. Responses in the ventral parts of the putamen and globus pallidus are large with a short latency (less than 5.0 msec) and appear to form a primary projection field. I n both monkey and phalanger, responses in the ventral pallidum during progressive penetration with a coaxial electrode show a “turnover” in bipolar records. Responses in the caudate nucleus in the phalanger t o amygdaloid stimulation are small and irregular in form. I n converse stimulation, the phalanger appears to possess no projections from caudate nucleus to amygdala, but stimulation of the globus pallidus evokes large responses in the lateral amygdala, although these may in part be antidromic (Fig. 3). A similar picture of evoked striatal responses from amygdaloid stimulation has been found in the monkey. Here contralateral pallidal responses occurred, but contralateral responses in the putamen were small and scattered. No contralateral responses were seen in the caudate nucleus. In contrast to the phalanger, ascending projections from the corpus striatum to the amygdala are much less powerful. The responses recorded here in the globus pallidus from single shock stimulation of the

16

FIQ.3. Amygdalo-striatal interrelations in the marsupial phalanger (A and B) and in the monkey (C, D, and E). Open circles indicate no responses, and the relative size of the closed circles indicates the relative amplitude of the primary response at each point. A. Distribution of responses in the ipsilateral basal ganglia of the phalanger to stimulation of the basolateral amygdala (as indicated in inset). Stimulus parameters, 12 volts, 2.0 rnsec square waves. B. Arrangement of points in ipsilateral basal ganglia of phalanger producing responses at the amygdaloid region indicated in inset, with the same stimulus parameters as in A. C. Responsive points in ipsi- and contralateral basal ganglia of monkey to stimulation of lateral amygdaloid nuclei (at lateral point in inset of amygdala). Stimulus parameters, 10 volts, 1.0 msec square waves. D. E. Responses in the basal and lateral amygdala respectively (at points indicated in inset of amygdala) to stimulation of ipsilateral and contralateral basal ganglia. It will be seen that reciprocity of amygdalo-striatal conduction is more obvious in the phalanger than in the monkey, where the preponderant pathway is caudally directed from amygdala to basal ganglia. Stimulus parameters, 10 volts, 1.0 msec square waves. Abbreviations: UP.,globus pallidus; PUT.,putamen.

THB RHINENCEPHALON AND BEHAVIOR DISORDERS

17

lateral amygdala (stimulus amplitude 10 volts, duration 1.0 to 2.0 msec in both phalanger and monkey) appear more extensive than those described by Gloor (1955a, b) in the cat. Gloor’s figures indicate responses limited to the medial part of the globus pallidus, with a latency of 7-14 msec, using stimuli with a duration of 5.0 msec (Figs. 4 and 5). Before turning specifically to problems of the functional significance of these descending streams of amygdaloid activity passing through the globus pallidus, attention may be directed to thalamic inflows to the corpus striatum. Cowan and Powell (1955) have described connections from the intralaminar and mid-line thalamic nuclei to the corpus striatum in the rabbit, particularly with the head of the caudate nucleus and the putamen. Although these findings have been attributed to incidental damage to the internal capsule (Nashold et ul., 1955), confirmatory evidence comes from the studies of Angeleri, Carreras, and Macchi cited above. In view of the role of the intralaminar thalamic systems in mechanisms of alerting and arousal, it is conceivable that these thalamostriatal connections, together with the influx through amygdalo-striatal connections, may be concerned in the elusive problem of willed movements. Buchwald and Ervin (1957) have aroused the lightly sleeping cat by tetanic stimulation of the amygdala and globus pallidus in a fashion resembling that from stimulation of the reticular activating system, Both arrest of movement and automatic movements accompanying an amygdaloid seizure are well known, and Carey (1957) has drawn attention to the “great loathness to move” in monkeys with lesions in the globus pallidus extending ventral to it and involving pathways from the orbital and more particularly from the temporal region. A similar hypokinesia follows lesions in inferior temporal cortical areas (Laursen, 1955; Adey, 1958), and it may be relevant that pathways from the entorhinal area also traverse this region ventral to the globus pallidus in the marsupial (Adey et al., 1956). While there is general agreement that stimulation of the corpus striatum does not produce movements in the absence of possible spread of the stimulus to the internal capsule (Rioch and Brenner, 1938), the participation of the globus pallidus in mechanisms of abnormal movements continues to attract attention, both in relation to tremor of rest, as in Parkinson’s disease (Cooper and Bravo, 1958) and in the tremor of movement associated with cerebellar lesions (Carpenter et ul., 1958). Since little is known of the effects of descending amygdaloid and pallids1 volleys on unit activity in the tegmental areas of the brain stem, my colleague, Dr. Buohwald, and I have examined unit behavior in the perisqueductal gray matter, dorsal tegmentum, red nucleus, and substantia nigra at the level of the rostra1 midbrain in the cat. These find-

18

W. R. ADEY

ings will be presented in detail elsewhere, but may be summarized here. We have selected for this study units which fired spontaneously and were facilitated by sciatic stimulation. The effects of paired shocks, separated by an interval of 50 msec, to amygdala and sciatic nerve, globus pallidus and sciatic nerve, and globus pallidus and amygdala, were each tested at repetition rates of 1-10 per second. I. Stimulation Lateral Amygdala

2. Stimulation Putamen

3. Stimulation Globus Pallldur

F I ~4.. Ipsilateral amygdalo-striatal interrelationships in the monkey (from the same experiment ria in Fig. 3). A. B. Responses in the globus pallidus and putamen, respectively, to stimulation of the lateral amygdala. C. D. Smaller responses in the basal and lateral amygdaloid nuclei, respectively, from stimulation of the putamen at the same point aa in B. E. Response in the basal amygdaloid nucleus from stimulation of globus pallidus at the same point as in A. Stimulus parameters, 10 volts, 1.0 m e c throughout.

In these experiments, stimulation of the sciatic nerve alone, a t rates of 3 per second or faster, caused a sustained rise in firing rate (up to 160-200 per second), although at 10 per second, firing was often not as fast as at 5 or 7 per second. While separate stimulations of the globus pallidus and amygdala produced variable effects, paired amygdaloidpallidal stimuli often reduced the firing rate to a low level. It ie of particular interest that amygdaloid and pallidal volleys were capable of modulating the sustained high frequency firing induced by sciatic


19

stimulation alone, converting the sustained high rate of firing into a rhythmic series of high frequency discharges. This modulation was greatest in the range 3-7 stimuli per second. Administration of chlorpromazine (1-2 mg per kilogram) markedly lowered the spontaneous firing rate, and also replaced the sustained increase in firing rate following repetitive sciatic stimulation by a high frequency burst with each stimulus and little or no discharge between stimuli. Amygdaloidsciatic or pallidal-sciatic pairing of stimuli further enhanced this modu1. Stimulation

Lateral

Amygdala

2. Stimulation Pallidum

3. Stimulation Putamen

c

E. Lat. Amyg. Response

5 Fro. 5. Contralateral amygdalo-striatal interrelationships in the monkey (from the same experiment as in Fig. 3). A. B. Responses in the putamen and pallidum, respectively, from stimulation of the lateral amygdala. C. D. Stimulation of the contralateral pallidum fails to evoke significant responses in the basal or lateral amygdaloid nuclei. E. A small response in the lateral amygdala follows stimulation of the putamen.

lating effect. Trifluoperazine (2 mg per kilogram) left basic firing rates unchanged and enhanced firing rates induced by paired pallidal-amygdaloid stimulation. These findings may bear on mechanisms of tremor and abnormal movement. Since the units tested were all responsive to sciatic stimulation, it is possible that they play a part in reticulospinal motor mechanisms. Chlorpromazine has been shown capable of producing clinically a Parkinsonian syndrome (though in doses considerably larger than those employed here) and trifluoperazine may produce in monkeys a

20

W. R. ADEY

catatonia or Parkinsonian tremors superimposed on a catatonic state (Courvoisier et al., 1958). Whatever may ultimately be resolved from the enigma of Parkinsonism, these unit studies would appear to draw attention to two important concomitants of tremorigenic mechanisms, namely, the possible role of the peripheral input and the possibility of a rhinencephalic participation. Many years have passed since the attention of Pollock and Davis (1930) was directed to this particular problem of the relationship of the integrity of the proprioceptive arc in relation to the tremor and rigidity of Parkinsonism. Their study provides a useful reference to earlier literature. Walshe (1924) had shown that intramuscular injections of procaine, so graded as to paralyze the afferent nerve fibers in the muscle and to leave the activity of motor nerve fibers unimpaired, abolishes or greatly reduces the rigidity of paralysis agitans. Pollock and Davis found that section of the posterior nerve roots in man abolished the rigidity and that the tremor was modified in rhythm, amplitude, and rate, with the appearance of irregularities in rate and amplitude. While these studies shed no conclusive light on the relationship between proprioceptive input and the tremor mechanism, it is probably also relevant that Halliday and Redfearn (1956), from studies of finger tremor in normal and tabetic subjects, were unable to exclude an instability of “servo” mechanisms in the reflex arc, and discuss the further possibility of tremor imposed on motoneurons by rhythmic bursts of impulses from higher sources, such as the reticular formation. As to a rhinencephalic participation in tremor mechanisms, the wellknown exacerbation of Parkinsonian tremor by emotional disturbances, and the natural occurrence of both in a single disease entity, will be discussed below.

111. Epileptic Phenomena in the Temporal lobe

The susceptibility of temporal lobe structures to seizure discharge, from mechanical, chemical, or electrical stimulation has long been known, and has been related in the hippocampus to an arrangement of its blood supply which appears to make it especially susceptible to anoxemia, carbon monoxide, and toxic agents (Meyer, 1956). A. ANIMALSTUDIES Much recent material relating to the initiation and spread of temporal lobe seizures is discussed in a symposium conducted by Gastaut (1954) and his colleagues and by Cadilhac (1955).


21

Cadilhac has drawn attention to propagation of hippocampal seizures in the cat and guinea pig to the mesencephalic reticular formation and to the medial geniculate body. Within the temporal lobe these secondary discharges are primarily in the piriform lobe and amygdala, and further spread involves the posterior ectosylvian gyrus, the middle ectosylvian gyrus and anterior and posterior sylvian gyri. Occasionally the seizure involves the whole of the lateral cortex of the hemisphere and the cerebellum. On the other hand, stimulation of the temporal cortex of the monkey, particularly in the anterior part of the temporal lobe, gives rise to after-discharge within the temporal cortex, frequently in the frontal granular cortex and face area of the motor cortex (Poggio et al., 1956). Subcortical propagation is consistently to the ipsilateral amygdala and hippocampus and, with further propagation, involves the septum, cingulum, and hypothalamus. The routes by which seizure activity, initiated in various p a r k of the temporal lobe, reaches the hypothalamus are by no means clear. Hypothalamic seizure activity can be elicited in the cat by stimulation of the amygdala, hippocampus, hippocampal gyrus, septum, and basal olfactory structures (E. W. Powell et al., 1957). Nevertheless, the hippocampal gyrus appears to bypass the hippocampus in its projection to the hypothalamus, a surprising finding if the Papez circuit does indeed provide a ready pathway for activity from the hippocampal gyrus to pass via the hippocampus and fornix to the hypothalamus. Green and Shimamoto (1953) have sectioned the fornix in the cat without interfering with the propagation of hippocampal seizures through the commissural fibers of the psalterium to the temporal lobe and other parts of the cerebrum, and they emphasize that the thalamus and hypothalamus are not essential to this propagation. E. W. Powell and associates (1957) found the ventromedial hypothalamic nucleus a focal area in receiving seizure discharges from basal temporal structures and septum. This nucleus receives projections from the amygdala in the monkey (Adey and Meyer, 1952a; Adey et al., 1958a). Powell et al. observed that stimulation at the same parameters of the ventral and dorsal portions of the hippocampus affect different parts of the hypothalamus in the cat. Thus, the ventral hippocampus projects to the ventromedial hypothalamic nucleus and mammillary body, whereas the dorsal hippocampus projects only to the mammillary body with an overall response smaller than from the ventral hippocampus. There appear to be important regional differences in the projections of dorsal and ventral hippocampal zones, a t least to the entorhinal area (Adey et al., 1957b). Projections from the caudal third of the rat’s hippocampus to the periventricular zone of the hypothalamus, and

22

W. R. ADEY

especially to the arcuate nucleus, have been described (Nauta, 1956). Powell et al. noted that amygdaloid seizures propagate rapidly into the ventromedial hypothalamic nucleus and appear to reach the mammillary body indirectly. These authors found the temporal and occipital lobes more effective in initiating hypothalamic epileptic discharges than other areas of the cortex. Experimental hippocampal seizures in animals have been induced by electrolytic lesions, alumina cream implants, and electrical stimulation. The characteristics of these discharges have been extensively described (Andy and Akert, 1955; Green and Adey, 1956; Youmans, 1956; Green et al., 1957a). Andy and Akert have drawn attention to the absence of altered functions in motor and sensory systems as long as the discharge is confined to the hippocampus, as shown by adequate and appropriate reactions to visual, auditory, and nociceptive stimuli. Spontaneous behavior may cease when there is spread of the discharge t o other subcortical and cortical structures. Bilateral seriatim removal of the temporal poles in the monkey leads to paroxysmal discharges in the EEG record (Kennard, 1957), with a marked increase in clinical seizures. Repetitive stimulation of medial temporal structures, including the amygdaloid region in the primates leads to an after-discharge from contralateral medial temporal structures, and is associated with ipsilateral movements involving face, jaw, and head, and with salivation, lacrimation, and tachycardia. Consciousness is altered (Baldwin et al., 1957). Epileptic seizures following electrolytic hippocampal lesions in cats seem to offer important evidence relating to the pathogenesis of psychomotor epilepsy in man, as described below. Green and co-workers (1957a) have described seizures in a series of such animals, and have noted that the temporal lobe epilepsy occurred chiefly in the first two postoperative weeks. Some seizures could be triggered by repeated clicks, a flashing light, or tactile stimulation. Since a t other times sensory stimulation was ineffective, they consider that an element of surprise might be important. The march of the seizure included a prodromal phase, with sudden head turning, staring, crouching, and peculiar arrested postures, before the onset of salivation and gagging, leading to clonic facial and limb movements. The brains of these animals showed extensive hippocampal lesions, strongly suggestive of damaged blood supply. The susceptibility of the hippocampus to damage from anoxia, carbon monoxide, and poisons (Meyer, 1956), attributable a t least partly to its precarious blood supply, has suggested an investigation of the effects of such simple manipulation of its environment as a raised carbon dioxide level. Dunlop (1957a, b) has studied the effects on excitability

23 of inhalation of carbon dioxide/oxygen mixtures containing 10 to 33% carbon dioxide. These mixtures exerted a differentially greater effect on conduction in the fornix-hippocampal pathway than in a nonspecific thalamocortical path. Moreover, these gas mixtures were capable of interrupting electrically induced hippocampal seizures, which recurred when administration of the carbon dioxide ceased. Brief reference may be made here to seizures initiated in parts of the limbic system lying outside the temporal lobe. Andy and Chinn (1957) have examined spontaneous and seizure activity in the cingulate gyrus of the cat. Spontaneous activity in this cortex is like that in the rest of the cortex, and differs from that in the hippocampus. “Spindle” trains, at 12 to 13 per second, associated with sleep appear very readily in the cingulate gyrus and may or may not be synchronous with those appearing in the cortical regions. Sleep spindles appearing during a cingulate gyrus after-discharge remain unaltered, whereas they disappear during a hippocampal after-discharge. The EEG pattern of after-discharge shows low-frequency waves and bursts a t 20 to 30 per second. This after-discharge is shorter than that from hippocampal stimulation. Unlike hippocampal stimulation, cingulate stimulation does not lead to general cortical activation. Cingulate seizures invariably propagate contralaterally. Threshold seizures induced in the hippocampus pass readily to the cingulate gyrus, but not in the converse direction. THE RHINENCEPHALON AND BEHAVIOR DISORDERS

1. Genesis and Intrahippocampal Propagation of Seizures

Slowly moving waves of D C potential, traveling longitudinally around the arch of the hippocampus a t speeds of 0.5 to 1.0 cm per minute, have been observed in the postictal phase by Liberson (1955), who suggests that the slow propagation involves an extrasynaptic mechanism of spread. It is possible that dendritic mechanisms may be involved. Using an ingenious technique to display vector magnitudes in horizontal and vertical recording dipoles during initiation and propagation of a hippocampal seizure, Green and Naquet (1957) and von Euler et a2. (1958) have examined the role of the hippocampal dendrites in evoked responses and after-discharges, Their results seem to show that following single volleys to the dorsal fornix, the distal parts of the pyramidal dendrites are not actively depolarized, but massive depolarization of the dendritic layer occurs after repetitive stimulation, and is apparently associated with the mechanism of seizure discharges t o which the hippocampus is particularly susceptible. The vector record a t the onset of a seizure shows that electrical discharges follow no orderly time sequence, but appear quite random. As the discharge progresses, the vec-

24

W. R. ADEY

tor record may become quite regular, showing that a high degree of synchronization has occurred between adjacent regions a few millimeters apart. Vector records obtained by von Euler, Green, and Ricci in two perpendicular planes indicate that a seizure spike may often sweep slowly as a wave of activity over the hippocampus in an anteroposterior direction with a velocity of about 10 cm per second. These peculiar horizontal movements do not accord with any known anatomical paths and are not related to any temporal features of the ordinary evoked potential. They conclude that when pathological electric activity occurs, a new kind of transmission is initiated in the dendritic layer, which a t this time is the region of inward current as seen in DC records. Seen in broad perspective, it is becoming increasingly apparent that many functional aspects of the rhinencephalon, including both its relationship to initiation and propagation of seizures, as well as its presumed role in such functions as emotional activity, arousal, and memory, probably resides in its complex interrelations with the diencephalon and brain stem, rather than in any intrinsic powers of integration that it may possess. This concept gains support from the effects of brain stem reticular stimulation in the monkey (Proctor et aZ., 1957), and is indeed the substance of many of the clinical studies cited below. Proctor and his colleagues observed changes in consciousness from stimulation in the mesencephalic tegmentum, red nucleus, and various areas of the reticular formation which were not dissimilar to seizures in humans with petit ma1 or psychomotor epilepsy. These disturbances in consciousness interfered with the monkey’s ability to recognize an unmatched object from a group of three objects. 2. Neuropharmacological Studies

Neuropharmacological studies in experimental temporal lobe epilepsy support concepts of the importance of rhinencephalic interrelations with the diencephalon in the manifestation of seizure discharges. Gangloff and Monnier (1957) have examined the action of anticonvulsant drugs by electrical stimulation of the cortex, diencephalon, and rhinencephalon (dorsal hippocampus) of the unanesthetized rabbit. Diphenyl hydanotoin acts on the diencephalon only, decreasing its excitability. Phenurone lowers the excitability of the diencephalon and cortex, and modifies the rhinencephalic after-discharge pattern, with a reduction in the number of spikes and lengthening of the slow wave components. Tridione depresses cortical and rhinencephalic excitability. Phenobarbital lowers the excitability of the rhinencephalon and diencephalon, but increases neocortical excitability. They conclude that temporal lobe epilepsy in-


25

volves both rhinencephalic and diencephalic factors, which are more susceptible to phenurone than the other drugs studied. I n scalp and cortical EEG records in patients with temporal lobe epilepsy, light sleep induced with Evipal (Kajtor et al., 1957) produced sinusoidal slow waves in the hippocampus, more prominent during waking. Hippocampal spike discharges were intensely activated by light sleep and during the period of waking up. It is of particular interest that awakening by strong peripheral stimulation was more effective than spontaneous lightening of anesthesia in evoking hippocampal spike discharge. These investigators observed an inverse relationship between sleep activation of hippocampal discharges and of temporal neocortical spikes in a case with hippocampal sclerosis.

B. HUMANSTUDIES IN TEMPORAL LOBEEPILEPSY Valuable reviews of the development of our present concepts in this field have been contributed by de Jong (1957) and Stevens (1957). Daly (1958) emphasizes that Hughlings Jackson’s concepts concerning uncinate fits were based partly on certain clinical observations, and also on the experiments of Ferrier, which suggested that smell and taste are localized in the uncus. However, Jackson (1899), in reviewing “the symptomatology of slight epileptic fits supposed to depend on discharge lesions of the uncinate gyrus,” adduces no direct evidence that the cortical centers for smell and taste adjoin one another. Magnus and coworkers (1952) and Penfield and Faulk (1955) have proposed that the insula and adjacent parts of the temporal lobe and uncus form a center for Edinger’s “oral sense,” and that epileptic discharge here may evoke gustatory experiences, oral and nasal sensations, salivation, and mastication. Psychological factors in initiation and inhibition of temporal lobe seizures have been investigated (Liberson, 1955; Rodin et al., 1955), with evidence that they may be inhibited by thinking about an unpleasant olfactory stimulus or a visual stimulus (Efron, 1957). Laughter occurring during seizures (so-called gelastic epilepsy) has been described in association with bitemporal EEG foci, and also in a case with an occipitotemporal tumor (Daly and Mulder, 1957). I n a further series of eleven such cases, Druckman and Chao (1957) have suggested the hypothalamus, rather than the temporal lobe, as the site of origin of these seizures. The differences in these considered opinions may support a concept of the essential interdependence of the rhinencephalic cortex and the diencephalon in the mediation of these seizures. More enduring psychological changes have been observed in the interictal period in patients with EEG abnormalities in the temporal

26

W. R. ADEY

lobe. In a series of patients with temporal spikes, the most frequent personality feature noted by Ervin et al. (1955) was an absence of stable character structure, with a high incidence of visceral, perceptual, aff ective, and ideational disorders. Weil (1955) has described paroxysmal depressive reactions lasting for periods up to several weeks in six cases with uncinate attacks and temporal lobe automatism. The episodes occurred before, during, or following the seizures, with a sudden onset, and were accompanied by olfactory hallucinations. They were refractory to psychotherapy. Gibbs (1958) has examined such factors as sleep and the age of the patient in detecting abnormal electrical activity in the temporal region and its relationship to abnormalities of behavior. He concludes that not only does sleep exacerbate the amount of subclinical seizure activity, but that most of the electrical ctbnormalities can be recorded only in sleep, with sleeping records almost essential to making a clear distinction between a mid-temporal and an anterior temporal lobe focus. H e finds that the type and locus of the EEG abnormality are largely a function of age. With increasing age, a diffuse disorder tends to disappear or become focal, and focal disorders tend to disappear, or t o move to more anteriorly located cortical areas. The mid-temporal focus is characteristically a childhood focus in his cases, whereas the anterior temporal focus characterizes adult abnormalities. The histopathology of hippocampal sclerosis, as seen in temporal lobes removed at operation, has been the subject of a continuing study by Meyer and Falconer and their associates, recently summarized by Meyer (1956). Within the hippocampus itself, loss of cells in the SSmmersector was a common finding. The hippocampal gyrus, hippocampus, uncus, or amygdala were involved in the majority of cases. The etiology was complex, with anoxia, vascular disorders secondary to trauma, meningoencephalitis, and tumor variously implicated. However, Meyer does not support the view that local anoxia at birth may have the exclusive significance initially attached to it by Penfield and Jasper (1954). Meyer points out that vascular accidents and other anoxic incidents in the first days of infancy, as in a series of epileptic crises during fevers, gastrointestinal infections, and allergic incidents, play an equal role. Thus, the autopsy findings in a child dying at the age of nine years of Still’s disease with a history of epileptic attacks beginning a year before death indicated severe lesions in the first temporal convolution and the medial temporal area of not more than a year’s duration. Whereas Penfield and his colleagues would regard mechanical herniation through the tentorium as a factor of prime importance in relation to birth accidents, Meyer considers that it can play a major part in disease


27

processes arising later in life, such as cerebral edema occurring in severe epileptic states and during other anoxic attacks. Support for the view that hippocampal sclerosis arises in the early years of life is given by Corsellis (1957) who found sclerosis in 8% of a series of cases without chronic brain disease, but considered this a terminal incident in all cases, whereas in the adult cases of cryptogenic epilepsy, the appearances were constantly those of a lesion of considerable duration, supporting the view that they arose in the early years of life. Gastaut (1953), on the other hand, has concluded that temporal lobe epilepsy is an affection of late adolescence, possibly arising as a sequel to encephalitis.

IV. Behavior Studies Following Rhinencephalic Stimulation and Ablation

Bilateral temporal lobectomy in the monkey leads to the well-known syndrome of Kliiver and Bucy (1939), characterized by a “psychic blindness,” or inability to comprehend the significance of environmental objects, tameness, strong oral tendencies, and hypersexuality. Many aspects of this picture have been reproduced in bilateral human temporal lobectomy (Terzian and Ore, 1955), with psychic disturbances far more severe than those following surgical interference with the frontal lobes.

A. CORTICAL ABLATION STUDIES The component aspects of the Kluver-Bucy Syndrome have been reproduced by more limited resections of different basal temporal structures. Indeed, the full syndome has been described in the monkey following seriatim resections limited to the temporal poles (Kennard, 1957). Kennard (1955-1956) has examined the effects on the temporal lobe syndrome of lesions elsewhere in the cerebral cortex of monkeys, including combined seriatim lesions of frontal association areas 9-12, temporal association areas, and anterior cingulate cortex. She concludes that the cingulate cortex and medial portion of the temporal pole should be considered as a functional unit directly related to more recently acquired frontal and temporal association areas. Certain aspects of the functional capacities and interrelations of the rhinencephalic and adjoining neocortical areas in visual discrimination in the monkey have been investigated by Mishkin, Pribram, and their colleagues (Mishkin, 1954; Mishkin and Hall, 1955; Mishkin and Pribram, 1954; Pribram and Barry, 1956; Pribram and Mishkin, 1955). In comparing the effects of lesions of the ventral temporal surface, including the inferior temporal, fusifom and posterior hippocampal gyri, with

28

W. R. ADEY

the results of resection of the hippocampus, they found that ventral temporal resections produced markedly greater impairment of visual discrimination than did control operations. The performance defect varied directly with the difficulty of discrimination, irrespective of the order of presentation of the discriminanda. Ventral lesions impaired the ability to discriminate visual size, and they considered that the decrement in performance was related to the situation determining the response, impairment occurring in tasks which differed from others previously learned only in the manner in which the identical discriminanda were presented. Since other data which they present also indicate a deficit following ventral resection merely by gradually reducing the differences in size between the stimuli, they conclude that no selective relationship between the impairment of visual discrimination and either of these two classes of variables is established by their data. Furthermore, their results may bring into question the usefulness of the distinction between “agnosia,” which might account for these findings, and “acuity loss,” which would explain the results on varying the size of the discriminands. A lack of visual comprehension, amounting to an agnosia, may explain this performance deficit, since changes in behavior following bilateral temporal pole damage (Kennard, 1957) may involve total behavior, with a blunting or alteration in responses to environmental stimuli. This problem has been extensively examined by Fuller and coworkers (1957) in dogs with pyriform-amygdala-hippocampallesions. The effect of the operation is to make the dog less responsive to stimuli in general, though not incapable of responding appropriately t o stimuli when they are persistent enough to elicit a response. They found no evidence that affective responses were more intense postoperatively. They found, furthermore, that contradictory results may be obtained depending on the circumstances in which the behavior is observed. Dogs are less timid postoperatively in reacting toward the handler. When reacting toward other dogs in a competitive feeding situation, however, they are no longer dominant but noncompetitive, which may be interpreted as more timid. Our findings in the baboon with ventral temporal resections involving the entorhinal area, and partly destroying the hippocampus but largely sparing the amygdala on one or both sides, support this point of view (Adey, 1958). These animals responded appropriately in an avoidance situation with markedly reduced emotional responses in comparison with normal animals. The dogs used by Fuller et al. showed gross impairment in a discrimination test, and the authors consider this as consistent with the opinion that there is a focus for visually guided behavior on the inferior aspect of the temporal lobe. Avoidance thresholds in the monkey following ventral rhinencephalic


29

lesions have been examined by Weiskrantz and Wilson (1958). Using three groups of animals, with orbitoinsulotemporal, amygdaloid, and control inferotemporal lesions, they found that the orbitoinsulotemporal groups showed less avoidance than the amygdaloid group, both in the first postoperative avoidance threshold and in terms of a general index of response and shock rates. Bilateral transection of the dorsal hippocampus near its rostral tip markedly retards conditioning of avoidance responses in the rat (Thomas and Otis, 1958a). Similar lesions produced significant defects in maze learning, in terms of number of trials to criterion, number of errors, and total time of learning (Thomas and Otis, 1958b). They suggest that this impaired performance may result from interference with mechanisms of secondary reinforcement. The aberrant sexual behavior following lesions of the amygdala and pyriform cortex in the cat have been studied by Green et al. (19.5710). They have examined the effects of administering sex hormones in normal and ablated animals, with evidence that the sex steroids may exert differing effects before and after puberty, with the possibility that changes in behavior following the first sexual experience may be correlated with subsequent reaction to hormones. Male hypersexual behavior was induced by lesions restricted to the pyriform cortex. Cataleptic changes were associated with lesions involving the rostral amygdala and basal ganglia. Hyperphagia followed lesions restricted to the basal and lateral parts of the amygdala. A similar hyperphagia has been described after amygdaloid lesions in the rat (Morgane and Kosman, 1957). The value of surgical intervention in appropriate human cases of focal unilateral temporal abnormalities is well recognized. Green et al. (1958) and Halstead (1958) have concluded that cognition was not markedly influenced by surgery. They found that although affective behavior was severely abnormal in all patients prior t o surgery, pcstoperative improvement in affect was markedly greater than the improvement in cognition. Intelligence as measured with the Wechsler-Bellevue rating is not permanently affected, although there is a deficit on verbal subtests in those with a left temporal excision during their period of postoperative dysphasia (Milner, 1958). There is evidence that the left temporal lobe contributes to the understanding and retention of verbally expressed ideas, whereas the right temporal lobe aids in rapid visual identification. Milner (1958) suggests that the hippocampus and hippocampal gyrus (either separately or together) play a crucial role in the retention of new experience, although the findings cited above from learning studies in animals do not unequivocally support this hypothesis. Behavior changes following ablation of areas 23 and 24 of the cingulate cortex in cat and monkey have been studied (Pechtel et al., 1958).

30

W. R. ADEY

Following operation, these animals displayed a minimal amnesia for previous learning, but showed an increased incidence of error and impaired precision in acquiring new skills. There was a tendency to diffuse and precipitate activity, and moderately increased aggressiveness toward other animals and man, with a lowered threshold of startle and fear, exacerbated by isolation. Maternal preemption of food a t the expense of a nursing baby was noted. There was minimal relief of phobic, regressive, or socially maladaptive behavior, but greater amenability to retraining. B. CORTICAL STIMULATION STUDIES Stimulation of the cat’s hippocampus has been studied in relation to acquisition and extinction of an instrumental response by Correll (1957), who concludes that stimulation during learning has no significant effect upon response speed, but results in a constant decrement in running speed. Stimulation during training has no effect on the number of responses to extinction, but stimulation maintained through extinction results in a significantly greater number of responses to extinction. Stimulation of the region of the dentate fascia and the subiculum under these conditions does not elicit a conditionable emotional response. Stimulation of deep cerebral structures in man, including the less accessible areas of the temporal cortex, has proceeded apace in recent years. Changes in memory function produced by electrical stimulation of the human temporal lobe are reported by Bickford et al. (1958). Electrical stimulation of the periamygdaloid area reproduces some of the features of temporal lobe epilepsy, although aggressive behavior is markedly absent. The responses elicited include fear, fright, visual hallucinations, depersonalization, and behavior suggestive of being startled, with an absence of recall of many of these experiences during stimulation (Chapman, 1958) . Pupillary dilatation, increased heart rate, and raised blood pressure can be elicited a t lower voltages in the absence of behavior changes.

C. BEHAVIOR CHANGWFOLLOWING SUBCORTICAL LESIONS An increase in both emotional reactivity and startle responses has been reported following septal lesions in the rat by Brady and Nauta (1955). No such effects followed habenular lesions. The changes were transient and all animals reverted to normal within 60 days. Although the habenular lesions produced no effect upon acquisition and retention of a conditional emotional response, extinction of the emotional response was significantly more rapid in the animals with habenular lesions than those with septal damage. These findings of increased emotional re-

31 activity following septal lesions in the rat appear to contrast with the effects of similar but more extensive lesions in the cat, involving posterior septum, fornix, and anterior thalamus, investigated by Bond et al. (1957). These cats showed impaired consciousness, hypokinesia, muscular plasticity, sustained grasping, the “jump” reflex, piloerection, marked hypothermia, and relatively persistent hyperglycemia. Rage reactions were conspicuously absent. Blood electrolyte examination revealed no consistent changes. Although the authors could not correlate their results with anatomically discrete nuclear masses or their specific projections, the highest incidence of abnormalities was associated with a combination of posterior septal, fornical, and anterior thalamic destruction, in regions already discussed in relation to rhinencephalic-hypothalamic and mesencephalic-reticulothalamocortical projections (Adey et al., 1958b). Subcortical mechanisms in the “searching” or “attention” response elicited by prefrontal cortical stimulation in unanesthetized cats have been examined by Jansen and associates (1955-1956). Stimulation of the medial prefrontal cortical region in freely moving cats resulted in a typical “attention” response, which was abolished by lesions in the anterior basal part of the ipsilateral internal capsule, or by bilateral thalamic lesions, involving the nucleus centralis medialis, nucleus paracentralis, and centralis lateralis. However, the searching response persisted after separate bilateral destruction of the cingulum, fornices, centrum medianum or dorsomedial thalamic nucleus, the habenulae, and striae medullares thalami. Also without effect were fairly extensive lesions of the anterior thalamic nuclei, caudate nuclei, hypothalamus, and mesencephalic tegmentum. Nature’s experiments in the realms of the degenerative neural diseases are rarely sufficiently controlled to afford detailed evidence of physiological systems or anatomical pathways. With the reservations that would seem inevitable in the interpretation of such material, it is, nevertheless, of great interest to note the recent findings of Gajdusek and Zigas (1957) and Zigas and Gajdusek (1957) in “kuru,” a degenerative disease among the native population of New Guinea. This disease appears germane to our discussions here, since it presents features of both Parkinsonism and emotional disorders, and might, therefore, be expected to shed light on common areas of involvement in the rhinencephalon, the basal ganglia, and the brain stem. The possible significance of such interconnections is discussed above. The first sympton in kuru is a locomotor ataxia, with awkward swaying of feet and a weaving gait. The accompanying tremor is usually irregular, Speech and intelligence are normal in the early stages of the THD RHINENCEPHALON AND BEHAVIOR DISORDERS

32

W. R. ADEX

disease, but dysarthria appears between the second and fourth months. As the disease progresses, but still in the early months of the illness, the patients become markedly emotional. Inappropriate excessive laughter and slowly relaxed smiles or paroxysmal hilarity alternate at times with moods of depression, and occasionally with moderately belligerent and aggressive behavior, This moodiness, more usually that of excessive euphoria than of depression, settles later into a gradual pattern of withdrawal, with a masklike facies resembling that of classic Parkinsonism. The flexed posture is characteristic of the disease. Preliminary histopathological examination (Zigas and Gajdusek, 1957) in kuru indicates a widespread degeneration, with the cerebellum and so-called extrapyramidal system most severely affected, and with further damage to the anterior horn cells, inferior olives, thalamus, and pontine nuclei. Sunderland (1958, personal communication) confirms these observations and points out that the cortex of the hippocampal gyrus, the hippocampus, island of Reil, and superior temporal gyrus are all normal, whereas the putamen and globus pallidus are the seat of a diffuse degeneration. The caudate nucleus and claustrum appear normal. Scattered degeneration is seen in tegmental areas at the level of the red nucleus, but the substantia nigra appears normal. Widespread changes occur in the Purkinje cells of the vermis. It is of particular interest that the emotional disorders seen here occur in the apparent absence of major disturbances in the hippocampal system. The association of emotional disorders with the Parkinsonian syndrome has not attracted a great deal of attention, although exacerbation of the tremor by emotional stresses is well known. Narabayashi et al. (1956) have drawn attention to the vegetative symptoms, such as hyperidrosis, salivary hypersecretion, and facial seborrhea, accompanying both senile and post-encephalitic forms of Parkinsonism, with the finding that these symptoms may be relieved by procaine-oil blocking of the globus pallidus. If, then, subcortical structures play a significant role in the mediation of affective responses, stimulation studies in chronic animals might be expected to reveal levels or zonal areas yielding maximal affective reactions. Hunsperger (1956) has reported the results of such studies in the cat. Affective reactions (hissing, laying back of ears, hunched back, pupillodilation, piloerection) were elicited from an unbroken field comprising portions of the central gray of the preoptic area and hypothalamus, and the central gray of the brain stem. Hunsperger describes two central responsive areas, one in the perifornical aone of the rostra1 hypothalamus, and the other in the middle portion of the midbrain central gray. Bilateral coagulation of the midbrain central flay led to a

THE) RHINENCEPHALON AND BEHAVIOR DISORDERS

33

temporary abolition of an affective defense reaction obtained by stimulation of the central zone of the hypothalamus. The reaction returned after 2 weeks. Bilateral coagulation of the posterior hypothalamus did not affect the reactions obtained by stimulation of the more rostrally and caudally situated portions of the active field. I n Hunsperger’s view, the central gray of the midbrain represents a relay station within the neuron paths for affective reactions in the brain stem. It is not merely traversed by the discharge pathway from the hypothalamic active field, but constitutes “a self-reliant acting system.” Self-stimulation studies by Olds (1956) and others in rats with electrodes implanted in subcortical structures have indicated very high press-rates (up to 5000 per hour) from paramedian areas of the hypothalamus and midbrain, with intermediate rates from rhinencephalic structures and still lower rates (at chance level) in classic sensory and motor pathways. Hunsperger’s finding of two key regions in subcortical structures for mediation of affective responses, one in the perifornical zone and the other in the periaqueductal gray of the midbrain, may be correlated with our studies indicating major projections from the entorhinal cortex to the central gray of the rostral midbrain (Adey et al., 1956,195713, 1958b; Adey, 1958). The hippocampus and entorhinal area may form part of a re-entrant pathway, receiving caudally directed activity via the fornix from reticular thalamic nuclei and the septum, and with activity passing from the entorhinal area to the midbrain tegmentum. Furthermore, the hippocampus appears to act as a “filter” on activity passing through it to the entorhinal area, with only a portion of the activity reaching it from the septum and thalamus ultimately passing to the entorhinal area. It may beesuggested that the rhinencephalon is set astride the brain stem at both the rostral diencephalic and midbrain levels. The central tegmental areas of the midbrain, the reticular thalamic, hypothalamic, and septa1 areas of the diencephalon from part of a “nonspecific” cortical activating system (ascending reticular system). Caudally directed activity from the entorhinal area may re-enter this activating system at the midbrain level, where it may further modify ascending streams of neural activity.

V. Neuropharmacological Studies of Rhinencephalic Functions

It is outside the scope of this review to discuss in detail the subject of psychotropic and psychotherapeutic drugs, a topic that has formed the basis of three recent symposia (Berger, 1957; Kety, 1957; Garattini and Ghetti, 1958), which have contributed a wealth of fascinating data.

34

W. R. ADEY

This account will deal primarily with the valuable additions to our knowledge of rhinencephalic functions made with the aid of the pharmacological tool. The susceptibility of rhinencephalic structures, and particularly the hippocampus, to seizure discharges and to damage by toxic agents and anoxia has been discussed above. Dunlop’s studies (1957a, b) have indicated that such simple manipulations of the hippocampal environment as the administration of raised carbon dioxide concentrations in the inspired air differentially depresses hippocampal excitability and can modify electrically induced hippocampal seizure discharges. Birchfield et al. (1958) have examined the alterations in blood gases during natural sleep and narcolepsy, with measurements of changes in arterial oxygen saturation, carbon dioxide tension, and p H before, during, and on awakening from sleep. I n normal subjects, retention of arterial carbon dioxide and a decrease in arterial oxygen content appeared with the earliest EEG changes correlated with drowsiness, and persisted through deep sleep. Narcoleptic patients, on the other hand, tended to show values for blood gases similar to those in normal subjects during sleep. These authors suggest that the respiratory function in man is directly dependent on the state of awareness, and that the mild hypercapnia and hypoxia observed in normal subjects while asleep and in narcoleptics while awake may be due to a decrease in afferent stimuli or a diminished sensitivity of the respiratory center. Inhalation of carbon dioxide was found by Otis and Thomas (1958) to have no effect on conditioned emotional responses in rats which had been made to “fear” the onset of a stimulus previously associated with painful shock. Inhalation of 10 and 30% carbon dioxide/oxygen mixtures, which produced profound autonomic reactions, including salivation, urination, and defecation, caused no discernible attenuation of the strength of the conditioned emotional response. Otis and Thomas point out that, since a comparable number of electroconvulsion treatments would have attenuated the conditioned emotional response, and since both electroconvulsion and carbon dioxide inhalation grossly stimulate the autonomic nervous system, it appears unlikely that stimulation of the autonomic nervous system would constitute the “common mechanism” proposed by Gellhorn to explain behavioral changes following various types of “shock” treatment. The effects of cingulumotomy, ablation of anterior cingulate cortex, and frontal lobectomy on the morphine withdrawal syndrome in monkeys have been investigated by Foltz and White (1957). Here, also, autonomic responses are a striking feature of the syndrome, with salivation, yawning, rhinorrhea, lacrimation, pilomotor activity, and diarrhea.

35 The modification of the effects of withdrawal was strikingly similar after all three frontal lobe lesions. Frontal lobotomy was slightly superior to cingulumotomy and to resection of the cingulate gyrus. The authors suggest that interruption of the cingulum, which is common to all three procedures, is the critical physiological lesion responsible for modification of the autonomic hyperactivity in morphine withdrawal. This bundle provides a pathway from the frontal lobe to the limbic system (Adey and Meyer, 1952b). Hallucinatory experiences are a common concomitant of irritable foci in the temporal lobe and, as Penfield and his colleagues have so amply demonstrated, can be evoked by electrical stimulation in certain cases. It is, therefore, of special interest that acetylcholine inhibitors, which are effective against the abnormal motor activity of Parkinsonism, have been shown by Pfeiffer et al. (1958) to be capable of producing a psychosis in normal persons, characterized by visual and auditory hallucinations. The relative infrequency of such hallucinations in Parkinsonian patients receiving therapeutic doses of these drugs adds interest to this observation. Although the role of acetylcholine as a central transmitter substance is still uncertain, Himwich and Rinaldi (1955-1956) have examined the possibility of using the brain stem reticular activating system for screening anti-Parkinsonian drugs, claiming that it is cholinergic in function, and is the site in the brain most sensitive to cholinergic and anticholinergic drugs. They emphasize that drugs which affect the alerting response also affect the caudally directed extrapyramidal control of motion. THE RHINENCEPHALON AND BEHABIOR DISORDERS

A. THEPOSSIBLE ROLEOF SEROTONIN AS A CEINTRAL TRANSMITTER STJBSTANC~ Attention has turned in recent years to the possibility that, in addition to acetylcholine, substances involved in protein metabolism may serve as transmitter substances, or, a t least, provide a background of tonic excitation or inhibition. Maclean et al. (1955-1956) have found that insulin convulsions, for 1 hour before administration of radioactive methionine, interfere with its uptake by the hippocampus. From interaction studies between reserpine, serotonin, and lysergic acid diethylamide (LSD), Brodie and his colleagues (Brodie et al., 1955; Pletscher e t al., 1955; Shore et al., 1955) suggested that serotonin (5-hydroxytryptamine) might function as a neurohumoral agent. Their hypothesis was based on the finding that reserpine potentiates the hypnotic effects of hexobarbital in mice, and that serotonin in large doses depresses mice and potentiates the action of hexobarbital in these animals. This potentiating action of serotonin was found to be antagonized by LSD. Follow-

36

W. R. ADEY

ing intravenous administration of reserpine, cerebral serotonin dropped by 15% in 30 minutes, and by 90% in 4 hours. The level remained low for 24 hours and returned to normal in 7 days. Low doses of reserpine phosphate produce spontaneous seizures and facilitate evoked seizures in the rhinencephalon without affecting the thalamocortical recruiting response and EEG arousal (Sigg and Schneider, 1957). The occurrence of cortical low-voltage fast activity with large doses of reserpine can be blocked by atropine, but atropine does not modify the rhinencephalic seizures evoked under these conditions. Two antiserotonins, BAS (the benzyl analog of serotonin) and BAB (the benzyl analog of bufotenin) have been studied by Rinaldi (1958) in their effects on spontaneous electric activity of the brains of man and rabbit. In the rabbit, BAS induced convulsive nonpropagated hippocampal discharges, while BAB produced continuous fast activity in the lenticular nucleus. In psychotic humans, a small proportion of subjects showed a decreased amplitude in the EEG and slowing of the dominant alpha activity with BAS. Examination of the serotonin content of the human brain (Costa and Aprison, 1958) indicates that the allocortex has a higher concentration than the isocortex. The greatest amount is in the mesencephalon (substantia nigra and red nucleus). Relatively large amounts also occur in the hypothalamus, and the lowest concentrations are in the cerebellum and the isocortex. Attempts to define more clearly the role of serotonin in cerebral metabolic pathways have not been successful, and current views on its actions in cerebral metabolism have been reviewed by Woolley (1958). It is his view that it plays a part in maintaining normal mental processes, and that interference with its action in the brain leads to mental disorders and neural dysfunction. Woolley points out that serotonin can be synthesized in the brain from 5-hydroxytryptophan1 and that it does not enter the brain readily from the periphery. Many drugs which cause human mental disturbances can be shown to be antimetabolites of serotonin, and these drugs may antagonize or mimic the contraction of the oligodendroglia produced by serotonin. A major difficulty in investigations of the intracerebral activity of substances such as serotonin (and particularly of y-aminobutyric acid, to be discussed below) lies in their failure to enter the cerebrum from the periphery, despite their indubitable synthesis within the cerebral cells. Until it becomes possible t o investigate the exceedingly subtle and dynamic aspects of their activity within the microcosm of an individual neuron, our knowledge of these substances is likely to remain largely inferential.


37

B. EVIDENCE CONCERNINQ THE EFFECT OF STRIATAL LESIONS ON THB ACTIONOF RESERPINE The subtlety of the action of drugs such as reserpine is exemplified by behavior studies which my colleague, Dr. Dunlop, and I have undertaken in a series of monkeys trained in an avoidance situation and then subjected to lesions in the corpus striatum. The results presented here are taken from single pre- and postoperative drug trials in an immature female monkey, (Macacus cynornolgus), but may be justified by reason of the high consistency in successive trials (four preoperative and three postoperative). The avoidance procedure involved conditioning to a light stimulus in a shuttle box. The light was presented for 2 sec and was repeated a t 15-sec intervals, if necessary, up to a maximum of 3 exposures. A painful shock was then administered t o the feet for one-tenth second if movement to the adjacent compartment had not occurred. The animal reached a 100% criterion (movement with the first light) in 200 trials over 4 days. Histological examination of the bilateral lesion prepared by DC coagulation indicated damage to the dorsal part of the globus pallidus and adjacent putamen on the right side, without involvement of the internal capsuIe. On the left side, the lesion has extensively involved the dorsal part of the globus pallidus and putamen, and has extended slightly into the internal capsule. The anteroposterior extent of the lesion on each side was about 3 mm at its maximum extent (Fig. 6 ) . Postoperatively there was a transient paresis of both forelimbs, which disappeared in 24 hours. At no time was a tremor discerned except during the administration of reserpine. The first postoperative trial in the avoidance situation was made 38 days after operation, and indicated a high retention of the learned response (98% movement with the first light). A comparison of the effects of chlorpromazine (1 mg per kilogram) on the pre- and postoperative behavior did not reveal any differential effects which might be attributed to the striatal lesion. There was a rapid fall in the response level to 4 to 12% with the first light within 1 to 2 hours, 80% recovery at 6 hours, and full recovery within 24 hours. With reserpine, however, there was a clear differential between preand postoperative performance with doses of 0.33 mg per kilogram (Fig. 6 ) . It will be seen from the graphs that the postoperative tests show an earlier and more sustained decrease in correct performance with the first light, and that during the fourth and fifth hours, there was an absence of even an escape response to shock in as many as 60% of the trials, as compared with 16% in the preoperative tests. While it would be tempting from these data to suggest that the striatal lesion

38

W. R. ADEY

MONKEY

0

I

CY4

2

1 4 HOURS

OPERATION II WR.'58

5 8 7 AFTER lNJfCTION

8

9

50

RESERPINE 0.33 ma/ kg POSTOPERATIVE (MAY)

40 0

:" a

10

24

.. ..

LLLS

/*

--d

L

a

LLL LL

.L

20 10

0 0

1

2

1 4 5 HOURS AFTCR

8

7

8

24

twccrm

Fia. 6. Effects of reserpine on avoidance behavior in the monkey before and after a bilateral lesion in the striatum. Extent of the lesion is indicated in upper diagram. Details of avoidance testing are described in the text. Graphs indicate performance a t hourly intervals after administration of reserpine, with the number of correct trials out of 60 for each hour. Key to graphic symbols: L, avoidance response on first light; LL, avoidance response on second light; LLL, avoidance response on third light; LLLS with arrow-superscript, escape response after three lights and shock; LLLS with bar-superscript, no escape response after three lights and shock. Abbreviations in lesion figure: CN, caudate nucleus; GP, globus pallidus; IC, internal capsule; OP, optic tract, PUT, putamen; TH, thalamus.

THPJ BHINENCEPHALON AND BEHAVIOR DISORDERS

39

may have potentiated the capacity of reserpine to inhibit an adequate response in an avoidance situation, it would seem more appropriate at this stage to view this type of evidence as a pointer to future research, which may take account of other explanations in an exceedingly complex situation. Weiskrantz (1958) has proposed that reserpine may act by blocking incoming sensory information, although Killam (1958) was unable to find altered thresholds for EEG or behavioral arousal elicited by stimulation of either the reticular formation or the diffuse thalamic projection system. Moreover, John et aZ. (1958) do not accept the interpretation that reserpine blocks the conditioned avoidance response by interference with sensory perception, with motivation to perform, or with motor coordination. They propose that its action may be attributed to an interference with the specific conditioned association between the stimulus and a directed evasion response. This they regarded as an interference with the symbolic qualities of learned associations. Human studies with chronically implanted intracranial electrodes (Monroe et al., 1957) have provided certain EEG correlations with psychotic behavior induced by d-LSD-25, Z-LSD-25, and mescaline. Increased beta activity in cortical and subcortical records seemed associated with anxiety. Paroxysmal activity induced in the hippocampus, amygdala, and septa1 regions was associated with the overt expression of psychotic behavior, whereas the spread of such paroxysmal activity until it became generalized in the cortex seemed to interfere with the full expression of psychotic behavior. Chlorpromaeine effectively blocked the behavioral response and also abolished or minimized the appearance of low-amplitude fast activity in cortical records as well as the paroxysmal subcortical discharges. Evarts (1958) points out that the most striking feature of experiments with LSD in unanesthetized cats is the essentially negative character of the results obtained with all but very large doses. In comparing the dose of LSD required to produce consistent electrophysiological effects with the amounts necessary to produce consistent behavioral effects in man (less than 1.0 pg per kilogram) , he found the ratio to be enormous. Also, in comparing the ratio of the dose necessary to produce consistent behavior changes in animals with those in man, the ratio was likewise enormous.

C. THEMODEOF ACTIONOF 7-AMINOBUTYRIC ACID (GABA) IN RHINENCEPHALIC MECHANISMS The unique occurrence of GABA in the brain of the mouse was first reported by Roberts and Frankel (1950) and its possible importance in cerebral metabolism has been reviewed by Roberts and Baxter (1958).

40

W. R. ADEY

It has become apparent that a n intimate relationship exists between cellular oxidations and the cerebral metabolism of GABA and glutamic acid, GABA is formed in the brain from glutamic acid by the action of glutamic acid decarboxylase. GABA transaminase catalyzes the transamination of GABA with a-ketoglutarate. Since a-ketoglutarate, glutamic acid, glutamic acid decarboxylase, GABA, and GABA transaminase are found in the central nervous system, Roberts points out that there exists the possibility of a shunt around the a-ketoglutarate oxidase system by which a-ketoglutarate could be withdrawn as glutamic acid (by the donation of the amino group of GABA), and the carbon chain of the GABA could enter the tricarboxylic acid cycle a t the succinate level. The synthesis of GABA from glutamic acid requires the presence of pyridoxal phosphate as a coenzyme t o the activity of glutamic acid decarboxylase. There is evidence that GABA may excite a tonic inhibitory influence on many cerebral neurons, and that convulsions will occur if it is removed from the cells. Killam and Pfeiffer and their co-workers (Killam, 1957; Killam and Bain, 1957; Pfeiffer et al., 1956) have shown that experimental seizures in man and animals occur from an acute pyridoxine deficiency induced by hydrazides, such as thiosemicarbazide. These seizures are not prevented by diphenylhydantoin or pentobarbital, but are prevented by pyridoxine, acetone, atrolactamide, phenurone, and trimethadione. GABA appears to have many characters in common with the inhibitory factor I, isolated from beef brain by Florey (1956) and the relationships have been reviewed by Elliott (1958) . The action of GABA in rhinencephalic seizures in the cat has been examined by Dasgupta et al. (1958). GABA was applied topically to the cortex and to the hippocampus, and also intraventricularly, either before or during seizures induced by electrical stimulation of the cortex, or stimulation of the hippocampus, or by depletion of intrinsic brain levels of GABA by administration of thiosemicarbazide. While pretreatment with GABA intraventricularly or topically prevented or reduced the duration of hippocampal seizures, application during such a seizure prolonged its duration. Cortical application of GABA was not effective in shortening seizures elicited by cortical stimulation. Hippocampal seizures induced by administration of thiosemicarbazide were inhibited by topical application of GABA and by intravenous pyridoxal. VI. Summary and Conclusions

Obviously, in a review of this nature, much of the material discussed skirts the problem of the relationships between brain and mind, and in


41

this field the reviewer, often with nai’vet6 and rarely with insight, will attempt generalizations where none exist and offer postulates where the quicksands of changing opinion and new vistas of knowledge all too rapidly consign to oblivion any but the most carefully constructed hypotheses. Nevertheless, it has become increasingly apparent that certain functions may be assigned to the deep structures of the temporal lobe, concerned with interpretative and symbolic aspects of peripheral stimuli, with a degree of assurance that scarcely seemed possible a decade ago. In Penfield’s view (1958), absence of the hippocampal gyrus and hippocampus on both sides makes any permanent recording of present experience impossible. Yet, we must tread warily for, as Penfield points out, we do not know whether “the ganglionic record of the stream of consciousness is located there,” or whether the hippocampal structures play an essential part in the laying down of the record elsewhere. The production of interpretive illusions by local epileptic activity or electrical activation suggests to Penfield that the perceptional cortex contains a mechanism which is employed in the interpretation of current experience. This interpretation would include comparison of the present with selected past experience. We have ventured in this review into the realm of willed movements and abnormalities of movement, observing that interreIations between the rhinencephalon and the deep centers of the thalamus and basal ganglia may play a part in such neural integration. Eccles (1953), on the basis of Sherrington’s conclusion that mind is not a form of energy, has evolved a theory as to how non-energy mind can act on matter, and looks to telepathy as supporting evidence. He uses the principle of uncertainty to show that a minute influence can act upon a synaptic junction and modify behavior. This point of view has been criticized by Lashley (1958) on the grounds that the principle of uncertainty is entirely irrelevant to the question of causal determination. In Lashley’s view it is a principle of unobservability and, as a basis for doctrines of will, it is in a class with the belief that the invisible face of the moon is made of green cheese. Lashley (1958) has drawn attention t o the role of the tonic background in the formation of associations essential to the learning process, stressing the view that the importance of repetition has been greatly exaggerated, and tbat an organization of associated neurons excited a t sub-threshold levels is an almost necessary consequence of the structure and known physiological properties of the cerebral cortex. The interplay of such systems might provide a mechanism for a large proportion of the selective activities of mind. As Jasper and Rasmussen (1958) have

42

W. R. ADEY

pointed out, it is particularly in the anterior deep temporal regions that the integration of body functions, visceral and somatic, is performed, and where complex intellectual functions important for conscious awareness of experience and rational responsiveness are achieved. ACKNOWLEDQMIONTS Certain studies presented here were carried out in collaboration with Dr. C. W. Dunlop in the Department of Anatomy, University of Melbourne. We wish to thank Professor Sydney Sunderland for his unfailing help and encouragement in this aspect of the work. REFERENCE8 Adey, W. R. (1956). Australasian Ann. Med. 5, 163. Adey, W. R. (1958). In “The Reticular Formation of the Brain” ( H. H. Jasper et al., eds.) Little, Brown, Boston, Massachusetts. Adey, W. R., and Meyer, M. (1952a). Brain 75, 358. Adey, W. R., and Meyer, M. (1952b). J . Anat. 86, 58. Adey, W. R., Merrillees, N. C. R., and Sunderland, S. (1956). Brain 79, 413. Adey, W. R., Segundo, J. P., and Livingston, R. B. (1957a). J. Neurophysiol. 20, 1. Adey, W. R., Sunderland, S,,and Dunlop, C. W. (1957b). EEG Clin. Neurophysiol. 9,309.

Adey, W. R., Rudolph, A. F., Hine, I., and Harritt, N. J. (1958a). J . Anat. 92, 219. Adey, W. R., Dunlop, C. W., and Sunderland, S. (1958b). J . Comp. Neurol. 110, 173.

Andy, 0. J., and Akert, K. (1955). J. Neuropathol. Exptl. Neurol. 14, 198. Andy, 0. J., and Chinn, R. McC. (1957). Neurology 7, 56. Angeleri, F., and Carreras, M. (1956). R i v . Neurobiol. 2, 645. Baldwin, M., Frost, L. L., and Wood, C. D. (1957). Neurology 7, 15. Berger, F. M. (1957). Ann. N . Y . Acad. Sci. 67, 671-894. Bickford, R. G., Mulder, D. W., Dodge, H. W., Svien, H. J., and Rome, H. P. (1958). Research Publa. Assoc. Research Nervous Mental Disease Sb, 227. Birchfield, R. I., Sieker, H. O., and Heyman, A. (1958). Neurology 8, 107. Blackstad, T. W. (1956). J . Comp. Neurol. 105, 417. Bodian, D. (1939). J . Comp. Neurol. 71, 259. Bodian, D. (1940). J . Comp. Neurol. 72, 207. Bond, D. D., Randt, C. T., Bidder, T. G., and Rowland, V. (1957). A.M.A. Arch. Neurol. Psychiat. 70, 143. Brady, J. V., and Nauta, W. J. H. (1955). J . Comp. and Physiol. Psychol. 48, 412. Bremer, F. (1935). Compt. rend. soc. biol. 118, 1235. Bremer, F. (1954). In “Brain Mechanisms and Comciousneea” (J. F. Delafresnaye, ed.) Blackwell, Oxford, England. Broca, P. (1878). R e v . Anthropol. 2, 385. Brodal, A. (1947). Brain 70, 179. Brodie, B. B., Pletscher, A., and Shore, P. A. (1958). Scknce 122, 968. Buchwald, N. A., and Ervin, F. R. (1957). EEG Clin. Neurophysiol. 9, 477. Bucy, P. C., and Kluver, H. (1956). J . Comp. Neurol. 103, 151. Cadilhac, J. (1955). “Hippocampe et Epilepsie.” DBhan, Montpellier, France. Cajal, 5. R. y (1911). “Histologie du systbme nerveux de l’homme et des vertBbr6s,” Vol. 2. A. Maloine, Paris, France. Carey, J. H. (1957). J . Comp. Neurol. 108, 67.


43

Carpenter, M. B., Glinsman, W., and Fabrega, H. (1958). Neurology 8, 352. Carreras, M., Macchi, G., Angeleri, F., and Urbani, M. (1954). Arch. intern. studi neurol. 1, 1. Carreras, M., Macchi, G., Angeleri, F., and Urbani, M. (1955). Boll. Soe. ital. biol. sper. 31, 1. Chapman, W. P. (1958). Research Publs. Assoc. Research Nervous Mental Disease 36, 258.

Cooper, I. S. and Bravo, G. J. (1958). Neurology 8, 344. Correll, R. E. (1957). J. Comp. and Physiol. Psychol. 50,624. Corsellis, J. A. N. (1957). Brain 80, 193. Costa, E., and Aprison, M. H. (1958). J. Nervous Mental Disease 126, 289. Courvoisier, S., Ducrot, R., and Julou, L. (1958). In “Psychotropic Drugs” (S.Garattini and V. Ghetti, eds.), pp. 373-391, Elsevier, Amsterdam, Holland. Cowan, W. M., and Powell, T. P. S. (1955). J. Neural. Neurosurg. Psychiat. 18, 267. Cragg, B. G., and Hamlyn, L. H. (1956). J. Anat. 90, 591. Cragg, B. G., and Hamlyn, L. H. (1957). J. Physiol. 135, 460. Daly, D. (1958). Neurology 8, 250. Daly, D. and Mulder, D. W. (1957). Neurology 7, 189. Dasgupta, S. R., Killam, E. K., and Killam, K . F. (1958). J. Pharmacol. Exptl. Therap. 122, 16a. de Jong, R. N. (1957). Neurology 7, 1. Dell, P. and Olsen, R. (1951). Compt. rend. SOC.biol. 145, 1088. Druckman, R. and Cheo, D. (1957). Neurology 7, 26. Dunlop, (3. W. (1957a). A m . J . Physiol. 190, 172. Dunlop, C. W. (195713). A m . J . Physiol. 191, 200. Dunlop, C. W. (1958). EEG Clin, Neurophysiol. 10, 297. Eccles, J. C. (1953). “The Neurophysiological Basis of Mind.” Oxford Univ. Press, London and New York. Efron, R. (1957). Brain 80, 251. Elliot Smith, G. (1910). Lancet i, 1-6; 147-153; 221-227. Elliot Smith, G. (1918). J. Anat. 63, 271. Elliott, K. A. C. (1958). Neurology 8, Suppl. 1, 98. Ervin, F., Epstein, A. W., King, H. E. (1955). A.M.A. Arch. Neurol. Psychiat. 74, 488. Evarts, E. V. (1958). Research Publs. Assoc. Nervous Mental Disease 36, 347. Florey, E. (1956). Can. J. Biochem. and Physiol. 34, 669. Flotz, E. L., and White, L. E. (1957). J . Neurosurg. 14, 655. French, J. D., Hernandez-PBon, R., and Livingston, R. B. (1955). J. Neurophysiol. 18, 74.

Fuller, J. L., Rosvold, H. E., and Pribram, K. H. (1957). J . Comp. and Physiol. Psychol. 50, 89. Gajdusek, D. C., and Zigas, V. (1957). New Engl. J . Med. 257, 947. Gangloff, H., and Monnier, M. (1957). EEG Clin, Neurophysiol. 9, 43. Garattini, S., and Ghetti, V., eds. (1958). “Psychotropic Drugs.” Elsevier, Amsterdam, Holland. Gastaut, H. (1953). Epilepsia 2, 29. Gastaut, H. (1954). In “Colloque sup les problhmes d’anatomie normale et pathologique poses par les dBcharges Cpileptiques (Marseille),” pp. 6-20. Les Editions Acta Medica Belgica, Brussels, Belgium. Gibbs, F. A. (1958). Research Publs. Assoc. Nervous Mental Disease 36, 278. Gerebtzoff, M. A. (1941-1942). J . belge neurol. psyehiat. 41-42, 199. Gloor, P. (1955a). EEG Clin. Neurophysiol. 7, 223.

44

W. R. ADEY

Gloor, P. (1955b). EEG Clin. Neurophysiol. 7, 243. Green, J. D., and Adey, W. R. (1956). EEG Clin. Neurophysiol. 8, 245. Green, J. D., and Arduini, A. A. (1954). J . Neurophysiol. 17, 533. Green, J. D., and Machne, X. (1955). Am. J . Physiol. 181, 219. Green, J. D., and Naquet, R. (1957). Extrait du volume publie B l’occasion du IV” Congrhs International d’Electroenc6phalographie et de Neurophysiologie clinique et de la VIII” RBunion de la Ligne Internationale contre 1’Epilepsie. pp. 225-249. (Paper presented a t Proc. Intern. Neurol. Congr., Brussels, 1967, published by Excerpta Medica, Amsterdam.) Green, J. D., and Shimamoto, T. (1953). A.M.A. Arch. Neurol. Psychiat. 70, 687. Green, J. D., Clemente, C. D., and de Groot, J. (1957a). A.M.A. Arch. Neurol. Psychiat. 78, 259. Green, J. D., Clemente, C. D., and de Groot, J. (1957b). J. Comp. Neurol. 108, 505. Green, J. R., Steelman, H. F., Duisberg, R. E. H., McGrath, W. B., and Wick, S. H. (1958). Research Publs. Assoc. Nervous Mental Disease 36, 295. Guillery, R. W. (1956). J. Anat. 90, 35. Halliday, A. M., and Redfearn, J. W. T. (1956). J . Physiol. 134, 600. Halstead, W. C. (1958). Research Publs. Assoc. Nervous Mental Disease 36, 478. Herrick, C. J. (1933). Proc. Natl. Acad. Sci. U.S. 19, 7. Himwich, H. E., and Rinaldi, F. (1965-1956). Y a k J. Biol. and Med. 28, 308. Hunsperger, R. W. (1956). Helv. Physiol. et Pharmacol. Acta. 14, 70. Jackson, J. Hughlings (1899). Lancet i, 79. Jansen, J., Jr., Andersen, P., and Kaada, B. R. (1955-1956). Yale J. Biol. and Med. 28, 331.

Jasper, H. H., and Rasmussen, T. (1958). Research Publs. Assoc. Nervous Mental Disease 36,316. John, E. R., Wenzel, B. M., and Tschirgi, R. D. (1968). Science 127, 25. Johnston, J. B. (1923). J . Comp. Neurol. 35, 337. Kajtor, F., Hullay, J., Farago, L., and Haberland, K. (1957). E E G Clin. Neurophysiol. 9,441.

Kennard, M. A. (1955-1956). Yale J. Biol. and Med. 28, 342. Kennard, M. A. (1957). Neurology 7, 404. Kety, 8. S., ed. (1957). Ann. N . Y . Acad. Sci. 66, 417. Killam, K. F. (1957). J. Pharmacol. Ezptl. Therap. 119, 263. Killam, K . F. (1958). I n “Psychotropic Drugs” (S.Garattini and V. Ghetti, eds.), pp. 244-251. Elsevier, Amsterdam, Holland. Killam, K. F., and Bain, J. A. (1957). J. Pharmacol. Exptl. Therap. 119, 255. Kluver, H., and Bucy, P. C. (1939). Arch. Neurol. Psychiat. 42, 979. Lashley, K. S. (1958). Research Pubb. Assoc. Nervous Mental Disease 36, 1. Laursen, A. M. (1955). J . Comp. Neurol. 102, 1. Liberson, W. T. (1955). Am. J. Psychiat. 112, 91 Lorente de N6, R. (1934). J . Psychol. and Neurol. 46, 113. Macchi, G., Angeleri, F., and Cameras, M. (1956). Atti SOC. italiana anat. XVII Convegno sociale, 1-7. Macchi, G., Carrerm, M., and Angeleri, F. (1955). Archiv. tal. anat. e embryol. 60, 413.

McLardy, T. (1955a) J. Comp. Neurol. 103, 305. McLardy, T. (1955b). J . Comp. Neurol. 103, 327. Maclean, P. D. (1955). Psychosomat. Med. 17,355. Maclean, P. D., Flanigan, S., Flynn, J. P., Kim, C., and Stevens, J. R. (19561956). Y a k J. Biol. and Med. 28, 380.


45

Magnus, O., Penfield, W., and Jasper, H. (1952). Acta Psychiat. Neurol. Scand. 27, 91. Magoun, H. W. (1952). Research Pubb. Assoc. Nervous Mental Disease 30, 480. Meyer, A. (1956). Acta Neurol. Psychiat. Belg. 1, 21. Meyer, A., and Beck, E. (1954). “Prefrontal Leucotomy and Related Operations,’’ The Seventeenth Henderson Trust Lectures. Oliver and Boyd, Edinburgh, Scotland. Milner, B. (1958). Research Publs. Assoc. Nervous Mental Disease 36, 244. Mishkin, M. (1954). J . Comp. and Physiol. Psychol. 47, 187. Mishkin, M., and Hall, M. (1955). J . Comp. and Physiol. Psychol. 48, 97. Mishkin, M., and Pribram, K. H. (1954). J . Comp. and Physiol. Psychol. 47, 14. Monroe, R. R., Heath, R. G., Mickle, W. A,, and Llewellyn, R . C. (1957). EEG Clin. Neurophysiol. 9, 623. Morgane, P. J., and Kosman, A. J. (1957). Nature 180, 598. Morin, F. (1950). J. Comp. Neurol. 92, 193. Moruzzi, G., and Magoun, H. W. (1949). EEG Clin. Neurophysiol. 1, 455. Narabayashi, H., Okuma, T., and Shikiba, S. (1956). A.M.A. Arch. Neurol. Psychiat. 75, 36.

Nashold, B. S., Hanbery, J., and Olszewski, J. (1955). EZG Clin. N e u r o p h y ~ o l7. , 609.

Nauta, W.J. H. (1956). J. Comp. Neurol. 104, 247. Olds, J. (1956). J. Comp. and Physiol. Psychol. 49,281. Otis, L. S., and Thomas, G. J. (1958). J . Comp. and Physiol. Psychol. 51, 205. Papez, J. W. (1937). Arch. Neurol. Psychiat. 38, 725. Pechtel, C., McAvoy, T., Levitt, M., Kling, A., and Massermann, J. H. (1958). J. Nervous Mental Disease 26, 148. Penfield, W. (1958). Research Publs. Assoc. Nervous Mental Disease 36, 210. Penfield, W., and Faulk, M. E. (1955). Brain 78,445. Penfield, W., and Jasper, H. (1954). “Epilepsy and the Functional Anatomy of the Human Brain.” Churchill, London, England. Pfeiffer, C . C., Jenney, E. H., and Marshall, W. H. (1956). EEG Clin. Neurophysiol. 8, 307.

Pfeiffer, C. C., Murphree, H. B., Jenney, E. H., Robertson, M. G., Randall, A. H., and Bryan, L. (1958). Federation Proc. 17, 403. Pletscher, A., Shore, P. A., and Brodie, B. B. (1955). Science 122, 374. Poggio, G. F., Walker, A. E., and Andy, 0. J. (1956). A.M.A. Arch. Neurol. Psychiat. 75, 350.

Pollock, L. J., and Davis, L. (1930). Arch. Neurol. Psychiat. 23, 303. Powell, E. W., Haggart, T. J., Goodfellow, E., and Niemer, W. T. (1957). Neurology 7, 689.

Powell, T. P. S., Guillery, R. W., and Cowan, W. M. (1957). J . Anat. 91, 419. Pribram, H. B., and Barry, J. (1956). J . Neurophysiol. 19, 99. Pribram, K. H., and Kruger, L. (1953). Ann. N . Y . Acad. Sci. 58, 109. Pribram, K. H., and Mishkin, M. (1955). J . Comp. and Physiol. Psychol. 48, 198. Proctor, L. D., Knighton, R. S., and Churchill, J. A. (1957). Neurology 7, 193. Renshaw, B., Forbes, A., and Morison, B. R. (1940). J. Neurophysiol. 3, 74. Rinaldi, F. (1958). J. Nervous Mental Disease 126, 272. Rioch, D., and Brenner, C. (1938). J . Comp. Neurol. 68,491. Roberts, E., and Baxter, C. F. (1958). Neurology 8, Suppl. 1, 77. Roberts, E., and Frankel, S. (1950). Federation Proc. 9, 219. Rodin, E. A., Mulder, D. W., Faucett, R. L., and Bickford, R. G. (1955). A.M.A. Arch Neurol. Psychiat. 74,365.

46

W. R. ADEY

Segundo, J. P., Arana, R., and French, J. D. (1955). J. Neurowrg. 12, 601. Shore, P. A., Silver, S. L., and Brodie, B. B. (1955).Science 122, 284. Sigg, E. B., and Schneider, J. A. (1957).EEG Clin. Neurophysiol. 9, 419. Stevens, J. R. (1957). A.M.A. Arch. Neurol. Psychiat. 77, 227. Terzian, H., and Ore, G. D. (1955).Neurology 5,373. Thomalske, G., Klingler, J., and Woringer, E. (1957). Acta Anat. 30, 865. Thomas, G. J., and Otis, L. 5. (1958a).J. Comp. and Physiol. Psychol. 51, 130. Thomas, G. J., and Otis, L. S. (1958b).J . Comp. and Physiol. Psychol. 51, 101. von Euler, C.,and Richi, G. F. (1958).J. Neurophysiol. 21, 231. von Euler, C.,Green, J. D., and Ricci, G. F. (1958). Acta Physiol. Scand. 42, 87. Walshe, F. M.R. (1924). Bruin 47, 159. Weil, A. A. (1955). J. Nervous Mental Disease 121, 505. Weiskrantz, L. (1958). I n “Psychotropic Drugs,” (9. Garattini and V. Ghetti, eds.), Elsevier, Amsterdam, Holland, pp. 07-72. Weiskrantz, L., and Wilson, W. A. (1958).J. Comp. and Physiol. Psychol. 51, 107. White, L. E.,Nelson, W., and Foltz, E. L. (1958).EEG Clin. Neurophysiol. 10, 361. Whitlock, D. G., and Nauta, W. J. H. (1950).J. Comp.Neurol. 106,183. Whitlock, D. G.,Arduini, A. A., and Moruzzi, G. (1963). J. Neurophysiol. 16, 414. Woolley, D. W. (1958). Research Publ. Assoc. Nervous Mental Disease 36, 381. Yakovlev, P., Hamlin, H., and Sweet, W. H. (1950). J. Neuropathol. Ezptl. Neurol. 9, 250. Youmans, J. R. (1950).Neurology 6, 179. Zigas, V., and Gajdusek, D. C. (1957). M e d . J. Australia 2, 745.

NATURE OF ELECTROCORTICAL POTENTIALS AND SYNAPTIC ORGANIZATIONS IN CEREBRAL AND CEREBELLAR CORTEX By Dorninick P. Purpura Paul Moore Research laboratory, Department of Neurological Surgery, College of Physicians and Surgeons, Columbia University, New York, New York

Part I. Brain Waves as Axon-Spikes, Cell, “Dendritic,” and Postsynaptic Potentials

I. General Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Electrical Responses of Excitable Cells . . . . . . . .......................... 111. Applied Axonology . . . . . . . . ..... ....................... IV. Cortical Potentials as Sync d, Spontaneous Fluctuations in Cell Membrane Polarization; the “Somatic Potential” Theory . . . . . . . . . . . . . . V. Brain Waves as “Graded Responses”; Current Theories of Cortical Elec-

48 50 53

.............................................. s” and Spikes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Dendritic Potential Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... C. “Postsynaptic Potential” Theory of Cortical Electrogenesis . . . . . . . . . . .

61

58

62 64

72

Part 11. Synaptic Origin of Cortical Surface Potentials I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Nature of Cortical Synapses . . . . . . . . . . . . . . . . . . . . . 80 111. Applied Synaptology . . . . . ............................. 82 A. Depolarizing Postsynaptic Potentials of Apical Dendrites . . . . . . . . . . . . . 83 B. Are Cortical Pyramidal Neurons Electrically Inescitable? C. Hyperpolarizing Postsynaptic Potentials of Apical Dendrites . . . . . . . . 91 IV. Functional Differences in Cortical Axosomatic and Axodendritic Synapses 97 A. Geometrical Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 98 B. Chemical Specificities . . . . . . . . . . . . . . . . . . . . . . . . 99 C. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 Part 111. Transmissional and Conductile Activity in Different Cortical Neuronal Organizations I. Relationship of Spontaneous and Evoked Potentials to Single Neuronal .......... Discharges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Surface Potentials and Discharges of Corticospin B. Evoked Potentials and Conductile Activity . . . . . . . . . . . . . . . . . . . . . . . . . . C. Factors Contributing to Nonrelatedness . . . . . . . . . . .......... D. Interpretation . . . . . . ............... .......... ...................................... ns in Cerebral Cortex . . . . . . . . . . . . A. Distribution and Dispersion of Cortical Activity Stimulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

101 103 103

107

48

DOMINICK P. PUBPURA

B. Antidromic Cortical Stimulation . . . . . . . . . . . . . . . . . . . . . . . . . .... C. Synaptic Distribution of Specific Afferent Activity . . . . . . . .... D . Synaptic Distribution of Transcallosally Evoked Activity . . . . . . . . . . . . E. Synaptic Distribution of “Unspecific” Thalamocortical Activity . . . . . . F. Distribution of Caudate-Cortical Activity ..................... G. Synaptic Distribution of “Reticulocortical” Afferents . . . . . . . . . . . . . . . . 111. Analysis of Different Synaptic Organiaations in Cerebellar Cortex . . . . . . . . A. Response to Surface Stimulation . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Responses to Afferent Stimulation ................ C. Effects of w-Amino Acids on Differ from the Same Locus .......................... D. On Localitation in the Cerebellar Cortex . . . . . . . . . . . . . . . . . . . . . . . . . . IV. General Considerations, Conclusions, and Summary ..................... A. Recapitulation .............................. ............. B. On the Properties of Apical Dendrites ............. C. Further Implications of the Theory .................................. References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

116 119 124 127 133 136 142 143 144 150 161 152 162 153 164 166

Part I. Brain Waves as Axon-Spikes, Cell, “Dendritic,” And Postsynaptic Potentials

1. General Introduction

Prior to the discovery that the bioelectric events associated with peripheral and central synaptic transmission differed from those underlying conduction in axons, there was little reason to suppose that the complex activity of the central nervous system could not be resolved exclusively in terms of the processes operating in peripheral nerves (Lucas, 1917). In order to fully appreciate the overwhelming influence which axonology exerted on all aspects of neurophysiology it is only necessary to recall that Sherrington’s comparison (1925) of central excitatory state, c.e.s., with Lucas’ local excitatory process in nerve was long considered “one of the most fruitful working hypotheses ever introduced in neurophysiology” (Lorente de N6, 1939, p. 418). Synapse physiology could not be treated any differently from axon physiology until the distinctive properties of synapses were exhaustively investigated. The knowledge which has accumulated in the past two decades concerning the fundamental differences between synaptic and conductile electrogenesis (cf. review by Grundfest, 1957b) has provided a suitable theoretical framework for reinterpreting c.e.8. and its counterpart, c i s . (central inhibitory state). Moreover, the relatively slow potential changes recorded from the surface or depths of the spinal cord, spinal roots, or interior of motoneurons accompanying the enduring states of excitation or inhibition are generally attributable to the special properties of synapses, not

NATURE OF ELECTROCORTICAL POTENTIALS

49

of axons. It would be remarkable, indeed, if the mechanisms underlying the production of slow waves in other parts of the neuraxis were qualitatively different. The central theme of this review will be devoted to the proposition that such is not the case. However, a more important objective is to define the sites of origin of postsynaptic potentials (p.s.p.’s) since these presumably constitute the major form of activity recorded from the cortical surface. Such an inquiry entails an analysis of the different properties of different parts of the neuron, recognition of the various forms of bioelectrical phenomena generated by excitable cells, and special consideration of the properties of cortical dendrites. The consequences of identifying slow waves of cortical neuronal aggregates with one of the two characteristic bioelectrical phenomena of neurons, i.e., the postsynaptic potentials, are of more than nosological significance. Recognition of the synaptic origin of such slow waves permits the introduction of pharmacological techniques to the study of specific electrocortical phenomena. Thus, the diverse forms of electrical activity recorded from the surface of the brain become amenable to analysis with selectively acting synaptic blocking or activating agents, which diff erentiate between depolarizing and hyperpolarising p.s.p.’s and provide information relating to their sites of origin. Information acquired in this fashion supplements that obtained with purely electrophysiological dissections of complex electrocortical potentials. But a further significance is attached to the hypothesis of identifying the vast majority of slow potentials of cortex with postsynaptic potentials which stems directly from the properties of the latter. For, if it can be adequately shown that such an identity is valid, then terms used to describe excitability changes in axons (i.e., refractoriness, etc.) are not applicable to descriptions of evoked cortical potentials. Similarly, since p.s.p.’s are essentially “standing potentials” of two kinds, depolarizing and hyperpolarizing, attempts to determine their origin in cortex in terms of classic volume conductor theory become extremely hazardous. The experimental data reviewed and interpreted below in the light of more recent knowledge appear sufficient to justify these assertions. To accomplish the stated objective of providing evidence in support of the synaptic nature of cortical potentials it will first be necessary to indicate the reasons why previous attempts to define the origin of cortical potentials in terms of the properties of peripheral nerves were of limited heuristic value. Only then will a general discussion of the differences in the properties between synapses and axons become relevant to the succeeding arguments. Such arguments can be expected to depend largely on the chief characteristic of synapses, i.e., their chemical sensitivity, in order to explain the alterations in cortical potentials produced by a

50

DOMINICK P. PURPURA

variety of pharmacological agents. That the latter will include some agents classified ordinarily as brain metabolites (e.g., amino acids) is of more than coincidental importance. For, since it can be shown that certain of these amino acids ordinarily found in brain exert profound effects on the functional activity of the cerebral and cerebellar cortex, the problem of determining whether or not cortical electrical activity is primarily metabolically or neurohumorally determined then becomes relatively insignificant. Finally, the use of amino acids as pharmacological tools which permit electrophysiological dissection of cortical synaptic organiaations provides an operational approach to the origin and nature of “brain waves.” This has previously not been possible with purely electrophysiological techniques.

II. Electrical Responses of Excitable Cells

As can be readily appreciated from prior surveys of the problems involved in determining the origin of brain waves (Bullock, 1945; Jasper, 1937) attempts to define the nature of diverse forms of electrical activity recorded from the surface of the brain have generally been based on interpreting the net activity of cortex in terms of the properties and behavior of the neuron. The problem of constructing hypotheses to account for the recorded surface activity would be relatively simplified if the excitable membrane enveloping the neuron possessed uniformly distributed, homogeneous, electrogenic characteristics. Such, however, is not the case. That the all-or-none (conductile) response of axons might not be representative of the major form of electrical activity constituting the surface cortical potentials was appreciated long before other varieties of bioelectrical activity of neurons were intensively investigated. This is detectable in the form of an assumption, implied, when not overtly stated, that neurons possessed rather specialized electrical characteristics in comparison to peripheral nerves. With the recognition that conductile activity was one of a number of electrical responses generated by neurons a more dynamic reconstruction of cortical potentials became possible. While it is obvious that the different electrical responses of neurons can satisfactorily account for brain waves, the problem which confronts the reviewer is to provide evidence as to which of these phenomena is predominantly recorded from the surface of the brain [leaving aside the question as to whether or not non-neural elements contribute anything directly to this activity as suggested recently by Tasaki and Chang (1958) 3.

51 To accomplish this, it is desirable to catalog, briefly, the general properties of excitable cells, if for no other purpose than to feel secure in the knowledge that during the subsequent excursions through the cortical mantle no bioelectrical phenomena are likely to be encountered which cannot be identified with one or another variety of electrical response characteristic of excitable cells. Two major varieties of electrogenesis are recorded from excitable cells following appropriate stimulation, all-or-none and graded responses. The latter may be divided into two subgroups of responses: (1) those electrically excitable; and (2) those which are not eIectricaIly excitable. A full account of the functional significance of these different varieties of electrical activity may be found in recent reviews by Fatt (1954), Fessard (1956), Bishop (1956, 1958), Grundfest (1957a, b, 1958b) and in the monograph by Eccles (1957). A thorough review of the electrochemical processes underlying the responses of excitable cells containing over 750 references to the subject has also appeared (Shanes, 1958a, b ) . All-or-none response: This response is characteristic of all axons so far investigated as well as of skeletal muscle fibers. The major ionic events and membrane processes associated with the axon spike are detailed elsewhere (Hodgkin, 1951, 1958; Hodgkin and Huxley, 1952). The response is triggered by a critical membrane depolarization initiating an increase in sodium conductance which results in further depolarization of the membrane and augmented inward sodium movement (“regenerative action”). This process is halted by another event, %odium inactivation,’’ the membrane potential recovering its resting or near resting value during a phase of increased potassium conductance. The all-or-none response propagates without decrement in a saltatory fashion a t a velocity determined largely by the physical characteristics of the axon. Axons are absolutely refractory to further electrical stimulation during an all-or-none response. Immediately thereafter, only graded responses may be evoked and then additional all-or-none responses with amplitudes determined by the magnitude of post-spike membrane conductance changes associated with the development of after-potentials. after-potentials": The all-or-none response is followed by afterpotentials, much smaller in size and much greater in duration, which vary in different axons. The complete sequence is a negative after-potential followed by a positive potential (cf. Gasser, 1939). Following the spike, intracellularly located microelectrodes record a slowed repolarization followed by a phase of “after-hyperpolarization” when cat motoneurons are antidromically or orthodromically activated (Coombs et al., 1955). (Frank and Fuortes (1955) however, did not observe an after-hyperpolarization following orthodromic stimulation of motoneurons.) The NATURE OF ELECTROCORTICAL POTENTIALS

52

DOMINICK P. PUBPURA

after-hyperpolarization of cat motoneurons attains a maximum of about 5 mv at 10-15 msec and gradually declines so that it is no longer detected after 100 msec. Ionic mechanisms associated with these afterpotentials have been reviewed by Shanes (1958a, b) . Electrically excitable graded responses: The classic local response of nerve is a depolarization which precedes the development of the all-ornone spike. Ionic mechanisms associated with it are poorly understood, but it may be due t o a process of selective sodium permeability like that initiating the all-or-none response (Hodgkin and Huxley, 1952). Alternative mechanisms have also been considered (Grundfest, 1957a; Kao and Hoffman, 1958). The local response of nerve propagates decrementally and may be graded over wide ranges especially during relative refractoriness of the axon (Hodgkin, 1938). It has approximately the time dimensions of the all-or-none spike. At axon terminals (both afferent and efferent) where specialized transitional membrane presumably occurs, the evidence for graded responsiveness has been reviewed (Bishop, 1956). It should be noted that many invertebrate muscle fibers conduct only graded potentials. During refractoriness, single mammalian heart muscle fibers also show graded responses which may vary in amplitude up to the full height of the normal action potential without becoming propagated (Cranefield and Hoffman, 1958; Kao and Hoffman, 1958). Electrically inexcitable graded responses: The evidence supporting the hypothesis that junctional transmission is signaled by the development of a graded, electrotonically propagated response in postsynaptic membrane which cannot be reproduced by electrical stimulation of the postsynaptic site has been marshaled and critically reviewed by Grundfest (195713). Postsynsptic potentials (p.s.p.’s) occur in membrane which is only chemically excitable. Transmitter agents acting on specific receptor sites presumably induce a nonselective increase in membrane conductance or a selective increase in potassium and/or chloride conductance. The former results in a depolarization, the latter in a hyperpolarization of the membrane. Membrane hyperpolarization is not necessarily a concomitant of inhibitory transmitter activity. At peripheral inhibitory junctions (crayfish stretch receptors) increases in membrane conductance may be sufficient to “short-circuit” the electrotonic spread of generator potentials (Edwards and Kuffler, 1957). At crustacean neuromuscular junctions inhibitory transmitters may increase membrane conductance and thereby prevent the transmitter induced depolarization at excitatory synapses (Fatt and Katz, 1953). Depolarizing and hyperpolarizing p.s.p.’s act on electrically excitable membrane components through electrotonic extension. In certain excitable cells which are not capable of supporting all-or-none responses, only p.s.p.’s may be gener-


53

ated. Not being electrically excitable, postsynaptic membrane is not affected by its own electrogenesis, hence the response is graded and without refractoriness. Absence of refractoriness implies that p.s.p.’s may be summated over wide ranges, amplitude and duration being functions of the stimulus. Of particular significance for later discussions is the fact that “unitary” p.s.p.’s recorded extra- or intracellularly from cat motoneurons are 10-15 msec in duration (Eccles, 1946, 1957). Further analysis of the properties of synaptic electrogenesis will be discussed below. In subsequent sections it will be shown that each of the responses noted above has, a t one time or another, been considered by different (or the same) investigators to be major fundamental “units” of electrical activity involved in the production of brain waves. However, it is now clear that the harmonious blending of conductile and graded activity in different organizations of neurons is responsible for the electrical activity recorded from the cortical surface. Some of the experimental data to be discussed have been treated differently in recent reviews (Albe-Fessard, 1957; Brazier, 1958a; Bremer, 1958; Buser, 1957). The reader should consult these in order to obtain additional reference material relating to studies on evoked cortical potentials and historical analysis.

111. Applied Axonology

There appears to have been little serious attention focused on the origin and nature of brain waves after the initial pioneering discoveries of Caton (1875), Danilewski (1891), Prawdicz-Neminski (1925), and Berger (1929) until Adrian and Buytendijk (1931) attempted to analyze the slow potentials recorded from the isolated brain stem of the goldfish, and Bartley and Bishop (1933a) began their investigations on the visual cortex of the rabbit. The provocative statement by Adrian and Buytendijk that rhythmic activity of the respiratory center in the goldfish was observable in the entire absence of sensory impulses seems to have obscured the fact that their original intention was to define the neural elements responsible for these slow potentials. Only two possibilities were recognized. Slow waves could be due to the summation of brief potential changes occurring repeatedly in different elements, or represent potential changes occurring at the same rate throughout the active part of the brain stem, increasing or decreasing pari passu with t h e potential change at the electrodes. Although Adrian and Buytendijk considered it more probable that. the slow potential waves represented slow depolarizing changes in cells

54

DOMINICK P. PURPURA

and dendrites, this hypothesis was categorically rejected somewhat later by Adrian and Matthews (1934). The introduction of more reliable recording techniques is noted to have been the major factor involved in permitting a final decision. The conclusions formulated by Adrian and his collaborators concerning the origin of potential waves in cortex became important guides to future investigations for a number of years. They are, therefore, of more than historical significance. Adrian and Matthews inferred that the slow cortical waves were summations of monophasic spikes of 10-100 msec duration occurring in some nervous constituent of the grey matter, cell bodies or dendrites, or both. Axons were excluded from consideration. Different neurons might produce action potentials of different duration, but none generating spikes of less than 10 msec duration. Granting these assumptions, the electrical activity of the cortex could be related to events of the same character as those taking place in active nerve or muscle fiber, i.e., to successions of brief action potentials repeated rhythmically with a frequency which could vary within wide limits (Fig. 1 ) . No modification in the electrical response of each unit apart from the change in frequency was recognized. Adrian and Matthews’ investigations led them to conclude that “we must abandon hope that the slow potential changes in the cortex might be an index of slow change of polarization in the nerve cells and so an index of the rise and fall of the excitatory state which preludes activity.” It is of interest to point out that Adrian’s early studies led him to suggest that: (1) the “spike” discharges of axons and their assorted after-potentials contributed nothing to the normal spontaneous electrical activity of cortex; (2) an electrical response of 10 msec or more in duration generated in the cell bodies or dendrites of cortical neurons was the basic unit of activity of which the slow waves were composed. I n further studies a similar 10 msec monophasic negative “superficial response” of cortex was described which was assigned to the activity of “nerve cells with dendrites running laterally, e.g., the horizontal cells of Cajal, or the superficial branches of apical dendrites of pyramidal cells” (Adrian, 1936). Other early concepts relating to the origin of brain waves were somewhat a t variance with those outlined above, but were, nevertheless, based on the notion that “cells or specialized parts of them rather than fibers must produce these large potentials” (Bartley and Bishop, 1933b, p. 180). An experiment apparently unique in the annals of cortical electrophysiology is worth relating for its historical significance (cf. Bullock, 1945). In an attempt to determine whether the record of a bundle of fibers could give rise to potentials of the magnitude recorded from the cortical surface, Bartley and Bishop (1933b) dissected a rabbit vagus


55

nerve (1/2 mm in diameter) and laid it across cortex, then into a cortical slit flush with the surface, then buried 1 mm deep and finally threaded the nerve through the lateral ventricle of the brain. Under these various conditions, stimulation of one end of the nerve resulted in action potentials which were insignificant compared to the potentials developed by the same nerve in air. The results of this experiment supported the

I?-

//

Response of units

Composite

- -----------

wave

PIO.1. Slow potentials viewed as summations of cell discharges. Upper left and right: potential gradients due to a wave of activity arising in a nerve cell and traveling down the axon, to show the production of monophasic waves. Below: potential gradients due to repeated asynchronous activity in a group of nerve cells to show the formation of slow potential waves. For simplicity it was assumed that all the nerve cells had the same effect on the recording system (from: Adrian and Matthews, 1934).

conclusion noted above. But it was later concluded that for analytical purposes only the cortex could be treated like a “very broad and very short nerve with elements conducting in both directions . . .” (Bishop, 1936). Although the shortest wave observed in the cortex (Bartley and Bishop, 1933b) was also approximately 15 msec in duration, one possible way to explain slow waves was t o treat them as envelopes of individual axon epikes.

56

DOMINICK P. PURPURA

While not denying the existence of potentials other than nerve axon spikes, Bishop (1936) attempted to reconstruct the slow electrical responses evoked in rabbit cortex by optic nerve stimulation from envelopes of axon potentials conducted in elements of different length oriented in various directions in cortex with respect t o the recording electrodes. The unit responses of cortex might make contributions to the record of any form, from a monophasic first phase, through all degrees of diphasicity to a monophasic second phase. Three factors emerged from these studies of particular relevance for later discussions. (1) The time occupied by an impulse started in the optic nerve in traversing the optic cortex was estimated a t about 200 msec. This represented the time during which successive activation of cortical elements gave rise to a continuous succession of potentials. (2) The response of cortex to a surface stimulus might be repetitive, the frequency of repetition being similar to the spontaneous cortical rhythm. (3) The repetitive response associated with “refractory states” of cortex was of the same duration as one cycle of the spontaneous or induced rhythm (200 msec). The last factor deserved special attention for while it was recognized that the individual elements involved in the evoked surface response had a refractory period no longer than that for nerve fibers, the long refractoriness of 200 msec was attributed to the time required for completion of the entire sequence of repetitive events occurring during a single progression of impulses in the optic pathway. At no time did Bishop equate “cortical refractoriness” with axon refractoriness. In the years which followed the first attempt to show how evoked and spontaneous responses of cortex could be viewed in terms of summated axon potentials, Bishop’s views underwent considerable alteration. First it was categorically stated that central nervous system activity was predominantly that derived from cells (soma and dendrites) rather than of axons (Bishop and O’Leary, 1942) and finally, the specific contribution of apical dendrites of pyramidal cells to this activity (Clare and Bishop, 1955) was recognized. Another formulation (Bishop, 1949) was characterized by modification of the original thesis and was based largely on interpreting the responses of cortex in terms of summations of monophasic cell spikes ; diphasic records representing the activity of two different groups of cells related by synaptic passage. Cells could be treated like a short length of axon polarized anodally and cathodally at the two ends, an interpretation based on the hypothesis proposed by Libet and Gerard (1941). The unit cell potential was interpreted as the resultant of unequal depolarization of the two ends of the cell, dendritic and axonal, during activity. This hypothesis appeared t o account reasonably for the alterations in various components of evoked cortical poten-


57

tials during the application of polarizing currents and was supported by data obtained from peripheral axons (Bishop and Erlanger, 1926). Although Bishop and his colleagues apparently recognized the necessity of modifying some of their views concerning the origin of cortical potentials there appeared to be little reason to alter, a t the time, in any great detail the sketches showing how slow potentials might be derived from an analysis of the statistical distribution of unit cell discharges (cf. Bishop, 1936, Fig. 4 and Bishop, 1949, Fig. 4 ) . Still it is of interest to note that records of striate cortex responses published after the introduction of improved recording techniques revealed clear distinctions between 1-msec sharp spikelike deflections and slower components of 5-10 msec duration (Bishop and O’Leary, 1938). The brief components were not observed in earlier records (Bishop and O’Leary, 1936). Although the slower responses could not be resolved into elements of the dimensions of axons spikes, it was felt that they could consist of temporal dispersions of such spikes (Bishop and O’Leary, 1938). It should be added, parenthetically, that the reconstruction of slow waves from units of briefer time dimension as suggested by Bishop and his collaborators were, and are today, perfectly valid and didactically useful demonstrations of the application of principles of electrophysiology to complex analytical problems. That alternative explanations of the origin of slow waves might also be equally plausible was always apparent to Bishop (cf. discussion, Bishop, 1936). Studies of nerve fibers had, of course, revealed phenomena other than those immediately associated with the production of the all-or-none discharges. I n particular, it had been known that as a consequence of activity, different fibers exhibited relatively prolonged oscillations in membrane potential (Gasser, 1933). Slow polarization changes in axons might contribute in some way to cortical surface potential changes as well as fluctuations in the excitability of cortical neurons. Accordingly, it was proposed (Jasper, 1936) that cortical cell potentials could be central fiber potentials of very slow time characteristics relative to those of peripheral fibers, their spontaneous discharges being dependent upon excitatory and metabolic conditions similar to those which were found to determine spontaneous discharge in isolated peripheral axons. Considerable data were at hand from studies of peripheral axons to justify these assertions (Monnier and Jasper, 1932). Like Adrian and Matthews (1934), Jasper assumed that brain waves could represent the repetitive spike-discharges of central neurons with slow but variable time characteristics. The most interesting feature of Jasper’s early hypothesis was the recognition that slow polarization changes, analogous to the after-poten-

58

DOMINICK P. PURPURA

tials in axons, acting in a graded and not all-or-none fashion might reflect changes in ‘‘cortical excitatory states”; slow positive polarization of the cortical surface representing a decrease or complete dropping out of the autonomous alpha rhythm (a decrease in cortical excitatory state) a surface negative polarization being associated with the return of rhythmic activity. The decrease in rhythmic activity accompanying the positive polarization change (depression) was believed to be associated either with anodal depression (i.e., a true inhibition) or cathode blocking (i.e., Wedensky inhibition). One advantage of this theory was its apparent usefulness in accounting for the rhythmicity of central neurons, as well as most of the electrical phenomena observed without assigning new mechanisms of action peculiar to cortical neurons. It must not be assumed that the hypotheses which have been briefly reviewed above constituted the general concensus of all physiologists (of. Davis, 1936) interested in cortical electrophysiology during the “axonological era” (circa 1922-1940). (Indeed, were this the case, the necessity for continuing this review would be subject to serious doubts.) But considering the rather overwhelming volume of information on the physiology of peripheral nerves available at that time, i t seemed reasonable to conclude without fear of contradiction, that “the greatest progress toward an adequate interpretation of cortical potentials (was) to be made by first applying principles derived from the known electrical properties of axons” (Jasper, 1936, p. 337). Like Bishop, Jasper was also among the first t o recognize somewhat later, the inadequacy of these axonological principles.

IV. Cortical

Potentials as Synchronized, Spontaneous Fluctuations in Cell Membrane Polarization; the “Somatic Potential” Theory

Although it seemed likely to some that cortical potentials might represent summations of single cell body “spikes” of 10-20 msec duration rather than of 1-2 msec axon-spikes, it appeared more reasonable t o others to reject all such hypotheses and to propose, instead, that cortical potentials might represent membrane potential fluctuations truly spontaneous in nature in the sense that no nerve impulses would be immediately necessary to their evocation (Libet and Gerard, 1941). This hypothesis appears to have been formulated primarily from observations relating to the persistence of spontaneous rhythms in a variety of isolated neural structures following “blocking of synaptic conduction by nicotine” (Libet and Gerard, 1938). The latter point deserves some special consideration for it repre-


59

sented a “crucial” test of the hypothesis that the spontaneous electrical rhythm of isolated structures (e.g., frog’s olfactory bulb) originated in single neurons and was not maintained by repetitive synaptic bombardment by trapped impulses in closed circuits (Lorente de N6, 1935). Nicotine was known to block synaptic transmission in autonomic ganglia, but no reference to a similar effect on central synapses had been reported up to the time of Libet and Gerard’s experiments (1938). Their results are noteworthy: “In the isolated olfactory bulb the neurons maintain their rhythms in the absence of impinging nerve impulses, for the rhythm is preserved, actually enhanced and made more regular, even after half an hour’s soaking in 0.5% nicotine.” The central stimulating action of nicotine, related to its ganglionic depolarizing (excitatory) effect is today well-known (Paton and Perry, 1953; R. M. Eccles, 1956). For a more detailed account of the electrophysiological mechanisms underlying the synapse blocking action of such depolarizing drugs, the reader is referred to other reviews (Grundfest, 1957c; Shanes, 1958a, b). What is of importance a t this time is to point out that in attempting to demonstrate central synaptic blocking effects with nicotine, Libet and Gerard actually demonstrated (olfactory bulb and cat spinal cord) that the primary pharmacological action of nicotine was to “excite” or activate depolarizing synapses. Nevertheless, it was concluded that brain rhythms were independent of synaptically mediated influences and no attempt was made to account for the augmentation of such rhythms occurring during the presumed “synaptic blockade” by nicotine. From these and other data, an hypothesis to account for cortical potentials was developed (Gerard and Libet, 1940) which proposed that the cell body of each neuron was polarized from dendritic to axonic poles (“somatic potential”), a d.c. potential maintained by the cell’s active metabolism. Horizontal sheets of cells, oriented vertically, could behave like the polarized membrane of a nerve fiber. Local depolarization resulting from the discharge of one cell or a few adjacent ones, would permit neighboring cells to discharge through the “leak” and so initiate a spreading wave of depolarization. Thus, electric currents flowing through intercellular fluids were considered a likely basis for neuron interaction. Synaptic activation of cells might contribute to a change in the over-all “somatic potential,” to alter it in one direction for excitation, and in the other for inhibition, otherwise the unevoked potential rhythms of central neurons were largely controlled by their own metabolism (Dubner and Gerard, 1939). Concerning the relationship of the spontaneous rhythms to cellular potentials, it was suggested that individual neuron potentials might be of a frequency and form identical with those of the recorded potentials. The latter were integrals of synchronized unit potentials,

60

WMINICK P. PURPURA

synchrony being achieved by neuronal interaction resulting from excitatory, intercellular currents. The effects of altering the physicochemical environment of central neurons on the spontaneous electrical rhythms were detailed and “impressive parallels’’ noted between these and similar effects on oscillating after-potentials of nerve. Somewhat earlier, Gesell (1926) had proposed an electrotonic theory of nerve cell discharge which was re-employed later (Gesell, 1940) primarily to account for the rhythmicity of respiratory center neurons. According to Gesell, metabolic gradients in neurons could give rise to electrotonic currents flowing within the cell from dendrites to axon hillock, outwards across axon hillock membrane, where repetitive discharges are initiated, and returning in the immediate environment back to the dendrites. In contrast to Gerard (1936; Gerard and Libet, 1940) and Gesell assigned an important role to synapses especially on dendrites as a source for the high metabolic activity of these elements. Gesell was convinced that self-engendered electrotonic exciting currents of metabolic origin might have a universal application to automatic biological activity in general, for “we envisage the electrotonic current as the master tool of integration . . . and would well be considered in the explanation of brain waves” (Gesell, 1940, p. 505). I n succeeding sections attention will be focused on a variety of hypotheses concerning the physiology of nerve cells (e.g., site of impulse initiation in motoneurons; inability of dendrites and/or cell bodies to support all-or-none discharges, etc.) which appear to have been of speculative interest to Gesell (1940). Gerard’s hypothesis influenced interpretations of brain wave patterns in various invertebrate and vertebrate species (Bullock, 1945). The “somatic potential” theory of cortical potentials was, indeed, a radical departure from “axon-spike” and “after-potential” theories outlined above. According to the “somatic potential” theory, intrinsic membrane potential fluctuations in individual cells determined their excitability and not prior conductile activity. More importantly, however, it was proposed that intercellular current “leaks” established on the basis of differential cellular polarization might be modified by synaptic influences, but were otherwise not dependent on the latter. One corollary of the “somatic potential” hypothesis is worth additional comment. If spontaneous, slow neuronal depolarizations occurred which could be modified by synaptic drives as well as physicochemical changes in the environment, then it was possible to conceive of slow potentials independently of all-or-none cellular discharges. “I take it as probable, therefore, that an electrically beating cell may discharge a single or brief volley of impulses down its axon a t some regularly repeated phase of the electric variation, but that this does not occur in the majority of


61

cases. The potential wave may mark an excitation state which rises and falls rhythmically, sometimes reaching the threshold for discharge on the rise or at the peak, but more often not" (Gerard, 1936, p. 295). It should be pointed out that as early as 1936 the existence of slow waves and spikes as two distinctly different bioelectric phenomena was reported in a commentary by Forbes (cf. discussion of Davis, 1936). Not until four years later, however, were these decisive experiments to be formally published in detail (Renshaw et al., 1940). The significance of Forbes' statement appears to have escaped the notice of some primarily interested in axonological interpretations of cortical potentials, but to Gerard, Forbes' remarks seemed to confirm the impression that slow, graded potentials as well as brief spikes could be generated by single cortical neurons. Summary. The earliest attempts to define the nature and origin of cortical potentials were based on two fundamentally different assumptions: (1) that slow potentials were summations of individual all-ornone spike discharges or after-potentials; or (2) slow potentials represented slow polarization changes in cells capable of sustained membrane potential fluctuations independently of synaptically mediated influences. To account for some troublesome experimental facts [such as the finding that the briefest waves recorded in cortex were of longer duration (1015 msec) than of axon-spikes] it was necessary to confer special properties on cortical neurons. Most difficulties could be overcome by assuming that individual discharges of cortical cells might be much longer in duration than axon-spikes. Additional discrepancies were eliminated by recognizing the possibility that cell bodies might spontaneously develop sustained graded depolarizations, as well as brief spikes. In all cases, however, it seemed possible to rely on the vast axonological literature to supply pertinent and sufficient information t o explain most of these special properties.

V. Brain Waves as "Graded Responses"; Current Theories of Cortical Electrogenesis

It is difficult to determine precisely when electrocortical potentials came to be viewed as more characteristic of cell potentials rather than summations of all-or-none responses. The problem for the historian would be considerably simplified if a temporally distinct sequence of events, vie., discovery, conjecture, trial, application, discovery, etc. . . . could be defined objectively. The analysis would be further facilitated if these events were viewed in retrospect, from a distance. But events are

62

DOMINICK P. PUItPURA

too immediate, focusing difficult and the picture is blurred. I n lieu of historical completeness, outlines must suffice and omissions are inevitable. Until the time when fact and fancy are more clearly defined, the origin of current concepts of the nature of brain waves can only be approximated. The indistinct pattern of development which presents itself to this reviewer appears to consist of the following elements: (1) the discovery that cortical neurons could generate slow waves as well as axon-like spikes; (2) recognition of the “special” properties of dendrites; (3) the discovery that slow potentials were associated with junctional transmission; and (4) application of principles of synaptic electrogenesis and dendritic potential theories to analyses of electrocortical activity. A. INDEPENDENCE OF “WAVES” AND SPIKES Few studies have been as significant as those reported by Renshaw et al. (1940). The discovery that slow waves as well as brief axon-like spikes were distinctly recordable from the immediate vicinity of small aggregates of neurons with microelectrodes appeared to define clearly one of the major problems in cortical electrophysiology. These observations have since been repeatedly confirmed (Brookhart et al., 1951; Li and Jasper, 1953) most recently by Mountcastle et al. (1957) (Fig. 2). Additional observations were recorded by Renshaw et al., whose significance wilI become apparent below. (1) Waves and spikes were not interdependent; (2) waves were not as localizable as brief spikes; (3) monophasic waves of 20-100 msec duration recorded from the surface and depths of the hippocampus following afferent stimulation were presumably due to the activity of vertically oriented pyramidal cells; polarity being dependent on differences in the location of synapses; and (4) “If the position of the active synapses determines the electric field set up by the pyramidal cells, the existence of a process more or less localized to the region of an active synapse and characterized by an electrical sign is indicated” (Renshaw et al., 1940, p. 102). The tentative assignment of slow potentials recorded with microelectrodes to events occurring at the synapses of active neurons carried with it no commitment as to the location of the activity, viz. in presynaptic terminals, postsynaptic elements, or both. (Note: The term “synaptic potential” to describe the slow potentials recorded electrotonically from the dorsal roots following afferent stimulation was introduced by Fessard and Matthews (1939) in a preliminary note appearing after the manuscript by Renshaw et al. (1940) was submitted for publication.) Although the distinction between pre- and postsynaptic potentisl activity has been considerably elucidated since 1940 (cf. below), some idea of the continuing doubts as to the origin of the localized


63

spontaneous slow potential activity in cortex recorded with extracellularly located microelectrodes (Fig. 2) can be appreciated by noting the commentary by Mountcastle et al. (1957), relating to the restricted origin of slow wave activity, the lack of association between spikes and waves, and failure of peripheral stimulation to modify these potentials. “All of these observations suggest that this event is the sign of local nonconducted oscillations in the membrane potential of the soma or dendrites of cortical cells, but we have no direct evidence that this is so, microns 1072 1092 1112 1116 1120 1124 1132 II36

1140

1152

I

I sec

I

FIQ.2. Microelectrode recording of local spontaneous slow waves of limited spatial extent with change in position of electrode. Depths below cortical surfarc indicated in microns for each record. Cat under ether-decamethonium bromidepentothal anesthesia; long time-constant recording. Note marked attenuation of potentials produced by 4 p displacement of microelectrode (1120-1124 p ) (from : Mountcastle et al., 1957).

and the chance that glial cells may be responsible for it must be considered” (p. 401). Without knowledge of the existence, in the central nervous system, of responses peculiar to synapses, Renshaw et al. (1940) concluded from histological examination of microelectrode tracts that localized slow waves evoked in response to afferent stimulation characterized the activity of the perikarya of pyramidal cells. It should be recalled that earlier Bishop and O’Leary (1938) rejected axon-spikes in favor of cell (soma and dendrites) discharges aa

64

DOMINICK P. PURPURA

the “unit” electrical component of brain waves. There occurred shortly thereafter another eventful “antidromic” progression which arose, quite logically, from a consideration of the distribution of potentials recorded from the bird optic lobes following optic nerve stimulation. When single shocks were applied to the optic nerve of ducks and geese, slow negative potentials were recorded from the surface of the optic tectum. To explain these potentials and their alterations during progressive displacement of the recording electrodes, O’Leary and Bishop (1943) thought that they might be postsynaptic potentials of 5 msec duration, or more, occurring in dendrites. The potentials were “essentially non-conducted or stationary.” Their general conclusions are well worth recording: “In previous attempts we have made to correlate the activity of the cortex with the depth in the cortex a t which the activity was recorded, we believe some of the difficulties we have met arose from thinking of the cortex as a system of short parallel axons conducting the usual impulses. It now seems probable that much of the recorded activity of the cortex should be assigned to dendrites rather than to axons . , . and that account should be taken of the fact that most of the dendrites extend through several cortical layers. The details of their responses may be as much a function of how they are activated as of their histological arrangement” (O’Leary and Bishop, 1943, p. 85). The two reports cited above by Renshaw et al. (1940) and O’Leary and Bishop (1943) have been rather arbitrarily selected as representative of the conjectural stage of development of current concepts of cortical slow wave activity. The experimental facts established that slow waves could in some manner be generated by cortical neurons. The speculations were further strengthened by: (1) demonstration of the sites of origin of slow waves in cortical neurons; and (2) application of newly acquired data on the possible nature of graded potential activity.

B. DENDRITIC POTENTIAL THEORY The reader should by now be familiar with the rigorous experimental attitudes of Bishop and his collaborators toward the problems of cortical electrophysiology. The clues that there might be something special about dendrites in the bird optic lobes did not appear sufficient to warrant further speculations without additional experimental demonstration, One approach was to return to the mammalian visual cortex where data had been accumulated relating to the processes involved in the progression of evoked activity through the various cellular layers (Bishop and O’Leary, 1936, 1938). Stimulating small fractions of cortex and recording from different depths, Bishop and Clare (1952b, 1953) analyeed more completely the distribution of brief spikes and slow waves gen-


65

erated by optic nerve volleys. The first spike was assigned to presynaptic activity in radiation axons, the second to the discharge of cortical cells (possibly in layer IV), and subsequent spikes to cells located higher in cortex. The initial slow surface positive wave associated with the cellular discharge was inferred to represent the activity of basal dendrites of these cells; subsequent slow negativity to antidromically conducted activity of apical dendrites. It was further suggested that: “apparently the apical dendrites usually fail to fire antidromically” upon afferent stimulation of cortex, but such conduction could be induced with intense afferent stimulation or following application of dilute strychnine solutions. The studies are of added significance in that they indicated the difficulty of attempting t o relate the source of potential components involved in the primary response t o conventional cell layers of the cortex. Unpublished observations were cited which called attention to the fact that potentials evoked from optic cortex via afferents other than optic radiations (via. from a point ventral to the brachium of the superior colliculus) differed completely from those evoked by radiation volleys. Whether or not the same cells were activated by the two modes of cortical activation was not known, but it seemed probable that “activation of the same cells of the cortex by way of different systems of endings should give rise to potentials of different form.” The reader is to note the similarity between this conclusion and that of Renshaw et al. (1940) recorded above. Additional information concerning the properties of dendrites was forthcoming from Chang (1951a, 1952, 1955a) and Tasaki et al. (1954b). Chang specifically assigned the 10-20 msec negative potentials of cortex (Adrian, 1936) evoked by surface stimulation “to the passage of nerve impulses along the apical dendrites of the pyramidal cells” (Chang, 1951a) and considered this response in many respects comparable to the soma response of motoneurons in the brain stem (Lorente de N6, 1947), on the basis of conduction velocity measurement. Recognizing the difficulty of accounting for the propagation of the directly evoked dendritic potential for distances up to 5 mm along the cortical surface, when the apical dendrites presumably did not extend beyond 2 mm (a figure now known to be incorrect, cf. Sholl, 1956), Chang (1951a) concluded that the potential must “involve the divergent branches of dendrites originating from the same shaft and must be compounded with indirectly initiated impulses.” No explanation was provided by Chang to account for the fact that his data clearly showed that a t distant points dendritic potentials might be of much greater magnitude than those recorded closer to the stimulating electrodes (Fig. 3,I). Having inferred that stimulation of the cortical surface resulted in

66

DOMINICK P. PURPURA

T I

¶ 4 6 8 1 0 1 P 1

FIQ.3. I. Series of sample records used by Chang for determination of conduction velocity of cortical dendritic potential. Records A-F were taken successively from six points on cortical surface of area 6 of monkey, each point being 1 mm further away from stimulating electrodes than its preceding one. Distance between stimulating and recording electrodes in each case is indicated under corresponding record (from: Chang, 1951a). (Note increased amplitude of responses at 3-4 mm from stimulating site.) 11. A. Distance-response graph for surface negative responses (cat). Ordinates : amplitudes of first surface-negative responses expressed w per cent of the amplitudes recorded at 2 mm from the stimulated point. Abscissa: recording distance from point of stimulation in mm; 100% amplitude and 2 mm distance are marked by dotted lines. Sequence of 0, 0, 0, A indicate trials with increasing stimulus strength in one experiment. (0): 10% response, ( 0 ) :20% response, ( 0 ) :60% response, ( A ) : 100% response. Stimulating and recording positions shown in diagram. 11. B. Plots of latencies (ordinates: msec) of first surface-negative response (measured to peak) against recording distance (abscissa: mm). Note shortening of latency at 8-10 mm with 100% responses ( A ) . (From Brooks and Enger, 1959.)

electrical excitation of dendrites and production of a 15-20 msec negative wave which propagated along dendrites a t approximately 1-2 m per second, Chang (1955a) later identified this potential with the surface negative component of the antidromic pyramidal response. According to

67 Chang, dendrites exhibited the following properties: sensitivity to hypoxia and local trauma, slow conduction and recovery processes, and an action potential of 15-20 msec; properties which clearly appeared to distinguish dendrites from axons. Chang’s conclusions were applied in a most fascinating way to certain theoretical problems relating to the physiological significance of activating pyramidal cells synaptically via their dendrites and cell bodies (Chang, 1952, 1955a, 1956) which will be discussed below. Tasaki and co-workers (1954b) attempted to dissect cortical and lateral geniculate responses to presynaptic stimulation. All-or-none impulse propagation in dendrites a t a relatively slow velocity was assumed to account for the 15-20 msec responses recorded from lateral geniculate neurons. These potentials were thought to be related to long duration responses recorded intracellularly from Mauthner cells (Tasaki et aZ., 1954a) following antidromic stimulation and were inferred to be dendritic potentials. In striate cortex similar slow potentials were assigned to impulse propagation along dendrites (although no records of these responses were published). Tasaki et al. (1954b) concluded that all other forms of activity, axon-spikes, cell body spikes, and synaptic potentials could not contribute to the production of brain waves. From their analysis, they suggested the following: “We propose, therefore, a hypothesis that brain waves such as Berger rhythms, are the summations of rhythmical dendritic potentials” (p. 422). The properties of dendrites enunciated by Chang and Tasaki and others differed from those detailed later by Clare and Bishop, (1955). Bishop’s theory of the function of cortical dendrites may be viewed in relation to a more general formulation of the “natural” history of the nerve impulse (Bishop, 1956). Bishop views the all-or-none conductile activity of axon as a special case of the general property of excitability developed chiefly under the necessity of action over distances. In contrast “the chief and most characteristic functions of nervous and other excitable tissue are performed by means of graded responses” (Bishop, 1956, p. 377). Thus, integrative activity of neuronal organizations in the cerebral cortex is largely carried out in a “more-or-less” (graded) fashion rather than in an all-or-none explosive manner. NATURE OF ELECTROCORTICAL POTENTIALS

1. Properties of Cortical Dendm’tes According to Bishop and Clare (1) When stimulated directly, dendrites can conduct a 15 msec negative potential upwards (antidromically) but, ordinarily, not downwards (Bishop and Clare, 1953). Downward conduction is inferred to be decremental, local, and not all-or-none, but can be propagated with repetitive stimulation (Clare and Bishop, 1955). (2) Absence of refractoriness is

08

DOMINICK P. PUBPURA

shown by responsiveness to a second stimulus before the first response subsides. (3) Following a 20 msec period of facilitation, apical dendrites exhibit depression which can be overcome by repetitive stimulation. (4) Dendritic potentials summate over wide ranges; repetitive stimulation produces sustained surface negative d.c. potential shifts (cf. also Goldring et al., 1958, and Purpura, 1958). (5) As a consequence of (4),dendritic depolarization accounts for waves of any form and frequency in the electrocorticogram. (6) Synaptic activation of the same dendrites at different loci via afferent paths composed of different fiber types may produce different varieties of evoked potentials, (7) “The essential features of central nervous system rhythmicity are predominantly functions of the excitability cycle of dendritic synapses, such as may be exhibited in relative isolation in the recruiting response of apical dendrites of cortex” (Clare and Bishop, 1956). 2. Nature of the “Dendritic Potential” According to Bishop and Clare

The graded, locally propagated potential observed in apical dendrites was believed to be related to two varieties of graded potentials; (1) the local response of axons from which an all-or-none spike arises, and (2) the postsynaptic potential. The time characteristics of dendritic potentials were recognized as being similar to p.s.p.’s. On the other hand, since data were already a t hand indicating that postsynaptic potentials were not electrically excitable (cf. Altamirano and Grundfest, 1954), Clare and Bishop suggested that dendritic potentials had characteristics similar to both types of graded responses. Synaptic activation of dendrites was merely one way in which dendrites could be excited. Electrical excitability of dendrites had presumably been demonstrated previously. I n later formulations (Bishop, 1956, 1958), more details of the exposition were presented and similarities in behavior noted between pyramidal cell dendrites and dendrites of crayfish sensory neurons (Eyzaguirre and Kuffler, 1955a, b). Although Chang, Tasaki, Clare, Bishop, and others viewed dendritic potentials in different ways, all were in agreement on two points: (1) apical dendrites were electrically excitable and capable of conducting an electrical response of much longer duration than the axon-spike; and (2) all believed that the properties of dendrites could readily account for most of the electrical activity recorded from the surface of the brain. Bishop (1956), in climaxing a long and extraordinarily fruitful preoccupation with the electrophysiology of the cerebral cortex, could state with assurance that “the chief physiological business of the nervous system is transacted in graded-response elements” (p. 396). No history of the development of concepts relating to the different


69

properties of different parts of the neuron would be complete without mentioning the speculations of Lorente de N6 (1934) on this subject which derived from an exhaustive study of the organization of synaptic pathways to pyramidal cells in the Ammon’s horn. Commenting on the possible functional significance of his observations that pyramidal neurons received afferents from different sources on different parts of the neuron, Lorente de N6 (1934) wrote “. . . each synapse sets only a subliminal (chemical or other) change able of summation and second, that the conduction through the synapses is not followed by a refractory period. The subliminal changes are summated first in the dendrites and then in the surrounding of the axon. When the change reaches threshold value, an explosive discharge through the axon takes place. The axonas well as any other nerve fibre-enters in a refractory state, but the cell body and dendrites do not do so, they continue receiving and adding subliminal changes until the threshold value is reached again and the axon has recovered. To consider the anatomy of the nerve cell leads then to a conception similar to Sherrington’s central excitatory state (c.e.s.) although with an important difference ; while Sherrington accepts that the refractory period is a property of the whole cell, we must accept that the refractory period is a property of only the axon and that the body and dendrites never become refractory. If this conception of the neuron is right, then it becomes easy to understand that impulses arrived at different dendrites or a t two points of a dendrite can be summated. As far as I can see this is the greatest problem of the physiology of the nerve cell” (p. 171). 3. Dendrites or Synapses? The hypothesis that “brain waves” represent largely summated dendritic potentials has tended to displace all previous theories of cortical electrogenesis, but it should be recognized that large areas of disagreement exist concerning the fundamental nature of graded dendritic potentials. The general problem may be succinctly expressed in the following way: Since postsynaptic potentials (p.s.p.’s) represent an important variety of graded electrogenesis, to what extent can they be identified with most, if not all, of the slow wave activity recorded from the central nervous system? How necessary is it to assume that the so-called “dendritic potential” seen in its purest form following stimulation of the cortical surface, represents a graded response with properties that are fundamentally different from those which characterize postsynaptic potentials recorded elsewhere? The answer to this question may become apparent by reconsidering some of the properties of cortical dendrites which have been inferred by Clare and Bishop. Dendrites may be

70

DOMINICK P. PURPURA

stimulated electrically or synaptically. In both instances, identical responses are evoked which reveal no discontinuities suggestive of additional processes related to the indirect mode of stimulation. That both modes of stimulating apical dendrites should result in similar, if not identical, responses requires that the electrogenic membrane of cortical pyramidal cell dendrites be so constituted as to permit an electrically inexcitable graded response (p.s.p.’s) a t one locus, and a t a different locus an electrically excitable graded response. Dendritic graded responses must be capable under certain conditions, of propagating antidromically to the dendritic tips, whereas when stimulated a t their tips, conduction downwards is ordinarily decremental. Whether dendrites conduct a 15 msec graded response more readily in one direction than another is of relatively little consequence. The chief problem is whether or not the 10-20 msec graded dendritic potential is always a postsynaptic potential, which is not to be confused with the problem of whether or not proximal portions of dendrites close to the soma are capable of supporting an all-or-none response. There is some evidence for (Cragg and Hamlyn, 1955) and against (Freygang, 1958) this latter possibility. The argument that an all-or-none response propagating antidromically from soma into dendrites can generate only graded potentials of long duration in the latter, might be immeasurably strengthened by the knowledge that such a sequence of events is not without precedence. Apical dendrites are inferred to have properties similar to those which characterize the terminals of sensory neurons (Bishop, 1956). Appropriate stimulation of the latter (by stretch or mechanical deformation) sets up a sustained electrogenesis (“generator potential”) which electrotonically propagates to more distally located spike generating membrane and there evokes repetitive discharges whose frequency is determined by the degree of stretch or deformation and consequently the magnitude of the generator potentials (Eyzaguirre and Kuffler, 1955a; Gray and Sato, 1953; and Loewenstein, 1958). The terminals of sensory nerves do not support an all-or-none spike. In the dendrites of crayfish stretch receptors some special form of non-axonal slow potential activity is inferred following antidromic stimulation (Eyaaguirre and Kuffler, 1955b). This is also presumably supported by the observation that generator potentials can be modified, though not wiped out, by antidromic stimulation. The suggestion that some form of conduction differing from axonal propagation might occur in the finer terminals of completely relaxed crayfish sensory neurons derived largely from an analysis of the time course of the spike after-negativity recorded in the soma and in the axon following paired antidromic stimuli. When recorded from the soma,

71 a second antidromic impulse invading the cell soma 3-4 msec after the first did not add a large after-negativity. This presumably indicated that the second impulse failed to penetrate the more distal finer regions of the dendrites due to the longer refractory period of dendritic terminals as compared with the cell soma and large dendrite portions. Hence, the after-negativity which survived such a short interval double invasion was attributed mainly to activity in the dendritic terminals remaining after the first impulse. I n contrast, in axons with a relatively long afternegativity, a second impulse a t a 3 4 msec interval added its own negative after-potential to that remaining after the first one. Although these results suggested to Eyzaguirre and Kuffler (1955b) that “excitation processes in the dendrites are relatively slow and contribute to the afternegativity of impulses recorded from the cell soma” (p. 134), the latter also cautiously pointed out that the long after-negativity might well be a special property of the soma region itself. In the Pacinian corpuscle there is now direct evidence that the unmyelinated terminal portions of these sensory terminals are incapable NATURE OF ELECTROCORTICAL POTENTIALS

TO GRID CHANNEL I

TO GRID CHANNEL 2

TO GRID CHANNEL I

Fm. 4. An antidromic impulse does not invade a sensory non-myelinated ending (Pacinian corpuscle, cat). A. A subthreshold mechanical stimulus produces a generator potential at active membrane sites below microelectrodes, El and En. The nerve ending is stimulated by the deflection of a pieso-electric crystal applied to the ending by means of a five glass stylus (St) El and Ea are connected to separate amplifier channels and are placed about 350 p apart in contact with the non-myelinated ending. Negativity at Et and EPsignaled by an upward deflection. Calibration: 50 pv; 1 mseo. B. The myelinated axon is stimulated electrically causing an impulse to travel antidromically towards the ending. The impulse can be traced as an all-or-nothing potential as far as the first node, but no active response is detected from the ending. Calibration: 200 pv; 1 msec (from Loewenstein, 1959).

72

DOMINICK P. PURPUBA

of supporting any variety of antidromically conducted activity. When impulses are fired from the myelinated segment into the terminals, impulse propagation fails somewhere at the first node (Fig. 4) (Loewenstein, 1959). Furthermore “the membrane of the non-myelinated ending is incapable of producing an all-or-nothing type of potential when stimulated mechanically” (Loewenstein and Rathkamp, 1958). Thus, receptor membrane of sensory terminals, like postsynaptic membrane, is electrically inexcitable (Grundfest, 195713). If, as Bishop suggests, apical dendrites are capable of supporting an electrically excitable graded response of long duration, then the properties of apical dendrites must be fundamentally different from those of sensory terminals and not a t all similar. The demonstration that generator potentials of sensory neuron terminals arise in electrically inexcitable membrane does not preclude the possibility that cortical apical dendrites may be capable of electrically excitable graded electrogenesis. Data to be presented below, however, make this likelihood improbable. C. THE “POSTSYNAPTIC POTENTIAL” THEORY OF CORTICAL ELECTROGENESIS The hypothesis that p.s.p.’s might represent the major form of activity recorded from the surface of the brain was largely derived from a consideration of the nature of slow potentials recorded in “less complexly” organized systems. (The term “less complex” is used as a matter of convenience to indicate amenability for analysis rather than simplicity of organization.) It had been known for some time (Gotch and Horsley, 1891) that afferent stimulation evoked catelectrotonic potentials in the dorsal roots of the spinal cord. Following the work of Gasser and Graham (1933) and Umrath (1933), Barron and Matthews (1938) systematically explored the nature of dorsal and ventral root catelectrotonus and concluded that the dorsal root potential was due to persisting depolarization of dorsal root fiber terminals, arising as a special kind of negative afterpotential following propagated impulses. Such a depolarization was believed to induce depolarimation of the motoneurons, which in turn, was detectable as a ventral root electrotonus. The dorsal root catelectrotonus evoked by stimulation of single dorsal root fibers was termed the “unitary synaptic potential’’ (Fessard and Matthews, 1939). An alternative suggestion was first proposed by Bonnet and Bremer (1938), who placed the site of origin of the catelectrotonus postsynaptically in neurons with which the dorsal root fibers entered into contact, the latter being in turn depolarized by the currents generated by this potential. This view was favored by Eccles (1939) and was supported later by new experimental data (Eccles and Malcolm, 1946) derived from experiments dealing with


73

the electrical responses associated with synaptic transmission in motoneurons (Eccles, 1946). A brief resume of the events preceding and following Eccles’ description of the synaptic potential of motoneurons is important to record. 1. Electrogenesis at Synapses

The almost simultaneous discovery that junctional transmission was associated with an electrical potential without parallel in muscle and nerve fiber (Gopfert and Schaefer, 1938; Eccles and O’Connor, 1939) may be viewed as one of the most significant electrophysiological events since the description of the all-or-none action potential of nerve. In the following years, Eccles described the synaptic potential of sympathetic ganglia (1943) and finally the synaptic potential of motoneurons (1946). I n all cases, it was inferred from the data that the arrival of an impulse a t presynaptic terminals, evoked a catelectrotonic potential in the postsynaptic membrane, which if of sufficient magnitude, resulted in the generation of an all-or-none conducted discharge in the cell or muscle fiber. Recorded extracellularly, excitatory, depolarieing p.s.p.’s all had similar properties (Fig. 5) and when techniques became available for their demonstration with intracellularly located microelectrodes, these properties were explored in detail (motor end-plate; Fatt and Katz, 1951; sympathetic ganglion, R. M. Eccles, 1955; cat spinal motoneurons, Brock et al., 1952). The latter report also contained a description of the properties of inhibitory postsynaptic potentials of spinal motoneurons. The modern history of electrophysiology ushered in by the employ of microelectrodes which permitted intracellular recording in motor endplates, muscle fibers, motoneurons, interneurons, dorsal root fibers, axons, as well as a host of other excitable cells, has largely clarified many of the problems of excitation, inhibition, impulse and receptor function as well as providing fundamental data on the biochemical and biophysical processes involved in the various electrogenic phenomena characteristic of excitable cells. (Numerous critical reviews on the subject are available for the student to pursue problems of special interest. Throughout this paper, a number of the more relevant critical reviews have been called to the reader’s attention. It should be noted that clarification, however, is not always synonymous with resolution of a problem!) The discovery that slow potentials recorded from the spinal cord were largely due t o p.s.p.’s generated in various groups of interneurons or motoneurons still raised the question as to what fraction of the dorsal or ventral cord and root electrotonus could be attributed to after-currents in dorsal root terminals or motoneurons. Persisting activity in dorsal root terminals was shown by Lloyd and McIntyre (1949) to account for

74

DOMINICK P. PURPURA

a fraction of dorsal root electrotohus. Recently a contrary view of the origin of dorsal root electrotonus has been taken by Wall (1958) who claims that action potentials in postsynaptic cells need not be assumed to account for the large negative dorsal root potentials. The latter’s work extends earlier observations of Howland et al. (1955) concerning the possible nature of “presynaptic” inhibition. Maps of potential “sources’’ and “sinks” in the immediate vicinity of dorsal root terminals suggest that marked depression of excitability apd blockade of impulse conduction a t terminals may play a role in reflex inhibition independent of postsynaptic effects.


75

The nature and possible functional significance of ventral root electrotonus appears to have been more amenable to analysis. The ventral root catelectrotonus following orthodromic stimulation of motoneurons largely reflects the p.s.p.’s of motoneurons. Coombs et al. (1955) report that orthodromic stimulation of motoneurons produces a significant “after-hyperpolarization” of motoneurons, but this has not been observed by Frank and Fuortes (1955). Antidromic stimulation of motoneurons results in a significant ventral root and cord anelectrotonus which has been ascribed by Lloyd (1951) to nonuniform behavior of different parts FIQ. 5. Electrical characteristics of extracellularly and intracellularly recorded postsynaptic potentials. A. End-plate potential recorded with surface electrodes from curarized frog sartorius muscle in situ. Amplitude of the response approximately 0.1 m v ; duration, 15 msec (from: Gopfert and Schaefer, 1938). B. Depolarizing postsynaptic potential in stellate ganglion of curarized cat recorded between ganglionic origin and distal end of cardiac nerve. Duration of response under these conditions is approximately 150 msec (from : Eccles, 1943). C. Depolarizing postsynaptic potential evoked by stimulus to 8th dorsal root, recording from 8th ventral root at cord origin. Reference electrode on distal cut end of the root. Cat, 80 mg/kg Nembutal. Duration of response approximately 10 msec (from: Eccles, 1946). D. Depolarizing postsynaptic potentials obtained in a biceps-semitendinosus motoneurone following stimulation of Group I, afferents. Intracellular recording (negativity downwards). Magnitude of the excitatory p9.p. is approximately 15 mv, duration 15 msec (cf. Eccles, 1957). E. Depolarizing postsynaptic potential of apical dendrites (cat) cerebral cortex following stimulation of cortical surface (negativity upwards). Duration, 20 msec (from: Purpura et aZ., 1957a). F. Hyperpolarizing postsynaptic potential recorded intracellularly from cortical neuron following stimulation of n. ventralis lateralis of thalamus. Rapid decay of potential is due to short coupling time-constant. Voltage calibration (vertical bar) 5 mv, time 10 msec. Duration of response under these conditions 15 msec (negativity downwards) (from : Branch and Martin, 1958). G. Surface positive hyperpolarizing p.s.p. of apical dendrites (cat cortex) “unmasked” following topical application of -y-aminobutyric acid. Further explanation in text. (From: Purpura et aZ.,’1957a.) H. Intracellularly recorded depolarizing p.s.p.’s set up in a gastrocnemius motoneurone by repetitive orthodromic volleys. The depolarizing synaptic electrogenesis is sustained during the repetitive stimulation which lasted approximately 110 msec (from: Brock et al., 1952). I. Sustained synaptic electrogenesis induced by repetitive stimulation (100/sec) of cortical surface. Stimulus duration 5 sec. Early summation of dendritic depolariaing p.s.p.‘s is halted by an opposing electrogenesis which is rapidly overcome. At cessation of stimulus (arrow) negative shift persists and slow oscillations in d.c. recording are observed. Maximum shift attains an amplitude of 0.5 mv. Further explanation in text (from : Purpura, 1958).

76

DOMINICK P. PURPURA

of the cell. The after-currents generated by antidromic stimulation of motoneurons is accompanied by inhibition or facilitation of orthodromically evoked reflex activity (Renshaw, 1941). Renshaw considered it likely that antidromic inhibition could be due to the polarizing actions of the after-currents or collateral activation of adjacent (inhibitory) elements. Lloyd (1951) attributed the depression of adjacent motoneurons by antidromic volleys to the after-currents only, On the other hand it has been demonstrated that antidromic stimulation of ventral roots produces collateral activation of interneurons (“Renshaw cells”) which in turn evoke hyperpolarizing p.s.p.’s in the same and adjacent motoneupons (Eccles et al., 1954). The fundamental question, therefore, appears to be: What is the functional significance of ventral root electrotonus? Brooks and Wilson (1958) have shown that antidromic inhibition of adjacent motoneurons is entirely attributable to collateral activation of inhibitory interneurons and not to ventral root electrotonus as proposed by Lloyd (1951). The work of Brooks and Wilson provides an example of the judicious use of appropriate pharmacological techniques t o dissect the various functional components of complex potentials. It was known that synapses between axon collaterals and “Renshaw cells” were cholinergic and inactivated by dihydro-p-erythroidine (Eccles et al., 1954). Simultaneous recording of ventral root electrotonus and repetitive discharges evoked by an antidromic stimulus revealed that abolition of recurrent activity by intravenous administration of dihydro-a-erythroidine was not accompanied by changes in ventral root electrotonus (Brooks and Wilson, 1958). Although the experiments defined the functional significance of recurrent inhibition (viz. to confine stretch reflexes t o the paths of afferent origin) the authors also proved that after-currents of motoneurons exerted no inhibitory actions on adjacent elements. ’ It must be concluded from a brief review of the data that slow potential activity in some afferent terminals may contribute to the overall electrotonic potentials recorded from the surface, depths, or roots of the spinal cord and may make contributions to the slow potential activity recorded from the surface or depths of the cortex as suggested by Li e t al. (1956a). Compared to the known characteristics of postsynaptic potentials, however, these contributions are likely to be of incidental importance. In the spinal cord the evidence overwhelmingly favors the hypothesis originally proposed by Bonnet and Bremer (1938) that the major portion of electrotonically transmitted activity recorded from the ventral and dorsal roots following repetitive afferent stimulation is a sustained postsynaptic electrogenesis.


77

2. Theoretical Application of Synaptology to Cortex An unwritten law formulated a t a time when neurophysiology was primarily concerned with the study of axon properties goes somewhat as follows: “One cannot tell a process from a potential.’’ The discovery that excitatory events a t synapses were associated with specific bioelectric signals (via. depolarizing, catelectrotonically propagated potentials generated in postsynaptic membrane by chemical transmittors) cast considerable doubt on the validity of this “principle.” If, as it was assumed, evoked slow potentials in spinal elements were largely due to processes operating at synapses, then why could it not be assumed that similar processes accounted for the slow potentials recorded from other parts of the central nervous system and, particularly, the cerebral cortex. Theoretical formulations soon followed this assumption (Bremer, 1949 ; Eccles, 1951) which differed in detail. Bremer’s observations on “slow waves” of the spinal cord seemed to leave “no doubt concerning the existence of rhythmical fluctuations in the electrical polarization of the synaptic surfaces of nerve cells, underlying and causing (by their catelectrotonic effect) the rhythmical discharges of centrifugal impulses” (1953, p. 42). But he admitted that “the extension of these observations on the spinal cord-where electrotonic recording of the synaptic potentials is feasible-to the situation in the cortex represents an extrapolation that has still to be justified experimentally” (Bremer, 1953, p. 42). Still this did not restrain speculation on how sustained synaptically induced depolarizations might give rise to slow waves in the cortex (Bremer, 1949). Bremer’s “auto-rhythmicity” theory of brain waves, however, differed significantly from that proposed by Libet and Gerard (1941). Bremer (1953) defined neuronal auto-rhythmicity “as the capacity to respond t o humoral and to synchronous nervous stimuli by a regular rhythmical succession of bioelectric potentials, associated with oscillations of excitability and (eventually) discharges of impulses.” It will be recalled that according to Libet and Gerard (1938) “spontaneous” fluctuations in membrane polarization could occur independently of any presynaptic stimulus. The repetitive response of neurons to the persisting stimulus of a synaptically induced membrane depolarization could account, according to Bremer, for all repetitive activity of neurons (viz. after-discharge, etc.) . Bremer denied that impulses circulating in closed re-entry circuits were of any significance as a determinant of repetitive neuronal activity. Bremer (1958) has extended some of his original views but otherwise they are essentially unmodified.

78

DOMINICK P. PURPURA

Perhaps because of a greater familiarity with the characteristics of postsynaptic potentials under various conditions of root recording (Eccles, 1946), “focal” recording (Brooks and Eccles, 1947a), or intracellular recording (Brook et al., 1952), Eccles (1951) stated that the “identification of the superficial response (of cortex) as a synaptic potential is not contraindicated by any other experimental evidence.” Basing his interpretations on data published by Chang (1951a) and Burns (1950, 1951), Eccles proceeded to calculate the synaptic delay involved in setting up the postsynaptic “dendritic potential,” proposed synaptic mechanisms to account for the surface positive response of cortex (Adrian, 1936), as well as a variety of evoked cortical potentials, and indicated in detail methods to be employed in proving these hypotheses. To explain the rhythmical nature of brain waves, Eccles (1951) proposed a compromise between the auto-rhythmicity hypothesis of Bremer (1938, 1949) and the closed-chain theory of Lorente de N6 (1935). He suggested that “the frequency of the rhythm would be set by two factors, on the one hand the time curve of recovery of the neuron, and on the other hand the intensity of synaptic bombardment, which is a function of the activity of closed neural chains that have no fixed circuits.” Thus, although Eccles considered p.s.p.’s the major variety of cortical surface activity, his theory was by no means monolithic. He proceeded to define the ways in which synaptic potentials, all-or-none discharges, and after-potential changes might all contribute to this activity. But, according to Eccles’ theory, there would be no need to assume the existence of slow potentials in dendrites different from p.s.p.’s to account for brain waves. Since the formulation of the p.s.p. and dendritic theory of brain waves, no serious attempts to evolve new hypotheses have been forthcoming. Additional conjectures have contained elements of many theories with some suggestions as to how a more fruitful approach t o the study of cortical electrophysiology might be pursued. Verzeano and Calma (1954) have called attention to the possibility that certain varieties of rhythmical slow potentials, vis. 8-l2/sec “spindle bursts” in barbiturized animals might be compounded of p.s.p.’s and summated after-potentials ; p. s. p.’s representing the initial waves of the burst and summated afterpotentials constituting later components of the burst. A more provocative and all-embracing proposal has been summarized as follows: “For radial potential fields of dipole type, long apical dendrites are the most probable sources in the cortex. For explanation of the brain rhythms slow processes of excitation with d.c. potential fields, pre-synaptic and local potentials, electrotonic spread and/or decremental conduction in the neuropil and dendritic plexuses in connection with interneuronal activity

NATURE OF ELCCTROCORTICAL POTENTIALS

79

must be assumed” (Jung, 1953). Such a statement could hardly have been possible without the realization that the electrogenic characteristics of excitable cells were far more complex than imagined when axons were considered samples of nervous tissue.

3. Summary The inadequacy of axonological interpretations of electrocortical potentials was demonstrated by three major discoveries: (1) slow potentials and brief axon-like spikes were two distinct and independently recordable bioelectric responses of neurons; (2) the properties of dendrites were distinctly different from axons; and (3) electrocortical activity largely represented the summated “graded potential” activity of cortical neurons rather than envelopes of all-or-none discharges. Further characterization of the nature of graded potentials indicated that they might be due to either electrical and synaptic stimulation of dendrites or only the latter, in which case, the ((unit”dendritic potential of 10-20 msec duration could be viewed exclusively as a postsynaptic potential. If dendritic potentials were postsynaptic potentials, then it followed that the spontaneous rhythms recorded from the cortical surface might be summated p.s.p.’s. Close examination of the experimental and theoretical developments of the period prior to 1953 fails to justify the pronouncement that “after over twenty years of study the neurophysiological basis of the rhythmic oscillations in electrical potentials which can be recorded from the cerebral cortex (remained) poorly understood” (Li and Jasper, 1953).

Part

II. The Synaptic Origin of Cortical Surface Potentials I. Introduction

The following sections are devoted to a further elaboration of the hypothesis that slow potentials recorded extracellularly from the cortical surface or in the vicinity of single or small groups of neurons, are postsynaptic potentials. Application of the general principles of synapse physiology (and pharmacology) not only supports the hypothesis, but extends it. Support is provided by observations on the effects of various pharmacological agents on spontaneous and evoked cortical potentials ; extension by the observations that hyperpolarizing as well as depolarizing p.s.p.’s are prominent components of electrocortical potentials. In addition, the fundamental differences between axodendritic and axosomatic electrogenesis will be detailed within the framework of a gen-

80

DOMINICK P. PURPURA

era1 discussion on the relationship of cortical synaptic and conductile activity and the organization of various afferent pathways to the cerebral and cerebellar cortex. The extent to which the general hypothesis can satisfactorily account for a variety of extant experimental observations will also be indicated.

II. Nature of Cortical Synapses

The morphology of synapses in the cerebral cortex has, until recently, escaped serious attention. The problem has been summarized in the following manner: “One of the major problems in cortical histology is the nature of the terminations of an axon on the cell to which it discharges its impulses. In the spinal cord many axons can be seen to have Specialized ends or terminal buttons of various kinds. In general, such specialized endings have not been seen in the cortex. They are either absent or fail to stain” (Sholl, 1956, p. 16). In Golgi or Golgi-Cox preparations, the dendrites of cortical pyramidal cells “exhibit” small club-ended filaments or spines (gemmules) which often appear to make contact with parallel-running axons. Cajal (1934) concluded after much consideration that the dendritic spines were not connections between axon and dendrite, and suggested that the club ending represented a local deposition of metal. Others, however, have developed special theories of cortical synaptology based on observations concerning the uniform, minimal separation of fibers seen in Bielschowski impregnations (cf. Bok, 1956). The apparent absence of “boutons terminaux” in the cerebral cortex and the close approximation of dendritic gemmules to axons has prompted the suggestion that cortical synapses might be different from spinal cord synapses in that the expanded terminal might be formed by the dendrites (postsynaptic element) and not by the presynaptic terminal (Sholl, 1956). Such articulations between pre- and postsynaptic elements are commonly observed at invertebrate junctions (Robertson, 1957). It is difficult to evaluate what influence the failure to demonstrate synaptic knobs on cortical neurons has exerted on interpretations of cortical function. But this influence cannot be denied. Berkley (1897) suggested that the heads of gemmules of cortical dendrites were principally the organs receptive of nerve impulses from other neurons. Chang (1952) considered the suggestion plausible, and extended the hypothesis to account for the ineffectiveness of presynaptic volleys to initiate discharges in dendrites. The ineffectiveness would derive from the possibility that postsynaptic excitation initiated in the


81

gemmules would be markedly attenuated by the higli ohmic resistance of the fine spines. In addition, the protrusions of the gemmules might prevent the synaptic knobs from touching the dendrite surface directly. Thus, axodendritic synapses of cortical pyramidal cells could never effectively discharge the neuron (Chang, 1952, 1956). The gemmule theory of dendritic synapses has been iiiore enthusiastically supported by investigators in the Soviet Union. Beritov (1956), like Chang, postulates that synaptic activation of dendrites is effected through the dendritic gemmules where weak local potentials are generated which electrotonically spread to the cell bodies. The presence of the thorns or spines prevents these currents from leaking out in the dendrites, but in the cells current leaks produce anelectrotonic depression or inhibition of the cells. Thus, according to Beritov, pyramidal cells are always inhibited by dendritic activation (in contrast to stellate cells which are always excited by dendritic activation). The fallacy of these arguments and similar ones proposed by Roitbak (1955) has been pointed out by Grundfest (1958a). Sarkissov (1956) asserts that the dendritic thorns have additional properties which permit thein t o expand and contract depending on metabolic and environmental changes. In this way the receptor surface of the dendrites may be varied under different conditions. It is not unreasonable for this reviewer to assume that interpretations of dendritic potentials as electrically excitable responses (Chang, 1951a; Clare and Bishop, 1955, 1957; and Roitbak, 1955) may have been conditioned in part by: (1) previous failures to demonstrate dense coverings of synaptic knobs on cortical dendrites; and (2) the classic picture of cortical neurons as seen in the usual Golgi preparations (cf. Chang, 1955a). There is now some evidence that synaptic end-feet are densely packed on cortical pyramidal cells (Armstrong and Young, 1957). The apical dendrites of cortical neurons in particular, appear to be covered with 80 or more 1 - 1 . 5 ~ synaptic knobs per 1 0 0 ~length of dendrite (Fig. 6) . Furthermore, electron microscopy reveals that cortical axodendritic and axosomatic synapses are fundamentally similar to those elsewhere in the central nervous systeni (Palay, 1956), although marked differences in the number, morphology, and location of boutons terminaux in the cerebral cortex are detectable with silver impregnation techniques (Smythies et al., 1957). Only particular kinds of boutons may, however, be demonstrable by this method (Smythies, 1958). From histological considerations it seems unlikely that stimulation of the cerebral cortex could directly excite apical dendrites if the latter are densely covered by synaptic end-feet (Fig. 6) and surrounded by bundles of presynaptic axon terminals. On the contrary, as will be shown

82

DOMINICB P. PURPURA

F I ~6.. Demonstration of synaptic end-feet in cerebral cortex by means of :I stain that preserves the mitochondria they contain. (1) End-feet around cell bodies and dendrites seen in section parallel to the pial surface; cat’s visual cortex. (2) Pyrsmidal cell and its apical dendrite from cingulate cortex of the cat (from: Armstrong and Young, 1967).

below, the surface negative response is more reasonably viewed as arising from stimulation of the presynaptic terminals with subsequent activation of dendritic synapses; the evoked response having all the characteristics of a postsynaptic potential.

111. Applied Synaptology

The properties which distinguish transmissional (synaptic) from conductile electrogenesis have been recently analyzed in terms of a general hypothesis that postsynaptic membrane is electrically inexcitable (Grundfest, 1957b). In the following analyses, the interpretations are compatible with the assumption that the properties of synapses account for the alterations in electrocortical potentials induced by a variety of physiological and pharmacological procedures.

NATURE O F ELECTROCORTICAL POTENTIALS

83

A. THE DEPOLARIZING POSTSYNAPTIC POTENTIALS OF APICAL DENDRITES The possible nature of the 10-20 msec graded, surface negative cortical response evoked by surface stimulation has been noted above (Part I). By inspection alone (Fig. 5) the reader may appreciate for himself why identification of the response as a postsynaptic potential was intuitively inferred (Eccles, 1951). The resemblance between the hyperpolarizing p.s.p.’s of cortical neurons (intracellularly recorded) and the hyperpolarizing p.s.p.’s of apical dendrites should also be noted (Fig. 5F, G.) . But mindful of the unwritten law that “one cannot tell a process from a potential,” other characteristics must be noted. One of these is shown in Fig. 51, via. the ability of 10-20 msec surface negative responses to summate following high frequency surface stimulation (Clare and Bishop, 1955; Goldring et al., 1958; and Purpura, 1958). In d.c. recordings sustained surface-negative cortical shifts are observed which have a striking resemblance to the sustained intracellularly recorded membrane depolarization induced by orthodromic stimulation of cat motoneurons (Fig. 5H). I n the latter, sustained electrogenesis is attributable to the excitatory p.s.p. Similar events have been shown for the end-plate potentials of muscle (Fatt and Katz, 1951) and p.s.p.’s recorded extracellularly (Eccles, 1943) and intracellularly (R. M. Eccles, 1955) from sympathetic ganglion cells. The cortical effects are satisfactorily explained in the following manner: Weak surface stimulation excites presynaptic cells or fibers terminating synaptically on apical dendrites (Eccles, 1951; Purpura and Grundfest, 1956). The evoked synaptic electrogenesis developing in electrically inexcitable membrane may accumulate and be sustained throughout the period of stimulation. Two differences between the cortical records and those obtained from cat motoneurons require additional comment (cf. Fig. 5H, I).Unlike the latter, the cortical records show early discontinuities and persistence of sustained negativity after cessation of a repetitive stimulus. These can be ascribed to the simultaneous development of (a) opposing hyperpolarizing synaptic potentials, and (b) the activation of additional elements in cortex which in turn contribute to sustained repetitive synaptic bombardment of apical dendrites (Purpura, 1958). Evidence for the existence of hyperpolarizing p.s.p.’s “masked” in the overt surface negative response will be described below. Having considered, as a preliminary, some of the characteristics of the surface negative potential of cortex, additional arguments can be summarized, some of which logically derive from those presented in Part I. If the surface negative potential of cortex is inferred t o be a p.s.p., then it must be generated in membrane which is electrically inexcitable.

84

DOMINICK P. PURPURA

Blockade of all cortical synapses with suitable pharmacological agents should reduce or abolish the dendritic potential at a time when the electrical excitability of cells or axons is still demonstrable. In artificially ventilated, succinylcholine-paralysed, unanesthetized cats, intravenous injection of d-tubocurarine chloride (3-5 mg per kilogram) results in a prompt reduction or abolition of the surface evoked “dendritic potential” (Fig. 7, I) which is correlated with loss of spontaneous electrocortical activity (Purpura and Grundfest, 1956). At the height of the synaptic blockade, subsurface stimulation or antidromic pyramidal tract stimulation fails to evoke surface negativity (Fig. 7, I) and only the direct pyramidal tract response can be elicited by strong surface stimulation. Correlation of the loss of synaptically relayed (Patton and Amassian, 1954) orthodromic pyramidal tract activity with the loss of antidromic surface negativity indicates that the latter is postsynaptically evoked in membrane which is not electrically excitable (Purpura and Grundfest, 1956).

The effects of high concentrations of d-tubocurarine on surface evoked responses reported by Purpura and Grundfest (1956) have been attributed by Brinley et al. (1958) to changes in stimulus strength and electrode contact resulting from the secondary brain shrinkage which accompanies hemodynamic alterations. The conclusion that electrode displacement is responsible for the changes observed by Purpura and Grundfest is presumably based on the observations that blood pressure changes alone do not result in immediate significant alterations in surface-evoked responses in agreement with the observations of Purpura and Grundfest (1956) and the earlier findings of Adrian (1936). The conclusions of Brinley e t al. (1958) are implausible for a number of reasons. During the period when surf ace-evoked negativity is markedly depressed or absent no changes are detected in evoked potential components representing directly excitable responses (Fig. 7, I), indicating that mechanical displacement of subsurface or pyramidal tract electrodes did not occur. If electrode-cortical surface contact was altered, this would be signaled by the appearance of a change in stimulus artifact as well as a change in the strength of a direct cortical stimulus. None of these changes are observed (Fig. 7, I). These and a variety of other data reported by Purpura and Grundfest (1956, 1957b) cannot be accounted for in terms of changes in electrode contacts, especially when, under certain specific conditions, similar marked hemodynamic changes are accompanied by augmentation of surface-evoked responses (cf. below). The observations of Purpura and Grundfest (1956) have been confifmed and extended by Fan and Feng (1967) who studied the parallel


85

FIQ.7. I. A. Blockade of dendritic response evoked by stimulation of cortical surface. Bipolar stimulating electrodes were about 1.0 mm away from recording lead on anterior suprasylvian gyrus. Indifferent electrode was in sub-cortical white matter. 1, initial response, entirely surface negative rising out of shock artifact; 2, 50 sec after injecting 3.0 mg/kg d-tubocurarine into femoral vein; 3, 20 sec later; 4, a t 5 min; and 5, a t 20 min. Horizontal bar, 20 msec. B. Electrical inexcitability of synaptically blockaded dendritic response. Stimuli applied 0.8 mm below cortical surface in anterior sigmoid gyrus and recording lead on surface directly above. Indifferent lead on bone over frontal sinus. 1, initial response is a positive deflection, followed by dendritic negativity; 2, 45 aec after injection of 2 mg/kg d-tubocurarine only positive component remained; 3, recovery after 90 sec. Horizontal bar 20 msec. C. Direct and synaptic components of antidromic and orthodromic activity in pyramidal system. 1-3, responses at cortical surface to stimulating pyramidal tract in medulla. Indifferent electrode on frontal sinus. 4-6, activity in tract on stimulating cortex. 1,4, initial responses. 2,5, 5 min after injection of 3.0 mg/kg d-tubocurarine. 3,6, 20 min later. Horizontal bar 10 msec for records 4-6 (from: Purpura and Grundfest, 1956). 11. Blockading effects of d-tubocurarine on surface negative dendritic responses evoked 2 mm and 5-6 mm from stimulating site. “Near” responses on left; “far” responses on right. Within 15 sec after intravenous administration of 3.3 mg/kg d-tubocurarine both responses are reduced and nearly abolished at 92 sec. Recovery is virtually complete after 13 min (from: Fan and Feng, 1957).

86

DOMINICK P. PURF’URA

effects of intravenous d-tubocurarine and other drugs on the surface response evoked in the immediate vicinity of the stimulating electrodes (“near” response) and that evoked 6 mm away (“far” response) (Fig. 7, 11).Fan and Feng’s observations call attention to the relationship between the “near” surface negative response and those recorded at varying distances up to 10-12 mm from the locus of stimulation. As noted earlier, Chang (1951a) attributed the responses recorded within 2-3 mm from the stimulating electrodes to direct electrical stimulation of dendrites. Those recorded beyond 3 mm were inferred to be compounded of directly and indirectly evoked activity. Inspection of his data (Fig. 3, I) clearly indicates that the “far” responses were of greater amplitude than the “near” responses. It has generally been claimed (Burns, 1951; Chang, 1951a) that progressive displacement of the recording electrode along the cortical surface from the stimulating site resulted in a linear reduction in the surface negative potential associated with a progressively increasing latency. The relationship between the “near” and “far” response has been viewed differently by Brooks and Enger (1959). According to the latter, the linear decay of the surface negative response obtains only with weak stimuli. Stronger stimuli produce re-enforcement of the response at 6-9 mm. With maximal stimulation, two points of reenforcement are detectable at 5 and 10 mm (Fig. 3, 11).The re-enforced responses apparently develop with shorter latencies than can be expected for uniform conduction in homogeneous elements. Brooks and Enger (1959) suggest from these and other data that slowly conducted surface-negative responses are due to direct or synaptic activation of pyramidal cells while responses of shorter latency are initiated postsynaptically in pyramidal neurons by conductors, presumably axons of stellate cells, with lengths of 5 mm. While new data are provided to explain some of the factors involved in the production of the “far” responses to surface stimulation (which are identified as p.s.p.’s) no new information concerning the specific nature of the “near” responses is suggested by their studies. One of the questions raised by studies dealing with the “propagation” of the surface negative response along the cortical surface seems justified. When does a “near” response become a “far” response or vice versa? Assuming that the “near” response is directly evoked in apical dendrites, the question then becomes, a t what distance does the surface negative response acquire all the characteristics of a p.s.p.? The work of Fan and Feng (1957) indicates that the “near” response is generated by a process that is not fundamentally different from that involved in the production of the “far” response. I n the hippocampus aff erents from the subiculum and entorhinal cortex constituting the direct temporo-ammonic or perforant pathway terminate on apical


87

dendrites of pyramidal cells and elements in the Fascia dentata (cf. Lorente de N6, 1934). Extracellularly located micropipettes inserted t o the level of the apical dendrites of hippocampal pyramidal neurons record 10-20 msec monophasic negative responses (Fig. 8 ) following stimulation of the perforant fibers (Renshaw e t al., 1940; Purpura, 1959a,b). Repetitive stimulation of the perforant pathway results in initial augmentation of the focal negativity (Fig. 8, 3).With continued repetitive stimulation brief deflections appear which increase in magnitude as the slow depolarizing process initiates conductile responses in a progressively increasing population of active neurons (Fig. 8, 4-12). In view of the fact that the 10-20 msec graded negative responses initiated in apical dendrites by afferents from the subiculum represents excitatory events prerequisite for cell discharge such slow waves are inferred to be depolarizing p.s.p.’s generated in dendritic elements of Ammon’s horn pyramidal neurons (Purpura, 1959a,b). Thus synaptic activation of apical dendrites in archicortex evokes excitatory p.s.p.’s which are in every respect identical with the so-called “direct” or “indirect” responses of neocortex evoked by surface stimulation. It is concluded, therefore, that the 15-20 msec “unit” negative responses of apical dendrites arise via synaptic activation of electrically inexcitable membrane and that both the “near” and “far” responses are p.s.p.’s (Purpura and Grundfest, 1956). The use of the term “direct cortical response” (Ochs, 1956, 1958) in conjunction with the superficial negative response of cortex evoked by a weak electrical stimulus has introduced considerable confusion. The surface negative response is generated in apical dendrites “indirectly,” viz., by electrical stimulation of conductile pathways which synapse on secondary elements. The term “direct” cortical response (DCR) is a misnomer when applied to the superficial negative response. When used to connote a response evoked by direct electrical stimulation it is applicable only in connection with activity initiated by stimulation of electrically excitable elements (which may activate, synaptically, other neurons involved in the production of an evoked response). Thus, the “direct cortical responses” of Ochs (1956, 1958) are similar to those shown by Landau and Clare (1956) which result from excitation of elements in the cortical depths by strong surface stimulation. The distinction between “direct” and “indirect” responsiveness has important implications clearly recognized for the first time by Claude Bernard. The conclusion that apical dendrites of pyramidal cells are electrically inexcitable is further supported by the observation that synaptically evoked activity in hippocampal pyramidal cells does not ordinarily propagate in the apical dendrites of these elements (von Euler et al.,

88

DOMINICK P. PURPURA

1958), contrary to the findings of Cragg and Hamlyn (1955), nor can somatofugal propagation be detected in neocortical dendrites (von Euler and Ricci, 1958). The upward shift of dipoles generated in cortex by peripheral and distant stimulation has been accounted for by serial synaptic activation of superficial portions of apical dendrites rather than by local circuit propagation of graded responses in dendrites (Amassian

FIO.8. Responses evoked from level of apical dendrites of Ammon’s horn pyramidal neurons (CAB region, Lorente de N6, 1934) following stimulation of perforant fibers in the subiculum. Extracellularly-located saline-filled micropipette 1.3 mm below ependymal surface. Low frequency stimulation, Ob/sec, (1) evokes focal negativity which increases in magnitude and duration during repetitive stimulation at higher frequencies, lO/sec ( 8 ) and 25/sec (5-6).Conductile discharges develop during rising phase of focal e.p.s.p.’s. When stimulus frequency is maintained a t lO/sec, precise latencies are maintained (7-18).Conductile responses gradually disappear during continued stimulation a t lO/sec (13-16). Slight depression of responses evoked with 0.5/sec stimulation observed 20 sec after 16. Calibration 100 cps, 1 mv (from Purpura, 1959b).

et al., 1955), but this hypothesis has been criticized by Mountcastle et

al. (1957).

B. ARE CORTICAL PYRAMIDAL NEURONSELECTRICALLY INEXCITABLE? The foregoing account reviews some of the indirect evidence that apical dendrites of cortical pyramidal cells are electrically inexcitable (Purpura and Grundfest, 1956). Irrespective of the manner in which they are evoked (viz. “direct” cortical stimulation, “antidromic” stimu-


89

lation, or afferent stimulation) the dendritic potentials of cortical pyramidal cells are presumably postsynaptic potentials. It seems reasonable to inquire whether the entire soma-dendritic membrane of the cortical pyramidal cell may be electrically inexcitable. Data suggesting electrical inexcitability of the soma-dendritic membrane of lateral geniculate neurons (Freygang, 1958) and spinal motoneurons (Freygang and Frank, 1959) have been obtained, the latter by means of concentric microelectrodes permitting simultaneous registration of intracellular and extracellular potentials. Freygang suggests that most of the soma-dendritic membrane of lateral geniculate cells is excited synaptically to produce a postsynaptic potential, but is not excitable electrically and hence cannot support a propagating spike. The hypothesis is based on the observations that following orthodromic stimulation : (1) the only reliable intracellularly recorded activity is from axons ; and (2) microelectrodes located extracellularly close to, or in contact with the soma-dendritic membrane record three phases of outwardly directed current followed by a phase of inwardly directly current. The first component is ascribed to synaptic activity, subsequent ones to activity in the initial axonal segment and a restricted region of high threshold membrane on the soma. Freygang’s independent analysis of the extracellularly recorded action potentials of lateral geniculate cells agrees with the interpretation proposed earlier by Tomita (1957) to account for the polarity of extracellular action potentials of single photoreceptor cells (Limulus). It had been known that the slow ommatidial potential was always negative when recorded extracellularly and positive (depolarizing) when recorded intracellularly (Hartline et aZ., 1952). Spikes superimposed on this internal positivity were initially positive (Hartline et al., 1952). Tomita confirmed these observations and extended them t o include an analysis of the extracellularly recorded spikes which were also found to be initially positive. Impulses recorded from the nerve strand were initially negative. Negative pulses were more effective than positive ones in eliciting action potentials from the nerve strand and vice versa when the stimulating electrode was within the ommatidium indicating that spikes were not generated in the cell body but a t a more proximal locus by an outwardly directly current through the membrane. From these data Tomita concluded “the cell body appears to lack the property t o discharge impulses.” Spikes initiated in the axonal processes of most types of lobster cardiac ganglion cells (“follower cells”) cannot invade the soma except as a small electrotonically conducted portion (Hagiwara and Bullock, 1957). However, in one type (“driver cells”) invasion of the soma is

90

DOMINICK P. PURPURA

possible (Bullock and Terzuolo, 1957). The supramedullary cells of blowfish are also capable of supporting all-or-none activity (Bennett et aZ., 1959). In the case of spinal motoneurons, intracellularly recorded action potentials generated by antidromic and orthodromic stimulation show discontinuities preceding the overshoot of the spike (Araki and Otani, 1955; Fuortes et al., 1957; Coombs et al., 1957a, b). It is generally accepted that the discontinuities signal stepwise activation of different components of the motoneuron, related, in part, to differences in threshold depolarization (cf. also Fatt, 1957a, b). The lowest threshold for impulse initiation obtains in the vicinity of the initial axonal segment. (Coombs et al., 1957b, claim that impulses are initiated in the latter a t 5-18 mv depolariaation, whereas the soma-dendritic membrane is said to have a threshold at 20-37 mv.) Fatt (1957a, b) has analyzed the extracellular and intracellular potentials generated by antidromic and, orthodromic stimulation of motoneurons and concludes that following both modes of activation, the soma is discharged first and then after a short delay, the dendrites. According to Fatt (1957a) dendrites of motoneurons conduct action potentials of 1 msec duration at a velocity of 0.7-1 m per second. According to Freygang and Frank (1959) intracellular and extracellular records of spinal motoneurons may now be viewed as indicating that orthodromically and antidromically evoked all-or-none discharges do not actively propagate in soma-dendritic membrane. The latter authors also consider the possibility that pressure effects on the membrane may play a significant role in determining their results. Freygang (1958) concludes “at present there is no evidence that the major portion of the soma-dendritic membrane is used for any purpose other than to receive synaptic excitation.” Extracellularly recorded action potentials from orthodromically activated cells in the lateral geniculate (Tasaki et al., 1954b), cerebellum (Granit and Phillips, 1956), and cerebral cortex (Martin and Branch, 1958; Phillips, 1959; Rayport, 1957) exhibit patterns similar to those recorded by Freygang (1958). Intracellularly recorded action potentials from Betz cells evoked by antidromic (Phillips, 1956; Martin and Branch, 1958) or orthodromic stimulation (Branch and Martin, 1958) exhibit characteristics identical to those recorded intracellularly from spinal motoneurons, suggesting that the process of impulse initiation in Betz cells is similar to that in spinal motoneurons (Phillips, 1959). Extracellularly located microelectrodes in the immediate vicinity of single antidromically activated Betz cells record large diphasic positive-negative spike-potentials with positive peaks of from 10-17 mv followed by much smaller, slower negative phases (Martin and Branch, 1968). It is not unreasonable to infer that if extracel-


91

lularly recorded action potentials of this nature indicate that all-or-none discharges do not actively propagate in the soma-dendritic membrane of certain neurons (Tomita, 1957; Freygang, 1958; Freygang and Frank, 1959) then the soma membrane of Betz cells may be electrically inexcitable. Thus, the possibility must be seriously entertained that Purpura and Grundfest (1956) have been too conservative in their conclusion that only the apical dendrites of pyramidal cells appear to be electrically inexcitable. (Since the dendrites comprise approximately 80-90% of the surface of the pyramidal cell (Sholl, 1956) the error may not be too disturbing !)

C. HYPERPOLARIZING POSTSYNAPTIC POTENTIALS OF APICALDENDRITES Following stimulation of cutaneous nerves, intracellular recordings from spinal motoneurons often reveal both depolarizing and hyperpolarizing p.s.p.’s (Eccles, 1957) which may alter membrane potential sufficiently to initiate or block impulse initiation somewhere in the vicinity of the initial axonal segment. Because of the heterogeneous origin of presynaptic fibers in the molecular layer of cerebral cortex, direct stimulation of these elements can be expected to generate hyperpolarizing, as well as depolarizing p.s.p.’s in apical dendrites. Under ordinary conditions surface stimulation results in a net depolarizing synaptic potential in apical dendrites, the hyperpolarizing p.s.p. being masked by a simultaneously developing depolarizing p.s.p. of greater magnitude (Purpura and Grundfest, 1956). Three methods are available to alter the normal pattern of algebraic summation: (1) the use of specifically acting pharmacological agents which block either depolarizing or hyperpolarizing synaptic electrogenesis; ( 2 ) the use of pharmacological agents in combination, one of which “selectively” protects one variety of synapse from the general blocking effects of a subsequently administered agent; and (3) physiological interactions which augment one or the other type of synaptic activity. (1) Selective blocking effects of w-amino acids o n apical dendritic synapses. The use of w-amino acids as powerfully acting synapse blocking agents was suggested, in part, by the observations of Hayashi and Nagai (1956). Purpura and co-workers (1957a, b) have shown that topical application of aliphatic w-amino acids containing five or less carbons are powerful blockaders of depolarizing synapses of apical dendrites, whereas those containing six or more carbon atoms selectively block hyperpolarizing dendritic synapses. The theoretical framework upon which the mode of testing the various amino acids is based has been described in detail (Purpura e t al., 1959b). The effects of some members of this series of aliphatic w-amino acids on surface dendritic p.s.p.’s of

92

DOMINICK P. PURPURA

WO.9. Effects of topical y-aminobutyric acid (GABA) on dendritic postsynaptic potentials evoked by surface stimulation of cerebral (upper channel) and cerebellar cortex (lower channel). Application of 2-3 drops of buffered 1% GABA to both recording sites results in immediate reversal of surface negative response in cerebral cortex (1) to surface positivity (21, but in cerebellar cortex abolition of surface negativity is not accompanied by “unmasking” of surface positivity (lower channel, 2) ; flushing both sites with warm Ringer solution results in immediate recovery of cerebral cortical response, and delayed recovery of cerebellar activity (3). Note “overshooting” of cerebral response during recovery. Cerebral surface negative response in (1) approximately 18 msec in duration. Explanations in text (from: Purpura et al., 1957a). cerebral and cerebellar cortex are shown in Figs. 9 and 10. The most powerfully acting o-amino acid which blocks depolarieing synapses is 7-aminobutyric acid (GABA) whereas the Cs member of the series, ( 0 aminocaprylic acid) exerts a maximal blocking action on hyperpolarieing synapses. In the cerebellar cortex GABA blockade of depolarieing p.s.p.’s of Purkinje cell dendrites evoked by surface stimulation does not result in subsequent unmasking of hyperpolarieing p.s.p.’s as in the case of the cerebral cortex (Fig. 9). Also, the longer chain w-amino acids (c-aminocaproic and o-aminocaprylic) augment cerebral cortical responses, but are without effect on the cerebellar responses (Fig. 10). Failure to reveal

FIG.10. Differential actions of e-aminocaproic (CB) and o-aminocaprylic acids on cerebral and cerebellar cortex. Records for the 2 compounds taken from 2 different experiments. Direct comparison of the augmentation produced by Coand CS cannot be made. 1,4: controls; 2,5: 1 min after application of agents to cerebral (upper trace) and cerebellar cortex; 3,6: 20 min after frequent rinsings of the cortical sites with warm Ringer solution. The striking augmentation of responsiveness observed in cerebral cortex (2,5) does not occur in the cerebellar cortex. Five superimposed records throughout. Time 20 msec (from: Purpura et al., 1957b).


93

hyperpolarizing p.s.p.’s in cerebellar cortex and absence of effects following application of the Ca and Cs 0,-amino acids is attributed to a relative paucity of hyperpolarizing synapses on superficially located Purkinje cell dendrites excited by surface stimulation (Purpura and Grundfest, 1956, 195713) (cf. below). The effects of all the w-amino acids are rapid, reversible, and reproducible. Those produced by GABA on cerebral cortex have also been described by Iwama and Jasper (1957) though given a different interpretation. Marrazzi et al. (1958) have also shown some transient effects of intracarotid GABA on transcallosally evoked responses (although these effects have been questioned by further studies of Purpura et al. (1958d,e) on the role of the blood-brain barrier in limiting penetration of systematically administration GABA and other w-amino acids into brain). Particular attention is called to the effects of GABA since this Oamino acid, normally present in brain in relatively high concentrations (Awapara et al., 1950; Roberts and Frankel, 1950; Udenfriend, 1950) has been shown by Baeemore et al. (1956) to be a constituent of Florey and McLennan’s (1955) Factor I, an “inhibitory” substance extractable from brain. McLennan (1958), unable to extract GABA from Factor I, suggests that the “drastic extraction procedures” used by Bazemore et a2. (1956) may have split GABA from a larger, parent molecule. It is beyond the scope of the review to detail the evidences relating to the general mode of action, structure-activity relations and effects of systemic as opposed to topical administration of the various amino acids. These data are presented in the following: Purpura (1958) ; Purpura and Grundfest (1959) ; Purpura et al. (1957a, b; 1958a, b, d, e; 1959a, b, c). The discovery that some amino acids are potently acting synapse blocking agents calls attention to the important role which these agents may play in the over-all functional activity of the brain. Some of these amino acids or their guanidino derivatives are normally present in brain in relatively high concentration. The metabolic pathways involved in their production and transformation are today being intensively pursued. Neuropharmacological techniques may provide suitable methods for facilitating the accumulation of biochemical data. One example of this is illustrated by the observation that y-guanidinobutyric acid, which is normally present in brain (Irreverra et al., 1957), when topically applied to the cerebral cortex produces marked augmentation of evoked dendritic p.s.p.’s. In animals with localized destruction of the blood-brain barrier (Purpura et al., 1958e) systemic administration of y-guanidinobutyric acid results in the development of paroxysmal activity confined to the region of blood-brain barrier loss. These observations suggest that in contrast to GABA which inactivates depolarizing dendritic synapses, the

94

DOMINICK P. PURFWRA

guanidino derivative may inactivate hyperpolarizing electrogenesis. In view of these observations it was considered possible that a conversion of GABA to guanidinobutyric acid might occur in the brain for some specific physiological function (Pisano and Udenfriend, 1958) and such was found to be the case. In addition to providing important data on the possible synaptic composition of different organizations involved in the production of evoked cortical potentials (cf. below) the effects of some amino acids on electrocortical activity require further elaboration. Of particular interest are the dramatic effects of some dicarboxylic amino acids (Bureg, 1956; Purpura et al., 1959b). Topical application of these amino acids rapidly abolishes evoked cortical activity, induces pial vascular dilatation and waves of spreading cortical depression, similar, if not identical to the “spreading depression” of Leio (1944). These results suggest that amino acids, like glutamic acid play a major role in the development of spreading depression induced by a variety of means. Should spreading depression be related to alterations in intra- or extracellular concentrations of some dicarboxylic amino acids, other factors such as have been reported, viz. swelling of apical dendrites (van Harreveld, 1957, 1958) and increase in cortical impedance during spreading depression (van Harreveld and Ochs, 1956) may be of secondary significance. Surface positivity unmasked in the “direct” cortical response is locally generated in apical dendrites (viz., a hyperpolarizing p.s.p.) (Fig. 11) but it is not always possible to determine from analysis of surface records alone whether a recorded potential which is positive to an “indifferent” reference point is a “source” for remotely located “sinks” or represents a local hyperpolarizing p.s.p. Analysis becomes more difficult when afferent pathways to cortex are activated. Under these conditions, p.s.p.’s of opposing varieties may be generated a t different loci in cortex. Thus, when large subsurface “sinks” appear in GABAtreated cortex during responses evoked by afferent stimulation, the problem of whether large surface positivities are hyperpolarizing p.s.p.’s or “sources” for subsurface “sinks” is not always readily resolved (Purpura et al., 195%). The externally recorded hyperpolarizing p.8.p. of the “direct” cortical response revealed by topical application of GABA has characteristics similar t o intracellularly hyperpolarizing p.s.p.’s of spinal motoneurons (Brock et al., 1952) and cortical neurons (Martin and Branch, 1958) (Fig. 5 F, G). Like the surface negative response of cortex, the surface positive response may be detected at sites remote from the stimulating electrodes, with latencies similar to those observed for the former (Jasper et al., 1958). Surface positive p.s.p.’s unmasked in GABA-treated


95

cortex recorded a t a distance from the site of weak surface stimulation cannot be interpreted as the spreading positive wave of Adrian (1936), which presumably signals progressive synaptic transmission through the

H

POMSEC

F I ~11. . Locus of synaptic blocking action of r-aminobutyric acid. Upper channel: dendritic depolarizing p.s.p. evoked by moderately strong cortical stimulus recorded with surface electrode ; lower channel : responses recorded with 10 p microelectrode adjacent to surface electrode and introduced stepwise into cortex (depths indicated on left). I n control records negativity attenuates rapidly between 200400 p, revealing a small positivity below 500 p. Following control recordings microelectrode is returned to surface and buffered 1% GABA is applied. Surface positivity “unmasked” by GABA also attenuates between 200-400 p and below this depth only a brief low voltage negativity is detected, presumably signaling some subsurface activity generated by the moderately strong surface stimulus. The hyperpolariring p.s.p., like the depolarizing p.s.p. (before GABA) is confined to the upper 0.5 mm of cortex. Calibrations: 0.2 mv (from: Purpura et al., 19590).

cortical depths (Burns, 1951), for the following reasons: (1) the surface positive p.s.p. is smoothly graded, lacks the “all-or-none” characteristics of the spreading deep response; and (2) is not abolished by very high doses of pentobarbital (Purpura et al., 1958a; 1959~).

96

DOMINICK P. PURPURA

(2) Use of pharmacological agents in combination. The existence of countervailing hyperpolarizing electrogenesis in the surface negative response of cortex was originally inferred from analyses of the effects of systemic d-tubocurarine in heavib heparinized preparations (Purpura and Grundfest, 1956, 1957b). Under these conditions, protection of depolarizing synapses by heparin (Cheymol et aZ., 1955) results in augmen-

IOOcps FIQ. 12. Interaction of surface-evoked dendritic p5.p.‘s and reticulocortical responses evoked .in snterior supraaylvian gyrus. Upper channel: surface response ; lower channel: subsurface responses recorded with a 100 p wire electrode located a t a depth of 1.5-1.8 mm (unanesthetired-paralyzed cat). 1 and 2: conditioning dendritic and testing reticulocortical responses, respectively. 3-6 : interaction at different intervals. In 3, a t approximately 12 msec, the reticulocortical negativity is completely absent, whereas the subsurface positivity is relatively unaffected. Reticulocortical response is of longer duration but reduced in amplitude somewhat later (4) and from 40-80 msec after the conditioning surface stimulus reticulocortical surface responses are reversed in polarity (5,6) (from : Purpura, 1958).

tation of surface evoked activity as a consequence of d-tubocurarine blockade of “unprotected” hyperpolarizing synapses. ( 3 ) Physiological interactions augmenting synaptic activity. Interaction of various evoked potentials often results in the “unmasking” of surface positive components in responses which are overtly surface negative in configuration (Purpura, 1958) (Fig. 12). Subsurface recordings may not reveal significant “sinks” for these surface positivities, indicat-

97 ing that, like the surface negativity of the unconditioned response, the “unmasked” positivity is close to the cortical surface and presumably represents a hyperpolarizing p.s.p. of apical dendrites. These data, supported by the direct demonstration of hyperpolarizing p.s.p.’s in cortical neurons (Branch and Martin, 1958; Phillips, 1956, 1959) indicate the necessity for rejecting the notion that “no clear synaptic inhibition has been observed in cortex” (Clare and Bishop, 1957). The dramatic occurrence of high amplitude positivities in the immediate vicinity of cortical elements activated by unspecific thalamocorticaI pathways has been observed by Li et al. (1956b). Such high amplitude positive responses are of much greater magnitude than the surface negative responses. The locally recorded responses may be accounted for if it is assumed that unspecific afferent5 activate inhibitory elements in cortex which generate hyperpolarizing p.s.p.’s. That inhibitory elements in cortex are activated by thalamic volleys has been demonstrated both indirectly (Li, 195610) and directly by means of intracellular recordings from cortical neurons (Fig. 5 F) (Martin and Branch, 1958). Thus, the extraordinarily localized nature of slow waves recorded with extracellularly located microelectrodes (cf. Renshaw et al., 1940, Fig. 5 ; Mountcastle et al., 1957) may be viewed as summated hyperpolarizing and depolarizing p.s.p.’s (Fig. 2). NATURE OF ELECTROCORTICAL POTENTIALS

IV. Functional Differences in Cortical Axosomatic and Axodendritic Synapses

The soma-dendritic surface of pyramidal cells in the cerebral cortex is now thought to be densely covered with synaptic end-feet (Fig. 6) (Armstrong and Young, 1957). Electron microscopy reveals no fundamental differences in morphology between axosomatic and axodendritic synapses in the mammalian central nervous system (Palay, 1956). But this is not to say that marked functional differences are not to be expected between axodendritic and axosomatic synapses. Caj a1 (1934) had long been aware of the possibility that synapses on different parts of the neuron might indicate important functional differences, but no formal characterization of the physiological properties of axosomatic and axodendritic synapses was attempted until Chang (1952) studied in detail the properties of apical dendrites of cortical pyramidal cells. As noted above Chang’s “paradendritic” (axodendritic) synapse was described as an articulation between axon terminah and gemmules or spines on dendrites. It should be recalled that Chang suggested that gemmules of high ohmic resistance separated the synapse from the dendritic surface, thus providing for weak and ineffectual stimulation of

98

DOMINICK P. PURPUBA

the latter, Because of this it was inferred that synaptic activation of dendrites could never discharge pyramidal cells. Pyramidal cell discharge was believed to be effected only through “pen-corpuscular” (axosomatic) synapses ; axodendritic activation serving to facilitate discharge. Other hypotheses developed along similar lines, but differing as to the functional consequences of synaptic excitation of dendrites have also been noted above (Beritov, 1956; Roitbak, 1955; Sarkissov, 1956). Two factors, reasonably supported by experimental data, can account for the difference in effectiveness of axodendritic and axosomatic synaptic activation to discharge pyramidal neurons without invoking special morphological factors: ( a ) the geometrical relation of activated synapses to the locus of impulse initiation, and ( b ) possible differences in the “functional” characteristics of the two varieties of synapses.

A. GEOMETRICAL CONSIDERATIONS Data are now available concerning the mode of impulse initiation in cortical pyramidal neurons (Phillips, 1959) which indicate that as in the case of spinal motoneurons (Araki and Otani, 1955; Fuortes et al. 1957; Coombs et al. 1957a, b) and lateral geniculate neurons (Freygang, 1958) the all-or-none spike is initiated somewhere in the vicinity of the initial axonal segment. In view of this it is likely that the soma-dendritic membrane of pyramidal neurons may not be capable of supporting an all-ornone response. Thus, the problem of defining the relative effectiveness of differently located synapses becomes a problem in “simple” geometry. Since they are generated in electrically inexcitable membrane via transmitter action, depolariaing and hyperpolarizing p.s.p.’s are “standing potentials.” It had generally been inferred that the conductance change endures only for the rising (or falling) phase of the synaptic potential, the decay of the potential representing the discharge of charged membrane capacity through the resting membrane resistance (Fatt and Kata, 1951; Coombs et a2. 1955; Fatt, 1957b). This concept now requires some modification (cf. Grundfest, 1957a). It has been shown, at least for invertebrate synapses, that conductance increases at the synaptic membrane do not appear to be transient, but rather to be enduring over most of the period of the synaptic potential (Hagiwara, 1958; Hagiwara and Tasaki, 1958). To what extent these observations are applicable to vertebrate synapses cannot be evaluated, but it is probable that the time course of the p.s.p. is not governed solely by the membrane time constant, which appears to be too short to account for the decay of the p.s.p. (Frank and Fuortes, 1956). P.s.p.’s, through electrotonic extension, affect the spike generating


99

membrane a t the initial segment. P.s.p.’s developing in the cell body presumably suffer less electrotonic losses in relation to their effect on electrically excitable membrane than those generated out on the tips of dendrites (Purpura and Grundfest, 1956). Between these two extremes, a wide variety of interactions between axosomatic and axodendritic p.s.p.’s are conceivable. Furthermore, since inhibitory as well as excitatory synapses are distributed over the entire soma-dendritic surface, the possible interactions of the two varieties of synaptic activity generated a t different geometrical sites permit of localized or generalized fluctuations in membrane potential which, if of sufficient magnitude, may initiate or block all-or-none discharge a t the initial segment. (Considering the relative distance of a large percentage of dendritic synapses from the cell body, it is doubtful whether the complex synaptic interactions occurring in apical dendrites can be adequately evaluated by a microelectrode in the soma!) Cortical pyramidal cells are apparently not discharged by depolarizing p.s.p.’s developing in their apical dendrites. Dendritic responses evoked by surface stimulation are not associated with discharge of cells in deeper layers of cortex, although facilitation of neuronal discharge may be detected (Clare and Bishop, 1955). To effect the discharge of pyramidal elements relatively strong stimuli must be employed which initiate “deep” activity (“Surface positive wave” of Adrian, 1936; Burns, 1951). Hyperpolarizing p.s.p.’s “unmasked” in apical dendrites by GABA do not affect the direct or synaptic excitability of corticospinal neurons (Purpura et al., 1957b). Pyramidal neurons in archicortex activated via their apical dendrites by fibers from the subiculum may be discharged by depolarizing p.s.p.’s developing in these elements but in order to do this it is likely that summation of p.s.p.’s generated at different dendritic loci is required as shown by the effects of sustained repetitive stimulation of the afferent pathway (Fig. 8).

B. CHEMICAL SPECIFICITIES Geometry alone may not account for physiological differences between axodendritic and axosomatic synapses. Apical dendritic synapses of pyramidal cells show marked sensitivity to w-amino acids, as noted above, whereas axosomatic synapses are not affected even by relatively prolonged application of high concentrations of the amino acids (Purpura et al., 1957b). These differences cannot be ascribed to failure of the amino acids to effectively penetrate to the regions where pyramidal cell bodies are located since it has been shown that systemic administration of w-amino acids in cortical regions of blood-brain barrier de-

100

DOMINICIC P. PURPURA

struction do not alter axosomatic synaptic activity at a time when axodendritic depolarizing synapses are completely inactivated (Purpura et al., 1958d, e) , Selective depression of evoked dendritic activity was also noted by Iwama and Jasper (1957). It has been suggested that GABA may function in part as an anti-excitatory transmitter in cortex a t dendritic synapses (Purpura et al., 1959b) by competing at the receptor sites with a transmitter activating depolarizing synapses. It is beyond the scope of this review to detail the comparative pharmacological actions of GABA and other w-amino acids on different varieties of synapses in different species. Still, it should be noted that in contrast to the predominantly blocking effect of GABA on depolarizing axodendritic synapses in cerebral and cerebellar cortex, the amino acid appears to duplicate the action of the inhibitory transmitter on the terminals of crayfish sensory neurons with respect to potential and conductance changes (Edwards and K d e r , 1957; Kuffler and Edwards, 1958). Similar conductance changes have been noted at crustacean neuromuscular synapses (Boistel and Fatt, 1959; Grundfest et al., 1959). Of considerable significance, however, is the fact that longer chain w-amino acids which block hyperpolarising p.s.p.’s in apical dendrites of cortex are without effect on invertebrate synapses (Grundfest et al., 1959). Whatever future role may be assigned to GABA and other naturally occurring synaptically active amino or guanidino acids, the demonstration that these “metabolites” exert selective actions on axodendritic synapses indicates that the latter possess biochemical characteristics distinctly different from axosomatic synapses (cf. Purpura et al., 1957a, b). It is unlikely that the differential effects of topically applied GABA (or systemically administered GABA following destruction of the bloodbrain barrier) can be attributed to the presence of a second “barrier” peculiar to axosomatic synapses which limits penetration of the amino acid to receptor sites. Such a “synaptic barrier” has been postulated to account for the relative ineffectiveness of the action of some drugs upon cholinergic synapses on collaterally activated “Renshaw cells” in the ventral horn (cf. Eccles et al., 1954; Curtis and Eccles, 1958a,b). The amino acids thus provide new and powerful tools for analyzing the composition of synaptic pathways involved in the production of various evoked potentials and for investigating the molecular configuration of different electrogenic membranes in different species (Purpura et al., 1959b). C. SUMMARY Cortical synapses do not appear to be fundamentally different, morphologically, from synapses found elsewhere in the central nervous sys-


101

tem. Evidence is reviewed which supports the hypothesis that the surface negative dendritic potential of cerebral cortex is composed of depolarizing and hyperpolarizing postsynaptic potentials generated in structures which are electrically inexcitable. In the cerebellar cortex, hyperpolarizing synaptic electrogenesis is poorly developed in the terminals of Purkinje cell dendrites. The functional characteristics of axodendritic and axosomatic synapses on cortical pyramidal neurons are defined in terms of their geometrical distribution and specific pharmacological properties.

Part 111. Transmissional and Conductile Activity in Different Cortical Neuronal Organizations

Two aspects of cortical electrophysiology become amenable to analysis as a consequence of identifying spontaneous and evoked slow potentials as algebraically summated postsynaptic potentials, generated in varying proportions in the soma-dendritic membrane of cortical neurons by complexly organized “systems” of excitatory and inhibitory elements: (1) the relation of slow potentials to unit discharges; and (2) the synaptic organizations in cerebral and cerebellar cortex involved in the production of potentials evoked in response to different activating pathways.

1. Relationship of Spontaneous and Evoked Potentials to Single Neuronal Discharges

Jung’s diagrammatic representation ( 1953) of the relations between unit discharges and slow waves (Fig. 13) summarizes much of the experimental data obtained on this subject since the initial investigations of Renshaw et al. (1940). Ordinarily no correlation between single neuron discharges and spontaneous slow wave activity is observed (cf. Renshaw et al., 1940; Li and Jasper, 1953; Amassian, 1953; Mountcastle et al., 1957) but occasionally neurons are encountered which discharge repetitively during specific phases of the surface potentials, especially when the latter have been modified by barbiturate anesthesia (Verzeano and Calma, 1954). Conditions are rendered more favorable for detecting specific correlations when experiments have been designed to: (1) specify the identity of the units under examination; and (2) examine the relationship of evoked potentials to discharges of identifiable and nonidentifiable cortical elements.

102

DOMINICK P. PURPTJRA

I.

Ilb

A

I

0

100

200

300

400

I

500msec 0.5 sec.

FIa. 13. Scheme of three different groups of single neuronal discharges classified after their relation to slower brain waves. I. Neuronal spike discharges independent of slow rhythms. 11. Spike discharges associated with certain phases of slow rhythms (potential relative to cortical surface): (a) on surface negative peak of the waves, (b) during the negative rising phase. 111. Initial high-frequency neuronal spikes corresponding with the surface positive primary response in cortical receiving areas after sensory stimuli (from : Jung, 1953).

A. SURFACE POTENTIALS AND DISCHARGES OF CORTICOSPINAL NEURONS Of all the neurons in the cerebral cortex, those most readily identifiable are the elements giving rise to the corticospinal tract. The accidental discovery that relatively high amplitude spontaneous waves in motor cortex were associated with pyramidal tract discharges (Adrian and Moruzzi, 1939) was, indeed, a significant event in neurophysiology. Tract discharges were not correlated with surface activity recorded from other regions of cortex, hence it was concluded that the slow waves recorded from the motor cortex were intimately related to processes responsible for generating the grouped discharges. The latter tended to correspond largely with the surface positive components of high amplitude slow waves. In cats prepared by electrocoagulation of the midbrain (“pyramidal cat”), further analysis indicated that unit tract discharges occurred only during cortical spindle bursts, but these might also be variable; discharges vanishing after the first few cortical waves of the


103

spindle train, or “waxing and waning” throughout the course of the spindle (Whitlock et al., 1953). Relayed pyramidal tract activity during spindle bursts was also observed by Brookhart and Zanchetti (1956). In unanesthetized preparations, no correlation between spontaneous low voltage fast waves recorded from the motor cortex and discharges of pyramidal units was detectable (Calma and Arduini, 1954). Under these conditions pyramidal units exhibited random discharges of 20-100 per second. During electrocortical arousal produced by electrical stimulation of midbrain reticular regions the discharges were reduced in frequency or abolished, although in some cases augmentation of pyramidal tract discharges was also observed.

B. EVOKED POTENTIALS AND CONDUCTILE ACTIVITY Considerably more information is obtained when the relations between single units and evoked potentials are analyzed. Peripheral stimulation is associated with reflex discharges of pyramidal units (Adrian and Moruzzi, 1939) which is related to the initial surface positive component of the primary evoked potential (Amassian et al., 1955). In unanesthetized, paralyzed preparations stimulation of the corpus callosum also evokes relayed pyramidal tract discharges (Purpura and Girado, 1959) during the early surface positivity of the cortical potential. The effects are presumably attributable to the manner in which callosal afferents engage cortical neurons (cf. below). Recruiting responses induced by repetitive medial or intralaminar thalamic stimulation evoke tract discharges in animals with midbrain lesions (Arduini and Whitlock, 1953) but not in unanesthetized, paralyzed preparations (Brookhart and Zanchetti, 1956). I n the latter, only “augmenting” thalamocortical responses are associated with relayed tract activity (Brookhart and Zanchetti, 1956; Purpura, 1958; Purpura et al., 1958c) suggesting a close correlation between the spontaneous spindle bursts and augmenting thalamocortical responses (Fig. 14). Tract discharges, however, are complexly related to augmenting potentials and will be discussed below. What is revealed by these studies is the important fact that slow waves in motor cortex may or may not be associated with the discharge of definitely identifiable elements, i.e., corticospinal neurons, but when such discharges occur, they are generally associated with the surface positive components of spontaneous or evoked cortical potentials. C. FACTORS CONTRIBUTING TO NONRELATEDNESS Examination of the relationship between evoked potentials and other elements in cortex reveals complexities which tend to dwarf those already mentioned for corticospinal neurons. Most of the units activated in

104

DOMINICK P. PURPURA

FIG.14. Relationship of spontaneous and evoked responses to relayed pyramidal tract activity. Simultaneous recordings from the anterior sigmoid gyrus (upper trace) and medullary pyramid (lower trace). Top : unanesthetised cat. Augmenting responses initiated by stimulation of nucleus ventralis lateralis. Note the relayed pyramidal volleys. Middle : similar recordings from another preparation. Recruiting response initiated by stimulation of nucleus reuniens. Note stability of pyramidal recording. Bottom: recordings taken during spontaneous spindle burst in an unanesthetized cat with mesencephalic thermocoagulation. Note the relayed spindle waves in the pyramidal recording. (Frequency response of the preamplifier altered to eliminate higher frequency components.) (From : Brookhart and Zanchetti, 1956.)

somatic sensory cortex are associated with the initial surface positive (deep-negative) component of the primary evoked response (Amassian, 1953; Per1 and Whitlock, 1955; Li et al., 1956a; Mountcastle et al., 1957) Fig. 15), but the precise location of the intracortical “sinks” responsible for the surface postivity has been disputed. Li et al. (1956a) recorded “sinks” 0.8-1.2 mm below the cortical surface, whereas Amassian et al. (1955) and Mountcastle et al. (1957) found them a t higher levels corresponding to where units driven by afferent stimulation were first encountered. Latency and frequency of unit discharges are dependent on the source and strength of the stimulus and “excitability” of cortex (via. quality and quantity of anesthesia). Variations in stimulus strength a t the periphery (Mountcastle et al., 1957) may cause marked


105

FIG.15. Responses of neurons of first somatic area of cat’s cerebral cortex recorded extracellularly with microelectrodes ( 2 4 p tip diameter), Records 1 and 2 illustrate slight variation in temporal relation of initially negative spike discharge to slow wave response, whereas 3 illustrates independence of slow wave of spike discharge when latter fails. (Unit recorded at depth 1050p; modality skin hair; electrical stimulation of skin, long timeconstant recording.) Record 5 shows response of another unit (depth 520 p ) to stimulation of deep fascia and independence of slow wave and spike when latter fails (4). In 6,a response pattern is shown in which early repetitive response occurs nearly a t end of evoked slow wave. (Depth 920p; modality skin hair, short time-constant recording.) Time lines, 1000 cps. (From: Mountcastle et al., 1957.)

alterations in response latency, number of impulses per response, and discharge frequencies of single elements. Similar events have been observed by Towe and Amassian (1958) and Li et al. (1956a). Unit discharges have been observed to extend in time throughout the duration of the subsequent surface negativity of the primary evoked responses (Li et al., 1956a) although this has rarely been observed by Mountcastle et al. (1957). In auditory cortex unitary discharges often appear after the summit or even after completion of the surface positive potential (Erulkar et al., 1956). Units discharging during transcallosally evoked responses have mean latencies of from 1-100 msec (Latimer and Kennedy, 1958). Fifty per cent of the units encountered by the latter showed increases in mean response latency with decreases in stimulus intensity ; most of these units fired after the surface positive peak of the evoked response.

LO6

DOMINICK P. PURPURA

The relation between slow waves and unit discharges becomes inore complex when corticipetal pathways are activated which apparently have a different functional distribution than those derived from primary projection systems. Units discharged by stimulation of a primary projection nucleus (ventralis postero-lateralis ; VPL) may be facilitated but not discharged by afferents from nucleus centromedianus (CM) (Li, 1956a). Low frequency (l/sec) stimulation of CM evokes slow surface negative potentials of 18 msec duration which are facilitated by VPL stimulation and accompanied by repetitive discharges of units previously unresponsive to both stimuli alone. The facilitatory action of CM on VPL units may be maximal a t 30-40 and 110-120 msec after the VPL stimulus, indicating a profound dispersion of activities through different cortical organizations of excitatory and inhibitory elements. Cyclic facilitation and inhibition of spontaneously discharging units by repetitive stimulation of unspecific thalamic nuclei (Li et aZ., 1956b) supports this assumption. The inhibitory effects of stimulating nucleus ventralis lateralis on spontaneously discharging units, or those evoked by cortical surface stimulation, persist from 80-400 msec (Li, 195613). These inhibitory interactions are associated with diphasic surface positive-negative potentials with characteristics identical to those evoked following stimulation of primary somatic sensory projection pathways. Thus, surface potentials do not betray the internal complexities of the synaptic organizations which are, in part, responsible for them. The inhibitory effects exerted by aff erents from nucleus ventralis lateralis on cortical pyramidal units are presumably mediated via interneurons and not directly on Bets cells (Li, 1958; Branch and Martin, 1958). In visual cortex the complex relations of unit discharges to primary evoked surface potentials may be attributed to the interaction of a variety of afferent pathways activated by retinal stimulation. At least five types of neurons have been described by Jung and his collaborators (cf. Jung, 1953). A-types show no response to light, but show spontaneous discharges. B-types, resembling retinal on-elements, are activated by light and inhibited by darkness (discharge during evoked surface positive wave of cortex). C-types are inhibited by strong flashes of light. D-types resemble retinal off -elements and are activated by darkness. E-types resemble retinal on-off -elements. Still another group of elements in striate cortex appears to discharge only in response to objects moving across the visual field (Hubel, 1957). The reactions of the various types of neurons do not occur independently of each other. The surface positive evoked potential associated with the discharge of B-neurons is also associated with inhibition of D-neurons. Thalamoreticular stimulation can facilitate or inhibit the discharge of units of different types (Baumgartner, 1958).

107 The interaction of specific and nonspecific excitations on single neurons in striate cortex has a counterpart in the interaction of evoked surface potentials (Jasper and Ajmone-Marsan, 1952). Additional details of the reciprocal interactions and relationships between unit discharges in visual cortex and surface responses have been described by Jung (1958). NATURE OF ELECTROCORTICAL POTENTIALS

D. INTERPRETATION

The correlation (or lack of correlation) between spontaneous or evoked slow potentials and unit discharges in cortex can be satisfactorily interpreted on the basis of the hypothesis that the slow potentials recorded from the surface or depths of the cortex are composed of depolarizing and hyperpolarizing postsynaptic potentials generated in complexly organized elements (Purpura and Grundfest, 1956). Li and Jasper (1953) were among the first to suggest, on the basis of their experimental data, that p.s.p.’s were most likely to account for spontaneous slow potentials recorded from the surface or depths of the cortex. Somewhat later, however, they attributed the deep negative (surface positive) component of the primary evoked potential in somatic sensory cortex to persisting activity at presynaptic terminals “amplified by Golgi type I1 cells,” (Li et al., 1956a). Their later conclusion was largely based on the observation that deep negativity was maximal between 0.8-1.2 mm in cortex and showed decremental electrotonic spread into the white matter along the radiation axons. Mountcastle et al. (1957) have rejected this hypothesis and have suggested that penetration damage may have resulted in inactivation of the upper cortical regions in the experiments of Li et al. (1956a). Mountcastle et al. (1957) suggest that the “deep negative wave is the integrated sign of the local post-synaptic responses of large numbers of cortical neurons” (p. 406). This supports the conclusion of most workers (cf. Bremer, 1958) concerning the surface positive component of the primary evoked response. The relation which obtains between synaptic electrogenesis and conductile activity in elements where this relation can be studied in detail (such as spinal motoneurons) indicates that generation of conductile activity in neurons is dependent on postsynaptic potentials producing a critical membrane alteration in the vicinity of the initial axonal segment (cf. above). The initiation of motoneuron discharge by graded monosynaptic afferent excitatory volleys is said to result from the development of transmitter potentiality among members of a pool (Lloyd and McIntyre, 1955a,b). According to Hunt (1955b) transmitter potentiality is a function of the number of excitatory synaptic knobs, the degree to which such knobs are aggregated on the motoneuron soma, and the intensity of action per knob. However, it should be noted that Hunt

108

DOMINICR P. PURPURA

(195513) considers the generalized postsynaptic potential as playing a supportive role in initiation of discharge. “It seems likely that a significant assemblage of active excitatory knobs can initiate an impulse without the necessary antecedent production of such a post-synaptic potential” (p. 849). In the cerebral cortex some elements are monosynaptically discharged by specific thalamocortical aff erents. Volley activation of these elements generates sustained depolarizing p.s.p.’s which may discharge the neuron once or repetitively, at some phase of the synaptically induced depolarization. With intracellular recordings these events are readily observed (Fig. 16) in cortical neurons (AlbeFessard and Buser, 1955). This constitutes the most consistent relation between slow waves and unit discharges that have been recorded in cortex (Amassian, 1953; Li et al., 1956a; Mountcastle et al., 1957; Towe

F I ~16. . Extracellular, A, and intracellular, B, potentials recorded from a pyramidal neuron in somatic sensory cortex following stimulation of homolateral nucleus ventralis posterior. 1, low gain and 2, high gain record. A, before and B, after impalement of neuron (signaled by 40 mv negative shift, trace 1). Note in B, 2 repetitive spikes ( 8 ) superimposed on slow depolarizing potential (p) and abrupt cessation of spikes associated with the development of hyperpolarizing potential ( 0 ) . Vertical bar (A), 20 mv for trace 1 ; 4 mv for trace 2. Time 100 msec. (From: AlbeFessard and Buser, 1966.)

and Amassian, 1958). But it must not be assumed that only depolarizing p.s.p.’s are generated in cortical neurons activated by thalamic aff erents. In some cortical elements short latency hyperpolarizing p.s.p.’s of considerable magnitude (Fig. 5F) are detectable with intracellularly located microelectrodes (Branch and Martin, 1958) and hyperpolarizing p.s.p.’s have been identified indirectly with pharmacological techniques in a variety of responses (Purpura, 1958; Purpura and Grundfest, 1957b). Thus, inhibitory p.s.p.’s may sufficiently “restore” membrane polarization to prevent all-or-none discharge of elements without necessarily affecting the magnitude of externally recorded p.s.p.’s [or for that matter intracellularly recorded excitatory p.s.p.’s (Terzuolo, 1958)I. As conductile activity is transynaptically induced in primary cortical elements the afferent volley is dispersed to second, third, eta. order


109

neurons in which p.s.p.’s and conductile activity must be initiated. With increasing dispersion through polysynaptic pathways in cortex, a variety of single unit interactions can be anticipated and indeed, have been reported. The relationship of evoked potentials to unit discharges becomes indistinct with increasing synaptic dispersion and, finally, only random correlations are detectable. In the case of the largest pyramidal cells of cortex with well-developed apical dendrites minor correlations between certain types of spontaneous cortical waves and unit discharges can be made (Adrian and Moruzzi, 1939; Whitlock et al., 1953). This may be attributed to the fact that the spontaneous waves recorded from the cortical surface presumably represent largely the integrated p.s.p.’s of pyramidal cell dendrites generated by conductile pathways from a variety of sources. Due to electrotonic losses p.s.p.’s must be amplified by reactivation at the same site (temporal summation) or other dendritic loci (spatial summation) to effect and affect neuronal discharge. If dendritic activation is associated with simultaneous axosomatic activation of the neuron the relationship between slow potentials and spikes may be more predictable. I n the absence of “amplification,” it is doubtful whether p.s.p.’s generated solely in apical dendrites may effectively initiate the discharge of pyramidal neurons (cf. Grundfest, 1958b; Purpura and Grundfest, 1956). I n addition to providing a theoretical perspective from which to view previous data on the relationship of slow waves to spikes, recognition of the existence of inhibitory and excitatory synaptic processes operating in cortex can account for other data on the discharge characteristics of single neurons. Intracellularly recorded discharges of pyramidal cells activated by thalamic stimulation are abruptly halted by increases in membrane polarization (Fig. 16) (Albe-Fessard and Buser, 1955). Beta cells (Phillips, 1956) and Purkinje cells develop similar increases in membrane polarization during repetitive activity, although Purkinje cell discharge may be halted by another process similar to “cathodal depression” (Granit and Phillips, 1956) or “Wedensky inhibition” (Grundfest, 1 9 5 7 ~ )I.n striate cortex the sudden increase or restoration of membrane potential which abruptly terminates unit discharges evoked by radiation stimulation have been interpreted as inhibitory phenomena by Tasaki and associates (1954b). The identification of hyperpolarizing p.s.p.’s in a variety of evoked potentials and in responses of single cortical neurons strongly argues against the possibility that sudden changes in discharge frequency, or cessation of unit discharges can be due to refractoriness or fatigue of cortical neurons which are capable, under appropriate conditions, of sustained discharges a t 1000/sec (Adrian and Moruzzi, 1939).

110

DOMINICK P. PURPURA

Apart from the observations noted above other non-correlations between unit spikes and slow waves have been reported which permit interpretations similar to those described. It is generally agreed that cortical depression produced by asphyxia or deep anesthesia (Adrian and Moruzzi, 1939; Brookhart et al., 1951; Li and Jasper, 1953; Schlag and Brand, 1958) rapidly abolishes unit activity in cortex, but does not interfere with the production of slow waves (although the latter may be markedly modified). In the spinal cord, asphyxia and deep anesthesia have been employed as pharmacological tools (Brooks and Eccles, 1947b) to separate transmissional activity from conductile activity as well as monosynaptic from polysynaptic reflexes (Brooks and Koizumi, 1953). The sequence of events observed during progressive asphyxiation consists of an early augmentation of monosynaptic reflexes followed by depression. Polysynaptic reflexes are blocked at a time when the monosynaptic reflexes remain. After 3-4 min., orthodromic propagation of a volley through the cord begins to fail, but focal synaptic potentials are still evoked (Brooks and Eccles, 1947b). Administration of high doses of pentobarbital sodium blocks monosynaptic reflexes but postsynaptic potentials are still recorded electrotonically from the ventral roots (Eccles, 1946). (Note: the p a p . shown in Fig. 5, C was obtained after 80 mg per kilogram pentobarbital; no reflex discharges are detectable.) Concentrations of pentobarbital up to 120 mg per kilogram are required to markedly depress the focally recorded p.s.p. of cat motoneurons (Brooks and Eccles, 1947a). It should be noted that the surface negative response of cortex [p.s.p.’s of apical dendrites (Purpura and Grundfest, 1956)] evoked by cortical stimulation is remarkably resistant to asphyxia (Adrian, 1936) and is still obtainable after 200 mg per kilogram dose of pentobarbital (Ochs, 1956). Like the depolarizing p.s.p. of apical dendrites, the hyperpolarizing p.s.p. “unmasked” in GABA-treated cortex is also resistant to barbiturates (Purpura et al., 1958a). Recently, some effects of asphyxia on spinal motoneurons and interneurons have been studied with intracellularly located microelectrodes (Kolmodin and Skoglund, 1959). The data on the “pharmacological” actions of asphyxia or barbiturates on spinal motoneurons indicate that p.s.p.’s eventually become ineffective in initiating conductile activity in elements involved in reflex responses. Small reductions in synaptic drive or increase in “resting” membrane potential of the order of 2 mv without change in the amplitude of the p.8.p. (Terzuolo, 1958) may block generation of orthodromically elicited spikes. Spontaneous discharges of units in cortex presumably arise as a consequence of considerable synaptic bombardment of cortical elements by a variety of conductile pathways (cf.

NATURE OF ELIWTROCORTICAL POTENTIALS

111

above). When cortical units are spontaneously active, the delicate interplay of excitatory and inhibitory p.s.p.’s developing at different loci on the soma-dendritic surface may produce membrane potential fluctuations which periodically erupt in regenerative activity. Reduction in background excitatory “synaptic noise” or increase in inhibitory “synaptic noise” can abruptly halt unit discharges (cf. Hunt, 1955a), but these events may not be signaled by significant alterations in externally recorded potentials which reflect the integrated p.s.p.’s of many elements. It is reasonable to suggest that the blockade of unit discharges by asphyxia and barbiturates is attributable to a critical reduction in net depolarizing synaptic electrogenesis or expressed in more classic terms, a decrease in the “central excitatory state” of the discharging unit. Applied to the cerebral and cerebellar cortex, these events may result in reduction or absense of detectable unit activity in the presence of well preserved, but altered surface potentials.

E. SUMMARY Spontaneously discharging units encountered in cortex ordinarily show no relation to summated p.s.p.’s recorded from the surface. Elements contributing to the corticospinal tract appear to discharge repetitively at some phase of the surface potential oscillations, but predictable relations between the latter and tract discharges are dependent on the amplitude, frequency, and polarity of the surface potentials, and most of all, on the “past history” of the preparation. In somatic sensory cortex, primary afferent volleys ordinarily discharge cortical neurons at some phase of the extracellularly recorded focal postsynaptic potential; latency of discharge being primarily dependent on intensity of synaptic drive. As the interval between arrival of the presynaptic volley in cortex and the initial discharge of units increases, predictable relations between evoked slow potentials and unit activity rapidly vanish, but inhibitory and excitatory interactions among differently organized units are detectable. The integrated activity recorded with surface or depth electrodes is that of p.s.p.’s of opposing varieties generated in the soma-dendritic membrane of different neuronal aggregates. The discharge of any one unit in the population depends on the “central excitatory state” of the unit, via. its net excitatory and inhibitory synaptic bombardment. Studies involving sampling of units in a population of cortical neurons activated or inhibited by different synaptic pathways can be viewed as attempts to define the relative magnitude of inhibitory and excitatory effects which one pathway exerts on another at “loci” where the synaptic pathways engage common elements. P.s.p.’s recorded from the surface or depths of the cortex do not betray the internal complexity of the

112

DOMINICK P. PURPURA

synaptic pathways which are responsible for them. Inhibitory and excitatory interactions among units can proceed in the absence of detectable alterations in the integrated synaptic activity of the population and conversely, changes in recorded slow potentials may not reflect changes in the net synaptic electrogenesis of individual neurons.

II. Synaptic Organizations in Cerebral Cortex

Introduction. Unit analyses do not permit a complete evaluation of the organization of excitatory and inhibitory elements which participate in the production of an evoked cortical potential. This is not to say, however, that inhibition or discharge of cortical elements may not be related to specific components of the evoked potential. If it is difficult to predict the average behavior of a population of elements from the “private life” of a single member of that population (Burns, 1955), it is impossible to determine from inspection of the evoked surface potential alone the nature of the separate organizations which constitute the parent population. Evoked cortical potentials are compounded of depolarizing and hyperpolarizing p.s.p.’s of elements synaptically interconnected by different conductile pathways. Activation of the same population of cortical neurons in different ways can result in entirely different surface evoked potentials and, conversely, similar if not identically “appearing” evoked cortical responses can result from activation of different neuronal organizations. These factors indicate that methods of other than single unit analyses must be employed to define more precisely the composition of different synaptic organizations. Valuable information can be derived from examination of extant data on the mode of engagement of cortical neurons by various afferent pathways, and on interaction patterns produced by activation of a neuronal population via the same or different pathway. Differences in synaptic organization may be further revealed by the effects which different corticipetal volleys exert on steady-potential changes in cortex and, finally, by the differential effects of pharmacological agents on various evoked potentials. A.

THE DISTRIBUTION AND DISPERSION OF CORTICAL ACTIVITYEVOKED BY SURFACE STIMULATION

1. Response to “Direct” Cortical Stimulation a. Superficial Response. The “simplest” response recorded from the cerebral cortex following surface stimulation is compounded of depolarizing and hyperpolarizing p.s.p.’s generated in apical dendrites by conduc-

113 tile pathways (Purpura and Grundfest, 1956,1957b). The conductile component is ordinarily not recorded with large surface electrodes. It is likely that the relay pathways are not of homogeneous composition and it is equally probable that considerably more activity is generated in elements close to the cortical surface by a stimulus sufficient to evoke dendritic p.s.p.’s than is recorded with a large surface lead. When paired stimuli are employed of equal magnitude, the effects of a conditioning stimulus on a tested response are dependent on stimulus strength and the past history of the cortex (vie. anesthesia, type and quantity; trauma; blood loss; etc.) . In the absence of anesthesia, surface responses in suprasylvian gyrus of cat show early summation (2-10 msec) to weak paired stimuli followed by short periods of depressed responsiveness (20-30 msec) and . the conditioning stimulus rapid recovery (Purpura et al., 1 9 5 7 ~ )When is increased, tested responses are facilitated for 30-80 msec (Purpura et al., 1 9 5 7 ~ ) Similar . activity cycles are recorded from the exposed human cortex (Purpura e t al., 1957e). Under conditions of moderately deep barbiturate narcosis depression following a conditioning stimulus is profound and of a long duration. The second response of a pair evoked by strong cortical shocks may then reverse to surface positivity when the testing response is evoked as long as 80 msec after the conditioning stimulus (Chang, 1951b). Similar periods of depressed responsiveness in lightly anesthetized preparations have been reported by Clare and Bishop (1955) who related the prolonged period of depression to the development of positive after-potentials in dendrites. In the cerebellar cortex surface stimulation evokes a depolarizing p.s.p. of Purkinje cell dendrites. Strong repetitive surface stimulation a t frequencies which markedly attenuate the responses evoked in cerebral cortex does not depress cerebellar responses, but, on the contrary, augments them (Purpura and Grundfest, 1957b). The difference in responsiveness of cerebral and cerebellar cortex to surface stimulation reflects a fundamental difference in synaptic organization of the elements involved in the production of surface p.s.p.’s. I n cerebral cortical responses evoked by surface stimulation, hyperpolarizing synaptic electrogenesis is a prominent component of the overt response but, in cerebellar cortex, superficial dendritic synapses activated by surface stimuli are of homogeneous type, i.e., depolarizing (Purpura and Grundfest, 1957b; Purpura et al., 1957a, 1959a). Differences in activity cycles of dendrites revealed under various operational conditions may be attributed to alterations in the conductile (relay) pathways. Since the dendritic responses are evoked in membrane which does not show “refractoriness” (Purpura and Grundfest, 1956), it is inferred that amplitude variations in the evoked response result from NATURE OF ELECTROCORTICAL POTENTIALS

114

DOMINICK P. PURPURA

inhibitory blockade of some fraction of the interneuronal relay pathway to dendrites, or relative increase in the magnitude of countervailing hyperpolarizing electrogenesis. With strong conditioning shocks activity evoked in the cortical depths may also be a source of the augmented inhibitory effects on more superficially located elements, but it must also be recognized that the production of subsurface negativity by a strong cortical shock will be reflected as a surface positivity (“source”) and contribute to a cancellation of some fraction of the surface negativity. In view of these factors, as well as others noted throughout this review, the terms “unresponsive” or “relatively unresponsive” more appropriately describe the excitability of a population of elements rather than “refractoriness” as originally pointed out by Marshall (1941). b. The “Deep” Response. With sufficiently strong cortical stimuli the superficial negative response is followed by a surface positive potential of from 30-100 msec duration (Adrian, 1936). The latter response has been most intensely studied in the isolated cortical slab (Burns, 1950, 1951), but it is important to point out that data obtained utilizing neuronally isolated slabs are not strictly applicable to the intact cortex. In intact cortex repetitive stimuli are required to evoke propagating activity in the cortical depths (Adrian, 1936), whereas in isolated cortex, single stimuli 3 0 3 5 % maximal for production of surface negative responses generally evoke a surface positive response which spreads without attenuation at a velocity of 10-20 cm per second throughout the slab (Burns, 1950, 1951). In neuronally isolated cortex a very small dose of an anesthetic agent, insufficient for surgical anesthesia, reduces the magnitude of the surface positive response. At full surgical anesthesia the latter is completely abolished and cannot be obtained with any stimulus strength. There is general agreement among all workers that the surface-positive response represents successive depolarization of deep-lying cortical elements with initiation of conductile and synaptic responses in cortical networks. Burns (1957) has developed a general hypothesis to account for the spreading deep response based on an analysis of the histological nature of the network responsible for the repetitive discharges which accompany the surface positive response of isolated cortex. Neurons in layers IV and V, but not the largest pyramidal cells of layer V are believed to comprise most of the elements responsible for the surface positive (“burst”) response (Burns et al., 1957). Eccles (1951) suggested that the surface positive response was due to the discharge of pyramidal cells synaptically activated via their apical dendrites. According to the latter when p.s.p.’s in apical dendrites of pyramidal cells reach a critical level, an impulse is generated which propagates over the surface of the pyramidal cell, then down its axon

115 and along axon collaterals [vie., the discharges of impulses in the pyramidal tract which accompany the surface positive response of motor cortex (Adrian, 1936; Adrian and Moruezi, 1939) 1. Collateral excitation of other elements in the cortical depths results in continued dispersion of activity in progressively increasing areas of the network. Eccles’ suggestion (1951) concerning the origin of the surface-positive response appears improbable for two reasons: (1) dendritic responses evoked by single surface stimuli are not accompanied by cellular discharges in the cortical depths (Clare and Bishop, 1955) ; and (2) blockade of depolarieing dendritic synapses of motor cortex with y-aminobutyric acid does not affect the direct or synaptic excitability of corticospinal neurons (Purpura et al., 1957b). It appears more likely that single strong shocks delivered to an isolated cortex (or repetitive stimuli to intact cortex) directly activate elements in the cortical depths which initiate the deep response. This is strengthened by the observation that the latency of first detectable pyramidal tract discharges evoked by surface stimulation is consistent with direct and not synaptic excitation of corticospinal elements (Patton and Amassian, 1954; Purpura and Grundfest, 1956). The discharge of pyramidal neurons produced by surface volleys is compounded of directly and indirectly relayed activity. While the latter is attributable to collateral activation of interneurons, it is possible that conductile pathways near the cortical surface, which terminate axosomatically on pyramidal cells, may also be excited by strong surface stimulation and contribute to the development of the surface-positive response. A satisfactory explanation to account for the observation that spreading surface-positive responses are extraordinarily difficult to elicit in intact cortex with single shocks (as compared with neuronally isolated cortex) has not been forthcoming. [Burns (1957) has noted that the surface positive burst response is never encountered in intact unanesthesized preparations.] Two possibilities may be considered: ( a ) in intact cortex the network of elements presumably involved in transmission of the surface positive response may be occupied by activity generated via intracortical, intercortical, and corticipetal afferents which results in “occlusion” or “refractoriness” in different parts of the network; or ( b ) inhibitory mechanisms activated by these aff erents limit the development of sustained propagated activity through the cortex. The discharge characteristics of single elements in intact cortex have been reviewed above. These data indicate that cortical neurons can be driven, under appropriate conditions, to discharge at extremely high frequencies. Only when these frequencies are maintained for relatively long periods, does all-or-none conduction fail for a variety of reasons. To explain the failNATURE OF ELECTROCORTICAL POTENTIALS

116

DOMINICK P. PURPURA

ure of propagating surface positive responses in intact cortex on the basis of “occlusion or refractoriness” of elements in the network, neurons must be “spontaneously” active with extraordinarily high discharge frequencies and, furthermore, this activity must be relatively sustained so that any superimposed stimulus encounters many elements in the network in a state of absolute refractoriness. These conditions have never been fulfilled in unit recordings from cortex. On the other hand, the alternative suggestion, that inhibitory mechanisms are active in intact cortex, is supported by a host of observations. Cortical shocks can inhibit spontaneously discharging units for periods up to 400 msec (Creutzfeldt et al., 1956). Pyramidal cell discharges may be abruptly halted by cortical, as well as thalamic volleys (cf. Martin and Branch, 1958; Branch and Martin, 1958; Li, 1956b, 1958; Phillips, 1956) and brain stem reticular stimulation ordinarily reduces or abolishes spontaneous pyramidal tract discharges (Calma and Arduini, 1954) or those evoked by motor cortex stimulation (Purpura, 1957a). These inhibitory processes, which have been shown to be prominent features of electrocortical activity (Purpura and Grundfest, 1957b) may reasonably account for the differences observed in intact and isolated cortex in response to strong surface stimulation. B. ANTIDROMIC CORTICAL STIMULATION Were it not for the fact that neurons in the central nervous system are richly endowed with axon collaterals which synaptically activate other elements (which in turn may excite or inhibit the parent element) there would be little necessity to discuss this method of studying cortical Bynaptic organizations. Relatively little is known about the synaptic organizations activated by Betz cell collaterals in comparison to what is now known about the pharmacology and physiology of recurrent pathways in the ventral horn (Renshaw, 1946; Eccles, 1957; Brooks and Wilson, 1958). In the cerebral cortex two distinct varieties of neurons are accessible to antidromic stimulation: (a) corticospinal neurons; and (b) neurons contributing axons to the corpus callosum. 1. Corticospinal Neurons

Antidromic stimulation of the pyramidal tract was employed by Woolsey and Chang (1948) to study the cortical origin of elements contributing to this pathway in the cat. The results obtained by the latter may have been conditioned by inadvertent stimulation of adjacent p i mary afferent pathways (Landau, 1956). Recent anatomical studies (Chambers and Liu, 1957), however, indicate that the origin of cortico-

117 spinal neurons is more widespread than indicated by Landau and are in close agreement with the findings of Woolsey and Chang (1948). When pyramidal tract stimuli are carefully controlled, antidromically evoked cortical responses consist of two brief positive responses (1.5-2 msec) followed by a 10-15 msec surface negativity. Chang (1955a) and Landau (1956) have attributed the latter to antidromic invasion of apical dendrites. Purpura and Grundfest (1956), on the other hand, view the surface negativity as an excitatory postsynaptic potential evoked by collateral excitation of interneurons which, in turn, synaptically engage apical dendrites. Their data are derived from observations on the synaptic blocking action of d-tubocurarine on antidromically evoked surface negativity, as well as the blocking action of strychnine (Purpura and Grundfest, 1957a) and y-aminobutyric acid on this component (Purpura et al., 1958a). The depressant effect of strychnine on antidromically evoked surface negativity has been confirmed by Landau (1956). Topically or intravenously administered strychnine also exerts a marked depressant action on dendritic activity evoked by stimulation of other pathways (Purpura and Grundfest, 1957a). These effects have been confirmed by Clare and Bishop (1957), though given a different interpretation. The data of Purpura and Grundfest (1957a, b) are consistent with the hypothesis advanced by Eccles et al. (1954) that strychnine preferentially blocks inhibitory synapses. However, a t high concentrations (which induce spontaneous paroxysmal activity in the cerebral but not cerebellar cortex following topical application) strychnine also blocks excitatory synapses superficially located on apical dendrites. Collateral activation of cortical interneurons by antidromic stimulation of the pyramidal tract has been proposed by Chang (1955b) to account for delayed evoked potentials which are markedly affected by topical strychnine, but these results have been attributed by Landau (1956) to activation of primary afferent projections. Chang (195510) showed that stimulation of the pyramidal tract below the site of a recording microelectrode in the tract evoked long latency (5-15 msec) returning tract discharges and related these to similar phenomena observed in ventral roots following antidromic stimulation of motoneurons (Renshaw, 1941, 1946). Collateral re-excitation of corticospinal neurons via interneuronal relays was also suggested by Phillips (1956). Cortical interneurons activated by Beta cell collaterals are also activated by other corticipetal pathways, in particular, those from lateral thalamic projection nuclei (Li, 1958; Branch and Martin, 1958) and a wide variety of interactions between “antidromic” and dromic stimulation of cortical elements may be anticipated. NATURE OF ELECTROCORTICAL POTENTIALS

118

DOMINICK P. PUBPURA

Examination of the surface negative component of the antidromic cortical response suggested to Chang (1955a) that it was identical with the dendritic potential evoked by surface stimulation. While it is true that both responses have the same general characteristics they are not produced at similar loci in cortex as evidence by Chang’s own data (cf. 1955a, Fig. 2). After (antidromic) pyramidal tract stimulation microelectrode recordings in cortex showed persisting negativity at 1.2 mm depth which contrasts sharply with the fact that dendritic potentials evoked by surface stimulation are ordinarily not detectable below 0.5 mm (Fig. 11).Results similar to those of Chang have been obtained by Fan and Feng (1957) who conclude: “The direct response and the antidromic response even though both representing the activity of the apical dendrites must nevertheless occupy different locations on this structure. It appears that the direct response set up in the terminal portion of the apical dendrites does not conduct towards the basal portion and that likewise the antidromic response reflecting the invasion of the impulse to the basal portion, does not travel upwards to the terminal portion.” These results are consistent with the interpretation that both responses represent “standing” potentials (p.s.p.’s) generated at different loci on apical dendrites (Purpura and Grundfest, 1956), but it should also be noted that in the case of “antidromic” activation of apical dendrites hyperpolarizing p.s.p.’s are not “unmasked” following blockade of “antidromic” surface negativity with GABA (Purpura et al., 1958a) in contrast to the effects of the amino acid on surface response to direct cortical stimulation (cf. above). Thus different synaptic organizations are involved in the production of similarly appearing surface negative responses (cf. Smith and Purpura, 1959). 2, Antidromic Stimulation of Corpus Callosum

Surface activity evoked by contralateral cortical stimulation or direct stimulation of the corpus callosum is compounded of antidromic and orthodromic responses (cf. below). One to four weeks after ablation of one hemisphere, stimulation of the corpus callosum antidromically activates callosal elements in the intact hemisphere (Feng and Fan, 1957). Under these conditions callosal volleys evoke responses which strongly interact with antidromic cortical responses. This suggests engagement of similar cortical interneuronal organizations by callosal and corticospinal neuron axon collaterals. Of considerable significance is the finding that antidromic callosal stimulation can result in relayed pyramidal tract activity. Further work on preparations similar to those described by Feng and Fan (1957) is required to define more clearly the nature of the interposed relay pathways to pyramidal cells.

119


C. SYNAPTIC DISTRIBUTION OF SPECIFICAFFERENT ACTIVITY The immediate synaptic events in cortex following activation of primary projection pathways have been noted above (cf. also Per1 and Whitlock, 1955). At some phase of the surface positivity (or focally recorded depolarizing p.s.p.) axosomatic discharge of cortical elements is detectable (Fig. 15) in almost all cytoarchitectonic layers following discrete peripheral stimulation (Mountcastle, 1957; Mountcastle et al., 1957). Later events associated with the development of surface negativity are attributable to synaptic activation of various loci on apical dendrites (Amassian et al., 1955; Purpura and Grundfest, 1956; von Euler and Ricci, 1958) (cf. Bishop and Clare, 195213; Li et al., 1956a, for different views). During, and for some time after, dendritic activation, dispersion of activity in cortex may result in excitation of inhibitory neurons which tend to limit discharge frequencies in elements initially fired by the afferent volleys (Albe-Fessard and Buser, 1955; Tasaki et al., 1954). Depending on the general state of “excitability” of the preparation, a primary afferent response is succeeded by a variable series of surface potentials or “after-discharges” first described in detail by Bartley and Bishop (1933a). The spontaneous or evoked 5/sec rhythm in striate cortex is associated with periodic changes in excitability in the optic pathway (Bishop, 1933; Bishop and O’Leary, 1936). Changes in cortical excitability following an afferent volley usually consist of a series of facilitation-depression-recovery cycles, but the depression phases are not complete and are “capable of being counteracted by an adequate excitation” (Clare and Bishop, 1952). “After-discharges” are also observed in other primary projection cortical areas (Adrian, 1941; Bremer, 1943; Chang, 1950, 1951b). The repetitive waves evoked in somatic sensory cortex by afferent stimulation have characteristics similar to the repetitive “spindle bursts” observed in lightly barbiturized preparations. In the latter, interactions between primary afferent responses and “spindle bursts” exhibit similar cyclic variations (Moruzzi et al., 1950). Under certain conditions afferent stimulation is followed by a continuous high-frequency discharge resembling the previously existing spontaneous activity (Bishop and Clare, 1952a). During this fast after-discharge, fluctuations in responsiveness to superimposed test stimuli may also be observed. Although it was originally thought that different populations of elements were involved in the production of the primary evoked potentials, and spontaneous after-discharges (Bishop, 1949), it is now believed that both phenomena are attributable to the activity of the same neurons (Bishop and Clare, 1952a). The prolonged repetitive activity generated in cortex by primary afferent volleys pne; I

120

DOMINICK P. PURPURA

sumably represents successive synaptic activation of excitatory and inhibitory elements which can also be activated from other sources (Brerner, 1953; Chang, 1953b; Per1 and Whitlock, 1955; Latimer and Kennedy, 1958), despite the fact that those neurons initially discharged by the primary afferent volley may be responsive to afferent stimulation from restricted sources (Mountcastle et al., 1957). Bremer and Bonnet (1950) have classified the after-discharges recorded in auditory primary projection cortex into “fast” and “slow” types. The “fast” type is believed to represent a transitory intensification of the spontaneous cortical activity (cf. Bishop and Clare, 1952a) reinforced by thalamocortical impulses. It is claimed that the “~low” after-discharge is due to responses of cortical interneurons to volleys of impulses discharged repetitively by thalamic and mesencephalic relay nuclei, but is not dependent on corticothalamic reverberation as suggested by Chang (1950). The fast after-discharge is abolished by barbiturates but is much more resistant to ether anesthesia; the slow after-discharge is inferred to characteriee a condition of functional depression of the brain (of. Bremer, 1958). Iwama and Jasper (1957) have noted enhancement of the sensory after-discharge in the absence of significant effects upon spontaneous electrical activity following topical application of y-aminobutyric acid. This suggests that although the after-discharge and spontaneous electrical activity are attributable t o the same elements in cortex, the after-discharge may represent the activity of a particular complexly organized group of elements. Grundfest (1958b) has interpreted the effects observed by Iwama and Jasper (1957) as representing the “unmasking” of hyperpolari~ing p.s.p.’s ordinarily counterbalanced by depolarizing p.s.p.’s. In the experiments of Purpura et al. (1957b) topical application of GABA significantly enhanced the amplitude of the spontaneous cortical rhythms during inactivation of superficially located depolarizing axodendritic synapses. It is abundantly clear from these studies that the electrocorticogram represents much more than the activity contributed by depoIariaing p.s.p.’s of apical dendrites. Further clues as to the nature of the synaptic organizations activated by lateral thalamic volleys are provided by observing the effects of repetitive stimulation of lateral thalamic nuclear regions which evoke typical “augmenting” responses in sensorimotor cortex (Dempsey and Morison, 1942~).Advantage can be taken of the fact that such responses are accompanied by relayed pyramidal tract volleys (Fig. 14) (Brookhart and Zanchetti, 1956). Simultaneous registration of cortical surface activity and pyramidal tract discharges reveals (Fig. 17) that the first stimulus of an 8/sec train evokes a short-latency (2 msec), short-dura-


121

FIG.17. Sequence of fast sweep records to show the alterations in evoked cortical surface (upper channel), and relayed pyramidal tract responses (lower channel) induced by 8 per second stimulation of nucleus ventralis anterior. Cortical responses recorded from anterior sigmoid gyrus, pyramidal responses recorded with pair of 100 fi Teflon coated wires located in medullary portion of corticospinal tract. A-D: first four responses of 8 per second stimulus (indicated by arrows). First response of stimulus train (A) evokes short-latency (2 msec) short-duration tract discharge associated with early surface positive component of cortical potential. All subsequent stimuli a t 8/sec evoke only long-latency tract discharges which increase in intensity during the augmenting (surface negative) component of the cortical potential. Further explanation in text. (From : Purpura, 1958.)

tion tract discharge during the initial surface positivity. When other surface components have subsided and the cortex is “apparently quiescent,” a second volley coming 125 msec later evokes delayed tract activity of considerably longer duration which persists during the surface positivity and is augmented during the subsequent surface negativity (Purpura, 1958). That the longer-latency responses are not attributable to “refractoriness” in pathways generating short-latency discharges is demonstrated by the observation that short-latency tract responses re-

122

DOMINICK P . PURPURA

turn when stimulus frequencies are increased beyond 25/sec (Fig. 18) (Purpura et al., 1 9 5 8 ~ )The . discharge patterns of cortical elements involved in the relayed tract activity are similar to those produced by callosal volleys and will be treated more fully below. Suffice it to say that the complexity of the synaptic pathways involved in the production of short- and long-latency tract discharges is only hinted at by the

FIG. 18. Relationship of thalamocortical evoked responses to relayed corticospinal tract activity in unanesthetized-paralyzed cat. Evoked cortical potentials (upper channel) following lateral thalamic stimulation recorded transcortically from anterior sigmoid gyrus. Pyramidal tract responses recorded with pair of 100 p Teflon coated wires located above decussation. Stimulus frequencies noted above records. Numbers 2-24 indicate seconds after cessation 2/sec stimulation in third row. Short-latency tract discharges disappear at medium stimulus frequencies (10/sec) and reappear at higher frequencies (2&50/sec). Note absence of short-latency discharges in immediate posttetanic period and appearance of very long latency discharges which decrease in latency, progressively, during reappearance of short latency activity. The transition from long- to short-latency relayed activity is not accompanied by significant alterations in surface evoked potentials (4-12). (From : Purpura et al., 1958c.)

overt cortical potentials. Long-latency inhibitory pathways as well as excitatory pathways must be activated to account for these interactions. During the surface negativities of the augmenting response, hyperpolarising p.s.p.’s are inferred to be present as indicated by the effects of topical GABA (Purpura, 1958). Stimulation of lateral thalamic regions which generate augmenting responses in cortex also produce hyperpolarizing p.s.p,’s in cortical interneurons (Branch and Martin, 1958). This

NATUBE OF ELECTaOCORTICAL POTENTIALS

123

suggests that inhibitory p.s.p.’s may be generated a t diflerent sites in the synaptic pathways involved in the thalamo-pyramidal responses, shown in Figs. 17 and 18. Further indication that inhibitory elements other than those synapsing on apical dendrites are involved, is shown by the fact that hyperpolarizing p.s.p.’s “unmasked” in augmenting responses by topical application of GABA do not affect the patterns of tract discharges evoked by thalamic volleys (Purpura, 1958). Topical application of 7-aminobutyric acid results in prompt abolition of the surface negative components of both augmenting and recruiting responses (Purpura, 1958) and the appearance of high amplitude surface positivities. The application of nonspecific depressants (viz. cocaine or trauma to the cortical surface) also results in reversal of the surface negative recruiting response t o positivity (Arduini and Terzuolo, 1951). This observation prompted the latter to conclude that the residual surface positivity was attributable to persisting activity in thalamocortical afferent terminals. The “similarity” of effects produced by GABA and cocaine is more apparent than real. Following application of GABA, surface positivity is considerably greater in magnitude than encountered following application of nonspecific depressants ; the effects are rapidly reversible and all axosomatically relayed activity in cortex is unimpaired. Lack of cortical depression is also demonstrable by the fact that sensory after-discharge and the spontaneous cortical rhythms may be accentuated in cortex treated with GABA (Iwama and Jasper, 1957; Purpura et al., 195713). The following conclusions derive from the above: (1) All components of the primary evoked cortical potential with the exception of the first brief spike, seen most clearly in striate (Bishop and Clare, 1952b) and somatic-sensory cortex (Per1 and Whitlock, 1955), are attributable to postsynaptic events. The conclusions of Chang and Kaada (1950) and Malis and Kruger (1956), that the early deflections in striate cortex responses represent activity conducted to cortex in radiation fibers of different conduction velocity, are ruled out by the fact that all components but the first brief deflection are abolished by intravenously administered d-tubocurarine (Purpura and Grundfest, 1956). These results also obviate the possibility that the spikes might represent an unequal slowing of corticipetal impulses in unmyelinated intracortical terminals of afferent fibers (Bremer and Stoupel, 1956, 1957). (2) During successive synaptic excitation of elements in cortex, slow potentials develop at different cortical sites. Synaptic activity in dendrites results in surface negativity. (3) The sequence of events which follow these indicate that other synaptic organizations are invaded which may also be activated by afferents from a variety of sources. (4) The elements in-

124

DOMINICK

P. PURPURA

volved in these organizations engage corticospinal neurons, and the patterns of activity which the latter exhibit as a consequence of afferent stimulation reveal some of the complexities of these networks.

D. SYNAPTIC DISTRIBUTION OF TRANSCALLOSALLY EVOKED ACTIVITY The question which should be raised prior to considering the synaptic distribution of callosal afferents in cortex is the following: to what extent do callosal aff erents engage synaptic organizations which are fundamentally different from those activated by primary projection pathways? In terms of the immediate synaptic effects which callosal volleys induce in cortical elements some differences should be anticipated on the basis of anatomical differences in the distribution of afferent fibers. Chang (19534 has proposed that callosal aff erents terminate exclusively on dendrites, whereas Nauta (1954) maintains that although axodendritic relations predominate, axosomatic articulations are also encountered at different depths. Nevertheless, there is general agreement that the anatomical distribution of callosal aff erents is quite different from specific thalamocortical afferents. Electrophysiological data, however, indicate that similar synaptic organizations may be activated by both afferent pathways. Transcallosally evoked responses exhibit a wide range of variability in amplitude in different cortical areas (Curtis, 19408, b). The reasons for this are not immediately apparent. I n the cat, maximal amplitude responses are consistently recorded in suprasylvian gyrus (Curtis, 1940a; Chang, 1953s; Peacock, 1957) where most electrophysiological analyses of this variety of evoked activity have been attempted. Mean response latencies of units activated by contralateral stimulation of homologous sites range from 1-100 msec (Latimer and Kennedy, 1958). Surface activity (suprasylvian gyrus) commences with a short latency spike presumably signaling antidromic and orthodromic axonal activity followed by slow surface positivity which reflects synaptic activation of deeper lying elements, as in the case of primary afferent stimulation. [Peacock (1957) does not rule out the possible contribution of persistent afferent terminal activity to the deep negativity.] The surface positivity is terminated in surface negativity of 10-20 msec duration. The latter has characteristics similar to the surface negative components of primary evoked potentials and antidromically evoked responses and presumably represents a p.s.p. of apical dendrites. During the surface negativity and for some time thereafter (100 msec), evoked unit activity is detectable. The fact that some of the units may also be driven by peripheral stimulation supports prior observations relating to the interaction between transcallosally evoked and peripherally evoked responses (cf . Bremer

125


et al., 1956; Chang, 1953b; Perl and Whitlock, 1955). I n auditory cortex conditioning callosal volleys facilitate responses evoked in primary and secondary receiving areas (Bremer, 1953). Facilitation of primary responses is maximal between 30-100 msec after conditioning contralateral cortical shocks; earlier, “depression” is observed. Interaction patterns in auditory cortex are similar to those observed with paired auditory clicks (Rosenzweig, 1950) (vh., depression-facilitation-depression) and also mimic the patterns observed in visual cortex following paired radiation volleys (Clare and Bishop, 1952). The distribution of potential “sources and sinks” in cortex evoked by contralateral cortical and primary projection pathway stimulation is remarkably similar following both modes of cortical activation (Perl and Whitlock, 1955). These results suggest that in primary sensory cortex, callosal afferents engage synaptic organizations in a manner which is not fundamentally different from the way. primary thalamocortical volleys are dispersed in cortex (cf. also Malcolm and Smith, 1959). In visual cortex, transcallosally evoked potentials exhibit characteristics similar to primary evoked responses (Bremer, 1955). Landau and Clare (1956) and Peacock (1957) claim that Bremer’s results may be due to spread of stimulating currents from one hemisphere to the opposite subcortical radiations. However, the hypothesis proposed by Bremer et al. (1956) that the functional distribution of interhemispheric projections in primary sensory cortex is similar to that of radiation activity, is supported by the observations of Purpura and Girado (1959).

In unanesthetized-paralyzed cats contralateral motor cortex stimulation or direct stimulation of the corpus callosum evokes discharges in the pyramidal tract (recorded above the decussation) Short latency (2-5 msec) discharges are evoked by low-frequency stimulation, whereas higher stimulus frequencies (1&25/sec) evoke longer-latency (6-8 msec) , longer duration discharges. At still higher frequencies (50-200/ sec) short-latency discharges are again produced (Purpura and Girado, 1959). The relationship between pyramidal tract discharges and cortically evoked potentials following corpus callosum stimulation are shown in Fig. 19. The significant facts revealed by this analysis are: (1) relayed tract activity is initiated and terminated prior to the full development of cortical surface positivity; (2) complex interactions within interposed synaptic organizations which constitute the callosalpyramidal relay pathways are not reflected in the evoked surface responses; and (3) the pattern of corticospinal tract activity evoked by callosal stimulation is similar to that produced by lateral thalamic stimulation (Fig. 18), although there may be fundamental differences

.

126

DOMINICK P. PURPURA

in evoked cortical potentials. These results strongly suggest that projections from lateral thalamic nuclei and callosal aff erents engage similar organizations in motor cortex. These data not only support Bremer’s

H

20msec FIG.19. Effects of repetitive stimulation of genu of corpus callosum at different frequencies pn relayed pyramidal tract activity (upper channel) and evoked cbrtical potentials (lower channel). (Unanesthetired-paralyzed cat.) Latter recorded tramcortically (anterior aigmoid gyrus) ; pyramidal tract responses recorded with pair of lOOp Teflon coated wires. 1, l/sec; 2, lO/sec; 3, 25/sec; 4, 5O/sec; 5, 100/sec; 6, 2OO/sec, Reduction ip stimulus frequency in similar steps shown in 7-11. Note disappearance of short-latency relayed tract activity at 10-25/sec stimulation and reappearance a t 50-200/sec. Disappearance of shorblatency discharges at medium stimulus frequencies (10-25/sec) is aarociated with appearance of longlatency discharges, and marked augmentation of surface evoked responses. At 50200/sec evoked responses attenuate. Following tetaniration synchronous tract discharges are absent (9-10) but 8-10 sec later, potentiation of relayed tract activity occurs without significant alteration in surface evoked potentials. Note also shortlatency tract discharges are associated with early positive components of surface potentials. (From : Purpura and Girado, 1959.) r

observations, but also provide a physiological basis for interpreting the effects of callosal sectioning on the interhemispheric transfer of learned kinesthetic patterns in cats (Stamm and Sperry, 1957).


127

E. SYNAPTIC DISTRIBUTION OF “UNSPECIFIC” THALAMOCORTICAL ACTIVITY Repetitive stimulation (6-12/sec) of mid-line and intralaminar thalamic nuclei evokes a series of responses of variable latency, polarity, and distribution in relatively widespread regions of cortex which characteristically increase in amplitude during the first 5-6 stimuli (Morison and Dempsey, 1942). The latter designated such evoked potentials as “recruiting responses.’’ The anatomical organization of thalamic systems involved in the production of cortical recruiting responses has been a subject of considerable controversy. The terms “nonspecific, unspecific, or diffuse” are used interchangeably to describe the physiological characteristics of certain thalamocortical projection pathways. They do not indicate that thalamic nuclear groups are in themselves “diffuse” or nonspecific. Quite the contrary, the “non-specific system is a very specific system with its particular properties which distinguish it from other systems. It is diffuse in its distribution because it overlays projections from the specific system. In the thalamus, however, it is a very specific set of neurons which are diffusely distributed throughout the thalamus” (Jasper, 1954, p. 109). The organization and functional significance of the diffuse projection system have been reviewed (Jasper, 1949; Jasper and AjmoneMarsan, 1952).The major problems have been to define the anatomical substrate and nature of the projection pathway responsible for thalamocortical recruiting responses. Concerning the cortical projection of recruiting responses (which must be clearly distinguished from augmenting responses) there is now some agreement that the rostral pole of nucleus centrum medianum (CM) projects primarily t o frontal, motor and anterior cingulate cortex. The dominant projection of anteroventral portions of the intralaminar system is to rostral cortical areas while dorsolateral portions project predominantly to posterior areas of cortex (Jasper et al., 1955). The problem of whether or not recruiting responses are evoked from primary receiving areas of cortex (cf. Morison and Dempsey, 1942; Jasper and Droogleever-Fortuyn, 1947; Stare1 and Magoun, 1951 ; Hanberry et d.,1954) has also been “resolved” and it now appears that prior discrepancies can be related, as is often the case, to different operational teohniques. Suffice it to say, that recruiting responses in primary sensory areas are obtainable by unspecific thalamic stimulation, but are ordinarily quite labile and of lower voltage than those evoked in nonsensory areas (Jasper et aZ., 1955).Since publication of the latter report, which represented the combined efforts of investigators who held some-

128

DOMINICK P. PURPURA

what dissimilar views on the subject of thalamocortical projections, no serious discrepancies in data have been reported. Recruiting responses evoked in cortex by repetitive stimulation of unspecific thalamic nuclei have been analyzed in terms of the sites of origin and intracortical distribution of various components of the evoked potentials. Arduini and Terzuolo (1961) suggested that the surface negative, but not the surface positive components were generated transynaptically in cortex. Verzeano and associates (1953) adopted the view that all recruiting phases of the responses were of cortical origin on the basis of the following observations: (1) no recruiting responses were recorded with an electrode in a defect created by local aspiration of cortex; when the electrode was introduced into the subjacent white matter only small deflections were encountered which could be attributed to pick-up from adjacent intact cortex; (2) topical strychnine enhanced both positive and negative components of the evoked potentials; and (3) the positive phase of the recruiting response required stronger stimulation than the negative phase. The “laminar” distribution of evoked potentials following unspecific thalamic stimulation is entirely different from that following specific thalamic stimulation (Li et al., 1956a, b). In the cortical depths high amplitude relatively localized positive potentials are encountered which are not reflected in negativities of similar magnitude above or below these sites. As a consequence of the apparent complexity of the synaptic distribution of unspecific afferents in cortex, excitatory, and inhibitory interactions between the latter and specifically evoked activities are demonstrable, as noted above. There are features of the recruiting responses which clearly distinguish them from other forms of responses which “grow with repetitive stimulation” (viz. augmenting responses) and these demand special consideration. The pioneering investigations of Morison and Dempsey are too well known to require extensive review. For the present purposes, it should be recalled that the latter considered augmenting responses, which were relatively localized in sensorimotor cortex, following stimulation of medial lemniscus-internal capsule relays (Dempsey and Morison, 1942c) to be distinctly different from recruiting responses evoked by mid-line thalamic stimulation (Morison and Dempsey, 1942). They also separated “repetitive” burst phenomena observed following primary afferent stimulation from the spontaneous 8-l2/sec “bursts” characteristic of barbiturized animals. The latter bursts were identified with recruiting responses (Dempsey and Morison, 1942a). The relationship between various kinds of evoked potentials was established largely by analyzing


129

cortical projection maps and determining the degree of interaction between different responses (Dempsey and Morison, 194213). Utilizing a fundamentally different approach to the problem of defining the relation between different varieties of recurring spontaneous and evoked cortical potentials (viz., by observing the effects which these responses exerted on the activity of corticospinal elements) Arduini and Whitlock (1953) concluded that similar cortical organizations were involved in the production of spontaneous “spindle waves” and recruiting responses. This was later denied by Brookhart and Zanchetti (1956) who identified “spindle bursts” with augmenting responses and suggested that the latter might involve different synaptic organizations in cortex than those responsible for recruiting responses. Brookhart and Zanchetti’s observations have been confirmed in this laboratory (Purpura, 1958; Purpura et al., 1 9 5 8 ~ )Although . the initial synaptic events which lead to the production of augmenting and recruiting response are different, the subsequent production of surface negative components of both responses may involve activity of the same elements (Brookhart et al., 1957). In an attempt to resolve the fundamental differences between responses which increase in amplitude and those which ordinarily decrease during repetitive stimulation of aff erent pathways, Clare and Bishop (1956) have formulated an hypothesis which considers that both responses represent activation of similar elements ; differences being attributed to differences in cortical distribution and pathway of activation. All responses which decrement with repetitive stimulation (vie. primary evoked potentials) are termed Type I responses, while those which grow during repetitive stimulation are classified as Type I1 (incrementing evoked responses). It is proposed that Type I responses are conveyed to cortex via rapidly conducting (myelinated) axons whereas Type I1 responses are evoked by afferents of smaller diameter and slower conduction velocity, Clare and Bishop maintain that the long latency of recruiting responses is attributable to slow conduction in unmyelinated fibers from thalamus to cortex. [Others have argued that the long latency is due to passage of impulses through successive nuclear synapses on route to cortex (cf. Jasper, 1949).] According to Clare and Bishop, both types of responses are initiated at synapses via two (or more) afferent paths of different fiber types, but incrementing and decrementing responses may be generated in identical dendritic elements. To explain the incrementing nature of the Type I1 responses it is proposed that “the excitability cycle” of one variety of dendritic synapse shows two periods of facilitation. After a first stimulus, a second is effective for 15 msec

130

DOMINICK P. PURPURA

(which is the duration of the first response). A period of depression or relative refractoriness follows. Approximately 100-150 msec later, this depression passes over into supernormal excitability, viz. the repetition of an initial stimulus causes an increased response during the supernormal phase. Type I synapses presumably do not have excitability cycles with supernormal periods, hence repetitive stimulation a t 8-l0/sec evokes only decrementing responses in apical dendrites. While there is general agreement that the surface components of evoked potentials (and spontaneous spindles) are generated in dendritic elements, available data indicate that the interposed synaptic organizations involved in initiating augmenting or recruiting responses are not identical. This is indicated, in part, by the different effects which both responses exert on corticospinal neurons. But a more serious defect of Clare and Bishop’s hypothesis is the proposal that dendritic synapses have different “excitability cycles.” If it is inferred that the differences are attributable to different properties of postsynaptic membrane of apical dendrites then the hypothesis is embarrassed by the fact that postsynaptic membrane lacks regenerative action and is therefore incapable of showing “refractoriness,” “supernormality,” or any other variety of excitability change characteristic of axons. Furthermore, since Clare and Bishop (1955) propose that dendrites develop graded action potentials following synaptic activation, if postsynaptic membrane has none of the properties which could account for refractoriness or supernormality, then the latter phenomena must be attributable to other electrogenic sites on dendrites which are not postsynaptic loci. But, it is proposed that the dendrites may be synaptically activated a t two different loci to give two fundamentally different responses. I n the final analysis this would require that nonsynaptic dendritic membrane be capable of supporting graded responses at one locus which are fundamentally different from those generated at “a not too distant” site in the same element. An alternative suggestion has been proposed by Purpura and Grundfest (1956) to account for the differences in recruiting and augmenting responses which is based, in part, on the observations of Brookhart and Zanchetti (1956). It is suggested that both varieties of response are produced by activating different synaptic organizations in cortex. Thalamic aff erents which evoke augmenting responses engage elements initially in the cortical depths which in turn axosomatically discharge corticospinal neurons. At later stages of the synaptic dispersion, p.s.p.’s are generated in apical dendrites, Thalamocortical aff erents generating rmruiting ”reqmmhd6 fiot.appear to have access to synaptic pathways involved in the relay of short- or long-latency pyramidal tract dis-


131

charges but are capable of evoking p.s.p.’s in the same population of apical dendrites that are activated by some components of the organization responsible for augmenting responses (Brookhart et al., 1957). Thus, the different properties of the two classes of responses may be accounted for by differences in the mode of engagement of interposed synaptic organizations and not to differences in the properties of dendritic synapses as proposed by Clare and Bishop (1956). That cortical synaptic organizations responsible for recruiting responses in cortex are different from those involved in the production of augmenting responses is demonstrable by techniques other than those already described. When observations are made on cortical d.c. potential changes which accompany afferent excitation of cortex (cf. O’Leary and Goldring, 1953) it has been found that repetitive stimulation (40-601 sec) of lateral and medial thalamic foci is accompanied by surface negative potential shifts of 0.3-2 mv (Brookhart et al., 1958; Arduini, 1958). The potential shift accompanying lateral thalamic stimulation often exceeds the maximal negativity of augmenting waves, produced by 8/sec stimulation, but the negative shift accompanying medial thalamic stimulation rarely reaches the maximum magnitude of the negativity of the recruiting waves evoked by 8/sec stimulation. Slow cortical potential shifts of opposite polarity are produced during paroxysmal activity evoked by repetitive stimulation of intralaminar and lateral relay thalamic nuclei (Goldring and O’Leary, 1957). Although the exact nature of the cortical-surface d.c.-potential changes is not as yet clarified (nor for that matter the nature of the pre-existing cortical d.c. potential) it is reasonable to view slow potential changes as steady states of activity in dendrites (Clare and Bishop, 1955; Goldring and O’Leary, 1957). Sustained electrogenesis in response t o repetitive stimulation is characteridtic of Postsynaptic membrane (Fig. 5) (cf. Grundfest, 1957b). If slow cortical surface potential changes induced by intense afferent bombardment represent sustained synaptic electrogenesis of opposing varieties then it is likely that qualitative and quantitative differences in steady potential changes produced by lateral or medial thalamic stimulation reflect differences in the site of generation and/or relative magnitude of the two kinds of sustained synaptic activity developing in differently organized cortical elements. The characteristics of recruiting responses in different cortical areas may be attributable to differences in the intrinsic organization of cortical elements and their mode of engagement by unspecific afferents. These differences can be demonstrated when recruiting responses are made to interact with spontaneously occurring epileptogenic “spikes” induced by

132

DOMINICIC P. PPURPURA

local cortical freezing. Under these conditions, recruiting responses are relatively unaffected by “spike” discharges occurring in somatic sensory cortex, but exhibit clear interaction with “spikes” induced in suprasylv i m “association” cortex (Fig. 20) (Smith and Purpura, 1959).

I

ss

0

ss

A

I

I

1

I

6

0: PS

I- I1I 1

H 100C

I1I I I I I I I I I I1 I I I I I I I I I 1 I I I I I I-11

IOen”

1 ooc H

FIQ.20. Interaction patterns of recruiting responses evoked by 8/sec stimulation of unspecific thalamic projections with spontaneously discharging “epileptogenic spikes” induced by local cortical freezing. A-C : three different preparations, unanesthetized-paralyzed cats. A: slow (1) and fast (2) records of “spikes” and recruiting responses evoked in anterior supraaylvian (es) gyrus. Note sudden “waxing” of recruiting response after each “spontaneous spike,” then rapid “waning.” Thalamic stimulus (indicated by signal marker below records) is continuous throughout. B: recruiting responses evoked in posterior sigmoid (pa) g y ~ do s not interact with focal “spikes” induced in this cortical location. C: cortical “spikes” induced in sa and ps by freezing lesions in both areas. Recruiting responses generated by thalamic stimulation are affected by “spikes” in as, but not in pa. Note in particular timing of “waxing and waning” and focal “spike” discharge in ss (Cl). (From: Smith and Purpura, 1969.)


133

Summary. Activation of cortical elements by unspecific aff erents from thalamus proceeds in a manner that is fundamentally different from the mode of engagement of cortical neurons by specific or other lateral thalamic afferents. Unspecific afferents do not appear to have access to relay elements in the cortical depths which activate corticospinal neurons, “Augmentation” and “recruitment” are largely dendritic synaptic phenomena; successive volleys delivered to interposed organizations of excitatory and inhibitory elements evoke p.s.p.’s in dendrites which constitute the surface potentials. Interaction studies indicate that some components of the cortical relay pathways to dendrites may be activated by afferents from medial and lateral thalamic nuclei. Spontaneous “spindle bursts” resemble augmenting not recruiting responses in terms of the effects on corticospinal elements which accompany these activities. Different synaptic organizations appear to be involved in the production of the initial components of augmenting and recruiting responses, but some interaction between these organizations is detectable during later stages of apical dendritic activation.

F. DISTRIBUTION OF CAUDATE-CORTICAL ACTIVITY Brief consideration of caudate-cortical relations illustrates the complex effects which are produced in cortex by stimulation of subcortical nuclei which have cortical, as well as pronounced extra-cortical, projections. In intact preparations, single stimuli to the head of the caudate evoke short- and long-latency responses in cortex which represent, predominantly, activation of apical dendrites (Purpura et al., 19581). Like unspecific thalamic afferents, caudate-cortical afferents do not engage elements which activate corticospinal neurons. A stimulus which evokes cortical responses also activates projection pathways to thalamic and presumably other subcortical relay nuclei (Stoupel and Terzuolo, 1954; Shimamoto and Verzeano, 1954) which in turn, condition caudate as well as cortical organizations to subsequently applied stimuli. A conditioning stimulus produces profound long-duration alterations in the synaptic organizations responsible, in part, for the evoked potential, but these alterations are not reflected in overt surface activity. The synaptic pathways activated by caudate volleys may be analyaed by comparing the effects of paired stimuli on evoked cortical potentials in intact and “telencephale isol6” preparations (Purpura et al., 1958f). In intact preparations conditioning stimuli evoke short latency, surface negative responses in pericruciate cortex (Fig. 21). Testing responses evoked a t varying intervals after conditioning responses exhibit the following alterations: ( a ) development of brief early positivity; ( b ) development of a second negative-positive component during the time

134

DOMINICK P. PURPURA

occupied by the positivity of the unconditioned response (Fig. 21,4-8) and finally, (c) after 400 msec loss of late negativity and recovery of the initial response. When paired stimuli are delivered t o the caudate nucleus in “telencephale isol6” preparations, activity cycles are dramatically altered (Fig. 22). Under these conditions, testing responses

1-6 M 7-9

H

8 1 SEC.

FIQ.21. Cortical effects of paired caudate stimuli in intact unanesthetizedparalyzed preparation. Responses evoked from rostra1 pole of anterior sigmoid gyrus. 1 and 2: conditioning and testing responses, respectively. 3-9: alterations in testing responses a t different shock intervals. Minimal facilitation of surface negatively early (3), later depression of initial negativity of testing responses ( 4 4 3 , development of early positivity and a late diphasic (negative-positive) response which was maximal at 126-250 msec and persisted for 300 msec. Five superimposed records throughout (from: Purpura et al., 1958f).

exhibit loss of early negativity and development of late negative-positive sequences (4-20 msec) after a conditioning stimulus (Fig. 22, A-C). Testing responses are absent after this initial period for as long as 200 msec. These and other data have been incorporated in a scheme which


135

provides a reasonable approach to the analysis of the contributions of various synaptic pathways involved in the production of the evoked potential changes (Purpura et al., 1958f). From these analyses, it has been inferred that in intact preparations at least 9 excitatory and 4 inhibitory pathways are activated by single caudate volleys. Thalamic

FIG.22. Alterations in caudate-cortical responses induced by previous activity in “telencephale isolP preparation. Conditioning and testing responses from peri-cruciate area shown, respectively, in F and G. A-E: at arrow, testing shock delivered at time (in msec) shown on right. “New” response occurs within 4 msec but rapidly disappears. Depression of testing responses persists as long as 200 msec. Compare interactions with those produced in intact preparations (Fig. 20). (From: Purpura, et al., 1958f.)

activity evoked by caudate stimulation is reflected in cortex in terms of 3 temporally distinct, excitatory synaptic events which are not detectable as overt potential alterations, but are only revealed by conditioningtesting procedures. The complex, but orderly, interactions of cortical potentials evoked by paired caudate volleys appear to be, as yet, without parallel in cortical electrophysiology (Figs. 21, 22).

136

DOMINICK P. PURPURA

G. SYNAPTIC DISTRIBUTION OF “RETICULOCORTICAL” AFFERENTS Since Magoun’s rediscovery (1944) of bulbospinal inhibition (cf. Sechenov, 1863) the complex neuronal organizations which constitute the f m a t i o reticularis or substantia reticularis have been intensively studied anatomically, physiologically, and pharmacologically. [No less than 38 reviews and 6 published symposia are cited in 682 references collated by Rossi and Zanchetti (1957). O’Leary and Coben (1958) have also critically reviewed most of the literature pertaining to the “reticular core.”] From present indications (cf. Fischgold and Gastaut, 1957; Brazier, 1959) it appears unnecessary to add that future investigations will define more clearly the role of reticular organizations in various forms of “higher nervous activity.” “Generalized” electrographic effects similar to those produced by brain stem stimulation (Moruzzi and Magoun, 1949) are observed in cortex following high frequency stimulation of the caudate (Shimamoto and Verzeano, 1954; Stoupel and Terzuolo, 1954), cerebellum (Moruzzi and Magoun, 1949), mid-line thalamic nuclei (Jasper, 1949), hypothalamus (Morison et al., 1943; Murphy and Gellhorn, 1945), as well as stimulation of neocortex (Bremer and Terzuolo, 1953), cingulate gyrus (Sloan and Jasper, 1950), other parts of the limbic system, and amygdala (Feindel and Gloor, 1954). When such effects are compared with those produced by strong afferent stimulation or brain stem reticular stimulation, it is possible that fundamentally different synaptic organizations may underlie the cortical electrographic effects observed in each instance. This appears to be evident already from the fact that high frequency mid-line thalamic stimulation produces different enduring potential changes than those evoked by midbrain reticular stimulation (cf. Arduini, 1968). In the following sections, only the effects produced by sensory or brain stem stimulation will be considered, but the reader must be aware that the term “reticulo-cortical projection” is used without commitment as to the exact origin, course, or composition of the projection pathway from brain stem to cortex. The problem of defining the mechanisms which underlie the development of cortical electrographic “arousal” following intense reticular stimulation can be somewhat simplified by examining: ( a ) the effects of reticular stimulation on identifiable elements in cortex; ( b ) the distribution of evoked reticulocortical potentials; ( c ) the interaction of the latter with other cortically evoked responses. ( a ) Effects of reticular stimulation. It is most certainly true, as Moruzzi (1953) has insisted, that results obtained by examination of a limited number of elements such as corticospinal neurons should not be

NATUBE OF ELECTROCORTTCA~Jh3“#N$fAL$ 137 generalized. However, it is equally true that because of the prominent role played by these elements in the production of surface potentials any consistent effects which are observed in their responsiveness to various modes of excitation should be considered highly indicative of a significant event. When all varieties of experiments dealing with the effects of repetitive sensory or reticular stimulation on pyramidal tract discharges are reviewed, the most consistent finding in all of these has been the observation that during electrographic arousal spontaneously discharging units in the pyramidal tract are abolished or reduced in frequency (Adrian and Momzzi, 1939; Whitlock et al., 1953; Calma and Arduini, 1964). In some instances increased frequency of spontaneous activity has been noted (Calma and Arduini, 1954; Dondey and Machne, 1955). Utilizing the “pyramidal” preparation of Whitlock et al. (1953), Brookhart and Zanchetti (1955) observed that high frequency diencephalic stimulation “stabilized” the responsiveness of corticospinal neurons to motor cortex stimulation. Purpura (1967a, b) has observed that the direct and indirect responsiveness of corticospinal neurons is reduced during and for a few seconds after intense brain stem stimulation in intact unanesthetized-paralyzed preparations. It has been proposed (Purpura, 1957a) that high-frequency brain stem reticular stimulation results in activation of elements in cortex capable of inhibiting the discharge of corticospinal neurons. (It is not known whether lower frequencies of stimulation or single volleys are capable of initiating or facilitating the discharge of corticospinal neurons.) Clearly, by the very nature of the intense afferent bombardment which is required to induce electrocortical activation, it is unreasonable to expect that the discharge of all units encountered by exploring microelectrodes will be abruptly silenced by long trains of high-frequency volleys from diffuse sources. ( b ) The distribution of evoked reticulocortical potentials. Analysis of the nature and distribution of evoked reticulocortical potentials suggests that different synaptic organizations are involved in their production. In unanesthetized preparations single volleys evoke diphasic positive-negative responses in rostra1 cortex which are compounded of depolarizing and hyperpolarizing p.s.p.’s (Purpura, 1958). In anterior and middle suprasylvian gyrus (“association” cortex) evoked potentials are initially surface negative, indicating that in these regions reticulocortical projections engage elements predominantly axodendritically. In barbiturized preparations long-latency discharges are evoked by medial brain stem reticular stimulation (Purpura, 1955) which have been identified with the “secondary discharges” of Forbes and Morison (1939). Like the responses obtained in the unanesthetised animal, the long-latency diffuse discharges evoked by reticular stimulation of differ-

138

DOMINICK P. PURPURA

ent polarity in different cortical regions, but the general characteristics of the potentials evoked in barbiturized preparations are otherwise remarkably dissimilar. Differences in the synaptic distribution of reticulocortical afferents in different cortical areas are revealed by interacting surface-evoked responses with reticulocortical potentials. It has already been noted above (Fig. 12) that prior surface activity evoked by direct cortical stimulation blocks development of reticulocortical responses in suprasylvian gyrus. The inhibitory interactions are accompanied by reversal in polarity of the tested reticulocortical activity, similar to the effect produced by topical GABA (Purpura, 1958). However, when the “direct” cortical response follows the reticulocortical response, marked augmentation of the tested surface response is observed. In rostra1 cortical areas, conditioning cortical stimuli do not significantly alter evoked reticulocortical responses. The pattern of interaction between reticulocortical and “direct” cortical responses in suprasylvian gyrus of cat is reproduced by interactions between caudate and “direct” cortical responses in pericruciate cortex. This suggests that both subcortical projection pathways may engage cortical neuronal organizations in different cortical regions to effect similar actions (Purpura, 1958). ( c ) The interaction of cortically evoked responses. Although reticulocortical activation blocks or reduces a variety of spontaneous and evoked cortical potentials, considerable difficulty is encountered in deciding whether or not the “inhibitory” effects are primarily cortical or subcortical. Reticulocortical blockade of recruiting responses (Moruzzi and Magoun, 1949; Jasper et al., 1955; Evarts and Magoun, 1957), augmenting responses (Gauthier et al., 1956), and primary responses evoked by thalamic stimulation (Desmedt and LaGrutta, 1957) has been reported. While some of these blocking effects may result from inhibitory actions at subcortical and cortical sites, responses evoked by cortical stimulation are also markedly affected. Reticular stimulation in unanesthetized-paralyzed cats results in prompt reduction of surface evoked dendritic potentials (Purpura, 1956). Following a 6-sec stimulus to the brain stem, the surface response to local stimulation exhibits cyclic changes in amplitude and duration and periodically a surface positivity appears (Fig. 23) (Purpura, 1958) . Following recovery, a second train of stimuli induces a similar sequence of changes in the tested evoked responses which persist for a comparable period of time. In any one preparation the cyclic events develop in a remarkably orderly fashion. Similar effects are observed on transcallosally evoked potentials which have been altered by topical application of y-aminobutyric acid. Tested responses under these conditions are composed of high amplitude

A

FIG.23. Sustained effects of reticulocortical activation on surface-evoked dendritic depolarizing PB.P.'S. Single denhit@ responses evoked every 2 sec from anterior suprasylvian gyrue in unanesthetized-succinylcholine paralyzed cat. The white bars above the fourth, fifth, and sixth reaponsea in A, B, and C indicate the duration of the brain stem reticular stimulation (200 cps). Stimulating electodes located in posterior tegmentum. The record shows a continuous series of 3-min duration. Note the complex change in evoked dendritic response after each stimulation of the brain stem, with recovery occurring at about 1 min after stimulus in A, B, and C. The periodic nature of the slow oscillation in responsiveness can be readily appreciated. (From: Purpura, 1957b.)

140

DOMINICK P. PURPURA

surface positivities which presumably represent summation of early axosomatic p.s.p.’s in the cortical depths and hyperpolarizing p.s.p.’s of apical dendrites (Fig. 24) (Purpura et al., 1969~).Brain stem stimulation tends to abolish most of the surface positivity, but following stimulation, cyclic alterations are observed which are completely reproducible. As yet, it is difficult to satisfactorily account for these sustained, cyclic events in specific detail; a difficulty which reflects ignorance of the complex manner in which synaptic organizations throughout the entire

Fra. 24. Sustained effects of reticulocortical activation on transcallosally evoked responses in supraaylvian gyrus, following topical application of 1% GABA. Transcallosally evoked potentials exhibit large GABA-induced surface positivity of 40-60 m e c duration and later negativity of similar duration. Responses evoked every 2 sec; continuous series, 2 min duration. After the third response in each row, stimulation of the midbrain tegmentum for 4 sec (200 cps) produced prompt loss of late GABA-induced surface positivity in the fourth and fifth responses. After cessation of the brain stem stimulus the testing transcallosally evoked responses exhibit cyclic alterations in amplitude and duration. Approximately 30 sec are required for the GABA effect to “break-through,” Note “spontaneous” reversal of GABA effect after brain stern stimulation between ninth and fourteenth response in each series and again 6 sec later (17, 18). (From : Purpura et al., 19590.)

cortex are activated by reticulocortical volleys. The aforementioned sustained effects produced by reticular stimulation are also reflected in steady-potential changes in cortex, but even these are quantitatively different in different cortical areas (Arduini et al., 1957). That such effects may come about by complex interactions between excitatory and inhibitory organizations of cortical neurons cannot be seriously challenged. It seems probable that when activated by an adequate stimulus, different organizations of neural elementa may interact for long periods

141 of time. In the case of reticulocortical activation, the adequate peripheral or brain stem stimulus is primarily frequency determined. [The fact that a variety of pharmacological agents are capable of reproducing the electrographic effects of peripheral or “direct” electrical stimulation of the brain stem (cf. review of Rossi and Zanchetti, 1957, and Dell, 1958) does not invalidate this assumption, but merely indicates that postsynaptic membrane of reticular formation neurons is exquisitely reNATURE OF ELECTROCORTICAL POTENTIALS

A=O.5/sec B=Z.O/ sec C: l 5 / s e c D=50/sec

1

SEC.

FIG.25. Alterations in evoked reticulocortical responses induced by brain stem reticular stimulation at different frequencies. Monopolar responses recorded from anterior syprasylvian gyrus. Low-frequency Od/sec stimulation evokes surface negative response followed by repetitive response. “Alternation” is observed at 2/sec (B) and 15/sec (C) but responses “break-through” every 125 msec in (C) whereas they are absent in (B) a t 600 msec intervals between stimuli. In (D) the evoked potentials are observed at shorter intervals and the electrographic pattern begina to take on the characteristics of “activation.” (From: Purpura, 1958.)

sponsive to various species of compounds (adrenergic, cholinergic, etc.) in addition to those which are endogenously liberated a t presynaptic terminals in different reticular organizations.] Significant clues as to the processes which are set into operation by a brain stem stimulus of progressively increasing frequency are illustrated in Fig. 25. (Purpura, Housepian and Grundfest, 1957d; Purpura, 1958). Low-frequency stimulation results in characteristic “alternation” of evoked responses, but

142

DOMINICK P. PURPURA

as the stimulus frequency increases, responses are evoked at progressively shorter intervals, and finally a t the “threshold” frequency for “activation” (BO/sec; Moruzzi and Magoun, 1949) all responses are reduced in amplitude but almost all are detectable. The blockade of alternate reticulocortical evoked responses a t 2/sec cannot be attributed to “refractoriness,” since with higher frequencies “breakthrough” occurs. For reasons that have been amplified above, reduction in responses at higher frequencies also cannot be explained in terms of “refractoriness.” Clearly, the only reasonable hypothesis which accounts for the observed effects (Fig. 25) is to assume that different synaptic pathways in cortex are capable of showing complex interactiona of an excitatory and inhibitory nature; the over-all effects of these interactions being reflected in a rapid sequence of dendritic depolarizing and hyperpolarizing p.s.p.’s of different proportions in apical dendrites. The existewe of interneurons which praduce inhibition and excitation interposed between reticulocortical aff erents and apical dendrites insures that the over-all effects of repetitive stimulation will represent admixtures of both kinds of synaptic electrogenesis. For reasons which are not immediately apparent, high-frequenoy reticulocortical stimulation is particularly effective in permitting inhibitory build-up in different cortical organisations. It is remarkable indeed, that the complex interactions described above have a parallel in the effects which different parameters of corticalathulation exert on single spontaneously discharging Betz cells (cf. Fig. 9 in Branch and Martin, 1958).

111. Analysis of Different Synaptic Organizations in Cerebellar Cortex

In previous sections differences in surface potentials recorded from a common locus have been analyzed in terms of: (1) the origin and distribution of various corticipetal afferents, (2) mode of engagement of cortical neurons, (3) subsequent synaptic dispersion of afferent volley, and (4) organiaation of excitatory and inhibitory components which constitute the interposed synaptic pathways to various parts of the same or different neuronal elements. Further analysis of surface potentials fi-om different loci in the cerebral cortex is facilitated by relating these factors t o basic differences in cytoarchitecture. Differences in neuronal organizations are inferred from architectural heterogeneity and such regional differentiation provides, in part, a basis for functional localization. c In oontrmt to the cerebral cortex, the cerebellar cortex is of little interest to the cartographer. Its homogeneity provides no clues to func-


143

tional localization and the physiologist is struck by the apparent lack of discreteness of afferent input (Bremer and Bonnet, 1951) that characterizes cerebral primary projection cortex (Mountcastle, 1957; Mountcastle et al., 1957). The mossy fiber-granule cell articulation represents the major mode of engagement of afferent fibers in cerebellar cortex, and the seemingly characteristic relations of granule cells to other elements, near and far, precludes “localization.” Arguments in favor of assigning a more important role to the climbing fibers in localization processes have been presented (Scheibel and Scheibel, 1954). Still, the conclusion of Jansen and Brodal (1954) is worth noting: “In view of the multifarious intra-cortical cerebellar pathways demonstrated by anatomical methods, it appears to be an almost hopeless task to attempt with available neurophysiological techniques to decipher precisely what goes on in the cerebellar cortex” (p. 379). The problem of “what goes on in the cerebellar cortex,” or more precisely, how a structure with uniform architecture and restricted mode of activation handles diverse signals differently is one of defining the physiological differences in synaptic organizations. There are, as yet, no morphological criteria for distinguishing excitatory from inhibitory neurons and methods must be sought which permit evaluating the relative contribution of their respective activities in any evoked potential complex. Attention has been focused in this review on the use of specifically acting pharmacological agents either alone or in various combinations. Application of these pharmacological techniques to the problem of defining cerebellar cortical organizations has provided significant clues as to the nature of intrinsic organizations which have not been forthcoming by examination of electrophysiological data alone (cf. Dow and Moruzzi, 1958).

A. RESPONSE TO SURFACE STIMULATION As in the case of the cerebral cortex, direct stimulation of the cerebellar cortex evokes a relatively simple surface negativity of approximately 10-20 msec duration which is detectable up to 5 mm along a single folium from the site of stimulation (Dow, 1949). The response is attributed t o synaptic activation of Purkinje cell dendrites by fibers excited in the molecular layer. It is reversibly blocked by intravenous d-tubocurarine (Purpura and Grundfest, 1956). Unlike surface responses of the cerebral cortex surface evoked cerebellar negativity does not include opposing surface positive, hyperpolarizing p. s. p.’s as revealed by various tests. The lack ,of hyperpolafiising ‘ synapses on superficial dendritic ter-

144

DOMINICK P. PURPURA

minals of Purkinje cell dendrites was initially inferred from pharmacological and physiological data (Purpura and Grundfest, 1957b) and subsequently confirmed by tests employing selectively acting synaptic blocking agents such as 7-aminobutyric acid (GABA) and the higher chain homologs (Purpura et al., 1957a, b). Topical application of GABA abolishes surface negativity in cerebellar cortex by inactivating excitatory synapses, but in contrast to cerebral responses, no surface positivity is “unmasked” during the blockade of depolarizing synaptic electrogenesis. Longer chain o-amino acids which are “convulsant” in cerebral cortex by virtue of their blocking effects on hyperpolarizing p.s.p.’s are inert on cerebellar cortex (Figs. 9, 10). Failure to significantly affect superficially evoked p. s. p.’s of Purkinje cell dendrites with agents which augment surface negativity in the cerebral cortex following topical application is attributable to the relative paucity of hyperpolarizing synapses on these structures activated by fibers in the molecular layer. Contrary to the views originally expressed by Purpura and Grundfest (1957b) concerning the relative absence of inhibitory synaptic electrogenesis in the cerebellar cortex of neuraxially intact preparations, it is now clear that although the superficial response to direct stimulation is exclusively a depolarizing p.s.p., activation of cerebellar dendrites via interposed elements driven by afferent volleys from various sources may evoke hyperpolarizing p.s.p.’s in these structures in intact, as well as decerebrate preparations, but inhibitory activity is always of greater magnitude in the latter. Whether or not a cerebellar evoked potential will be compounded of dendritic depolarizing and hyperpolarizing p.s.p.’s depends largely on: (1) the source of the afferent volley; (2) cortical projection area; and (3) relative complexity of the interposed projection pathway (Purpura et al., 1959a).

B. RESPONSES TO AFFERENT STIMULATION Cerebellocerebellar projections. Different areas of the cerebellar cortex are connected to each other by “associati4n” fibers similar to those observed in the cerebral cortex, (e.g., stimulation of anterior vermal cortex results in relatively discrete evoked potentials in posterior vermal regions (cf. Dow and Moruzzi, 1958). Topical application of o-amino acids are without effect on cerebellocerebellar responses (Purpura et al., 1959a) from which it may be inferred that in the cerebellar cortex “association” pathways do not utilize axodendritic activation of Purkinje cells. This is in striking contrast to the manner in which the association pathways of the cerebral cortex activate distant elements, as noted above. Olivocerebellar projections. These have been studied in detail by a

145

NATURE OF ELECTROCORTICAL POTENTIAGI

number of workers (cf. Dow and Moruzzi, 1958). Like cerebellocerebellar projection pathways, olivocerebellar responses are evoked by stimulation of rapidly conducting pathways which are of relatively simple composition. The characteristic responses, triphasic and of short duration, are unaffected by topical application of o-amino acids (Fig. 26A).

2

I

A

H

Smsec

6

3

H

20msec

C

H

IOmsec

Fxo. 26. Differential effects of GABA and e-aminocaproic acid (C,) on responses evoked in paramedian lobule by different pathways of stimulation. Three Merent neuraxially intact, succinylcholine-paralyzed preparations. Note different time scales of the recordings. A: stimulation of contralateral inferior olive. B : paired stimuli delivered to contralateral mid-pontine reticular formation. C : stimulation of contralateral posterior sigmoid gyrus. In each experiment the records of column 1 represent control responses, of 2, responses after application of buffered GABA (1%) and, 3, after applying Co. Olivocerebellar responses are unaffected by GABA and CS; pontocerebellar responses are moderately affected, especially the testing response of the pair; cerebrocerebellar responses are markedly affected by GABA and Ce. (From: Purpura et 02, 1959a.)

Pontocerebellar projections. With increasing complexity of the afferent projection pathway, evoked responses become compounded of inhibitory as well as excitatory activities which may be revealed by topical application of o-amino acids and/or paired conditioning testing stimuli. Mid-pontine stimulation evokes surface-positive responses in contralateral paramedian lobule (Jansen, 1957) (Fig, 26B, 1). Activation of the pontocerebellar pathway produces long lasting alterations in synaptic organizations not reflected in an overt potential change, but demon-

146

DOMINICK P . PURPURA

strable by a second stimulus dclivered 40 msec after the conditioning activity. Under these conditions surface positivity is lost and a 20 msec surface negativity is prominently displayed (Fig. 26B, 1 ) . The latter is abolished by topical GABA (Fig. 26B, 2 ) , but relatively unaffected by e-aminocaproic acid (Fig. 26B, 3 ) . Increase in amplitude and duration of the surface positivity of the Conditioning response is also observed (Fig. 26B, 2 ) . (It is of considerable interest to note that the appearance of surface negativity in a response originally surface positive produced by thc interaction of paired responses (Fig. 26B, 1) may be viewed as the counterpart of Fig. 12 in which the interaction of cortical-reticulocortical responses resulted in the appearance of surface positivity in an evoked response initially surface negative!) The interaction between paired volleys shown in Fig. 26B, 1 cannot be attributed to “refractoriness” since with closely spaced stimuli, the second evoked response of the pair is only slightly reduced in amplitude (Purpura et al., 1959a). Activity cycles of this nature cliaracterizcd by reappearance of evoked potentials at short intervals with disappearance a t longer intervals between stimuli are commonly observed in cerebellar cortex following stimulation of a variety of afferent pathways (AlbeFessard and Szabo, 1954; Jansen, 1957). They are also observed in the interaction between paired reticulo-cortical (cerebral) rcsponses (Fig. 12) (Purpura, 1958), but never in responses evoked by stimulation of primary projection pathways to cerebral cortex. Inhibitory processes which underlie these interactions presumably have their origin in the pontine reticular system as previously suggested (Purpura and Grundfest, 195713). This is demonstrated by the fact that vertical displacement (1 mm) of the stimulating electrode in the pons can effectively eliminate the inhibitory interactions (Fig. 27). As in the case of caudate-cortical projections (Purpura et al., 1958f) excitation of parallel excitatory and inhibitory pathways of different duration of action can result in a variety of complex synaptic interactions. Spinocerebellar projections. In decerebrate preparations, peripheral nerve stimulation produccs a well-known sequence of potentials in 110molateral anterior lobe which has been analyzed by Grundfest and Campbell (1942), Carrca and Grundfcst (1954), and others (cf. Dow and Moruzzi, 1958) . Potentials evoked by forelimb and hindlimb stimulation and recorded from the same locus in homolateral anterior lobe are of similar configuration and the effects of topically applied 0-amino acids are identical in both cases (Fig. 28). GABA abolishes early negativity and augments the amplitude and duration of the prominent surface positivity. e-Aminocaproic acid abolishes all surface positivity and an extraordinarily high-amplitude surface negativity ensues. The effects

NATURE OF EL~TROCORTICALPOTENTIALS

147

produced by craminoaaproic acid are entirely comparable t o those produced by topical application of strychnine (Bremer and Bonnet, 1953; Jansen, 1957; Pupilli et aE., 1956). It is inferred from the reconstruction (center, Fig. 28) that spinal afferents edgage cerebellar elements axo-

I

I

0.1 s e c Fxo. 27. Different physiological effects produced in correlation with apparently identical cerebellar activities. Neuraxially intact, succinylcholine-paralyzed preparation. Monopolar recording from surface of paramedian lobule. Stimuli delivered to contralateral upper pontine reticular formation. 1, 2 : cerebellar responses t o single and paired stimulations. The testing response in 2 was not greatly affected by the conditioning activity which preceded by about 60 msec. 3, 4: responses at same recording point evoked after stimulating electrodes were raised about 1 mm in the pons. The single response (3) and the conditioning activity were not distinguishable by magnitude or time course from their homolog in 1 and 2. The testing stimulus following the conditioning activity by 60 msec in this case elicited only a small response. 5, 6: the stimulating electrodes were lowered again to the original site and produced the same effects as seen in 1 and 2. (From: Purpura et al., 1959a.)

somatically ant1 axodendritically, but that a late surface inhibitory p.s.1). presumably prevents the development of sustained excitatory bombardment of Purkinje cell dendrites. Sinall though it is in comparison to the profound excitatory events, the inhibitory action is strategically

148

DOMINJK!K P. PURPURA

FIQ. 28. Analysis of postsynaptic potentials contributing t,o t,lie cerebellar responses evoked by stimulation of radial and sciatic nerves. Decerehrate preparation, all records made from one recording site on the anterior vermia, near ccreheller midline. 1-4 : stimulation of homalateral radial nerve. 5-8: of homolateral sciatic nerve. Initial control responses 1 and 6. Responses after topical application of buffered 1% GABA (Cd, 2, 6: after applying s-aminocaproic acid (G), 3 and 7: after rinsing cerebellar cortex 4 and 8. Center figure: Superposed records of control response, of that after GABA and after Cs produced by stimulating the radial nerve. The surface positive potentials are composed of a large component (horieontal bars) which represents chiefly axosomatic ps.p.‘s and a smaller potential (large dots) which is contributed by hyperpolarizing (inhibitory) dendritic ps.p.’s. Both sum to make up the response seen on applying GABA. I n the response of the untreated cortex only a small surface negative excitatory dendritic ps.p. is generated (small dots). When the inhibitory synapses are blocked by C. a second normally inhibited synaptic system is brought into activity and this gives rise to a large surface negativity shown b y the diagonal lines. (From: Purpura et al., 1959a.)

located in the synaptic pathway, and thus effectively limits full dispersion of the afferent volley. Cerebrocerebellar projections. Like spinocerebellar responses, potentials evoked by stimulation of contralateral cerebral cortex reflect the activities of excitatory and inhibitory organirations in cerebellar oortex.

149 The relative proportion of these activities vanes in different cerebellar cortical regions (Purpura et al., 1959a). The electrophysiological details of the evoked potentials have been recently explored by Jansen (1957) NATURE OF ELECTROCORTICAL POTENTIALS

Fro. 29. Analysis of components in response of paramedian lobule evoked by stimulating contralateral pericruciate cerebral cortex. In each sequence the evoked potentials are preceded by a surface negativity induced by stimulation of the cerebellar cortex at a site 4 mm from the recording electrode. 1 and 4: control responses. 2: after applying buffered 1% GABA; 3: reversal of drug action b y riming cortex with warm Ringer solution; 5: after applying 1% e-aminocaproic acid ( C d ; 6: reversal of drug action. The initial surface-evoked negativity waa eliminated with GABA but unaffected by C,.The second surface positivity of the cerebrocerebellar evoked response was augmented by GABA (2) and eliminated by Co (5). The latter action waa accompanied by development of prominent negativity. Center: Analytical reconstruction showing components involved in these responses. Superposed records of control response, of that after GABA, and after CB. The surface positive potentials are composed of an initial component (horizontal bars) which represents axosomatic ps.p.’s and a large component (large dots) which is contributed by hyperpolarizing (inhibitory) dendritic ps.p.’s. A surface negative depolarizing dendritic component (small dots) is also shown which is “unmaaked” during blockade of the hyperpolarizing dendritic p.s.p.’s by Ce. (From: Purpura et al., 1959a.)

and earlier by Curtis (19404, Dow (1942), Hampson (1949), and Scabo and Albe-Fessard (1954). Examples of the effects produced by o-amho acids on evoked responses in paramedian lobule by contralateral cerebral cortical Stimulation are shown in Fig. 26C and Fig. 29. In the latter, the

’

150

DOMINICK P. 'PURPURA

responses are preceded by surf ace negativities evoked by direct stimulation of the cerebellar cortex. The signifiaant alterations induced in the cerebrocerebellar evoked responses by the amino acids are: (1) increase in duration and magnitude of surface positivity by GABA; (2) abolition of late positivity and development of early negativity following c-aminocaproic acid. Analysis of the evoked potentials under these conditions (center Fig. 29) indicates that the response evoked by cerebral cortical stimulation in paramedian lobule is compounded of a t least Sve varieties of p.s.p.'s, the most prominent of these being early depolarizing and late inhibitory p.s.p.'s of relatively large amplitude generated in Purkinje cell dendrites by interposed excitatory and inhibitory elements. (IF RESPONSES C. EFFECTSw "-AMINOACIDSON DIFFERENTVARIETINS EVOKED FROM THE SAMELocus Surface responses evoked by stimulation of different afferent pathways are often recorded from a common locus in the cerebellar cortex. Considering the limited possibility that differences in evoked potentials can be aljtributed to the manner in which various afferents engage cerebellar cortical neurons, it is likely that some degree of selectivity is conferred by virtue of the specific connections which first order elements make with other intrinsic neurons1 organizations and the extent to which other parallel excitatory and inhibitory pathways are activated by corticipetal aff erents. These factors are illustrated by comparing the effects of "-amino acids on responses evoked from paramedian lobule by stimulation of different afferent pathways. Neither GABA .nor c-aminocaproic acid alter olivoparamedian responses, but minimal effects are noted on pontoparamedian evoked potentials (Fig. 26). On the other hand, responses evoked by cerebral cortical stimulation and recorded from the same locus are markedly affected by the @-aminoacids (Fig. 26, C, Fig. 29). Jansen (1957) observed a similar differential effect of topical strychnine on these three varieties of responses evoked from paramedian lobule. Data are insufficient as yet to permit construction of "circuit diagrams" to explain the differential action of w-amino acids on these cerebellar evoked potentials. Still, certain conclusions are permissible concerning the nature of the surface responses. The striking differences observed following application of GABA and c-aminocaproic acid can be attributed to the relative degree to which different pathways gain access (via climbing fibers or other relays) to Purkinje cell dendrites. Axodendritic depolarizing and hyperpolarizing postsynaptic potentials are prominent components of responses evoked by cerebral (or peripheral nerve) stimulation as revealed by the effects of selectively acting


151

o-amino acids. But the failure of e-aminocaproic acid to affect “directly” evoked surface negativity indicates that inhibitory synapses on Purkinje cell dendrites are not activated by elements in the molecular layer directly excited by surface stimuli. Pharmacological data strongly suggest that the dendrites of Purkinje cells possess functionally different synaptic “zones” which are activated by differently organized relay elements, a8 in the case of the pyramidal cells of cerebral cortex. The latter presumably differ from Purkinje elements, however, by the presence of hyperpolarizing synapses on their dendritic terminals.

D. ON LOCALIZATION IN THE CEREBELLAR CORTEX The overwhelming complexity of the interactions among afferent responses in cerebellar cortex has encouraged the tendency t o define functional localization in the cerebellar cortex in terms of efferent corticonuclear projeetions. Indeed, there is very good evidence that longitudinal corticonuclear zones are organized in principle on a point-topoint basis (cf. Jansen and Brodal, 1954; Chambers and Sprague, 1955a, b). If such a precise relationship exists between cerebellar c o r t q and the various roof nuclei, it is not unreasonable to inquire how “diffuse” inputs are translated into relatively selective outputs. Clearly the integration and subsequent output selectivity is a function of the intrinsic synaptic organizations which are interposed between aff erents (mossy and climbing fibers) and Purkinje cells. On the basis of the data presented above, it is inferred that “localization” is reflected in terms of the composition of the interposed organizations activated by various afferents. Despite the apparent sharing of some cortical elements by differeiit afferent pathways, the evoked potentials may be differentially affected by o-amino acids. This can only be ascribed to differences in the organization of interposed elements which terminate synaptically on the same or different Purkinje cells. Whereas in the case of the cerebral cortex, functional localization is based on diversity of synaptic organizations within the framework of different cytoarchitectural patterns in which some modalites appear to be discretely represented in columnar fashion (Mountcastle, 1957), functional localization in the cerebellar cortex seems to depend exclusively on diversity in synaptic organiaations in view of the homogeneity of its architecture. Summurg. Differences in the response of cerebellar evoked potentials to topically applied o-amino acids are reviewed from the standpoint of the degree to which excitatory and inhibitory synaptic organizations are activated by different afferent pathways. Hyperpolarizing p.s.p.’s as well as depolarizing p.s.p.’s of Purkinje cell dendrites are inferred to be prominent in responses to peripheral nerve or cerebral cortical

152

DOMINICK P. PURPURA

stimulation, but the relative proportion of both varieties of synaptic electrogenesis is dependent on the source of the afferent projection, as well as the site of registration of the evoked potentials. The problem of functional localization in the cerebellar cortex is briefly reviewed on the basis of these observations. IV. General Considerations, Conclusions, and Summary “We leave it an open question whether any theoretical interpretation of cortical phenomena is better than none, relying for justification in this endeavor on numerous and respectable precedents.” Gi. H. BISHOP,1936

A. RECAPITULATION The necessity to endow cortical neurons with special properties in order to account for the diverse forms of electrical activity recorded from the surface of the brain is viewed as having arisen, in the past, from an incomplete knowledge of the electrical responses of excitable cells. It is now possible to account satisfactorily for this activity in terms of two major varieties of electrogenesis, conductile and synaptic, the latter arising in membrane which is chemically, but not electrically excitable. The degree to which conductile and synaptic responses are .detectable in cortical surface recordings is dependent on the source and size of an afferent volley as well as the manner in which afferent excitation of cortical neurons is effected. Synchronous activation of cortical elements by primary projection and interhemispheric pathways or by antidromic stimulation of the pyramidal tract elicits discharges of axonspike dimensions which are recordable from the cortical surface. Stimulation of other pathways (unspecific thalamocortical, caudate-cortical, etc.) ordinarily does not evoke sufficiently intense discharges in cortical neurons to permit their detection in surface electrograms. Because of temporal dispersion and algebraic summation, conductile activity is not reflected in the spontaneous cortical rhythms. Brain waves are adequately accounted for as summations of postsynaptic potentials of two varieties, depolarizing and hyperpolarizing, generated in apical dendrites and deeper lying elements via complexly organized excitatory and inhibitory interneurons. Evoked cortical potentials represent the integrated synaptic activities of differently organized elements. “Source-sink” dipoles generated in cortex by stimulation of corticipetal afferents are attributable to the development of p.s.p.’s a t various loci. Since the latter are essentially “standing waves” the apparent migration of dipoles in cortex is due to


153

successive synaptic activation of elements in different loci by conductile pathways. The existence of hyperpolarizing, as well as depolarizing, p.s.p.’s in evoked potentials can, however, render interpretation of the distribution of cortical dipoles extremely hazardous. Analyses of different varieties of data indicate that the major determinants of electrocortical activity are the following: (1) the specific organization of neuronal connections; (2) the temporal and spatial distribution of their synaptic actions ; (3) the proportions of depolarizing and hyperpolarizing p.s.p.’s; and (4) their magnitudes. The use of selectively acting pharmacological agents permits, in certain instances, defining the proportion and relative magnitudes of axodendritic depolarizing and hyperpolarizing p.s.p.’s evoked by different afferent pathways. When evoked potentials are compounded of axosomatic excitatory and inhibitory p.s.p.’s, the latter are indicated, in part, by their insensitiyity to o-amino acids. By relating evoked potentials to specific kinds of neuronal discharge (viz. corticospinal neurons or elements responding selectively to various corticipetal afferents) further clues are provided concerning the composition of certain neuronal organizations, as well as the sites of origin of evoked synaptic potentials. Surface potentials with similar characteristics can represent the integrated synaptic activity of entirely different neuronal organizations and the same neuronal population activated in different ways can give rise to evoked potentials with entirely different characteristics.

B. ON THE PROPERTIES OF APICALDENDRITES Data are reviewed in considerable detail relating to the properties of apical dendrites, their role in the production of slow waves recorded from the cortical surface, and the functional differences between axodendritic and axosomatic synapses. There is general agreement that dendrites are incapable of supporting all-or-none activity, but whether or not apical dendrites can generate graded responses other than p. s. p.’s is a problem which requires further elucidation. The hypothesis that the apical denprites are electrically inexcitable is inferred from pharmacological as well as electrophysiological data. Additional data derived from other neurons in the mammalian central nervous system suggest that most of the cell body, in addition to the dendrites of pyramidal cells may be electrically inexcitable. One consequence of electrical inexcitability of apical dendrites has an important bearing on the relationship between conductile activity and slow waves in the cerebral and cerebellar cortex. All-or-none discharges are evoked in neurons under ordinary circumstances by p.s.p.’s affecting the spike-generating components of the membrane (presumably

154

DOMINICK

P.

PURPURA

in the vicinity of the initial axonal segment). Because of electrotonic losses, p.s.p.’s generated in apical dendrites cannot be expected to be as effective in initiating or inhibiting spike discharges as p.s.p.’s developing in the soma. Indeed, it is unlikely that axodendritic activation of pyramidal cells can effect the discharge of these elements, but dendritic activation may affect or modulate excitability. When different sites on neurons are activated by different pathways, neuronal discharges may or may not be accompanied by significant alterations in surface potentials. These factors are taken into account in reviewing the relationship of p.s.p.’s to conductile activity in the cerebral cortex. The general hypothesis that the dendritic membrane of cortical neurons is essentially postsynaptic in type and therefore only sensitive to chemical activation is shown to have important implications from the standpoint of the application of neuropharmacology and neurochemistry to cortical electrophysiology. The demonstration that cortical axodendritic postsynaptic potentials can be markedly affected by certain “metabolites” such as the “-amino acids (or guanidino acids) indicates that characterization of certain neuronal organizations on the basis of their specific responses to pharmacological agents is likely to provide more information concerning the composition of different synaptic pathways than can be obtained with electrophysiological techniques alone. This is amply illustrated in the section dealing with the use of w-amino acids to facilitate analysis of different synaptic organizations in cerebellar cortex. Recently, the comparative pharmacology of hippocampal synaptic organizations has also been studied with the w-amino acid compounds (Purpura and Grundfest, 1959).

C. FURTHER IMPLICATIONS OF THE THEORY The hypothesis that “brain waves” represent largely the summated p.s.p.’s of superficially located elements in cortex carries with it the important corollary that most of the synaptic and conductile activity of different cortical organizations is not reflected in surface electrograms. The empirical relationship of various brain wave patterns ,to specific clinical syndromes is too well-known to require comment. The astute electroencephalographer niay often be able to extract useful information from the complex electrical tracings observed from the surface of the skull or cortex which points to one or another pathophysiological process. It may even be possible to relate certain varieties of activity to alterations in metabolic states and, indeed, there are some indications that the development of learning or memory traces is associated in some way with changes (specific?) in surface or depth electrograms. But mindful of the fact that such electrograms reflect the integrated synaptic activ-


155

ities of whole populations of coinplexly organized excitatory and inhibitory elements, it cannot be argued that the sudden appearance or disappearance of one variety of activity in a certain structure represents a “mechanism” involved in the establishment (or disestablishment) of complex behavioral activities. Furthermore, spontaneous or evoked potentials in different central structures are generated in postsynaptic elements as a consequence of their chemical (transmitter) activation. The distant effects which one neuronal organization exerts on another via conductile pathways can be reproduced by chemical activation of that organization without involvement of its usual afferent projections. Similarly, the accumulation within a given organization of certain metabolic agents may produce synaptic effects on that organization (recordable as local electrographic alterations) without involvement of afferent or efferent conductile pathways. Only when the synaptic relations between different elements in a particular neuronal organization are known, as well as the proportion of depolarizing and hyperpolarizing p.s.p.’s evoked by a pathway afferent to the organization will i t be possible to deduce the nature of the relationship between one organization and another. Since these are today largely unknown, there is little necessity to engage in any discussion of the possible mechanisms involved in the production of different electrographic patterns (via. “spindle bursts,” alpha-, beta-, and delta-rhythms, “K-complexes,” %pikes-and-waves,)’ etc.). Suffice it to say) that from the data reviewed above concerning the fundamental differences between specific, augmenting, recruiting, caudate-cortical, interhemispheric, and reticulocortical evoked responses, it is likely that different electrographic patterns may also be amenable to analysis in terms of differences in subcortical-cortical synaptic relations. ACKNOWLEDGMENTS The work reported by the author represents for the most part the results of a series of investigations carried out in collaboration with Dr. Harry Grundfest and our associates, Dr. Martin Girado and Dr. Edgar Housepian. The author’s work has been supported, in part, by grants from the Donner Foundation, Paul Moore Neurological Research Fund, and National Institute of Neurological Diseases and Blindness (B-1312 C). Dr. Grundfest’s work has been supported by the Muscular Dystrophy Associations, National Institutes of Health ( 3 8 9 4 ; C J , National Science Foundation (NSF-G-2030), and United Cerebral Palsy Research Foundation.

REFERENCES Adrian, E. D. (1936). J . Physiol 88, 127. Adrian, E. D. (1941). J . Physiql 100, 159. Adrian, E. D., and Buytendijk;F. J. J. (1931). J . Physiol. 71, 121.

186

DOMINICK P. PURPURA

Adrian, E. D., and Matthews, B. H. C. (1934). J. Physiol. 81, 440. Adrian, E.D.,and Moruzzi, G. (1939). J. Physiol. 97, 153. Albe-Fessard, D. (1957). J. physiol. (Paris) 49, 521. Albe-Feseard, D.,and Buser, P. (1955). J. physiol. (Paris) 47, 07. Albe-Feesard, D.,and Szabo, T. (1954). J. physiol. (Paris) 46, 225. Altamirano, M., and Grundfest, H. (1954). Tram. A m . Neurol. Assoc. 79, 35. Amassian, V. E. (1953). EEG Clin. Neurophysiol. 5, 416. Amassian, V. E., Patton, H. D., Woodbury, J. W., Towe, A., and Schlag, J. E. (1955). EEG C l h . Neurophysiol. 7, 480. Araki, T., and Otani, T. (1956). J. Neurophysiol. 18, 472. Arduini, A. (1958). In “Reticular Formation of Brain” (H. H. Jasper et al., eds.), p. 333-351. Little, Brown, Boston, Massachusetts. Arduini, A., and Terzuolo, C. (1951). EEG Clin. Neurophysiol. 3, 190. Arduini, A., and Whitlock, D. G. (1953). J. Neurophysiol. 16, 430. Arduini, A., Mancia, M., and Mechelse, K. (1957).Arch. ital. Biol. 95, 127. Armstrong, J., and Young, J. 2. (1957). J . Physiol. 137, 1OP. Awapara, J., Landua, A. J., Fuerst, R., and Seale, B. (1950). J. Biol. Chem. 187, 35. Barron, D.H., and Matthews, B. H. C. (1938). J. Physiol. 92, 270. Bartley, 8. H., and Bishop, G. H. (1933a). Am. J. Physiol. 103, 169. Bartley, 9. H., and Bishop, G. H. (1933b).A m . J . Physiol. 103, 173. Baumgartner, G. (1958).EEG Clin. Neurophysiol. 10, 195. Bazemore, A., Elliott, K. A. C., and Florey, E. (1950). Nature 178, 1052. Bennett, M.V.L., Crain, S. M., and Grundfest, H. (1959).J . Gen. Physiol. (In press). Berger, H. (1929). Arch. Psychiat. 87, 527. Beritov, I. (1956). Abstr. XX Intern. Physiol. Congr., Brussels p. 88. Berkley, H. J. (1897).Johns Hopkim Hosp. Repts. 6, 1. Bishop, G. H. (1933).Am. J. Physiol. 103, 213. Bishop, G. H. (1930). Cold Spring Harbor Symposia Quant. Biol. 4, 305. Bishop, G. H. (1949). EEG Clin. Neurophysiol. 1, 421. Bishop, G. H. (1956). Physiol. Revs. 36, 370. Bishop, G. H. (1968). EEG Clin. Neurophysiol. Suppl. 10, 12. Bishop, G. H., and Clare, M. H. (1952a).EEG Clin. Neurophysiol. 4, 321. Bishop, G. H., and Clare, M. H. (195213).J . Neurophysiol. 15, 201. Bishop, G. H., and Clare, M. H. (1953).J. Neurophvsiol. 16, 1. Bishop, G. H., and Erlanger, J. (1920).A m . J. Physiol. 78, 030. Bishop, G. H.,and O’Leary, J. L. (1930). A m . J. Physiol. 117, 292. Bishop, G. H., and O’Leary, J. L. (1938).J. Neurophysiol, 1, 391. Bishop, G. H.,and O’Leary, J. L. (1942). J . Cellular Comp. Physiol. 19, 315. Boistel, J., and Fatt, P. (1959). J. Physiol. (London) 144, 170. Bok, S. T.(1950). Progr. in Neurobwl. I . Neurochem. p. 13. Bonnet, V.,and Bremer, F. (1938). Compt. rend. soc. biol. 127, 806. Branch, C. L., and Martin, A. R. (1958). J. Neurophysiol. 21, 380. Brazier, M. A. B. (1958a). J. Nervous Mental Disease 126, 303. Brazier, M. A. B., ed. (1959). “First Conference on Central Nervous System and Behavior.’, Josiah Macy, Jr. Foundation, New York. Bremer, F. (1938). “L’activitB Blectrique de 1’Bcorce cBrBbrale.” Hermann and Cie, Paris. Bremer, F. (1943). Arch. inter. physiol. et biochem. 53, 53. Bremer, F. (1949). EEG Clin. Neurophysiol. 1, 177. Bremer, F. (1953). “Some Problems in Neurophysiology.” Athlone, University of London.


167

Bremer, F. (1955). Proc. SOC.Exptl. Biol. M e d . 90, 22. Bremer, F. (1958).Physiol. Revs. 38, 357. Bremer, F. and Bonnet, V. (1950). E E G Clin. Neurophysiol. 2, 389. Bremer, F.,and Bonnet, V. (1951). J. physiol. (Paris) 43, 665. Bremer, F., and Bonnet, V. (1953).Folia Psychiat. Neurol. 56, 438. Bremer, F.,and Stoupel, N. (1956). Arch. internat. physiol. et biochem. 64, 234. Bremer, F.,and Stoupel, N. (1957).Arch. ital. Biol. 95, 3. Bremer, F.,and Terzuolo, C. (1953). Arch. internat. physiol. et biochem. 61, 86. Bremer, F., Brihaye, J., and Andre-Balisaux, G. (1956). Schweiz. Arch. Neurol. Psychiat. 78,31. Brinley, F. J., Jr., Kandel, E. R., and Marshall, W. H. (1958). Trans. A m . Neurol. Assoc. 83, 53. Brock, L. G., Coombs, J. S., and Eccles, J. C. (1952). J. Physiol. (London) 117, 431. Brookhart, J. M., and Zanchetti, A. (1955).Federation Proc. 14, 20. Brookhart, J. M., and Zanchetti, A. (1956).E E G Clin. Neurophysiol. 8, 427. Brookhart, J. M., Moruzzi, G., and Snider, R. S. (1951).J. Neurophysiol. 14, 181. Brookhart, J. M., Arduini, A., Mancia, M., and Moruzzi, G. (1957). Arch. ital. Biol. 95, 139. Brookhart, J. M., Arduini, A., Mancia, M. and Moruzzi, G. (1958).J. Neurophysiol. 21, 499. Brooks, C. M., and Eccles, J. C. (1947a).J. Neurophysiol. 10,251. Brooks, C. M.,and Eccles, J. C. (194713). J. Neurophysiol. 10, 349. Brooks, C. M.,and Koiaumi, K. (1953). I n “Ciba Foundations Symposium on the Spinal Cord,” pp. 63-74.Little, Brown, Boston. Brooks, V. B., and Enger, P. S. (1959).J. Gen. Physiol. 42, 761. Brooks, V. B., and Wilson, V. J. (1958). Science 127, 472. Bullock, T.H. (1945). Yale J . Biol. and M e d . 17, 657. Bullock, T.H., and Terzuolo, C. A. (1957).J. Physiol. (London) 138, 341. Burei, J. (1956).J . Neurochem. 1, 153. Burns, B. D. (1950).J. Physiol. (London) 111, 50. Burns, B. D. (1951). J. Physiol. (London) 112, 156. Burns, B. D. (1955). E E G Clin. Neurophysiol. 7, 483. Burns, B. D. (1957). “Mammalian Cerebral Cortex.” Edward Arnold, London. Burns, B. D., Grafstein, B., and Olszewski, J. (1957). J. Neurophysiol. 20, 200. Buser, P. (1957).J. physiol. (Paris) 49, 589. Cajal, 8. Ramon y (1934).Trabajos Lab. Invest. Biol. Univ. Madrid 26, 1. Calma, I., and Arduini, A. (1954). J. Neurophysiol. 17, 32. Carrea, R. M.E., and Grundfest, H. (1954).J. Neurophysiol. 17,208. Caton, R. (1875). Brit. M e d . J. 11, 278. Chambers, W.W.,and Liu, C.-N. (1957).J . Cellular Comp. Neurol. 108, 23. Chambers, W.W.,and Sprague, J. M. (1955a).J. Comp. Neurol. 103, 105. Chambers, W.W.,and Sprague, J. M. (195513). A.M.A. Arch. Neurol. Psychiat. 74, 853.

Chang, H.-T. Chang, H.-T. Chang, H.-T. Chang, H.-T. Chang, H.-T. Chang, H.-T. Chang, H.-T.

(1950).J. Neurophysiol. 13, 235. (1951a).J. Neurophysiol. 14, 1. (1951b). J. Neurophysiol. 14, 95. (1952).Cold Spring Harbor Symposia Quant. Biol. 17, 189. (1953a).J . Neurophysiol. 16, 117. (195313).J. Neurophysiol. 16, 133. (1955a). J . Neurophysiol. 18, 332.

158

DOMINICK P. PURPURA

Chang, H.-T. (1955b). J. Neurophysiol. 18, 452. Chang, H.-T. (1956). In “Problems of the Modern Physiology of the Nervous and Muscle Systems,” pp. 43-50. Acad. of Sciences, Georgian SSR, Tbilisi. Chang, H.-T., and Kaada, B. (1950). J. Neurophysiol. 13, 305. Cheymol, J., Bovrillet, F., and Levassort, C. (1955). J . physiol. (Paris) 47, 132. Clare, M. H., and Bishop, G. H. (1952). EEG Clin. Neurophysiol. 4, 311. Clare, M. H., and Bishop, G. H. (1965). EEG Clin. Neurophysiol. 7, 85. Clare, M. H., and Bishop, G. H . (1956). EEG Clin. Neurophysiol. 8, 583. Clare, M. H., and Bishop, G. H. (1957). J. Neurophysiol. 20, 255. Coombs, J. S., Eccles, J. C., and Fatt, P. (1955). J. Physiol. (London) 130, 291. Coombs, J. S., Curtis, D. R., and Eccles, J. C. (1957a). J. Physiol. (London) 139, 198. Coombs, J. S., Curtis, D. R., and Eccles, J. C. (195713). J. Physiol. (London) 139, 232. Cragg, B. G., and Hamlyn, L. H. (1955). J. Physiol. (London) 129, 608. Cranefield, P. F., and Hoffman, B. F. (1958). Physiol. Revs. 38, 41. Creutzfeldt, O., Baumgartner, G., and Schoen, L. (1956). Arch. Psychiat. Nervenkrankh. 194, 597. Curtis, D. R., and Eccles, R. M. (1958a). J. Physiol. (London) 141, 435. Curtis, D . R., and Eccles, R. M. (1958b). J . Physiol. (London) 141, 446. Curtis, H. J. (1940a). J. Neurophysiol. 3, 405. Curtis, H. J. (1940b). J. Neurophysiol. 3, 414. Exptl. Biol. Med. 44, 664. Curtis, H. J. (1940~).Proc. SOC. Danilewski, B. (1891). Physiologischeskom. Sbornike. 2, 77. Davis, H. (1936). Cold Spring Harbor Symposia Quant. Biol. 4, 285. Dell, P. C. (1958). In “Reticular Formation of the Brain” (H. H. Jasper, et al. eds.), pp. 365-379. Little, Brown, Boston. Dempsey, E. W., and Morkon, R. S. (1942a). Am. J. Physiol. 135, 293. Dempsey, E. W., and Morison, R. S. (1942b). Am. J. Physiol. 135, 301. Dempsey, E. W., and Morison, R. S. (194%). A m . J. Physiol. 138, 283. Desmedt, J. E., and LaGrutta, G.(1957). J . Physiol. (London) 136, 20. Dondey, M., and Machne, X. (1955). Rev. neurol. 93, 486. Dow, R. S. (1942). J. Neurophysiol. 5,121. Dow, R. S. (1949). J. NeurophysioL 12,245. DOW, R. S., and Moruzzi, G. (1958). “The Physiology and Pathology of the Cerebellum.” Univ. Minnesota Press, Minnesota. Dubner, H. H., and Gerard, R. W. (1939). J. Neurophysiol. 2, 142. Eccles, J. C. (1939). Ann. R e v . Physiol. 1, 363. Eccles, J. C. (1943). J. Physiol. (London) 101, 485. Eccles, J. C. (1946). J. Neurophysiol. 9, 87. Eccles, J. C. (1951). EEG Clin. Neurophysiol. 3, 449. Eccles, J. C. (1957). “The Physiology of Nerve Cells.” Johns Hopkins Press, Baltimore, Maryland. Eccles, J. C., and Malcolm, J. L. (1946). J. Neurophysiol. 9, 140. Eccles, J. C., and O’Connor, W. J. (1939). J. Phydol. (London) 97, 44. Eccles, J. C., Fatt, P., and Koketsu, K. (1964). J. Physiol. (London) 126, 524. Eccles, R. M. (1955). J. Physiol. (London) 130, 572. Eccles, R. M. (1956). J. Pharmacol. Exptl. Therap. 118,,26. Edwards, C., and Kuffler, S. W. (1957). Federation Proc. 16, 34. Erulkar, S. D., Rose, J. E., and Davies, P. W. (1956) Johns Hopkins Hosp. Bull. 99, 55.


159

Evarts, E. V., and Magoun, H. W. (1957). Science 125, 1147. Eyzaguirre, C., and KuWer, S. W. (1955a). J . Gen. Physiol. 39, 87. Eyzaguirre, C., and Kuffler, S. W. (1955b). J . Gen. Physiol. 39, 121. Fan, S. F., and Feng, T. P. (1957). Acta Physiol. Sinica 21, 423 (in Chinese). Fatt, P. (1954). Physiol. Revs. 34, 674. Fatt, P. (1957a). J . Neurophysiol. 20, 27. Fatt, P. (1957b). J . Neurophysiol. 20, 67. Fatt, P., and Katz, B. (1951). J . Physiol. (London) 115, 320. Fatt, P., and Katz, B. (1953). J . Physiol. (London) 121, 374. Feindel, W., and Gloor, P. (1954). EEG Clin. Neurophysiol. 6, 389. Feng, T. P., and Fan, S. F. (1957). Sci. Sinica (Peking) 6, 159. Fessard, A. (1956). Abstr. X X Intern. Physiol. Congr., Brussels p. 35. Fssard, A., and Matthews, B. H. C. (1939). J . Physiol. (London) 95, 39P. Fischgold, H., and Gastaut, H., ed. (1957). EEG Clin. Neurophysiol. Suppl. 6. Florey, E., and McLennan, H. (1955). J . Physiol. (London) 129, 384. Forbes, A., and Morison, B. R. (1939). J . Neurophysiol. 2, 112. Frank, K., and Fuortes, M. G. F. (1955). J . Physiol. (London) 130, 625. Frank, K., and Fuortes, M. G. F. (1956). J . Physiol. (London) 134, 451. Freygang, W. H., Jr. (1958). J . Gen. Physiol. 41, 543. Freygang, W. H., Jr., and Frank, K. (1959). J . Gen. Physiol. 42, 749. Fuortes, M. G. F., Frank, K., and Becker, M. C. (1957). J . Gen. Physiol. 40, 735. Gasser, H. S. (1933). Cold Spring Harbor Symposia Quant. Biol. 1, 138. Gasser, H. S. (1939). J . Neurophysiol. 2, 361. Gasser, H. S., and Graham, H. T. (1933). Am. J . Physiol. 103, 303. Gauthier, C., Parma, M., and Zanchetti, A. (1956). EEG Clin. Neurophysiol. 8, 237. Gerard, R. W. (1936). Cold Spring Harbor Symposia Quant. Biol. 4, 292. Gerard, R. W., and Libet, B. (1940). A m . J . Psychiat. 96, 1125. Gesell, R. (1926). Proc. SOC.Exptl. Biol. Med. 24, 263. Gesell, R. (1940). Ergebn. Physiol. 43, 477. Goldring, S., and O’Leary. J. L. (1957). EEG Clin. Neurophysiol. 9, 577. Goldring, S., O’Leary, J. L., and King, R. B. (1958). EEG Clin. Neurophysiol. 10, 233.

Gopfert, H., and Schaefer, H. (1938). Arch. ges. Physiol., Pfliiger‘s 239, 597. Gotch, F., and Horsley, V. (1891). Phil. Trans. Roy. SOC.London B182, 267. Granit, R., and Phillips, C. G. (1956). J . Physiol. (London) 133, 520. Gray, J. A. B., and Sato, M. (1953). J . Physiol. (London) 122, 610. Grundfest, H. (1957a). Progr. in Biophys. and Biophys. Chem. 7, 1. Grundfest, H. (1957b). Physiol. Revs. 37, 337. Grundfest, H. (1957~).Ann. N. Y . Acad. Sci. 66, 537. Grundfest, H. (1958a). I.. “Reticular Formation of Brain” (H. H. Jasper et al., ed.), pp. 47-87. Little, Brown, Boston. Grundfest, H. (195813). EEG Clin. Neurophysiol. Suppl. 10, 22. Grundfest, H., and Campbell, B. (1942). J . Neurophysiol. 5, 275. Grundfest, H., Reuben, J. P., and Rickles, W. H., Jr. (1959). J . Gen. Physiol. 42, 1301.

Hagiwara, S. (1958). J . Gen. Physiol. 41, 1119. Hagiwara, S., and Bullock, T. H. (1957). J . Cellular Comp. Physiol. 50, 25. Hagiwara, S., and Tasaki, I. (1958). J . Physiol. (London) 143, 114. Hampson, J. L. (1949). J . Neurophysiol. 12, 37. Hanberry, J., Ajmone-Marsan, C., and Dilworth, M. (1954). EEG Clin. Neurophysiol. 6, 103.

160

DOMINICK P. PURPURA

Hartline, H. K., Wagner, H. G., and MacNichol, E. F., Jr. (1952). Cold Spring Harbor Symposia Quant. Biol. 17, 125. Hayashi, T., and Nagai, K. (1956). Abstr. X X Intern. Physiol. Congr., Brussels p. 410.

Hodgkin, A. L. (1938). Proc. Roy. SOC.8126, 87. Hodgkin, A. L. (1951). Bwl. Revs. Cambridge Phil. SOC.26, 339. Hodgkin, A. L. (1958). Proc. Roy. SOC.8148, 1. Hodgkin, A. L., and Huxley, A. F. (1952). J. Physiol. (London) 117, 500. Howland, B., Lettvin, J. Y., McCulloch, W. S., Pitts, W. and Wall, P. D. (1955). J . Neurophysiol. 18, 1. Hubel, D. H. (1957). Federation Proc. 16, 63. Hunt, C. C. (1955a). J. Gen. Physiol. 38,801. Hunt, C. C. (1955b). J . Gen. Physiol. 38, 813. Irreverra, 8.F., Evans, R. L., Hayden, A. R., and Silber, R. (1957). Nature 180, 704. Iwama, K., and Jmper, H. H. (1957). J. Physiol. 138, 365. Janaen, J., Jr. (1957). Acta Physiol. Scand. 41, Suppl. 143. Jamen, J., and Brodal, A. (1954). “Aspects of Cerebellar Anatomy.” Tanum, Oslo. Jasper, H. H. (1936). Cold Spring Harbor Symposia Quant. Biol. 4, 320. Jasper, H. H. (1937). Psychol. Bull. 34, 411. Jasper, H. H. (1949). EEG Clin. Neurophysiol. I , 416. Jasper, H. H. (1954). In “Brain Mechanisms and Comciousness” (J. F. Delafresnaye, ed.) Blackwell, Oxford. Jasper, H. H., and Ajmone-Marsan, C. (1952). Research Publs. Assoc. Research Nervous Mental Disease 30, 492. Jasper, H. H., and Droogleever -Fortuyn, J. (1947). Research Publs. Assoc. Research Nervous Mental Diseases 26, 272. Jaaper, H. H., NaquBt, R., and King, E. (1955). EEG Clin. Neurophysiol. 7, 99. Jasper, H. H., Gonzales, S., and Elliott, K. A. C. (1958). Federation Proc. 17, 79. Jung, R. (1963). EEG. Clin. Neurophysiol., Suppl. 4,57. Jung, R. (1958). I n “Reticular Formation of Brain” (H. H. Jasper, et al., eds.), pp. 423-434. Little, Brown, Boston. Kao, C. Y.,and Hoffman, B. F. (1958). A m . J. Physiol. 194, 187. Kolmodin, G. M., and Skoglund, 0. R. (1959). Acta Physiol. Scand. 45, 1. Kuffler, 8. W., and Edwards, C. (1958). J. Neurophysiol. 21, 589. Landau, W. M. (1956). E E G Clin. Neurophyswl. 8, 445. Landau, W.M., and Clare, M. H. (1956). EEG Clin. Neurophysiol. 8, 467. Latimer, C. N., and Kennedy, T. T. (1958). Federation Proc. 20, 93. LeZo, A. A. P. (1944). 3. Neurophysiol. 7, 391. Li, C.-L. (1955a). J . Physiol. (London) 131,115. Li, C.-L. (1966b). J . Phyeiol. (London) 133, 40. Li, C.-L. (1958). I n “Reticular Formation of Brain” (H. H. Jasper, et al., eds.), pp. 459472. Little, Brown, Boston. Li, C.-L., and Jasper, H. H. (1963). J. Physiol. (London) 121, 117. Li, C.-L., Cullen, C., and Jasper, H. H. (1956a). J. Neurophysiol. 19, 111. Li, C.-L., Cullen, C., and Jasper, H. H. (1956b). J . Neurophysiol. 19, 131. Libet B., and Gerard, R. W. (1938). Proc. SOC.Exptl. Biol.Med. 38, 886. Libet, B., and Gerard, R. W. (1941). J . Neurophysiol. 4, 438. Lloyd, D. P. C. (1951). J . Gen. Phyaiol. 35, 289. Lloyd, D. P. C., and McIntyre, A. K. (1949). J. Gen. Physiol. 32,409. Lloyd, D. P. C., and McIntyre, A. K. (1955a). J . Gen. Physiol. 38,771. Lloyd, D. P. C., and McIntyre, A. K. (1955b). J . Gen. Physiol. 38,789.


161

Lorente de N6, R. (1934). J . Psychol. Neurol., Leipzig 46, 113. Lorente de N6, R. (1935). Am. J. Physiol. 113, 524. Lorente de N6, R. (1939). J . Neurophyaiol. 2, 402. Lorente de N6, R. (1947). J . Cellular Comp. Physiol. 29, 207. Loewenstein, W. R. (1958). J . Gen. Physiol. 41, 825. Loewenstein, W. R. (1959). Ann. N. Y .Acad. Sci. (In press.) Loewenstein, W. R., and Rathkamp, R. (1958). J . Oen. Physiol. 41, 1245. Lucas, K. (1917). “The Conduction of the Nerve Impulse.” Longmans, London. McLennan, H. (1958). Nature 181, 1807. Magoun, H. W. (1944). Science 100,549. Malcolm, J. L., and Smith, I. D. (1959). J . Physiol. (London) 144, 257. Malis, L. L., and Kruger, L. (1956). J . Neurophysiol. 19, 172. Marrassi, A. S., Hart, E. R., and Rodrigues, J. M. (1958). Science 127, 284. Marshall, W. H. (1941). J . Neurophyaiol. 4, 25. Martin, A. R., and Branch, C. L. (1958). J . Neurophysiol. 21, 368. Monnier, A. M., and Jasper, H. H. (1932). Compt. rend. aoc. biol. 110, 549. Morison, R. S.,and Dempsey, E.W. (1942). A m . J . Physiol. 135, 281. Morison, R. S., Finley, K. H., and Lothrop, G. N. (1943). A m . J . P h y b l . 139, 410. Moruzzi, G. (1953). EEG Clin. Neurophysiol. Isuppl. 4, 221. Morussi, G., and Magoun, H. W. (1949). EEO Clin. Neurophysiol. 1, 455. Morussi, G., Brookhart, J. M., Niemer, W. T., and Magoun, H. W. (1950). EEQ Clin. Neurophyaiol. 2, 29. Mountcastle, V. B. (1957). J . Neurophyaiol. 20, 408. Mountcastle, V. B., Davies, P. W., and Berman, A. L. (1957). J . Neurophyaiol. 20, 374.

Murphy, J. P., and Gellhorn, E. (1945). J . Neurophysiol. 8, 341. Nauta, W. J. H. (1954). Anat. Record 118, 333. Ochs, 8. (1956). J . Neurophysiol. 19, 513. Ochs, 8.(1958). J . Neurophysiol. 21, 159. O’Leary, J. L., and Bishop, G. H. (1943). J . Cellular Comp. Physiol. 22, 73. O’Leary, J. L., and Coben, L. A. (1958). Physiol. Revs. 38, 243. O’Leary, J. L., and Goldring, 8. (1953). EEG Clin. Neurophysiol., Suppl. 4, 52. Palay 8. L. (1956). J . Biophys. Biochem. C y t d . 2 (Suppl.), 193. Paton, W. D., and Perry, W. L. M. (1953). J . Physiol. 119, 43. Patton, H., and Ammian, V. E. (1954). J . Neurophyaiol. 17, 345. Peacock, S. M. (1957). J . Neurophysiol. 20, 140. Perl, E. R., and Whitlock, D. G. (1955). J . Neurophysiol. 18, 486. Phillips, C. G. (1956). Quart. J . Exptl. Phyaiol. 41,58. Phillips, C . G. (1959). Quart. J . Esptl. Physiol. 44, 1. Pisano, J. J., and Udenfriend, S. (1958). Federation Proc. 17, 403. Prawdicz-Neminski, W. W. (1925). Arch. gee. Phyawl. Pfiuger‘s 209,363. Pupilli, C. G., Fadiga, E., and von Berger, G. P. (1956). Arch ad. biol. (Bologna) 40,541.

Purpura, D. P. (1855). J . Neurophysiol. 18, 246. Purpura, D. P. (1956). Science 123,804. Purpura, D. P. (1957a). In “Third Conference on Neuropharmacology” (H. Abramson, ed.), pp. 297422. Josiah Macy, Jr. Foundation, New York. Purpura, D. P. (1957b). Ann. N . Y.Acad. 8ci.66,515. Purpura, D. P. (1958). In “Reticular Formation of Brain” (H. H. Jasper, et a l , eds.), pp. 436-487. Little, Brown, Boston. Purpura, D. P. (1969a). EEQ Clin. Neurophyhl. 11,400.

162

DOMINICK P. PURPURA

Purpura, D. P. (1959b). In “Second Conference on Central Nervous System and Behavior” (M. A. B. Brazier, ed.). Josiah Macy, Jr. Foundation, New York (in press). Purpura, D. P., and Girado, M. (1959). Arch. ital. Biol. (In press.) Purpura, D. P., and Grundfest, H. (1956). J. Neurophysiol. 19, 573. Purpura, D. P., and Grundfest, H. (1957a). EEG Clin. Neurophysiot. 9, 162. Purpura, D. P., and Grundfest, H. (1957b). J. Neurophysiol. 20, 494. Purpura, D. P., and Grundfest, H. (1959). Federation Proc. 18, 123. Purpura, D. P., Girado, M., and Grundfest, H. (1957a). Science 125, 1200. Purpura, D. P., Girado, M., and Grundfest, H. (1957b). Proc. SOC.Exptl. Biol. Med. 95, 791. Purpura, D. P., Housepian, E. M., and Grundfest, H. (1957~).Federation Proc. 16, 102. Purpura, D. P., Housepian, E. M., and Grundfest, H. (1957d). EEG Clin. Neurophysiol. 9, 179. Purpura, D. P., Pool, J. L., Ransohoff, J., Frumin, M. J., and Housepian, E. M. (1957e).EEG Clin. Neurophysiol. 9,453. Federation Proc. 17, 126. Purpura, D. P., Girado, M., and Grundfest, H. (1958~). Purpura, D. P., Girado, M., and Grundfest, H. (1958b). Science 127, 1179. Purpura, D. P., Girado, M., and Grundfest H. (1958~).Proc. A m . Acad. Neurol (In preparation.) Purpura, D. P., Girado, M., Smith, T. G., Jr., and Gomez, J. A. (1958d). Proc. Soc. Exptl. Biol. Med. 97, 348. Purpura, D. P., Girado, M., Smith, T. G., Jr., and Gomez, J. A. (1958e). EEG Clin. Neurophysiol. 10, 677. Purpura, D. P., Housepian, E. M., and Grundfest, H. (1958f). Arch. ital. Biol. 96, 145. Purpura, D. P., Girado, M., and Grundfest, H. (1959a). J. Gen. Physiol. 42, 1037. Purpura, D. P., Girado, M., Smith, T. G., Jr., Callan, D. A., and Grundfest, H. (1959bb).J. Neurochem. 3, 238. Purpura, D. P., Girado, M., and Grundfest, H. (1959~).EEG Clin. Neurophysiol. 11 (in press). Rayport, M. (1957). Federation Proc. 16, 104. Renshaw, B. (1941). J. Neurophysiol, 4, 167. Renshaw, B. (1946). J. Neurophysiol. 9, 191. Renshaw, B., Forbes, A., and Morison, B. R. (1940). 3. Neurophysiol. 3, 74. Roberts, E. and Frankel, S. (1950). J. Biol. Chem. 187, 55. Robertson, J. D. (1957). Progr. in Neurobiol. I I . p. 1. Roitbak, A. I. (1955) “Bioelectric Phenomena of the Cerebral Cortex’’ (In Russian). Academy of Sciences, Georgian SSR, Tiflis, Russia. Rosenaweig, M. R. (1950). A m . J. Physiol. 163, 746. Rossi, G. F. and Zanchetti, A. (1957). Arch. ital. Biol. 95, 199. Sarkissov, S. (1956).Abstr. XX Intern. Physiol. Congr., p. 426. Scheibel, M. E., and Scheibel, A. B. (1954). J. Comp. Neurol. 101, 733. Schlag, J., and Brand, H. (1958) EEG Clin. Neurophysiol. 10, 305. Sechenov, I. (1863). “Physiologische Studien uber die Hemmungs-Mechanismen fur Reflextatigkeit des Ruckenmarks im Gehirne des Frosches.” Hirschwald, Berlin. Shanes, A. M. (1958a). Phormacol. Revo. 10, 59. Shanes, A. M. (1958b). P h a r d o l . Revs. 10, 165. Sherrington, C. 5. (1925). Proc. Roy. Soc. B97, 619. Shimamoto, T., and Verzesno, M. (1954). J . Neurophysiol. 17, 278.


163

Sholl, D. A. (1956). “The Organization of the Cerebral Cortex.” Wiley, New York. Sloan, N., and Jasper, H. H. (1950). EEG Clin. Neurophysiol. 2, 317. Smith, T. G., Jr., and Purpura, D. P.(1958).EEG Clin. Neurophysiol. 10,363. Smith, T. G., Jr., and Purpura, D. P. (1959). EEG Clin. Neurophyaiol. in press. Smythies, J. R., Gibson, W. C., and Purkis, V. A. (1957). J Comp. Neurol. 108, 173. Smythies, J. R. (1958). Personal communication. Stamm, J. S., and Sperry, R. W. (1957). J. Comp. Physiol. Psychol. 50, 138. Starzl, T.E.,and Magoun, H. W. (1951). J. Neurophysiol. 14, 133. Stoupel, N.,and Termolo, C. (1954). Acta Neurol. Psychiat. Belg. 54, 239. Szabo, T., and Albe-Fessard, D. (1954). J. physiol. (Paris) 46, 528. Tasaki, I.,and Chang, J. J. (1958).Science 128, 1209. Tasaki, I., Hagiwara, S., and Watanabe A. (1954a).Japan J. Physiol. 4, 79. Tssaki, I., Polley, E. H., and Orrego, F. (1954b).J. Neurophysiol. 17, 454. Terzuolo, C. (1958). Federation Proc. 17, 162. Tomita, T. (1957).J. NeurophysioZ. 20, 245. Towe, A. L.,and Amassian, V. E. (1958). J. Neurophysiol. 21, 292. Udenfriend, S.(1950).J . Biol. Chem. 187,65. Umrath, K. (1933).Arch. ges. Physiol., Pfliiger’s 233, 357. van Harreveld, A. (1957). Am. J. Physiol. 191, 233. van Harreveld, A. (1958).Am. J. Physiol. 192, 457. van Harreveld, A., and Ochs, S. (1956).A m . J. Physiol. 187, 180. Verzeano, M., and Calma I. (1954).J. Neurophyswl. 17, 417. Verzeano, M.,Lindsley, D. B., and Magoun, H. W. (1953). J . Neurophysiol. 16, 182. von Euler, C. and Ricci, G. F. (1958). J . Neurophysiol. 21, 231. von Euler, C.,Green, J. D., and Ricci, G. (1958). Acta Physiol. Scand. 42, 87. Wall, P.D.(1958).J. Physiol. (London) 142,l. Whitlock, D. G., Arduini, A., and Moruzzi, G. (1953). J. Neurophysiol. 16, 414. Woolsey, C.N.,and Chang, H,T. (1948).Research Publs. Assoc. Research Nervous Mental Disease 27, 146.


CHEMICAL AGENTS OF THE NERVOUS SYSTEM By Catherine 0. Hebb The A. R. C. lnrtitute of Animal Phyriology, Babraham, Cambridge, England

I. Introduction ... ......... ..... ................... 11. Chemical Morphology of Nerve Cells . . . A. Subcellular Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Chemical Organization of the Neuron 111. Intracellular Storage and Transport of ............ secretion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. Synthesis and Storage of Acetylcholine .............................. B. Noradrenaline .................. C. Neurosecretion .................. IV. Chemical Transmiss of Mammals . . . . A. Excitatory Transmitter Agents ......................................

165

176 176 180 181 184 184 . . . . . . . . . . 186

V. Conclusion References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

I. Introduction

Classically the development of the theoretical concepts of neurophysiology has depended little on evidence of the chemical nature of nervous tissue or of the mechanisms concerned in its metabolic activity. It is true, too, that compared with the advances made in the biochemical study of other tissues and organisms, notably liver, muscle and yeast, the biochemistry of nerve has lagged far behind. In recent years, however, there has been much more general interest in this subject and, as a consequence, much more work is being done on the chemistry of nervous tissue, in particular on the chemistry of mammalian brain, than was the case formerly. One factor in stimulating this interest was undoubtedly the advent of the theory of the chemical transmission of nerve impulses. The indications that some neurons are able t o elaborate and secrete special chemical agents naturally focussed attention on the metabolism of these substances, and extensive investigation of the eniymatic mechanisms responsible followed. At the outset workers were preoccupied with the identification and properties of the enzymes directly concerned in the me166

166

CATHERlNE 0. HEBB

tabolism of the transmitters but it will be recognized that a problem of equal importance is to determine how the activity of these systems fits into the metabolism of the cell as a whole; furthermore because of the particular morphology of the neuron with its long cell process, the axon, the mechanisms by which the active materials are stored and transported intracellularly also require elucidation. Thus, the problem of the chemical transmitter substances and the analogous problems of neurosecretion should now be considered within the wider context of the chemical organization of neurons as a whole. The chemical organization of cells in general has come to be an important subject of biochemical research and, with the great success already achieved in this direction, biochemistry appears to be entering on a new phase of its history. From the rapid progress already made it is becoming clear that the spatial arrangement of enzyme systems within living cells is of paramount importance in regulating their metabolism. Consequently, subcellular structure, the study of which has been facilitated by the ultracentrifuge and the invention of micrascopes of high resolving power, has come to have great significance in biochemical research. Although the subcellular organization of nervous tissue has not yet been studied in detail, the data accruing suggest that the application of the new techniques will be highly rewarding in this field of study. And, one may reasonably hope that their pursuit will eventually come t o provide a secure basis for the understanding of the events which are associated with the formation and liberation of the chemical agents of nerve impulse transmission. In presenting my ideas about this subject I have begun by giving an account of the ultrastructure and chemical organization of the neurons. This is to provide what I trust will be useful background for the topic dealt with in Section 111, namely, the intracellular storage and disposition of the known transmitter agents. Section I11 also deals with neurosecretion as a related phenomenon. Section IV takes up the question of central transmitter agents, and acetylcholine and other candidates for this role are discussed. Finally the question of chemical inhibition will be considered.

II. Chemical Morphology of Nerve Cells

A. SUBCELLULAR STRUCTURE The description of the nerve cell given by Ramon y Cajal (1909) on the basis of his own researches and those of many distinguished histologists who had preceded him (Remak, Ranvier, Bethe, Retzius, Schultze,

CHEMICAL AGENTS OF THE NERVOUS SYSTEM

167

and Nissl among many others) is still valid today. According to Cajal the neuron, which typically has two types of processes, the axon and dendrites, arising from the perikaryon or cell body, is enclosed in a membrane and contains within its cytoplasm in addition to the nucleus, the Nissl bodies, the Golgi apparatus, the neurofibrils, and the mitochondria. The only structural entities not in this list which have been detected by the modern techniques of microscopy and ultracentrifugation are the microsomes, a term here used to denote cellular inclusions, either granular or vesicular, the diameter of which is less than 0.211. They are, therefore, well below the limits of resolution of light microscopy. Between the time they were first described and the time when nerve tissue had been re-examined by methods capable of resolving its ultrastructure, doubt had been very frequently expressed about the reality of the organelles which had been detected by histological methods. With the exception of the nucleus, each of the organized structures Cajal described has at various periods been thought by some to be an artifact. Such views are no longer tenable. The wealth of corroborative detail provided by the new techniques, including phase contrast and interference microscopy, have established their existence beyond doubt. I n this the electron microscope has certainly played a most important part. It has not only provided independent confirmation of the existence of these intracellular structures, it has also described their morphology in such infinitely finer detail that it brings a new dimension into thinking 8,bout structure-function relations; and at the same time it has drawn attention to the high degree of organization which exists within the cell as a whole. Entities such as the mitochondria, other granules, and filaments are no longer to be regarded as islands in an amorphous sea of cytoplasm but as parts of a very complex structural system into which all of the cell constituents are welded. An illustration of this is the description which has been given by Palay and Palade (1955) of the Nissl bodies. These are now seen to be far more complex in structure than could have been imagined earlier. As described in classical histology they appear as aggregates or clusters of densely staining material interspersed by channels and lacunae of clear cytoplasm, an appearance which has been likened to a mosaic by Cajal, who also pointed out that their absence from the axon distinguishes this cell process from the dendrites. With the electron microscope the over-all pattern is similar but at the higher magnifications permitted it is seen that they are part of the endoplasmic reticulum which seems to correspond to the ergastoplasm or basophilic cytoplasm of other tissues (Porter and Kallmann, 1952). This extends through the cytoplasm

168 CATHERINE 0. HEBB with which it is continuous as a membranous network enclosing a series of canals, vesicles, and tubules, and characteristically associated with granules which are in rows running parallel and near to the membranes that enclose the inner spaces of the system. The Nissl bodies are threedimensional expansions of this network in which the reticulum is condensed and arranged in a characteristic way, and which are embedded in a matrix of cytoplasm. The small granular component of the system, which is responsible for the basophilic staining reactions of the Nissl bodies and is generally attributed to nucleic acids (pentose nucleic acids), forms part of the microsomal fraction of the tissue. In addition to the endoplasmic reticulum, Palay and Palade describe another system of membranes similar in architecture which they call the agranular reticulum because it is characterized by the absence of granules in the vicinity of the membranes. Since intermediate forms are present, the two systems may not be separate entities although, for descriptive purposes they are treated as such. It seems probable that part of the agranular reticulum corresponds to the Golgi apparatus of the earlier histologists (see Fernhndez-Morh, 1957). Sjostrand’s work (1956)also supports this conclusion but he has criticized the terminology and some of the conclusions of Palay and Palade. The neurofilaments or neurofibrils have been, of all the histological structures of the neuron, the most frequently attributed to artifacts. However their existence is also confirmed by the electron microscope; and the only wonder is that they should have been detected at all by the light microscope, since they are only 60-100A. in width (that is of the order of one hundredth of a micron). They are present in all parts of the cytoplasm forming what Palay and Palade have described as a “loose feltwork of fibrils” which in the perikaryon runs in many different directions but is probably axially orientated in the axon. The appearance and distribution of mitochondria in nervous tissue have been described in a number of studies in which the electron microscope was employed. They are very numerous in the cell body and are also present in the dendrites and axon; at the axon terminals they are especially numerous and are there associated with large numbers of microvesicles or microsomes (FernSndez-Morhn, 1957; see also de Robertis and Bennett, 1955). In the axon they may occupy the total diameter of the axoplasmic space (Gasser, 1955, 1956) and are oriented axially; in the perikaryon they are sometimes associated with the Nissl bodies in addition to being scattered in abundance through the rest of the cytoplasm. Fewer mitochondria are present in the region of the axon hillock than elsewhere in the perikaryon of many nerve cells, but they appear


169

to be numerous in the same area of the olfactory bipolar cells described by Gasser (1956). Although on an average rather smaller than those of other tissues, the mitochondria in nervous tissue have the characteristic structure of other mitochondria. The cristae mitochondriales may run either vertical to or parallel with the long axis; in some they may appear as somewhat swollen invaginations of the inner membrane. There are other organelles of the same order of size which do not have any well defined inner structure and are more like vesicles in appearance (see Palay and Palade, 1955; Andrew, 1956). The cytoplasmic vacuoles referred to by Hogeboom and Kuff (1955) may belong to or include structures of this kind. Since mitochondria are sensitive to osmotic changes, differences in shape may be evidence of their functional state rather than of true morphological variation. Under the general heading of microsomes it is possible to include any granule or vesicle with a diameter of less than 0 . 2 ~since the name was originally chosen simply to denote small bodies (Claude, 1949). There has been a tendency by some to equate only the basophilic particles of the cytoplasm which are rich in RNA (ribonucleic acid) with the microsomes; but most authors have not made this distinction (for discussion see Palade, 1955). An essential difficulty in making any classification of microsomal particles as yet is that the many different components of the fraction obtained by ultracentrifugation have not been identified by electron microscopy. As Hogeboom and Kuff (1955) point out, this fraction is far from homogeneous and contains elements which are difficult to relate to the original structure of the tissue from which they are derived. It seems probable, however, that of the organized structures of the nerve cell those which chiefly contribute to the microsomal fraction will be the granules and membranes of the Nissl bodies and Golgi apparatus and a number of other formed elements-not associated with these bodies-which are both granular and vesicular and which have been identified in various parts of the neuron including the axon and its terminals. Much emphasis has been placed on the presence of large numbers of microvesicles (diameter = 200-1000 A.) in the region of the synapse; and some have thought they may be the carriers of the transmitter substances such as acetylcholine and sympathin (see de Robertis and Bennett, 1955; and Fernhdez-Morkn, 1957). Other evidence to be considered later, to some extent, conflicts with this view. Dense granules approaching the dimension of the smallest mitochondria and microvesicles of the size found at the synapse have both been observed in the nodal axoplasm of frog nerve by Robertson (1957).

170

CATHERINE 0. HEBB

From a more general point of view, it seems legitimate to regard the mitochondria and the submitochondrial granules and vesicles as potentially mobile organelles within an otherwise static system comprising the cell membrane, the nucleus, and such structures as the Nissl bodies and Golgi apparatus. In referring to the synaptic microvesicles as possible “vehicles” for transmitter agents, Fernhndez-Morh ( 1957) presumably is thinking of them in a similar way. The mobility of mitochondria in other tissues has been shown by phase contrast cinematography (see Fr6d6ric and ChBvremont, 1952). In the nerve cell some system of internal transport appears to be a requirement if only for the nutrition of the processes which are so far removed from the perikaryon and the seat of its chief metabolic activity; this requirement is even more evident for nerve cells to which secretory functions have been ascribed. Generally, the transport of secretory materials has been explained in terms of Weiss’ concept of the flow of axoplasm (Weiss, 1947). His work, supported by other studies reviewed earlier (see Hebb, 1957a), strongly suggests that in the axon there is a pressure gradient directing flow away from the cell body. The original idea was put forward to account for the transport of axoplasm without special reference to any cytoplasmic organelles; but it seems equally applicable to a nonhomogeneous system comprising large and small particulate elements. As Weiss (see Weiss and Hiscoe, 1948) points out, the axon may have a volume several hundred times that of the perikaryon and with normal wear and tear new protein must be supplied to its distal parts. His experiments provide fairly critical evidence that the replacement originates from synthesis in the cell-body. The suggestion now made is that more than protein synthesis may be involved and that the transport of all materials depends upon their association with some type of carrier. Further, it may be necessary to think of this transport system in terms of growth rather than of flow.

B. CHEMICAL ORGANIZATION OF THE NEURON The technique which has been most useful in the study of the chemical systems of the cell has been differential centrifugation. In principle this is a method of microdissection which makes it possible to segregate and isolate different classes of cell organelle by virtue of the fact that they sediment at different rates depending upon their shape, size, and density. The first step is to prepare a tissue homogenate; and for this the most useful suspending medium is isotonic sucrose (for mammals 0.3 and not 0.25 M sucrose, as is usually stated, is isotonic). The sucrosesuspended homogenate is then spun down for varying periods a t successively higher speeds; and the sediment from each centrifugation collected


171

as a separate fraction. The method is usually used to prepare three particulate fractions, the nuclear fraction, the large granule or mitochondrial fraction, and the small granule or microsomal fraction as well as the soluble fraction which is the final supernatant free of all formed elements. To reduce contamination between fractions the precipitates are washed, re-suspended, and re-centrifuged a t the same speed one or more times; or additional centrifugations a t other speeds may be interposed; or again, the spiming down of the precipitates may be done in swing-out buckets instead of angle-heads so that they can be collected as several separate layers. Usually, whatever technique is adopted, the more homogeneity in respect of particle size that is achieved, the less complete is the recovery of the substance that is being studied. It has been emphasized, however, that data about the intracellular distribution of enzymes or other substances carry weight only if the recovery from separatq.fractions is reasonably close to the amount present in the unfractionated material (Schneider and Hogeboom, 1951). As a first approach, therefore, it is possibly safest to obtain an over-all balance sheet from less pure fractions and, when this has been done, go on to examine individual fractions with more care. It has for long been clear that both the large and the small granule fractions contain more than one kind of particle. For example, the mitochondrial fraction contains many particles which are similar to one another in many of their staining reactions, but which differ from one another in their enzyme complement (for discussion see Lindberg and Ernster, 1954) and dimensions (ratio of smallest to largest may be of the order of 50). Hogeboom and Kuff (1955) have demonstrated that particles of different sizes belonging to the microsomal fraction have different biochemical characteristics. A refinement in sedimentation technique which has been of considerable use in separating the different classes of particles from such disperse fractions has been the specific gravity or density gradient technique (Kuff and Schneider, 1954). This has now been applied to the mitochondria1 fraction of a number of tissues with very valuable results; and has recently been used in the partial isolation of granules carrying neurohumors (Blaschko et al., 1957; Hebb and Whittaker, 1958). Cell particulates of liver have been more extensively studied than those derived from other tissues; and their biochemical characterization is the basis for many of the current generalizations about the intracellular geography of metabolic systems. Investigation of brain particulates hss lagged far behind. In such wofk as has been done on them the greatest attention has been paid to the “large granule” fraction, to com-

172

CATHERINE 0. HEBB

parison of its properties with those of the mitochondria of other tissues (especially the liver) and with those of whole homogenates of brain. Much of the information gained from this work has been reviewed by authors contributing to a recent volume edited by Richter (1957) which should be consulted for more detailed coverage of the subject than can be given here. From the articles dealing specifically with cytochemistry, those contributed by Pope and Hess (1957) and by Coxon (1957) , two general conclusions emerge. The first is that the chemical organization of the neuron is similar in principle to that of any other cell. In particular, the mitochondria1 fraction is the site of the important oxidation systems of the cell, of phosphorylation coupled with oxidative reactions, and of the components of the electron transport system, both coenzyme I and I1 (DPN and TPN) and the cytochrome system; the nucleus is found to contain all of the cellular DNA (deoxyribonucleic acid) and, in some unspecified way, is concerned with the microsomal fraction and the mitochondria in protein Synthesis; the microsomes (in this respect equated with the Nissl granules) contain a large proportion of the cell ribonucleoprotein which is probably catalytically conoerned in protein synthesis, possibly in transpeptidations and, as Lindberg and Ernster (1954) have put it, in “protein-shaping.” In these dispositions the neuron is entirely similar to other cells. Where it has been found to differ from other cells is in the inability of its mitochondria to oxidize octanoic acid, aspartic acid, and alanine (Brody and Bain, 1952) and in their ability to bring about the complete oxidation of glucose (Gallagher et al., 1956). This leads to the second general conclusion, namely, that the peculiarities of its mitochondrial metabolism “mirror the peculiarities of brain in vitro metabolism” (Pope and Hess, 1957), that is, as it has been determined on brain homogenates. One exception to this may be the oxidation of fats. The finding of Brody and Bain (1952) that rabbit brain mitochondria did not oxidize octanoic acid could occasion little surprise since homogenates and acetone-dried powders made from brain have not been found to oxidize fatty acids. Furthermore, these preparations have little or no FAAE (fatty acid activating enzyme activity) and an essential preliminary step in fatty acid oxidation is their activation, i.e., by esterification with coenayme A (CoA). The evidence therefore favors the view expressed by a number of writers that the inability of the neuronal mitochondria to oxidize fatty acids, which sets them apart from the mitochondria of all other tissues, is in keeping with the inability of the tissue as a whole to oxidize fatty acids. It now appears that these views must be revised. In the first place,


173

evidence from various laboratories (see Rossiter, 1957) shows that respiring slices of rat brain can oxidize octanoic and palmitic acids as well as the triglyceride of lauric acid. Secondly, there are the recent findings by Vignais and associates (1958) that preparations of rat brain can activate and oxidize a variety of fatty acids including palmitate, stearate, butyrate, octanoate, myristate, and laurate. The rate of oxidation is highest for palmitate (about one-fifth the rate observed in liver) and for stearate; while it is smaller but still significant for the other substances. An important result of the investigation is that these oxidations are not demonstrable with whole homogenate, as previous workers had found, but only with separated brain fractions of which the most active is the supernatant. This has nearly twice the activity of the mitochondria and microsomes combined. Of the last two fractions the mitochondria are slightly the more active. The authors also find that the mitochondria1 but not the supernatant system prefers palmitate to palmityl-CoA and oxidizes it more rapidly. This last observation suggests that there may be a permeability barrier which permits palmitate to enter the mitochondria more rapidly than its coenzyme A derivative and, if this is so, that a t least a portion of the activating enzyme is intra-mitochondrial. That would conform with the position of the acyl-activating enzymes in other tissues including AAE, (acetate-activating enzyme) which promotes the formation of acetyl- and propionyl-CoA and is, therefore, analogous in action to FAAE, although obviously of wider functional significance since acetylcoenzyme A is one of the key substrates in the Krebs cycle and essential for a variety of syntheses involving simple acetylations (e.g., the formation of acetylcholine). It is possibly significant that the species on which Vignais et al. made their observations was the rat while the earlier experimental evidence for the absence of fat oxidation by nervous tissue was obtained on guinea pig brain slices and on rabbit brain mitochondria. It would be of interest, therefore, if experiments similar to theirs were now to be done on nerve tissue of other species as well. Vignais et al. obtained most of their results from the brains of young rats but they state that the brain particulates of older rats have similar properties. This is a necessary control because, as Lynen (1957) has emphasized, the ability of the nervous system to synthesize fatty acid-a process also involving acyl-CoA formation-diminishes with age. It is possible, however, that in the rat, where other growth processes continue, albeit very slowly, much longer (relative to their life span), this capacity to oxidize fat is related to a continuation of brain development or to a reparative capacity not found in other kinds of mammals.

174

CATHERINE 0. HEBB

The fat metabolism of the brain is of additional interest because of the high phospholipid turnover of nerve tissue. The possible role of phospholipid metabolism in fatty acid oxidation has been discussed by Dawson (1957), who has pointed out their possible significance as structural units-of the mitochondria-indirectly controlling fatty acid oxidation. Since phospholipids form some part of all the cell structures, the nucleus, membranes, mitochondria, microsomes, etc., as well as of t h ? myelin which is a complex of cholesterol, cerebrosides, proteins, and phospholipids the dominant one being sphingomyelin (see Dawson, 1957) , it is clearly not a simple problem to determine the geography of their metabolism. Presumably glial cells have a large part in it. The mitochondria-whether from neuronal or glial tissue is not determinedhave a function in the synthesis of phospholipids since Rossiter et al. ( 1957) have recently shown that Ps2-labeled cytidine diphosphate choline is incorporated into the lecithin formed both by whole brain homogenates and by brain mitochondria1 preparations (see also Quastel, 1957).

A question of interest is the source of the fatty acids which are built into the phospholipids. Lynen’s work (1957) shows that there is active fat synthesis and oxidation during development of the nervous system and it suggests also that a moderate amount of lipid synthesis may be proceeding continually but more slowly in adult brain tissue. Perhaps this is just enough to renew sphingomyelin of the myelin sheath and the phospholipids built into expendable organelles of the neuron such as the mitochondria. With the breakdown of phospholipids, the constituent fatty acids are apparently released still attached t o coenzyme A so it is conceivable that, depending upon the site of the breakdown, these fragments may re-enter the phospholipid cycle and hence reduce to a minimum the need for the synthesis of new fatty acids. Another metabolic distinction of brain mitochondria is their ability to oxidize glucose completely (Gallagher et al., 1956). Exactly what part they play in anaerobic glycolysis is, however, uncertain (see Pope and Hess, 1957; and see Aldridge, 1957). It is known, however, that the mitochondria contain all the enzymes of the Krebs citric acid cycle and are able to oxidize pyruvate through this pathway. I n this cycle a key metabolite is acetyl-CoA. From present evidence it appears that in addition to other mechanisms, this may be produced in brain mitochondria both by direct acetylation, by the action of AAE (see above) and by the “citric” enzyme which forms it from citrate in the presence of ATP (adenosine triphosphate) and coenzyme A. These appear to be the two main pathways by which acetyl-CoA is supplied fop in vitro acetylation of choline (Balfour and Hebb, 1952).


175

It has also been mentioned that mitochondria oxidize glutamic acid but not two other amino acids, alanine and aspartic acid. The preference for glutamic acid is not surprising in view of its possible importance as a precursor of transmitter substances (see Section IV) and the part which it may play in protein synthesis. The intraneuronal synthesis of protein appears to proceed at relatively high rates, though not so rapidly as in glandular tissue; it is confined to the cell body and possibly the proximal parts of the dendrites because of the conjunction there of two tissue structures, the nucleus and the Nissl bodies which are essential to it. Mitochondria play some part in protein synthesis too, possibly in the production of certain key peptides (y-glutamyl peptides) which are utilized in further synthetic steps by the microsomes (see Lindberg and Ernster, 1954). Data reviewed by Waelsch (1957) however, show that the highest protein turnover and the shortest half-life of proteins is in the microsomal fraction of brain homogenates; that this fraction incorporates labeled amino acid more rapidly than any other fraction and so it appears, as, indeed, much other evidence indicates, that the chief seat of protein synthesis is the granular component of the Nissl bodies. The nucleic acid protein of these bodies, which accounts for their basophilia and which is active in the synthesis, is in turn secreted by the nucleus. It thus seems that the protein of all the mitochondria and microsomes as well as of other structures is derived from the Nissl bodies. How the granular or microsomal protein of these bodies is transformed into mitochondria1 protein can only be a matter for speculation as yet; but one possibility is that some of the microsomes found in the Nissl bodies are transformed into the larger particles by a process of growth or accretion. Indeed, it may be better to regard a t least some of the microsomes as structural units rather than as the workshops in which these are fashioned. There are a number of indications that mammalian brain mitochondria differ from other mitochondria not only in their metabolism but also in their structure. Thus, in this laboratory it is our experience that they are not as osmotically sensitive as other mitochondria in the sense that deformation does not occur so easily in solutions that are hyperor hypotonic; further, Tapley and Cooper (1956) find that brain mitochondria, like spleen and testis mitochondria, scarcely show any swelling in response to thyroxine while those of liver, kidney, and heart do; and in addition, Berger and co-workers (1956) find a difference in the reBponse of brain mitochondria and liver mitochondria to chlorpromazine which may depend upon a permeability or membrane difference.

176

CATHERINE 0. HEBB

111. lntracelluiar Storage and Transport of Neurohumoral Agents: Neurosecretion

A. SYNTHESIS AND STORAGE OF ACETYLCHOLINE It has long been known that much of the ACh (acetylcholine) occurring in mammalian brain and nerve is in some way bound to a granular component of the tissues (see Feldberg, 1945). The nature of this binding and the type of tissue particle, whether a macromolecule or a tissue organelle, has been a matter for much speculation (see Birks and MacIntosh, 1957). De Robertis and Bennett (1955) , del Castillo and Katz (1956), and Fernhndez-Mor&n (1957), among others, have been attracted by the possibility that ACh is contained in the synaptic microvesicles or microsomes, on average 0.02 to 0 . 0 6 ~in diameter. Various other authors (for references see Hebb, 1957a, and Hebb and Whittaker, 1958) have suggested that ACh may be contained in the mitochondria or at least in a particle that is osmotically sensitive. The speculation that ACh and possibly other neurohumors are carried in the synaptic microvesicles is attractive because the ejection and breakdown of such microcarriers would adequately account for the quanti~edrelease of ACh at the neuromuscular junction which has been demonstrated by del Castillo and Katz (1956). These authors suggest, however, that the vesicle breaks down within the cell in contact with the nerve membrane which becomes permeable at the point of contact and thus permits the extracellular ejection of its charge of ACh. Fernhndez-Morh (1957) on the other hand, appears to favor the view that the microvesicles, while still intact, carry the charge of ACh or other neurohumors through the membrane and then release it. Apropos of this suggestion it is perhaps worth recalling the claim of Carey et al. (1946) to have demonstrated the discharge of neurosomes or neurogenic granules from motor nerve endings. However, it is well known that ACh synthesis only slowly declines in the terminal parts of an axon after it has been sectioned (Hebb and Waites, 1956) and that repetitive stimulation for several houra does not lead to a loss of their ability to release ACh. Since the ACh particle probably contains choline acetylase as well as the ester, it follows that these particles must for the most part remain inside the nerve (see later discussion). Although these theoretical considerations suggest that ACh is in the synaptic microvesicIes they are not at first sight supported by cytochemical experiment. In an investigation of brain fractions separated by differential centrifugation, Hebb and Smallman (1956) found that choline acetylase, the enzyme which synthesizes ACh, is mainly in the


177

large granule fraction. Of the total enzyme activity of rabbit brain some 50 to 70% was associated with the mitochondrial fraction, while the microsomal fraction was consistently one of the least active. The remainder of the activity was in the nuclear and final supernatant or soluble fractions. From these experiments and subsequent experiments by Hebb and Whittaker (1958) it appears that the nuclear fraction activity is high in proportion to the amount of tissue which remains incompletely homogenized, while the soluble enzyme tends to be higher if homogenization is too severe. In spite of these variations however, the mitochondrial fraction contains the greatest activity. Further, it has been shown that ACh and choline acetylase probably reside in the same particle-the evidence for this is the correlation between their distributions in extensively fractionated material; that in this position ACh is inactive and may be equated with the “bound” ACh; that in the same position choline acetylase is also inactive; and that the release of ACh and activation of enzyme can be partially effected by procedures such as freezing and thawing, dilution from hypertonic to isotonic solution, prolonged storage a t 0 to 10°C in isotonic solution, or wholly effected by treatment with ether. All of the bound ACh can also be released by heating in acid, which of course is not practicable to test in the case of the enzyme. These observations lead to the conclusion that the enzyme is “inactive” and the ester “bound” because of a membrane barrier preventing exchange of substrates and reaction product between the inside of the particle and the surrounding fluid (of the incubate). The observations also suggest that the partial release of ester and enzyme, which is effected by the procedures described, occurs because some particles may be more labile than others and that these possibly represent an aging part of the total population of similar particles. By centrifugation of the large granule fraction over a density gradient (0.3, 0.8, 1.0, 1.2, and 1.6 M sucrose), Rebb and Whittaker were also able to show that the ACh particles are almost certainly not identical with the true mitochondria and can be nearly completely separated from them. Thus, in the density gradient tubes the ACh and choline acetylase activity are in the 0.8 to 1.2 M layers and highest in the 1.0 M layer while the activity of succinic dehydrogenase, a mitochondrial enzyme, is highest between the 1.2-1.6 M layers and at a minimum in the 0.8 to 1.2 M layers. These experiments have since been confirmed in further unpublished experiments. In other experiments since carried out by Whittaker (personal communication) he has examined with the electron microscope subfractions of brain mitochondrial preparations isolated by similar techniques and finds that there is also a morphological difference between the

178

CATHERINE 0. HEBB

succinic dehydrogenase particles and those which appear in the layers that have the high ACh and acetylase activity. The succinic dehydrogenase layer contains particles which have the typical appearance of mitochondria with cristae mitochondriales ; but the other particles are more vesicular in type, lacking any organized internal structure, and are smaller being usually between 0.03 and 0.3 p in size. Some of these particles are, therefore, of microsomal dimensions and so could be those already identified in the synaptic axoplasm by electron microscopy. There are, nevertheless, two difficulties about accepting this conclusion. The first is the constancy with which the ACh particles (this is here used to denote particles that contain choline acetylase as well as ACh) appear in highest concentration in the mitochondria1 fraction; and if this is obtained by two centrifugations, one at approximately 160,000 g mid, the other at 450,000 g min, and the two precipitations analyzed separately, the higher acetylase activity is found in the precipitate brought down at the lower spinning force and presumably containing the higher size range of particles. The second difficulty is that the microsomal fraction from whole brain has not only a low absolute activity it has a low relative specific activity (relative to protein nitrogen) which is evidence against the view that microsomal particles contribute much of the ACh activity (Hebb and Whittaker, unpublished experiments). However, in the experiments referred to there was one exception; this was on the caudate nucleus of the sheep. In this tissue although there was a low absolute activity in the microsomal fraction (about 10% of the total) the relative specific activity was the highest for all the fractions. Since on other evidence (Hebb, 1957c) the choline acetylase and ACh of the caudate nucleus are probably derived from cholinergic nerve endings it may be that at such sites all the enzyme and neurohumor are in fact contained in microvesicles. That is to suggest of course that at other sites the particles are larger and that in the course of development and transport to the nerve endings, by division or some other mechanism, they give rise to the small synaptic vesicles (see also Birks and MacIntosh, 1957). Possibly this size of particle represents the true functional unit throughout; and the larger forms which are presumed to occur because of their sedimentation characteristics are the result of coalescence with one another. A way of testing this might be found in redetermining the sedimentation rates of the small particles which have been isolated by density gradient. They should sediment at rates characteristic for microsomes if they are the residual parts of a larger particle isolated in the first stages of the fractionation. Further it would possibly be rewarding to make a more g min = relative gravitational force x time in minutes.


179

careful regional examination of the nervous system so as to obtain preparations in each of which some one part of the neuron, the cell body, the mid-axon, or the axon terminals, of cholinergic nerves is the chief component. The idea that smaller organelles can be formed from larger ones is not wholly without experimental foundation. Mitochondria1 fragments produced by supersonic vibration and by detergent agents acting on niitochondrial suspensions (Abood and Alexander, 1957; McMurray et al., 1958, Kielley and Bronk, 1958) are capable of carrying on oxidative phosphorylation with a number of substrates, and according to the first two groups of workers the disrupted fragments appear in suspension in the form of shall vesicles (0.014.1 p in diameter were obtained from brain by Abood and Alexander; and others 0.05-0.5p in diameter were obtained from rat liver by McMurray et al.). It is of interest too that whole brain mitochondria1 fragments obtained by detergent treatment seem to maintain the integrity of their metabolic systems rather better than do similar extracts of liver. Thus, the arrangement of metabolic systems in mitochondria is compatible with their splitting up into smaller functional units; and it is possibly relevant that Leon and Cook (1956) have claimed to observe the spontaneous transformation of large mitochondria into pairs of smaller mitochondria. Whether these findings can be applied with any profit to the organelles which carry ACh is another question. It is tempting indeed to conclude that the ACh vesicles are among those already identified in the synaptic axoplasm and the value of Whittaker’s recent work which brings this into the realms of possibility should be acknowledged; but i t is obviously necessary that some more direct method of identifying the AChcarrying particles be devised if this conclusion is to be substantiated. Choline acetylase and ACh are found in the particulate not only of mammalian but of avian brains as well (Hebb and Whittaker, 1958) ; the enzyme has also been detected in the large granule fraction from the brain of the grass snake, Natrix natrix; the frog, Rana temporaricr,; and wrasse, Labrus bergylta, a teleost fish. In certain insects, however, choline acetylase may be in the soluble fraction (Smallman, 1957). A possible explanation for this last observation is that the particles carrying the enzyme are exceptionally fragile and consequently disrupted in the course of the procedures used t o isolate them. As mentioned earlier, the mammalian particulate enzyme is not fully active unless treated with organic solvents of which ether is the most efficient. It is also necessary to treat avian brain homogenates to make them active. Such treatment however has no activating effect on the activity of reptilian, amphibian, or teleost brain homogenates. I n

180

CATHERINE 0. HEBB

the case of wrasse brain untreated homogenates synthesize about 5,500 pg per gram fresh brain per hour under optimal conditions which is about five times the output of the rabbit brain. The rate is slightly depressed by ether treatment (Milton, 1958). On the other hand, it is even more steeply depressed by the presence of sodium ions in the incubate, an effect which appears to be due to the reduction of choline entry into the particles (Hebb and Milton, unpublished work). Wrasse particulates have a special interest in that they seem to be more osmotically sensitive than are mammalian particulates ; and this may be connected with the fact that the membrane of their choline acetylase particles is evidently more permeable to substrates and cofactors than are the analogous particles derived from mammalian brain. In vertebrates, then, so far as is known the enzyme system for the production of ACh is enclosed within a tissue organelle of some kind. It is this arrangement which would appear to secure the functional specificity of the enzyme which otherwise might not act specifically for the acetylation of choline (see Hebb, 1957a) ; and at the same time this is an arrangement which prevents intracellular diffusion of ACh and segregates the ester from intracellular enzymes such as cholinesterase that might destroy it.

B. NOWRENALINE AND ADRENALINE TISSUH~ STORAGE Compared to ACh the catechol amines have a very limited distribution in brain and have known transmitter functions only in peripheral adrenergic nerves and chromaffin tissue. In the adrenal medulla adrenaline and noradrenaline are stored in separate particles between 0.1 and 0 . 6 in ~ diameter, each amine apparently in conjunction with just sufficient ATP to unite with them chemically and so to act as a binding agent within the particles (Blaschko et al., 1957; von Euler, 1957, who also gives earlier references; Schiimann, 1958a). These are distinct from the mitochondria. Blaschko et al. find that the particles containing the amines are in the large granule fraction but are much heavier than mitochondria and on density-gradient centrifugation appear in the 2.00 to 2.25 M sucrose layer. Noradrenaline in sympathetic nerves is also in the large granule fraction and apparently combined with ATP (Schumann, 1958b; see also von Euler, 1957). This binding with ATP may be true of other pharmacologically active amines (Born et al., 1958). Like ACh, adrenaline is found in the cell bodies as well as in the axons of the neurons which secrete it. It is presumed, therefore, that its carrier-particles originate in the perikaryon. According to von Euler it is probable that these particles contain an enzyme system for the formation

CHEMICAL AGENT8 OF THE NERVOUS SYSTEM

181

of noradrenaline from a precursor, presumably dopamine. Amine oxidase, which under certain conditions may be responsible for the destruction of noradrenaline (Kunteman et al., 1958) but is possibly not always so (Brown and Gillespie, 1957), is in the true mitochondria1 fraction and is not present in the granule fraction that contains the catechol amines of the medulla (Blaschko et al., 1957).

C. NEUBOSECRETION During a period of some 25 years the Scharrers (E. and B. Scharrer) have built up an impressive documentation of evidence to show that there are neurosecretory cells, i.e., nerve cells with a glandular function, in the nervous systems of both invertebrates and vertebrates. The evidence for their ideas has been most completely reviewed in English in a paper they published in 1954 (Scharrer and Scharrer, 1954) which deals with the presence of neurosecretory cells in two hypothalamic nuclei, the supraoptic and paraventricular nuclei. There is histological evidence for the existence of neurosecretory cells in some other parts of the vertebrate nervous system but the hypothalamic cells have been those most intensively investigated. The method used in the detection and investigation of neuroseoretion as described by the Scharrers is histological, possibly one should say histochemical as well except that the basis of the staining reactions of the neurosecretory material is not well understood. The first point of importance, however, is that on histological criteria, the cells of the hypothalamic nuclei (those mentioned above) show in common with a number of invertebrate nerve cells, the characteristics of both neuron and glandular cell. They have Nissl bodies, neurofibrils, and the type of nucleus and cell processes characteristic of neurons ; but like glandular cells they show, on occasion, accumulations of granular material that may fill the whole cell and obscure the Nissl bodies, or they may exhibit vacuolation and reduced numbers of granules. In addition, their nuclei undergo changes in shape and position of a kind typically seen in gland cells; and finally, colloidal masses of material that are formed in the cell body or perikaryon can also be detected in the axon and outside of the cell as well. The general theory developed by the Scharrers is that the material found in the hypothalamic cells consists of the hormones eventually secreted from the posterior lobe of the pituitary (there is evidence for that identification) combined with some carrier substance, that it passes along the axon and so comes to be stored in the bound form in the nerve endings lying within the posterior lobe of the pituitary. The speed of

182

CATHERINE 0. HEBB

movement of the material along the axons is estimated to be about the same as the rate of axonal flow estimated by Weiss and Hiscoe (i948), in crustacean nerves, about 3 mm per day. Supporting evidence for the theory is that an interruption or severance of the nerve tracts from the nuclei to the hypophysis the secretory material is dammed up on the proximal side of the section. More recently, Andersson and Jewel1 (1957) have provided additional evidence for this transport system by experiments on dogs in which they have shown that excessive hydration, a condition under which the release of antidiuretic hormone would not be Eiimulated a t all, leads to a great accumulation of the neurosecretory material in the nerve tract between the hypophysis and the supraoptic and paraventricular nuclei. This has been confirmed for the cat by Wethington (1957). There are thus very obvious analogies between the events which occur in these neurosecretory neurons and the events suspected to occur in the neurons which release ACh and noradrenaline a t their endings. If one is right about what happens in these, the cholinepgic and adrenergic neurons, then they too can be regarded as neurosecretory. The fact that they have not been identified as such by histology is, as the Scharrers point out, probably due only to the circumstance that the type of material they secrete does not form aggregates of a kind which is detectable by microscopy. A possible additional reason is that owing to their nearly continuous activity they do not have the same chance to accumulate material that the hypophysial nerves which may not be so frequently stimulated, have. In function there is, of course, a very real difference. ACh and noradrenaline have an action which in normal circumstances is strictly localized to structures immediately under the nerve endings, and are destroyed or withdrawn from action immediately after their local effect has been produced. Thus, they are “local” as opposed to hormones which act “par distance” by circulatory distribution. The posterior pituitary hormones, the polypeptides vasopression and oxytocin belong to this latter category. For them there is no “polypeptidase” that is equivalent in function to true cholinesterase. Otherwise it would appear, however, that the neurosecretory nerve is like the cholinergic or adrenergic nerve; impulses generated in the membrane of the cells of the hypothalamus probably by ACh (see Abrahams and Pickford, 1956) are conducted along the axon and by some unknown means liberate the neurosecretory substances presumably by the same mechanisms that lead to the release of ACh from the endings which innervate the neurosecretory cells themselves.


183

A study of neurosecretory processes in the hypothalamic cells by cytochemical methods would seem to offer many advantages. The fact that the hypophysial material is so readily identifiable by its staining reactions as well as being identifiable by its hormonal actions suggests that cytochemical analysis of its distribution would proceed towards success with greater rapidity than have the experiments on the ACh particles ; and similarly electron microscopic studies of the hypophysial nerves might also yield fresh information about the storage and transport of neurohumors. A start along these lines has recently been made by Palay (1957). Under the heading of neurosecretion it is possible that one should also consider the production and transport of true cholinesterase. This is an intracellular constituent of cholinergic neurons and is present also in the postsynaptic structure innervated by cholinergic neurons. Although some of the postsynaptic enzyme may be derived from other sources, there is good reason to believe that a part although not all of it comes from, and is produced by, the cholinergic neuron (see Hebb, 1957a; Lewis and Hughes, 1957). Thus, the presence of cholinesterase in the axons of cholinergic nerves can be explained in the same way as the presence of choline acetylase. Both are in transit. It is of interest in this connection that interruption of nerves by section which has been shown to lead to an accumulation of neurosecretory material in neurosecretory nerves also leads in cholinergic nerves to an accumulation of both true cholinesterase and choline acetylase proximal to the section (see Hebb and Waites, 1956). As has been stressed the process of neurosecretion as described by the Scharrers is only demonstrable in selected groups of cells. Among these they have not included the Purkinje cells of the cerebellum. However, recently these cells have been so described by Shanklin et al. (1957) while the term “neurocrine” which has a slightly different connotation has been applied to them by Mosinger (1951). Shanklin et al. in classifying the Purkinje cells as neurosecretory have built up quite a complex picture of their activity. According to their ideas the cells produce a hormone of unknown function and type which passes through the cell wall and eventually into the blood stream. These ideas seem to be somewhat tenuously based on a very complex histochemical and histological picture and unlike the carefully built up theories of the Scharrers cannot be correlated with any known functional attributes of the cells in question, Thus, while the evidence itself is of interest in contributing more information about the constitution of the Purkinje cells the theoretical interpretation does not at the moment seem justified.

184

CATHERINE 0. HEBB

IV. Chemical Transmission in the Central Nervous System of Mammals

A, EXCITATORY TRANSMITTER AGENTS It remains to consider what is the present status of the theory of chemical transmission in the central nervous system (CNS) . Acetylcholine is still the only substance to establish itself with any certainty as a central chemical transmitter but, as pointed out earlier, biochemical and histochemical evidence indicates that only a proportion of the central neurons are cholinergic (Hebb, 1957a, c). The simplest test of whether a neuron is of that type is to determine its content of choline acetylase since the absence of choline acetylase from a neuron means that it cannot be cholinergic. Additional evidence that is useful in tracing pathways can be obtained from the distribution of true cholinesterase. By histochemical methods this enayme can be detected in the perikaryon of cholinergic neurons and concentrated a t the membrane of cell bodies of cholinoceptive neurons (Hebb, 1957b, c) . Studies of this kind show that second order sensory and the terminal motor neurons, both cranial and spinal, are consistently cholinergic; probably, too, the fourth neurons in the sensory pathways are all cholinergic. Other potentially cholinergic neurons are the aff erents to the caudate nucleus including some that arise from the neocortex. The neocortex where it is well developed can have few neurons of this type and comparative data on choline acetylase suggest that its evolutionary enlargement has been due to the multiplication of non-cholinergic neurons (see Fig. 1). It will be noted, however, that enlargement of the cortex in some species (e.g., the dog) has not led to so great a reduction in cholinergic neurons as in other species (e.g., the pig). In all mammals the cerebellum has a low concentration of choline acetylase; what there is can be accounted for as ennyme in the axom and ending8 of the second order afferent8 (the mossy fibers) which enter the cerebellum chiefly through the superior and inferior peduncles: and fanning out in the white matter end in the granule layer of the cortex. The granule cells, the basket cells, and the Purkinje cells all appear to be non-cholinergic. It is, therefore, of special interest that the) excitatory principle isolated and studied by Crossland and his colleagues: (Crossland and Mitchell, 1956; and Crossland, 1957) which is discussed below was first obtained from this organ. Since ACh cannot be the excitatory transmitter a t all central sy-. napses other possible candidates have been canvassed. OnIy those which! are natural constituents of brain tissue have a serious claim for atten-

’


185

tion. On this basis some choline esters other than ACh may be considered. Propionylcholine present in cattle spleen (Banister et al., 1953) is probably also present in brain (Hosein, 1957a; Hebb, unpublished experiments). Butyrylcholine although not extracted from brain is found to be synthesized by homogenates in the presence of eserine without substrate addition (Henschler, 1956), while indoleacetylcholine has been reported by Gruner and Kewitz (1955) to be present in mammalian brain as well. One may doubt, however, that these esters have a transmitter function since their concentrations in the brain are low and they

Rabbit

Fro. 1. The concentration of choline acetylase in the cortex of seven mammals including man. The density of stippling of the left half of each diagram is proportional to the concentration of choline acetylase, numerical values for which are indicated on the right side as micrograms ACh synthesized per gram dried brain per hour. Except in the rabbit and guinea pig the striate cortex has the lowest concentration of enryme (for additional values see Hebb, 1957a).

are all preferentially hydrolyzed by pseudocholinesterase, an ensyme which is situated chiefly in glial tissue. So it seems more likely that they are intermediates in extraneuronal metabolism. I n passing, it may be worth noting however that pseudocholinesterase is also present within the nucleus of many nerve cells. At that site it could have no obvious function in excitatory processes or in the metabolism of the neuron as a whole ; but the possibility that pseudocholinesterase has a function in transmission is suggested by the experiments of Desmedt and La Grutta (1955, 1957), who have shown that doses of anticholinesterase, which

186

CATHERINE 0. HEBB

inhibit brain pseudocholinesterase selectively, produce an arousal reaction (EEG) in cats. Beyond this, however, there is no evidence to connect pseudocholinesterase with the excitation of neurons. Other possible transmitter substances which are natural constituents of brain including histamine, noradrenaline, adenosine derivatives, substance P, and serotonin (5-hydroxytryptamine) have been discussed by Crossland (1957) who concludes that “the evidence is inadequate to provide solid support for the claims of any of them to be considered as transmitter agents.” Since then no evidence has been produced to make it necessary to revise this dictum. The cerebellar excitatory substance, CES, described by Crossland and Mitchell (1956) is in a somewhat different category because it has not yet been chemically identified and is to be regarded not as evidence for the existence of a particular transmitter but as evidence for the presence of some transmitter which is not acetylcholine. Hosein suggests that the active principle of CES may be y-butyrobetaine (Hosein, 1957b). Like y-butyrobetaine it is destroyed in hot acid, a treatment which differentiates it from ACh which is stable in acid and unstable in alkali. Apart from chemical tests, CES is to be distinguished from ACh on the basis of its distribution; it is highest in such structures as dorsal roots and the optic nerve which in most mammals contain neither ACh nor choline acetylase (see Hebb, 1957a). Its high concentration in optic nerves is of interest in view of an earlier observation that high concentrations of optic nerve on incubation, in a medium designed for ACh synthesis, form a bio-active substance other than ACh (Hebb, 1955). Except for its excitatory effect on the frog rectus muscle, this substance was not examined in detail. However, it seems unlikely that it is identical with CES which is said not to have any muscle-stimulating activity. Nor could it have been y-BB (y-butyrobetaine) since this has a curare-like effect on striated muscle. The identity of the betaine compound with CES must also be questioned because it has smooth muscle-stimulating properties (Hosein and McLennan, 1958). Nevertheless, the possible contribution made by y-BB to CES, which may be a complex of more than one active principle, seems to be worth further investigation.

INHIBITION B. CHEMICAL Central inhibition was first demonstrated experimentally and described by Sechenov in 1863 (for discussion see Babkin, 1949). Subsequently Sherrington analyzed the phenomenon of somatic motor inhibition further and pointed out (see Sherrington, 1947) that since there are no inhibitory nerves supplying the skeletal muscles, as there are in the caw of visceral muscle, somatic inhibition must be of central origin;


187

and the lowest (or most peripheral) level a t which it can be initiated is the anterior horn cell or its cranial homologue. Clearly, however, inhibition could be set up in any of the synapses linking the successive neurons of a nerve pathway provided that inhibitory neurons also contribute endings to the synaptic pool. From the experiments of Purpura and co-workers (1957a, b) it appears that there be many such inhibiting neurons in. the cerebral cortex, but relatively few in the cerebellum. The question to be considered is, that granting the existence of these neurons, how is their inhibitory action transmitted. Does it depend upon a chemical transmitter? Recent evidence strongly suggests that it does. According t o Eccles (1957) hyperpolarization of the neuron (measured by an internal electrode) is associated with, or synonymous with, an inhibitory state while excitation is associated with a reversal of this condition. These are changes which can be effected by selective alterations in the membrane that can, in turn, be produced by chemical transmitter substances, some of which are excitatory and some of which are inhibitory (see Hebb, 1957c, for an account of some evidence on this question). In a discussion of these it should first of all be noted that ACh, which is normally thought of as transmitting excitation, apparently acts as an inhibitory transmitter on heart muscle (Hutter, 1957) and so it is just conceivable that it has a similar action on some central synapses. However, it does not contribute to the action of Florey’s Factor I which can be isolated from mammalian brain and spinal cord (but not from peripheral nerves) and acts as an inhibitory antagonist to ACh on stretch receptor neurons of the crayfish (for references see Bazemore et al., 1956). Although it appears that more than one compound may contribute to the actions of Factor I (Elliott and Florey, 1956), one which seems to have most of its properties is 7-aminobutyric acid, a compound which can be formed from glutamic acid and is present in brain tissue. GABA (y-aminobutyric acid) may, therefore, be the transmitter substance of central inhibiting neurons. It may be noted, however, that Factor I probably contains other inhibitory substances since, in the discussion which followed the presentation of a paper on the inhibition of muscle contraction in crustacea, Wiersma (1958) recently stated that the action of the inhibitory nerve fiber could be mimicked by Factor I preparations but not by GABA. Following identification of the inhibitory properties of GABA, an identification independently confirmed by the investigations of Hayashi and Nagai (1956) and of Hayashi and Suhara (1956), a very considerable amount of work has been done on the study of its central actions (Purpura et al., 1957a, b; Purpura and Grundfest, 1957; Iwama

188

CATHERINE 0. HEBB

and Jasper, 1957; Marraaai et al., 1958, among others). Marrazai et al., who have also studied the central inhibitory actions of serotonin which are antagonized by chlorpromazine, suggest that the inhibitory action of GABA should be tested by a metabolic inhibitor interfering with the reversible reaction that converts glutamic acid to GABA. Purpura et al. have concentrated more on the mode and site of action of injected GABA and some of its chemical analogues, The conclusions to be drawn from their work seem to be that GABA has most of the right properties to be a central inhibitory substance; that it acts by blocking excitatory synapses; and that its actions can to some extent be mimicked by other substances of a similar chemical species. The problem now is to find out more about its metabolism and to discover more about the precise regional distribution of the enzymes which catalyze its synthesis (from glutamic acid) and its destruction. Moreover, following the general argument presented in the second section of this article an investigation of its intracellular distribution would be informative. Hosein has recently carried out a number of analyses of brain tissue and in particular has made an analysis of “bound” ACh (Hosein, 1957a, b, and unpublished work). On the basis of these studies he has concluded that bound ACh may consist at least in part of acetyl-carnityl CoA which he thinks corresponds to the “F” component of Banister et al. (1953). This, however, is only part of a somewhat elaborate metabolic scheme in which he has linked up the metabolism, on the one hand, of glutamic acid with the formation of inhibitory compounds including GABA and p-hydroxy-GABA with the metabolism, on the other hand, of a series of betaine esters which he finds have excitatory properties. The common intermediate is acetyl-CoA. Some of the same compounds have been studied by Hayashi and his colleagues (Hayashi and Nagai, 1956; Hayashi and Suhara, 1956) and it seems possible that more progress will be made in the discovery of central chemical transmitters by a further study of these metabolites. At the same time it must be pointed out that Hosein’s work is open to a number of serious criticisms. For example, his identification of the chromatographic “F” component is not supported by the work of either Henschler (1957) or Whittaker (1956). Again, the bound ACh referred to by Feldberg (1945) and described by subsequent workers (see Hebb and Whittaker, 1958) has usually been identified by rigorous tests that exclude all save other choline esters of which pyruvylcholine, if it exists, is the only known one that might be improperly identified as ACh. In spite of these and other criticisms, however, Hosein’s work is valuable in bringing to no-


189

tice new compounds with transmitter properties and i t is to be hoped that he will continue his investigations.

V.

Conclusions

The purpose of this article has been to examine without special regard to their merits the theories of the chemical transmission of nerve impulses and of neurosecretion within the wider context of the over-all metabolism of the neuron. Neurosecretion should not perhaps be used as a term to describe only the histochemically-demonstrable secretory processes of nerves such as those of the hypothalamo-pituitary system. The analogies are sufficient to indicate that similar processes are involved in the production, transport, and secretion of ACh from other nerve endings. So these, too, may also be called neurosecretory nerves. Whether all nerves are to be regarded in the same light is a different problem. The mammalian CNS is presumed to contain a fairly large number of cholinergic nerves ; but ACh cannot be the universal central transmitter; and while there are many possible candidates to suggest for the role of transmitter in other central neurons some of the ones which have been discussed have not been specifically identified with any pathway. It should be mentioned, too, that it is still largely assumption that ACh functions as a central transmitter. Strong as the circumstantial evidence is, the presence of ACh and its metabolizing enzymes and the correlations established by pharmacological data, no experiment has, as yet, been done to prove the central function of ACh beyond doubt. Although such proof may not be required for personal conviction, it remains a challenging gap in our knowledge. In this same connection I have, for long, felt that the general acceptance of the chemical transmission theory so far as i t concerns ACh should not be taken as carte blanche to apply the idea to all neurons. It may well be that all do transmit by the neurosecretory principle; but the experimental reasons for making the assumption that neurons work in this way are derived mainly from research on the neuromuscular junction, sympathetic ganglia, and the anterior horn cells. It is clearly dangerous, however, to assume that these provide all possible models of synaptic structure; and if the structure of some central synapses does differ then transmission across them may be governed by different principles. Even so mechanisms of the type by which it is thought the trans-

190

CATHERINE 0. HEBB

mitter agents are produced and transported within the cell may be common to all neurons; whether or not they serve the process of transmission they are evidently necessary for the normal metabolism of the axon. Considerable reliance has been placed here on Weiss’ theory of axoplasmic flow and modern evidence seems to confirm his view that the synthesis of much of the material of the axon takes place in the perikaryon and so must be moved from the one site to the other. Whether it is proper to think of this as flow, the term used above is, perhaps, open to question. It may be that the movement of subcellular organelles occurs through some other mechanism ; or that the renewal of distally-situated constituents of the axon depends upon some continued process of growth in which the extraneuronal elements, the glia, and cells of Schwann participate. Yet they cannot be wholly responsible for the renewal of the axon since interpretation of classic degeneration experiments is fortified by modern experiments showing that nerve section dams up on the proximal side a number of enzymes that are normally found in the cell body and over the whole length of the axon and is also followed by s multiplication of mitochondria in the perikaryon. It would be of interest now to learn whether this is also reflected in an increase of true mitochondria in the proximal part of the axon. Another area for speculation, if the theory of intracellular transport is accepted, is how this is regulated. One might be tempted to postulate some relation between the rate of production and axonal transport of material on the one hand and the level of activity of the nerve on the other. However, the experiments which show that neurosecretory material accumulates in the hypophysial nerves when they are inactive suggests that if there is such a mechanism it is not unduly sensitive. Possibly, however, this field is worth further exploration. In the course of this discussion I have contrived to ignore many of the technical problems and sometimes a downright lack of evidence for some parts of the argument. For example, one of the difficulties and an important one for cytochemical analysis is the technical problem of distinguishing between glial and neuronal metabolism in tissue slices, homogenates, and their fractions. This problem is perhaps not insoluble. Lowry (1957) and Giacobini (1957) have both found ways in which to examine the metabolism of individual cells, and no doubt future studies will be greatly aided by microtechniques of the kind they describe. Histochemistry in the case of some enzymes, is an additional aid in this problem but negative histochemical evidence, in general, carries little weight; and it is only positive evidence of the presence of a metabolite or enzyme in any tissue that can be regarded as meaningful. In the meanwhile the decision as to the part played by glial components in in

.


191

vitro metabolism must be inferred from indirect evidence. The conclusions reached, as with many other problems, is a matter of judgment rather than of experimental certainty. ACKNOWLEDOMENTS I wish to thank Dr. E. Hosein, Dr. V. P. Whittaker, and Dr. B. N. Smallman for access to some of their unpublished work. In addition, I wish to thank Mre. T. Hartridge, Miss M. Leonard, and D. W. Butcher for their assistance in the preparation of this manuscript.

REFERENCES Abood, L. G., and Alexander, L. (1957). J. Biol. Chem. 227, 717. Abrahams, V. C., and Pickford, M. (1956). J. Physiol. (London) 131, 712. Aldridge, W. N. (1957). Biochem. J . 67, 423. Andemon, B., and Jewell, P. A. (1957). J. Endocnnol. 15, 332. Andrew, W. (1956). Intern. Rev. Cytol. 5, 147. Babkin, B. P. (1949). ‘‘Pavlov, a Biography.” Chicago Univ. Presa, Chicago, Illinois. Balfour, W. E., and Hebb, C. (1952). J. Physio2. (London) 118, 94. Banister, J., Whittaker, V. P., and Wijesundera, S. (1953). J. Physiol. (London) 121, 55.

Baaemore, A. Elliott, K. A. C., and Florey, E. (1956). Nature 178, 1052. Berger, M., Strecker, H. J., and Waelsch, H. (1956). Nature 177, 1234. Birks, R. I., and MacIntosh, F. C. (1957). Brit. Med. Bull. 13, 157. Blaschko, H., Hagen, J. M., and Hagen, P. (1957). J. Physiol. (London) 139, 316. Born, G. V. R., Ingram, G. I. C., and Stacey, R. S. (1958). Brit. J . Pharmacol. 13,62. Brody, T. M., and Bain, J. A. (1952). J . Biol. Chem. 195, 685. Brown, G. L. and Gillespie, J. S. (1957). J. Physiol. (London) 138, 81. Carey, E. J., Downer, E. M., Toomey, F. B., and Haushalter, E. (1946). Proc. Soc. Exptl. Biol. Med. 62, 76. Claude, A. (1949). Harvey Lectures, Ser. 43, 121. Coxon, R. V. (1957). In “Metabolism of the Nervous System’’ (D. Richter, ed.), p. 303. Pergamon Press, New York. Crossland, J. (1957). In “Metabolism of the Nervous System” (D. Richter, ed.), p. 523. Pergamon Press, New York. Crossland, J., and Mitchell, J. F. (1956). J . Physiol. (London) 132, 391. Dawson, R. M. C. (1957). Biol. Revs. Cambridge Phil. Soc. (London) 32, 188. del Castillo, J., and Kata, B. (1956). Progr. in Bwphys. and Biophys. Chem. 6, 122. de Robertis, E. D. P., and Bennett, H. S. (1955). J . Biophys. Biochem. Cylol. 1, 47.

Desmedt, J. E., and La Grutta, G. (1955). J. Physiol. (London) 129, 46. Desmedt, J. E., and La Grutta, G. (1957). J. Physio2. (London) 136, 20. Eccles, J. C. (1957). “The Physiology of Nerve Cells.” Johns Hopkins Press, Baltimore, Maryland. Elliott, K. A. C., and Florey, E. (1956). J. Neurochem. 1, 181. Feldberg, W. (1945). Physiol. Revs. 25, 596. Fernhdez-Morhn, H. (1957). In “Metabolism of the Nervous System.” D. Richter (ed.), p. 1. Pergamon Press, New York.

192

CATHERINE 0. HEBB

FrBdBric, J., and Chhremont, M. (1952). Arch. biol. (Lidgel 63, 109. Gallagher, C. H., Judah, J. D., and Rees, K. R. (1956). Biochem. J. 62, 436. Gamer, H. (3. (1955). J . Uen. Physiol. 38, 709. Gasser, H. 8. (1956). J . Gen. Physiol. 39, 473. Giacobini, E. (1957). J. Neurochem. 1, 234. Gruner, G., and Kewitz, H. (1955). Naturwksenschaften 42, 628. Hayashi, T., and Nagai, K. (1956). Abstr. XX Intern. Physiol. Congr. Brussels p. 410. Hayashi, T., and Suhara, R. (1956). Abstr. X X Intern. Physiol. Congr. Brussels p. 410. Hebb, C. (1955). Quart. J . Ezptl. Physiol. 40, 176. Hebb, C. (1957a). Physiol. Revs. 37, 2. Hebb, C. (1957b). Acta Physiol. et Pharmacol. Neerl. 6, 621. Hebb, C. (1957~).Nature 180, 627. Hebb, C.,and Smallman, B. N. (1956). J . Physiol. (London) 134, 385. Hebb, C., and Waites, G. M. H. (1956). J. Physiol. (London) 132, 667. Hebb, C., and Whittaker, V. P. (1958). J . Physiol. (London) 142, 187. Henschler, D.(1956). 2. physiol. Chem. Hoppe-Seyler’s 305, 97. Henachler, D. (1957). 2. physiol. Chem. Hoppe-Seyler‘s 309, 276. Hogeboom, G. H., and K d , E. L. (1955). Federation Proc. 14, 633. Hosein, E.A. (1957a). Rev. can. biol. 16, 490. Hosein, E.A. (1957b). Rev. can. biol. 16, 491. Hosein, E.A., and McLennan, H. (1958). Personal communication from Dr. Hosein. Hutter, 0. F. (1957). Brit. med. Bull. 13, 176. Iwama, K., and Jasper, H. (1957). Rev. can. bwl. 16, 492. Kielley, W.W., and Bronk, J. R. (1958). J . Biol. Chem. 230, 521. K&, E.L., and Schneider, W. C. (1954). J. Biol. Chem. 206, 677. Kuntzman, R., Mead, J. A. R., Brodie, B. B., and Shore, P. A. (1958). J. Pharmacol. Exptl. Therap. 122, 41A. Leon, H. A., and Cook, 8.F. (1956). Science 124, 123. Lewis, P. R., and Hughes, A. F. W. (1957). I n “Metabolism of the Nervous System” (D. Richter ed.), p. 511. Pergamon Press, New York. Lindberg, O., and Ernster, L. (1954). “Handbuch der Photoplasmaforschung. Protoplasmatologia.” Band 111, A.4. Springer, Vienna, Lowry, 0. H. (1957). In “Metabolism of the Nervous System” (D. Richter, ed.), p. 323. Pergamon Press, New York. Lynen, F. (1957). In “Metabolism of the Nervous System” (D. Richter, ed.), p. 381. Pergamon Press, New York. McMurray, W. C., Maley, G. F., and Lardy, H. A. (1958). J . Biol. Chem. 230, 219. Marraezi, A. S., Hart, E. R., and Rodriguez, J. M. (1958). J. Pharmacol. Erptl. Therap. 122, 48A. Milton, A. S. (1968). J . Physiol. (London) 142, 25P. Mosinger, M. (1951). Compt. rend. 233, 982. Palade, G.E. (1966). J . Bwphys. Biochem. Cytol. 1, 59. Palay, 8. L. (1957). In Wltrastructure and Cellular Chemistry of Neural Tissue” (H. Waelsch, ed.), p. 31. Cassel & Co., London. Palay, S. L., and Palade, G. E. (1955). J. Biophys. Biochem. Cytol. I,69. Pope, A., and Hem, H. H. (1967). In “Metabolism of the Nervous System” (D. Richter, ed.), p. 72. Pergamon Press, New York. Porter, K. R., and Kallmann, F. L. (1952). Ann. N. Y . Acad. Sci. 54, 882. Purpura, D.P., and Grundfest, H. (1957). J. Neurophysiol, 20, 494.


193

Purpura, D. P., Girado, M., and Grundfest, H. (1957a). Science 125, 1200. Purpura, D. P., Girado, M., and Grundfest, H. (1957b).Proc. SOC.Exptl. Biol. Med. 95, 791. Quastel, J. H. (1957). In “Metabolism of the Nervous System” (D. Richter, ed.), p. 267. Pergamon Press, New York. Ramon y Cajal, S. (1909). “Histologie du systeme nerveux de l’homme et des vertebres,” 1st ed. Maloine, Paris. Richter, D., ed. (1957). “Metabolism of the Nervous System.” Pergamon Press, New York. Robertson, J. D. (1957). J . Phy9iol. (London) 137, 8. Rossiter, R. J. (1957). In “Metabolism of the Nervous System’’ (D. Richter, ed.), p. 355. Pergamon Press, New York. Rossiter, R. J., McLeod, I. M., and Strickland, K. B. (1957). Can. J . Biochem. Biophys. 35, 11. Scharrer, E., and Scharrer, B. (1954). Recent Prog. in Hormone Research lo, 183. Schneider, W. C., and Hogeboom, G. H.(1951). Cancer Research 11, 1. Schiimann, H.J. (1958a). Arch. exptl. Pathol. Pharmakol. Naunyn-Schmiedeberg’s 233, 237. Schumann, H.J. (1958b). Arch. exptl. Pathol. Pharmakol. Naunyn-Schmiedeberg’s 2a3, 296. Shanklin, W. M., Issidorides, M., and Nassar, T. K. (1957). J . Comp. Neurol. 107, 315. Sherrington, C . S. (1947). “The Integrative Action of the Nervous System,” 2nd ed. p. 101. Cambridge Univ. Press, London and New York. Sjostrand, F. S. (1956). Intern. Rev. Cytol. 5, 456. Smallman, B. N. (1957).Personal communication from Dr. Smallman. Tapley, D. F., and Cooper, C. (1956). Nature 178, 1119. Vignais, P. M.,Gallagher, C. H., and Zabin, I. (1968). J . Neurochem. 2, 283. von Euler, W. 8. (1957).In “Metabolism of the Nervous System” D. Richter (ed.), p. 543. Pergamon Press, New York. Waelsch, H. (1957). I n “Metabolism of the Nervous System” (D. Richter, ed.), p. 431. Pergamon Press, New York. Weiss, P. (1947). Abstr. XVZZ Intern. Physiol. Congr. Oxjord p. 101. Weiss, P.,and Hiscoe, H. B. (1948). J . Exptl. 2001.107, 315. Wethington, F. (1957). Rev. can. biol. IS, 639. Whittaker, V. P. (1956).Bwchim. et Biophys. Acta 22, 590. Wiersma, C. A. G. (1958).J. Physiol. (London) 142, 22P.


PARASYMPATHETlC NEUROHUMORS; POSSlBLE PRECURSORS AND EFFECT ON BEHAVIOR' By Carl C. Pfeiffer Division of Basic Health Sciences, Emory University, Atlanta, Georgia

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .................. 11. The Role of Acetylcholine in Brain Function ........................ 111. Effect of Autonomic Agents on the Conditioned Avoidance Response (CAR) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...................... IV. Effect of Tranquilizing Drugs on Conditione voidance Response . . . V. Effect of Tertiary Amine Parasympathetic Stimulants on the CAR . . VI. Effect of Arecoline in Schizophrenic Patients ........................ VII. Failure of Acetylcholine or Methacholine to Inhibit the CAR

B. Biochemical Studies . . . . . . . .

........................

196 197 199 200 201 202

207

........... XII. Choice of a Deanol Salt for Clinical Trial . . . XIII. Possible Modes of Action of Deanol . . . . . . . . XIV. Deanol in Clinical Disorders A. Chronic Fatigue States .......................... . . . . . . . . . . 223 B. Periodic Headaches . . . . . . . . . . ........................ 224 .................. G. Miscellaneous . . . . . . . . . . . . . . . . . . . . . . . . . . .

.................

XVII. Critique of Acetylcholine as the Parasympathetic Neurohumor . . . . . . XVIII. The SAR of Muscarinic geners ................................ ............... XIX. Congeners of Deanol Designed to Increase Effect . . . . . . . . M. A Theory for the Mode of Action of Benac ........................ XXI. Summary . . . . . . . . . . . . . . References ...............................................

231 233

a38 240 241

'This study waa supported in part by Grant M-876from the National Institute of Mental Health, National Institutes of Health, U. S. Public Health Service, and grants from the Geschickter Fund for Medical Research, Inc. and Riker Labore tories, Inc. 196

196

CARL C. PFEIFFER

I. Introduction

This review is written from the viewpoint of a clinical pharmacologist who has maintained an intense interest in the structure-activity-relationships (SAR) of drugs and has personally treated or supervised the treatment of many schizophrenic patients since 1937. SAR is the abbreviation of structure-activity-relationship coined by S. Loewe (1950 ; Loewe and Adams, 1947) to facilitate communication in this field which deals with the relationship of chemical structure to pharmacological action. In the opinion of the reviewer, the term originated earlier by C. D. Leake, “biochemomorphology,” was as good a term for the subject 8s “SAR,” but was needlessly complex, and, hence, over a period of years has not gained common usage. Schueler (1950) suggests “chemobiodynamics’’ as a broader term than “biochemomorphology” but the reviewer prefers SAR. Both biochemicals, new drugs, and older drugs have been tested for their effect in schiaophrenia. The author’s concept of the schizophrenias is that these occur as a group of diseases of varying biochemical etiologies and, therefore, any correlation of improvement or deepening of the schizophrenic state should be vigorously explored to ascertain the possibility that biochemical subclasses of the schizophrenias may be definitely established. At present, the diagnosis is based on the patient’s history, chronicity of the disorder, behavior, motor patterns, and verbal exhaust. A clinical syndrome which in the past has been mimicked by vitamin deficiency, intoxication and infection may, therefore, in the future be proven to have still other subdivisions and biochemical etiologies. Various model hypotheses will be explored, elaborated, and used without attempt to document these hypotheses in all instances by laboratory data in the experimental animal. These “giant steps” are of course inexcusable from the standpoint of exact science, but tolerable from the following viewpoints: (a) schizophrenia has not been produced or diagnosed in the experimental animal, and (b) biological science is so enormously complex that any model hypothesis which allows the investigator to step forward more boldly may be productive as long as the investigator remembers that the model hypothesis is only an educated guess. Ordinarily reviewers are content to state the span of years which their opus covers with a benign footnote “See earlier reviews for adequate coverage of the entire field.” The history of the concept of acetylcholine-like substances is not so ancient but that a critical reviewer can partially probe the depths to suggest rearrangement of the jig-saw

PARASYMPATHETIC NEUROHUMORS AND BEHAVIOR

197

puzzle and resurrect for scrutiny those findings which are ordinarily omitted from textbooks because they do not fit into the present orderly scheme. Such an attempt will be made for parasympathetic neurotransmitters in the final portion of this paper,

II. The Role of Acetylcholine in Brain Function

The presence of acetylcholine (ACh) , its esterase, or its acetylase in the brain has been adequately documented by in vitro bio-assay methods by numerous investigators (see reviews by Feldberg, 1945 and Hebb, 1957). The studies thus far leave much t o be desired in regard to human brain function and any possible role of ACh. Thus, in animals and also man the biologically assayed ACh-like substance appears to be low in cortex, cerebellum, and white matter and high in the basal nuclei of the rhinencephalon. The caudate, amygdaloid hippocampus and hypothalamus appear to have the highest content of ACh. These areas also contain the highest levels of 5-hydroxytryptamine (5-HT), epinephrine (Epi) , and norepinephrine (Nor-epi) . Feldberg’s scholarly and exhaustive review (1945) of the mode of action of acetylcholine in the central nervous system (CNS) provides a superb background in this subject. Feldberg draws no conclusions but does suggest, that since the acetylcholine content in the cortex of animals decreases with phylogenetic development, that acetylcholine activity in man and higher mammals may be more primitive and limited to certain central nuclei or synapses. He notes that the evidence for a role of acetylcholine as a neurohumoral agent is as good (in 1945) for the central nervous system as for the peripheral areas of the body, such as ganglia and neuroeffector cells. Richter and Crossland (1949) find the acetylcholine content of young rats killed by immersion in liquid air to vary with the state of activity of the nervous system in microgams per gram as follows: light pentobarbital anesthesia, 1.76; sleep, 1.44; normal, 1.25 ; excited, 0.07; electrically stimulated, 0.55; and convulsing, 0.56. These changes were statistically significant and confirmed the findings of Tobias and coworkers (1946). The return to normal acetylcholine levels after convulsant activity requires only 100 seconds according to Richter. Elliott et al. (1950) also find that pentobarbital, ether, or chloralose anesthesia increases the acetylcholine content of rat and cat brain. Picrotoxin and pentylenetetrazol lower the acetylcholine content of the anesthetized cat but not the normal cat. Because of our findings that neither methacholine nor acetylcholine

198

CARL C. PFEIFFER

pass to the brain to inhibit the conditioned avoidance response (CAR), we will not review the animal experiments wherein central effects were obtained from intravenous or intraarterial injection of acetylcholine. These effects were probably either hypotensive, anoxic, or peripherally mediated reflex effects. Numerous workers have injected acetylcholine into the cerebrospinal fluid of man and animals. In the experiments by Dikshit (1934), Silver and Morton (1936),and Henderson and Wilson (1936),sleep rather than stimulation was the predominating effect. In more recent experiments, Sherwood (1952) found catatonia and agitation in cats injected intraventricularly with 0.5 to 2.0 mg of acetylcholine, 3 to 7 mg of methacholine, or 0.05 to 0.1 mg of carbachol. Feldberg and Sherwood (1954) report a similar syndrome with diisopropyl fluorophosphate (DFP) or eserine injections. Sherwood further found that lesions on the border between the upper tegmentum and the posterior hypothalamus, which were subliminal insofar as catatonia was concerned, made the cat much more sensitive to acetylcholine-induced catatonia. In interpreting these data, the reader should be cognizant of the possibility that acetylcholine applied to the cell surface of neurones may have limited access to the site of normal action and therefore an opposite effect may occur such as sleep rather than CNS stimulation. Thus, deano12 produces a slow developing CNS stimulation in both man and animals. As a further example, methacholine mist, if inhaled, precipitates asthmatic symptoms, whereas deanol taken orally does not; several asthmatic patients have been benefited by this therapy and relapse when the deanol is substituted by placebo therapy. Grob et al. (1947) found that when DFP was used in the treatment, of myasthenic patients, they, after a time, refused the medication because of the nightmares, mental confusion, and hallucinations. Rowntree st al. (1950)injected schieophrenic and manic patients and normals with DFP to ascertain the effect of increased levels of acetylcholine in these disorders. The schizophrenics showed pallor and nicotinic tremors and were uniformly made worse by the treatment, whereas the manic patients were somewhat improved. This therapeutic failure in schizophrenics of higher levels of cerebral acetylocholine, which may have been obtained by the inactivation of cholinesterase, can probably be ascribed to an equal affinity of DFP for the muscarinic receptors where the excess acetylcholine would have a drug effect. Dogs chronically poisoned with DFP show, predominantly, a constant nicotinic tremor which may seriously interfere with their food and water intake. This might be expected if the DFP acted not only on the cholinesterase but Generic name for 2-dimethylaminoethanol “Deaner,” Riker Laboratories.


199

also on the muscarinic receptors. The nicotinic receptors were not inhibited and the excess ACh would then produce predominantly nicotinic effects. Cohen and associates (1944) have reviewed the intravenous use of acetylcholine in the treatment of schizophrenia. The severe cardiovascular effects of this therapy would indicate that the anoxia of temporary cardiac standstill probably accounted for most of the benefit. Since Cohen et al. report cardiac standstill for periods of 20 to 50 seconds the permeability of the blood-brain barrier may be influenced by the anoxia so that more of the choline administered in the 400- to 600-mg dose of acetylcholine may pass to the brain. Phillips and Hutchinson (1954) treated 211 psychoneurotic patients with intravenous acetylcholine, of which 104 compulsive neurotic patients showed an improvement rate of 70%. They consider the treatment to be safe. Birkhauser (1941) finds no difference in the cholinesterase content of the brains of normals and schizophrenics. He reports, however, an increased level of monoaminoxidase in the globus pallidus of schizophrenics. Hebb and Smallman (1957) report that 50-70% of brain choline acetylase is located in or on the mitochondria1 fraction. Cerebral acetylcholine, if synthesized from choline, might, therefore, be limited in manufacture because of the several membranal barriers which this molecule, with its charged nitrogen atom, must traverse. Williams (1955) has shown that liver mitochondria contain choline oxidase which is relatively inactive when surrounded by a substrate of choline. When the mitochondrial membranes are ruptured then the oxidation of choline proceeds rapidly. According to Nachmansohn and Wilson (1951) neither the liver nor the kidney contain discernible amounts of choline acetylase.

111. Effect of Autonomic Agents on the Conditioned Avoidance Response (CAR)

Gantt and Freile (1944)report that both epinephrine and ACh affect the responses of the conditioned dog. In the neurotic animals epinephrine is less disturbing than in the non-neurotic animals. Acetylcholine improves the performance of the neurotic dogs and improves the differentiation between excitatory and inhibitory processes in the non-neurotic animals. Funderburk and Case (1947) find in the conditioned cat that eserine (0.25 mg per kilogram), but not pilocarpine or neostigmine, will inhibit the CAR. They further find that atropine or magnesium ion will corn-

200

CARL C. PFEIFFER

pletely prevent this effect of eserine. They ascribe the inhibition of the CAR to the cerebral acetylcholine which accumulates after eserine inhibits brain cholinesterase. Gellhorn (1953) finds that treatment with atropine or thyroxine tends t o restore the CAR when it has been inhibited by insulin or electrically induced convulsions. Jacobsen et al. (Jacobsen and Sonne, 1955; Jacobsen and Skaarup, 1955) show that the CAR is facilitated by the synthetic atropine-like compound benactyzine. Kosman and Gerard (1955) report that large doses of epinephrine inhibit the CAR. Slater and Jones (1958) find that Butamoxane, a compound modeled after the adrenergic blocking drug piperoxan, will inhibit the CAR and produce other typical pharmacological effects usually found with chlorpromasine or reserpine. Cook and Weidley (1957) report that LSD-25 (1.5 mg per kilogram) or 5-HT (10 mg per kilogram) will inhibit the CAR.

IV. Effect of Tranquilizing Drugs on the Conditioned Avoidance Response

Numerous investigators have studied the inhibition by reserpine and chlorpromazine of various types of conditioned avoidance responses in animals (Guha et al., 1954; Cook et al., 1955; Smith e t al., 1957; Pfeiffer and Jenney, 1957). This inhibition occurs a t a dosage that is a small fraction of the lethal dose of chlorpromasine. This inhibitory effect is evidenced by reserpine, 11-demethoxyreserpine, and rescinnamine, all of which are active tranquiliring drugs. Azacyclonol (Frenquel) and meprobamate are completely inactive (Pfeiffer et al., 1957b). 5-Hydroxytryptamine (5-HT) in doses of 25 mg per kilogram produces marked symptoms in the rat, but it does not inhibit the CAR. This is in contrast to the findings of Cook and Weidley. . Since many of the pharmacological data in animals and clinical studies in patients show that the active alkaloids of Rauwolfia, and, to a lesser extent, chlorpromasine, have a persistent acetylcholine-like effect, we chose to investigate the effect on the CAR of various muscarinic-like drugs that will pass the blood-brain barrier freely. The tertiary amines arecoline, pilocarpine, and eserine were, therefore, studied. Atropine, a tertiary amine, passes the blood-brain barrier freely and would protect the animals from the peripheral effects of arecoline, eserine, and pilocarpine, but it cannot be used since it would also protect the brain from the muscarinic effects of these drugs. Atropine methylnitrate (Eumydrin) is a quaternary nitrogen analog of atropine that probably does not pass the blood-brain barrier. Methyl atropine, 10 to 20 mg per kilogram intraperitoneally, was, therefore, used to protect the


201

rats from the peripheral effects of the parasympathetic stimulants while the brain was unprotected, and we could, therefore, observe the effect of cerebral muscarinic stimulation in animals that were not salivating, defecating, or anoxic from pulmonary edema.

V. Effect of Tertiary Amine Parasympathetic Stimulants on the CAR

Arecoline, in a dose of 2 mg per kilogram subcutaneously, specifically inhibits the CAR down to 5% of normal performance within 6 minutes after dosing. Complete recovery occurs in 20 minutes. Atropine (3 mg per kilogram) will completely protect the rat from the effect of arecoline. Eserine (physostigmine), in a dose of 0.25 mg per kilogram subcutaneously, inhibits the CAR to 30% of normal performance in the 20 to 40 minute period, and complete protection from the eserine effect is provided by pretreatment with atropine. Pilocarpine, in an average dose of 9 mg per kilogram subcutaneously, produces a reduction of the CAR to 3576 of normal performance a t the 20 minute interval after dosing. This effect is almost completely prevented by pretreatment with atropine. Theoretically, atropine should modify the known inhibition of the CAR that is produced by small doses of chlorpromazine or reserpine. This antagonism, however, must depend on the selective affinity of the two competing blocking drugs for the receptors in the CNS. We know that the pharmacological effect of both chlorpromazine and reserpine lasts longer than does the effect of atropine. This could mean a greater afbity for receptors. We were, thus, not too disappointed when we could not demonstrate any statistically significant effect of atropine (10 mg per kilogram intraperitoneally) on the inhibition of the CAR produced by reserpine (0.75 mg per kilogram subcutaneously). Similarly, we could not demonstrate any statistically significant effect of intraperitoneal atropine (10 mg per kilogram) on the inhibition of the CAR produced by a 4-mg per kilogram dose of chlorpromazine given subcutaneously. Cook and Weidley (1957) report that LSD-25 (0.1 mg per kilogram) antidotes the inhibitory effect of 5-HT, reserpine, and chlorpromazine, but not morphine, on the CAR. Jenney and H d y (1958) do not find any antidotal effect of LSD-25 (0.5 mg per kilogram), mescaline (10 mg per kilogram), or epinephrine (2.0 mg per kilogram) on the typical inhibition of the CAR produced by a subcutaneous injection of arecoline (2 mg per kilogram) in the methyl atropinized rat. As stated before, atropine sulfate, 3 mg per kilogram given intraperitoneally, does antidote the arecoline effect completely.

202

CAI& C. PFEIFFER

VI. Effect of Arecoline in Schizophrenic Patients

Since these three drugs (arecoline, pilocarpine, and eserine) inhibit the CAR in a manner similar to chlorpromazine and reserpine, cautious clinical trial wa8 started in schizophrenic patients who were protected from the peripheral effects by means of methyl atropine in a subcutaneous dose of 1 to 3 mg. Fulcher et al. (1957) report that arecoline (15 mg subcutaneously) produces a “lucid interval” comparable to that produced by the inhalation of 30% carbon dioxide or the intravenous injection of amobarbital. Eserine, in a dose of 5 mg subcutaneously, produces a similar alerting effect but this is too slow to be convincingly dramatic. Pilocarpine was used in subcutaneous dosage of 20 mg without seeing any effect in the single schizophrenic patient in which it was tried.

VII. Failure of Acetylcholine or Methacholine to Inhibit the CAR

A subcutaneous dose of methacholine (0.5 mg per kilogram) will produce salivation, lacrimation, and defecation in the normal rat. After methyl atropine (10 mg per kilogram), a 100-fold larger dose of methacholine (50 mg per kilogram) can be given without any evident drug effect in the rat. This large dose is also without effect on the CAR. The rats show none of the tremor or hyperreflexia which is characteristically produced by the tertiary amine parasympathetic stimulants. Similarly, rats protected with methyl atropine are given neostigmine (0.1 mg per kilogram) to prevent hydrolysis of ACh and then given ACh (10 to 25 mg per kilogram). This combined treatment with three quaternary nitrogen containing drugs has no inhibitory effect on the CAR. The rats do show salivation, muscle fasciculations, and a nicotinic tremor which differs from that produced by arecoline or eserine and is probably a peripheral rather than a central effect of either ACh or neostigmine.

VIII. Limited Passage of Quaternary Amines across the Blood-Brain Barrier

Crum-Brown and Fraser (1869) made methyl iodide derivatives of many of the tertiary amine alkaloids. They discovered that these quaternary nitrogen derivatives no longer produced central effects but produced peripheral actions characterized by curare-like paralysis and ganglionic blockade. Numerous workers since then have determined that quaternary nitrogen analogs do not easily pass the blood-brain barrier

PABASYMPATHETIC NEUROHUMOW AND BEHAVIOR

203

and exert only a peripheral pharmacological action. Thus, tetraethylammonium is convulsant when applied to the exposed cortex of the brain, but Gellhorn et al. (1952) finds that ganglionic paralyzing doses have no effect on central transmission of nerve impulses. Similar findings apply to other quaternary nitrogen ganglionic blocking compounds and to d-tubocurarine. Koelle and Steiner (1956) find that an effective peripheral anticholinesterase which is a thiocholine phosphate ester has no effect on brain cholinesterase while the tertiary amine derivative inhibits both peripheral and cerebral cholinesterase. Mayer and Bain (1956) have studied the distribution of a fluorescent convulsant compound which, in its tertiary form, produces convulsions and stains the nucleus and nucleolus of selected neurones. The quaternary nitrogen derivative stains only the endothelial cells associated with the blood-brain barrier and is not convulsant. Choline has usually been considered as the precursor of ACh and no serious consideration has hitherto been given to the transport of choline across the various membranal barriers to the mitochondria which contain 70% of the cerebral choline acetylase. Therefore, possible tertiary amine precursors of choline, including deanol, have been studied extensively in these laboratories.

IX. Literature Survey on Deanol (2-Dimethylaminoethanol)

Deanol is an old and well-known compound. The reported studies may be most conveniently divided into (a) nutritional, (b) biochemical, and (c) pharmacological. Some of the nutritional and pharmacological studies are in the nature of chronic and acute toxicity studies which will help substantiate the premise that deanol is a nontoxic compound with vitamin properties similar to those of choline and may occur naturally in the body. Choline Chloride CHa\ + CH1-N-CHpCHsOH

Deanol

~~>N-cH,--cH,oH

c&/ c1 A. NUTBITIONAL STUDIES Various nutritional studies compare deanol to choline as a dietary factory. A thorough search of the literature discloses no assays of foods for their deanol content. The highest level of choline is found in fish eggs. Guggenheim (1951) summarizes the distribution of free choline as follows: the blood of man, dog, pig, rat, and mouse contains 0.06 to 0.1

204

CARL C. PFEIFFER

mg%. (The menstrual period increases the level in blood 2- to 15-fold and the concentration in sweat 30- to 100-fold.) The guinea pig has levels in blood of 2-12 mg%, brain 2-3.5 mg%, heart muscle 5 mg%, spleen 2-25 mg%, liver 5-20 mg%, stomach 1 0 3 0 mg%, testes 40 mg%, seminal vesicle fluid 70 mg%, and viscera 1-12 mg%. The level of free choline in trout eggs at 70 mg% equals that of seminal vesicle fluid. In normal human pregnancy, the free choline blood level falls from 0.19 to 0.14 mg% while in toxemia of pregnancy the level is reduced to 0.12 to 0.09 mg%. The serum choline level of women is reported to be 5-fold greater in summer than in winter. Nutritionists report that in some species, such as the rat, aminoethano1 (AE) (of dietary origin or derived from serine) can be methylated by methyl donors (such as ckoline or methionine) to monomethylaminoethanol (MAE), dimethylaminoethanol (deanol) , and choline (Guggenheim, 1951). It is improbable that deanol can be obtained from the demethylation of choline. The end product of methyl donation, dimethylglycine, is derived from choline by the demethylation of the oxidation product, betaine (Dubnoff, 1949 a, b; Muntz, 1950; Pilgeram et al., 1953) (see Section B). The question, then, arises as to how well these various intermediates and particularly deanol can substitute for choline as a dietary factor. Jukes and Dornbush (1945) find that a Neurospora “choline-less” mutant which cannot utilize betaine, AE, or methionine can grow as well on deanol as on choline. Noland and Bauman (1949) find that neither deanol, AE, nor methionine will substitute for choline in the diet of the German cockroach. Betaine, however, does substitute for choline. In chicken nutritional studies Jukes and Oleson (1945) find that 0.2% deanol will substitute for 0.1% choline as a dietary factor, but the 28 day weight gain is 145 gm per chick compared to 194 gm with choline and 112 gm in the choline deficient chick (see also Jukes et al., 1945). Shaefer and co-workers (1951) find that chicks fed diets supplemented with vitamin Bla and folic acid are unable to use betaine, methionine, or AE as precursors of choline. When deanol is used as a replacement for choline, perosis is prevented and growth is almost equivalent to choline controls. Demers and Bernard (1950), using a purified diet deficient in choline (18% casein) in ducklings, show that betaine and AE are without effect, while deanol, like choline, stimulates growth and prevents perosis. Both the chick and the duck do not appear to be able to methylate AE to any extent. Johnson et al. (1955) find that a soy protein diet containing a total of 0.8% methionine, 2% serine, and 2.2% glycine will completely protect baby pigs from choline defioiency in the presenoe of adequate amounts

205


of vitamin B12. Choline deficiency results in the absence of vitamin B12 or in the presence of vitamin BIZ with the glycine content reduced to 1.6%. Vitamin B12 is required to synthesize methyl groups from glycine. Deanol has a sparing action on methyl synthesis while AE greatly increases the methyl group demand, causing choline deficiency even in the presence of vitamin B12 and glycine a t the 2.2% dietary level. Reid (1955) placed numerous groups of 2- to 3-day old guinea pigs on a choline-deficient diet which contained 30% vitamin-free casein, 7.3% fat, four types of carbohydrates, 6% salt mixture, and liberal amounts of the known vitamins except choline. The guinea pigs were observed and weighed for a period of 6 weeks and the survivors were then sacrificed, autopsied, the organs weighed, and the tissues examined. These studies provide excellent 6-week chronic toxicity studies in the growing guinea pig. The data are interpreted as follows: assuming that the guinea pig eats 10 gm per 100 gm body weight, then 1.91 gm of deanol in the diet would represent a dose of 19.1 mg per 100 gm of guinea pig or approximately 190 mg deanol per kilogram body weight per day. TABLE I

GUINEAPIG6 Choline in diet (gm/kg) 0.1 2.0 0.0 0.0 0.0 0.0 0.0 0.0

0.0

ON

CHOLINE-DEFICIENT DiET

Supplement in diet (gm/kg) None None Deanol 1.91 Methionine 6.54 Betaine 1.68 Aminoethanol 1.35 N-Methylaminoethana113.22 Dimethylglycine 2.21 Deano10.55

Body Weight Survival 6/24 56/56 8/8

1/8 4/8

3/8 2/8 7/8 8/8

0 wk. 2 wk. 4 wk. 6 wk.

103 103 103 103 103 103 103 103 103

124 155 151 121 134 133 140 129 142

136 242 253 140 160 144 150 153 228

166 322 337 152 180 184

-

186 310

Deanol proves to be an excellent substitute for choline. With 0.55 gm per kilogram of diet growth is similar to that obtained with larger amounts of choline chloride. With 1.91 gm per kilogram, which is the molar methyl equivalent to 2.0 gm per kilogram of choline chloride in the diet, growth is equal to that obtained with choline chloride. It also appears to be equal in value to choline chloride when used to supplement diets partially deficient in choline. Similarity in the effect of deanol and choline is seen at autopsy in the appearance and weights of selected organs.

206

CABL C. PFEIFFEB

TABLE I1 EFFECTOF DXFFERBNT COMPOUNDS ON ORGANWEIQHTBQ Compound Controle Choline chloride Aminoethanol Aminoethanol (N-methyl)

Deanol

Adrenal mg/100 gm

Heart mg/lOO gm

Diet gm/kg

Liver gm/100 gm

0.0 2.0 0.75

4.0 f 0.8

6.9 f 0.3 6.6 f 0.6

1.3 f 0.2 0.9 0.06 1.0 0.06

123 f 14 48 f 1.2 102 f 9

406 f 52 317 f 14 392 f 8

3.22 1.91

8.7 f 1.6 6.8 f 0.6

1.7 f 0.2 0.98 f 0.06

92 f 60 f

571 f 72 354 f 27

Kidneys gm/100 gm

+

+

4 2.6

0 Eight guinea pige in each group on diet 6 weeks. Weight gain is in grams per 100 gm body weight.

Reid thus shows that deanol will substitute for choline as a growth factor and vitamin for guinea pigs on a choline deficient diet. Betaine, AE, and MAE will not. In contrast to deanol, MAE is toxic as shown by survival of only 2 out of 8 guinea pigs, and a significant increase in liver, kidney, and heart weights at autopsy. Choline deficiency in guinea pigs characteristically produces an anemia, with concomitant decrease in hemoglobin, and muscular weakness; but infiltration of the liver or hemorrhagic kidneys, such as occur in the rat, are seldom seen. Much of our textbook material on choline and its precursors is based on rat nutritional studies rather than on chicken or guinea pig studies. The albino rat, in contrast to the guinea pig and chick, is able to methylate AE to deanol and choline and seems to thrive on a choline-deficient diet if the casein level is above 5%. This probably reflects the methionine contents of casein. Apparently, soybean protein contains less methionine and is the protein of choice for the production of “choline deficiency” in the rat. Numerous authors (Sinclair, 1930; Sure, 1940; Fries et al., 1940; and Artom and Fishman, 1943) find that the choline requirement of suckling rats is high because of the great Iieed for phospholipid synthesis in the immature rat. If the mother is placed on a choline-poor diet, the offspring stop growing in 13 days and die in several days of a typical paralysis. This may be corrected by giving 8 supplement of 15 mg of choline per day to the mother or offspring. Deanol has not been studied in choline-deficient lactating animals. Hove et al. (1954) find that the rabbit requires 0.13% of dietary choline for normal growth. Methionine is of some value as a substitute for choline. MAE plus methionine is ineffective. Vitamin B12 and folic acid are adjuvants with MAE but not with AE. Symptoms of choline deficiency in the rabbit are anemia, an increased icteric index, cirrhosis

207


of the liver, and urinary excretion of porphobilinogen. Hove et a2. did not study deanol. Summary: Nutritional studies. Choline is an essential vitamin which occurs in many foods and is needed for phospholipid synthesis and the synthesis of acetylcholine. Animals on a choline-deficient diet can utilize deanol and grow normally. Deanol has not as yet been demonstrated to occur in natural foods. Choline cannot be a precursor of deanol since, in donating its methyl group, choline is first oxidized to betaine which in turn donates the methyl group and leaves dimethylglycine. Neither betaine nor dimethylglycine will substitute for choline as a growth factor for animals on a choline-deficient diet. MAE will not substitute for choline and is chronically toxic. Deanol has a sparing action on methyl synthesis while AE greatly increases the methyl group demand. Only the rat can methylate AE to choline; the guinea pig, rabbit, pig, and fowl apparently cannot perform this methylation.

B. BIOCHEMICAL STUDIES I n contrast to the nutritional studies, the biochemical studies on deanol utilize individual animals for specific tests of labeled compounds or use tissues in vitro to study biochemical reactions. Again, the albino rat which is apparently unique in its ability to methylate AE to choline has been the most widely used species. Pilgeram (1955) states that those animals, such as the rat, which can methylate AE to choline and phosphatidyl choline are not susceptible to experimental atherosclerosis, whereas the monkey, rabbit, guinea pig, and fowl, species which deal poorly with AE, are easily susceptible to experimental atherosclerosis. Deanol is primarily a precursor of choline and not a product of its demethylation. Dubnoff (1949a, b) , Muntz (1950), and Pilgeram et al. (1953) show the following reactions in methyl donation by choline. (cH~)&-cH?-cH,-oH

Choline

02 + + (CH~)~=N-CH,-C=O

-+

\4

Betaine

(CHr)-N-CH*-C=O

+ -CHI

AH Dimethylglycine

Dubnoff (1949a, b) finds that no methionine is formed when choline and homocysteine are incubated with livers of the rabbit, guinea pig, or chick which contain little or no choline dehydrogenase. Muntz, using rat

208

CARL C. PFEIFFER

liver homogenates incubated with homocysteine and P - l a b e l e d choline, shows a high concentration of N15 in the dimethylglycine fraction, whereas none is found in the deanol. Artom and Crowder (1950) also fail to find any activity in the deanol fraction when liver slices are incubated with C14-methyl-labeled choline in the presence of deanol as a carrier. In further studies, Artom and Crowder (1949) find that, after the administration of a larger dose of deanol to rats, detectable quantities are found in the lipids as well as in the aqueous extracts of muscle and liver. I n contrast, after single large doses of choline, no deanol can be detected in tissues (see Table 111). TABLE 111 DEANOL IN AQUEOUS AND LIPIDEXTBACTS OF LIVERSOF RATSON CHOLINE-DEFICIENT DIETO ~~

A

~

Micromoles deanol in liver Substance

Route

Water Deanol Deanol Deanol Deanol Choline Choline Choline

Oral Oral Oral 1.P.c I.P. Oral Oral I.P.

(I

Hours aft.er dosageb Aqueous extract 6 3 6 3 6 3 6 6

Lipid extract

0 47

0 20

35 110 25

70 29 90

0 0 0

0 0 0

Artom (1962).

* All doses 1-2 millimoles per rat.

c

I.P. = Intraperitoneal.

These data indicate that deanol is redistributed from the aqueous phase to the lipid phase in the 3- to 6-hour period. Artom and Crowder are unable to extract deanol from the phospholipids of rat tissues under various conditions such as restriction of methyl donors in the diet. Artom concludes that under physiological conditions deanol is formed in tissues but is probably methylated to choline very rapidly. Levine and Chargsff (1951) also fail to detect deanol in their analyses of nitrogenous components of phospholipids by paper chromatography. When deanol is compared to choline for its effect on the oxygen consumption of liver slices, an increase is found with both compounds but the effect of choline is twice that of deanol. With liver homogenates, deanol does not increase oxygen consumption while choline still produces 50% of the oxygen consumption seen with liver slices. Wells (1954) finds


209

that deanol inhibits choline oxidase so that choline oxidation is reduced to 10 to 20% of the control values. Deanol is more active in this regard than is a,a-dimethyltriethylcholine, or 2-amino-2-methyl-1-propanol. Williams (1955) shows that choline oxidase is contained in the mitochondria of the liver cell. When the mitochondria1 wall is intact the mitochondria oxidize choline poorly, presumably because of limited transport of choline through the membrane. Disrupted mitochondria oxidize choline rapidly. The observations by Fishman and Artom (1944), that the level of lecithin in the liver and other tissues of young rats is lower when the animals receive a choline-low diet from weaning, provides indirect evidence for the role of choline in lecithin synthesis. The importance of choline and deanol in the synthesis of phospholipids is shown by their rate of turnover. When choline is fed, the half-life of phosphatide choline is found by Boxer and Stetten (1944) to be 6 days (daily replacement of choline = 3.9 mg per rat per day). When no choline is fed the choline is conserved so that the half-life becomes 18 days (daily replacement of choline = 1.3 mg per rat per day). P32tracer studies also indicate that choline and deanol play a role in phospholipid synthesis. Perlman and Chaikoff (1939) find an increased formation and turnover of phospholipids in the liver of rats fed choline as compared to the controls. The increase in phospholipid is proportional t o the amount of choline fed. Deanol has been found by Artom e t al. (1949) to be about 75% more effective than choline in incorporating PSz into the phospholipids of the livers of rats on 5% casein diet. D u Vigneaud e t al. (1946) gave rats 50 mg per day of deuterium-labeled MAE and deanol. They find MAE a t this dosage is lethal but deanol is well tolerated by rats on a diet which includes homocysteine. Deanol prevents fatty livers and hemorrhagic kidneys but growth is not equal to that of the choline rats. They find 43% of the labeled deanol methyl groups incorporated in choline in 3 weeks. Thus, deanol is slowly converted t o choline in the rat. Steensholt (1949) finds that d-methionine is more active than Zmethionine in methylating AE or deanol to choline in rat liver slices. Stekol e t al. (1955) find that choline is not synthesized by the addition to ethanolamine of 3-methyl groups (methionine) but by the transfer of a methyl group from methionine to deanol as the direct acceptor. This process does not require folic acid as a cofactor although folic acid is required for the synthesis of deanol from AE. Stekol et al. (1956) use liver slices from folk acid deficient rats to show that choline formation is restored to normal by the addition of either deanol or citrovorum factor, but not by the addition of AE or MAE. The transfer of the methyl group from methionine does not involve folic acid, but is substrate specific for

210

CAaL 0. PFEIFFEX

deanol. In tumor tissue, deanol is the limiting factor in the synthesis of choline. Stekol et al. (1958) further show that activated methionine (Sadenosyl-1-methionine) is inactive in vivo but manyfold more active in vitro than methionine when slices and homogenates are employed. Transport of S-adenosyl-1-methionine in. vivo is questioned, whereas l-methionine in vivo is probably synthesized to activated methionine after transport. Korey and associates (1950) studied the acetylation of deanol in vitro in the presence of choline acetylase from the squid ganglia. The rate of acetylation is equivalent to that of choline but the acetate ester formed is much less active biologically than acetylcholine. Summary :Biochemical studies. The most important known biochemical functions of choline and deanol are their lipotropic actions which are evidenced by their prevention of fatty liver, incorporation into phospholipids such as lecithin, and their prevention of hemorrhagic kidneys. Choline and deanol are of importance in the kidney development of young animals, in the development of normal lactation, in the prevention of perosis or slipped tendons in fowl, and finally in the synthesis of acetylcholine. The requirements of choline are greatest in young animals, presumably because of the great need for phospholipid synthesis. Atherosclerosis may be more common in animal species which cannot methylate aminoethanol to choline. Biochemical studies show that deanol is a precursor of choline rather than a degradation product. Methyl donation by choline depends on choline oxidase which converts choline to betaine and dimethylglycine plus a donated methyl group. Deanol competes for the en~ymecholine oxidase and thus inhibits the oxidation of choline. Choline passes the mitochondria1 membrane poorly. Numerous biochemists find that deanol is methylated to choline in the body. Both choline and deanol go to phospholipids but deanol causes a more rapid incorporation of Psa into lipid fractions. Deanol is acetylated by choline acetylase as efficiently as choline. STUDIES C. PHARMACOLOGICAL The first recorded pharmacological study on MAE is that of Dakin (1905) who reports 10 mg per kilogram intravenously in the rabbit to be without effect on the blood pressure. Hunt and Taveau (1911) apparently first studied deanol and report that 3 mg per kilogram is devoid of muscarinic or nicotinic action on the blood pressure of the curarized cat anesthetized with ether. Fuchs (1938), however, finds both MAE and deanol to be hypotensive in the dog in doses of 10 to 20 mg per kilogram. Stehle et al. (1936) studied the effects of AE, MAE, deanol, and choline on the blood pressure of the amobarbitalified cat in doses of 10


21 1

mg per kilogram. Both AE and MAE produce a slight pressor effect while deanol is depressor. Choline shows greater depressor effect but this depressor effect is followed immediately by a pressor component. Acetylation greatly augments the depressor effect of each analog. Of the acetate esters tried, only acetylcholine has a nicotinic pressor action after atropinization. The Bovets (Bovet and Bovet-Nitti, 1947) report on the comparative hypotensive effect of nitrate esters of AE, MAE, deanol, choline, and other derivatives. Their studies on the four analogs administered intravenously in the chloralosed dog are shown in Table IV. TABLE IV COMPARATIVE HYPOTENSWE EFFECT OF NITRATE ESTERS Compound

AE nitrate MAE nitrate Deanol nitrate Choline nitrate

Minimal hypotensive dose (mg/kg)

Relative vasodilation

0.1 1.o 0.6 0.01

1.0 0.01 0.02 100

The hypotensive action of choline nitrate is completely prevented by 2-mg per kilogram dose of atropine sulfate, whereas the other three compounds are still hypotensive after this dose of atropine. Burkhalter et al. (1957) report that the acetate ester of deanol or the corresponding p-methyl derivative are as effective as acetylcholine in preventing the usual decrease (Jones et al., 1956) in cholinesterase activity in tissue cultures of chick embryo intestinal smooth muscle cells. Rate of hydrolysis is not a determining factor since deanol acetate or the corresponding p-methyl analog are hydrolyzed enzymatically a t one-half the rate of acetylcholine. Choline or the reversed carboxy analog of acetylcholine are ineffective. Solandt and Best (1939) studied vagus inhibitory action in rats on a choline-poor diet. A standard stimulus reduces heart rate 25% in the choline-deficient rat and 70% in normal rats. When the deficient rats are fed choline, the inhibition increases from 25% to 55%. Kraata and Gruber (1949) find no difference in spontaneous movement of uteri from normal and choline deficient rats. These uteri respond normally to acetylcholine and epinephrine. Krayer and co-workers (1946) studied MAE and deanol in the heartlung preparation of the dog because the erythrophleum alkaloids are tricyclic esters of these amino-alcohols. These esters exhibit an inotropic

212

CARL C. PFEIFFER

effect on heart muscle, whereas erythrophleic acid, the nitrogen-free fraction, has no such action in doses 100 times greater than its MAE ester. TABLE V DEANOL EFFECT AQAINST PENTOBARBITOL-INDUCED FAILURE IN A FEMALE Doc44 Systemic Cardiac output flow (ml/ (ml/ Min min) min) 0 7 17

400 400 410

20 20 30

19

Heart rate (beats/ min)

Stroke vol. (ml)

420 420 440

144 144 146

2.92 2.92 3.01

R.Aur. L.Aur. Heart pressure pressure work (mm (mm (kg-m/ H20) H20) min) 52 52 54

86 86 88

0.655 0.656 0.686

205 281 330

0.617 0.470 0.460

123 130 160 225 290 360

0.826 0.720 0.658 0.668 0.610 0.490

Pentobarbital Sodium (100 mg) Injected

35 60 68

280

200 180

50 90 90

330 290 270

140 124 120

2.35 2.34 2.25

75 83 85

Deanol HCl (300mg) Injected

70 71 73 82 90 95 100 4

Total output (ml/ min)

400 350 300 275 240 170

120 120 140 138 130 130

520 470 440 410 370 300

118 118 120 118 122 118

4.40 3.98 3.66 3.48 3.03 2.54

60 50 75 85 90 110

Heart-lung preparation (10.5 mg) with a volume of 660 ml waa used.

In Table V the dose of deanol (300 mg) dramatically restores the barbiturate-failed heart for a period of one-half hour. The smallest dose which produces this effect is 50 to 100 mg. Amounts rapging between 800 and 1500 mg are tolerated before cardiac irregularities are produced. The margin of safety is aO-fold, which is much higher than with veratrum or the erythrophleum alkaloids. Neither atropine nor nicotine prevents the effect of deanol. Of the compounds tested, deanol is most active while MAE, diethylaminoethanol, and AE are active in descending order. Choline and betaine are inactive while tetramethylammonium and tetraethylammonium have some activity. The acetate ester of deanol has a marked muscarinic effect on the heart and at the dosage used no inotropic effect is elicited. These authors conclude that deanol acts on the heart in a manner similar to epinephrine or digitalis to improve work output. I n contrast to acetylcholine, the action is not influenced by atropine or nicotine. None of the esters are more effective than deanol. I n a second publication from the same laboratory Uhle and co-workers (1966) studied


213

other esters of deanol for their inotropic effect on the heart and report that the succinic, glutaric, adipic, and pimelic acid esters have, on a molar basis, 5 to 10 times the activity of deanol. Goldin et al. (1948) studied the toxicity of deanol (apparently the alkaline base) administered intraperitoneally to the swiss albino mouse. The LD60 is reported as 2.62 mM per kilogram or 233.2 mg per kilogram of base. They find that 2-dimethyl-p-chlorethylamine,which has an LD60 of 1.95 mM per kilogram, produces cerebellar ataxia, while deanol is devoid of this cerebellar degenerative action. We find the intraperitoneal LD60 of deanol hydrochloride (calculated as the base) to be 1.143 0.33 gm per kilogram for the mouse (see Table VI) . Death is apparently due to respiratory paralysis. TABLE VI LDws Salt HCl HCl HC1 Lactate Lactate Lactate Acid tartrate Acid tartrate Acid tartrate Acid tartrate p-Acetylaminobenzoate p-Acetylaminobenzoate

OF

VAEIOUS DEANOL SALTS^

Animal

Route

Gm/kg f S.E.

Slope

Mouse Mouse Rat Rat Rat Rat Mouse Rat Rat Rat Mouse Mouse

Oral 1.P.b Oral Oral Subc.c I.P. Oral Oral Subc. I.P. Oral I.P.

3.380 f 0.086 1.143 f 0.033 4.074 f 0.170 3.361 f 0.121 1.732 f 0.122 0.804 f 0.025 3.105 f 0.154 2.590 f 0.074 1.098 f 0.092 0.459 f 0.041 3.918 f 0.105 1.020 f 0.057

18.3 18.4 18.4 14.3 6.9 16.6 10.5 16.3 6.2 6.7 20.7 11.5

All expressed as gm/kg of deanol base. I.P. = Intraperitoneal. c Subc. = Subcutaneous.

0

b

Ting and Coon (1951) report that deanol neutralized with hydrochloric acid has an LD60 of 2.08 gm per kilogram (base) when given subcutaneously to mice. I n similar trials, neutralized 2-diethylaminoethapol has an LD60 of 1.61 gm per kilogram. Cornatzer (1954) postulates that since the guinea pig, in contrast to the rat, is lacking in choline oxidase, one therefore might expect a difference in the acute toxicity of choline in the two species. Two dosage levels are used for each species, namely 450 artd 600 mg per kilogram. The drugs are given as the base 'intraperitoneally to rats (100-110 gm) and guinea pigs (200-300 gm). Deanol is a much stronger base than choline and therefore the short lethal period of 30 minutes which Cornatzer reports may have been, in

214

CARL C. PFEIFFER

TABLE VII DIFFERBYNCES IN Acnm TOXICITY OF CHOLXNE AND DEANOL IN Two SPECIES

Compound ~~~~

Choline Deanol Choline Deanol

Dose mg/k ~~~

460 460

600 600

Guinea pigs

Rats

No.

% Mortality

No.

% Mortality

29 76

45 28 39 22

20 71 74 77

~~~

14 24 20 25

60 68

part, due to the effect of the highly alkaline deanol on the peritoneal surfaces, He concludes that deanol is slightly more toxic than choline. For purposes of comparison, Hodge et al. (Hodge and Goldstein, 1942; Hodge, 1944) report the following LD60safor choline (hydro) chloride administered intraperitoneally (2%): 320 mg per kilogram for the mouse and 670 mg per kilogram for the rat. For oral toxicity, Neuman and Hodge (1945) used male and female rats ranging in weight from 76 to 343 gm. For high concentrations of 670 and 500 mg per millimeter, the LDso is 3.4 gm per kilogram; for lower concentrations (200 to 400 mg per milliliter), the LDliois 6.1 gm per kilogram. Summary: Pharmacological studies. Deanol is pharmacodynamically less active than choline in its immediate effect on blood pressure. Deanol produces an inotropic effect on the heart. Deanol is not chronically toxic to the brain as is the P-chlorethyl analog. Deanol is reported to be slightly more toxic than choline when given intraperitoneally to guinea pigs but less toxic than choline by the intraperitoneal route in mica. The LD60-r of choline are tabulated for comparison. The pharmacological studies on deanol are fewer than the rather extensive nutritional and biochemical studies. X. Pharmacological Studies on Deanol

Deanol in oral doses up to 1.0 gm per kilogram does not inhibit the CAR of the rat but an effect can be found in young rats learning the CAR. This is a significantly quicker response similar t o that reported by Jacobsen et al. (Jacobsen and Sonne, 1955; Jacobsen and Skaarup, 1955) for benactyzine. Also, rats on chronic dosage of deanol have an altered response to the inhibitory effect of arecoline when compared to rats on chronic dosage of choline. Chemical congeners of deanol which have a more reactive functional group than the hydroxy group will produce a specific inhibition of the CAR. Thus, (CHs) 2NCH&H&3H will inhibit


215

the CAR down t o 10% of normal performance when given in an intraperitoneal dose of 0.88 mM per kilogram (92.4 mg per kilogram). This inhibition is not influenced by atropine (10 mg per kilogram). The compound, (CH,) z N C H ~ C H ~ N Hwill Z , completely inhibit the CAR after an intraperitoneal dose of 3.5 mM per kilogram (308 mg per kilogram). This compound produces a continuous rage reaction in the rats and the inhibition of the CAR extends beyond 8 hours with gradual and complete recovery in 24 to 48 hours (Jenney, 1958). In single large doses deanol is of little pharmacological interest in that it is less active than choline as a parasympathetic stimulant. However, when the mice or rats are placed on a continued high daily intake of deanol they show greater emotionality, and eventually they will have spontaneous epileptiform seizures and will be susceptible to audiogenic seizures (Pfeiffer et al., 1957a). I n mice and rats a seizure usually begins with clonic movements of the head accompanied by squeals and, in rats, marked salivation. The head is thrown back; the animal assumes a sitting position with forelegs elevated as if for face washing. The whole body may be involved in myoclonic jerking during which there appears t o be no loss of consciousness. The seizure may end abruptly at this point or progress into violent clonic activity, running movements, loud squealing, aimless biting, and loss of consciousness followed by a period of postictal depression of 2 to 15 minutes. The seizure is primarily clonic in nature but occasionally may progress to one with a tonic flexor component. This seizure activity is completely reversible when deanol is discontinued. Three groups of 10 or more mice on a normal diet have been treated with tap water (controls) choline (0.03 M ) , and deanol (0.03 M ) in their drinking water for a 2- to 3-month period. Table V I I summarizes the convulsant thresholds to electroshock, pentylenetetrazol (Metrazol) (Orloff et al., 1949; Jenney and Pfeiffer, 1956), strychnine, and amphetamine (Lasagna and McCann, 1957). Choline raises slightly the threshold to electroshock but the rise is greater with deanol and significantly different from that of choline. This rise may not be owing to deanol but rather t o the constant seizure activity which provides chronic postictal depression. Choline has no effect on the pentylenetetrazol threshold, but deanol makes the mice twice as sensitive to pentylenetetrazol which is a highly significant difference. Choline has no effect on the strychnine threshold, whereas deanol elevates the strychnine threshold 30%, which is again significant. Deanol in high dosage in the mouse lowers the threshold for pentylenetetrazol and elevates the threshold for electroshock and strychnine convulsions. This spectrum of action differs from that of reserpine and chlorpromazine, both of which lower the threshold of pentylenetetrazol and electroshock.

216

C A W C. PFEIFFER

TABLE VIII THE CONVULSANT THRESHOLDS IN MICE TREATBID WITH 0.03 CHOLINE OR DEANOL FOR 2 TO 3 MONTHS

Water controls

0.03 M Choline

0.03 M Deanol

Convul. dose 50 f stand. error in milliamperes

6.86 f 0.22

7.75 f 0.260

8.91 f 0.32b

Pentylenetetrazol (0.5% T.I.V.I.G)

Mean convul. dose f stand. error, mg/kg

92.1 f 4.7

88.1 f 6.3

47.5 f 6.6b

Strychnine sulf. (0.005% T.I.V.1.e)

Mean convul. dose f stand. error, mg/kg

1.01 f 0.025

1.00 f 0.027

1.30 f 0.10’

dl-Amphetamine

LDw f stand. error, mg/kg

12.2 f 2.4

16.7 f 2.4

Convulsant

Threshold units

Electroshock (0.3 sec duration)

>30.0

0 Indicates statistically eigniicant difference from water controls and deanol treated group. * Indicates atatistically significant differencefrom both choline and water controls. 0 T.I.V.I. = timed intravenous infusion a t the rate of 0.06 ml/lO sec.

The lowered threshold to pentylenetetrazol occurs with doses of 300 to 500 mg per kilogram per day, and the effect is completely reversible when the deanol is discontinued or the daily dosage reduced. TABLE I X P~PNTYL~NBITETRAZOL THRESHOLDS AFTER DISCONTINUATION OF DEANOL IN MICE Threshold (mg/kg) After 3 months on 0.03 M Offdean013 days Off dean019 days Off dean0121 days Off dean0130 days

Water (mg/kg) Choline (mg/kg) Deanol (mg/kg) 71.4

-

73.3

-

-

-

-

27.3 40.2 44.3 55.2 76.0

Mice on choline are not significantly different from the controls in their sensitivity to amphetamine, whereas those on deanol are significantly less sensitive to the lethal effect of amphetamine. As a precursor to intracellular or bound acetylcholine, one might expect deanol-treated animals to show differences in their sensitivities to dtubocurarine, decamethonium, atropine, and neostigmine. Table X summarizes the thresholds in groups of mice treated chronically for a 2- to

217

PARASYMPATHETIC NPJUROHUMOBS AND BEHAVIOB

TABLE X EFFECT ON THRESHOLDS OF OTHERDRUQS IN MICE TREATED WITH 0.03 M CHOLIN~ OR DEANOL Dl'W

Threshold

Controls

Choline

d-Tubocurarine (intravenously)

Paralyzing dose: 60 f stand. error flg/kg Paralyzing dose: 60 f stand. error Ilg/kg Deaths at 60

62.2 f 2.4

60.0 f 3.4

51.6 f 1.7"

343 f 17.1

426 f 20.0"

316

Decamethonium (intravenously) Atropine-SO, (intravenously) Neostigmhe-HBr (intravenously) 0

Deanol

+ 16.0

2/16

3/17

10/16"

0/10

6/10"

6/10"

w/kg

Deaths at 100 ag/kg

Indicates statistically significant difference from other groups.

3-month period with tap water, 0.03 M choline, and 0.03 M deanol (which is equivalent to 300 to 500 mg/kg/animal/day) . Choline-treated animals are not different from the controls in their sensitivity to d-tubocurarine, whereas deanol-treated animals are significantly more sensitive. The choline-treated animals are significantly less sensitive to decamethonium, whereas the deanol-treated animals are slightly more sensitive but not significantly so. Choline-treated animals are not more sensitive to intravenous atropine, whereas the deanol-treated animals are significantly more sensitive. Both the choline- and deanol-treated mice are slightly more sensitive to intravenous neostigmine. d-Tubocurarine paralyzes muscles by stabilizing the end-plate potential while decamethonium depolarizes the end-plate. It is, therefore, of interest to find that chronic choline administration does not affect d-tubocurarine action while the threshold for decamethonium is raised, Conversely, deanol in chronic dosage lowers the thresholds for both d-tubocurarine and decamethonium although only the d-tubocurarine effect is statistically significant. Atropine, a tertiary amine, is more lethal in the deanol mice, while neostigmine, a quaternary amine, is more lethal in both the choline and deanol animals than the controls (Table X ) . We are at a loss to explain those changes in terms of classic pharmacological knowledge except for the possibility that acetylcholine within the body cells may be quite different in its pharmacological effect from the response which is ordinarily seen when acetylcholine is applied to exterior surface of the cell. If deanol is a precursor of acetylcholine, one might expect to evoke an antidiuretic response in rats by the standard test. Blackmore (1968)

218

CARL C. PFEIFFER

has used 50 and 75 mg per kilogram doses of deanol in comparison with a 75 mg per kilogram dose of choline and compared the results with water alone (Table XI). It is obvious from these data of Blackmore that, for a given dose, deanol is more antidiuretic than is choline and that both doses of deanol are significantly antidiuretic in the rat. TABLE XI COMPARATIVE ANTIDIURETIC EFFECTOF DEANOLAND CHOLINE IN FASTEDRATS*

Group

Dose

Group A

Water 25 ml/kg

Group

1

Group

{

Group

Water 25 ml/kg Dean0160 mg/kg Water 25 ml/kg Dean0176 mg/kg Water 25 ml/kg Choline 75 mg/kg

5-Hour excretion of water load (% of load f standard deviation)

Statistical analysis

56.1 f 11.7% 33.1 f 16%

A vs. D p A vs. C p

< 0.05 < 0.001

26.6 f 13%

A vs. B p

< 0.001

43.2 f. 13.5%

C vs. D p

< 0.01

0 Measurementa taken for a period of 5 h o w after dosing. Rats were housed 2 to a metabolic cage and each group contains data from 9 to 16 trials.

In studies on renal hemodynamios in trained unanesthetized dogs, Blackmore (1958) shows by the usual clearance methods that doses of 1 to 5 mg per kilogram given intravenously do not significantly alter glomerular filtration rate or renal plasma flow.In contrast to his findings in rats, changes in urine output did not occur. Preliminary studies on electrolyte excretion indicate that deanol does not change the excretion of sodium, potaasium, or chloride when given in single intravenous doses. Kiplinger and associates (1958) have studied deanol for its cardiac action and find that it antidotes the depressant effect of pentobarbital in the dog heart-lung preparation where the lethal dose of pentobarbital is raised from 135 to 270 mg per kilogram by deanol therapy. This confirms the studies of Krayer et al. (1946). Deanol also antidotes the depressant effect of quinidine or sodium fluoroacetate on the myocardium. Their findings are consistent with the hypothesis that deanol may act as a precursor of acetylcholine in myocardial cells. Konigsmark et al. (1958) and Killam et al. (1958) show that deanol in the curarized cat produces low voltage fast activity in the EEG suggestive of mild stimulation. The threshold for EEG arousal is lowered for

PARABYMPATHETIC NEUROHUMORS AND BEHAVIOR

219

electrical stimulation of the reticular formation and possibly the thalamus. The increased activity of the thalamus alone is insufficient, to increase the low voltage fast EEG activity in the absence of ascending influences from the reticular formation. Increased tonic activity of the entire reticular formation appears to be the mechanism of deanol. They further find that deanol antagonizes the depressant effect of barbiturates on the reticular formation. The effect of deanol is variable and thus supports the view that biochemical adjuvants may, if combined with deanol, produce a more constant effect. The action of deanol is similar to amphetamine and dissimilar to eserine which causes prompt arousal of the EEG even in cats having sections through the midbrain reticular formation. Eserine, therefore, does not selectively stimulate the reticular formation. Since eserine inhibits cholinesterase and, thus, presumably increases cerebral acetylcholine levels, these authors conclude that deanol may not produce its CNS stimulant effect by acting as a precursor of cerebral acetylcholine. One must, however, attempt to differentiate between a drug effect of eserine and eserine acting as an anticholinesterase. This differentiation is thus far impossible. The antidiuretic effect of deanol in rats and the antidotal effect against pentobarbital in the heart and brain can be rationalized as evidence for the conversion of deanol into acetylcholine in the body. That deanol is methylated to choline in other tissues has been shown by various groups of biochemical investigators. Groth et al. (1958) compare the time course distribution of radioactivity from deanol-1,2-C14, choline-C14H3,and choline-1,2-C14 in the acid soluble, lipid, and protein fractions of mouse brain and liver. The mice received 0.1-0.15 mg per kilogram dose of the labeled compounds. Both choline compounds provide a peak of 4% of the label as expired C1402 within 30 minutes. Deanol labeled as C1402 also appeared in a peak at 30 minutes but was 0.6% of the administered radioactivity. The 24-hour urinary excretion of label from choline-C14H3 is 36% of the administered dose while only 12% of the deanol label appears in the urine in 24 hours. The amount of radioactivity from deanol appearing in the acid soluble fraction of brain attains a constant level in 30 minutes and is retained over the 24-hour period of measurement. Choline-C14Hs or choline-1,2-C14 are incorporated into the acid soluble a t only 50% of the rate of deanol incorporation. In addition radioactivity from choline is progressively lost from the acid soluble fraction. Although the brain lipids become more quickly labeled when choline is administered, deanol is over-all more effective. Radioactivity from choline is incorporated a t a much higher rate into brain protein than is deanol which may perhaps

220

CARL C. PFEIFFER

indicate a more rapid breakdown of choline into smaller fragments. They conclude that although choline serves as a more immediate source of CNS choline in the mouse, deanol becomes within 12 to 24 hours a comparable and perhaps more effective precursor of intracellular choline. These findings may explain the ineffectiveness of single doses of deanol when compared to the convulsive stimulation found with chronic dosage.

XI. Trial of Deanol in Human Subjects

By 1967 we had accumulated s a c i e n t pharmacological data on deanol, in both animals and patients, to study deanol for its possible stimulant effect in medical student volunteers. Deanol, 10 mg of base in tablets (as the acid tartrate salt), was compared to an identical placebo tablet under double blind test conditions. Prior to the experiment, group Rorschach tests, group learning of nonsense words, and questionnaires were completed. The questionnaire objectively tabulated changes in mood; energy; sleep, bowel, and urinary habits; libido; thirst; muscle tone ; appetite ; mental concentration; headache incidence; and the physiology of the nose, eyes, mouth, and hands. In addition, blood pressure, pulse, body weight, hsnd-steadiness, muscle power, and vital capacity were measured and recorded weekly. Gastric analysis was done in 8 subjects and determinations of blood cholesterol, protein-bound iodine, and cephalin flocculation tests were done on 19 subjects during the 12 weeks of the experiment. The dose was regulated at 1 tablet per day for the first week and 2 tablets per day for the second week. The total daily dose was taken in the morning of each day. After 2 weeks the subjects were allowed to increase or decrease their daily dosage with the top limit set at 3 tablets per day. Treatment was continued “doubly blind” for a 6-week period during which time 17 students received deanol and 18 students received placebo therapy. After 6 weeks the double blind portion of the experiment was terminated and all students were placed on deanol therapy and the experiment was continued uncontrolled for an additional 6-week period. Statistical analysis of the controlled data of the first 6 weeks of the experiment showed the following significant changes when the deanol group was compared with the placebo group: (1) increase in muscle tone in the deanol group p < 0.05; (2) increased ability for mental concentration p < 0.02; and (3) changes in sleep habits p < 0.01. In most instances, the change in sleep habit was “less sleep required.” Others reported sounder sleep with earlier, clear-minded awakening. No signifi-


221

cant changes occurred in heart rate, blood pressure, muscle strength, hand-steadiness, vital capacity, and body weight, or the levels of fasting gastric acidity, protein-bound iodine, blood cholesterol, and cephalin flocculation tests of liver function. The weekly questionnaire and observations were continued during the last 6 weeks of therapy when 35 students were on deanol in doses of 10 to 30 mg of base as the lactate, acid tartrate, or p-acetamidobenzoate salts. During this time 25 of the 35 subjects noted a definite stimulant effect. Five subjects thought the effects were indefinite and the remaining 5 subjects could not discern any stimulant action. The subjects who noted CNS stimulation with deanol had many interesting findings to report. Most subjects reported greater daytime energy, attentiveness at lectures (but greater intolerance of poor lecturers) , sounder sleep with a reduction in the hours of sleep needed, and better ability to concentrate on writing of papers or studying. A decrease occurred in normal apprehension prior to and during examinations. Many reported a more affable mood and outspoken personality. Two subjects reported that they were able to stop smoking without difficulty, whereas prior attempts had been unsuccessful. Two subjects reported that amphetamine taken at examination time was devoid of its customary stimulant action. Several subjects noted an increase in their tolerance to alcoholic beverages and freedom from “hangover” depression or headache. Two subjects reported an apparent increase in sensitivity of the retina as evidenced by sneezing in bright sunlight and awareness of reflections from their spectacle lenses. Since then, 2 patients have reported that contact lenses, which were formerly tolerated for a 12-hour period, can now only be worn for a 2-hour period while on deanol (they both prefer deanol t o their contact lenses). A large initial dosage of deanol can produce dull occipital headaches. Continued overdosage produces increased tone of the anti-gravity muscles such as the neck, masseter, and quadriceps groups. Insomnia also occurs with overdosage but is relieved by reduction in the daily morning dose. In man, CNS stimulation by deanol differs from that produced by the amphetamines in that deanol stimulation develops slowly over a period of 2 weeks. A druglike let-down or after-depression does not occur and the stimulant action, when fully developed, lasts at least 24 hours. After 3 to 4 weeks the reports from our volunteers indicate that an amphetamine-like stimulation without the motor component of amphetamine is continually present. The animal data show further that amphetamine is less toxic in the chronic deanol mice.

222

CARL C. PFEIFFER

XII. Choice of a Deanol Salt for Clinical Trial

In trials with oral doses of deanol salts in volunteer subjects, the degree of cerebral stimulation varied with the salt of deanol used. Of the three salts used the lactate was the most stimulant, the p-acetamido benzoate salt next best, and the tartrate the least effective. That this may be correlated with the degree of absorption from the gastrointestinal tract is plausible from the ratios of the oral to parenteral LDao-.in animals (see Table XI1 and Table VI) . TABLE XI1

RATIOOF LDws OF ORAL/PARENTBIRAL DOSES IN MICE AND RATS Oral/I.P. ratio of HC1 salt (mouse) Oral/I.P. ratio of PAAB salt (mouse) Oral/I.P. ratio of Lactate salt (rat) Oral/I.P. ratio of Tartrate salt (rat)

2.96

3.84 4.18 6.64

Since these LDaovsare all calculated on their content of deanol base, the lethality must be influenced by the acid used to neutralize the base. The hydrochloric acid salt has the best absorption (lowest ratio) while acid-tartrate has the poorest absorption (highest ratio). This might reflect either tartaric acid toxicity by I.P. route or lack of absorption of the acid-tartrate salt by oral route. Even the lactate salt shows a ratio significantly higher than the hydrochloride and, since lactic acid has no inherent toxicity, the possibility of the salt influencing intestinal absorption becomes more plausible. Lasslo (1957) finds that the acid-tartrate salt of deanol will not evolve deanol when heated to 100OC. and pressure is reduced to 0.06 mm Hg. The acetate and p-acetamidobenzoate salts lose all the deanol base. It is therefore postulated that organic acids may in some instances have a more firm binding to deanol than the usual ionic bond-perhaps in the case of hydroxy acids a hydrogen bond is formed. Gero and Reese (1956) find that 2-diethylaminoethanol is hydrogen-bonded to various amino acids and simple peptides and thus similar hydrogen-bonding might occur with deanol. Furthermore, Harington (1953) finds that tartaric acid influences the absorption from the intestine of hexamethonium. By using 24-hour urinary excretion as the index of oral absorption he finds only 1.4% of the hexamethonium absorbed when given as the acid tartrate salt, whereas 18.7% is absorbed when the bromide salt is used and 16.7% is absorbed when the chloride salt is used.


223

Three factors recommending the p-acetamidobenzoate salt of deanol for clinical use are (1) the nonhygroscopic nature of this salt, (2) the absorption from the gastrointestinal tract, and (3) the fact that p-acetamidobenzoic acid is nontoxic and is known to occur in the body.

XIII. Possible Modes of Action of Deanol

(1) Deanol in mice is not oxidized as is choline and the over-all retention of this molecule by the brain is greater than choline. Since deanol is a precursor of choline the mice made convulsant with deanol probably have a higher level of ACh. We have attempted to assay the ACh levels of mouse brain by use of the leech and frog rectus muscle preparations but have been unsuccessful. We are now using the guinea pig ileum bioassay on brain concentrates but data are not yet available. (2) If acetylcholine levels are not raised by deanol therapy then the stimulant effect may be owing in some unexplained manner to the increased incorporation of phosphate into brain lipids. We know that deanol is more active than choline in this incorporation of phosphate. (3) If acetylcholine is not the cerebral parasympathetic neurohumor then the stimulant effect might be owing to a sparing action on the unknown neurohumor.

XIV. Deanol in Clinical Disorders

A. CHRONIC FATIGUE STATES The best result obtained with deanol therapy is in the treatment of chronic fatigue and mild to moderate depression in the neurasthenic patient. The drug produces an increase in drive and in physical energy. It produces greater affability and a more out-going personality. People with insomnia find they can sleep well while on deanol therapy. In one study of over 100 patients presenting various complaints of psychosomatic origin, especially of exhaustion and depression, a majority report that deanol therapy increases their “energy” and decreases their depression (Lemere and Lasater, 1958). The maintenance dosage is from 10 to 50 mg (average 25 mg) taken as a single oral dose after breakfast. Approximately 70% of these patients express a preference for deanol over other previously prescribed drugs. Dependency or tolerance does not develop and most patients spontaneously discontinue the medication as they feel better. Other unpublished reports confirm the efficacy of deanol in mild depression and chronic fatigue states. The data indicate that a good re-

224

CARL C. PFEIFFER

sponse may be expected in about 70% of such patients. No significant side effects or toxic reactions are reported. Deanol is found to be effective also when given to counteract the drowsiness caused by tranquilizing drugs, Paradoxically, in some nonhospitalized neuropsychiatric patients, deanol therapy has been reported to relieve anxiety even better than did previous medication with tranquilizing agents. B. PERIODIC HEADACHES Numerous investigators have reported data on patients with tension or relaxation headache treated with daily doses of deanol as large as 75 mg per day. A beneficial effect was obtained in 70%. In patients having migraine headaches, beneficial results were obtained in over 80%. Thus, between 70 and 80% of patients with headaches are benefited. These patients also show the usual alerting effect of deanol and may in addition show weight gain, greater affability, and relief of dizziness. In patients with hypertensive headaches, the blood pressure level is not changed with deanol therapy but the headaches are prevented.

C. BEHAVIOR DISORDERB OF CHILDREN Oettinger (1958) reports on the use of deanol in behavior disorders of children. His average dose was 50 mg per day, usually given in the morning but occasionally divided and given half in the morning and half after the noon meal. The report is based on continuous treatment for 3 to 9 monthe. The group includes 83 boys and 25 girls. Seventy-six per cent of the girls have fair to good improvement in their behavior and 66% of the boys have similar improvement. Twenty per cent of the total group show no change and 12% have increaeed behavioral problems on deanol therapy. In addition, in a group of 17 epileptic children treated with deanol, 6 have increased behavior problems, 1 no change, 4 fair, and 7 good improvement in behavior. The seizure activity of these patients show 7 increased (grand mal) , 9 no change, and 3 good relief of seizures (petit mal). Two patients had increase in major seizures and a decrease in minor seizures. This effect on seizure activity correlates with the findings in animals and the observations of Pfeiffer et al. (1957a). Oettinger concludes that deanol is not as uniformly effective in seizure control as is amphetamine, but in behavior disorders it is superior to meprobamate, captodiamine, the phenothiazines, and reserpine. Deanol is also superior in some aspects to the amphetamines in that anorexia, tachycardia, insomnia, and early morning drug let-down does not occur. The children on deanol therapy show a decrease in hyperactivity, a lengthening of attention span, decreased irritability, better scholastic

PARASYMPATHETIC NEUROHUMOBS AND BEHAVIOR

225

ability, less enuresis, and, occasionally, an increase in intelligence quotient.

D. EFFECT IN PETIT MALAND PSYCHOMOTOR EPILEPSY AND NARCOLEPSY Several investigators have reported that deanol has beneficial effects in petit ma1 and psychomotor epilepsy. In a study now in progress, maintenance doses of 50 mg of deanol per day are given to patients having psychomotor epilepsy and no component of grand ma1 epilepsy, Many of these patients are reported to be showing an improvement in personality. Deanol is effective in two-thirds of the narcoleptic patients and superior to other stimulants in this disorder. Deanol has caused no untoward effects in unmixed psychomotor epilepsy, but an increase in attacks has been reported in 2 cases of grand ma1 epilepsy. The drug is not recommended in the presence of this latter disorder. E. EFFEOT IN SCHIZOPHRENIA Seventy schizophrenic patients at a state institution have received continuous deanol therapy for an average of 9 months (Murphree et al., 1957). Chronic schizophrenic patients receiving 250 mg per day have shown increased motor activity and increased verbal responsiveness within the first 2 weeks of therapy. Because the male patients in this group lost weight during the first month of therapy, the deanol dosage was reduced until all patients were maintained on 60 mg per day. Thereafter, improvement was gradual but definite over the next 9 months of therapy. Others also have reported improvement in some schizophrenic patients. Pennington (1957) notes that 5 female schizophrenics improved markedly in interest, in attentiveness to personal care and hygiene, and in their work. This investigator reports also that somatic delusional trends are decreased in the entire group of 30 patients. Several investigators report marked improvement in cases of catatonia, and some chronic schizophrenics, who had obtained a maximal response to tranquiliring drugs, show further improvement when deanol is added to their therapeutic regimen. Ambulatory, non-institutionalized schizophrenics show a quicker response than the chronic institutionalized patients. Some of these have been treated covertly in their home environment with good results. F. EFFECT IN SEVERE DEPRESSION In severe depressions deanol therapy is not as effective as electroconvulsive therapy or the monoaminoxidase inhibitors. Because of the usual urgency of these syndromes, the slow biochemical action of deanol

226

CARL C. PFEIFFER

ie undesirable. In comparisons of iproniazid with deanol the former produces a more rapid effect in the severely depressed patient. Deanol can be used as an adjuvant to iproniazid to decrease the possibility of liver damage. In mild depressions, deanol is effective in 70% of the patients treated, and the patient usually notes a slight improvement within a week and the maximal effect in 3 to 6 weeks. G . MISCELLANEOUS The role of deanol as a possible precursor of acetylcholine and its action on nervous tissue in particular suggest its use in functional bowel distress including spastic constipation and mucous colitis. Several asthmatics have reported benefit with deanol therapy, and further trial seems indicated in bronchial asthma in which psychogenic factors are important. An anti-inflammatory action has been reported for deanol (Kuehl et al., 1957) in animal tests. These findings have been confirmed by Cronheim (1958) and are borne out in part by the original chronic toxicity experiments which show a significant decrease in eosinophil levels of both male and female rats. A slight but Significant drop in hematocrit and hemoglobin might also indicate that deanol produces a hemodilution. Myasthenic patients have not responded to deanol therapy.

XV. Biochemical Adjuvants to Deanol Therapy

If deanol is the precursor of cerebral acetylcholine then various biochemicals such as the water soluble vitamins and methyl donors might enhance the antischizophrenic effect. To date we have tried 18 possible adjuvants for periods of 1 to 4 weeks in patients selected for their ability to respond to other therapies such as electroconvulsive therapy, etc. Only four compounds appear to have an added beneficial effect (see Table XI11 and Table XIV) . Pantothenic acid is contained in coenzyme A (CoA)and therefore might be'expected to have an enhancing effect. Furthermore, Len5 (1952) found this vitamin to be somewhat effective in schizophrenia. He still uses it but always concomitantly with tranquilizing drugs (Lena, 1956). With a normal diet a deficiency of pantothenic acid is unlikely. However, with a deficient diet and pantothenate antagonists, normal volunteers become nervous and irritable, have slowed mentation, and adrenal cortical insufficiency (Bean et aZ., 1955; Thornton et al., 1955). Pyridoxal phosphate is active as the prosthetic group of the coenzyme needed for decarboxylation, transamination, and desulfhydrase activity in the body. It appears to be a definite adjuvant to the deanol therapy

PABASYMPATHETIC NEUBOHUMOFtS AND BEHAVIOB

227

TABLE XI11 , VITAMINAND BIOCHEMICAL CO-FAOTOFW WHICHARE IMPORTANT IN THE METHYLATION AND ACETYLATION OF DEANOL TO ACETYLCHOLINB IN TEE BODY Methyl donation Methionine Folic acid Betaine (produced by choline oxidase) CHa 0

\

CH-N--CHAHr-OH deanol

HO!-CHr

Choline acetylase

+ Coemyme A-acetylation

8-mercaptoethylamineo Pantothenic acid 8-alanine Pantoic acid

1

a

Cyancobalamin keeps thiol group reduced. Thyroxine increases liver and brain coenzyme A activity. Pyridoxine provides : (1) Cysteine from methionine (2)8-mercaptoethylamine from cysteine. TABLE XIV

BIOCHEMICALS TRIEDAS POSSIBLE ADJUVANTSTO DEANOL THERAPY

No. of patienta 23 7 7 3 20" 4 6

1 2" 10

3 6 1

3 1 70a 8"

ga 20

Biochemical Choline dihydrogen citrate Folic acid Folic acid LGlutamic acid L-Triiodothyronine I-Methionine Z-Cysteine 1-Tryptophan d-Calcium pantothenate 8-Alanine Somatatrophic hormone Nicotinic acid Thiamine Uridine Cytidine Pyridoxine HC1 Cyancobalamin &Mercaptoethylamine a-Tocopherol (Vit. E)

Dose 1-6 gm/day 5 mg/dw 10 mg/day

3.0 gm/day 100 pglday (1-700 pg/day) 4.5 gm/day 2.0 gm/day 4.6 gm/day 25 mg/day 2 gm/day 10 rg/day 2 gm/day 200 mg/day 1.0 gm/day 1.0 gm/day 10 mg/day 100 d d a y 300 mg/day 100 mg/day

a Biochemicals which apparently increase the clinical effectiveness of deanol in chronic sckophrenia.

228

CARL C. PFEIFFBIII

of schizophrenia, headache, minor epilepsies, and chronic fatigue states. Pyridoxine is necessary for the formation of cysteine from methionine and for the decarboxylation of cysteine to form 8-mercaptoethylamine. Cyancobalamin, in moderate dosage of 100-200 pg per week, appears to have a definite enhancing effect in the treatment of headache, schizophrenia, and depression, but not in petit ma1 or psychomotor epilepsy. Its biochemical action is probably that of keeping the sulfhydryl group of CoA in a reduced and active state. 8-Mercaptoethylamine (cysteamine) is the active end-group of CoA and is derived from decarboxylation of cysteine in the body. This compound has been studied extensively as a protectant against radiation damage, Oral doses of 100 mg given three times per day are without toxicity and appear to have an enhancing effect in schizophrenics when given over a period of 3 months. Other sulfur-containing compounds such as cysteine and dimercaprol (BAL) are ineffective. Brice (1935), Altschule et al. (1952), and Martens et al. (1956) report that glutathione blood levels are abnormally low in schizophrenics and that glutathione in enormous dosage by intravenous infusion has a temporary antischizophrenic effect (Altschule et al., 1957) . Various workers (Quastel and Wales, 1938; Georgi et al., 1948; Smith, 1954; and Muller, 1955) have found the benzoic acid test of liver function to be abnormally low in the schizophrenic in that less hippuric acid is formed and excreted than in normal individuals. The conjugation of benzoic acid with glycine to form hippuric acid is CoA-linked, and, therefore, this lowered synthesis may indicate a specific decrease in tissue CoAactivity rather than poor liver function. Unfortunately, CoA occurs only in tissues and is not present in the blood so that CoA activity in schizophrenics could only be measured by tissue biopsies or indirectly as by tests of the ability of the schizophrenic to acetylate p-aminobenzoic acid or the sulfonamides. These simple tests for acetylation have not been explored. Tabachnik and Bonnycastle (1954) show that thyroid function is intimately linked with CoA activity. Thyroidectomized rats have decreased activity which can be restored to normal by thyroid feeding. This may account for the occasional anti-schizophrenic effect of added thyroid. In our studies L-triiodothyronine may have had an enhancing effect on deanol therapy. The unconfirmed reports of Torda and Wolff (1946) on choline acetylase are of interest. Of the amino acids which increase the action of choline acetylase, they list I-arginine, I-lysine, 2-cystine, 2-methionine, I-cysteine, 2-histidine, I-tryptophan, and the peptides carnosine and

229 glutathione. Of these, the dipeptide carnosine would appear to be the most active. Carnosine has not been tried in schizophrenia. The drugs and naturally occurring substances which have this effect are the methylated xanthines, caffeine and theobromine, adenylic acid, a-tocopherol, and urea, while the inhibitors of choline acetylase are salicylic acid, benzaldehyde, phenol, a- and 8-naphthol, quinoline, piperidine, indole, skatole, and camphor. Other naturally occurring substances which inhibit are adenine acetate, alloxan, ammonia, guanine, and uric acid. The last compound may be of interest because of its occurrence and excretion in primates and almost total lack of occurrence in lower species. Were those clever, witty, and active individuals who made history in spite of their gout ever subject to periods of schizophrenia? Does caffeine exert its stimulant effect by promoting acetylcholine synthesis? From this review of possible adjuvants one can conclude that the schizophrenic should do best on a high protein diet supplemented by a multivitamin tablet which contains 2 mg of pyridoxine. A daily dose of 50 pg of L-triiodothyronine and a weekly injection of 1 mg of cyancobalamin may also help. PABASYMPATLfBrmC IVBIUIIOHUMORS AND BEHAVIOR

XVI. Hallucinatory Effect in Man of Acetylcholine Inhibitors

In man, many of the symptoms produced by the hallucinogen LSD-25 are not unlike those of atropine. These symptoms are facial flush, rise in blood pressure, dilation of the pupils, and rise in body temperature. The hallucinatory effect of atropine in overdosage is well known-a toxic delirium occurs with doses of 5 to 10 mg (Gordon and Frye, 1955). We have shown that muscarinic stimulation of the brain produces Parkinsonian tremor in animals and the control of this tremor clinically with atropine or scopolamine is a time honored procedure. We have used a group of normal volunteers at the Atlanta Penitentiary. These men are a drug-sophisticated group and can differentiate accurately between 0, 25, 50, and 100 pg of LSD (DeMaar et aZ., 1958). The tests of atropine derivatives were done under double blind test conditions except that commercial tablets were used, but the tests were done in groups of four trials so that the subject did not know the nature of the tablets. Abood et al. (1958) find that JB-318, the tertiary amine analog of “Piptal,” and JB-336 (see Fig. l ) , the N-methyl derivative of JB-318, produce a model psychosis which is characterized by visual and auditory hallucinations when these compounds are given in oral doses of 10 to 20 mg. We have not been able to give doses of JB-336 larger than 9 mg be-

230

CARL C.

PFEIFFER

cause of the severity of the mental effects. Jenney and Healy (1958) find that JB-336 is ten times more active than atropine in preventing the usual effect of arecoline (2 mg. per kilogram) in the CAR of rats. A similar possible acetylcholine antagonist, MER-16, produces severe LSD-like effects when given in oral doses of 150 mg. The hallucinatory effect persists for 3 days and is characterined by repeated waves of depersonalization, visual hallucinations, and feelings of unreality. Because of these interesting findings with acetylcholine antagonists, we studied most of the compounds which are presently used for the therapy of Parkinsonism (see Table XV) . Without exception, when adequately large doses are used, these atropine congeners produce an effect which the subjects liken to LSD-25. These drugs are more effective than either 0 OH

Citrate salt

JB.468 CH3

MER.16

OH

Hydrochloride salt

CH3

Fw.1. Chemical eitructures of the experimental hallucinogens.

scopolamine or atropine in producing hallucinatory effects. Scopolamine was not more active as an hallucinogen than atropine, but the subjects did report a sedative and sleep-producing effect which lasted 3 days with a 2-mg dose of scopolamine. The simple alcohol of JB-336, called compound JB-468, did not produce hallucinations in the dosage used which was 100 mg of the citrate salt given orally. The subjects did report a sedative action of JB-468 and this compound may be analogous in its pharmacological action to Laborit’s compound SCTZ, which is reported to have a sleep producing effect. Summary. Synthetic atropines, both marketed and unmarketed, are more active hallucinogens than either atropine or scopolamine. Some of these, such as Artane, produce hallucinatory effects which


23 1

last 24 to 48 hours. The recommended daily dosage of many of these anti-tremor drugs produces hallucinations in single trials in these normal subjects. The Parkinsonian patient may possibly be more resistant to the hallucinatory effect than is the normal subject, and the recommended dosage schedules would tend to produce tolerance to the hallucinatory effect. TABLE XV DOUBLEBLINDSTUDYOF EXPERIMENTAL HALLUCINOGENS AND ANTI-TREMORDRUGS FOR THEIR LSD-25-LIKE EFFECTI N NORMAL HUMAN SUBJECTS Type Experimental hallucinogens

Anti-tremor drugs

b

Drug

Dose (mg)

JB-336 JB-336 JB-468 MER-16 Artene Artane Pagitane Panparnit Cogentin Kemadrin Parsidol Disipal Benadryl Atropine Scopolamine

6 9 100 150 5 10 5 25 2 15 50 100 100 2 2

Results 6/6” = 75 LSD 3/3 = 100-150 LSD 0/6 = Sedative effect only 8/8 = 150+ LSDb 6/8 = 25-50 LSD 5/5 = 100-150 LSDc 6/8 = 25-50 LSD 4/8 = 25-50 LSD 3/8 = 25-50 LSD 3/8 = 25 LSD 3/8 = 25 LSD 2/8 = 25 LSD 2/8 = 25 LSD 1/8 = 25 LSD 1/8 = 25 LSD

Read 6 out of 6 subjects likened this to 75 pg LSD-25. Effect lasts 3 days. Bradycardia in all subjects and extrasystoles in 2.

The newer synthetic anti-tremor drugs have fewer peripheral side actions but the central side actions, such as hallucinations, are exaggerated and should be anticipated in therapy.

XVII. Critique of Acetylcholine as the Parasympathetic Neurohumor

Sigmund Freud once wrote “The concepts I have summarized here I first put forward only tentatively, but in the course of time they have won such a hold over me that I can no longer think in any other way.” The present reviewer is in much the same mental state in regard to his approach to the SAR of acetylcholine (Pfeiffer, 1948). Therefore, the reader should keep in mind that with the Freudian concept the art has

232

CARL

0. PFEIFFER

far outstripped the science, which fact is a grave deterrent to psychiatric advance in some areas of the world. Model hypotheses which deal with SAR may, on vigorous promotion, actually deter alternate novel approaches by other inquiring minds. The following speculations may occasionally sound apodictic, whereas the real spirit behind the written word is that of James Thurber--“Speculations when confined to certainties are eased of their wonder and warmth.” The basis for the establishment of acetylcholine as the parasympathetic neurohumor has been furthered mainly by: (1) maximal nicotinic (N) and muscarinic (M) potency of ACh; (2) extraction from the spleen of the horse or ox (Dale and Dudley, 1929; Gollwitzer-Meir and Kruger, 1934) ; (3) bio-assay on various tissues in comparison to congeners (Chang and Gaddum, 1933); (4) extraction from the adrenal glands of the horse (Feldberg and Schild, 1934); (5) extraction from mammalian brain tissue of ACh synthesized in vitro (Stedman and Stedman, 1937) ; and (6) bio-assay of rat brain by use of the frog rectus (Richter and Crossland, 1949). Some of these workers use sodium acetate to salt out the neurohumor which may allow acetate ester formation with choline. We, therefore, still lack the fine proof of paper chromatography and other tests to establish acetylcholine as a neurotransmitter. Banister et a2. (1953) finds propionyl choline in mammalian tisaues and Nachmansohn and Wilson (1951) report an “enzyme formed factor” which is slightly more potent than acetylcholine. More precise data are needed on the structure of parasympathetic neurohumors which might occur in human tissuetiparticularly the brain. Careful chemical extraction and identification of the acetylcholine-like substance which is reported to occur in the human placenta might be profitable (Chang and Gaddum, 1933). From SAR studies one would expect that the a-carbon atom of any naturally occurring acid which is esterified with choline should be asymmetric. A 30-fold difference in activity occurs between the D- and L-isomers of hyoscyamine (atropine is DL-hyoscyamine) . Neither acetic nor propionic acids.have such an asymmetric carbon atom. This substituent is most likely a hydroxy group. The possible acids may therefore be lactic, glyceric, or others with optical isomerism in the a-carbon atom. SAR studies provide more serious incongruity in regard to the choline portion of the ACh molecule. (1) Various names have been suggested to differentiate serum (pseudo) cholinesterase from ACh (true) cholinesterase (CHE). Serum CHE will hydrolyze many choline esters while true CHE will hydrolyze most specifically p-methyl ACh (methacholine) . This suggests a /3-substitution on one of the parasympathetic transmitters. (2) Greig and Howell (1948) studied the CHE inhibition produced by the d- and 2-isomers of the analgesio drugs methadone and

PARASYMPATHETIC NEUROHUMORB AND BEHAVIOB

233

isomethadone. The former has an a-substitution and the latter has B 8-substitution which respective substituents provide the optical isomerism. For serum CHE the isomeric ratio (Pfeiffer, 1956) is only 2 for both methadone and isomethadone. For ACh (true) CHE the isomeric ratio for d- vs. Z-methadone is 12 while for the p-substituted isomethadone the isomeric ratio is 30. This again indicates the need for a p-substituent on the parasympathetic neurohumor which is hydrolyzed by true CHE. (3) Finally, if in cerebral tissue the pharmacological effect of epinephrine may be that of modulation of the ACh effect to produce a transient degree of atropine-like action, then the p-hydroxy group on the epinephrine molecule (which has an isomeric ratio of 20) should be mirrored by a p-hydroxy or isosteric group on the choline molecule. Summary. Various unknown parasympathetic neurohumors may be more important in human physiology than the two esters which are thought to occur in mammalian tissue, namely acetyl and propionylcholine. SAR analysis indicates that an asymmetric a-carbon atom may occur in the acid fragment and that a p-substituent (possibly hydroxy) on a choline-like compound might be present to provide an asymmetric p-carbon. The neurotransmitters may of course be represented by a family of compounds rather than by a single compound. The compound should have greater muscarinic effect than acetylcholine which should occur with a 8-carbon substitution (see next Section).

XVIII. The SAR of Muscarinic and Nicotinic Ends of Acetylcholine Congeners

A study of the pharmacological action of simple congeners of acetylcholine leads to the suggestion that the ester end is mainly responsible for muscarinic activity (M) while the onium head represents predominantly the nicotinic activity (N). This suggestion is made in full cognizance of the possibility that by so doing we are converting the jig-saw puzzle of acetylcholine into a see-saw puzzle. Minor modifications of either end of the ACh molecule shift M and N effects as though they were truly in pharmacological balance (Pfeiffer and DeMaar, 1957). Selected pairs of singly substituted ACh derivatives from the early study of Simonart (1934) are used to illustrate this point (see Example 1)Compound A, which is a choline ether, has comparatively much greater loss of muscarinic effect than nicotinic effect because the predominant change is at the M end of the molecule. If a p-methyl group is added (to the muscarinic half) then a 10-fold increase in M effect occurs accompanied by a 5000-fold decrease in N effect. The methyl ether is not hydrolyzed by cholinesterase so the usual explanation for the in-

234

CARL C. PFEIFFER

EXAMPLE 1

(B) CH~+CH

CH&(CH,),

1/10

1/1o,OOo

AH8

creased potency of methacholine, namely that of greater difficulty in hydrolysis, cannot be a factor. EXAMPLE 2. (Simonart,1932) Compound Acetylcholine 0 (A) Et-

8

-0-CH&H2&(CH& 0

8--O-CH&Hz&(CHa)a

(B) Propyl~~~

~

Muecarinic activity

Nicotinic activity

1

1

1/5

2.6

1/100

6

~

In Example 2, compound A which is propionylcholine has been written so as to emphasize the fact that an ethyl group has replaced an active methyl group on the muscarinic side of the ACh molecule. The expected decrease in M effect is accompanied by an increase in N effect. In compound B (butyrylcholine) the methyl of ACh is replaced by a propyl group which further diminishes M effect and increases N effect. EXAMPLE 3. (Simonart, 1932) Compound

Muscarinic activity

Nicotinic activity

2.5

1/100

1

Example 3 shows that in identical propionylcholines a 8-methyl (near M end) inoreases M and decreases N effect. An a-methyl (near N end)

235

PARASYMPATHETIC NEUROHUMORB AND BEHAVIOR

increases N (compared to B) and decreases M effect. Compared to A the nicotinic effect is not even restored. This may be governed by the proposed requirement for a p-substituent (see previous Section). EXAMPLE 4. (Edwards and Marsh, 1951) Compound

Muscarinic activity

Nicotinic activity

1

1

1/28

3/4

1/150

314

1/75

2/5

(E) C H ~ - C H ~ O - C H ~ C H ~ & ( C H , ) ~ 1/15

114

(A) Acetylcholine (B) WAmyl-&(CHa)a

0

(c)

1

C H ~-cH,cH,&(cH~),

0

II

(D) C H ~ C - C H ~ C H , C H ~ & ( C H ~ ) ~

In this series (Example 4) of congeners the onium head is kept constant while the ester end is varied. The N effect is relatively constant, from 1/4 to 3/4 the activity of ACh, while the M effect varies from 1/15 t o Compound E, which has the greatest M effect, has the least N effect. This same balance can be noted in other congeners of ACh where the authors do not provide a concise comparison to ACh. Fellows and Livingston (1942). Compound

Muscarinic activity

Nicotinic activity

100

1.6

Compound A has both nicotinic and muscarinic actions. When an ether oxygen is introduced as in the furane ring of compound B the muscarinic action increases 10-fold while the nicotinic action decreases to I/a that of compound A. A similar phenomenon occurs with the a-methyl congeners. Muscarinic activity 1

6

Nicotinic activity

236

CARL C. PFEIFFER

The ether oxygen reverses the M and N potencies. Note again that the a-methyl does not cause the N potency of compound C to become greater than that of compound A. These data are mainly from the anesthetized cat or dog where the depressor response is recorded as muscarinic and the pressor response l0,OO

1001

\

\

I

I

I I

I01

I

I

I

II I

I

I

‘I

II

I(

I I

I

‘II I

L 1 2

‘II

I 4

L 6

8 10 COMPOUNDS

I2

A 14

-

&

16

18

20

E’ro. 2. Acetylcholine congeners are graphed according t o increase in muscarinic potency, The corresponding nicotinic potency decreases.

after atropine is the nicotinic effect. From data in the literature (Bovet and Bovet-Nitti, 1948; Jacob et aE., 1952; Kogl et al., 1957; and Fraser, 1957) one can obtain nineteen compounds which have been studied on blood pressure for depressor (M)effect and in the atropinized animal for pressor (N) effect. These compounds have been arranged in increasing order of (M) potency in order to portray graphically the inverse pro-

237

PARASYMPATHETIC NEUBOHUMORS AND BEHAVIOR

TABLE XVI DATAOF COMPOUNDS TAFIULATE)D ON

No.

Name

1. Choline phenyl ether

THE)

X-AXIS OF FIQURE 2 Muecarink Nicotinic activitya activity"

Formula QOCH&H&(CHa)a

0

3000

0 0 2. n-Butyrylcholine 3. n-hyl-trimethylammonium 4. 4Hydroxy-n-amyl-

tmethylammonium

CdL-

6. 6-Hydro~+t-amyl

\CHzCHd (CH,),

CHaCHzCH&H*CHn&(CH&

8

20

76

25

3000MxKl

OH CHdHCH&H&HJ(CHl)l OH

5. PHydroxy-n-2,3-pentane

trimethylammonium

&/

26

C&H-CH=CH-CH&

(CHa),

HOCH~CH~CHPHPH,~~ (cH,)~

1600u")

25

trimethylammonium (pentylcholine)

0 0 7. Acetyl-a-methylcholine

C H r y \CHrCH-&(CH$,

50

lo00

LHa

0 0 8. Propionylcholine

CH:CH!/

200

\CHzCH8&(CH&

9. 4Hydroxy-n-2,3-penOH tynyltrimethylammoI nium iodide CHAHC=C-CHa$ 10. Choline vinyl ether

500-

(CH,) 8

HZC=CHOCH&H$(CH:)a

16002500 2000-

2000

3000

lo00

>ACh

lo00

lo00

1000

International Review of Neurobiology Volume 1

International Review of Neurobiology, Volume 6

International Review of Neurobiology Volume 25

International Review of Neurobiology Volume 38