History of the Synapse
Dedicated to Bernard Katz and to the memory of John Newport Langley
History of the Synapse Max R.Bennett University of Sydney, Australia
harwood academic publishers Australia • Canada • France • Germany • India • Japan • Luxemburg Malaysia • The Netherlands • Russia • Singapore • Switzerland
This edition published in the Taylor & Francis e-Library, 2005. “To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.”
Copyright © 2001 OPA (Overseas Publishers Association) Amsterdam N.V. Published by license under the Harwood Academic Publishers imprint, part of the Gordon and Breach Publishing Group. All rights reserved. No part of this book may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying and recording, or by any information storage or retrieval system, without permission in writing from the publisher. Printed in Singapore. Amsteldijk 166 1st Floor 1079 LH Amsterdam The Netherlands British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN 0-203-30254-0 Master e-book ISBN
ISBN 0-203-34506-1 (Adobe eReader Format) ISBN 90-5823-233-6 (Print Edition) Cover: Hippocampal neuron stained for MAP2 together with synaptotagmin. Courtesy M. Matteoli and © Cell Press. Cover design by Lee McLachlan.
Contents
Introduction
ix
The early history of the synapse: from Plato to Sherrington
1
1.1
Plato and Aristotle: ‘vital pneuma’ is necessary to initiate organ function
1
1.2
Galen: pneuma is conducted and transmitted from nerve to muscle
3
1.3
Descartes: the replacement of pneuma by mechanical corpuscles
4
1.4
Borelli: a corpuscular description of conduction and transmission
6
1.5
Fontana: nerves are composed of many cylinders along each of which conduction occurs
8
1.6
Galvani: electricity is conducted and transmitted not corpuscles
9
1.7
Matteucci and du Bois-Reymond: transient electrical changes are conducted (the action potential)
12
1.8
Helmholtz: the action potential has a finite velocity
15
1.9
Kuhne and Auerbach: identifying the structure of nerve endings on muscle and neurons
16
1.10
Cajal: nerve endings are not continuous with the cells on which they impinge
18
1.11
Sherrington: the adoption of the word ‘synapse’
22
Emergence of the concept of transmitter release at peripheral and central synapses
26
2.1
Research on the synapse in the laboratories of Sherrington and Langley before the Great War
26
2.2
Sherrington’s concept of the inhibitory and excitatory states of central synapses
28
2.3
Lucas, Adrian, and the electrical concept of the inhibitory state of central synapses
29
2.4
Loewi, Dale and Eccles examine the inhibitory state at autonomie neuromuscular junctions
32
2.5
Eccles develops the electrical concept of the excitatory state at autonomie synapses
32
2.6
Katz, Kuffler and Eccles establish the motor endplate as the paradigm synapse for electrophysiology
35
2.7
Eccles elucidates the electrical signs of the inhibitory and excitatory states of central synapses
36
2.8
Katz’s concept of quantal transmitter release at the motor endplate and the vesicle hypothesis
39
2.9
Conclusion: the establishment of Sherrington’s concept of the synapse in the central nervous system and central synaptic transmission
41
The discovery of acetylcholine and the concept of receptors at synapses
43
3.1
Introduction
43
3.2
Claude Bernard and curarization: the notion of an intermediate zone between nerve and muscle
44
3.3
Paul Ehrlich and the idea of the ‘receptive side chains’ of cells
46
3.4
John Langley and T.R.Elliott: the emergence of the concept of chemical transmission between sympathetic 46 nerves and smooth muscle
3.5
The action of curare and John Langley’s development of the idea of transmitter receptors
51
3.6
The Langley-Ehrlich receptor theory
54
Chapter 1
Chapter 2
Chapter 3
vi
3.7
The discovery of acetylcholine and its physiological action at autonomic neuroeffector junctions
54
3.8
The physiological action of acetylcholine in autonomic ganglia
59
3.9
The identity of acetylcholine as the transmitter substance at somatic neuromuscular junctions
60
3.10
The discovery of the physiological action of single acetylcholine receptors
61
3.11
Conclusion
64
The discovery of adrenaline and the concept of autoreceptors at synapses
65
4.1
Introduction: the discovery of noradrenaline as a transmitter
65
4.2
Early observations leading to the idea of autoreceptors
65
4.3
Direct experimental evidence for autoreceptors
68
4.4
Identification of presynaptic adrenergic autoreceptors different from postsynaptic adrenergic receptors
71
4.5
Presynaptic adrenergic autoreceptors in the central nervous system
72
4.6
Evidence that endogenous autoreceptor mechanisms exist
72
4.7
The ionic basis of the action of alpha 2 adrenoceptors
73
4.8
Conclusion
76
The discovery of amino acid transmission at synapses in the central nervous system
77
5.1
Introduction
77
5.2
Identification of excitant and depressant amino acids
77
5.3
Glycine accepted as an inhibitory transmitter in the spinal cord
78
5.4
The emergence of GAB A as an inhibitory transmitter in the brain
81
5.5
L-Glutamate as a neurotransmitter: synaptic excitation, ion fluxes and neurotransmitter transporters
84
5.6
NMDA receptors: the first amino acid receptor identified at central excitatory synapses
86
5.7
Non-NMDA receptors at excitatory synapses
87
5.8
GAB A receptors
87
5.9
Conclusion
87
Monoaminergic synapses and schizophrenia: the discovery of neuroleptics
90
6.1
Introduction
90
6.2
Chlorpromazine
90
6.3
Haloperidol
91
6.4
The dopamine hypothesis for neuroleptics
91
6.5
Dopaminergic projections in the brain
92
6.6
Identification of the D1-like and D2-like dopamine receptors
92
6.7
Determination of different classes of dopamine receptors
95
6.8
Clozapine
95
6.9
Distribution of D1 and D2 receptors in the striatum of schizophrenics
98
Chapter 4
Chapter 5
Chapter 6
6.10
Mixed aminergic actions of the neuroleptics: serotonin and dopamine receptor blockade
102
6.11
Cellular and molecular mechanisms of action of dopamine receptors
102
6.12
The time course of action of neuroleptics on dopamine receptors and the emergence of antipsychotic effects 103
vii
6.13
Conclusion
104
The discovery of transmitters other than noradrenaline and acetylcholine at synapses in the peripheral nervous system
105
7.1
Introduction: J.N.Langley, H.H.Dale and non-adrenergic, non-cholinergic (NANC) transmission
105
7.2
Parasympathetic neuromuscular junctions in the gastrointestinal tract: mechanical studies
109
7.3
Parasympathetic neuromuscular junctions in the gastrointestinal tract: electrophysiological studies
116
7.4
NANC transmission: the new autonomic paradigm
121
7.5
Contemporary views on the identity of NANC inhibitory transmitters
127
7.6
Ionic mechanisms involved in generating the IJP
132
7.7
The secretion of NANC transmitters responsible for the IJP
133
7.8
NANC excitatory transmission in the gastrointestinal tract
134
7.9
Conclusion
135
Development of the concept of a calcium sensor in transmitter release at synapses
136
8.1
Introduction
136
8.2
Calcium is necessary for the release of transmitter
136
8.3
Electrophysiological evidence that calcium is necessary for the release of transmitter: the concept of a calcium sensor for secretion
137
8.4
The calcium action potential
140
8.5
Are calcium movements across the nerve terminal necessary for evoked secretion?
143
8.6
Direct evidence for calcium entry across the nerve terminal membrane during an impulse
147
8.7
Calcium channels in the nerve terminal
149
8.8
Identification of the calcium sensor molecule
158
8.9
Conclusion
162
The discovery of quantal secretion and the statistics of transmitter release at synapses
164
9.1
Introduction
164
9.2
Evoked quantal release as a binomial or Poisson variate
165
9.3
Kinetics of release of a quantum
176
9.4
Maximum likelihood estimation of parameters in statistical models of quantal release
186
9.5
Autocorrelation function used to detect quantal release
190
9.6
Model discrimination: statistical methods for discrimination between different statistical models of transmitter release
193
9.7
Appendix
195
The discovery of long-term potentiation of transmission at synapses
216
10.1
Introduction: the hippocampus, memory and long-term potentiation (LTP)
216
10.2
LTP at synapses in the brain
219
10.3
The induction of associative LTP in the brain
220
10.4
The maintenance of associative LTP
226
10.5
Biochemical pathways implicated in the maintenance phase of associative LTP
229
Chapter 7
Chapter 8
Chapter 9
Chapter 10
viii
10.6
Evidence that associative LTP is involved in memory
230
10.7
The induction of non-associative LTP
233
10.8
Summary and Conclusion
233
Emergence of the concept of synapse formation molecules
238
11.1
Introduction
238
11.2
Synapse formation in muscle
238
11.3
Synapse formation molecules in muscle and the elimination of polyneuronal innervation
243
11.4
Elimination of polyneuronal innervation during muscle development described by dual-constraint theory
247
11.5
Elimination of polyneuronal innervation during reinnervation of muscles described by dual-constraint theory
255
11.6
Elimination of polyneuronal innervation and establishment of topographical maps in muscle described by 262 dual-constraint theory
11.7
Identification of the synapse formation molecules
271
11.8
Synapse formation in autonomic ganglia
279
11.9
Elimination of polyneuronal innervation in autonomic ganglia
279
11.10
Elimination of polyneuronal innervation during reinnervation of ganglia
280
11.11
Identification of synapse formation molecules in autonomic ganglia
281
11.12
Conclusion
282
Epilogue
286
References
288
Illustration Acknowledgements
328
Acknowledgements
332
Index
334
Chapter 11
Introduction
This work is an attempt to provide a history of those discoveries concerning the identification and function of synapses which provide the foundations for research during this new century. It is written in the conviction that errors in the development and application of contemporary concepts to the understanding of synapses arise if there is failure to probe the origins of the scientific paradigm at present in use. The idea that blood vessels are the means by which muscles are activated lasted some five hundred years, from Aristotle in the fourth century BC to Galen in the second century AD, when it was shown that nerves are the means of communication to muscle. The concept that the ventricles of the brain, containing the psychic pneuma identified by Aristotle, are the sites at which sensations, thinking and memory are experienced lasted for a much longer period. The most posterior ventricle was taken as initiating the flow of psychic pneuma to the nerves and hence to muscles for their activation. This concept lasted for over fifteen hundred years, from shortly after Galen to Thomas Willis at the end of the seventeenth century. Is the historical development of these facts, involving both the discoveries and thoughts of men of genius, to be regarded as an oddity divorced from ‘the common-sense’ which we now bring to the solution of problems regarding the workings of synapses? The first chapter considers the wonderful story, evolving over two and a half thousand years, of how progress was made in the establishment of a conceptual structure that would allow the synapse to be identified at the beginning of the twentieth century. It was founded on the idea of conduction and transmission to muscle by Aristotle, the identification of nerves as providing the conducting medium by Galen and on Galvini’s discovery that it is electricity that is conducted, not Cartesian corpuscles. The concept of the synapse which was accepted for most of the twentieth century, was certainly not in place at the end of the nineteenth century through the work of Sherrington, as is popularly accepted. Rather, he highlighted the fact that a mechanism must be sought by which conduction could proceed from one excitable cell to another following the discovery of Cajal that neurones are separate cells. However, this difficulty had been a focus of physiological enquiry for centuries, particularly to Galen and to Descartes’ contemporary Borelli. In naming this region of apposition between excitable cells ‘the synapse’ Sherrington helped focus the physiological enquiry which was to be so brilliantly brought to fruition by his mentor at Cambridge, John Newport Langley. The genius of Langley is exemplified in the experiments he performed. These first drew attention to the similarities between the effects of sympathetic nerve stimulation on autonomic effectors and that of extracts of the adrenal glands, leading to the audacious speculation of his student Elliott that electrical conduction is transmitted chemically and that in the case of sympathetic nerves this chemical is adrenaline (Chapters 2 and 3). However, a reading of the papers of that period (the beginning of the twentieth century) points up that it was Langley’s discovery of transmitter receptors at the somatic neuromuscular junction that was the pivotal work in establishing the concept of the synapse. Langley championed the idea of the receptor in the face of fierce opposition, particularly from the founder of immunology, the chemist Paul Ehrlich. It was Langley’s work that gave us the modern concept of the synapse, namely an area of apposition between a nerve ending and another cell at which a chemical substance is released from the ending onto a ‘receptive substance’ found on the cell. All of these definitive concepts which gave ‘substance’ to the abstract term ‘synapse’ were established in a period of about eight years, principally from the one laboratory! The subsequent identification of acetylcholine as the chemical released from motor nerve terminals had to wait several more decades (Chapter 3). It was primarily the laboratory in London of Dale, a former student of Langley, together with that of Loewi in Germany that was then responsible for generalising the principle of chemical transmission at peripheral synapses. The study of central synaptic transmission did not reach a level of sophistication comparable to that at peripheral synapses for nearly half a century after the period when the idea of transmitter substances and their receptors was first conceived. The 1950s were a remarkable period for central synaptic transmission. First amino-acids were identified as likely central transmitters (Chapter 5) and then neuroleptic agents were synthesised and shown to act at central monoaminergic synapses in ways which were to have a profound impact on the alleviation of mental suffering, such as in schizophrenia (Chapter 6). It was also at this time that the mechanism of synaptic transmission was greatly illuminated due to introduction of the glass microelectrode for studying the same junction that gave us the concept of the receptor, namely the somatic neuromuscular junction. Katz and his colleagues showed that transmitter release occurs in units (Chapter 9) and that the probability of release
x
of these units is controlled by calcium ions. The search for the calcium sensor upon which these ions act has become a major research program of contemporary neuroscience (Chapter 8). The paradigm established by Langley, Elliott, Dale and Loewi, namely that adrenaline and acetylcholine are the transmitters at peripheral autonomic and somatic synapses, was shown to be incorrect in the early 1960s. At this time evidence was provided through application of glass microelectrode techniques to autonomic effectors, that at least two other transmitters must be active in synaptic transmission. The synapses using these transmitters were termed non-adrenergic noncholinergic (NANC) (Chapter 7). The subsequent discovery that these NANC transmitters include nitric oxide, neuropeptides and ATP has substantially contributed to our understanding of the process of chemical transmission. As the pace of research on synapses accelerated in the 1970s, three further discoveries were made that have now become principal foci of research. The first of these was the phenomenon of long-term potentiation of transmitter release whereby some nerve terminals greatly increase their efficacy of transmission following a high-frequency train of impulses, with this increased efficacy maintained for very long periods of time in excess of hours (Chapter 10). The second was the discovery that synapse formation molecules exist, of the kind originally envisaged by Langley at the beginning of the century, and that these are laid down on muscle cells by the motor-nerve terminal during early development (Chapter 11). This initiated a research program to identify such molecules, which is another area of intense activity amongst the contemporary synaptic neuroscience community. Finally, the 1970s saw the first analysis of transmitter receptors at the single receptor level by Sakmann and Neher (see Chapter 3). Suitably, they studied the receptors of the somatic neuromuscular junction to introduce their technique of patch-clamping, the same preparation that Langley had used seventy years earlier to give us the idea of transmitter receptors in the first place. This history provides a personal view of the process by which new concepts concerning the workings of the synapse have developed. The emphasis is naturally then on those aspects of synaptic transmission that have fascinated me over a forty-year period of research and brought me so much pleasure. I have written this book in the hope that others might share in the excitement of synaptic physiology and perhaps even help them in placing the development of their own concepts and research in an historical perspective. It will be clear to some that I have over emphasised those areas of research to which I contributed, such as the discovery of synapses that are non-adrenergic non-cholinergic (NANC) in the peripheral nervous system (Chapter 7), the discovery of the calcium action potential (Chapter 8), as well as the discovery that synapse formation molecules exist at the somatic neuromuscular junction (Chapter 11). On the other hand very important areas of research have been left out completely, such as the delivery of neurotrophic molecules to nerve terminals and the modification of proteins at synapses that determine the efficacy of transmission. I ask the readers indulgence in these distortions of perspective. If this work should seem worthwhile to some then these omissions will be rectified in the future. It will be a pleasure to receive from readers their own views of the events chronicled here which might then lead to adjustment of my perspective on this great subject. I have attempted an intellectual history of the synapse, charting the development of concepts and insights regarding synaptic structure and function that can be gleaned from reading primary sources, namely those research papers and occasionally other scholarly works in which the research was originally announced. I apologise to the many students of synaptic transmission whose work has not been quoted where they may see it as relevant. However, I have had to make difficult judgements as to the extent to which the literature should be referenced on a particular topic. This work does not provide much in the way of detail concerning the controversies surrounding the emergence of new ideas and claims of priority of discovery, issues that often involve the dramatic clash of personalities that can make the narrative of a history come ‘alive’. However, this detail has been provided in a few cases where it is necessary to understand the issue under consideration. Such cases arise in the disagreement between Langley and Ehrlich concerning the possibility of transmitter receptors at the beginning of the twentieth century (Chapter 3) as well as in the arguments during the middle of that century concerning the idea that there must be several transmitters in the peripheral nervous system other than just noradrenaline and acetylcholine (Chapter 7). It emerges again in the incisive arguments that the biophysicists such as Adrian and Lucas used in the 1920s to argue for electrical transmission in the central nervous system and that Eccles used at that time in favour of such transmission in the peripheral nervous system, all in opposition to the arguments of Sherrington on the one hand and Dale on the other (Chapter 2). Such disagreements emerge again concerning the concept that either a fluid or a set of corpuscles must pass from nerve into muscle at the site of transmission in order for a muscle to shorten, championed in their different ways by Aristotle, Galen, Descartes and Borelli. This idea was negated by the definitive experiments of Swammerdam in the eighteenth century and by the experiments of Galvani soon after showing that it is electricity that is conducted and transmitted at this junction, not fluid or corpuscles. Some will argue that there is an excessive number of illustrations in this history, which is not balanced by either an appropriate depth of analysis of the experiments that are highlighted nor of sufficient attention to many aspects of the history of the synapse that have been neglected. All I can say in reply is that I have tried to keep as close to the raw data of the experiments as possible. This has seemed best accomplished through figures explaining either the technical approach or the experimental finding. However no excuse is offered for failing to present a more comprehensive history, which I hope to produce at some time in the future, especially after receiving the criticisms of my fellow neuroscientists.
xi
The chapters in this book were previously published as essays on the history of various aspects of synaptic function in response to requests from my colleagues. This soon turned into a fascinating intellectual journey, which provided me with the foundations for a subject that I have found so personally satisfying. I hope the reader is as stimulated as I have been in surveying our efforts to understand the workings of the synapse.
1 The Early History of the Synapse: from Plato to Sherrington
1.1 Plato and Aristotle: `vital pneuma' is necessary to initiate organ function In the beginning there were four elements, fire, air, water and earth. Various proportions of these composed the blood, muscle, bone, tendons and nerves from which the body or soma was constructed. The ingredients of the blood determined intelligence so that the heart was the basis of the intellect and of mental life. This pre-Socratic idea of the fifth century BC, principally due to Empedocles, was developed further by Democritus who considered that each of the four elements was composed of a different kind of particle. He argued that the psyche or soul was composed of the lightest, fastest moving and most nearly spherical particles which are to be found throughout the body, especially concentrated in the brain. Particles of a lesser quality were to be found in the heart giving it a role in emotion whilst the most coarse particles were located in the liver, responsible for functions such as lust. Plato, in the fourth century BC, assigned specific geometrical shapes to each of the four kinds of particles. In addition he confronted the problem which these pre-Socratic ideas presented of how to relate the psyche to the body. A living thing for Plato was matter properly arranged to permit effectual intervention of the soul. Following Democritus, he claimed that there were three different kinds of psyche, namely that concerned with rational thought and behaviour which was associated with the head, that involved with passion and the emotions associated with the breast and the heart therein, and that concerned with desires which was associated with the liver. Only the rational psyche was immortal (Finger, 1994; Gross, 1998; White, 1996). The problem of what form the association between soul and body took was formulated in terms of the geometrical principles that played such a large role in Plato’s cosmology (Conford, 1956). As he considered the fundamental units of the elements themselves to be geometrical figures, such as the triangle, so the body composed of these elements must ultimately be thought of in mathematical terms. It was the appropriate organisation of these geometrical figures from which the body was ultimately composed that allowed the bonding of the soul to the body. It was only through such bonding that the soulbody complex could manifest life-as-action. Plato placed this bonding in what is now called the nervous system. In his ‘Timaeus’ he describes the soul as bonded to a substance that is found in its purest form in the cranial and spinal cavities where it appears as ‘marrow’, or what is now called brain and spinal cord. The marrow is the primary life stuff in which ‘were fastened the bonds of life by which the soul is bound to the body’. This marrow is not composed of the four elements, or rather of the elementary geometrical figures which make up the elements, but of specially well-formed examples of those triangles which are the common components of these elements (Hall, 1975). Thus Plato, following Pythagoras, developed the notion of the body as a temporary receptacle for a separate soul, which was associated with rationality, located in the marrow or nervous system, and which could pass from one body into another at death. The other kinds of soul, those associated with the emotions and desires and therefore with the heart and liver, were not capable of this transmigration. Physiological function of an organ was considered in terms of the associated psyche or soul giving life to the propensity of the organ to carry out its function. In this way organs came to be seen as possessing faculties or propensities to carry out a physiological act that was energised by the psyche (Plato, 1892). Aristotle, in the fourth century BC, developed radically different concepts concerning the functions of the body and soul that were to have a profound influence on physiological thought concerning the activation of organs and of muscle for two thousand years [see Aristotle, 1984, Fig. 1.1]. The Aristotlean concept of the soul will be considered in some detail here as Aristotle’s ideas are so different from those of Descartes, which embody a dualism like that of Plato which still dominates thinking on these issues to this day (Cottingham, 1995; Everson, 1996). For Aristotle the soul or psuche or psyche was the form of the thing under consideration. This form constituted the reason for a thing being as it is and could be considered as providing explanations for what it is made of (the material cause), what actually makes it (the efficient cause), what shape is used to identify it (the formal cause) and the ultimate reason for its existence (the final cause). Thus in the case of a muscle, the material cause is the fibres that it is made of, the efficient cause is the grouping of the fibres in relation to each other, the formal cause is that this grouping is done to a particular design in order
2
EARLY HISTORY OF THE SYNAPSE
Fig. 1.1. Theories of the nervous system before Descartes: conduction and transmission of psychic pneuma. (A) This anonymous fifteenth-century drawing illustrates pre-Cartesian brain theories, which followed the views of Aristotle. The senses of touch and taste are shown connected to the heart, while the boxes on the head are the ‘cerebral cells’ where mental faculties such as memory and fantasy are located. Anonymous, 15th Century, Bayerische Staatsbibliothek. Munich (from Posner, 1994). (B) A 15th century illustration due to Gregor Reisch, showing the four routes of communication connecting the organs of taste, smell, seeing and hearing to the anterior cerebral ventricle. This is divided into sensus communis, fantasia and imaginativa. The vermis connects it to the second ventricle, with the two faculties called cogitativa and estimativa, whereas the faculty spoken of as memorativa is in the third. The curving lines around the ventricles can be interpreted as representations of the brain’s convolutions (from Biblioteca Nazionale Centrale, Florence; see also Corsi, 1991; The Enchanted Loom). (C) In the 4th century A.D. and for several hundred years to follow, the faculties of the mind were thought to be housed in four ventricles of the brain as in (B). A 15th Century illustration, with indistinct lettering, designed to illustrate the 1494 edition of Aristotle’s de Anima. Four regions of the brain are labelled: sensus communis, virtus cogitativa, virtus imaginativa, and memoria (Courtesy of the Incunabula Collection at the National Library of Medicine, Bethesda; from Brazier, 1984).) (D) Leonardo da Vinci’s localization of the sensus communis. In the lower left of the figure his own words state: ‘Where the line a-m is intersected by the line c-b there the meeting place of all the senses (sense commune) is made’. (Courtesy of the Royal Collection © 2000, Her Majesty Queen Elizabeth II; this work later attributed to Cesare da Sesto).
to produce a muscle of a particular shape and the final cause is the fulfilment of the purpose of the muscle, which is to contract and produce, say, the movement of a limb. In this way an organ’s form or psyche is not material but is inherent in the organ and cannot exist separate from the organ. If then the constituents which make up the form are specified, so is the soul or psyche. In this way Aristotle lays stress on the activities of living things and on the distinction between ‘living’ and ‘dead’
HISTORY OF THE SYNAPSE
3
rather than in the distinction emphasised by Plato between ‘mental’ and ‘physical’. ‘Mind’ does not figure largely in Aristotle’s work, perhaps because of his emphasis on the psyche, that is on the activities of living things such as organs. If the psyche disappears from an organ it then ceases to be such a thing except in name only. The loss of psyche, of the soul of a living thing, means it ceases to exist (Hall, 1975). The concept of the psyche of an organ was not abandoned until Descartes. Meantime, the effect of Aristotle’s ideas was to lead to the search for the form of an organ so that scholars sought to identify the psyche of an organ (Clarke, 1995). This had two effects: first, it lead enquiry away from the mechanical workings of the organ; second, it placed emphasis on the final cause component of the psyche, that is the potential of an organ to carry out its function, of how the potentiality of the psyche of an organ could be realised, a problem which will now be considered. For Aristotle the heart was the central organ of perception, rather than the brain as postulated by Plato. If an animal’s perception gives rise to action, that is to the contraction of muscle leading, for example, to locomotion, it will occur as follows (Everson, 1996): ‘if the region of the origin (i.e. the heart) is altered through perception and thus changes, the adjacent parts change with it and they too are extended or contracted, and in this way the movement of the animal necessarily follows’. According to Aristotle, perception occurs in the heart with its particular psyche (Fig. 1.1 A). Perception is not an activity that involves two different substances, as later suggested by Descartes, for the affections of the psyche are common to the psyche of the body. ‘It is apparent that all the affections of the psyche are with the body…in all these the body undergoes some affection.’ The central sense organ is therefore the heart, which is connected to the individual sense organs (Fig. 1.1 A & 1.1C). When these are affected by their objects, the affections pass through the blood stream to the heart. Thus the movements around the heart bring about the movements of the limbs by acting through the blood stream. An organ ceases to be an organ if separated from the body as its psyche no longer exists. It is only as a consequence of being part of the body that its psyche is intact, which includes the final cause or ultimate reason for the organs existence. Given that the heart is the centre of perception and of the appetites and responds to these by initiating animal motion it is responsible for the activation of the muscular organs. The key question which now arises is how does the psyche of the heart conduct and transmit through the blood stream information to the psyche of the muscular organs which is responsible for their final cause, that is contraction. The emphasis of Aristotle on the natural world rather than the Platonic mathematical world led him to consider the most likely method for conduction and transmission from the heart to a muscle through the blood stream in terms of the elements. To the organs whose substance was made up of the four elements (fire, air, water and earth) he introduced a fifth element. This element was not restricted to this world only but also belonged to the stars and heavens, so that it permeated the entire universe: he named it the ‘ether’. The concept of the ether was to have a major impact on both the physical as well as the biological sciences. Aristotle considered that the ether element was taken into the body during breathing and conveyed from the lungs to the heart in which it was transformed to ‘vital pneuma’ or ‘vital heat’ (Peck, 1943). This vital pneuma was then distributed from the heart throughout the body by blood vessels, where it was able to mediate between the psyche of the heart and the psyche of the organs including muscles. It is then vital pneuma that is conducted from the heart along the blood vessels to be transmitted to the muscles and in so doing initiating their final cause, contraction (Hall, 1975). Aristotle had in one brilliant stroke introduced a means for mediating between the psyche of the heart and the psyche of muscles by introducing a fifth element that was not just of this world but seemed to possess a heavenly property associated with the stars. But the great contribution here is the introduction of the concept of a substance of a kind, be it of somewhat mysterious qualities, which had to be conducted to an organ to allow it to function, even though this function was taken to be simply the ability of the organ to release its propensities to action as dictated by the final cause of its psyche. 1.2 Galen: pneuma is conducted and transmitted from nerve to muscle Galen and his students, in the second and third centuries AD, greatly refined the concept of the conduction of pneuma to the organs of the body (see Galen, 1978; Galen, 1968; Galen, 1962). They retained Aristotle’s conceptual scheme with the four elements constituting the tissues and organs of the body, and a fifth composing the vital pneuma acting as a mediator for the psyche to give life to the organ and allowing it to release its propensities for action. Erasistratus argued that the pneuma of the inspired air became vital pneuma as it passed from the bronchioles of the lungs via the intrapulmonary veins to the pulmonary vein and into the heart. The heart on dilation sucked in the pneuma from the pulmonary vein and on contraction forced the vital pneuma to the rest of the body through the arteries. Blood is carried by veins not arteries. The vital pneuma which reaches the brain in this way is converted to ‘psychic pneuma’ there from which it travels outwards along nerves (Galen, 1821– 1833). The brain, rather than the heart, becomes once more the centre of perception in this scheme. Galen had already established that nerves arise from the brain and spinal cord, that conduction of psychic pneuma is necessary in these nerves for sensation and motor action for if they are cut or damaged there is no sensation or movement, and that there are two classes
4
EARLY HISTORY OF THE SYNAPSE
of nerves, one motor (if damaged no motor action) and the other sensory (if damaged no sensation). It was therefore estab lished that sensitive psyche possessed its own nerve supply as did the locomotor psyche. These observations and speculations of Galen and his students set the stage for the consideration of the mechanism of conduction of psychic pneuma along motor nerves and of the transmission of pneuma into muscle. Galen comments ‘All muscles require to receive a nerve from the brain or from the spinal cord and this nerve is small to behold but by no means slight in power’. Three possibilities for the conduction of the effects of psychic pneuma were entertained: one, that the psychic pneuma flows along the nerves like a liquid along a conduit; second, that the psychic pneuma in the brain pushes the pneuma resident in the nerve so that some is released at the ends of the nerves; finally, that there is only a flow of ‘potency’ through the psychic pneuma that is resident in the nerve. This last is akin to nerve conduction as we understand it today. However Galen did not speculate further on which of the three modes of psychic pneuma conduction was most likely to occur. The next problem concerns that of the transmission of the psychic pneuma into the muscle necessary for the muscle psyche, in Aristotle’s scheme, to realise its final cause and so contract. All that Galen says on this is that transmission must be such as to allow the psychic pneuma to reinforce and initiate the muscle’s intrinsic propensity to contract, that is, to achieve its final cause. However, he did entertain the possibility that this might occur by the psychic pneuma being pushed out of the end of the nerve (Hall, 1975). It was this idea that was to pave the way for a revolution in the approach to conduction and transmission, which followed 1300 years later, and is due to Descartes. 1.3 Descartes: the replacement of pneuma by mechanical corpuscles The great contribution of Descartes (1596–1650; Fig. 1.2A) was to dismantle the concept due to Aristotle two thousand years earlier that all manifestations of life, such as locomotion, nutrition, and sensation are to be attributed to the psyche; engagement of the causal entity then leading to the expression of the inherent capacity of a particular organ to be expressed. As he pointed out (Descartes, 1662): The error is that, from observing how all dead bodies are devoid of heat, and consequently of movement, it has been thought that it is the absence of the soul which has caused these movements and this heat to cease; and thereby, without reason we have come to believe that our natural heat and all the movements of the body depend on the soul. What, on the contrary, we ought to hold is that the reason why soul absents itself on death is that this heat ceases and that the organs that operate in moving the limbs disintegrate. The psyche, as elaborated by Aristotle, was abandoned. This opened up for enquiry the mechanisms by which organs move and heat is produced, cessation of which leads to death. It made transparent the fact that the idea of each organ possessing a psyche, which had prevented the development of physiology for two thousand years, was merely a means of declaring an ignorance concerning the mechanisms of how a particular organ functioned. The loss of the psyche as the causal agent meant that psychic pneuma was no more, leaving open the questions of what is conducted along nerves and transmitted into muscle and how does conduction and transmission occur. To these questions Descartes gave detailed answers, based on his new mechanistic philosophy. In this the body consists of a set of corpuscularly constituted mechanically interacting parts, so that the ultimate level of analysis concerns corpuscular motion. Each part of the body can be activated by the transfer to it of motion that is ultimately derived from heat, which itself is just the agitation of particles engaged in fermentation. Descartes thought this took place in the heart and therefore involved blood particles. This description has a modern ring about it, except of course for the placing of heat generation in the heart. In Descartes scheme large blood particles when they reached the brain were used to nourish it whereas fine blood particles were transformed into a different kind of particle that could be used by the brain for the purposes of conduction along the nerves leaving the brain and spinal cord. This different kind of fine particle to that found in the blood he referred to as animal spirits. Such a name tends to remind one of the psychic pneuma, but in Descartes case the animal spirits were fine particles and accessible to physiological enquiry. Descartes own dissections of the nervous system in his early twenties led him to describe nerves as hollow tubules with a sleeve like double outer sheath, the inner and outer membranes of the sheath being continuous with the inner and outer meninges of the brain. Each nerve tubule contained a central marrow of longitudinal fibrils, surrounded by animal spirits moving outward from the brain, the animal spirits being composed, as we have commented, of highly volatile material particles derived from the blood. In Descartes own words (Descartes, 1972; Fig. 1.3 A): Now in the same measure that spirits enter the cavities of the brain they also leave them and enter the pores or conduits in its substance, and from these conduits they proceed to the nerves. And depending on their entry or their mere
HISTORY OF THE SYNAPSE
5
tendency to enter some nerves rather than others, they are able to change the shapes of the muscles into which these nerves are inserted and in this way to move all the members. Conduction in nerves involves the passage of small particles derived from the heart. Transmission is due to these particles leaving the ends of the nerves and entering the muscle. In order to make sure that the reader is aware of the mechanism of conduction and transmission that he is proposing Descartes comments: But to make you understand all this distinct, I wish to speak to you first of the fabric of the nerves and the muscles, and to show you how from the sole fact that the spirits in the brain are ready to enter into certain of the nerves they have the ability to move certain members at that instant. This description is worth quoting at some length as the 1662):
first detailed account of conduction and transmission (Descartes,
Observe in Fig. 1.3 A [Fig. 1.3 A of this book] for example, nerve A whose external membrane is like a large tube containing several other small tubes, b, c, k, l, and so on, composed of a thinner, internal membrane; and observe that these two membranes (outer and inner) are continuous with the two, K(pia) and L(dura), that envelop the brain MNO. Observe also that in each of the little tubes there is a sort of marrow composed of several very fine fibrils which come from the actual substance of the brain N and whose two extremities end one at the internal surface of the cavities of the brain and the other at the membranes and flesh on which the tubule containing them terminates. But because this marrow is not used to move the members, it will suffice for now that you know that it does not completely fill the tubes containing it but leaves room enough for animal spirits to flow easily through them from the brain into the muscle whither these little tubes, which should be thought of as so many little nerves, make their way. Descartes goes on to say, with respect to Fig. 1.3B: And consider that although there is no evident passage through which the spirits contained in muscle D and E can leave them except to go from one to the other nevertheless because their particles are very small and indeed because they are made incessantly finer through the force of their agitation, some always escape across the membranes and flesh of the muscle while others return through the two nerve-tubes bf and cg to replace those that escape. It will be noted that Descartes retained the basic Galenic idea that the heart was the source of the material used to allow conduction by the nerves, after its transformation in the brain. In the case of Galen and his students that material passed from the heart as vital pneuma, was transformed in the brain to psychic pneuma whence it was used by the nerves that leave the brain and spinal cord for conduction. For Descartes, coarse and fine particles in the blood leave the heart and are sorted by the brain in such a way that the coarse particles are used to nourish it whereas the small particles cease to have the form of blood and become animal spirits. These are able to enter the ‘pores and conduits’ of the brain from which they are guided eventually into appropriate nerves to mediate a particular motor action. The Galenic and Cartesian schemes are very similar except that in the latter we are dealing with definite particles, sorted in a definite way by the brain, with the properties of the particles and their passage in the brain and nerves open to further physiological enquiry. It seems likely that Descartes conceived that conduction of the particles occurs by the mechanism favoured by Galen for psychic pneuma, namely that the particles are forced out of the peripheral end of the nerve in the muscle as a consequence of the entry of particles into the central end of the nerve. As for transmission, it probably required the direct entry of particles from the nerve endings into the muscle cells on which they impinge. However Descartes does not make these points explicit, commenting in relation to the nerves that animal spirits flow… easily through them from the brain without specifying whether this is to be thought of as a travelling wave in time. Indeed it was taken to be a wave of infinite velocity until the experiments of Helmholtz in the nineteenth century. As for transmission from nerve to muscle, he comments in relation to the nerve fibres in the muscle that animal spirits enter therein they cause the whole body of the muscle to inflate and shorten and so pull… while on the contrary, when they withdraw, the muscle disinflates and elongates again. This certainly seems to imply that there is direct flow of animal spirits into the muscle that causes the inflation, although that is not specified and explained in detail until the work of Borelli a few years later.
6
EARLY HISTORY OF THE SYNAPSE
Fig. 1.2. The principal contributors to our understanding of nerve conduction and transmission in the 17th and 18th centuries: from corpuscles to electricity. (A) R.Descartes (1596–1650) (from Bennett, 1997, with permission). (B) Giovanni Alfonso Borelli (1608–1679) (from Brazier, 1959, with permission). (C) Felice Fontana (1730–1805) (from Bentivoglio, 1996, with permission). (D) Luigi Galvani (1737–1798) (from Brazier, 1959, with permission).
1.4 Borelli: a corpuscular description of conduction and transmission The Cartesian hypothesis concerning the mechanism of conduction and transmission was taken up and elaborated on in great detail in the new tradition of physiological enquiry by Borelli (1608–1679; Fig. 1.2B), a young contemporary of Descartes who outlived him by some thirty years. Borelli’s (1670) description of the animal spirits used for conduction by the nerves follows closely that of Descartes (1664):
HISTORY OF THE SYNAPSE
7
Fig. 1.3. The elucidation and refutation of the corpuscular theory of transmission in the 17th and early 18th centuries. (A) The Cartesian model of the nerves cast a long shadow into the 19th century. Descartes (L’homme, 1664) conceived of motor impulses conveyed in the space between the pipes and outer sheath while sensory impulses were conveyed in the inner pipes. For a full description of this figure see the text (from Descartes, 1662). (B) On the left is the Cartesian model of how the nerves proceed to a muscle and control its shortening. In Descartes words ‘Next observe how the tube or little nerve bf proceeds to muscle D, which I assume to be one of those that move the eye, and how it there divides into several branches composed of a loose membrane which can extend, enlarge and shrink according to the quantity of animal spirits that enter or leave it, and whose branches of fibres are so arranged that when animal spirits enter therein they cause the whole body of the muscle to inflate and shorten and so pull the eye to which it is attached, while on the contrary, when they withdraw, the muscle disinflates and elongates again. Observe further that in addition to the incoming nerve-tube bf there is still another, namely ef, through which the animal spirits can enter muscle D, and another, namely dg, through which they can leave it. And quite similarly that muscle E, which I assume is used to move the eye in the contrary direction receives animal spirits from the brain through nerve-tube cg from muscle D through dg, and sends them back toward D through ef’ (from Descartes, 1662). On the right is the original sketch by Descartes illustrating ‘the canals by which the spirits of one muscle can pass into that which opposes it’. Valves in the canals, ‘i’ can open or shut as required for reciprocal innervation (from Brazier, 1984). (C) Croone’s diagram depicting the route (EFG) by which nervous fluid flows from the brain (H) to little bladders which inflate the muscle to expand from contour ABCD (solid line) to AQV (dotted line). On the right is a diagram of the direction of forces when bending the elbow (from Brazier, 1959; Croone, 1665). (D) Swammerdam’s experiments including the one by which he proved that muscles were not swollen by an influx of nervous fluid when they contracted. Fig. V is of an experiment to show the change in shape of a muscle when stimulated by pinching its nerve. Fig. VI illustrates the pulling together of the pins holding the tendons when the muscle contracts. Fig. VIII is the crucial one in which a drop of water is imprisoned in the narrow tube projecting from the vessel enclosing the muscle. Further description of this experiment is given in the text (Biblia Naturae, Amsterdam, 1738; from Brazier, 1959; Swammerdam, 1737).
In the animals, besides liquids such as blood, there is another extremely spirituous fluid substance which is the direct motive cause of the animal body. This appears from the effects of this substance. This spirituous humour is not wind or air but has a liquid consistency such as spirit of wine. It is generated from blood in the brain and diffused by the nerves.
8
EARLY HISTORY OF THE SYNAPSE
All modern authors admit this point. The exact structure and composition of the nervous juice, although unknown, can be surmised somewhat from its motions through the nerves. The mechanism of conduction of the animal spirits along the nerve is due to the compression of the spirits at the central end leading to secretion of spirit at the peripheral end (Descartes, 1664): The spongy cavities of these nervous fibres thus are conceived as being always soaked and filled up to turgescence by some juice or spirit transmitter from the brain. In a bowel full of water and closed at both ends, impulse at an extremity compressed and slightly percussed is instantly transmitted to the other extremity of the turgid bowel. The adjacent elements of the liquid are aligned in a long row. By pushing and shaking each other, they transmit the movement to the extremity of the bowel. Similarly, as a result of some slight compression, jolt or irritation at the origins of the canals of the nervous fibres which are in the brain itself, these fibres thus shaken and activated must secrete some drops of this juice which swells their internal spongy substance, into the fleshy mass of the muscles. Transmission involves the movement of the spiritious juice from the nerve endings in the muscle directly into the muscle cells (Descartes, 1670): The distal orifices of these nervous fibres are scattered everywhere in the mass of the muscle although they are open, the spongy structure itself with which the fibres are provided plays the role of valvules. Indeed droplets hanging from wet sponges do not flow out. A shaking force is required to express them. This may be the cause why the nervous juice is secreted and instilled in all the mass of the muscle by order of the will. The cause and mechanism by which nervous juice is instilled in the muscles with a convulsive force by an order of the will and produces their instantaneous swelling, are deduced from what was said above. Contraction will continue as long as the cause of the bursting is present i.e. the instillation of nervous juice. When it stops, the turgescence of the muscles disappears, as light disappears when the flame which continuously renews it is removed. In summary, Borelli conceives conduction and transmission thus (Descartes, 1670): Consequently, this slight motion of the spirits provoked by the will in the brain can shake or excite the fibres or spongy ducts of some nerves turgid with spirituous juice. As a result of this convulsive irritation which shakes all the length of the nerves, some spirituous droplets can be expressed and spilled from the orifices of their extremities into the corresponding muscle. This results in the boiling and bursting by which muscle is contracted. At the end of the seventeenth century, William Croone summarised for the Royal Society of London the revolution in understanding of conduction and transmission that had taken place that century, involving rejection of the concept of psychic pneuma for that of juices consisting of corpuscles. Fig. 1.3C shows his diagram of the mechanism of conduction along motor nerves and transmission to muscle. What is now called the nervous fluid flows from the brain along the motor nerves to inflate small bladders in the muscle which cause it to expand and shorten (Croone, 1665). 1.5 Fontana: nerves are composed of many cylinders along each of which conduction occurs The composition of the nerves along which conduction occurs was illuminated in the eighteenth century. At its beginning the Dutch microscopist Antoni van Leeuwenhoek (1632–1723) used his one lens microscope to give a description of the composition of nerves, commenting that Often and not without pleasure, I have observed the structure of the nerves to be composed of very slender vessels of an indescribable fineness, running length-wise to form the nerve (Fig. 1.5A). These vessels were taken to be hollow tubes, in agreement with the Cartesian concept that animal spirits flowed in nerves (van Leeuwenhoek, 1685; van Leeuwenhoek, 1717). The relation between these hollow tubes and the nerve was spelt out in detail through the nerve dissections of Felice Fontana (1730–1805; Fig. 1.2C). These were performed, after immersing the nerve threads in water, with very sharp needles under a magnification of X 700, and allowed Fontana to claim that (Fontana, 1760; Fig. 1.4B): The basic structure of nerves is as follows: a nerve is formed of a large number of transparent, uniform, and simple cylinders. These cylinders seem to be fashioned like a very thin, uniform wall of tunic which is filled, as far as one can see, with transparent, gelatinous fluid insoluble in water. Each of these cylinders receives a cover in the form of an outer sheath which is composed of an immense number of winding threads. A very large number of transparent
HISTORY OF THE SYNAPSE
9
cylinders together can form a nerve so small that it is barely visible but which shows the white bands on the outside. Several of these nerves together form the larger nerves seen in animals. I am fully convinced by my own observations, which I repeated many times with the same result, that the cylinders I have described are the simple and first organic elements of nerves, for I have not succeeded in dividing them further, no matter what investigations I carried out with the help of the sharpest and finest needles. I could easily tear and break them here and there; but they always remained indivisible. I could strip them off their sheaths and separate the winding cylinders of which they are formed, although they were very small. The primitive nerve cylinder then appeared transparent, homogeneous, and of equal diameter everywhere. He goes on to say that: After having dissected a very small nerve and its minimal nervous threads made of the different nervous primitive cylinders I have extensively dealt with in my work, I succeeded in stripping from the inner sheath, or rather from the tortuous threads, some nervous primitive cylinders. These were transparent, homogeneous, not empty, as I had found them in previous occasions. As to the constituents that made up the cylinders, that were ‘not empty’, Fontana in 1781 describes the microscopic features of the axoplasm extruded from the cut end of an axon as (Fontana, 1781): …glutinous, elastic, transparent material, insoluble in water, that decomposed itself into very little round grains of a diameter four or five times less than a red blood globule’. ‘I am not sure that Physiologists would be willing to consider those little grains as animal spirits, and the mechanical principle of all movements. This hypothesis would not explain the instantaneous speed of animal movements, since those little grains seem too lazy to move inside the nerve, where they form instead a viscous and inert glutine. Animal movements would be easier to explain considering that such grainy material is elastic, and continuous along all the nervous canal, as the observation in fact demonstrates. The movement could be transmitted at the moment that would follow a mechanical alteration of the nerve or any of its parts. These descriptions of the larger nerves as composed of smaller nerves which are not divisible further and which contain ‘a glutinous, elastic, transparent material’ has modern resonances. However Fontana produced this description in the year that Galvani began his most important discoveries. These were to identify electricity as the conducting material for nerves rather than the Cartesian corpuscles of fine particles derived from blood. Fontana then adheres still to the Cartesian animal spirits and so has difficulty in reconciling the size of the lazy particles in the nerve cylinders with that of the speed of animal movement. He then comes to emphasise the possibility favoured by Galen that it is the extrusion of the particles at the peripheral ends of the nerves following the entry of particles at the central ends of the nerves that provides the appropriate speed for nerve action (Fontana, 1767; Fontana, 1775). 1.6 Galvani: electricity is conducted and transmitted not corpuscles Borelli had commented in relation to the idea of the flow of a nervous fluid in nerve to muscle that All modern authors admit this point. Indeed when Boerhaave produced the first figure of the neuromuscular junction in the early part of the eighteenth century (Boerhaave, 1735; Fig. 1.4C), it emphasised continuity between the nerve ending and muscle, as expected if there was to be a direct flow of nervous fluid into the muscle required for muscle shortening. This whole conceptual scheme was dealt a major setback with the brilliant physiological experiments of Swammerdam, published in 1738. These showed that muscles were not swollen by an influx of nervous fluid during contraction. In this work, illustrated in Fig. 1.3D (VIII), he placed a muscle with its nerve supply in a narrow tube which was then filled with water in such a way that water could be expelled from the tube if the muscle swelled on contraction (Swammerdam, 1737). A wire attached to the muscle nerve in the tube was then pulled on to excite the nerve and contract the muscle. The result was unequivocal, muscle contraction did not lead to the expulsion of water from the tube, so that muscle swelling could not have taken place. Nervous fluid could not, by flowing directly into a muscle, cause contraction. The whole concept of a nervous fluid, consisting of small Cartesian corpuscles, was now thrown into doubt. What could be the nature of the substance that was conducted by nerves? The science of electricity emerged in the sixteenth and seventeenth centuries. William Gilbert (1544–1603) had constructed the first electroscope, consisting of a suspended needle that was attracted to static electricity, so that it turned on being brought near a piece of rubber amber. This apparatus allowed the amount of attraction due to the static electricity to be given in quantitative terms according to the extent of deflection of the needle. Gilbert used the electroscope to detect static electricity in a number of other rubbed objects consisting of glass, wax and sulphur. This led to the invention by Otto von
10
EARLY HISTORY OF THE SYNAPSE
Fig. 1.4. Anatomical identification in the 17th±19th centuries of nerve fibres and their junctions with muscle. (A) Leeuwenhoek’s drawing of a small nerve (ABCDEF) composed of many ‘vessels’ in which ‘the lines or strokes denote the cavities or orifices of these vessels’. This nerve is surrounded in part by five other nerves (one of which is labelled G), in which only ‘external coats’ are represented (from Bentivoglio, 1996; van Leeuwenhoek, 1685). (B) Nerve fibres drawn by Fontana. The drawing illustrates ‘a nerve torn with a needle, to determine the continuity of the primitive nervous cylinders’, a indicates the ‘two ends of the nerve’, c,n,o indicate ‘several of the primitive cylinders’ (from Bentivoglio, 1996). (C) Boerhaave’s concept of the neuromuscular junction. He believed that the nerve (EC) flowed directly into the substance of the muscle (HB) (from Boerhaave, 1735; Brazier, 1984). (D) Schematic summary view of the mammalian neuromuscular junction (from Kuhne, 1869; Shepherd, 1991).
Guericke (1602–1686) of the frictional machine which was constructed from a sulphur ball mounted on a spindle and rotated by hand to generate large quantities of static electricity. The opportunities for the discovery of animal electricity were in place with the subsequent invention of the Leyden jar for storing static electricity by Petru van Musschenbroek (1692–1761; Fig. 1.4A). Indeed speculations that electricity might compose the Cartesian animal spirits were made at this time by the mathematician Christian August Hausen (1693–1743). Luigi Galvani (1737–1798: Fig. 1.2D) discovered ‘animal electric fluid’, a phrase reminiscent of the ‘animal spirits’ used by Descartes in his mechanical description of nerve conduction. This story begins on the famous occasion during which one of Galvani’s collaborators touched with a lancet the exposed nerve of a frog muscle near a frictional machine (Fig. 1.5C),
HISTORY OF THE SYNAPSE
11
which occasionally sparked giving rise to the transfer of charge by induction to the frog’s nerve and thereby a twitch contraction. Galvani investigated this phenomenon further using the apparatus shown in Fig. 1.5C (Galvani, 1791). This consisted of a frog’s exposed spinal cord-leg preparation suspended in a sealed jar by means of a wire passed through the spinal cord and then through a seal at the top of the jar; lead shot was present in the bottom of the jar (right hand side of Fig. 1.5C). A wire was then strung across the ceiling to pick up the charge from a frictional machine and convey it to the wire from which the spinal cord was strung, as shown in Fig. 1.5C. This apparatus allowed for the unequivocal demonstration that when the machine sparked the legs twitched. From this Galvani concluded that frog nerves conduct electricity. Animal spirits had become electricity. Galvani devised a number of other experimental procedures in the years 1781 to 1791 which showed the existence of electrical conduction in nerves (Galvani, 1783; Galvani, 1791; Galvani, 1794). In one experiment he used a pair of jars in one of which there was enclosed a frog spinal cord-leg preparation suspended over lead shot as before by means of a fine iron wire; this wire then lead into another jar which in turn had a layer of lead shot, together with a coil of attached wire to collect the discharge from the frictional machine, as shown in Fig. 1.5B. This discharge was accompanied by sparking in the upper jar and twitching of the frog’s legs in the lower jar, due to what Galvani described as the passage of ‘electric fluid’ down the wire in the upper jar and down the spinal cord and nerves into the leg muscles. His conclusion from these experiments was that there must be a nervous ‘electric fluid’. That this electric fluid must flow along individual nerves and not just the spinal cord was confirmed by work in which the sciatic nerve of one leg of a frog was dissected and used in the experimental apparatus instead of the spinal cord. In this case the leg twitched on discharge of the friction machine as had been the case with the isolated spinal cord-leg preparation. An investigation of Galvani’s in 1794, which was to have far reaching repercussions in the following century in the hands of du Bois-Reymond, involved experiments that were to lead to the discovery of ‘animal electricity’, that is the existence of electricity generated by nerve and muscle itself (Galvani, 1791). In this experiment Galvani placed the severed end of a nerve, belonging to a leg-muscle preparation, on the intact portion of the nerve and obtained movement of the leg (Fig. 1.5E). This showed the existence of electrical potential in nervous tissue and that electrical flow could occur in nerves as a consequence of the nerves producing a potential. He published this work anonymously as ‘Dell’ uso e dell’ attivita dell’ Arco conduttore nelle contrazioni dei muscoli’ (On the application and activity of the Arco conduttore in the contraction of muscle). In 1797 Galvani showed that if he allowed one nerve of a nerve-leg preparation to fall from a glass rod on which it was suspended onto the cut region of another nerve from the same frog then the legs moved (Galvani, 1797). It was not then necessary that the same nerve be used to excite itself, but that any injured nerve would suffice. Indeed electricity could be led by a suitable conductor from the cut end of the spinal cord where the potential was generated to the leg directly in order to obtain a twitch (Fig. 1.5D). One of Galvani’s most famous demonstrations of the flow of electricity in nerve involved the observation that frog’s legs twitched when hung from brass hooks to an iron railing even in the absence of a thunderstorm. Galvani interpreted this as due to the generation of animal electricity rather than, as Volta was later to show, to the flow of current between dissimilar metals connected in a circuit. However Volta went further and attempted to analyse all of Galvani’s experiments as an artefact due to this phenomenon (Volta, 1918a; Volta, 1918b). It was left to Alexander von Humboldt (1769–1859) to confirm Galvani’s experiments and show that they occurred independently of any current flow due to dissimilar metals being incorporated into the experimental design, something which Galvani himself had shown (see Fig. 1.5E). Galvani met Volta’s criticism by cutting both sciatic nerves of a frog where they leave the spinal cord. He then lifted the cut end of one nerve with a glass rod so that it touched the other nerve with its cut end. When this occurred the muscle of the touched nerve contracted. Fig. 1.5 (F3) shows one of von Humboldt’s experiments in which he applied a charged tube of glass to the nerve of an isolated frog nervemuscle preparation and obtained a contraction. Fig. 1.5 (F6) shows another experiment in which he turned the cut end of the nerve against the muscle and obtained a contraction, a variation of the Galvani experiment in which the cut end of a sciatic nerve was placed on another sciatic nerve (von Humboldt, 1797). None of these experiments were open to the kinds of criticism that Volta had aimed at Galvani. As we have seen, at the end of the seventeenth century nerves were thought to conduct animal spirits rather than psychic pneuma, with the former envisaged as corpuscular in nature, derived from fine particles of blood. Galvani died at the end of the eighteenth century, by which time he had shown that nerves could conduct electricity and further that the potential for generating electricity could be found in nerve and muscle itself (Bresadola, 1998). It was generally accepted after this work and that of von Humboldt that the conduction of electricity in nerve was like the way in which metallic wire conducts voltaic electricity. Animal spirits had become electricity.
12
EARLY HISTORY OF THE SYNAPSE
Fig 1.5. The emergence of the concept in the late 18th century that electricity is conducted by nerves. (A) Two early Leyden jars in the collection from the Boerhaave Museum, Leyden. (Photograph by courtesy of the Boerhaave Museum). (B) Galvani’s sketch of his preparation of inverted flasks containing the frog’s nerve muscle preparation from an experiment dated December 10, 1781 (from Brazier, 1984). (C) This figure shows Galvani’s frictional machine, a Leyden jar, and a wire strung across the room to collect the charge (from Brazier, 1984; Galvani, 1791). (D) An artist’s depiction of Galvani’s favourite preparation published as part of the first illustration to the famous Commentary published in 1791 (from Brazier, 1984). (E) The critical experiment by Galvani on muscle contraction in the absence of all metals (from Brazier, 1984). (F) Von Humboldt’s experiments in which he demonstrated contraction of nerve-muscle preparations in the absence of any metals. His Fig. 3 depicts a frog nerve-muscle preparation to which he applied a tube of glass (x), producing a contraction. His Fig. 6 shows an experiment in which he turned back the nerve against the muscle without interposing the glass rod (from Brazier, 1984; von Humboldt, 1797).
1.7 Matteucci and du Bois-Reymond: transient electrical changes are conducted (the action potential) The triumph of nineteenth century physiology was to take Galvini’s discoveries and show that the nervous primitive cylinders of Fontana possess a potential across their membranes that could give rise to a propagating transient potential change, the action potential.
HISTORY OF THE SYNAPSE
13
As is so often the case in the history of neurophysiology, the development of concepts concerning electricity in nerve and muscle at the beginning of the nineteenth century was dependent on advances in instrumentation. Gilbert’s use of a suspended needle to detect electricity permitted only slight deviations from the meridian because of the earth’s magnetic field. In 1820 Schweigger, following Oersted’s discovery that a magnet tends to set itself at right angles to a loop of bent wire carrying an electric current (Oersted, 1820), designed the galvanometer. In this instrument many turns of wire were wound on a rectangular frame inside which a compass needle was placed that was balanced on a vertical pivot or in some cases suspended from a thread. Leopold Nobili in 1825 manufactured the first astatic galvanometer in which he wound two coils of wire on the rectangular frame of Schweigger in opposite directions, so as to cancel the effects of the earth’s magnetism. Nobili used this instrument in 1827 to detect currents passing up the body of a frog away from the legs towards a cut spinal cord and in this way made the first measurement of animal current, or as he called it the ‘intrinsic current’. However he attributed the current to a thermoelectric effect caused by the unequal cooling of nerve and muscle produced by evaporation rather than due to an intrinsic biological phenomenon (Nobili, 1824; Nobili, 1828). Carlo Matteucci (1811–1865; Fig. 1.6A) used the Nobili galvanometer to great effect on isolated nerve-muscle preparations. Although not well known he may be considered to be a founding father of electrophysiology. Matteucci showed that a twitching muscle generated current sufficient to stimulate the nerve of another muscle laid across it and so produce a twitch in the other muscle. Importantly he detected current flow between the cut end of a muscle and the intact end. These currents were correctly interpreted as generated by the muscles themselves (Matteucci, 1842a; Matteucci, 1942b). This was emphasised by his experimental technique of preparing a pile of sectioned frog’s thighs arranged in a series so that the intact surface of one thigh was in contact with the sectioned surface of the next one. The currents generated were in proportion to the number of thigh sections in the pile. However Matteucci’s most important observation was that the current between the cut end of a muscle and the intact end declined during a tetanus caused by strychnine, that is there was a negative variation in the current. Thus excitability was associated with a decrease in the potential that gives rise to the current. Although Matteucci was unable to detect with his instruments a negative variation in the nerve current, his observations laid the ground work for the emergence of the concept of the action current and of the action potential (Matteucci, 1838; Matteucci, 1844; Matteucci, 1848; see also Matteucci & Humboldt, 1843). du Bois-Reymond (1818–1896; Fig. 1.6B), confirmed Matteucci’s experiments on nerve-muscle preparations, and on muscles isolated from their nerve supply, calling current flow in the latter case ‘muscle current’. Most importantly the negative variation or ‘negative Schwankung’ of the muscle current during a tetanus was shown in 1843 to be produced by other means than strychnine, for instance by direct faradic stimulation. du Bois-Reymond, using more sensitive instrumentation than that available to Matteucci, was able to detect in 1834 the negative variation in nerves as well as muscle. The concept of the action potential with its action current showing up as a negative variation was clearly envisaged by du Bois-Reymond (1841; 1842). He hypothesised that a resting potential existed between the middle of muscle cells at positive potential and that of the tendons at negative potential: it was this potential which decreased during stimulation so that a negative variation was recorded. He went on to develop the concept of ‘electromotive particles’ or ‘electrical molecules’ (du Bois-Reymond, 1877). These possessed a positive charge in their middle and a negative charge at each of the polar regions. He postulated that these were situated along the length of the surface of muscle cells and nerve fibres and that it was these that gave rise to the polarisation of the cells (Fig. 1.6D). At rest these molecules were postulated to be arranged in an ordered longitudinal array, so that if a nerve or muscle was sectioned transversely this gave rise to the muscle or nerve current between the injured regions and the intact surface. An electrical stimulus was envisaged to perturb the ordered longitudinal array, producing an electrotonic disturbance leading to the initiation of the negative variation. In this ‘molecular hypothesis’ muscle and nerve fibres are composed of strings of so-called peripolarelectric molecules each of which possess an equator corresponding to the electropositive metal zinc and two poles corresponding to the electronegative metal copper. The current attributed to the internal potential difference thus created could be led off by placing one end of a conductor on the ‘natural longitudinal section’ of a nerve or muscle and the other end on the ‘natural cross section’; in this case the longitudinal section acted as the positive pole and the cross section as the negative pole. The term ‘natural cross section’ as applied to muscle refers to the tendon covered ends of the muscle, regarded as prisms or cylinders, while the term ‘natural longitudinal section’ refers to the lateral surface of these prisms or cylinders. The corresponding artificial cross section and longitudinal section are obtained by dividing the muscle length-wise or crosswise. In this sense the proximal cross section is the upper one, and the distal cross section the lower one. The same applies mutatis mutandis for the nerve. The negative variation involved the discharge of this electromotive force, an idea that clearly presaged the concept of the resting membrane potential and its depolarisation during the action potential (du Bois-Reymond, 1860; du Bois-Reymond, 1845; for a recent detailed account, see Piccolino, 1998). Matteucci had discovered the negative variation in muscle that accompanies activity and du Bois-Reymond that in nerve. Although it seemed very likely that this negative variation of electrical polarisation was the animal spirit of Descartes, it was still endowed with the mysterious property that it could travel at infinite velocity. This was accepted by all the leading physiologists of the first half of the nineteenth century. For example in 1846 E.Weber summarised his observations on the
14
EARLY HISTORY OF THE SYNAPSE
Fig 1.6. Identification of the action potential as the electrical means of conduction. (A) Carlo Matteucci (1811–1865) (from Brazier, 1959). (B) Emil du Bois-Reymond (1818–1896) (from Brazier, 1959). (C) H.von Helmholtz (1821–1894) shown as a young man when he made his greatest contribution to the understanding of impulse conduction in nerve (from Brazier, 1959). (D) Schemata of du Bois-Reymond’s postulated method for transmission at the motor end plate (from du Bois-Reymond, 1845). (E) Helmholtz’s apparatus for measuring the time course of muscle contraction and the propagation velocity of the nerve impulse. On the left, Figure 1 shows the entire apparatus; on the right, Figure 2 shows the arrangement when the nerve w is attached and more than one point on the nerve can be stimulated (from Cahan, 1993; Helmholtz, 1850a). (F) Helmholtz’s muscle curve (from Cahan, 1993; Helmholtz, 1850a).
conduction of the action current in muscle nerves with the comment that (Weber, 1846): When one stimulates a muscle through a motor nerve its movement occurs at the same moment that is the movement begins and ends with the stimulus. Muller’s comment was that (Muller, 1840; Muller, 1851): the time the stimulus takes to travel to the brain and back is infinitely small and unmeasurable. The existence of such a phenomenon as a travelling wave with infinite velocity left the mechanism of conduction opaque to further analysis, rather in the way that the idea of the psychic pneuma had until the time of Descartes. This impasse was broken through the experimental skill of Helmholtz (1821–1894; Fig. 1.6C).
HISTORY OF THE SYNAPSE
15
1.8 Helmholtz: the action potential has a finite velocity In 1848 Helmholtz began, in his own words (Cahan, 1993): …to study the processes occurring in the simple contraction of a muscle; by such an action I mean one that results from a stimulus of vanishingly small duration. I have now finished building my frog-tracing machine and have already carried out a few tracing experiments on mica sheets. Instead of the frog muscles, I inserted a spring. The weight hung from it, oscillated up and down, and recorded its movements. The traces are much prettier than the earlier ones, very fine and regular [see Fig. 1.6F]. The previously unknown fact that in animal muscles too, as in the case in much longer time intervals in organic muscles, the energy of the muscle does not develop completely at the moment of an instantaneous stimulus. Rather, in most cases after the stimulus has already ceased, it increases gradually, reaches a maximum, and again subsided. [Fig. 1.6F]. The force of the muscle was not strongest directly after the stimulation, but rather increases for a time and then falls. In October 1849 he set out to give a more accurate account of this apparent delay between the electrical stimulus and the muscle’s response, with its implications for a finite velocity of conduction of the nerve action potential. To this end he used a method for measuring small time intervals based on the fact that the length of the arc through which the magnetic needle of a galvanometer moves when a transitory current passes through its coil is proportional to the duration of the current. The time interval in question was then measured by ensuring that this movement was led to the deflection of a mirror. Helmholtz next arranged his apparatus so that the beginning and end of the time interval marking the currents duration coincided with the time interval that began with the application of the stimulus and ended with the onset of the muscle’s mechanical action. The latter was obtained by the mechanical action of the muscle lifting a weight that then placed a break on the electrical current. It is worthwhile analysing the apparatus that Helmholtz used for this experiment, both for the beauty of its design and for the unequivocal outcome that it led to, namely the accurate measurement of the velocity of the action potential. With reference to Fig. 1.6E, and following the description in Cahan (1993): …a muscle was hung suspended from a screw I to which was attached a series of screws and contact surfaces that would break the flow of the current when the muscle raised the weight suspended on the scale pan K. The apparatus was placed in a container with humidity enriched air in order to prevent the muscle from drying out; this set up remained in a usable state for three to four hours. At a certain point following stimulation of the nerve w (see the enlargement on the right hand side of Fig. 1.6E), where v is the current carrying wire to the nerve, the energy of the muscle would equal the load suspended from its lower end. After that point, any increase in the energy of the muscle would elevate the load a little and separate point m from the point n on the apparatus; if, however, weights were put on the scale pan K, such that the muscle was acted upon by an additional overload, then the stimulated muscle could raise the combined weight only if its energy (elastic Spannung) equalled the sum of the weights of the load and the overload. Helmholtz arranged his apparatus so that the current, whose time interval was to be measured, would break when the elastic Spannung of the muscle increased by an amount sufficient to raise the weight of the overload. With this approach Helmholtz discovered that the time interval between a stimulus applied to the nerve and the moment when the muscle produced enough force to lift the overload depended on the distance between the point of stimulation of the nerve and the muscle. A method for measuring the velocity of propagation of the action potential was now available. On 29 December 1849, Helmholtz measured the velocity of propagation as 30.8 metres per s. The curves obtained in these experiments (Fig. 1.6F) showed that the velocity was finite and the apparatus (Fig. 1.6E) that this velocity could be measured (Helmholtz, 1850a; Helmholtz, 1950b; Helmholtz, 1850c; see also Olesko & Holmes, 1993). Helmholtz had made the great discovery which transformed the nervous system from consisting of cylinders through which animal spirits flowed with infinite velocity to one which was amenable to quantitative measurement for the testing of hypotheses (Helmholtz, 1851; Helmholtz, 1852). He knew that as long as physiologists insist on reducing the nerve effect to the propagation of an imponderable or psychic principle, it will appear unbelievable that the velocity of the current should be measurable. Although the mechanism by which this propagation of the action potential occurred was not indicated by these experiments, Helmholtz proposed the first hypothesis to be tested, namely that the process was the same as that of the conduction of sound in the air and in elastic matter or the burning of a tube filled with an explosive mixture. J.Bernstein (1839–1917) began the use of quantitative measurement of nervous phenomena to investigate the mechanism of propagation of the action potential in 1868 with his measurement of the time course of the potential, its latency, rise-time and decay (Bernstein, 1868). This led him to his famous theory that the membrane of nerve and muscle is normally polarised at rest with an excess of negative ions on the inside and of positive ions on the outside. The action potential then becomes a self-
16
EARLY HISTORY OF THE SYNAPSE
propagating loss of this polarization or a depolarisation. Injury currents arise as a consequence of the loss of the polarization at some point in the membrane due to its injury. Bernstein was led to this theory by comparing the negativity of the current arising from an injured nerve and that of the action current (Bernstein, 1871). As the former was attributed to the inside of the tissue becoming continuous with the outside it was conjectured that a transient exposure of the inside to the outside occurs during the action current (Bernstein, 1912). The fact that the action potential could conduct in the orthograde or anterograde directions followed from this theory of the polarization of the nerve and muscle cell membrane. Perhaps the most original contribution of Bernstein was his theory of the origin of the polarization of the membrane of muscle and nerve, which was based on Nernst’s concept of the diffusion potential developed at about the same time (Nernst, 1888; Nernst, 1908). In this theory the potential arose as a consequence of the high permeability of the membrane to potassium compared with other ions. Given that the concentration of potassium is higher inside than outside a negative polarization of the inside of the membrane with respect to the outside is generated. 1.9 Kuhne and Auerbach: identifying the structure of nerve endings on muscle and neurons The advent of superior histological stains allowed the first descriptions of neurons to be given in 1836–1837 by C.G.Ehrenberg for single nerve cells in the leech nervous system (Fig. 1.7A) and by J.Purkinje for the large nerve cells of the mammalian cerebellum named after him (Fig. 1.7B; see also Purkinje, 1837a; Purkinje, 1837b; Purkinje, 1937c). Sixteen years later A.Kolliker (1817–1905; Fig. 1.8A) showed that the nerve fibres of Fontana originated from nerve cells with a description of the beginnings of the acoustic VIII nerve, or in his words in relation to Fig. 1.7C: Nerve-cell with the origin of a fibre (from the acoustic VIII nerve) of the Ox; a, membrane of the cell; b, contents; c, pigment; d, nucleus; e, continuation of the sheath region of the nerve-fibre; f, nerve-fibre. In the same year of 1852 he described how motor nerves originated from the anterior horn nerve cells of the spinal cord (Fig. 1.7D). The question then arose as to the relationship between these motor nerves and the muscle cells on which they impinge. R.Wagner showed in 1847 that the terminal branches of nerves going to the electric organ of the electric ray split into ever finer branches when they entered the electric organ until nothing was left of them in the fine-grained parenchyma of the organ (Wagner, 1847). He extended this by analogy to the nerve supply of muscle. W.Kuhne (1837–1900) described histological differences between the end of the motor nerve and the muscle cell on which it abuts in frogs in 1862, namely at the end-plate (Fig. 1.4D; see also Kuhne, 1862; Kuhne, 1888; Kuhne, 1869). However this did not illuminate the functional problem of how the action potential was transmitted from motor nerve to muscle any more than did the diagram of the motor endplate by Boerhaave’s some one hundred and forty years earlier (Fig. 1.4C) indicate how animal spirits left the nerve to enter the muscle. Boerhaave had simply acquiesced in the current physiological concept of Descartes, developed by Borelli, that animal spirits passed directly from the nerve into the muscle to increase the muscle volume and so shorten it. Kuhne, likewise, took the current physiological paradigm of the action potential and suggested that the action current of the nerve invaded the muscle at the endplate. This idea was developed in some detail by W.Krause in 1863, who drew attention to the similarity between the nerve endings in muscle and the electric plate in the organ of the electric catfish, which was taken to act as a Leyden jar. He argued that the nerve ending, the nerve endplate, charged like an electric plate when the nerve was excited, giving to the contractile substance of the primitive muscle fascicle an electric shock so stimulating it to contract. Once it had been established that nerve fibres originate from nerve cells the question arose as to the relationship between the nerve fibre endings in the nervous system and nerve cells. Given that the histological methods using silver staining as well as microscopical techniques current at that time were not able to give an appropriately detailed account of the relationship between nerve ending and muscle at the endplate, they were certainly not able to illuminate the question of the relationship between nerve ending and nerve cell. This area of study was ripe for speculation, without necessarily furthering understanding, and as a result controversy followed. In 1865 O.Deiter (Deiters, 1865) showed that dendrites (or as he called them protoplasmic processes; these were named dendrites by W.His in 1889, (Waldeyer, 1891)), in addition to nerve fibres (or as he called them axis cylinders; these were termed axons by Kolliker in 1896; Kolliker, 1856a, b), arose from the nerve cell (named the neuron by Waldeyer in 1891), a fact that he illustrated beautifully with his drawings of neurons dissected from the ox spinal cord (Fig. 1.9A). He noted in passing that the dendrites possessed trumpet-like expansions on their surfaces attached to very fine fibres (Fig. 1.9A), which he speculated could represent the input to the dendrites. In 1898 L.Auerbach showed, with silver stain, the ‘end bulbs’ of nerve fibres on the surface of neurons in the facial nucleus, which he unequivocally identified as such (Fig. 1.9B; Auerbach, 1898). He also subscribed to the Cartesian/Borelli idea that there was continuity in the propagation from nerve to target cell, commenting that (Auerbach, 1898):
HISTORY OF THE SYNAPSE
17
Fig. 1.7. Identification in the first half of the 19th century of the neuron as the cell body giving rise to the nerve fibre or axon. (A) Single neurons from the leech nervous system representing one of the first two illustrations of a neuron (from Ehrenberg, 1836). (B) The large corpuscles of the cerebellum, which became known as Purkinje cells after their discoverer, giving the other first illustration of a neuron. This was also the first published view of the cellular composition of the histological layers within a brain region. From below: fibres, granules, large corpuscles (Purkinje cells), molecular layer (from Purkinje, 1837a; Shepherd, 1991). (C) A nerve originates from a cell body. A nerve cell with its nerve in Kolliker’s Handbuch der Gewebelehre des Menschen (1852). In his words: ‘Nerve cell with the origin of a fibre from the acoustic VIII nerve of the Ox; a, membrane of the cell; b, contents; c, pigment; d, nuclear, e, continuation of the sheath region of the nerve fibre; f, nerve-fibre…’ (from Kolliker, 1896; Shepherd, 1991). (D) Another nerve cell from Kolliker’s (1832) textbook for comparison with Fig. C. This is a ‘large nerve cell with processes from the anterior cornua (horn) of the spinal end in man’ (from Shepherd, 1991).
As I understand that theory, the axon terminals exert their effects on the cell surface of the ganglion cell by means of close contact of the end bulbs, without intervention of any intermediate substance.
18
EARLY HISTORY OF THE SYNAPSE
Fig. 1.8. Major figures at the end of the 19th century responsible for the emerging definition of the synapse. (A) From left to right, Guilio Bizzozero (a friend of Golgi), Albrecht Kolliker (1817–1905) and Camillo Golgi (1844–1926) at Golgi’s home in Padua in 1887 (from Prof. P.P.C.Graziadei; from Shepherd, 1991). (B) Santiago Ramon y Cajal (1852–1934) (from the frontispiece in Cajal, 1995). (C) Charles Scott Sherrington (1858–1952) (from Granit, 1966).
1.10 Cajal: nerve endings are not continuous with the cells on which they impinge The use by C.Golgi (1842–1926; Fig. 1.8A) in 1886 of a silver stain in which potassium bichromate and silver nitrate are applied to produce a black impregnation of neurons did not help resolve the nature of the region of nerve fibre endings on cells. On the one hand Golgi claimed that he could detect intracellular neurofibrils in nerves with his silver technique and that these extended from the end bulbs of the nerves into the neurons on which they ended, providing continuity for the transmission of the action potential (Golgi, 1886a; Golgi, 1886b). No better illustration of the implications of this continuous
HISTORY OF THE SYNAPSE
19
reticular network of neurofibrils through nerve fibres, end bulbs and neurons is provided than that of Golgi’s drawing from his stained material of a transverse section through the hippocampus (Fig. 1.9C). Golgi emphasises in the legend to this figure that the axons ending on the granule cell neurons (upper right part of the figure in the fascia dentata; axons of the pyriform pathway) merge with the dendrites of the granule cells and that the nerves which emerge from the granule cell neurons (the mossy fibre axons, shown converging to form a single nerve in area CA3) merge with the dendrites of the CA3 pyramidal neurons. In his words (Golgi, 1886b): The diagram illustrates particularly the mode by which a fascicle of nerve fibers comes in relation to the small ganglion cells of the fascia dentata. Between the fascicle of nervous fibers still maintaining themselves as individual elements and the nervous prolongations (axons) of the small cells, exist a complicated network, occupying a semicircular area, which, especially toward the deep part, has indeterminate borders. It is on entering into this network, that, ramifying, a part of the nervous prolongations (axons) lose themselves, as well as the fibers deriving from the fascicle. The latter, issuing from the semicircle formed by the fascia dentata, traverse the zone of the grey layer of the convolution, occupied by the bodies of the cells which belong to this layer, and go to join the fibers of the Alveus and Fimbria. The conceptual framework here is that of Descarte/Borelli for the neuromuscular junction, namely with continuity between nerves and the structures on which they end. However in 1866, when Golgi stated that continuity existed between nerves, quite different conclusions were being reached by other histologists (Finger, 1994). W.His suggested, on the basis of Kühne’s description of nerve endings on muscle fibers (Kuhne, 1862; Kuhne, 1888), that ‘the motor endplates give the indisputable example of transmission of a stimulus without continuity of substance’ (His, 1886). In 1887, F.Nansen, using Golgi’s silverstaining technique on the nervous system of invertebrates, concluded that ‘a direct combination between ganglion cells, by direct anastomosis of the protoplasmic process, does not exist’ and that ‘the branches of the nervous processes do not anastomose’ (Nansen, 1887). Finally, A.Forel in 1887 used the Golgi technique to show that after a lesion ‘total atrophy is always confined to the processes of the same group of ganglion cells, and does not extend to the remoter elements in merely functional connections with them’ (Forel, 1887). None of these observations gave clear evidence one way or the other as to whether nerve endings are continuous with the cells on which they impinge, although the research of Forel on the effects of a lesion paved the way for the definitive work on this problem by a remarkable neurohistologist. S.Ramon y Cajal (1852–1934; Fig. 1.10B) learnt of Golgi’s silver staining technique in 1887 and applied it to blocks of the cerebellum in 1889. From this earliest work Cajal claimed that the terminal baskets of the stellate neurons could be seen to envelop the Purkinje neurons on which they ended without any sign of their being in continuity with the neurons. He developed from this the neuron doctrine, namely that each neuron is an independent cell that does not anastomose with surrounding cells (Cajal, 1995; Clarke, 1995). This doctrine is deservedly attributed to Cajal for two reasons: one is the trenchant way in which he defended it against the contrary claims of other neurohistologists, which he did by using their experimental techniques to show how their claims were based on artefacts; the other was his definitive degeneration studies in which he showed that the loss of a particular neuron type could leave behind the synaptic terminals that impinge on it, indicating that the latter were not in continuity with the former. Examples of his forceful style in relation to the defence of his work claiming that nerve terminals can be shown in silver stained material to abut but not anastomose with neurons are as follows (Cajal, 1995): Gerlach concluded that certain axons anastomose at their endings with the tips of dendrites, and, thus that central axonal arborizations do not end freely but instead merge with dendrites. Because Golgi thought that central axonal arborizations do not end freely but anastomose instead, his hypothesis is actually based on Gerlach’s theory. To demolish the theory, it was necessary to show by direct observation in the adult brain that axonal arborizations terminate freely, and in the final analysis to do so under conditions that no one could object to because observations were in embryonic material or because material was improperly stained. We were the first, in 1888, to demonstrate unequivocally and irrefutably that terminal arborizations end freely. One need only recall the varied and often profound alterations that occur in dendrites stained with Ehrlich’s method giving varicosities, cyanophilic masses and abnormal thickenings, which may condense or fuse with one another. When this happens the resulting images look so much like anastomoses that they may readily be mistaken for them. In the embryonic and adult spinal cord, in the cerebellum, cerebral cortex, Ammons horn, striatum, and olfactory bulb, in the autonomic nervous system, in the spinal ganglia, retina, and elsewhere, the terminal arborizations of axons and dendrites invariably end absolutely freely—a fact that can be demonstrated equally well by the Golgi and the Cox methods. He therefore concluded that (Cajal, 1995):
20
EARLY HISTORY OF THE SYNAPSE
Fig. 1.9. The search during the second half of the 19th century for nerve endings in the brain. (A) The concept that protoplasmic processes (b) and the axis cylinder (a) are different prolongations of the same cell was represented by Deiters in 1865 in a neuron dissected from the spinal cord of an ox (from Max Schultze, 1870). Deiters thought that input to the protoplasmic processes (dendrites) was via fine fibres connecting by means of a trumpet-like expansion (arrows added by another author) (from Deiters, 1865; Jacobson, 1993; Schultze, 1870). (B) The first representation of synaptic endings (Endknopfchen) in the central nervous system (facial nucleus; reduced silver preparation; paraffin section). From Leopold Auerbach (1898). He concluded that this was evidence in support of the contact theory of nerve connections (from Auerbach, 1898; Jacobson, 1993). (C) A diagram by Golgi of the nervous elements of the hippocampus and fascia dentata (from Golgi, 1886a; Shepherd, 1991). (D) A diagram by Cajal illustrating the structure and connections of the hippocampus and fascia dentata. A, retrosplenial area; B, subiculum; C, Ammon’s horn; D, dentate gyrus; E, fimbria; F, cingulum; G, angular bundle or dorsal hippocampal commissure (crossed temporoammonic path); H, corpus callosum; K, recurrent collaterals from pyramidal cells to the stratum lacunosum of Ammon’s horn (Schaffer collaterals); a, axon entering cingulum; b, cingulum fibres ending in the retrosplenial area; c, fibres of the perforant or direct temporoammonic path; d, perforant fibres of the cingulum; e, plane of dorsal perforant path fibres; g, subicular cell; h, pyramidal cells in field CA1 (regio superior) of Ammon’s horn; i, ascending (Schaffer) collaterals of large pyramidal cells; r, collaterals of alvear fibres (from Cajal, 1995).
…the only opinion that is in harmony with the facts (is) that nerve cells are independent elements which are never anastomosed, with by means of their protoplasmic expansions (dendrites) or by the branches of their prolongations of Deiters (axons), and that the propagation of nervous action is made by contacts at the level of certain apparatuses of dispositions of engagement.
HISTORY OF THE SYNAPSE
21
In 1892 Cajal gave his description of the nerve networks of the hippocampus using the Golgi silver technique, six years after Golgi himself had described the network (Fig. 1.9C). Cajal’s work could not be more different (compare Fig. 1.9D with 1.9C in which the sections through the hippocampus are oriented in the same way, with the fascia dentata uppermost). Here neurons are clearly circumscribed according to the neuron doctrine that each neuron is a separate cell, rather than in continuity with other cells through a neurofibrillar network that joins them all together at the site of the end bulbs. The neurofibrils were intracellular to the neuron according to Cajal (Fig. 1.10E), and did not extend out of the terminal boutons into the neurons on which they abutted (Fig. 1.10G). With Golgi and many other neurohistologists these neurofibrils were both intracellular and intercellular, joining the terminal boutons to the neuron on which they ended, and so giving rise to a reticulum of neurofibrils that gave continuity to the entire neural network in places such as the hippocampus (Fig. 1.9C). On technical grounds it is not easy to say unequivocally that Cajal had objective proof for his doctrine. This was particularly the case when considering the speculations which he and other neurohistologists engaged in concerning the role of the neurofibrils in the conduction of the action potential. For example Cajal comments that (Cajal, 1995): The existence of a conductive pathway in the cytoplasm was also postulated on the basis of observations made with the Nissl method… So that the reader may judge the extent to which the discovery of the neurofibrillar network justifies these assumptions, we shall reproduce a drawing [Fig. 1.10F published long ago and based on the work of Bethe]. Neurofibrils are the sole conductors of neuronal activity. They form bundles in the dendrites and axon, and course between the Nissl bodies as they cross the perikaryon on their way from one process to another without anastomosing among themselves. Long neurofibrils, most of which converge on the axon, are not the only type found in the cell. Cajal went much further in this conjecturing, for which there was not a scintilla of physiological evidence, that the arrangement of the neurofibrils within a single neuron is such that the conduction of the action potential could only occur from dendrites to soma to axon (Fig. 1.10F). This then led him to place arrows of action potential flow on so many of his drawings summarising the results of silver staining of a particular block, such as that of the hippocampus (Fig. 1.9D). The fact that these have ended up being approximately correct does not mean that Cajal should be credited for their discovery, which would be for profoundly wrong reasons, namely based on the conduction of action potentials by neurofibrils. It might be commented on in passing that Cajal’s work was not subject to review until towards the end of his life, as it was published privately. In summary then the neuron doctrine could not be considered to be definitively supported by this silver stain work. An entirely different conclusion may be reached when considering Cajal’s work on degeneration of neural centres which does give definitive support to the neuron doctrine. Again the major evidence was provided by the relationship between terminals on Purkinje cells in the cerebellum and the state of the terminals when the Purkinje cells degenerate (Cajal, 1995): As an example of a convincing and well known case let us mention the disappearance of the Purkinje cells in general paralysis with maintenance of the basket and stellate cells of the molecular layer. This persistence, revealing the independence of the baskets and the cells they surround, can also be produced experimentally by sectioning the axons of the Purkinje cells at the level of the granular layer or even below as is shown in Fig. 1.10B (compare with the normal in Fig. 1.10A). This remarkable conservation of the baskets, despite the disappearance of the cells in connection with them…’ ‘The basket cells resist pathological influences much more that do the Purkinje cells. The baskets in young traumatised animals appear nearly normal, even at the level of regions where the cells have disappeared (Fig. 1.10B(D)).’ ‘Baskets of the Purkinje cells…are perfectly formed in animals from twenty to thirty days old, while they are constantly altered in the vicinity of wounds, although they never react so actively and energetically as the axons of the Purkinje cells. The lesions most commonly found are as follows: (a) Baskets whose descending appendices have terminal balls. As can be seen in Fig. 1.10C(A), the Purkinje cells have been resorbed, and the descending branches of the baskets, notably thickened and intensely stained, end in a terminal ball or in a series of clubs. One may entirely agree with Cajal in his comment that (Cajal, 1995): The neuron doctrine is compatible with the well-documented phenomenon of secondary degeneration in neural centers. In fact, if neurons were not completely independent, it would be impossible to account for the precise localisation of degeneration following ablation of cell groups or fiber tracts. Cajal’s speculations concerning the physiological role of neurofibrils had led him to his polarisation of the neuron doctrine, namely that action potential flow was only from terminal bulb to dendrite (or sometimes soma), and then from soma to axon. However it was a physiologist, namely Sherrington, who supplied the experimental findings which showed that the region of contact between the end bulb and neuron might only allow the direction of action potential transmission in one direction through the end bulb.
22
EARLY HISTORY OF THE SYNAPSE
Fig. 1.10. Evidence at the end of the 19th century that nerve terminals are not in continuity with the cells on which they impinge. (A) The endings of cerebellar basket cells in the albino mouse viewed using the Golgi method. A, Purkinje cell stained with osmic acid; B, basket cell, a, b, pericellular axonal ramifications forming baskets; C, axon (from Cajal, 1995). (B) The endings of cerebellar basket cells in a cat (twenty-five days old) twenty-four hours after axotomy of the Purkinje cells, which destroy them. A, almost normal Purkinje cell; B, Purkinje cell undergoing atrophy and granular in appearance; D, baskets surrounding the empty spaces previously occupied by Purkinje cells that are now destroyed (from Cajal, 1959). (C) The endings of cerebellar basket cells in a rabbit (two months old) thirty hours after axotomy of the Purkinje cells. A, terminal clubs; B, molecular layer; a, transversal fibres of this zone (from Cajal, 1959). (D) Motor endplates on skeletal muscle fibres during reinnervation of the muscle by the motor nerve. Different stages of restoration are shown. A, nerve; B, fibre that gives rise to several plates; E, H, plates as yet without ramifications; C, F, plates with a well developed arborization (from Cajal, 1959). (E) The concept of a neurofibrillary network within an individual neuron, that was taken to conduct impulses by Cajal and others. Shown is a giant pyramidal cell (10-day-old dog). A, is a cell with a pericellular plexus. A, axon; B, summit of the axon hillock; F, dendritic branch with a single neurofibril (from Cajal, 1995). (F) Diagram due to Cajal showing the presumed direction of conduction of impulses along the neurofibrillary network with a cortical pyramidal cell. A, axon; B, nucleus; a, channels or pathways for the neuronal currents of the impulses; b, Nissl bifurcation cones; c, nuclear hood; d, recurrent path of currents in a process bordering the axon; e, elongated Nissl body. Arrows indicate the direction of current flow (from Cajal, 1995). (G) Terminal boutons surrounding a funicular neuron in the spinal cord (from Cajal, 1995).
1.11 Sherrington: the adoption of the word `synapse' Sherrington (1858–1952; Fig. 1.8C), as a consequence of his work on spinal reflexes in the 1890s, had reached the conclusion that transmission of the action potential across the end bulbs of sensory nerve terminals to neurons in the spinal cord involved different principles to that of the conduction of the action potential along nerve fibres (see Liddell, 1960).
HISTORY OF THE SYNAPSE
23
The nerve centre exhibits a valve like function, allowing conduction to occur through it in one direction only. How securely the circuits of the nervous system are valved against regurgitation is shown by the Bell-Magendie law of the reactions of the spinal nerve roots. It was work on spinal reflexes that had led to the idea of the ‘valve like function’ of the region of apposition between the end bulbs and the neuron, and so distinguished this region from the rest of the nerve fibre. This is reminiscent of Descartes model of the nervous mechanism of reciprocal inhibition of voluntary movement which was based on the idea of a flow of animal spirits in hollow nerve fibres; in this case valves could differentially alter the flow through anastomoses between nerves to antagonistic muscles, e.g. lateral and medial rectus muscles of the eye (see Fig. 1.3B). As Michael Foster said in 1901 (Foster, 1897): If we judge Descartes from the severe standpoint of exact anatomical knowledge, we are bound to confess that he, to a large extent, introduced a fantastic and unreal anatomy in order to give clearness and point to his exposition… If we substitute in place of the subtle fluid of the animal spirits, the molecular changes which we call a nervous impulse, if we replace his system of tubes with their valvular arrangements by the present system of concatenated neurons… Descartes’ exposition will not appear so wholly different from the one which we give today. How far then did the work on spinal reflexes in the late nineteenth century allow for a new principle to be enunciated concerning the operation of the region where end bulbs impinged on neurons, different from the speculations of the seventeenth century? Sherrington’s emphasis on some kind of discontinuity at the region of apposition between end bulb and neuron mostly rests on the results obtained from degeneration studies. He says (Sherrington to Schafer, 1897a; Sherrington to Schafer, 1897b): The evidence of Wallerian secondary degeneration is clear in showing that that process observes strictly a boundary between neurone and neurone in the reflex arc. The characteristics distinguishing reflex-arc conduction from nervetrunk conduction may therefore be largely due to inter-cellular barriers, delicate transverse membranes. He goes on to comment that: the characteristics distinguishing reflex arc conduction from nerve-trunk conduction may therefore be largely due to intercellular barriers, delicate transverse membranes… If the conductive element of the neurone be fluid and if at the nexus between neurone and neurone there does not exist any actual confluence, there must be a surface of separation. Even should a membrane visible to the microscope not appear, the mere fact of non-confluence of the one with the other implies the existence of a surface of separation. Such a surface might restrain diffusion, bank up osmotic pressure, restrict the movement of ions, accumulate electric charges, support a double electric layer, alter in shape and surface tension with changes in difference of potential, alter in difference of potential with changes in surface tension and in shape, or intervene as a membrane between dilute solutions of electrolytes of different concentration or colloidal suspensions with different sign of charge. It would be a mechanism where nervous conduction, especially if predominantly physical in nature might have grafted upon it characteristics just such as those differentiating reflex—arc conduction from nerve—trunk conduction. For instance, change from reversibility of direction of conduction to irreversibility might be referable to the membrane possessing irreciprocal permeability. In Foster’s textbook of 1897 he goes on to comment on the nervous impulse ‘sweeping along’ the axon of one neuron until it is (Foster, 1897): brought to bear through the terminal arborisation on the dendrites of another neuron where the lack of continuity between the material of the arborisation of the one cell and that of the dendrite (or body) of the other cell offers an opportunity for some change in the nature of the nervous impulse as it passes from one cell to the other. There is no doubt that the results of Wallerian degeneration pointed to the likelihood of end bulbs possessing membranes as did the rest of their parent nerve fibre. The problem then presented itself of how such a membrane might relate to the membrane of the underlying neuron membrane on which the end bulb impinged. It was to this region then that the irreversibility of nerve transmission must be ascribed and an explanation sought. It was the histological work on Wallerian degeneration together with the physiological discovery of the irreversibility of transmission that indicated the special nature of this region. It was clear that this region deserved a name that might focus the attention of experimenters and so help delineate its properties. In Foster’s textbook of 1897 Sherrington provided the name (Foster, 1897):
24
EARLY HISTORY OF THE SYNAPSE
So far as our present knowledge goes, we are led to think that the tip of a twig of the arborescence is not continuous with but merely in contact with the substance of the dendrite or cell-body on which it impinges. Such a special connection of one nerve cell with another might be called a ‘synapses’. The origins of this use of the word ‘synapsis’ or as it became ‘synapse’ can be found in letters Sherrington wrote to his colleagues Sharpy-Schafer in 1897 and Fulton in 1937, in reply to enquiries concerning the derivation of the word. To the former he wrote [see Sherrington to Schafer, 1897a; Sherrington to Schafer, 1897b; Sherrington to Fulton, 1937): As to nomenclature—its sole object is I take it clearness combined with brevity… Definition is wanting when a penny has to pass for 5 and 3 farthings as well as for 4: the one symbol is then too little… All I think we ought to be careful not to do is ‘commit barbarisms,’ e.g., impossible adjectival form, using prefixes and affixes with false signification or in impossible ways—that simply adds new terms which like other ‘monsters’ can’t live long, and may be misleading during life, does all of us harm as giving the impression of carelessness or ignorance’. ‘As to ‘junction’ I feel we are less easily reconcilable. If a Latin form capere not jungere should be the root. The mere fact that junction implies passive union is alone enough to ruin the term… I think it does not want the gift of prophecy to foretell that it [the word junction] must become more and more obviously inapplicable as research progresses. Synapse, which implies a catching on, as e.g., by one wrestler of another—is really much closer to the mark. But I am not a bit wedded to the word: if you could suggest an English word containing the notion which is not already overburdened with applications. I have been trying to find one but cannot. Conjunction is even worse than junction. Sherrington wrote to Fulton that (Sherrington to Fulton, 1937): You enquire about the introduction of the term ‘synapse’; it happened thus. Michael Foster had asked me to get on with the Nervous System part (Part III) of a new edition of his ‘Textbook of Physiology’ for him. I had begun it, and had not got far with it before I felt the need of some name to call the junction between nerve-cell and nerve-cell (because that place of junction now entered physiology as carrying functional importance). I wrote him of my difficulty, and my wish to introduce a specific name. I suggested using ‘syndesm’ ( ). He consulted his Trinity friend Verrall, the Euripidean scholar, about it, and Verrall suggested ‘synapse’ (from ), and as that yields a better adjectival form, it was adopted for the book. The concept at root of the need for a specific term was that, as was becoming clear, conduction which transmitted the impulse along the nerve fibre could not—as such— obtain at the junction, a membrane there lay across the path, and conduction per se was not competent to negotiate a cross-wise membrane. At least so it seemed to me then, perhaps A.V.Hill and Gasser and Bishop could tell us differently today. I do not know when the term ‘synapsis’ was introduced for a phase of the karyokinetic process. Neither Foster nor I knew of it in that connection. I fancy Salvin Moore, a cytologist, put it forward. He once told me he had not known the term was in use in physiology. I think that your proposed synaptic knobs would be very clear and helpful. Pace Verrall’s memory (Verrall was a delightful and charming man). ‘Synapsis’ strictly means a process of contact, that is, a proceeding or act of contact, rather than a thing which enables contact, that is, an instrument of contact. ‘Syndesm’ would not have had the defect, that is, it would have meant a ‘bond’. The credit for the word ‘synapse’ then goes to a classical scholar at Cambridge (for a detailed outline of this claim, see Tansey, 1997). Although the word ‘junction’ was abandoned by Sherrington as appropriate to describe the functional relationship between the end bulb and neuron it was preserved for that between motor nerve and muscle at the endplate. Here Wallerian degeneration had also indicated the discreteness of the nerve terminal from the muscle at the endplate, suggesting that neither Boerhaave nor Kuhn were any more correct than Golgi in ascribing continuity between the end of nerve terminals and the cells on which they impinged. The research of Tello, working in Cajal’s laboratory, was particularly persuasive on this issue, as it showed the postjunctional endplate apparatus was intact in frog muscle after denervation, and that reinnervating nerve fibres could be found at different stages of terminal formation on these regions of the muscle (Fig. 1.9D; see Cajal, 1928). Sherrington’s prescient comment that (Sherrington to Fulton, 1937): …conduction per se was not competent to negotiate a cross-wise membrane was followed by the caveat that At least so it seemed to me then, perhaps A.V. Hill and Gasser and Bishop could tell us differently today. Sherrington’s claim that conduction could not per se negotiate the synapse was soon challenged by K.Lucas and later his colleague E.D.Adrian. They produced credible biophysical explanations of how conduction per se could negotiate a cross-
HISTORY OF THE SYNAPSE
25
wise membrane. Their theory showed how even the process of inhibition at synapses could function without the necessity of evoking any new principles other than those involved in the conduction of the action potential. Cajal’s use of arrows showing the direction of action potential conduction on his drawings of silver-stained neurons seem to independently support Sherrington’s notions, with conduction only possible in one direction across the synapse. But of course Cajal’s arrows are placed according to an erroneous idea of conduction by neurofibrils. Too much importance has been placed in the history of neuroscience on Cajal’s arrows and Sherrington’s introduction of the work ‘synapse’. These researchers made great contributions to the delineation of the types of neurons to be found together with their spatial relationships on the one hand and to that of the excitatory and inhibitory processes that these neurons participate in on the other. However one must turn to the research of other investigators in order to find the observations which warrant the use of the words ‘junctions’ and ‘synapses’. Such research was supplied in the twentieth century, and is the subject of a more contemporary history (see Chapters 2 and 3).
2 Emergence of the Concept of Transmitter Release at Peripheral and Central Synapses
2.1 Research on the Synapse in the Laboratories of Sherrington and Langley before the Great War Over ninety years ago Charles Sherrington gave his Silliman Lectures at Yale University which were later published as The Integrative Action of the Nervous System (Sherrington, 1906). In that great work Sherrington laid the conceptual foundations for much that was to dominate research on the central nervous system for the rest of the century. Sherrington had begun his studies on the central nervous system at Cambridge in the physiological laboratory of John Langley and later at Liverpool. In his Silliman Lectures Sherrington pointed out that: In view, therefore, of the probable importance physiologically of this mode of nexus between neurone and neurone it is convenient to have a term for it. The term introduced has been ‘synapse.’ Sherrington had already defined the synapsis in Foster’s Textbook of Physiology some ten years earlier (Sherrington, 1897). He went on to say in the Silliman Lectures that: The neurone itself is visibly a continuum from end to end, but continuity, as said above fails to be demonstrable where neurone meets neurone—at the synapse. There a different kind of transmission may occur. The delay in the gray matter may be referable, therefore, to the transmission at the synapse. Regarding how synapses operate, he said: It would be a mechanism where nervous conduction, especially if predominantly physical in nature, might have grafted upon its characters just such as those differentiating reflex-arc conduction from nerve-trunk conduction. Sherrington had developed these ideas as a consequence of his studies on the reflex contractions of muscles following stimulation of muscle and skin afferents. His summary diagram of the place of excitation and inhibition in reflex pathways for flexor activation and extensor inhibition, shown in Fig. 2.1, is a masterpiece of fruitful speculation. This diagram not only indicates the concept of excitatory and inhibitory synapses, developed clearly by 1908, but draws the experimentalist into Sherrington’s line of inquiry as to what other nervous pathways may be delineated by this approach: in particular, how is the information transferred at the nerve terminal across the synaptic gap in the drawing, and what is the mechanism of inhibition. A research program entirely different from Sherrington’s was directed by his mentor J.N.Langley at Cambridge. In 1903 Langley, who had introduced Sherrington to neurophysiology (Langley & Sherrington, 1884), was at that time laying the foundations for our understanding of the chemical nature of transmission at synapses. In 1901 Langley published a remarkable paper (Langley, 1901) showing that stimulation of the sympathetic component of the autonomic nervous system, which Gaskell and he had already defined, resulted in changes in the effectors that in many cases could be mimicked by application of suprarenal extract (adrenaline). In Langley’s words: I have formerly divided the autonomic nervous system into sympathetic, cranial, sacral and enteric. It is a noteworthy fact that the effect of supra-renal extract in no case corresponds to that which is produced by stimulation in normal conditions of a cranial autonomic or of a sacral autonomic nerve…. It is equally noteworthy that the effects produced by supra-renal extract are almost all such as are produced by stimulation of some one or other sympathetic nerve. In many cases the effects produced by the extract and by electrical stimulation of the sympathetic nerve correspond exactly (see Fig. 2.2A)…. It is hardly possible to avoid the conclusion that in these cases the extract acts directly on the unstriated muscle, and if this is so, it is probable that in all cases the action is direct. The theory of direct action cannot, however,
EMERGENCE OF THE CONCEPT OF TRANSMITTER RELEASE AT PERIPHERAL AND CENTRAL SYNAPSES
27
Fig. 2.1. Sherrington's 1906 diagram `indicating connections and actions of two afferent spinal root-cells (dorsal root ganglia), and ' in regard to their reflex influence on the extensor and flexor muscles of the two knees. Flexor and extensor muscles of the knee joint on the right (R) and left (L) sides are shown together with the inputs to the spinal cord by a cutaneous afferent ( ) and a muscle spindle afferent ( ). The reflex pathways postulated show flexor (F) excitation (+) and extensor (E) inhibition (−) ipsilaterally and flexor (F) inhibition (−) and extensor (E) excitation (+) contralaterally. In Sherrington’s 1906 words, ‘the sign +indicates that at the synapse which it marks the afferent fibre ( (and ( ) excites the motor neurone to discharge activity, whereas the sign —indicates that at the synapse which it marks the afferent fibre (and ) inhibits the discharging activity of the motor neurones. The effect of strychnine and of tetanus toxin is to convert the minus sign into a plus sign’.
be regarded as more than provisional until it is shown experimentally that the inhibitory action of suprarenal extract on certain unstriated muscle, and its stimulating action on salivary gland cells take place in the absence of nerve-endings. These points I propose to consider in a later paper. These experiments were carried out by Langley’s student T.R.Elliott who concluded in a note to The Journal of Physiology (owned and edited by Langley) in 1904 that: Adrenalin might then be the chemical stimulant liberated on each occasion when the impulse arrives at the periphery (Elliott, 1904a). The idea of chemical transmission at the synapse, and indeed of receptors on the effector organ for receiving the chemical substance released by the nerves, was already a central part of the research program in Cambridge physiology under Langley (Langley, 1906). This research was furthered in 1906 by W.E.Dixon. Working in the Cambridge physiology laboratory on the effect of suprarenal extracts on the lung (Brodie & Dixon, 1904) Dixon decided to perform an experiment in which he took an extract of a dog’s heart that had undergone vagal stimulation and applied it to the exposed heart of a frog, obtaining an interruption of the heart beat (Fig. 2.2B). This work was similar in design to Otto Loewi’s famous 1921 experiment (see Fig. 2.3A) some 14 years later, establishing the idea of chemical transmission in the heart unequivocally (Loewi, 1921; Dale, 1934). Henry Dale had observed these experiments of Dixon’s. Dale came across acetylcholine accidentally in 1914, as a constituent of a particular sample of ergot. Here he describes his finding: I was led to make a detailed study of its action. This, I think, gave the first hint that acetylcholine might have an interest for physiology. Then I was struck by the remarkable fidelity with which it reproduced the various effects of parasympathetic nerves, inhibitor on some organs and augmentor on others—a fidelity which I compared to that with which adrenaline reproduces the effects of the other, true sympathetic, division of the autonomic system (Dale, 1914b) At the beginning of the century Sherrington had already placed both excitatory and inhibitory synapses at center stage in the integrative behaviour of the spinal cord. Furthermore, Langley’s school had shown that synapses at the autonomic neuroeffector junctions were likely to operate by the secretion of a chemical substance, which in the case of the sympathetic was related to adrenaline. Research over the next 25 years attempted to unravel the principles of operation of synapses within a conceptual framework that originated in the great research schools formed by Langley and Sherrington.
28
HISTORY OF THE SYNAPSE
Fig. 2.2. The evolution of the idea of chemical transmission at synapses. (A) Langley’s 1901 table showing the effect of suprarenal extract (adrenaline) in the cat and rabbit arranged roughly in the order of the amount of extract required per body weight to produce an obvious effect. Langley notes that: ‘I have formerly divided the autonomic nervous system into sympathetic, cranial, sacral and enteric. It is a noteworthy fact that the effect of supra-renal extract in no case corresponds to that which is produced by stimulation in normal conditions of a cranial autonomic or of a sacral autonomie nerve. It is equally noteworthy that the effects produced by supra-renal extract are almost all such as are produced by stimulation of some one or other sympathetic nerve. It is hardly possible to avoid the conclusion that in these cases the extract acts directly on the unstriated muscle, and if this is so, it is probable that in all cases the action is direct. (B) Unpublished record from a 1906 experiment by W.E.Dixon showing the beat of the exposed heart of a frog. At the first mark, extract from a dog’s heart that had been inhibited by vagal stimulation was applied; at the second mark, atropine was applied (from Dale, 1934b). This is the first known record of an attempt to determine if a nerve secretes a substance that, when placed on another organ, will mimic the effects of nerve stimulation to that organ.
2.2 Sherrington's concept of the inhibitory and excitatory states of central synapses John Eccles was born in 1903 and he entered Melbourne University Medical School at the very young age of 15, in the year that saw the end of the Great War, and later won a Rhodes Scholarship to Oxford in 1925 to work with Sherrington. Eccles entered an intellectual environment on synaptic transmission that was now dominated by Loewi’s recent experiments (Loewi, 1921) indicating that chemical transmission occurred between the vagus and the heart, and Dale’s work indicating a role for
EMERGENCE OF THE CONCEPT OF TRANSMITTER RELEASE AT PERIPHERAL AND CENTRAL SYNAPSES
29
acetylcholine in synaptic transmission (Dale, 1914; Dale, 1910; Loewi, 1935) (Fig. 2.3A). Eccles entered Oxford the year J.N.Langley died at the age 73 after completing a six-hour experiment in Cambridge. He arrived at a time when Sherrington was engaged in research with Liddell on the characteristics of the myotatic reflex (Liddell & Sherrington, 1924; Liddell & Sherrington, 1925) and with Creed on the flexion reflex (Creed & Sherrington, 1926). Sherrington had just produced a masterly summary of work on inhibition, in which he concluded that: In relation to inhibition at the synapse that it might be mediated by an agent, moreover, one whose existence lies outside the intrinsic properties of pure nerve-fibre and with a, so to say, more chemical mode of origin and function than the nerve impulse per se (Sherrington, 1925). This comment was made only four years after the experiments of Loewi. Eccles joined Creed in his first experimental work, which was on the subject destined to dominate his research life for over 40 years: the mechanism of inhibitory synaptic transmission (Creed & Eccles, 1928). Then in 1929 he joined Sherrington in a technical improvement of the torsion myograph (Eccles & Sherrington, 1930a) in preparation for a collaboration (that lasted but a few years, 1929–1931) concerned with research on the flexion reflex and inhibition (Eccles & Sherrington, 1930b; Eccles & Sherrington, 1931), These experiments were to see the last flowering of Sherrington’s scientific genius at the age of 75. The work on the ipsilateral spinal flexion reflex introduced Eccles to the technique of stimulating first with a just threshold conditioning volley, then at later intervals with a subsequent test volley in order to tease out the time course of the central excitatory state (Eccles & Sherrington, 1930a) (Fig. 2.4A). This approach, when applied to the mechanism of monosynaptic transmission in the spinal cord, gave a very precise measure of the time course of the central excitatory state or, as we now know, the excitatory postsynaptic potential (Fig. 2.4). With hindsight it might be expected that chemical transmission would seem to be the most likely mechanism for determining the central excitatory state (c.e.s.) and the central inhibitory state (c.i.s.), based on the experiments of Langley and his school, along with those of Loewi and Dale. This was certainly not the case, as is shown in the next section. 2.3 Lucas, Adrian and the electrical concept of the inhibitory state of central synapses Towards the end of Langley’s career, the Cambridge School of Physiology came to be dominated by those such as Keith Lucas and E.D.Adrian who were introducing electrophysiological techniques into the study of how impulses conduct in excitable tissue. Lucas had published a Physiology Monograph entitled The Conduction of the Nervous Impulse in which he argued that central inhibition might be brought about by the interference of high-frequency discharges in the nerves as they approach their synaptic connections on neurones (Lucas, 1917). In this way the refractory state of the axon following an impulse could operate to produce inhibition. This idea was followed up in detail in 1924 by Adrian (Fig. 2.5A), who was skeptical about the recent research of Loewi and Dale, commenting: It appears, then that the fluid coming from the stimulated organ reproduces the characteristic effects of the vagus or sympathetic on different tissues, though whether every detail of the nervous effect is copied by the fluid remains an open question. The nature of the ‘vagus substance’ is uncertain. If these results can come to be generally accepted we shall have a new and extremely interesting picture of the action of the autonomic system. The difficulty is that the effects seem to be capricious and are not easily reproduced. Some observers have failed to satisfy themselves that they occur at all outside the margins of the experimental error. In view of this uncertainty we can only wait until there is a more general agreement as to the experimental basis of the humoral theory (Adrian, 1924). In expanding on the Lucas theory of the electrical basis of spinal cord inhibition, Adrian went on to say that: If it (the humoral theory) is correct, the explanation of peripheral inhibition resolves Itself into that of (a) the secretory mechanism which produces the inhibiting substance whenever impulses pass along certain fibres to the muscle, and (b) the way in which an inhibiting substance, adrenalin for instance, exerts its effect on the muscle fibre… If we compare the present theory (the electrical theory of Figure 2.5A), or some modification of it, with the view which supposes that inhibition is due to the production of a special substance which blocks the excitatory paths to the motor neurone, it will be seen that there is actually not very much difference between them. If an inhibiting substance is produced, its production must be almost instantaneous and it must disappear very rapidly; what the present theory assumes is that the ‘substance’ is to be identified with the refractory state and that sustained inhibition is due to a series of refractory states and not to a steady production of an inhibiting substance (Adrian, 1924).
30
HISTORY OF THE SYNAPSE
Fig. 2.3. The first apparently unequivocal demonstration of chemical transmission at a synapse and characterisation of the accompanying synaptic potential. This was performed for the vagal inhibition of the heartbeat. (A) Otto Loewi’s 1921 original record in which at 1 the heartbeat is shown in normal ringer; at 2, the decline in the heartbeat is due to the addition of a ringer that had been in contact with another heart whose vagus had been stimulated for 15 minutes; at 3, the heartbeat is normally in the presence of a ringer from another heart in which the vagus had not been stimulated; finally, in 4, atropine was added in a normal ringer, increasing the heartbeat. (B) Curve showing the extent of inhibition of the heartbeat (on the vertical axis) due to a single stimulus in an experiment performed by Brown and Eccles (1934). A single stimulus is applied to the vagus nerve and the lengthening of each cardiac cycle (i.e., the amount by which it exceeds a normal cycle) is expressed as a fraction of the normal cycle (of about 305 msec) on the ordinate; the abscissa gives the interval between the vagal stimulus and the end of that particular cycle. There is a latent period of rather more than 100 msec before a volley in the postganglionic fibres produces an inhibitory effect on the pacemaker. (C) Intracellular records of the hyperpolarisation in the arrested frog’s heart due to a single vagal volley by del Castillo and Katz (1957). Note that the latency between the vagal volley and the hyperpolarisation is several hundred msec. Calibration is (small vertical bar) 1mV and indicates the moment of stimulation, which occurs 400 msec before the hyperpolarisation commences.
Adrian arrived at this conclusion as a consequence of experiments performed with Bronk on the frequency of discharges in motoneurones accompanying reflex and voluntary contractions. It was not until 1929 that Adrian felt able to abandon the idea that central inhibition could be explained along the lines suggested by Lucas (i.e., by the depressant effects produced by highfrequency impulse discharges) (Adrian & Bronk, 1929). Adrian carried great authority on matters concerned with electrical activity in nerves at the time Eccles arrived at Oxford in 1925. He had just successfully recorded for the first time the trains of
EMERGENCE OF THE CONCEPT OF TRANSMITTER RELEASE AT PERIPHERAL AND CENTRAL SYNAPSES
31
Fig. 2.4. The time courses of the central excitatory states (c.e.s.) and central inhibitory states (c.i.s.) compared with those of the excitatory and inhibitory postsynaptic potentials. A, time course of the central excitatory state (c.e.s.) determined using muscle reflexes. It shows the reflex response of the tibialis anticus muscle to two stimuli (each one of which alone is just sub-threshold) to medial gastrocnemius nerve and lateral gastrocnemius nerve at various intervals, as measured by Eccles and Sherrington (1930). The abscissa gives the time, approximately in msec, and the ordinate the tension in grams. Right of zero shows the interval by which the stimulus to the lateral gastrocnemius nerve is leading; left of zero shows the interval by which stimulus to the medial gastrocnemius nerve is leading to. Each point plotted shows the tension developed at the indicated interval between stimuli. The excitatory state, set up by the conditioning volley, lasts for about 15 msec and is called the ‘central excitatory state’ (c.e.s.). B, time course of the central excitatory state (c.e.s.) and central inhibitory state (c.i.s.) determined by measuring the compound action potential in a muscle nerve, a, the extent to which the response to a test reflex is enhanced by a prior conditioning reflex (facilitation) at the different intervals given in the abscissa, as measured by Lloyd (1946). The facilitation of the biceps reflex by afferent volleys in the semitendinosus nerve and of one head of the gastrocnemius by afferent volleys in nerves to the other head are given. Conditioning volleys of near reflex threshold strength were used. The relative facilitation, expressed in percent maximum, is plotted as a function of time and gives the c.e.s.; this is similar to that given by the method used by Eccles and Sherrington in A. b, the extent to which the response to a test reflex is inhibited by a prior conditioning reflex at the different intervals given in the abscissa, as measured by Lloyd (1946). The inhibition of the reflex of tibialis anterior by weak volleys to the gastrocnemius afferent nerve are given. The ordinate gives the degree of inhibition, in percent of maximum, to the time interval between volleys on the abscissa. This relative inhibition, expressed in percent maximum, is plotted as a function of time and gives the c.i.s. C, intracellular potentials recorded by Brock et al. (1952) from a biceps semitendinosus motoneuron due to two afferent volleys in the biceps semitendinosus nerve; note that the dorsal root spikes accompanying each volley are shown beneath the intracellular recordings. These excitatory postsynaptic potentials give the electrical signs of the central excitatory state measured by Eccles and Sherrington in 1930 and shown in A; note the similar time course. D, intracellular potentials recorded by Brock et al. (1952) from a biceps semitendinosus motoneuron due to a single volley in a quadriceps nerve, of increasing size downward; note the increasing size of the dorsal root spikes shown beneath the successive intracellular recordings. These inhibitory postsynaptic potentials give the electrical signs of the central inhibitory state shown in B, b; note the similar time course.
nerve impulses traveling in single sensory and motor nerve fibers which, according to A.L.Hodgkin, ‘marks a turning point in the history of physiology’ (Hodgkin, 1978).
32
HISTORY OF THE SYNAPSE
2.4 Loewi, Dale and Eccles examine the inhibitory state at autonomic neuromuscular junctions The chemical or electrical nature of synaptic transmission in either the peripheral or the central nervous system was an entirely unsettled issue when Eccles arrived at Oxford. It seemed likely, despite Adrian’s skepticism, that vagal inhibition of the heartbeat was chemical and that the ‘vagus substance’ or ‘Vagusstoff’ was acetylcholine. However, it must be remembered that Otto Loewi and Henry Dale did not win the Nobel Prize for their investigations until 1936 and Loewi still felt constrained to defend this idea as late as 1935. Eccles, after completing his last work with Sherrington in 1931 on spinal cord inhibition, sought to delineate the characteristics of this chemical inhibition; he used the only appropriate preparation available at the time, namely the vagus to the heart. Together with G.L.Brown he determined that the time course of the inhibitory state (analogous to the c.i.s.) following a single stimulus to the vagus nerve was of the order of a second or more and did not arise for over 100 msec after a stimulus (Brown & Eccles, 1934) (Fig. 2.3B). This determination was supported 20 years later with the introduction of intracellular recording of the inhibitory junction potential in the heart by del Castillo and Katz (del Castillo & Katz, 1957) (Fig. 2.3C). Brown and Eccles commented that: If the vagal volley is set up late in a cardiac cycle, that cardiac cycle is not inhibited, the latent period of the inhibition being usually 100 to 160 ms. Of this amount the conduction time to the region of the pacemaker probably only accounts for about 10 ms, ie. the greater part of the latent period appears to occur after the arrival of the inhibitory impulses at the nerve fibres of the pacemaker. It is probable that most of this time is occupied in the liberation of the acetylcholine substance and its diffusion to the point of its action (Brown & Eccles, 1934). This slow time course of the only reasonably well established chemical synapse seemed to set the temporal characteristics of chemical transmission. This was particularly so for other synapses at which nerve terminals were shown to secrete acetylcholine. 2.5 Eccles develops the electrical concept of the excitatory state at autonomic synapses Eccles then turned his attention to the only readily accessible synapse on neurons for which acetylcholine was known to be released on nerve stimulation, namely that in sympathetic ganglia (Feldberg & Gaddum, 1934; Eccles, 1937) (Fig. 2.6A). It was only natural that he should first approach the problem of defining the excitatory state in the ganglion by the same methods developed to study the time course and other characteristics of the c.e.s. of motoneurones in the spinal cord (Fig. 2.4A). Examination was made of the interaction of submaximial volleys to each of two preganglionic inputs to a ganglion delivered at different intervals apart in test-conditioning pairs, and the ganglionic action potential measured (Fig. 2.6B(a)). In this case the time course of compound action potentials was being determined rather than the time of reflex contractions of muscles used to determine the c.e.s. of motoneurons (Fig. 2.4A). Compound action potential waveforms had to be subtracted in the manner indicated in the legend to Fig. 2.6B(a) before the c.e.s. could be estimated; this figure shows that the test compound action potential seems to be elevated compared with the compound action potential in the absence of a conditioning impulse. This elevation occurs from the earliest times for the test-conditioning interval, then declines to zero at an interval of 4.5 msec, increases again, and reaches a peak at 17.6 msec from which it slowly declines over the next 40 msec or so to zero. The time course of these events is indicated in Figure 2.6B(b) which shows the early fast phase of the potentiation of the test compound action potential followed by the later developing slow component. The fast phase, termed the ‘detonator response’ (Eccles, 1937) was not affected by anticholinesterases as is vagal inhibition of the heart; furthermore, it is much faster than the action of acetylcholine on the heart (Fig. 2.3B). The detonator response was not then attributed to the action of acetylcholine but rather to the action currents in the preganglionic nerve terminals that produced a potential response in the ganglion cells; the later phase was identified as the excitatory state. This analysis led to the proposition that: On present evidence however, it seems that the action-current hypothesis offers a more probable explanation for direct synaptic transmission, the acetylcholine liberated in sympathetic ganglia possibly having a secondary excitatory action as already suggested (Eccles, 1936). The analysis of synaptic transmission from the postganglionic nerves to the smooth muscle of the nictitating membrane also revealed an excitatory state that had an initial fast component followed by a slower late component (Monnier & Bacq, 1935; Eccles & Magladery, 1936; Eccles & Magladery, 1937). In this case the anti-adrenaline drug 933F (piperidonethyl-3benzodioxane) blocked the slow electrical response and the associated contraction but not the fast response, suggesting that the latter might also indicate the signs of electrical transmission (Eccles, 1936; Eccles & Magladery, 1936). Similar
EMERGENCE OF THE CONCEPT OF TRANSMITTER RELEASE AT PERIPHERAL AND CENTRAL SYNAPSES
33
Fig. 2.5. Electrical theories of central inhibition. (A) Scheme due to Adrian (1924) based on an idea of Keith Lucas (1917; see his Fig. 22) in which inhibition occurs as a consequence of the refractory state left in a nerve following an impulse. In this diagram, A is the excitatory pathway and B the inhibitory pathway converging on the motoneuron M; the shaded areas conduct with a decrement. By making the decremental path from B to M longer than that from A, we can account for the fact that an impulse from B can never succeed in reaching M, whereas impulses from A can do so provided the path is given time for complete recovery between each impulse. With such an arrangement, an impulse from B, though not itself exciting M, would leave the common pathway in a refractory state, absolute or relative, which would hinder the passage of impulses from A for a time depending on the rate of recovery of the path and the extent of the decrement in it. A rapid succession of impulses from B would produce continuous inhibition by never giving time for the complete recovery of the common pathway. If the impulses from B also pass by a more direct route, dotted in the figure, to the antagonistic motoneuron M , the periods of inhibition of M would synchronize with the discharge of motor impulses from M . (B) Scheme due to Brooks and Eccles (1947) of how electrical inhibition could occur in the spinal cord. The diagram indicates current flow at a schematic synapse of a Golgi cell G on a motoneuron M according to this electrical theory of inhibition. E shows the excitatory line to M, and I the inhibitory line that subliminally excites G and so generates the current flow producing an electrotonic focus on M. According to this theory, an intracellular electrode would record a brief positively going electrical field at X.
difficulties were arising with the parasympathetic innervation of the bladder. In this case only the slow phase of contraction could be blocked by atropine and mimicked by applied acetylcholine; this left the fast phase to be accounted for in terms of an electrical component of transmission (Henderson & Roepke, 1934). In order to escape this awkward fact, Dale and Gaddum (1930) suggested that the concentration of adrenaline secreted at the sympathetic postganglionic nerve terminal was very high, giving the fast response, and was therefore insensitive to 933F; the slow reaction was attributed to the escape of adrenaline and its secondary diffusion from the synaptic cleft to act on other cells that were sensitive to 933F. As late as 1937 Dale was reiterating that:
34
HISTORY OF THE SYNAPSE
Fig. 2.6. Claims for electrical and chemical transmission at synapses. (A) Demonstration of how stimulation of the cervical sympathetic causes liberation from the ganglion of a substance pharmacologically identified by Feldberg and Gaddum (1934) as acetylcholine. The top panels show the response of the frog’s beating heart and the bottom panels that of contraction of leech muscle as a consequence of adding Ringer’s fluid collected from a ganglion during preganglionic stimulation (A); adding Ringer’s fluid with different concentrations of acetylcholine (B and C), and adding Ringer’s fluid from an unstimulated ganglion (D). Note that the fluid from the stimulated ganglion has identical effects on slowing the beating heart and contracting the leech muscle, as does acetylcholine. (B) Electrophysiological evidence involving the study of action potentials interpreted by Eccles (1937) as showing an electrical component to synaptic transmission. In (a), single submaximal stimuli were applied to two different preganglionic nerve branches at various intervals apart; the continuous lines show the ganglionic action potentials produced by a second stimulus at the indicated intervals and the broken lines show the action potential set up by the second stimulus alone. It will be noted that when the two volleys are simultaneous, spatial facilitation is maximal (i.e., it is effective in producing a discharge from the largest number of ganglion cells); this number is still large at the interval of 2.3 msec. At 4.5 msec, effective summation occurs in very few, if any, ganglion cells. At intervals longer than 4.5 msec, spatial and temporal facilitation again develop and the time course of decay of this second facilitation wave is due to the excitatory state of the ganglion cells, which would now be called the excitatory postsynaptic potential. In (b) the time course of the ‘detonator response’ is shown and the excitatory state derived from experiments such as those in (a); the former is attributed to electrical transmission at the synapse and the latter to chemical transmission using acetylcholine to give the synaptic potential (from Eccles, 1936). (C) Electrophysiological evidence involving the direct study of synaptic potentials in the absence of action potentials interpreted by Eccles (1943) as showing that only chemical transmission occurs in ganglia, (a) shows a single synaptic potential, recorded with extracellular electrodes, which lasts for about 100 msec. The faster curve is a theoretical estimate of the time course of transmitter action that gives rise to the synaptic potential, based on the ‘local potential’ theory of A.V.Hill. Ordinates for transmitter actions are in arbitrary units, (b) shows the theoretical times for the decline in the amount of transmitter remaining within a sphere of either radius 2 µm (B) or 1 µm (D), as well as a cylinder of radius 2 µm (A) or 1 µm (C) following the transmitter’s instantaneous deposition (from Ogston, 1955). The curves show that the concentration declines with a similar time course to that of the time course of transmitter action given in (a), suggesting that free diffusion of acetylcholine out of a synaptic region with the dimensions of 1 µm or 2 µm could account for the observed results. (D) First synaptic potentials recorded with an intracellular electrode from a mammalian sympathetic ganglion (from R.Eccles, 1955).
This antagonist (atropine) cannot similarly intervene, when acetylcholine is liberated front nerve endings in immediate contact with, or even inside the cell membrane (Dale, 1937).
EMERGENCE OF THE CONCEPT OF TRANSMITTER RELEASE AT PERIPHERAL AND CENTRAL SYNAPSES
35
Dale thought that the fast phase of contraction of smooth muscle on nerve stimulation was due to the high transmitter concentration reached at the nerve endings, with the slow contraction attributed to its later diffusion to other muscle cells (Dale, 1937). He had often used different renditions of this argument to escape the unpalatable fact that neither adrenaline nor acetylcholine seemed suitable candidates for transmission to some smooth muscles. For example, in 1934 Dale said that: There are some parasympathetic effects, such as the action of the vagus on the intestine, and the vaso-dilator action of parasympathetic nerves in general, which are resistant to atropine, though the otherwise similar actions of injecting or applying acetylcholine are readily abolished by it… Gaddum and I suggested that in such cases the nerve impulses liberate acetylcholine so close to the reactive structures that atropine cannot intervene, whereas it can prevent its access to them when it is artificially applied from without (Dale, 1934). Following the discovery of inhibitory junction potentials in smooth muscle in 1963 (Bennett et al., 1963) that are mediated by nonadrenergic noncholinergic (NANQ) synapses (Bennett et al., 1966) we now know of many synapses that do not conform to Dale’s paradigm. Eccles developed his ideas on electrical transmission in Oxford between 1934 and 1937. He then returned to Australia in 1937 to become Director of the Kanematsu Memorial Institute of Pathology at Sydney Hospital. His first studies there were concerned with the possibility of electrical transmission at the somatic neuromuscular junction. In 1936 G.L.Brown had shown that close interarterial injection of acetylcholine into the cat’s gastrocnemius gave repetitive impulse firing and contraction of the muscle (Fig. 2.7B). Protection of the metabolism of acetylcholine with eserine allowed it to appear in a perfusate after stimulation of motor nerves (Dale et al., 1936; Brown et al., 1936). By 1938 both Eccles and O’Connor (1938) as well as Gopfert and Schaefer (1938) had recorded the endplate potential with extracellular electrodes in curarized muscles (Fig. 2.7C). This was probably the first time a synaptic potential had been observed without distortion due to electrotonic conduction. The possibility that the endplate potential was due to the release of acetylcholine was not grasped. In 1939 Eccles and O’Connor used the method of applying conditioning-test volleys to mammalian motor nerves and recording the impulses in a muscle in order to determine the time course of the excitatory state. They concluded that a nerve impulse exerts two excitatory actions at the neuromuscular junction: (1) Newborn muscle impulses are set up by a brief excitatory action probably no more than 1 msec in duration and analogous to the detonator action described for synaptic transmission. (2) The much more prolonged end-plate potential is set up independently of the newborn impulses, but if the growth of these impulses is sufficiently delayed, it appears to aid in their growth to the fully propagated size. It is analogous to the N wave and the associated central excitatory state of synaptic transmission, and analogous responses have also been described at the neuromuscular junction of smooth muscle (Eccles & O’Connor, 1939). Thus the test-conditioning volley approach used so successfully to determine the time course of the c.e.s. for motoneurons now led to the erroneous conclusion that a very fast response, much faster than the endplate potential recorded at the neuromuscular junction in curarized preparations, was the primary means of transmission. 2.6 Katz, Kuffler and Eccles establish the motor endplate as the paradigm synapse for electrophysiology J.N.Langley and Keith Lucas initiated the studies of A.V.Hill on the biophysics of nerve and muscle at Cambridge in 1909, and Hill in turn supervised the first research of Bernard Katz on electrical excitation and conduction of the nerve impulse at University College London in 1935 (Hill et al., 1936a,b; Katz, 1978). The frog isolated gastrocnemius sciatic nerve preparation was used by Katz in his investigations of Hill’s theory of excitation in order to provide an index for the duration of maintained nerve excitation (Katz, 1936). However it was the Cambridge zoologist Carl Pantin, who had acted as a guide to A.L.Hodgkin’s studies, that was responsible for the first explicit research by Katz on neuromuscular transmission (Katz, 1936). In this work it was shown that magnesium ions could block neuromuscular transmission in crabs. By 1939 Katz was using the frog isolated sartorius-nerve preparation after curarization (Katz, 1939) to confirm the work of Gopfert and Schaefer that showed: A small non-conducted potential change is to be found in the myoneural region, which reaches a maximum 4 msec after arrival of the nerve impulse, and then falls at a slow rate, similar to the electrotonic potential (Gopfert & Schaefer, 1938).
36
HISTORY OF THE SYNAPSE
It was natural that Katz should have wanted to continue research on this ‘small non-conducted potential change’ or endplate potential, using the frog sartorius-nerve preparation, when he joined Eccles and Stephen Kuffler at the Sydney Kanematsu Institute in 1939 (Fig. 2.7A). Rather than working on the innervation zones of the cat’s soleus muscle with Eccles, Katz did the following: I ganged up with Stephen Kuffler, and I was very pleased when we succeeded in getting hold of some nice Australian tree frogs whose sartorius muscles proved to be very suitable for the experiments we wanted to do, and this kept me busy and moderately happy for two years (Katz 1986). The work of Katz, Kuffler, and Eccles in Sydney (Eccles & Kuffler, 1941; Eccles et al., 1942) marks the beginning of a new era in synaptic physiology after the one begun 50 years earlier by Langley and Sherrington. It is characterised by the use of progressively more sophisticated electrical techniques to probe the mechanism of synaptic transmission. The first experiments of Eccles, Katz, and Kuffler did not involve test-conditioning volleys to estimate (the time course of the excitatory state but rather concentrated on the properties of the extracellular signs of the endplate potential made subthreshold by a suitable dose of curare (Figure 2.7D). They showed, using an analysis provided by A.V.Hill (1933), that the time course of the underlying transmitter action lasted for only a few msec. As Eccles, Katz, and Kuffler stated: Thus it seems that most of the declining phase of the e.p.p. is a passive decay of a negative membrane charge after the depolarizing agent has ceased to act. The earlier suggestion, therefore, that the decline of the endplate potential follows the time course of a passively decaying electrotonic potential is confirmed (Eccles et al., 1941). They stated further: By making plausible assumptions it is shown that the observed curare and eserine actions are reconcilable with the hypothesis that acetylcholine is responsible for all the local potential changes set up by nerve impulses (Kuffler, 1942). Eccles then abandoned the hypothesis of an early ‘detonator’ electrical response followed by a slower endplate potential due to the secretion of acetylcholine. In the following year, Eccles used an analysis similar to that applied to the neuromuscular junction when he evaluated extracellular recordings of the synaptic potential in curarised sympathetic ganglia. This led to the conclusion that (Fig. 2.6C(a)): The results conform well with the postulate of a single depolarizing agent…it is concluded that most and possibly all of the evidence for the detonator action may now be attributed to the brief transmitter action (Eccles, 1943). The time course of transmitter action following a single impulse here could be shown to conform to free diffusion of acetylcholine from the synaptic cleft (Ogston, 1955) rather than to the hydrolysis of acetylcholine by cholinesterase (Fig. 2.6C (b)). Katz returned to A.V.Hill’s laboratory in 1946. Using the frog sartorius muscle nerve preparation once more, and as a result of the introduction of the microelectrode in 1949 by Ling and Gerard (Ling & Gerard, 1949), he was able to confirm with Fatt that the endplate potential alone initiated the muscle action potential (Fatt & Katz, 1951); (compare Fig. 2.7D with Fig. 2.7E). This idea was soon generalised for the nervous system when Rosemary Eccles (Fig. 2.6D) showed that the synaptic potential in sympathetic ganglia alone initiated the action potential (Eccles, 1955). 2.7 Eccles elucidates the electrical signs of the inhibitory and excitatory states of central synapses The introduction of the microelectrode also allowed for the first time an investigation of whether the c.e.s. and c.i.s. of a motoneuron could be described in terms of synaptic potentials, and also whether these synaptic potentials were likely to be due to a ‘detonator’ electrical effect or the secretion of a transmitter. Lloyd (Lloyd, 1946) had already utilized the testconditioning volley method described 16 years earlier by Eccles and Sherrington (1930) to determine this c.e.s. He used the stretch reflex and the more accurate method of electrical recording from the muscle nerves rather than muscle contraction. Lloyd’s results for the time course of the c.e.s. of motoneurons were similar to those of Eccles and Sherrington (compare Fig. 2.4B(a) with Fig. 2.4A); he also gave the time course of the c.i.s. using this method (Fig. 2.4B(b)). Eccles still thought it possible that electrical transmission might account for the c.i.s. of motoneurons; he provided a Golgi cell model of this as late as 1947 (Brooks & Eccles, 1947) (Fig. 2.5B), even though he had abandoned the idea of electrical transmission in the peripheral nervous system. It must be remembered that at this stage the chemical transmitters acting on motoneurons were
EMERGENCE OF THE CONCEPT OF TRANSMITTER RELEASE AT PERIPHERAL AND CENTRAL SYNAPSES
37
Fig. 2.7. The electrical and chemical analysis of synaptic transmission at the neuromuscular junction. (A) S.W.Kuffler, J.C.Eccles and B.Katz in 1941 at the time of their experiments on elucidating the nature of the endplate potential. They are shown in Martin Place, Sydney, walking from the Kanematsu Memorial Institute of Pathology at Sydney Hospital to catch a tram to the University of Sydney to give a lecture in the Physiology Department (B) First recordings of the electrical and mechanical responses of the cat’s gastrocnemius muscle to close intra-arterial injections of two different concentrations of acetylcholine, taken by G.L.Brown and communicated to J.C.Eccles (1936). (C) First endplate potentials to be recorded with extracellular electrodes in curarized (a) cat soleus muscle (Eccles & O’Connor, 1938) and (b) frog sartorius muscle (Gopfert & Schaefer, 1938). (D) Endplate potential recorded with an extracellular electrode in a curarised frog sartorius muscle by Eccles et al. (1941); the numbers on the records refer to the distance in mm from the pelvic end of the muscle. (E) Tracings of endplate potentials recorded with an intracellular electrode from the frog sartorius by Fatt & Katz (1951); the numbers refer to different distances in mm of the recording electrode from the endplate (compare with D above).
unknown, research having shown that neither acetylcholine nor adrenaline were likely to be secreted at inhibitory or excitatory synapses. The first intracellular recordings of synaptic potentials in motoneurons were awaited with considerable interest. Eccles had recently left Dunedin and taken up the foundation Chair of Physiology at the John Curtin School of Medical Research in Canberra. While there, he, Brock, and Coombs published their classic paper ‘The recording of potentials from motoneurones with an intracellular electrode’ in 1952. As in the peripheral nervous system, the c.e.s. and the c.i.s. were
38
HISTORY OF THE SYNAPSE
Fig. 2.8. The discovery of transmitter quanta and synaptic vesicles. (A) Spontaneous miniature endplate potentials in frog sartorius muscle treated with 10−6 prostigmine bromide (Fatt & Katz, 1952). (B) Distribution of amplitudes of spontaneous miniature endplate potentials from an endplate in the frog sartorius muscle treated with prostigmine; there are 800 miniature potentials in the sample (Fatt & Katz, 1952a). (C) Histogram showing distribution of amplitudes of spontaneous miniature potentials and endplate responses at a low calcium endplate in the frog sartorius muscle. The lower part shows a continuous curve that has been calculated according to the hypothesis that the responses are built up statistically of units whose mean size and amplitude distribution are identical to those of the spontaneous potentials. The expected number of failures are shown by the arrows. Abscissae: scale units mean amplitude of spontaneous potentials (0.875 mV) (del Castillo & Katz, 1954a). (D) Electron micrograph of a section of an area of complex axonal entanglement from the neuropile of the earthworm. Numerous profiles of axonal membranes can be distinguished, varying in density from place to place. Mitochondria (M) and endoplasmic reticulum are distinguishable in several places. Y denotes an area of specialized axonal contact identified as synaptic in nature. Numerous synaptic vesicles (SV) are seen in the presynaptic neuron (PRSN), whose profile is of irregular outline. The profile of the postsynaptic member (PSN) is identified as a section through a finger-like axonal projection indenting the presynaptic axon, producing puckerings or folds (FO) in the presynaptic axonal membrane. Enlarged 25 000x. (de Robertis & Bennett, 1955). (E) Electron micrograph of catecholamine-containing granules in synaptic vesicles of sympathetic nerve terminals in the rat vas deferens. The distance between nerve and muscle membranes is about 25 nm (Richardson, 1962; 79). (F) Vesicle hypothesis as first enunciated (del Castillo and Katz, 1956). A diagram is shown of a nerve-muscle junction, with several features described after Robertson’s (1956) electron micrographs of the junction. N, nerve terminal; M, muscle fiber. In the lower part, an enlarged part of the nerve terminal is shown, containing ‘ACh-carrier corpuscles,’ as then described by del Castillo and Katz, or synaptic vesicles. Release of ACh is supposed to occur as a result of critical collisions between these synaptic vesicles and the membrane. This is indicated formally by labelling certain ‘critical spots’ on the surface of both.
identified with monophasic synaptic potentials, one in the depolarising direction (Fig. 2.4C) and the other in the hyperpolarising direction (Fig. 2.4D). Furthermore, these potentials had time courses similar to those predicted for the c.e.s. and the c.i.s. in the original work of Eccles and Sherrington in 1930 (compare Fig. 2.4C with Fig. 2.4A) and with that of Lloyd in 1946 (compare Fig. 2.4D with Fig. 2.4B(b)). On observing the inhibitory postsynaptic potential, Eccles concluded that:
EMERGENCE OF THE CONCEPT OF TRANSMITTER RELEASE AT PERIPHERAL AND CENTRAL SYNAPSES
39
…the potential change observed is directly opposite to that predicted by the Golgi-cell hypothesis, which is thereby falsified… It may therefore be concluded that inhibitory synaptic action is mediated by a specific transmitter substance that is liberated from the inhibitory synaptic knobs and causes an increase in polarization of the subjacent membrane of the motoneurone (Brock et al., 1952). So concluded the long saga that established the supremacy of chemical transmission at peripheral and central synapses. It enabled Eccles finally to show that Sherrington’s c.e.s. and the c.i.s. of motoneurons were due to chemical transmission (Fig. 2.10A). In hindsight, Keith Lucas and E.D.Adrian, the founding fathers of the biophysical approach to the study of impulse conduction, had inappropriately tried to transfer knowledge gained on conduction to the theoretical analysis of the mechanisms of transmission. Langley, Loewi, and Dale had exerted a counteracting influence that was founded on a tremendous amount of ingenious experimentation on transmission. However, it was the analysis of the endplate potential offered by the biophysicists A.V.Hill and Bernard Katz that finally clarified the whole matter. 2.8 Katz's concept of quantal transmitter release at the motor endplate and the vesicle hypothesis The discovery of spontaneous miniature endplate potentials in the frog sartorius muscle by Fatt and Katz revolutionised our understanding of neurotransmitter release (Fatt & Katz, 1952a) (Fig. 2.8A). Although the endplate potential in response to a nerve impulse had been identified in 1938, it was not possible to observe spontaneous endplate potentials without intracellular microelectrodes. The amplitude-frequency distribution of the spontaneous potentials was approximately Gaussian (Fig. 2.8B), although Fatt and Katz did note that: …there is indication of several discharges of about twice the mean amplitude, and of one isolated discharge of three or four times the mean size (Fatt & Katz, 1952a), They attributed this to the coincidence of two (or three) unitary discharges that could not be resolved given that detection was only possible down to 5 msec; the calculated chances of units occurring at such small intervals apart supported their conclusion. This question concerning the composition of the unitary discharges is still a matter of great interest. It was natural to consider if the endplate potential was composed of these spontaneous unitary discharges. Del Castillo and Katz showed that this was likely to be the case. They determined that the amplitude-frequency distribution of the endplate potential under conditions of low transmitter release could be built of units whose mean size and amplitude distribution were identical to those of the spontaneous unitary discharges (Fig. 2.8C). They concluded that: statistical analysis indicates that the end-plate potential is built up of small all-or-none quanta which are identical in size and shape with the spontaneous occurring miniature potentials (del Castillo & Katz, 1954). Del Castillo and Katz noted that the statistical analysis failed under conditions of reasonably high-evoked transmitter release, which they thought may occur because some synaptic units respond more readily than others. With the application of more refined electrophysiological techniques it is now known that there is indeed nonuniformity in the probability of secretion of quantal unit at different release sites within a nerve terminal. Attempts to both measure this non-uniformity and see if the Katz statistical paradigm for quantal secretion holds for different peripheral and central synapses constitute a major research effort at this time. The use of the microelectrode to study neurotransmitter release was complemented by the development of refined biochemical techniques for determining the constituents of nerve terminals, as well as by introduction of the electronmicroscope. The major biochemical contributions came from the Karolinska Institute in Stockholm, where U.S. von Euler, who had trained with H.H.Dale in London in 1934, first showed definitively in 1946 that the catecholamine noradrenaline was a transmitter at sympathetic nerve terminals (von Euler, 1946). In a letter to Dale in 1945, von Fuler commented that: perhaps it will interest you to hear about the sympathetomimetic substance in the spleen which I have been working with lately. It appeared that ordinary alcoholic extracts of cattle spleen contain the somewhat surprising amount of some 10 mg adrenaline pressor equivalent per kg. After perfusion the active substance was found to differ somewhat from adrenaline, and, on the basis of your admirable analysis with Berger in 1910 of the action of sympathomimetic amines, it emerged that it resembled definitely more an amino-base like nor-adrenaline than adrenaline or methylated compounds (see Blaschko, 1985).
40
HISTORY OF THE SYNAPSE
Fig. 2.9. The calcium dependence of neurotransmitter release. (A) Evidence that calcium is required for the release of acetylcholine (ACh) at sympathetic preganglionic nerve terminals. Both contraction of the cat’s nictitating membrane (above) and the cat’s blood pressure (below) were used to assay acetylcholine in the venous effluent collected during perfusion of the superior cervical ganglion during the following corresponding periods. A–C, Perfusion with normal Locke’s solution containing eserine: A, no stimulation; B, maximal preganglionic stimulation, 10 per sec; C, at the arrow, injection of 2 mg KCI. D–F, perfusion with Ca-free Locke’s solution containing eserine: D, no stimulation; E, 10 min later, maximal preganglionic stimulation, 10 per sec, producing no further contraction of the nictitating membrane; F, at the arrow, injection of 2 mg KCL G, time signal, and effect of 0.005 µg of Ach (Harvey & Macintosh, 1940). (B) Relationship between calcium concentration and amplitude of endplate potential in the frog sartorius muscle. Each symbol gives the results for a different magnesium concentration (open circles, 0.5 mm; crosses, 2.0 mM; filled circles, 4.0 mM). The coordinates are logarithmic, giving straight lines with a slope of approximately 4 (Dodge & Rahamimoff, 1967). (C) Suppression of transmitter release during a large ‘positive voltage step’ of the presynaptic membrane potential of the giant synapse in the stellate ganglion of the squid, treated with tetrodotoxin to block nerve impulses. The presynaptic terminal is loaded with tetraethylammonium ions. Blocks 1 to 8 show increasing pulse intensity. In each block of records, the bottom trace shows the presynaptic voltage step, the middle trace shows postsynaptic response, and the top trace monitors the current pulse. There is a progressive suppression of ‘on’-response and replacement by ‘off’ response, as presynaptic voltage is increased from 100 to 200 mV (records 4 to 8), indicating that, if the movement of calcium ions into the terminal is blocked by depolarization to 200 mV, then transmitter release fails to occur until after the depolarization is removed (Katz & Miledi, 1967b).
In 1953, Hillarp, Lagerstedt, and Nilsson in Lund, Sweden (Hillarp et al., 1953), as well as Blaschko and Welch in Oxford, showed that catecholamines such as noradrenaline were stored in granules within the adrenal medulla. It was not long before von Euler together with Hillarp (von Euler & Hillarp, 1956) showed that noradrenaline was also stored in the particulate fraction of sympathetic nerves. Noradrenaline was therefore likely to be stored in granules within sympathetic nerve terminals.
EMERGENCE OF THE CONCEPT OF TRANSMITTER RELEASE AT PERIPHERAL AND CENTRAL SYNAPSES
41
This concept of the packaging of neurotransmitters in particles or granules within nerve terminals was greatly enhanced by the first electron microscope images of the terminals by Palade (1954) as well as de Robertis and Bennett (1954) (Fig. 2.8D). De Robertis and Bennett commented in 1955 that: A granular or vesicular component, here designated the synaptic vesicles, is encountered on the presynaptic side of the synapse and consists of numerous oval or spherical bodies 20 to 50 nm in diameter, with dense circumferences and lighter centers. Synaptic vesicles are encountered in close relationship to the synaptic membranes (de Robertis & Bennett, 1955). At sympathetic nerve terminals these synaptic vesicles were seen to actually contain the granules of catecholamines (Fig. 2.8E), as predicted by von Euler and Hillarp. By 1956 the quantal unit of transmitter release had been discovered at the motor endplate, along with the storage of catecholamines in granules within sympathetic nerves, and synaptic vesicles within both central and peripheral nerve terminals. That year del Castillo and Katz enunciated the vesicle hypothesis, attributing quantization of transmitter release to its association with synaptic vesicles (Fig. 2.8F). In their own words: Recent electron microscope studies (Robertson, 1956) have shown that the motor nerve terminals contain a fairly dense population of microsomes, granules or vesicles, of less than 0.1 µ diameter, which may well be the intracellular corpuscles to which ACh is attached. It has been known since Loewi’s investigations that most of the ACh which is present in ‘homogenised’ nerve tissue can only be extracted into an aqueous solution after chemical destruction of the cell proteins. It appears then that the discharge of ACh from a nerve terminal requires the disruption of more than one diffusion barrier: first the release from its intracellular attachment, and secondly a passage through a nerve membrane. One might suppose that when a ‘critical’ collision occurs between an infracellular ACh-carrier and the membrane of the nerve terminal, the two barriers are opened simultaneously and the ACh-contents of the carrier particle are suddenly discharged. This picture, though purely speculative, is nevertheless in accord with recent experimental findings; it takes account of the evidence discussed below that the release of ACh from nerve terminals occurs in multi-molecular units or ‘quanta’ and of the evidence, already cited, for the bound state of intracellular ACh content (del Castillo & Katz, 1956). The mechanism by which the ‘contents of the carrier particle are suddenly discharged’ is perhaps the major focus of research on neurotransmitter release at the present time. Katz next tackled the problem of determining the necessary and sufficient conditions for the ACh-contents of the carrier particle to be suddenly discharged. It had been known since the work of Locke reported in 1894 that calcium was necessary for transmission at the neuromuscular junction (Locke, 1894). The reason for this, as Harvey and Macintosh showed in 1940, was that the release of acetylcholine at nerve terminals required calcium (Harvey & Macintosh, 1940).This transmitter was only released in the perfused superior cervical ganglion of the cat upon stimulating the preganglionic nerves in the presence of calcium (Fig. 2.9A). Kuffler and Eccles had shown by 1942 that the amplitude of the endplate potential was affected by the calcium concentration in accord with the observations of Harvey and Macintosh on the calcium-dependence of acetylcholine release. However, it was not until 1967 that Katz’s laboratory produced two most important observations on how calcium might govern transmitter release. The first was due to Dodge and Rahamimoff, who showed that the endplate potential in low concentrations of calcium ions increased as the fourth power of the calcium concentration (Dodge & Rahamimoff, 1967) (Fig. 2.9B). This observation gave rise to the idea that the cooperative action of about four calcium ions is necessary for release of each quantal packet of transmitter by the nerve impulse. The second observation was due to Katz and Miledi, who determined that the site of this cooperative action of calcium ions was on the inside of the nerve terminal rather than on the outside (Katz & Miledi, 1967b) (Fig. 2.9C). The basis of this cooperative action of calcium ions on the inside of the nerve terminal membrane to release the contents of synaptic vesicles is a theme of much current research. 2.9 Conclusion: the establishment of Sherrington's concept of the synapse in the central nervous system and central synaptic transmission While Katz was researching the mechanism of neurotransmitter release, Eccles and his colleagues in Canberra explored synaptic mechanisms at successively higher levels of the central nervous system. Studies were carried out on inhibition of Purkinje cells in the cerebellum (Eccles et al., 1966) (Fig. 2.10D), on thalamic-cortical relay cells (Eccles, 1969) (Fig. 2.10B) and on the CA3 pyramidal cells in the hippocampus (Andersen et al., 1963) (Fig. 2.10C). This work was revolutionary in as much as it supplied a functional microanatomy of the synaptic connections to be found in these different nerve centers. In addition, it
42
HISTORY OF THE SYNAPSE
Fig. 2.10. Inhibitory pathways in the central nervous system elucidated by Eccles (1969) and colleagues. (A) Synaptic connection involved in the inhibitory action produced in extensor motoneurons by afferent volleys in the Group 1b afferent fibers from the Golgi tendon organs in extensor muscles (Eccles, 1969). (B) Pathway to the sensory-motor cortex for cutaneous fibres from the forelimb; inhibitory neurons are shown in black in both the cuneate nucleus and ventrobasal nucleus of the thalamus. Note that the inhibitory pathway is of the feed-forward type in the cuneate nucleus and feed-back type in the thalamus (Eccles, 1969). (C) Results of recording from a CA3 hippocampal pyramidal cell in response to commissural (Com), septal (Sept), and local (Loc) stimulation, a, responses recorded by a microelectrode penetrating CA3 following local stimulation, b, graph in which the size of the positive waves of the responses to commissural, septal, and local stimulation is plotted against depth, with the positivities measured at a time indicated by the stippled line in (a) c, a CA3 pyramidal cell, semi-diagrammatically drawn to scale to facilitate comparison with (b). The arrows indicate the extracellular flow of current generated by the inhibitory postsynaptic potential (from Andersen et al., 1963). (D) Perspective drawing of a cerebellar folium to show the synaptic connections of the inhibitory interneurons. The cerebellar cortex is seen to be divided into three layers: molecular layer (ML) Purkinie cell layer (PL), and granular layer (GL). The input to the cortex is by two types of fibre: mossy fiber (MF) and climbing fiber (CF). Single examples are shown of four types of interneurons: granule cells (GrC), Golgi cells (GoC), basket cells, and outer stellate cells (SC). Also shown are two Purkinje cells, one (PC) with its dendritic ramifications, and both axons (PA), one with two collaterals (PAC) ending on the Golgi cell and the basket cell. The mossy fibre shown with numerous branches and thickenings at the sites of its synapses on granule cell dendrites, so forming the glomeruli (Glo). Collaterals of the climbing fiber (CF) are shown making synapses on the Golgi cell and basket cell. The axons of the granule cells bifurcate to give rise to the parallel fibers (PF) in the molecular layer. Arrows show directions of normal propagation in: the mossy fiber, climbing fiber, and its collaterals; the Purkinje axons and collaterals; and the axons of the interneurons BC, SC, and GoC. (Eccles et al., 1966).
provided insights into the ionic basis of inhibition at different synapses and the role this inhibition plays in controlling the excitability of the major neuron types. In this way the research program initiated by Sherrington’s concepts of inhibition and excitation in 1906 was brought to fruition.
3 The Discovery of Acetylcholine and the Concept of Receptors at Synapses
3.1 Introduction The idea of the ‘receptive substance’ or receptors as we now call them, was developed by John Langley of Cambridge 90 years ago (Fig. 3.1 A). Between 1901 and 1905 Langley laid the foundations for the idea of chemical transmission with his student Thomas Elliott (Fig 3 1B) through their investigations on sympathetic neuroeffector transmission. In an extraordinary act of creative ability, Langley then carried out a series of
Fig. 3.1. The founding fathers of chemical transmission at synapses. (A) J.N.Langley (1852–1925), Fig 20.4 in Finger (1994). (B) T.R.Elliott (1877–1961), portrait facing p.53 in Dale (1961) (C) H.H.Dale (1875–1968), portrait facing p.77 in Feldberg (1970). (D) O.Loewi (1873–1961), Fig. 20.5 in Finger (1994).
44
HISTORY OF THE SYNAPSE
Table 1. Chronological table of significant events in the history of receptors 1844 1866 1899 1901 1901 1904 1905 1905 1906 1906 1914 1921 1926 1929 1934 1936 1936 1970 1976
Curare paralyses rabbits without affecting the heart Curare acts on an intermediate zone between nerve and muscle Supra-renal extract (adrenaline) contracts and relaxes different smooth muscles Supra-renal extract (adrenaline) contracts or relaxes different smooth muscles as does stimulation of their sympathetic nerve supply Nicotine stimulates sympathetic ganglion cells directly Adrenaline acts at the junction between nerves and smooth muscle cells not on nerve terminals Nicotine stimulates skeletal muscles directly and this is blocked by curare The concept of a ‘receptive substance’ on skeletal muscles first described The ‘receptive substance’ shown to provide the receptor for alkaloids such as nicotine and curare Acetylcholine synthesised and shown to have powerful effects on the circulation Acetylcholine has similar actions on smooth muscles and cardiac muscle as stimulating the vagus nerve ‘Vagusstoff’ is released by the vagus nerve and controls the heart beat Physostigmine potentiates the effects of applied acetylcholine on the heart Acetylcholine shown to be a natural constituent of horse and ox spleen; likely to be ‘Vagusstoff’ Acetylcholine is released in autonomic ganglia on nerve stimulation; likely to be the transmitter Acetylcholine collected in venous fluid from skeletal muscles on nerve stimulation Acetylcholine injected into the arteries of skeletal muscles initiates contraction Acetylcholine applied at the endplate gives membrane noise due to the opening of channels Acetylcholine receptor channels give electrical signal that may be recorded directly
Bernard Vulpian, 1866 Lewandowsky, 1899 Langley, 1901 Langley, 1901 Elliott, 1904 Langley, 1905 Langley, 1905 Langley, 1906 Hunt and Taveau, 1906 Dale, 1914a Loewi, 1921 Loewi and Navratil, 1926b Dale and Dudley, 1929 Feldberg and Gaddum, 1934 Dale, Feldberg and Vogt, 1936 Brown, Dale and Feldberg, 1936 Katz and Miledi, 1970b Neher and Sakmann, 1976
investigations between 1905 and 1907 on the somatic neuromuscular junction that established the idea of transmitter receptors. This historical review traces the development of Langley’s ideas over this period, especially in relation to the concept of the ‘chemoreceptor’ developed by Paul Ehrlich. The review then examines how this work was applied by a number of investigators to place the concept of transmitter substances and their receptors on a firm foundation for the modern molecular approaches to the delineation of receptor types and their function. In order to assist the reader, a chronological table of significant experiments in the history of receptors is provided (Table 1), together with a list of the major contributors to these experiments (Table 2) and the agents they used to delineate the receptor concept (Table 3). 3.2 Claude Bernard and curarization: the notion of an intermediate zone between nerve and muscle In June of 1844 Claude Bernard wrote in his experimental note book that: A poisoned arrow obtained from a friend who had connections with South American natives was thrust into the subcutaneous tissue of a rabbit at the internal part of the thigh and maintained there for 30 seconds. The animal was then observed. At first, nothing happened. But after six minutes it became totally paralysed: no reflex movements were observed on pinching the rabbit, although the heart continued to beat. The animal subsequently died and at autopsy it was not possible to find any lesion capable of explaining paralysis and death (Fessard, 1967). Table 2. Scientists who contributed significantly to the idea of Name
Location
Dates of research
C.Bernard J.Langley T.Elliott H.Dale O.Loewi W.Feldberg G.Brown
Paris Cambridge Cambridge London Austria London London
1844–1883 1874–1908 1904–1905 1914–1936 1921–1926 1934–1936 1936–1937
Mentor M.Forster J.Langley
H.Dale H.Dale
THE DISCOVERY OF ACETYLCHOLINE AND THE CONCEPT OF RECEPTORS AT SYNAPSES
45
Table 3. Definition of some of the agents used to delineate the receptor concept Agent
Site of action
Muscle type
Curare Pilocarpine Supra-renal extract Adrenaline Nicotine Acetylcholine Vagusstoff Eserine Acetylcholine
Nicotinic receptors Muscarinic receptors
Skeletal Smooth and cardiac Smooth and cardiac Smooth and cardiac Skeletal Smooth and cardiac Cardiac
Nicotinic receptors Muscarinic receptors Muscarinic receptors Cholinesterase Nicotinic receptors
Skeletal muscle
Antagonist Atropine
Curare Atropine
Curare
Although this observation had been made in 1811 by Brodie, who later went on to show that curarized animals could be maintained alive on an artificial respirator, the advent of Bernard into this area of research brought a keen experimental mind to bear on the problem of the action of curare. Bernard constructed a galvanic stimulator for exciting either nerves or muscle fibres which allowed him to carry out investigations on curare that led him to report that: Electrical stimulation of a motor nerve in a curarized frog has no effect, whereas its muscles contract when directly stimulated. This pointed to curare acting on nerves rather than generally acting as some kind of anaesthetic. In order to test whether both the motor and sensory nerves were affected by curare, Bernard designed an experiment in which a ligature was passed around the waist of a frog, so that the lower limbs were isolated from the rest of the body, except for the sciatic nerve, as shown in Fig. 3.2A. Bernard then reported the following observations: Curare is introduced under the skin of the back. It poisons the anterior part of the body and prevents movement there; but sensation in this part is conserved, for stimuli applied to this paralysed portion cause energetic reflex movements in the isolated posterior half. Curare is thus a poison which not only produces physiological separation of nerves and muscles, but also separation of two major kinds of nervous manifestations. It suppresses movement but has no action on sensation; so that in a way it dissects out the neuromotor system and separates it from the muscular system, the sensory nervous system, and other tissues. He then designed an experiment that is illustrated in Fig. 3.2B: here electrical stimulation was applied to the sciatic nerve lying in a bath of curare, as shown in V, and contraction of the muscle outside the curare bath was present. On the other hand, when the muscle was placed in the curare bath as shown in V, stimulation of the nerve outside the bath did not give rise to contraction. The obvious conclusion to this experiment would seem to be that some junctional structure between the nerve and the muscle had been affected by curare. However Bernard did not reach this conclusion. Fessard (1967) has conjectured as to why Bernard did not follow the appropriate deduction from his observations. Perhaps Bernard was concerned that he was dealing with organs that were separated from the body and not subjected to the ‘milieu interieur’ and circulation of the blood? It seems that Bernard was persuaded of the idea that the action of curare should be related to the circulation of the blood, perhaps as Bernard himself suggested through an alteration of the gas exchange between the blood and the air in the lungs or the tissues of the capillaries. Bernard then turned to the experiment illustrated in Fig. 3.2C. The technique involves a kind of close arterial injection, as illustrated by the insert. Curare is injected into the artery supplying a muscle, so that it does not come into contact with the nerve trunk; furthermore there is an outlet in the vein which prevents the curare containing blood from reaching the central nervous system. This beautiful experiment seemed to Bernard to show that curare acted on the nerve terminals within the muscle. There is no mention in his books of the notion that curare might work at a junction formed between the nerve and the muscle although in his notebooks there is mention that ‘curare must act on the terminal plates of motor nerves’ and that ‘Curare does no more than interrupt something motor which puts the nerve and the muscle into electrical relationship for movement’ (Fessard, 1967). These quotes suggest that he had envisaged the notion of a neuromuscular junction, although this was never pursued in his formal statements to be found in his books concerning these experiments. The explicit claim that curare does not act on motor-nerve terminals, but rather on some intermediate zone between nerve and muscle was left to Vulpian (1866) in his Lecons sur la Physiologie Generale et Comparee du Systeme Nerveux. The nature of this intermediate zone was next investigated by histologists, seeking to find the site at which curare works. Chief amongst these at this time
46
HISTORY OF THE SYNAPSE
was Bernard’s student Kuhne, however it is now clear in retrospect that the histological approach to this problem had to await the advent of ultrastructural techniques, nearly a century later. As to the proper development of the functional approach, that might have been pursued immediately. Although there were no technical limitations to such an approach, another forty years had to pass before the problem was elucidated by Langley. 3.3 Paul Ehrlich and the idea of the `receptive side chains' of cells In 1885 Ehrlich presented his thesis to the University of Leipzig in which he described for the first time his ‘side chain theory’ of cellular action. The protoplasm of a cell was considered to be a giant molecule incorporating a central structure responsible for the specific activity of a particular cell type (such as a muscle cell or a neurone), which possessed chemical side chains. The side chains were envisaged as carrying out processes common to all cells. For example one such side chain might be involved in the process of oxidation, following which the chain had to be regenerated by the cell. Two years later, in 1897, Ehrlich elaborated this idea into his influential side chain theory of immunity. He postulated that a ‘receptive side chain’ of a particular cell, for example one involved in nutrition, has an atom group which by mere coincidence possessed specific combining properties for a particular toxin, such as tetanus toxin. The normal function of the side chain is lost once the toxin binds to the group, triggering the cell to produce a large number of such side chains. Many of these excess side chains then break off from the cell and are so released into the blood stream. Here they act as antibodies or antitoxins, combining with the toxin in the blood stream and so preventing it from combining with cells. Ehrlich in this work likened the relation between toxin and receptive side chain, which by 1900 he referred to simply as ‘receptor’, to that between a ‘lock and a key’. In his Croonian Lecture to the Royal Society of London in 1900, Ehrlich specifically excluded his receptor theory for the actions of toxins as being applicable to the action of drugs on cells. He came to the conclusion that drugs are not bound firmly to cells like toxins as most of the former are easily extracted from tissues by solvents. Thus toxins are bound to the protoplasmic molecule by chemical union whereas pharmacological drugs are not as they do not possess appropriate groups. It follows that they are not capable of eliciting the production of antibodies. If alkaloids, aromatic amines, antipyretics, or aniline dyes be introduced into the animal body it is a very easy matter, by means of water, alcohol, or acetone, according to the nature of the substance, to remove all these things quickly and easily from the tissues… We are therefore obliged to conclude that none of the foreign bodies just mentioned enter synthetically into the cell complex; but are merely contained in the cells in their free state. The combinations into which they enter with the cells, and notably with the not really living parts of them are very unstable, and usually correspond only to the conditions in solid solutions, while in other cases only a feeble salt-like formation takes place. The conclusion reached by Ehrlich then in 1900, and reiterated in 1902, was that pharmacological substances do not possess the necessary atomic groups which would allow them to combine with the appropriate groups of the cell protoplasm (Ehrlich & Morgenroth, 1900). The ‘lock and key’ concept did not then apply to the interaction of drugs with cells, so that the ‘receptor’ concept did not apply in this instance. However by 1907 Ehrlich had completely changed his mind on this issue, even introduced the word ‘chemoreceptor’ to describe the interaction of drugs with cells. What had happened in the five years between 1902 and 1907 to change his mind on this issue was largely due to the work during this period of the laboratory of Langley, which will now be described. 3.4 John Langley and T.R.Elliott: the emergence of the concept of chemical transmission between sympathetic nerves and smooth muscle In 1899 Lewandowsky observed that supra-renal extract causes in cats dilation of the pupil, withdrawal of the nictitating membrane (Fig. 3.3 A), separation of the eyelids and protrusion of the eyeball. Lewandowsky suggested that the extract acted directly on the smooth muscle and not on the nerve endings in the muscle as he obtained the same results with the extract after excision of the superior cervical ganglion and degeneration of the postganglionic nerves as in the normal animal. This was an extraordinary insightful interpretation, which formed the basis for the subsequent comprehensive study of the effects of supra-renal extract by Langley. In 1901 he inquired into the effects produced by supra-renal extract in the cat and rabbit on different organs, and arranged them in order as regards the amount of extract required per body weight to produce an obvious effect, as shown in the Table of Fig. 3.3B. This table shows that the extract in some cases contracts smooth muscles of a particular organ and in other cases relaxes the muscle. Langley had already, in 1898, defined the autonomic nervous system which he divided into sympathetic, cranial, sacral and enteric components. In his 1901 paper he makes the historic remarks:
THE DISCOVERY OF ACETYLCHOLINE AND THE CONCEPT OF RECEPTORS AT SYNAPSES
47
Fig. 3.2. The experiments of Claude Bernard and John Langley establishing the concept of `receptive substances' at somatic neuromuscular junctions. (A) A frog preparation illustrating an experiment by Claude Bernard. A ligature passed around the waist of the frog isolated the lower-limbs from the rest of the body, except for the sciatic nerve trunk. The experiment is described in the text. (Reproduced from Lecons 1883 edition, Fig. 26, p. 345; from Fig. 26 in Bernard, 1883; reproduced by Fessard as Fig. 2, p.111 in Grande & Visscher, 1967). (B) Illustration of an experiment with curare carried out by Claude Bernard (reproduced from Lecons 1883 ed., Fig. 23, p. 329; from Fig. 23 in Bernard, 1883; reproduced by Fessard as Fig. 3, p.112 in Grande & Visscher, 1967). (C) A drawing made by Claude Bernard to illustrate one of his experiments on neuromuscular curarization (see text). Reproduced from the original Note-book ‘Cahier rouge’ (also recently reproduced in Cahier de Notes, 1965, p. 76). Inset: an explanatory scheme of the drawing that is described in the text. (From pg. 76 in ‘Cahier de Notes’, 1965, Bernard; reproduced by Fessard as Fig. 4, p. 113 in Grande & Visscher, 1967). (D) Frog killed by destroying the whole of the central nervous system. Contraction of the muscles of the forelimbs caused by nicotine. (From Fig. 7 in Langley, 1906). (E) Nicotine injected into the abdominal cavity of a frog, whose spinal cord and brain had been destroyed. For details of the experiment see the text. (From Fig. 9 in Langley, 1906). (F) Fowl, anaesthetised with morphia and A.C.E. mixture, balanced on its thorax in a V-shaped piece of wood. The neck and legs hang down and are flaccid, the eyes are shut. (From Fig. 1 in Langley, 1906). (G) The same fowl as in F. Two minutes after injection of 5 mg of nicotine into the jugular vein. The injection caused a gradual and fairly quick extension of the legs, retraction and twisting of the neck, and opening of the eyes. In order to show the eyes, the beak was held when taking the photograph. The fowl was unfastened throughout, and the injection caused no general movement nor any decrease of the anaesthesia. (From Fig. 2 in Langley, 1906).
It is a noteworthy fact that the effect of supra-renal extract in no case corresponds to that which is produced by stimulation in normal conditions of a cranial autonomic or of a sacral autonomic nerve. It does not produce the effect of stimulating the third nerve on the eye, nor of the vagus on the stomach or the heart, nor the effect of stimulating the
48
HISTORY OF THE SYNAPSE
pelvic nerve on the bladder, the rectum, the anus, or the generative organs. It is true that it causes a free secretion of saliva, but the secretion is not accompanied in its first stages by increased vascularity such as is caused by stimulation of the chorda tympani of Jacobson’s nerve. It is equally noteworthy that the effects produced by suprarenal extract are almost all such as are produced by stimulation of some one or other sympathetic nerve. In many cases the effects produced by the extract and by electrical stimulation of the sympathetic nerve correspond exactly. Having made these observations, and the fact that the effects of the extract persist after denervation of the organs, as Lewandowsky had first observed, Langley reached the conclusion that: …the difference in action on different autonomic tissues must depend upon their intrinsic differences. However, at this time Langley did not comment on the possiblity that the sympathetic nerves exerted their effects by the release of a substance equivalent to suprarenal extract. In 1904, Langley’s student Elliott reported experiments that showed even further the parallel effects of sympathetic nerve stimulation and of supra-renal extract (now identified as adrenaline by Takamine, 1901) on autonomic effectors. He showed that stimulation of the sympathetic nerves causes the sphincter at the junction between the small and large intestine to contract at the same time inhibiting the circular muscle in the wall of the ileum and colon adjoining the sphincter. Adrenaline produced the same effect as sympathetic nerve stimulation, thus contracting the sphincter (Fig. 3.3C) and relaxing the circular muscle of the surrounding ileum. These observations emphasised the parallel actions of adrenaline and of sympathetic nerve stimulation on the smooth muscle of different organs, and in this case within an organ. In that same year Elliott carried out an extensive study of the parallel actions of sympathetic nerve stimulation to the smooth muscles of different organs and that of the action of adrenaline on these, such as inhibition of the stomach by the splanchnic nerves (Fig. 3.3D and 3.3E), examining all apparent exceptions to this rule by previous investigators (not including Langley) and came to the conclusion, published in 1905, that: In all vertebrates the reaction of any plain muscle to adrenalin is of similar character to that following excitation of the sympathetic (thoracico-lumbar) visceral nerves supplying that muscle. The change may be either to contraction or relaxation. In default of sympathetic innervation plain muscle is indifferent to adrenalin. A positive reaction to adrenalin is a trustworthy proof of the existence and nature of sympathetic nerves in any organ. Sympathetic nerve cells with their fibres, and the contractile muscle fibres are not irritated by adrenalin. Elliott was therefore led to conclude that since some plain muscles that do not receive a sympathetic innervation are not affected by adrenaline, then the contractile apparatus cannot be the site of action of this substance. Furthermore, as Lewandowsky, Langley and Elliott himself had shown that the actions of adrenaline were not dependent on an intact sympathetic nerve supply, then it was concluded that adrenaline did not exert its effects through the nerve supply. Elliott was then led to the important conclusion that: The stimulation takes place at the junction of muscle and nerve (Fig. 3.3F). The irritable substance at the myoneural junction depends for continuance of life on the nucleoplasm of the muscle cell, not of the nerve cell. However nowhere in this classic paper of 1905 is there any mention that stimulation of the muscle by the nerve involves the release of a chemical substance, let alone that this substance in the case of sympathetic nerves is adrenaline. Yet in a proceedings note to the Physiological Society of 1904 Elliott makes his claim for chemical transmission at sympathetic nerve terminals and that this might be adrenaline. In that famous note he presents the evidence in favour of his two hypotheses as follows (Elliott, 1904a): 1. the effect of adrenalin upon plain muscle is the same as the effect of exciting the sympathetic nerves supplying that particular tissue. 2. [the] medulla and the sympathetic ganglia have a common parentage’, (see Kohn, 1903a, b). 3. …the facts suggest that the sympathetic axons cannot excite the peripheral tissues except in the presence and perhaps through the agency, of the adrenalin or its immediate precursor secreted by the sympathetic paraganglia. 4. Adrenalin does not excite sympathetic ganglia when applied to them directly, as does nicotine. Its effective action is localised to the periphery. 5. …even after such complete denervation, whether of three days’ or ten months’ duration, the plain muscle of the dilatator pupillae will respond to adrenalin.
THE DISCOVERY OF ACETYLCHOLINE AND THE CONCEPT OF RECEPTORS AT SYNAPSES
Fig. 3.3. Elliott and Langley establish the concept of adrenaline as a transmitter at the autonomic neuromuscular junction. (A) Blutdruck und Membrana nictitans. Injection of 1ccm of extract of the adrenal bodies. (From Fig. 2 in Lewandowsky, 1899). (B) The effects produced by supra-renal extract in the cat and rabbit may be arranged roughly in the order shown as regards the amount of extract required per body weight to produce an obvious effect. (From the Table in Langley, 1901). (C) Cat. Vagi cut. Injection of 0.3 mgm. adrenalin into external jugular vein. A is the record of the ileo-colic sphincter under pressure of 15 cm. B gives the period of injection. The figures 12, 14, 16 indicate the blood-pressure in cms, given in the upper trace. Bottom trace gives time marker in seconds (not detectable in the original figure). (From Fig. 6 in Elliott, 1904b).
In summary then: Therefore it cannot be that adrenalin excites any structure derived from, and dependent for its persistence on, the peripheral neurone. But since adrenalin does not evoke any reaction from muscle that has at no time of its life been innervated by the sympathetic (for example the absence of action on the muscle of the bronchioles and of the pulmonary blood vessels, as shown by Brodie and Dixon, 1904), the point at which the stimulus of the chemical excitant is received, and transformed into what may cause the change of tension of the muscle fibre, is perhaps a mechanism developed out of the muscle cell in response to its union with the synapsing sympathetic fibre, the function of which is
49
50
HISTORY OF THE SYNAPSE
Fig. 3.3. (D) Inhibition of cat’s stomach by splanchnic nerves. Ether. Vagi and splanchnic cut within thorax and placed on shielded electrodes. Artificial respiration. Record of volume change of stomach under constant pressure of 7 cm. water. Stimulation of splanchnics caused rise of blood pressure and relaxation of stomach by 30 c.c. Period of stimulation of the splanchnics is given by the second trace from the bottom. Bottom trace gives time marker in seconds (barely detectable in original figure). (From Fig. 6 in Elliott, 1905). (E) From same experiment as in D. Injection of .18 mgm. adrenalin completely relaxed the stomach (second trace). Tone did not return until the vagi were again stimulated. Period of application of adrenalin given by the second trace from the bottom. The uppermost trace gives the blood pressure. The numbers on the traces are almost illegible: 210 mm Hg is indicated from above a base of 70 mm Hg. Bottom trace gives time marker in seconds (not clear in original record). (From Fig. 8 in Elliott, 1905). (F) Diagram of doubly innervated muscle-nerve system. (1) the sympathetic motor ganglion cell, (2) its axon, and (3) the nerve ending; (4) the myoneural junction; (5) the contractile muscle fibre. (1a) and (4a) the corresponding parts of the inhibitor mechanism. To simplify the diagram the motor myoneural junction (4) is represented as spatially separated from the inhibitor (4a). (From Fig. 10 in Elliott, 1905).
to receive and transform the nervous impulse. Adrenalin might then be the chemical stimulant liberated on each occasion when the impulse arrives at the periphery. We therefore have for the first time a succinct statement of the concept of chemical transmission but also identification of a transmitter substance. Although Langley undoubtedly supplied the intellectual environment in the laboratory for the development of these hypotheses, there is no sign in his papers to this time (1904) that he had joined together the set of
THE DISCOVERY OF ACETYLCHOLINE AND THE CONCEPT OF RECEPTORS AT SYNAPSES
51
numbered observations indicated above to arrive at the conclusion that chemical transmission is most likely to occur at sympathetic nerve terminals and that the transmitter is adrenaline. Langley was always loathe to speculate and develop hypotheses. His great experimental career consists of generating a formidable set of facts that lead inexorably to a conclusion. On no occasion did he draw a diagram of the kind shown in Fig. 3.3F from Elliott (1905) that so brilliantly concentrates one’s interest on the neuromuscular junction. Indeed this figure may be compared for its fruitful prescience with that of Sherrington’s figure of 1906 in the Integrative Action of the Nervous System, showing the monosynaptic connections of the motor and sensory nerves in the spinal cord. Langley’s reticence meant that he missed out generating the brilliantly fruitful hypotheses of Elliott. It may also be that Elliott himself was dissuaded by Langley from continuing down the path of elaborating his ideas further, as following the note of 1904 there is no mention by Elliott in the very substantial paper of 1905 of either the chemical transmission hypothesis or the possibility that adrenalin is a transmitter. 3.5 The action of curare and John Langley's development of the idea of transmitter receptors By 1904 it was clear that adrenalin acted on those smooth muscles that received a sympathetic innervation and that this action was independent of the nerve supply to the muscles. Elliott did not elaborate further on his concept of chemical transmission in his 1905 paper that there is a: mechanism developed out of the muscle cell in response to its union with the synapsing sympathetic fibre, the function of which is to receive and transform the nervous impulse, (Fig. 3.3F). However there were undoubtedly discussions in the Cambridge physiological laboratory concerning his hypotheses. This is made to some extent explicit by Langley in his first paper on the actions of curare on striated muscle in 1905 in which he says in the introduction: Elliott brings forward further and most striking evidence that adrenalin stimulates tissues which are stimulated by sympathetic nerves and these only. This leads him to look on adrenalin as acting on some substance common to sympathetic nerves. He finds, however, that degeneration of the nerves does not diminish the action of adrenalin, and as he considers that the axon endings degenerate, the substance affected by adrenalin must be in trophic connection with the muscle. This as I have pointed out above is, I think, the same as saying that it is part of the muscle. But in view of the close relation of adrenalin to sympathetic nerves, and because he considers it improbable that the varying action of adrenalin can be due to intrinsic differences in the muscle, he concludes that when sympathetic nerves unite with unstriated muscle they cause the formation in it of a new substance, the myo-neural junction, and it is this which is acted upon by adrenalin. Now supposing that nervous connection does cause in the muscle the formation of a new substance, this does not make the new substance any the less part of the muscle. The fundamental fact of Elliott’s view is then, I think, the same as mine, viz. that adrenalin acts directly on muscle. The concept of Elliott’s ‘new substance’ therefore had a major influence on how Langley designed his experiments concerning the manner by which curare acted. These were not only based on the conceptual framework of Elliott but also on Langley’s own discovery that nicotine stimulates sympathetic nerve cells by a ‘direct action upon them’ (Langley, 1901). Furthermore, it must not be forgotten that Langley’s first experiments in 1874, while still a student at Cambridge under the guidance of Michael Foster, involved an investigation into the actions of atropine and pilocarpine (muscarine like) on the secretion of saliva by the submaxillary gland. He found that these had opposite effects and in his full paper on this antagonism, published in 1878, there is the comment that: …we may, I think, without much rashness, assume that there is a substance or substances in the nerve endings or gland cells with which both atropine and pilocarpine are capable of forming compounds. On this assumption then the atropine or pilocarpine compounds are formed according to some law of which their relative mass and chemical affinity for the substance are factors. So Langley, some thirty years or more before the experiments of Elliott or for that matter of his own on curare, was already developing the idea of pharmacological agents forming compounds with the substances in cells. The concept of the receptor is clearly present in these early formulations, which are in contrast to those of Ehrlich in 1900 mentioned above. Furthermore, the observations upon which these conjectures were developed were made well before Ehrlich began his research in 1878. In 1905 Langley showed that injection of nicotine into the vein of an anaesthetised fowl led to gradual stiffening and extension of the hindlimbs over a couple of minutes due to tonic contraction of the red muscles (Figs. 3.2F and 3.2G). This
52
HISTORY OF THE SYNAPSE
effect still occurred after section of the sciatic and crural nerves, so that it did not involve the nerve supply. In order to provide quantitative details of this effect, Langley took measurements of the gastrocnemius muscle in the fowl following injection of nicotine into the vein, without interfering with the muscle’s blood supply and after cutting the sciatic and crural nerves. The results showed that the muscle contracted for several minutes (Fig. 3.4A upper). Langley then injected curare about one minute after the beginning of the nicotine-induced contraction: the muscle then relaxed (Fig. 3.4A lower). Repeating this experiment after cutting the sciatic and crural nerves several weeks previously so as to allow their peripheral extensions to fully degenerate, did not alter the results of injecting nicotine and curari, as shown in Fig. 3.4B. Langley (1905) commented on these experiments that: I conclude then that nicotine acts upon the muscle substance, and not upon the axon-endings. It has been shown above that curari acts upon the same substance as nicotine. It follows then that curari acts upon the muscle substance and not upon the axon-endings. Since, in the normal state, both nicotine and curari abolish the effect of nerve stimulation, but do not prevent contraction from being obtained by direct stimulation of the muscle or by a further adequate injection of nicotine, it may be inferred that neither the poisons nor the nervous impulse act directly on the contractile substance of the muscle but on some accessory substance. Since this accessory substance is the recipient of stimuli which it transfers to the contractile material, we may speak of it as the receptive substance of the muscle. This is the first occasion on which the phrase ‘receptive substance’ or as we now simply call it ‘receptor’ was used. In developing this concept, Langley (1905) makes his indebtedness to Elliott quite clear when he comments that: The subsequent work mentioned in the Introduction, and especially that of Elliott on the action of adrenalin, made the issues clearer He goes on to say that: In my view, the myo-neural junction is a part of the receptive substance localized in the neighbourhood of the axonending. Although Langley (1907) went on to give detailed descriptions of how nicotine acts to block transmission at the myo-neural junction, especially at high concentrations (Fig. 3.4E), and of the fact that the effects of nicotine are only found in those parts of individual muscles where nerve endings are found, his concept of the receptive substance was fully matured by the time he gave a Croonian lecture to the Royal Society of London in 1906. In that lecture he presented most elegant tracings of how contractions of the fowl’s gastrocnemius due to nicotine are blocked by curari (Fig. 3.4C). In addition he gave graphic demonstrations of the effects of nicotine when injected into the abdominal cavity of frogs whose brain and spinal cord were destroyed. These passed from a state in which the muscles are flaccid, so that when the limbs are raised they at once fall, to a condition in which there is maximum flexion of the forelimbs, as shown in Fig. 3.2D (see also Fig. 3.4D). The same experiment when performed on toads gives rise to a cataleptic condition in both the forelimbs and the hindlimbs. The contraction of the flexors and extensors of the arm are about equal so that there is little movement or no movement. The forelimbs can then be moved about almost as if made of lead, and stay with but slight return movement in any position in which they are placed consistent with the arrangement of the joints and ligaments (Fig. 3.2E). These cataleptic conditions are completely abolished by sufficient doses of curari. The conclusion is reached that: The mutual antagonism of nicotine and curari on muscle can only satisfactorily be explained by supposing that both combine with the same radicle of the muscle, so that nicotine-muscle compounds and curari-muscle compounds are formed.’ ‘Since neither curari nor nicotine, even in large doses, prevents direct stimulation of muscle from causing contraction, it is obvious that the muscle substance which combines with nicotine or curari is not identical with the substance which contracts. It is convenient to have a term for the specially excitable constituent, and I have called it the receptive substance. Langley (1906) concludes his lecture by drawing attention to the fact that if the set of experiments on muscle with nicotine and curari are carried out on the excitability of nerve cells in sympathetic ganglia, then analogous results are obtained (Langley, 1901). Thus: ‘I conclude that the substance affected by the poisons is a special receptive substance and not the fundamental substance of the cell.’ As to transmission between nerve endings and smooth muscle as well as glands, the arguments outlined above concerning the action of adrenalin that have been developed in particular by Elliott indicate that:
THE DISCOVERY OF ACETYLCHOLINE AND THE CONCEPT OF RECEPTORS AT SYNAPSES
53
Fig. 3.4. The experiments of John Langley that established the existence of `receptive substances' at the somatic neuromuscular junction. (A) Effects of curari and nicotine. Upper panel shows a contraction produced by a large dose of nicotine, applied for a time given by the second trace from the bottom. The bottom trace gives a time marker in 10s intervals. There is no calibration of the size of the contractions. Lower panel shows a similar contraction annulled by curari; the upper curve gives the blood pressure. (From Figs. 4 and 5 in Langley 1905). (B) Denervated muscle. Effect of nicotine and of curari after nicotine applied at times indicated in the trace second from the bottom. Time in sees, omitting every tenth is shown in the bottom trace, but is not clear in the original figure. There is no calibration of the size of the contraction. (From Fig. 7 in Langley 1905). (C) Abolition by curari of the contraction in the gastrocnemius muscle of the fowl caused by nicotine. The line second from the bottom gives the period of application of the drugs. The lowest line marks intervals of 10 seconds. No calibration is given of the size of the contraction. (From Fig. 4 in Langley 1906). (D) Frog. Brain and spinal cord destroyed. A thread was tied to the manus and connected with an unweighted lever, so that flexion of the arm caused a rise of the lever; 1 c.c. of a 1 per cent, nicotine injected into the abdominal cavity at the time shown by the signal in the bottom trace (the ‘1cc1pc nicotine’ is printed on the trace). Time marked in 10 seconds. No calibration is given of the size of the upper trace. (From Fig. 8 in Langley 1906). (E) Contraction of sartorius with nicotine compared with that due to direct stimulation (top trace). Effect of stimulating with make and break induction shocks before and after nicotine: 0’—Muscle in Ringer’s fluid (this is not shown in the original figure). 2’— Stimulate first with make and then with break shock, raise lever to base line. 7’—pour .005 per cent, nicotine into muscle chamber, beginning and end of pouring is marked. 11’—Stop drum, and run off nicotine. 11½’ and 12½’—Stimulate with make and break shocks. Bottom trace, time in 1s intervals. Period of application of make and then break shocks and of nicotine indicated in second bottom trace. No calibration is given of the size of the contraction in the upper trace. (From Fig.1 in Langley, 1907).
The legitimate statement from the premises is that it does not act on any muscle substance or on any nerve substance outside the limits of the myoneural junction. As regards the localisation of the receptive substance, strong evidence that this occurs to a considerable extent is afforded by the action both of adrenalin and of chrysotoxin on tissues which have a double nerve supply, but the evidence cannot be regarded as conclusive.
54
HISTORY OF THE SYNAPSE
It seems likely, despite some of Langley’s arguments to the contrary (viz. that Elliott adopted the theory that nerve and unstriated muscle are continuous), that Elliott had conceived of Langley’s receptive substance, but without giving it a name for he comments: But since adrenalin does not evoke any reaction from muscle that has at no time of its life been innervated by the sympathetic, the point at which the stimulus of the chemical excitant is received, and transformed into what may cause the change of tension of the muscle fibre, is perhaps a mechanism developed out of the muscle cell in response to its union with the synapsing sympathetic fibre, the function of which it is to receive and transform the nervous impulse. One cannot do better in this early part of the history of the receptor than to quote the conclusion of Langley’s great Croonian lecture: In the foregoing account we have seen reason to believe that in each of the three great types of connection of the peripheral end of an efferent nerve with a cell it is some constituent of the cell substance which is stimulated or paralysed by poisons ordinarily taken as stimulating or paralysing nerve endings. This theory adds to the complexity of the cell. It necessitates the presence in it of one or more substances (receptive substances) which are capable of receiving and transmitting stimuli, and capable of isolated paralysis, and also of a substance or substances concerned with the main function of the cell (contraction or secretion, or, in the case of nerve cells, of discharging nerve impulses). The Croonian lecture of 1906 on the antagonism between curare and nicotine giving rise to the concept of the transmitter receptor brings to a conclusion the research program embarked on by Langley some thirty years earlier while still a student at Cambridge working on the antagonism between atropine and pilocarpine. The antagonism of this set of drugs is seen as acting on the protoplasmic substance or substances in the muscle or effector organ, and does involve the combination of an alkaloid with protoplasm, in contradistinction to the suggestions of Ehrlich. Following these experiments of Langley, Ehrlich accepted the idea of the receptor for alkaloids such as nicotine and curare. 3.6 The Langley-Ehrlich receptor theory It is fascinating to trace the productive interaction of the ideas of Ehrlich and Langley over this period from 1878 to 1908, beginning as each did from quite different research programs. In the case of Ehrlich, his research was concerned with drug resistance as a consequence of studies on the chemotherapy of trypanosomes. The receptive side chain concept was developed in order to give a theoretical underpinning to his work on the chemotherapy of such micro-organisms, in particular in the use of substances that act in a manner that is largely irreversible, such as the arsenicals. For Langley, the starting point involved his research on the effects on muscle and nerve of alkaloids such as nicotine, curare, atropine, pilocarpine, strychnine and adrenaline, all of which produce an action that is relatively reversible compared with the actions studied by Ehrlich. Nevertheless by 1908 Langley could say: My theory of the action is in general on the lines of Ehrlich’s theory of immunity. I take it that the contractile molecule has a number of ‘receptive’ or side-chain radicles and the nicotine by combining with one of these causes contraction and by combining with another causes twitching… To which Ehrlich commented in 1914: For many reasons I had hesitated to apply these ideas about receptors to chemical substances in general, and in this connection it was, in particular, the brilliant investigations by Langley, on the effects of alkaloids, which caused my doubts to disappear and made the existence of chemo-receptors seem probable to me. Thus was conceived the tremendously fruitful concept of the receptor. 3.7 The discovery of acetylcholine and its physiological action at autonomic neuroeffector junctions In 1906 Hunt and Taveau synthesized acetylcholine and reported that:
THE DISCOVERY OF ACETYLCHOLINE AND THE CONCEPT OF RECEPTORS AT SYNAPSES
55
…as regards its effect upon the circulation, it is the most powerful substance known. They went on to say: we have not determined the cause of the fall of blood pressure from acetyl-cholin, but from the fact that it can be prevented entirely by atropine, I am inclined to think that it is due to an effect upon the terminations of the vagus in the heart. In the same year, Dixon (1906) gave a description of his experiments on the vagus inhibition of the heart, and stated that he was of the opinion that: the heart contains a substance—‘pro-inhibitin’, which, as a result of vagus excitation, is converted into a chemical body —‘inhibitin’. This substance, combining with the heart muscle, results in cardiac standstill. These suggestions were extraordinarily prescient of Loewi’s experiments on the identity of acetylcholine as the inhibitory transmitter released by the vagus in the heart subsequently carried out in the early 1920s (see below). In 1914, Ewins showed that extracts of certain specimens of the fungus ergot contained an active principle, which he identified as acetylcholine. Much of this work of Ewins was done at the instigation of Dale (Fig. 3.1C), as a consequence of his knowledge of the work of Hunt and Taveau (1906) who had already shown that synthesized acetylcholine had powerful depressor activity, indeed Dale reported to Elliott: We got that thing out of our silly ergot extract. It is acetylcholine and an most interesting substance. It is much more active than muscarine, though so easily hydrolysed that its action, when it is injected into the blood-stream, is remarkably evanescent, so that it can be given over and over again with exactly similar effects, like adrenaline. Here is a good candidate for the role of a hormone related to the rest of the autonomic nervous system, I am perilously near wild theorising… I shall be surprised, however, if this principle, once identified, does not turn up in all sorts of tissueextracts (see Letter, Dale to Elliott, 11 Dec, 1913, Contemporary Medical Archives Centre, Wellcome Institute, GC/42 ‘T.R.Elliott’; quoted in Tansey 1991). It was then of considerable interest to see if the depressor effect of acetylcholine was its principal effect. In 1914, Dale determined to carry out a systematic study of the effects of acetylcholine on various organs of the autonomic and somatic nervous systems (Dale, 1914a, b). In that work he showed that following the injection of acetylcholine into an animal, there were considerable responses elicited in a number of organs: cat’s blood pressure declined (Fig. 3.5A); there was complete cessation of the heart beat in frogs (Fig. 3.5B) and the rabbit small intestine contracted vigorously (Fig. 3.5C). Dale was alert to the fact that these effects are those produced by stimulation of the vagus nerve, and in a sense constitute a complementary set of effects to those observed by Langley and Elliott with respect to adrenalin and sympathetic nerve stimulation. He comments that: The question of a possible physiological significance, in the resemblance between the action of choline esters and the effects of certain divisions of the involuntary nervous system, is one of great interest, but one for the discussion of which little evidence is available. Acetylcholine is, of all the substances examined, the one whose action is most suggestive in this direction. The fact that its action surpasses even that of adrenine both in intensity and evanescence, when considered in conjunction with the fact that each of these two bases reproduces those effects of involuntary nerves which are absent from the action of the other, so that the two actions are in many directions at once complementary and antagonistic, gives plenty of scope for speculation. The main problem at this time in making a physiological claim for acetylcholine was: On the other hand, there is no known depot of choline derivatives, corresponding to the adrenine depot in the adrenal medulla, nor, indeed, any evidence that a substance resembling acetylcholine exists in the body at all. So Dale concludes in this paper which gives the first systematic account of the actions of acetylcholine in the peripheral nervous system that: Acetylcholine occurs occasionally in ergot, but its instability renders it improbable that its occurrence has any therapeutic significance’.
56
HISTORY OF THE SYNAPSE
Fig. 3.5. Henry Dale and Otto Loewi establish acetylcholine as a transmitter substance at autonomic neuromuscular junctions. (A) Cathether. Plethysmograph records from intestine and limb with calibration in c.c. Carotid blood-pressure shown with calibration in mmHg. Injection of 0.001 mg acetylcholine indicated by the second trace from the bottom. Bottom trace is time marker in 10s intervals. Third trace from the top is not specified (From Fig. 3 in Dale, 1914a). (B) Perfused heart of frog, recorded by suspension-lever. Tracings read from right to left, the vertical line in each case indicating change from pure Ringer’s solution to similar solution containing a choline-ester (the vertical line is about 7 beats from the right in each case), a Acetylcholine 1 in 100 millions, b Acetylcholine 1 in 200 millions, c Acetylcholine 1 in 500 millions, d Acetylcholine 1 in 1000 millions, e Nitroso-choline 1 in 100 000. 1 Nitroso-choline 1 in 1 million. Bottom trace is time in seconds. There is no calibration of the size of the contraction (From Fig. 11 in Dale, 1914a) (C) Loop of Rabbit’s small Intestine in 50 cc of Tyrode’s solution. At A 0.01 mg synthetic Acetyl-choline. At B 0.01 mg of Acetylcholine from ergot, added to the bath. At R, R, fresh Tyrode’s solution. Bottom trace is time marker in 10 s intervals. No calibration is given for the size of the contraction (From Fig. 13 in Dale, 1914a).
There the matter rested until after the first World War. In 1921 Loewi (Fig. 3.1D) reported from Austria his experiments on the heart indicating that chemical transmission occurred between the vagus nerve and the heart, mediated by a substance which he called ‘Vagusstoff’. Fig. 3.5D shows his original record, in which the heart beat in 1 is in normal Ringer, whereas the subsequent decline in the heart beat in 2 is due to the addition of Ringer that has been in contact with another heart whose vagus had been stimulated for 15 minutes; at 3 the heart beat returns to normal as a consequence of it being exposed to a Ringer that had been in contact with another heart for which
THE DISCOVERY OF ACETYLCHOLINE AND THE CONCEPT OF RECEPTORS AT SYNAPSES
57
Fig. 3.5. (D) Frog heart contractions. At 1. Ringer. 2. Ringer from a heart that received 15 min of vagal stimulation. 3. Ringer from a heart that did not receive stimulation. 4. Addition of 0.1 mg atropine. Bottom trace is not specified and there is no time or contraction calibration given. (From Fig. 1 in Loewi, 1921). (E) Frog heart contractions (a) The testing material is heart extract (1:20 dilution) plus acetylcholine (dilution 1:100,000). 2 is inactivated heart extract plus acetylcholine (dilution 1:1000). 3 is heart extract plus acetylcholine again, (b). The starting material is again heart extract (dilution 1:20) plus acetylcholine (dilution 1:100,000). 1 is heart extract inactivated by standing for 30 min plus acetylcholine (dilution 1: 750). 2 is heart extract inactivated by standing for 90 min plus acetylcholine (dilution 1:1000). 3 is heart extract inactivated by standing for 90 min plus acetylcholine (dilution 1:150). Bottom trace is not specified and no time or contraction calibrations are given. (From Fig, 5 in Loewi & Navratil, 1926a). (F) Frog heart contractions. 1=Ringer. 2=Vagusstoff (alcoholic extract from the heart). 3=Acetylcholine (dilution 1:100 million). Bottom trace not specified and no time or contraction calibrations given. (From Fig. 4 in Loewi & Navratil, 1926a). (G) Effect of stimulating the heart nerve on contraction of the frog’s heart in relation to the effects of applied acetylcholine and physostigmine given at the times indicated. In the left-hand panel, samples were taken from the heart at the time indicated by I and another taken at the time indicated by II (Inhaltsentnahme). In the right-hand panel the effects of applying these samples (I and II) to another heart are shown. Bottom trace is not specified and no time or contraction calibration is given. (From Fig. 3 in Loewi & Navratil, 1926b). (H) Effect of eserine on the frog’s heart. The caption reads: ‘Heart extract with Eserine without. Left standing for 2 hours.’ Bottom trace not specified and no calibrations given for the time of contraction. (From Fig. 5 in Loewi & Navratil, 1926b). (I) Frog heart, a=Acetylcholine (dilution 1 :million). I=Bovine blood plus acetylcholine. II=Heated bovine blood plus acetylcholine. 1=Inactivated bovine blood plus acetylcholine. 2=Heated bovine blood plus acetylcholine. (From Fig. 7 in Engelhart & Loewi, 1930). (J) Comparison of purified spleen extract (S.E.) and acetylcholine (A.C.) solution on the blood-pressure of a cat under ether. Calibration of the blood pressure is given in mm Hg. The amount of A.C. given is 0.01 mg. About 0.4 cc of the spleen extract gave the same result as that of 0.01 mg of A.C. (From Fig. 1 in Dale & Dudley; 1929)
the vagus was not stimulated; at 4 atropine was added to the normal Ringer and this increases the heart beat. By 1926 Loewi and Navratil had produced sufficient evidence to mount a persuasive case that Vagusstoff was acetylcholine: application of very low concentrations of acetylcholine to the heart greatly decreased the heart beat (Loewi & Navratil, 1926a; Fig. 3.5E and
58
HISTORY OF THE SYNAPSE
3.5F). These authors went on to show that inactivation of esterases in the heart for acetylcholine, using eserine (physostigmine; Fig. 3.5G, 3.5H and 3.5I; Loewi & Navratil, 1926b; Engelhart & Loewi, 1930) greatly potentiated the decrease in the heart beat brought about by the exogenous application of acetylcholine. This at last provided an explanation for the problem that Dale had referred to with regards to acetylcholine, that: when it is injected into the blood-stream, it is remarkably evanescent. Although it had been known for some time that physostigmine ‘sensitized’ the heart to stimulation of the vagus, this demonstration by Loewi and Navratil that physostigmine acted to potentiate the actions of exogenous acetylcholine led directly to the concept by these authors that the heart possesses an endogenous esterase for acetylcholine, for which they later coined the term ‘cholinesterase’. Dale continued in the late 1920s to try and find a natural source of acetylcholine in the body like that provided for adrenaline in the adrenal medulla. He commented to a friend in 1929: We are still struggling with the acetylcholine problem, which I mentioned to you when I saw you in the autumn. I am more and more convinced that the thing is there to be found, if only we can overcome the technical difficulties’ (Letter, Dale to Richards, 22 Mar 1929, Archives of the National Institute for Medical Research, File 647; quoted in Tansey, 1991). The breakthrough occurred in 1929, when Dale and a chemist Dudley discovered that acetylcholine was a natural constituent of both horse and ox spleens (Fig. 3.5J), thus giving the long sought after ‘depot of choline derivatives’ in the body. This then provided the necessary impetus for once more examining the role of acetylcholine as the mediator of the effects of transmission from parasympathetic nerves to the effectors of the autonomic nervous system as well as at other sites of transmission in the body. For as they state in their paper (Dale & Dudley, 1929): But there has been a natural and proper reluctance to assume, in default of chemical evidence, that the chemical agent concerned in these effects, or in the humoral transmission of vagus action, was a substance known, hitherto, only as a synthetic curiosity, or as an occasional constituent of certain plant extracts… It appears to us that the case for acetylcholine as a physiological agent is now materially strengthened by the fact that we have now been able to isolate it from an animal organ and thus to show that it is a natural constituent of the body. They go on to say: We feel, however, that its definite isolation from one organ has so far altered the position that, when an extract from, or a fluid in contact with the cells of, an animal organ can be shown to contain a principle having the actions, and the peculiar instability, of acetylcholine, it will be reasonable in future to assume the identification. On such lines a physiological survey of its distribution should be practicable. Similarly, when there is evidence associating some physiological event with the liberation of a substance indistinguishable from acetylcholine by its action, the presumption that it is, indeed, that ester will be strengthened by the knowledge that acetylcholine occurs in the normal body. By 1930, then, three decades of research had shown firstly that acetylcholine, either synthesized or extracted from ergot, had dramatic depressor effects (Ewins, 1914; Hunt & Taveau, 1906; Dale, 1914b); secondly, that stimulation of the vagus to the heart released a substance that seemed to mediate transmission and which had properties remarkably similar to that of acetylcholine (Loewi, 1921), and finally that acetylcholine occurred naturally in mammals (Dale & Dudley, 1929). The question that came to dominate the 1930s was whether acetylcholine could be detected in the overflow from other organs than the heart during nerve stimulation, and also whether the effects of such stimulation could be mimicked by the close arterial injection of acetylcholine into these organs. Loewi had already shown that the effects of exogenous acetylcholine on the heart or of stimulation of the vagus could be greatly enhanced if esterases for acetylcholine in this organ were first inhibited using physostigmine or eserine (Fig. 3.5G, H and I; Engelhart & Loewi, 1930). In 1930, Matthes, working in Dale’s laboratory, showed that the destruction of acetylcholine in the blood was due to the action of an esterase and that the action of this enzyme could be inhibited by eserine. Thus inhibition of acetylcholinesterases was a necessary requirement of any attempt to show that acetylcholine was released from a particular nerve ending, either by using an intravenous injection of eserine or by adding it to the organ bath. Shortly after this work of Matthes, Feldberg arrived in Dale’s laboratories as a refugee from Germany to Great Britain. He introduced the eserinized leech muscle preparation which had previously been shown to be exquisitely sensitive to applied acetylcholine by the German pharmacologist Fuhner in 1918. Feldberg was inspired to use this approach because of the experiments of Loewi and Navratil (1926b) on the actions of physostigmine in potentiating the effects of applied acetylcholine on the frog’s heart. In the hands of Dale and Feldberg (1934) the eserinized leech muscle was
THE DISCOVERY OF ACETYLCHOLINE AND THE CONCEPT OF RECEPTORS AT SYNAPSES
59
Fig. 3.6. The identification of acetylcholine as a transmitter substance in autonomic ganglia. A. Effect of fluid collected from stimulated cat sympathetic ganglia on contraction of the frog’s heart and of eserinised leech muscle. Upper panel: frog’s heart (Straub’s method). Lower panel: leech muscle treated with eserine. (A), fluid collected from the ganglion during the stimulation. (D), control fluid. (B), (C), acetylcholine (15 and 30 mg per litre respectively). No time base or contraction calibration is given in the original figure. (From Fig. 6 in Feldberg & Gaddum, 1934). (B) Effect of eserine on nictitating membrane contraction due to stimulation of the cervical sympathetic. Responses to equal groups of submaximal shocks to cervical sympathetic; (A) and (B) before, (C) and (D) during perfusion of eserine 1 in 106. Time calibration is 10 ms; no contraction calibration is given. (From Fig. 6 in Feldberg & Vartiainen, 1934). (C) The evaluation of esterase activity in normal and decentralized sympathetic ganglia using rabbit’s jejunum contraction as test. Numbers give the contractions to the following amounts of acetylcholine: 1, 1.5 µg; 2, 2 µg; 7, 1.8 µg; 8, 1 µg; 10, 1.5 µg; 11, 1.7 µg; 12, 1.9 µg; 14, 2µg. Numbers 9 and 13 give the contraction in the presence of 10µg of acetylcholine plus an extract of normal sympathetic ganglia. Numbers 4 and 5 give the contraction in the presence of 3 µg and 2 µg of acetylcholine respectively in the presence of extract of denervated sympathetic ganglia. No time or contraction calibration is given. (From Fig. 1 in Brucke, 1937).
first used to show that acetylcholine appeared in the venous blood after stimulation of the vagus nerve to the stomach. This was followed by experiments that showed acetylcholine also appeared after stimulation of the splanchnic nerves to the suprarenal glands (adrenal medulla; Feldberg et al, 1934). 3.8 The physiological action of acetylcholine in autonomic ganglia In 1934 experiments were also begun to see if acetylcholine could be detected at neuronal synapses in addition to neuroeffector junctions. To this end Feldberg and Vartiainen (1934) were able to show that: …when the superior cervical ganglion of the cat is perfused with warm, oxygenated Locke’s solution containing a small proportion of eserine, acetylcholine appears in the venous effluent whenever the cervical sympathetic nerve is effectively stimulated, and only then The assay for this acetylcholine was either the frog’s heart (Fig. 6A, upper panel) or the leech muscle treated with eserine (Fig. 3.6A, lower panel). They regarded these observations as (Feldberg & Gaddum, 1934): …support for the theory that the mechanism by which each impulse normally passes the synapse consists in the liberation of a small quantity of acetylcholine This discovery of the release of acetylcholine in autonomic ganglia then raised the possibility that the ‘Vagusstoff’ of Loewi was in fact liberated from ganglia in the heart rather than at the neuroeffector junction. Whilst this would certainly be the case, Feldberg and Gaddum (1934) were persuaded of the view that most likely the Vagustoff collected by Loewi came from both the synapses in the intramural ganglia as well as the neuroeffector junction, although no very effective argument was offered in defence of this proposition. One caveat in these experiments on the ganglia was that Eccles had in the same year shown that the electrical signs of the action potential set up in the postganglionic nerve trunk of the superior cervical ganglion by volleys in the preganglionic trunk
60
HISTORY OF THE SYNAPSE
were depressed by eserine rather than potentiated as would be expected as a consequence of an enhanced transmission through the ganglion on inactivation of acetylcholinesterase with eserine (Eccles, 1934). Later in 1934, Feldberg and Vartiainen offered a vigorous defense of the idea that acetylcholine is the transmitter substance in autonomic ganglia. Using the nictitating membrane of the cat as a measure of the postganglionic volleys in response to stimulation of the preganglionic supply to the ganglion, they were able to show that 18 impulses delivered in 10s gave rise to a greatly potentiated response of the membrane after the ganglion had been perfused with eserine (Fig. 3.6B(C) and 3.6B(D)) compared with that in the absence of eserine (Fig. 3.6B(A) and (B)), leading them to conclude that: …these new items of evidence entirely support the conception of transmission at ganglionic synapses by liberation of acetylcholine. The idea that eserine has a depressant action on cholinesterase, which may be located at nerve terminals in ganglia, so that eserine potentiates the effects of endogenously released acetylcholine, was shown to be very likely when Brucke (1937) found high concentrations of cholinesterase in the superior cervical ganglion. He showed that this mostly disappeared on section and degeneration of the preganglionic nerves to the ganglion. These results are illustrated in Fig. 3.6C, where the contractile responses of the rabbit’s jejunum to different concentration (in gamma units) of acetylcholine alone (1=1.5; 2=2; 6=2; 7=1.8; 8=1; 10=1.5; 11=1.7; 12=1.9; 14 =2), or acetylcholine together with extract of normal ganglia that therefore contains cholinesterase (9 and 13 =10) or acetylcholine plus extract of denervated ganglion plus eserine (4=3; 5=2) are shown. It will be noted that much higher concentrations of acetylcholine and extract of normal ganglion had to be used in order to get responses comparable to that of acetylcholine alone, indicating the effects of cholinesterase in this case (Fig. 3.6C). The consensus opinion at the end of the 1930s was that acetylcholine acted as the transmitter of impulses in autonomic ganglia. Eccles provided the main continuing resistance to this idea with some persuasive arguments that are detailed elsewhere (Bennett, 1994). 3.9 The identity of acetylcholine as the transmitter substance at somatic neuromuscular junctions Although the concept of the transmitter receptor was developed primarily in relation to striated muscle, as detailed above, identification of acetylcholine as the transmitter substance that acts on these receptors came relatively late, well after the establishment of acetylcholine as the transmitter from the vagus nerve to the heart. In the late 1920s and early 1930s the problem of the relationship between motor nerves and muscle revolved around questions relating to their relative excitability and the actions of agents thought to exert effects at the junction between nerve and muscle, such as curare, had on this excitability. Lapicque had at the beginning of the century carried out a series of experiments on the excitability of nerve and muscle in which he had defined the chronaxie and rheobase of the strength-duration curve for setting up excitation in these tissues, as are now described in many text books (Lapicque & Lapicque, 1906; Lapicque, 1926). In these works he developed the concept that nerve and muscle possessed the same chronaxie which he defined as isochronism. This was challenged by Rushton (1930) who showed, following the work of Lucas (Lucas & Mines, 1907), that in general muscle possessed two different excitabilities, one associated with the intramuscular nerves and the other with the muscle fibres themselves (Rushton, 1930, 1932), so that nerve and muscle did not possess isochronism. The possibility that nerve terminals and the muscle or endplate could be brought into isochronism by the action of acetylcholine was then entertained as involved in the direct transmission of the nerve impulse into the muscle. This proposition was put forward as agents, such as acetylcholine, could shorten the chronaxie (Fredericq, 1924) whereas curare lengthened the chronaxie (Fig. 3.7A; Lapicque, 1934). The heated arguments used in the considerable controversy between Lapicque and Rushton outlined in their papers in the Journal of Physiology seemed to offer at this time the possibility for an esoteric interaction between the action of agents known to affect transmission at the endplate and the electrical properties of this region of the muscle. These electrical controversies concerning isochronism in relation to the mechanism of transmission between nerve and muscle declined in the second half of the 1930s. First, using the eserinized leech preparation, Dale et al. (1936) showed that considerable quantities of acetylcholine could be collected in venous fluids from the cat gastrocnemius muscle following nerve stimulation (Fig. 3.7C). This also occurred if the muscle was directly stimulated, even in the presence of curare, but not if the muscle had previously been denervated. Dale and his colleagues were careful at this time not to claim this showed acetylcholine, released at nerve terminals, transmitted the impulse from the terminals to muscle. They were still open to an interpretation: That the propagated disturbance in the nerve fibre is directly transmitted to the effector cell, but that the latter cannot accept it for further propagation unless sensitized by the action of the acetylcholine, which appears with its arrival at the nerve ending. Such an hypothesis might be stated in terms of Lapicque’s well-known conception, by supposing that the
THE DISCOVERY OF ACETYLCHOLINE AND THE CONCEPT OF RECEPTORS AT SYNAPSES
61
action of acetylcholine shortens the chronaxie of the nerve cell, or of the motor end-plate of the muscle fibre, so that it is momentarily attuned to that of the nerve. Later in 1936 Brown, Dale and Feldberg showed that injection of acetylcholine directly into the empty arteries of a normal mammalian muscle could, if given with adequate rapidity, cause contraction of the muscle at not less that half the speed of a maximal motor nerve twitch (Fig. 3.7B). Furthermore, eserine caused the response to stimulation of the nerve supply to be converted from a simple twitch to a repetitive response with a maximum tension twice that of the normal twitch, a result reproduced and examined further by Bacq & Brown (1937; Fig. 3.7F). No mention is made any longer of Lapicque’s theories, especially given the effect of close intrarterial injection of acetylcholine mimicking the normal twitch response of the muscle to nerve stimulation. Rather they suggested that: …acetylcholine, is liberated by arrival of the nerve impulses at the nerve ending, and destroyed during the refractory period by a local concentration of cholinesterase. The facts supporting this hypothesis are: (1) that acetylcholine, when suitably injected into the muscle, produces the kind of contraction which the transmitter should produce; (2) that acetylcholine, identified as such, is liberated by impulses at motor nerve endings; and (3) that a suitable dose of eserine causes the muscle to give a short, waning, tetanic response to a single, synchronous volley of nerve impulses Many of these actions of applied acetylcholine on muscle were subsequently examined by measuring the rate of impulse firing in the muscle by Brown (1937). He found that close arterial injection of acetylcholine into a denervated muscle gave rise to a quick initial contraction (Fig. 3.7E) that was accompanied by a burst of action potentials in the muscle. He comments that: The facts presented give incidental support to the suggestion of a concentration of cholinesterase at the mammalian motor nerve endings. That this is the case was shown by Couteaux and Nachmansohn (1940) who determined the concentration of cholinesterase in the middle portion of the guinea pig’s gastrocnemius. This work showed that cholinesterase was indeed found at relatively high levels in the central region of the muscle where the nerve endings are located (Fig. 3.7G). By 1940 acetylcholine was believed to be the agent of transmission between motor nerve and muscle. However it should be noted that Eccles, who together with O’Connor (1938) had just recorded for the first time the electrical signs of transmission at the endplate (the so called ‘endplate potential’; see also Gopfert and Schaefer, 1937), did not come around to this opinion until he worked on motor nerve transmission to muscle with Katz and Kuffler a few years later (Eccles et al., 1941). Indeed Loewi at the time of being awarded the Nobel Prize with Dale in 1936 was not persuaded of chemical transmission at the motor endplate. Nevertheless, most students of transmission accepted by 1940 that the ‘receptive substance’ of Langley could be identified as a receptor for acetylcholine, although the full flavour of the opposition to the concept of chemical transmission from the time of Loewi’s experiments after the Great War up to the 1950s can only be gauged by reading a first hand account of the controversies (for example that of Bacq, 1983). 3.10 The discovery of the physiological action of single acetylcholine receptors The study of the physiological action of single acetylcholine receptors began in 1970 with the discovery by Katz and Miledi (1970b) of membrane noise at the endplate in response to the steady action of acetylcholine from a micropipette. They hypothesised that during such a steady application: the statistical effects of molecular bombardment might be discernible as an increase in membrane noise, superimposed on the maintained average depolarisation. This is what in fact they observed with an intracellular electrode as can be readily ascertained by inspection of Fig. 3.8A. A simple relationship was then used that connects the size of the elementary voltage (a) due to the opening of a single channel as a consequence of acetylcholine binding to a receptor to the average depolarisation (V) and the root mean square value of its fluctuation (E), namely . Applied to the results of Fig. 3.8A this gave a value for the elementary event of 0.29 µV. In 1971 the
62
HISTORY OF THE SYNAPSE
Fig. 3.7. The identification of acetylcholine as the transmitter substance at the somatic neuromuscular junction. (A) Strength-capacity curves on a toad’s gastrocnemius; O, before curare; I, II, III, after a small dose of curare. In the upper right-hand corner, variation of the chronaxie with the time in hours as abscissa. (From Fig. 2 in Lapicque, 1934). (B) Spinal cat. Record from denervated gastrocnemius, 14 days after nerve section. Perfusion with Locke’s solution. Effect of 1 µg of Ach. by close arterial injection. No time calibration is given. (From Fig. 5 in Brown et al., 1936). (C) Effect of venous fluids from gastrocnemius muscles of cat, in 75 percent. Dilution, on contraction of eserinized leech muscle. (A) and (B) from denervated muscle, (A) resting, (B) during stimulation; (C) and (D) from normal muscle, (C) resting, (D) during stimulation. No time or contraction calibration is given. (From Fig. 5 in Dale et al., 1936). (D) Contractions of decerebrated cat gastrocnemius 20 days after nerve section, in response to close arterial injections of (A), 2–5 µg Ach., (B), 50µg carabaminoylcholine and (C), 21 min. later 10µg Ach. (From Fig. 5 in Brown, 1937). (E) (A), (B) and (C), decerebrate cat, contractions of denervated (12 days) gastrocnemius in response to 0.25 µg, 1.0 µg and 2.5 µg Ach. by close arterial injection respectively. (D), spinal cat, contraction of denervated gastrocnemius in response to 20 µg Ach. by ‘distant’ injection into inferior mesenteric artery. (From Fig. 1 in Brown 1937). (F) Spinal cat, 9 days after lumbosacral sympathectomy. Contractions of gastrocnemius in response to maximal shocks to nerve at 10 sec. intervals. (A), arterial injection of 5 pig eserine; (B), 10 min. later; (C), arterial injection of 20 µg eserine. No time or contraction calibration is given. (From Fig. 4 in Bacq & Brown, 1937). (G) Concentration of choline esterase in the middle portion of the interior section of a guinea pig’s gastrocnemius cut in 11 slices of similar thickness and weight. Each horizontal line corresponds to one slice and indicates its weight in % of the total weight. Abscissae: Region from which the tissue was obtained in terms of order of consecutive slices. Point 50 corresponds to the center region where the nerve endings are situated. Ordinate: Choline esterase in mg Ach hydrolyzed per hour by 100 mg of fresh tissue. (From Fig.1 in Couteaux & Nachmansohn, 1940).
same authors determined the approximate time course of the elementary conductance change underlying the elementary event by recording the extracellular voltage fluctuations due to the bombardment of receptors with acetylcholine. In this case they
THE DISCOVERY OF ACETYLCHOLINE AND THE CONCEPT OF RECEPTORS AT SYNAPSES
63
Fig. 3.8. Discovery of the physiological action of single acetylcholine receptors at the somatic neuromuscular junction. (A) Intracellular recordings of membrane potential from end-plate region of frog muscle fibre. In each block, the upper trace was recorded on a low gain d.c. channel (10 mV scale); the lower was simultaneously recorded on a high gain a.c. coupled channel (0.4 mV scale). The records in the upper row are controls (no ACh); the lower row shows membrane noise during ACh application, by diffusion from a micropipette. In the lower records, the increased distance between a.c. and d.c. traces shows upward displacement of the d.c. trace because of ACh-induced depolarization. Two spontaneous m.e.p.p.s are also seen. (From Fig. 1 in Katz & Miledi, 1970b). (B) Power spectrum of intracellularly recorded ACh noise. Temperature 5.5°C. Linear plot of E2/ f, (see text), in relative units against frequency in Hz in the upper part; double-log plot in lower part. (From Fig. 1 in Katz & Miledi, 1971 a). (C) Schematic circuit diagram for current recording from a patch of membrane with an extracellular pipette. VC, Standard two— microelectrode voltage clamp circuit to set locally the membrane potential of the fibre to a fixed value. P, Pipette, fire polished, with 3 to 5 mm diameter opening, containing Ringer’s solution and agonist at concentrations between 2×10−7 and 6×10−5 M. d.c. resistance of the pipette: 2 to 5 Mohms. The pipette tip applied closely on to the muscle fibre within 200 µm of the intracellular clamp electrodes. VG, Virtual ground circuit, using a Function Modules Model 380K operational amplifier and a 500 Mohm feedback resistor to measure membrane current. The amplifier is mounted together with a shielded pipette holder on a motor-driven micromanipulator. V.Bucking potential and test signal for balancing of pipette leakage and measuring pipette resistance. (From Fig. 1 in Neher & Sakmann, 1976). (D) Oscilloscope recording of current through a patch of membrane of approximately 10 mm2. Downward deflection of the trace represents inward current. The pipette contained 2×10−7 M SubCh in Ringer’s solution. The experiment was carried out with a denervated hypersensitive frog cutaneous pectoris (Rana pipiens) muscle in normal frog Ringer’s solution. The record was filtered at a bandwidth of 200 Hz. Membrane potential: −120 mV. Temperature: 8°C. (From Fig. 2 in Neher & Sakmann, 1976).
ascertained the power spectrum of the acetylcholine induced noise, that is the relationship between () and the frequency (Fig. 3.8B). This gave an average time constant of the elementary event of about 1 ms when the event is treated as decaying exponentially and a net charge transfer across the open channel of about 5×104 univalent ions, with a channel conductance of about 10−10 Siemens.
64
HISTORY OF THE SYNAPSE
Direct recording of the electrical signs of individual acetylcholine receptor channels was made by Neher and Sakmann in 1976. They introduced the technique of recording from a small membrane area of the muscle, so as to decrease background noise (Fig. 3.8C). The tip of a glass pipette of 3 to 5 mm diameter, with fire polished edges, was connected up in the circuit shown in Fig. 3.8C after being filled with Ringer’s solution and an acetylcholine receptor agonist, in this case suberyldicholine (SubCh). This pipette was then applied to the surface of a muscle fibre, denervated so as to ensure an abundance of receptors all over the surface of the muscle and subjected to enzyme treatment for the digestion of connective tissue and the basement membrane. Discrete conductance changes like those shown in Fig. 3.8D could only be resolved if the conductance between the pipette interior and the bath decreased by a factor of at least four after the pipette came into contact with the muscle membrane. Inspection of Fig. 3.8D shows that the amplitude of the single channel conductance is about 3.4 pA, giving a channel conductance of 28×10−12 Siemens. This conductance is about the same as that determined by Katz and Miledi (1973) for the agonist SubCh using the noise method of Fig. 3.8A, namely 28.6×10−12 Siemens. 3.11 Conclusion The saga of the concept of the receptor has been followed from its beginnings in the hands of Langley and Ehrlich to the triumph of recording the electrical signs of the opening of a single acetylcholine receptor channel. This work took almost exactly a century to accomplish, from the experiments of Langley in 1874 on pilocarpine and atropine to those of Neher and Sakmann in 1976. The structure of the receptor molecule was also opened up by the discovery in the late 1960s by Lee and his colleagues of toxins that could irreversibly bind to the receptor (Lee & Chang, 1966; Lee, 1972), and so allow for its isolation. But that story takes us too far from the main theme of this chapter, which has been the establishment of the reality of the transmitter receptor.
4 The Discovery of Adrenaline and the Concept of Autoreceptors at Synapses
4.1 Introduction: the discovery of noradrenaline as a transmitter The presser effects of suprarenal extract were first shown to exist by Oliver and Schafer (1895) in 1894, with the first full paper describing their results appearing in 1895 (Fig. 4.2A). In 1899 Abel (1899) obtained an extract from the suprarenal gland, which was a mono-benzoyl derivative, that was less active than crude extracts of the gland, and which he named ‘epinephrin’. Later, Abel (Fig. 4.1 A) produced a crystalline epinephrine-hydrate, which he thought had the same action as the gland extract. However, this was not the case, and it was up to an industrial chemist in 1901, Takamine (1902), to take Abel’s hydrate and purify it to the active principal. This he patented as ‘Adrenalin’, which he had Parke, Davis & Co. market. Presumably Takamine chose the word ‘Adrenalin’ as the patent name because the physiological community in Britain had been using the name ‘adrenaline’ to refer to the active principal of the gland for some years. At this time, 1901, Langley showed that stimulation of the sympathetics results in changes of effectors, which in many cases can be mimicked by application of suprarenal extract (Langley, 1901). Then Elliott in 1904, working in Langley’s laboratory, published his much-quoted note suggesting that adrenaline might be the chemical stimulant liberated on each occasion when the impulse arrives at the periphery (Elliott, 1904a). In that same year Stolz (1904) and Dakin (1905) independently synthesised adrenaline. In 1910 Barger and Dale showed that the effects of sympathetic nerve stimulation are more closely reproduced by the injection of sympathomimetic primary amines than by adrenaline or secondary amines (Barger & Dale, 1910) (Fig. 4.2B). They studied a large number of synthetic amines related to adrenaline and termed their action ‘sympathomimetic’, without establishing that noradrenaline is the sympathomimetic transmitter, although coming very close to doing so. After the First World War, in 1921, Cannon (Fig. 4.1B) together with Uridil reported that stimulation of the sympathetic hepatic nerves in mammals releases an adrenaline-like substance (which they called sympathin) that increases blood pressure and heart rate (Cannon & Uridil, 1921) (Fig. 4.2C). However adrenaline when injected into the body was shown to accelerate heart rate and simultaneously dilate some vascular beds while constricting others; on the other hand sympathin accelerated the heart without dilation of vascular beds and produced an increase in total peripheral resistance and diastolic blood pressure. Again the possibility, following the observations of Barger and Dale (1910), that sympathin might actually be demythelated adrenaline (noradrenaline) was not pursued, although Bacq (1974) repeatedly advanced this idea. The identification of the sympathomimetic substance as noradrenaline in mammals was further delayed by Loewi’s observation in 1921 that an accelerator substance is released from the toad’s heart on stimulation of the sympathetic nerves (at least those in the vagus that predominate over the parasympathetics in summer; Fig. 4.2D). He concluded that this might be adrenaline, which it is in the case of toads (Loewi, 1921). It took another twenty-five years before the identity of the sympathomimetic substance in mammals was established by von Euler in 1946. He showed that highly purified extracts of sympathetic nerves and effector organs resemble noradrenaline by all criteria available (Fig. 4.2E) and proposed noradrenaline as the sympathetic transmitter (Euler, 1946). Subsequently Ahlquist in 1948 examined the effects of adrenaline and noradrenaline as well as isoproterenol on a variety of target tissues (Fig. 4.2F). He concluded that the difference in action of these catecholamines could be explained by the presence of two types of receptor for the catecholamines, namely alpha and beta (Ahlquist, 1948). 4.2 Early observations leading to the idea of autoreceptors In 1931 Cannon and Bacq reported an increase in the sympathin in the blood stream, as determined by its effects on the heart, following stimulation of the sympathetic nerves to the colon or to the hind quarters (Cannon & Bacq, 1931). Subsequently Jang showed that the responses of several organs to sympathetic nerve stimulation could be potentiated by appropriate concentrations of the adrenergic receptor blocking drugs 933F, yohimbine and ergotoxine (Jang, 1940). That such blocking drugs may enhance the effects of sympathetic nerve stimulation was further elaborated by Holzbauer and Vogt in 1954, who
66
HISTORY OF THE SYNAPSE
Fig. 4.1. The discovery of adrenaline. (A) John J.Abel (1857–1938). Taken from Dale, 1939. (B) Walter Bradford Cannon (1871–1945). Taken from Finger, 1994.
showed that the responses of the dibenzyline pretreated uterus to exogenous adrenaline were greatly enhanced over control untreated preparations (Holzbauer & Vogt, 1956). However, it was not until 1957 that a thorough analysis was made of this phenomenon whereby alpha adrenergic blocking drugs potentiate the overflow of noradrenaline and the contractile response of organs to sympathetic nerve stimulation. Brown and Gillespie stimulated the sympathetic nerves to the spleen of the cat and determined the amount of noradrenaline in the venous blood during different frequencies of stimulation and in the presence of different alpha adrenergic blocking drugs (Brown & Gillespie, 1957). Following 200 impulses, noradrenaline output could just be detected at 10 Hz, with maximum output occurring at 30 Hz (Fig. 4.3A). They showed that the low level of output of noradrenaline at 10 Hz was not due to the breakdown of the catecholamine by monoamine oxidase. Furthermore the level at 10 Hz could be increased to the high level found at 30 Hz if adrenergic blocking drugs such as dibenamine were present in the perfusion fluid (Fig. 4.3A). This led them naturally to the conclusion that: The constancy of the output at frequencies between 1 and 30/sec after dibenamine would be explicable on the ground that the dibenamine had prevented the destruction of the transmitter. The known effect of dibenamine is to block the tissue receptors for noradrenaline, and we must conclude therefore that combination with the receptors is a necessary prelude to the destruction and removal of liberated noradrenaline. In 1961, an article by Koelle in Nature dramatically challenged the conceptual framework in which the mechanisms of synaptic transmission might be considered. He suggested, on the basis of observations showing that the threshold dose of intra-arterially injected carbamylcholine into the superior cervical ganglion was greatly elevated following denervation of the ganglion, that: …acetylcholine mobilised by a conducted presynaptic impulse may act at two sites: postsynaptically, as outline above, but prior to that at the presynaptic terminal itself, to bring about the release from the terminal of sufficient additional acetylcholine or of some other neurohumoral agent to initiate the characteristic postsynaptic effect (Koelle, 1961). Other neurohumoral agents might include noradrenaline, for at that time a theory was in currency that cholinergic mechanisms might be involved in the release of catecholamines (Burn & Rand, 1959). Fig. 4.3B summarises Koelle’s conjectures: acetylcholine is released from vesicles (open circles) and acts back on the nerve terminal to release more acetylcholine containing vesicles (open circles) in the case of cholinergic synapses; in the case of adrenergic nerve terminals acetylcholine acts back on the terminal to release noradrenaline (filled circles). It is interesting that there is now considerable evidence to support the idea that acetylcholine released at motor-nerve terminals during a train of impulses does act back to accelerate the release of further acetylcholine (Wessler, 1992). In the case of adrenergic nerve terminals, there appear to be both nicotinic and muscarinic receptors that can modulate catecholamine release, either increasing or decreasing the release depending on the organ being considered, although it is not clear if these receptors have a physiological role (Fredholm, 1995). Koelle’s idea that acetylcholine acts as a ubiquitous intermediary in synaptic transmission, independent of the identity
THE DISCOVERY OF ADRENALINE AND THE CONCEPT OF AUTORECEPTORS AT SYNAPSES
67
Fig. 4.2. Identification of adrenaline and noradrenaline as transmitter substances released by sympathetic nerves. (A) Suprarenal extract exerts presser effects. Effect of injecting extract of 0.2 gram of dog-suprarenal into a 16 kilo dog. A, auricle; B, ventricle; C, femoral artery; D, injection of suprarenal (watery extract), the line D also indicates 0 pressure; E, time in 0.5 sec. Note delayed inhibition. Taken from Oliver & Schafer, 1895. (B) Injection of sympathomimetic primary amines closely reproduces effects of sympathetic nerve stimulation. At A, 2 c.c of N/100_ phenylpropylamine injected. At B, 2 c.c. of N/100 phenylethylamine injected. At C 2 c.c. of N/100 p. hydroxyphenylethylamine injected. Taken from Barger & Dale, 1910. (C) Evidence that stimulation of the sympathetic hepatic nerves releases an adrenaline-like substance. Average increments of heart rate in 5 second intervals in 5 cases of hepatic-nerve stimulation for 30 seconds (continuous line). Also average increments of heart rate in the same intervals from reflex adrenal stimulation. Taken from Cannon & Uridil, 1921. (D) Evidence that the perfusate of a toads heart collected during a period of stimulation of the vagus nerve when the sympathetic component dominates gives rise to an increased ionotropic effect. 1. Ringer from a 20 min normal period. 2. Ringer from a 20 min accelerans-stimulation period. 3. Ringer. Taken from Loewi, 1921. (E) Effect of purified extracts of sympathetic nerves resemble effects of noradrenaline. Blood pressure of the chloralosed cat is shown. 1, 0. 06 g splenic nerves, cattle, extract treated with Fuller’s earth. 2, 0.5 µg nor-adrenaline-HCI. 3, 1 mg dihydroxynor-ephedrine. 4, 1 µg adrenaline. Then between 4 and 5, an amount of 0.6 mg dihydro-ergotamine/kg i.v. 5, 0.25 g splenic nerves. 6, 2 µg nor-adrenaline-HCI. 7, 4 µg dihydroxy-nor-ephedrine. 8, 3 µg adrenaline. Time 30 seconds. Taken from Euler, 1946. (F) Comparative action of amines on mean arterial pressure of cats, dogs, and rabbits. Ordinates: pressure in mm. Hg; abscissae: time marks at 10 sec. intervals. The amines were injected intravenously and the key in the box indicated the amines injected. Cats: average results for 6 cats with dosage of 0.1 cc. M/1000 solution per kgm. The box gives the average results for 3 cats after dibenamine, dosage, 0.1 cc per kgm. of the concentrations shown. Dogs: average results from 12 determinations in 6 dogs with dosage of 0.05 cc. of a M/1000 solution per kgm. Rabbits: average results from 10 determinations in 6 rabbits with dosage of 0.1 cc. of a M/1000 per kgm solution. Taken from Ahlquist, 1948.
of the transmitter that finally exert an effect on the postsynaptic receptors (Fig. 4.3B), was not supported by further work, but his challenging concepts did open up the possibility that the action of adrenergic and cholinergic blocking drugs might be on the nerve terminal as well as on the postsynaptic membrane.
68
HISTORY OF THE SYNAPSE
4.3 Direct experimental evidence for autoreceptors The year 1971 saw the publication of a large number of papers on different peripheral and central organs that established the existence of presynaptic receptors that could control transmitter release during a short train of impulses in a nerve terminal. The work of Brown and Gillespie (Fig. 4.3A) was not reinterpreted in the light of Koelle’s conjectures (Fig. 4.3B) as evidence for a presynaptic effect of noradrenaline on its own release as Koelle had emphasised that it was always acetylcholine that exerted this effect. Furthermore, with the subsequent discovery in the 1960s of the catecholamine uptake pumps in the terminals of adrenergic neurones (Iversen, 1967); other interpretations of the observations of Brown and Gillespie were popular. Chief amongst these was that alpha adrenoceptor blocking agents, such as phenoxybenzamine, increased overflow of noradrenaline as a consequence of blocking neuronal uptake of the catecholamine. In addition, such blocking agents maintain blood flow in the spleen during sympathetic nerve stimulation by preventing vasoconstriction; this alone would be expected to enhance the diffusion of noradrenaline from the nerve terminals after it had been liberated and so increase overflow. A direct test of the blood flow argument was made by Kirpekar and Puig in 1971. They perfused spleens with a constant outflow pump during the application of phenoxybenzamine or phentolamine and then stopped the flow for different times to test whether the affects of the drug on overflow of noradrenaline could be attributed to its preventing vasoconstriction. Their argument was that if the increase in transmitter output after phentolamine treatment was mainly due to its action in blocking vasoconstriction and thus improving flow rates, then restriction on flow should cause a much greater percentage reduction in phentolamine treated than in untreated spleens. Fig. 4.3C shows that this was not the case, as restricting the blood flow for 30 s resulted in the removal of about the same percentage of noradrenaline in the control spleens as in those treated with phentolamine. These results lead to the conclusion that the alpha receptor blocking drugs do not enhance overflow as a consequence of blocking vasoconstriction (Kirpekar & Pulg, 1971). Kirpekar and Puig observed that noradrenaline released during nerve stimulation appeared to be removed during the course of the stimulation whereas Blakeley and Brown (1964) showed that the uptake of noradrenaline infused into the spleen was not affected by nerve stimulation. In order to reconcile these observations Kirpekar and Puig suggested that: …the noradrenaline released by nerve stimulation acts on these alpha sites of the presynaptic membrane to inhibits its own release. In the presence of adrenergic blocking agents the transmitter output would be enhanced, since the inhibitory alpha sites on the nerve terminals are blocked. These are probably not the same sites through which noradrenaline is taken up into the nerve endings (Kirpekar & Pulg, 1971). Farnebo and Hamberger independently came to the same conclusion in 1971 as a consequence of their studies of the overflow of noradrenaline from nerve-stimulated rat irides. In this work alpha adrenoceptor antagonists such as phenoxybenzamine and phentolamine, used at concentrations that do not affect amine uptake into nerve terminals, increased the overflow of the catecholamine (Fig. 4.3D), whereas alpha adrenoceptor agonists like clonidine decreased the overflow. These authors carefully excluded the possibility that these agents changed the rate of metabolism of the amine by, for example, monoamine oxidase or catechol-o-methyltransferase. They suggested that a possible explanation for their results was inhibition of noradrenaline release via alpha-adrenoceptors on the nerve terminal (Farnebo & Hamberger, 1971). Starke in the same year reached much the same conclusion from his studies of the effects of alpha-adrenoceptor agonists on the overflow of noradrenaline from the isolated perfused rabbit heart, observations that were particularly interesting because of the lack of postsynaptic alpha receptors in the heart. Phenylephrine produced a dose-dependent inhibition of noradrenaline released by sympathetic nerve stimulation (Fig. 4.3E). This lead Starke to conclude that: …adrenergic nerve terminals are endowed with structures related to the alpha adrenoceptive sites of effector cells. On reaction with the alpha stimulants, e.g. with liberated noradrenaline, these neuronal alpha receptors mediate the inhibition of noradrenaline liberation; in the presence of the alpha blockers, this restriction is attenuated (Starke, 1971). This work on the heart was extended several years later by Langer, Adler-Graschinsky and Giorgi in 1977, in an investigation that took advantage of the absence of postsynaptic alpha receptors to determine if blockade of prejunctional alpha receptors on the sympathetic nerve terminals would lead to an expected increase in the overflow of noradrenaline accompanied in this instance by an increase in the response of the organ. They showed that an alpha-receptor antagonist (phentolamine) could produce both a significant increase in the response of guinea-pig atria to nerve stimulation (see Fig. 4.4F), and at the same time enhance the overflow of noradrenaline fourfold (Langer et al., 1977). Interestingly enough at this time (1971) evidence was also growing for autoreceptors at cholinergic and even GABAergic synapses in the central nervous system. Polak showed that atropine could exert a stimulating effect on the release of acetylcholine due to incubating cortical slices of rat brain in high potassium solutions (Fig. 4.3F). As this stimulating action of atropine could be inhibited by adding muscarinic agonists such
THE DISCOVERY OF ADRENALINE AND THE CONCEPT OF AUTORECEPTORS AT SYNAPSES
69
Fig. 4.3. The existence of receptors on nerve terminals. (A) Discovery that adrenergic blocking drugs enhance the overflow of noradrenaline during stimulation of sympathetic nerves to the spleen. Individual results for the output/stimulus of noradrenaline after dibenamine, superimposed on the graph showing the average output in untreated animals (solid line). Taken from Brown & Gillespie, 1957. (B) The idea, due to Koelle, of nerve terminal autoreceptors and of muscle autoreceptors. Upper diagram shows released noradrenaline acting on a target smooth muscle cell as well as on its own terminal to release more noradrenaline. Note that the smooth muscle cells themselves are shown releasing transmitter onto themselves. Lower diagram shows released acetylcholine acting on a postsynaptic neuron as well as back onto the terminal from which it was released. The autoreceptors shown in both the adrenergic and cholinergic synapses accelerate release. Taken from Koelle, 1961. (C) Evidence that the increased overflow of noradrenaline from the spleen on nerve stimulation in the presence of an adrenergic receptor blocker is not due to decreased vasoconstriction enhancing overflow. Effect of flow-stop on noradrenaline (NA) overflow expressed as a percentage of the control output without flow stop is shown. Left and right hand parts of the figure show experiments in normal spleens and spleens treated with phentolamine. Nerves were stimulated at 10 Hz for 20 seconds. Flow was stopped for the duration of stimulation +30, 60 or 120 seconds. Control output was 208±60 ng in normal spleens. In spleens treated with phentolamine control output without flowstop was 499 ±208 ng. Taken from Kirpekar Pulg, 1971. (D) Effect of alpha adrenoceptor antagonists on the overflow of noradrenaline from nerve-stimulated irides. Efflux of tritium from irides of untreated rats superfused with drug-free buffer solution (left) or buffer solution containing phentolamine 10−6M (right). The irides were preincubated with [3H]-NA, 10−7M, for 30 min before superfusion and stimulation. Each point represents mean±S.E.M. of four experiments. Taken from Farnebo & Hamberger, 1971. (E) Alpha adrenoceptor agonists decrease the noradrenaline overflow from the heart. Influence of phenylephrine on the output of noradrenaline during sympathetic nerve stimulation from isolated rabbit hearts. Means±S.E.M, n=number of experiments. Significance of difference from controls was evaluated by Student’s t-test. Taken from Starke, 1971. (F) Muscarinic antagonists can enhance the release of acetylcholine. ACh content (expressed as nmol/g initial weight of wet tissue) of cortical slices and incubation media after 1 h incubation with and without atropine and oxotremorine. (Hatched histogram bars), ACh extracted from the tissue slices; (open histogram bars), ACh in the incubation medium. The figures on top of the columns give the number of observations from which the mean values were obtained. Note that the concentration of atropine in A is ten times less than in B as indicated in the abscissa. The concentrations of oxotremorine are different for each histogram bar as indicated in the abscissa. Taken from Polak, 1971. (G) Effect of alpha-adrenoceptor antagonists on the release of transmitter by sympathetic nerves as measured by excitatory junction potentials. Effect of phenoxybenzamine (10 µg ml−1) on the amplitude of the excitatory junction potential in the mouse vas deferens as stimulation of the sympathetic nerves at 10 Hz. (a) control (b) in the presence of phenoxybenzamine. Taken from Bennett, 1973.
70
HISTORY OF THE SYNAPSE
Fig. 4.4. Identification of presynaptic autoreceptors as different from postsynaptic receptors. (A) The autoreceptor hypothesis. Negative feedback mechanism for norepinephrine released by nerve stimulation. Response (R) of the effector cell mediated through alpha-receptors ( ), and through beta-receptors ( ). Note that, irrespective of the nature of the postsynaptic adrenergic receptor, the presynaptic receptor involved in the regulation of transmitter release is an alpha-receptor. Taken from Langer, 1974. (B) The result of removing the target organ or its nerve supply on the overflow of noradrenaline on exposure to an alpha receptor antagonist. Effects of phentolamine on the overflow of 3H-transmitter induced by 60 mM K+ in the rat submaxillary gland. Ordinates a and c: Fractional release: total 3H-released divided by that remaining in the tissue at the moment of the addition of K+. For estimation of fractional release the spontaneous outflow was subtracted from the total overflow of radioactivity elicited by exposure to 60 mM K+. Ordinates B and D: Ratio of the fractional release of radioactivity by the second potassium depolarization (S2) to that induced by the first one (S1). Abscissa: S1 and S2 indicate the two depolarization periods induced by exposure to 60 mM K+ with an interval of 37.5 min between the two stimulation periods. (Open histogram bars) normal glands; (Vertical striped histogram bars) atrophied glands (15 days after duct ligation). Where indicated phentolamine (Phent), 3.1 µM was added 30 min before S2. Shown between parentheses are the number of experiments per group. Shown are mean values ±S.E.M. * P<0.01 ; ** P<0.005 when compared against the corresponding controls (Cont). Taken from Fillinger, Langer, Perec & Stefano, 1978. (C) Decrease of tritiated-dihydroergocryptine in the denervated heart (6-OHOA treated). Binding of 3H-DHE and 3H-QNB in heart ventricle after 6-hydroxydopamine (6-OHDA) treatment. The Kd and Bmax presented are means (±S.E.M.) derived by Scatchard analysis from 5, 6 or 7 separate experiments, each of 6 to 9 concentrations, determined in duplicate. 1 P<0.05, 2 P<0.002 when compared to the corresponding controls (2-tailed unpaired Student’s t-test). Taken from Story, Briley & Langer, 1979. (D) Differences in the effectiveness on phenoxybenzamine in blocking presynaptic and postsynaptic alpha receptors. Effects of low concentrations of phenoxybenzamine (PBA) on the responses and on noradrenaline overflow elicited by nerve stimulation in the perfused spleen. Left ordinate: ratio of responses (increase in perfusion pressure in mmHg) between the second period of stimulation (S’) and the corresponding control (S) for each frequency of stimulation. Right ordinate: ratio of transmitter overflow (ng of noradrenaline) between the second period of stimulation (S’) and the corresponding control (S) for each frequency of stimulation, a: stimulation at 5 Hz for 60 sec (0.1 msec, supramaximal voltage), b: stimulation at 30 Hz for 10 sec (0.1 msec, supramaximal voltage). C: control (n=5) PBA: 8.8×10−10M (n=5); PBA: 2.9×10−9 M (n=8). Shown are mean values ±S.E.M. of the mean. * P<0.005; ** P<0.001 when compared with the corresponding control. Taken from Dubocovich & Langer, 1974. (E) Phenoxybenzamine does not block uptake of noradrenaline at concentrations at which it enhances overflow of noradrenaline. Effect of phenoxybenzamine in inhibiting alpha postsynaptic receptors mediating pressure responses and increasing exocytotic release by nerve stimulation. Ordinates: left, percent inhibition of pressure responses, o—o; right, DBH activity, ratio of overflow, (stimulated +phenoxybenzamine)/(stimulated-phenoxybenzamine), o____o. Shown are the mean values ±S.E.M. of five experiments. Taken from Cubeddu, Barnes, Langer & Weiner, 1974. (F) Blockade of prejunctional alpha receptors on sympathetic nerve terminals in the heart, which lacks postsynaptic alpha receptors, potentiates the overflow of noradrenaline. Potentiation by phentolamine of the positive chronotropic effects of accelerans nerve stimulation in isolated guinea pig atria. Increase in atrial rate induced by nerve stimulations S1–S9: at 0.5 Hz, for 10s, supramaximal voltage. The interval between each nerve stimulation was 20 min. Phentolamine (0.31 µM) was added to the medium 10 min before S5 and kept until 10 min after S7. In the absence of phentolamine the responses to nerve stimulation were sustained from S1 to S8. Mean values ±S.E.M. of 11 experiments are shown. *P<0.05 compared with S1 Taken from Langer, Adler-Graschinsky & Giorgi, 1977.
THE DISCOVERY OF ADRENALINE AND THE CONCEPT OF AUTORECEPTORS AT SYNAPSES
71
as oxotremorine and methacholine to the incubation medium, Polak suggested that atropine probably acted by inhibiting an inhibitory effect of the released acetylcholine on further release (Polak, 1971). At this time also Johnston and Mitchell showed that the GABA receptor antagonist bicuculline potentiated the electrically evoked release of GABA from cortical slices, without affecting the uptake of the amino acid. They commented that: …bicuculline might ‘disinhibit’ a presynaptic inhibitory action of GABA on its own release as has been proposed for the potentiation of central acetylcholine release by atropine (Johnston & Mitchell, 1971). By the end of 1971 there was a growing consensus that autoreceptors exist on nerve terminals. The first electrophysiological analysis of the actions of the alpha adrenoceptor antagonist phenoxybenzamine on excitatory junction potentials due to the stimulation of sympathetic nerves was carried out in 1973: this showed that during a short train of impulses there was a spectacular increase in the size of the junction potentials in the presence of the antagonist compared with controls (Fig. 4.3G), as expected if the noradrenaline released during the train acts on autoreceptors to decrease the release of transmitter by impulses (Bennett, 1973). The autoreceptor hypothesis was formally described by Langer in 1974 (Fig. 4.4A), in his summary of the research carried out in 1971 that had established this idea: In studies on transmitter overflow carried out with 3H-norepinephrine, it was discovered that under control conditions a significant fraction of the transmitter released by nerve stimulation is collected as metabolites (Langer, 1970; Su & Bevan, 1970; Tarlov & Langer, 1971; Langer & Vogt, 1971). Since phenoxybenzamine prevented the metabolism of 3H-norepinephrine released during nerve stimulation, inhibition of extraneuronal uptake by this drug was postulated as another factor contributing to the increase in transmitter overflow obtained in the presence of phenoxybenzamine (Langer, 1970; Langer & Vogt, 1971). Yet, when a quantitative analysis was carried out on the overflow of total radioactivity obtained in the presence of phenoxybenzamine it was concluded that, in addition to inhibition of neuronal and extraneuronal uptake of norepinephrine, phenoxybenzamine increases the output of norepinephrine elicited by nerve stimulation (Langer, 1970). A similar conclusion was reached by Starke, Montel and Schumann (1971) for phenoxybenzamine and for phentolamine, because these drugs increased transmitter overflow elicited by nerve stimulation in concentrations which did not inhibit either neuronal or extraneuronal uptake… The increase in transmitter release obtained by exposure to phenoxybenzamine was observed in the range of concentrations of the drug eliciting blockade of the alpha-receptors. However, a causal relationship between the block of the responses of the effector organ by phenoxybenzamine and the increase in transmitter release was excluded because similar results were obtained in isolated atria (Langer et al., 1971) and in the perfused rabbit heart (Starke et al., 1971), where the adrenergic receptors that mediate the response of the effector organ are of the beta type. These results led to the hypothesis of a presynaptic regulation of norepinephrine release through a negative feedback mechanism mediated by adrenergic alphareceptors. According to this hypothesis, the norepinephrine released by stimulation, once it reaches a threshold concentration in the synaptic gap, would activate presynaptic alpha-receptors, triggering a negative feedback mechanism that inhibits further release of the transmitter (Fig.4.4A; Farnebo & Hamberger, 1971; Starke, 1971). Much later some direct tests were made of the existence of presynaptic alpha receptors by either removing the effector organ whilst leaving the nerve supply intact or by removing the nerves and maintaining the effector organ. In the former kind of experiment the rat submaxillary gland was ligated so that it degenerated, leaving the nerve supply intact; it was then shown that the release of noradrenaline from these nerves on depolarisation by high potassium could be greatly potentiated with an alpha receptor blocking drug (phentolamine; Fig. 4.4B) (Fillinger et al., 1978), indicating the existence of alpha receptors on the nerve terminals. In the other kind of experiment 6-hydroxydopamine was used to denervate rat heart ventricles, and the extent of binding of 3H-dihydroergocryptine to alpha adrenoceptors in membranes from the heart ventricle ascertained. It was found that this binding was significantly decreased in the denervated heart, a decrease attributed to the loss of nerve terminals (Fig. 4.4C) (Story et al., 1979). 4.4 Identification of presynaptic adrenergic autoreceptors different from postsynaptic adrenergic receptors In 1974 it was still not clear whether the autoreceptor at sympathetic nerve terminals was to be taken as equivalent to the postsynaptic receptor or not. In that year Langer and his colleagues showed that the two receptors were different. First, Dubocovich and Langer carefully determined the concentration of phenoxybenzamine which just produced a decrease in the response of the spleen to sympathetic nerve stimulation due to postsynaptic alpha receptor blockade in relation to the
72
HISTORY OF THE SYNAPSE
concentration that produced an enhanced overflow of noradrenaline as a consequence of the blockade of presynaptic alpha receptors. They found that the drug was much more effective, by thirty fold, in blocking the postsynaptic receptors than the presynaptic receptors (Fig. 4.4D), and so concluded that it is possible that the pre- and postsynaptic alpha receptors are not identical (Dubocovich & Langer, 1974). One caveat in this work, mentioned above, is the action of phenoxybenzamine in blocking the uptake of noradrenaline, as well as that of presynaptic and postsynaptic receptors. In order to address this issue, Cubeddu, Barnes, Langer and Weiner showed that the concentration of phenoxybenzamine necessary to block the uptake of noradrenaline in the spleen was in excess of 0.3 µM, a concentration at which there is full postsynaptic alpha receptor blockade and a very substantial increase in the overflow of noradrenaline during nerve stimulation (Cubeddu et al., 1974) (Fig. 4.4E). Thus the differences in the potency of phenoxybenzamine for the presynaptic and postsynaptic receptors could not be attributed to complications arising from the effect of this drug on amine uptake. These observations led Langer to propose the existence of the alpha 2 adrenoceptor to demarcate the presynaptic adrenoeceptor from the postsynaptic adrenoceptor, or in his words: Perhaps the postsynaptic alpha-receptor that mediates the response of the receptor organ should be referred to as alpha 1, while the presynaptic alpha-receptor that regulates transmitter release should be called alpha 2 (Langer, 1793). Although it is now known that there are also presynaptic alpha 1 receptors (Starke et al., 1989) and indeed postsynaptic alpha 2 receptors, it seems that the alpha 2 receptor (Timmermans & van Zwieten, 1981) is the dominant type of adrenoceptor on nerve terminals. 4.5 Presynaptic adrenergic autoreceptors in the central nervous system The delineation of two classes of alpha adrenoceptors in the peripheral nervous system in 1974, one associated most often with the effector (alpha 1) and the other with the nerve terminal (alpha 2), raised the question as to whether these two classes exist in the central nervous system, and if so what is their location? The overflow of noradrenaline from slices of rat brain neocortex depolarised with moderately high potassium solutions was shown by Dismukes, De Boer and Mulder (1977) in 1977 to be decreased by alpha adrenoceptor agonists such as oxymetazoline and to be potentiated by alpha adrenoceptor antagonists like phentolamine (Fig. 4.5A). The action of oxymetazoline was especially effective in decreasing the overflow of noradrenaline if the calcium concentration was lowered (Fig. 4.5B), indicating that perhaps its action on the presynaptic alpha receptor involved a decrease in the calcium available for the exocytotic machinery in the nerve terminal. Although Dismukes and colleagues provided some evidence that the alpha receptor in their study was indeed presynaptic, as for example there was no affect of tetrodotoxin on the release process ruling out the participation of a sodium action potential neuronal loop in the alpha receptor mediated process, more direct proof that the alpha receptor in the cortex is presynaptic on nerve terminals was required. This was forthcoming with the study of Langen, Hogenboom and Mulder on superfused synaptosomes from rat cortex: noradrenaline could be released from these with high potassium stimulation, and alpha adrenoceptor agonists such as noradrenaline, oxymetazoline and clonidine inhibiting the release (Fig. 4.5C). This was taken by these authors to substantiate the hypothesis that the alpha receptors mediating a local negative feedback control of noradrenaline release are localized on the varicosities of central noradrenergic neurones (DeLangen et al., 1979). However the idea that the peripheral paradigm involving a presynaptic alpha 2 receptor and a postsynaptic alpha 1 receptor could be simply transferred to central adrenergic neurones turned out not to be correct. In the locus coeruleus there are noradrenergic cell bodies that possess alpha 2 receptors. Cedarabaum and Aghajanian used single unit recording from these neurones, coupled with microiontophoretic application of noradrenaline (NE) and the alpha 2 agonist clonidine (CLON) as well as of the alpha receptor antagonist piperoxane (PIP) and the beta adrenoceptor agonist isoproterenol (ISO) (Cedarbaum & Aghajanian, 1977). Their results showed that the adrenergic neurones of the locus coeruleus possess adrenoceptors that have the characteristics of alpha 2 adrenoceptors (Fig. 4.5D), a result later confirmed using the pure antagonist for central alpha 2 receptors, idazoxan (Fig. 4.5E) (Freedman et al., 1984). 4.6 Evidence that endogenous autoreceptor mechanisms exist Although the existence of presynaptic alpha 2 adrenoceptors was well established by the middle of the 1970s, it was not at all clear that these receptors mediated an endogenous effect of released noradrenaline, that is whether autoinhibition occurred physiologically. For example, Fuder, Muscholl and Spemann were able to show in 1983 that noradrenaline overflow from the perfused rat heart was inhibited by the alpha 2 adrenoceptor agonist oxymetazoline, as expected, but that the alpha 2 adrenoceptor antagonist yohimbine neither increased the overflow nor affected the heart rate or its development of tension
THE DISCOVERY OF ADRENALINE AND THE CONCEPT OF AUTORECEPTORS AT SYNAPSES
73
although it did antagonize the effects of oxymetazoline (Fuder et al., 1983). Thus the existence of presynaptic alpha 2 adrenoceptors on the sympathetic nerve terminals in the heart were confirmed, but it was suggested that these receptors did not participate in an endogenous negative feedback of released noradrenaline. Kalsner pointed out that noradrenaline and phenoxybenzamine had quite different effects on the overflow of tritiated noradrenaline from the vas deferens on nerve stimulation, with noradrenaline depressing the overflow by about 60% at different test frequencies (0.5 to 10 Hz) for 10 impulses, whereas phenoxybenzamine increased the overflow to a diminishing extent with increasing frequency over 10 impulses (Kalsner, 1980). Many of these caveats concerning the endogenous action of autoreceptors were explained away by the work of Storey, McCulloch, Standford-Starr and Read in 1981 who established the conditions that must prevail in order for the autoreceptor machinery to be engaged. Phentolamine had no effect on the release of tritium labelled noradrenaline from guinea-pig atria during a train of 4 impulses at 2Hz whereas that evoked by a train of 16 pulses at 2 Hz was increased by more than 2.5 fold. These observations suggested to them that noradrenaline does not act on the autoreceptors if the concentration is too low as at the beginning of the train and that a minimum interval is required between the beginning of a train and when the autoreceptor activation is just detectable (>1.5s, that is the duration of 4 pulses at 2 Hz). Furthermore the effects of phentolamine on the release of tritium labelled noradrenaline evoked by 4 impulses at 0.125 Hz and 2 Hz was not significantly altered, but that evoked by 4 impulses at 0.25 Hz, 0.5 Hz and 1Hz was significantly enhanced. The observations indicate that there is a minimum time interval for the effects of alpha 2 adrenoceptor activation to be detectable (namely >1.5s, that is the duration of 4 pulses at 2 Hz) and a limited period during which transmitter release evoked by one pulse can affect that evoked by subsequent pulses (namely <8s, the interval between two pulses at 0.125 Hz) (Storey et al., 1981). 4.7 The ionic basis of the action of alpha 2 adrenoceptors It was mentioned above that the action of oxymetazoline in lowering the release of noradrenaline from nerve terminals in the cortex was enhanced by decreasing the calcium concentration (Dismukes et al., 1977), suggesting that perhaps the mechanism of action of the presynaptic alpha receptor in decreasing transmitter release involves a decrease in the calcium influx responsible for the process of exocytosis of noradrenaline. This idea was subsequently supported by experiments in which the overflow of noradrenaline from slices of rat occipital cortex, induced by depolarisation with high potassium solutions, was greatly enhanced by blocking the alpha receptors with phentolamine (Fig. 4.5F); this potassium induced effect in the presence of phentolamine was greater than that due to the introduction of a calcium ionophore in the presence of the alpha receptor antagonist, indicating that voltage-dependent calcium channels are involved in the action of the alpha receptor (Gothert et al., 1979). It was not until 1989 that Elliott, Marsh and Brown obtained direct evidence for the proposition that the presynaptic alpha receptor in the peripheral nervous system (the alpha 2 receptor) was likely to exert its effects in decreasing transmitter release by depressing voltage-dependent calcium channels that mediate the exocytosis of noradrenaline release. These authors showed that they could obtain a calcium action potential in the cervical sympathetic nerve trunk to the superior cervical ganglion, treated with tetrodotoxin to block sodium currents and 4-aminopyridine to reduce the potassium currents, and that this calcium action potential could be reduced by alpha 2 receptor agonists, effects blocked by the alpha 2 antagonist phentolamine (Fig. 4.6A). As the reduction in the calcium action potential by the alpha receptor agonists was unaccompanied by hyperpolarization of the nerves, it was unlikely that the changes in the calcium channels were secondary to potential changes produced by a potassium dependent hyperpolarization of the nerves (Eliott et al., 1989). Investigations were also made in the 1970s of the ionic mechanisms involved in the action of alpha receptors found on cell bodies of adrenergic neurones. Catecholamines produce a hyperpolarization of the cell bodies of sympathetic ganglion cells (De Groat & Volle, 1966; Haefely, 1969), an effect that was shown to increase on removal of external calcium ions or following reduction of external potassium ions (Fig. 4.6B) (Brown & Caulfield, 1979), similar to that already reported for the effects of these ions on alpha-mediated inhibition of noradrenaline release (Dismukes et al., 1977; Marshall et al., 1978). In a direct test of the ionic basis of the hyperpolarization produced by adrenergic agonists of the alpha 2 receptor subtype in neurones, Williams, Henderson and North made intracellular recordings from catecholamine sensitive neurones in slices of the locus coeruleus and showed that the hyperpolarization produced by pressure ejection of clonidine or noradrenaline could be reversed to a depolarization on membrane polarisation to −110 mV (Fig. 4.6C): as this is close to the potassium equilibrium potential, the result points to the activation of alpha 2 adrenoceptors on cell bodies giving rise to an increase in a potassium conductance (Williams et al., 1985). The mechanisms that could link alpha 2 adrenoceptors to changes in either voltage-dependent calcium or potassium channels were first investigated in detail at the beginning of the 1980s. By this time it had been discovered that islet-activating protein (IAP), a pertussis toxin, could abolish receptor-mediated inhibition of cAMP accumulation in rat pancreatic islets (Katada & Ui, 1981). This occurred as a consequence of IAP inactivating an inhibitory GTP binding regulatory protein normally responsible for inhibiting adenylate cyclase (Cooper, 1982). IAP was shown to inhibit the effects of the alpha 2 adrenoceptor agonist clonidine on sympathetic nerve terminals in the vas deferens (Fig. 6D), suggesting that the presynaptic alpha 2
74
HISTORY OF THE SYNAPSE
Fig. 4.5. Presynaptic adrenergic autoreceptors in the central nervous system. (A) Overflow of noradrenaline from slices of rat brain neocortex is modulated by alpha adrenoceptor agonists and antagonists. Effects of oxymetazoline and phentolamine on 3H-NA (noradrenaline) release from brain cortical slices, induced by electrical pulses or various K+concentrations. After labelling by incubation with 3H-NA slices were superfused and exposed to one of the depolarizing stimuli, either in the absence (control) or presence of one of the drugs investigated. Data are expressed as means ±S.E.M. for the number of determinations in parentheses. Statistical significance of differences (vs. control): * P<0.001; ** P<0.01; n.s.: 0.05
THE DISCOVERY OF ADRENALINE AND THE CONCEPT OF AUTORECEPTORS AT SYNAPSES
75
Fig. 4.6. The ionic basis of the action of alpha 2 adrenoceptors acting as autoreceptors. (A) Calcium action potential in the cervical sympathetic trunk is reduced by alpha 2 adrenoceptor agonists. Effect of increasing concentrations of (a) phentolamine, (b) yohimbine and (c) prazosin on the inhibition of Ca-spikes produced by 10 µM (−) noradrenaline (solid rectangles below trace) and 10 µM adenosine (open rectangles below trace). (Three experiments). Taken from Elliott, Marsh & Brown, 1989. (B) Noradrenaline hyperpolarizes the cell bodies of sympathetic ganglion cells and this is modulated by changes in external potassium and calcium ions. Effects of changing (a) external K+ concentration and (b) external Ca2− concentration on the hyperpolarizing responses to 10 µM (−) noradrenaline. Responses in column (i) are controls in normal Krebs solution; responses in column (ii) and (iii) were obtained 30 min after solution changes. Taken from Brown & Caulfield, 1979. (C) Evidence that the hyperpolarization due to clonidine acting on catecholamine—sensitive neurons in slices of locus coeruleus is due to an increase in potassium conductance. Noradrenaline-induced outward current. A single-electrode voltage clamp was used to measure membrane currents induced by noradrenaline applied by pressure ejection (20 ms/35 kPa, at arrows), (a) Top trace is membrane voltage and bottom trace is membrane current. Noradrenaline was applied twice, and between applications the recording mode was changed from current clamp (zero applied current) to voltage clamp (holding potential −60 mV). Noradrenaline hyperpolarized the membrane by about 25 mV. The outward current produced by a second application was about 150 p(A). Note the identical time course of current and voltage responses, (b) Membrane currents recorded at the holding potentials indicated beside each trace (mV). In each of the eight traces, noradrenaline was applied by pressure as in (a). Arrows have been omitted in all but the first trace for the clarity. Note the progressive reduction and the eventual reversal of the noradrenaline current with membrane hyperpolarization. Taken from Williams, Henderson & North, 1985. (D) Pertussis toxin inhibits the effects of the alpha 2 adrenoceptor agonist clonidine on sympathetic nerve terminals. The influence of pertussis toxin (IAP) treatment on contractile responses to electrical stimulation in rat vas deferens. Control tensions and dose-response curves for clonidine were determined at the end of the first and second incubation periods by adding 5×10−9, 10−7 M clonidine to the organ bath. The electrical stimulation applied was 1 Hz, 1 ms at the maximal voltage at 7 min intervals. * Significantly different from controls, vehicle-treated P<0.002. Taken from Lai, Watanaber & Yoshida, 1983. (E) Pertussis toxin (IAP) enhances the electrically evoked release of noradrenaline and decreases the facilitatory effects of the alpha 2 autoreceptor antagonist yohimbine. Effect of IAP-pretreatment on the yohimbine induced facilitation of the electrically stimulated overflow of tritium from rabbit hippocampal slices incubated with 3H-noradrenaline. Slices were pretreated, incubated with 3H-noradrenaline, superfused and stimulated. Yohimbine (0.01, 0.1 and 1 µmol/1) was added 15 min prior to the second stimulation period. The effect of yohimbine is expressed as the ratio (S2/S1 of the overflow of tritium evoked by the two stimulation periods. The evoked overflow of tritium of the first stimulation period (S1, expressed as % of tissue tritium) was 3.01±0.24% (n=15) for the control slices (open columns) and 3.93 ±0.30% (n=12; P<0.0001 vs. control slices) for the IAP-treated slices (hatched columns). Means of the S2/S1 ratios ±SEM are given, numbers of experiments in the columns; significant differences between untreated and IAP-treated slices: ** p<0.01; **** p<0.0001. Taken from Allgaier, Feuerstein, Jackisch & Hertting, 1985.
adrenoceptor exert their effects on voltage-dependent ion channels through a pertussis-toxin sensitive GTP binding protein, that is a G protein (Lai et al., 1983). A direct test of the effects of pertussis toxin (IAP) on alpha 2 adrenoceptor mediated overflow of noradrenaline was made on hippocampal slices: IAP treatment of the hippocampus enhanced the electrically evoked release of noradrenaline and decreased the facilitatory effects of the alpha 2 adrenoceptor antagonist yohimbine as
76
HISTORY OF THE SYNAPSE
well as reducing the inhibitory effects of the alpha 2 adrenoceptor agonist clonidine (Fig. 4.6E) (Allgaler et al., 1985). These observations then pointed to an inhibitory guanine-nucleotide-binding protein of a presynaptically adenylate cyclase coupling the activation of the alpha 2 adrenoceptor to decreases in the opening of voltage-dependent channels. However at the end of the 1980s there seemed to be a number of difficulties in the way of this idea. One of these was that drugs which elevate cAMP levels generally have much smaller facilitatory effects on noradrenaline release than do alpha adrenoceptor antagonists (Starke et al., 1989). Furthermore, pertussis toxin treatment does not affect the action of alpha 2 adenoceptor agonists and antagonists on stimulation induced noradrenaline release in a large number of sympathetically innervated organs (Musgrave et al., 1987). This casts some doubt on the generality of the proposition that presynaptic alpha 2 adrenoceptors work through a pertussis toxin sensitive G protein. 4.8 Conclusion The pivotal year in autoreceptor research was 1971 when a number of observations established this idea for the adrenergic synapse. First, Kirpekar and Puig showed that the effects of adrenergic blocking drugs on the increased overflow of noradrenaline could not be attributed to these drugs preventing vasoconstriction (Kirpekar & Pulg, 1971). Second, Farnebo and Hamberger excluded the possibility that the increased overflow in the presence of these agents was due to their changing the rate of metabolism of noradrenaline (Farnebo & Hamberger, 1971). Finally, Starke showed that there was an increase in overflow of noradrenaline from the heart in the presence of alpha-adrenoceptor blockers, an organ that does not possess postjunctional alpha adrenoceptors (Starke, 1971). Then in 1974 Langer and his colleagues characterised the presynaptic alpha receptor as pharmacologically different to the postjunctional receptor and named it the alpha 2 receptor (Langer, 1974). The importance of this receptor in modulating the release of catecholamines from central as well as peripheral nerve terminals was established by the late 1970’s and the mechanism of its action delineated in the early 1980s. At this time the generality that all synapses possess autoreceptors, no matter what transmitter type utilised, was established.
5 The Discovery of Amino Acid Transmission at Synapses in the Central Nervous System
5.1 Introduction In the late 1950s a number of investigations were carried out concerning the effects of amino acids on the excitability of neurones in the cerebral cortex and spinal cord. Dusser de Barrenne introduced the method in 1933 of applying certain substances, such as strychnine, directly to the cortex of animals and observing the changes in behavior of the animal as well as of electrical discharges from other parts of the cortex. Hayashi refined this technique and in a series of papers beginning as far back as 1952 showed that acidic amino acids such as glutamate cause clonic convulsions when applied directly to the cortex of a dog (see Hayashi, 1952, 1956; for a review see Takagaki, 1996). He also showed, in several different mammalian species, that 4-aminobutyrate (gamma-aminobutyric acid, GABA) acted on the brain, and suggested that this might be the inhibitory transmitter (Hayashi & Nagai, 1952). This seemed particularly persuasive as both GABA and the activity of GABAsynthesizing enzyme (L-glutamate carboxy-1-lyase, glutamate decarboxylase, GAD, EC 4.1.1.15) had been known to be present at particularly high levels in the mammalian brain (Awapara et al., 1950; Roberts et al., 1950; Udenfriend, 1950; Roberts & Frankel, 1951). Subsequently both van Harreveld (1959) and Purpura et al. (1959) used the approach of applying substances to the cerebral cortex and observing the resultant changes in field potentials in order to produce a systematic study of the effects of amino acids on excitability. Van Harreveld (1959) discovered that extracts of rabbit pallium cause contractions of crustacean muscles in high dilutions and that the active principle was glutamic acid, suggesting that this may be an excitatory transmitter. This appeared to be especially the case given that Robbins (1959) had shown the excitant effects of glutamic acid on crustacean muscle, and concluded that ‘The possibility that L-glutamic acid is the E transmitter at this neuromuscular junction is compatible with the results’. Robbins (1959) also noted that GABA acted to inhibit muscle contractions in crustacean muscle, as others had noted (Blazemore et al., 1956), and that this effect was blocked by picrotoxin (Van der Kloot et al., 1958). Florey (1957) had already concluded that GABA was the transmitter at inhibitory synapses and Robbins concluded that ‘The most likely mode of action is that GABA combines with the I receptor in the post-junctional membrane’. The interpretation of data from similar experiments using mammalian brain is complex, given the temporal changes in the effectiveness of the topically applied substances due to their concomitant diffusion through the tissue over time and their removal by various processes in the cortex. Nevertheless, the observations on the interaction of GABA and picrotoxin on the cortex were particularly interesting and Purpura et al. (1957) showed that the action of GABA could be competitively antagonised by picrotoxin (Fig. 5.1 A). The introduction of the iontophoretic application of the amino acids directly to neurones in the spinal cord to test the effects of these agents on excitability by Curtis and Phillis in 1958, followed the development of the iontophoretic technique for the application of drugs to the neuromuscular junction by del Castillo and Katz (1955) as well as to spinal cord Renshaw cells by Curtis and Eccles (1958a, b). Subsequent introduction of multibarelled electrodes even allowed simultaneous iontophoretic application of a test substance and its antagonist onto single neurons in brains of living anaesthetised animals—a much more elegant way of studying amino acids than the simple topical application to the cerebral cortex. This initiated a new period, now forty years old, in the study of the role of amino acids in central nervous system excitability, which this chapter reviews. 5.2 Identification of excitant and depressant amino acids The new level of resolution in the study of the role of the amino acids introduced by the iontophoretic technique soon produced a great deal of information about their actions. This was to become invaluable in subsequent years as the identification of antagonists for the amino acids and for the endogenous transmitters became available. At the end of the 1950s it was clear that GABA had a depressant effect on motoneurones (Fig. 5.1B) and Renshaw cells (Curtis et al., 1959a)
78
HISTORY OF THE SYNAPSE
and that glutamic acid had an excitant effect on interneurones in the dorsal horn of the spinal cord (Fig. 5.1D; Curtis et al., 1970a). The subsequent systematic study of the effects of structurally related amino acids on motoneurones, Renshaw cells and dorsal horn interneurones resulted in the classic paper of Curtis and Watkins (1960) which gave in two tables the structure-activity relationships of both the excitatory amino acids (Fig. 5.2) and of the depressant amino acids (Fig. 5.3). Inspection of these two tables immediately reveals that L-glutamic and L-aspartic acids have a profound excitant action on interneurones and that N-methyl-DL-aspartic acid (NMDA) also produces a high level of excitability (Fig. 5.2). In contrast, GABA produces a profound level of depression as does glycine, although to a lesser extent (Fig. 5.3). Comparison between the related excitatory and depressant amino acids shows that there is a correlation between the potency of the excitatory activity of any acidic amino acid and that of the depressant activity of its -decarboxylation product. Curtis and Watkins (1960) showed that this is such that the relative activity of a particular excitant may be accurately predicted from a knowledge of the relative depressant activity of its -decarboxylation product and vice versa. It follows that glutamic and aspartic acids, the two most active of the excitatory agents, lead respectively to GABA and -alanine. What then is to be made of the action of these amino acids at the end of the 1950s in relation to the identification of the chemical transmitters of the central nervous system? The ubiquitous nature of the effects of the amino acids on a wide range of neurone types and in different parts of the central nervous system led to the belief that they could not be identified as specific chemical transmitters. Thus Curtis et al. (1959a) commented in their paper on the depression of spinal neurones by GABA that ‘All evidence indicates that these substances have a non-specific depressant action upon the whole surface membrane of neurones, both the chemically activated subsynaptic regions and the remaining electrically excited post-synaptic membrane’. The same authors claim in relation to their work on chemical excitation of spinal neurones at this time that: …these observations suggest that glutamic and aspartic acids have a non-specific excitatory action upon certain spinal neurones, a finding which is in accordance with earlier observations upon the non-specific depressant actions of the related monoamino-monocarboxylic acids’ (Curtis et al., 1959b) so that ‘It is considered that this action is nonspecific and unrelated to excitatory synaptic action (Curtis et al., 1960). In the first half of the 1960s the relative potency of excitatory amino acids was confirmed, with special emphasis on NMDA (Fig. 5.2F) and glutamic acid (Fig. 5.2B; Curtis & Watkins, 1960). However, the ubiquitous nature of their action was emphasised at this time, so that ‘Unlike choline esters, however, members of the glutamic acid group of excitants have actions which are nonspecific with regard to neuronal type; they excite many functionally different types of neurone, including both cholinoceptive and non-cholinoceptive cells in the spinal cord (Curtis et al., 1960), brain stem (Curtis & Koizumi, 1961), and cerebral cortex (Curtis & Watkins, 1961; Phillis & Krnjevic, 1961), as well as cells of the lateral geniculate nucleus which have been shown to be a site of action of drugs possessing an indole nucleus (Curtis & Davis, 1962). Furthermore, there is evidence (Curtis, 1962) that the equilibrium potential for the conductance change, induced in the membrane of spinal motoneurones by L-glutamic acid, differs from that associated with excitatory transmitter action.’ So that ‘The present results support the conclusion that an amino acid-surface interaction is involved and further elucidate the three-dimensional geometry of the site of action. The non-specificity of the excitatory action is in accord with earlier results and suggests that the site of action is a fundamental structural component of the membranes of all central neurones’ (Curtis & Watkins, 1963). 5.3 Glycine accepted as an inhibitory transmitter in the spinal cord The second half of the 1960s saw a radical break with the perspective that amino acids exerted non-specific actions. This was due to a detailed study of the depressant actions of glycine on spinal motoneurones. Thus Werman, Davidoff and Aprison (Werman et al., 1967) carried out experiments in which they recorded from lumbar motoneurones with an intracellular electrode and examined the effects of iontophoretic application of glycine. They showed that glycine always produced a hyperpolarization of the membrane consequent on an increase in membrane conductance. This blocked antidromic action potentials invading the soma and decreased the amplitude and time course of the excitatory postsynaptic potential (Fig. 5.4A; Werman et al., 1967). An observation of particular interest was that the amplitude of the inhibitory postsynaptic potential decreased with increasing hyperpolarization of the membrane due to the application of increasing amounts of glycine (Fig. 5.4B; Werman et al., 1967). These authors observed that the inhibitory postsynaptic potential reached its equilibrium potential (where it is of zero amplitude) as the hyperpolarization due to glycine reached its maximum value (Fig. 5.4B; Werman et al., 1967), suggesting that they both have identical equilibrium potentials. Furthermore, changes in the equilibrium potential brought about by changing the intracellular environment of the motoneurones recorded from produced similar changes in the amplitude and polarity of the inhibitory postsynaptic potential and the glycine potential. It had already been argued by Werman (1965) that this is a very stringent test of the identity of action of a particular transmitter candidate. Werman and his colleagues (1967) then concluded that
THE DISCOVERY OF AMINO ACID TRANSMISSION AT SYNAPSES IN THE CENTRAL NERVOUS SYSTEM
79
Fig. 5.1. GABA and bicuculline. (A) Interaction of GABA with picrotoxin and strychnine. Simultaneously recorded responses to independent stimulations of homologous points of right (upper trace) and left (lower trace) anterior suprasylvian gyrus, the cerebral hemispheres being disconnected by callosal section: 1, initial dendritic postsynaptic potentials; 2, 30 sec after application of GABA (3 drops, 10−2 w/bv) to each side. Picrotoxin (2 drops of 3×10−3 w/v solution) was then applied to right cortex and strychnine sulfate (2 drops of 5×10−3 w/v) to the left; 3, 20 sec later; 4, 1 min later, after another application of the drugs; 5, 2 min later and 30 sec after another application of GABA. Whereas the effects on the strychninized (left) cortex were minimal, marked antagonism to picrotoxin is seen; 6, 15 min after repeated washing of the cortex with Ringer’s solution. Recovery of previously inactive strychninized side is the more rapid, probably denoting persistent blockade of inhibitory synapses. (From Fig. 2 in Purpura et al., 1957). (B) Left-hand column of traces: Potentials recorded extracellularly near BST motoneurones in the L7 segment, A before, B during and C after GABA was passed from the electrode. The L6, L7 and S1 dorsal roots were divided and the BST nerve was stimulated at an intensity five times the threshold stimulus of the large or alpha motor axons. The antidromic spike potential of the small motoneurone is marked with an arrow. Calibration, 1 mV; time, msec. Three right-hand columns of traces: Responses of a single Renshaw cell to synaptic stimulation (A–C) and to iontophoretically applied acetylcholine (D–F) recorded from the central barrel of a five-barrel electrode. A, D, control responses; B, E, during application of GABA, using a cationic current of 50 mmA in B, 30 mmA in E; C, F, 10 sec after the cessation of the respective applications of GABA. Calibration, 0.5 mV for all records; time, 10 msec. (From Figs. 9 & 10 in Curtis et al., 1959). (C) Histograms of the probability of firing of a Purkinje cell located 350 mm beneath the surface of the cerebellar vermis. A computer, triggered once per 1.8 s, analysed the number of action potentials occurring in 128 intervals, each lasting 4 ms (the first 512 ms of each sweep), for 60 sweeps. Approximately 150 ms after the beginning of each sweep the cerebellar surface 2 mm along and 1 mm across the folium was stimulated electrically with a bipolar electrode (0.8 mA, 0.2 ms pulse, arrowed). Throughout the period of observation the mean firing frequency of this cell was maintained between 60 and 80 spikes per second by the electrophoretic injection of DL-homocysteate. a, Control; b, 11 min after the intravenous administration of bicuculline, 0.2 mg/kg; c, 2 min after a further dose of 0.2 mg/kg; d, 4 min after a further dose of 0.2 mg/kg. (From Fig. 2 in Curtis, et al., 1970(a)). (D) Effects of bicuculline and strychnine on the inhibition of a spinal Renshaw cell by glycine and GABA. The rate of firing of the neurone (ordinates) was maintained by the continuous electrophoretic ejection of DL-homocysteate (15nA). Glycine (5 nA) and GABA (8 nA) were
80
HISTORY OF THE SYNAPSE
THE DISCOVERY OF AMINO ACID TRANSMISSION AT SYNAPSES IN THE CENTRAL NERVOUS SYSTEM
81
The findings previously reported which indicate that glycine is concentrated in spinal interneurones (Aprison & Werman, 1965; Graham et al., 1967), and the present results which demonstrate the similarity between the actions of glycine and the inhibitory transmitters raise the strong possibility that glycine is an inhibitory transmitter in cat spinal cord. A critical test of this hypothesis came with the realisation that the actions of strychnine on the spinal cord, which had been shown by Bradley et al. (1953) to block inhibition of monosynaptic reflexes by the recurrent inhibitory pathway, suggested that it might be a specific antagonist of the actions of glycine. This seemed likely to be the case for there was no effect of strychnine on membrane potential changes due to the iontophoresis of GABA as there was for glycine (Fig. 5.4C; Curtis et al., 1967a, b). This showed that The failure of strychnine to affect the action of GABA indicates that glycine and GABA apparently interact with different postsynaptic receptors, as was previously inferred from a study of the relationships between structure and activity of excitant and depressant amino acids (Curtis et al., 1967b). However, these authors were still hesitant to claim that glycine was the inhibitory transmitter to interneurones and motoneurones of the spinal cord, which Werman and his colleagues had strongly suggested was the case. Curtis and his colleagues suggested that if strychnine merely reduces the potassium conductance at activated inhibitory synapses, these amino acids may not necessarily interact with the same receptor as the transmitter.’ But ‘the failure of strychnine to reduce the hyperpolarizing post-synaptic inhibition of certain supra-spinal neurones, which presumably also involves an increased permeability to potassium ions, however, suggests that the interfering effect on spinal inhibition is most probably at the transmitter receptor site (Curtis et al., 1967b). Thus by 1968 there was a growing consensus that glycine was the inhibitory transmitter to both interneurones (Fig. 5.4D) and to motoneurones (Fig. 5.4C) in the spinal cord. This was the first identification of an amino acid as a central transmitter. 5.4 The emergence of GABA as an inhibitory transmitter in the brain The identification of glycine as an inhibitory transmitter acted as a kind of psychological release on the possibility that the other major depressant amino acid, GABA, could also be such a transmitter rather than as previously supposed ‘that these substances have a non-specific depressant action upon the whole surface membrane of neurones’. Indeed the comment was made as early as 1967 that ‘The failure of strychnine to affect the action of GABA indicates that glycine and GABA apparently interact with different postsynaptic receptors’ (Curtis et al., 1967). So that if glycine is an inhibitory transmitter acting on its own receptor why not GABA? As we have seen, the inhibitory effect of GABA and GABA-like compounds applied topically to the surface of the cerebral cortex upon seizures produced by stimulation of the motor cortex (Hayashi, 1959), as well as the inhibition by GABA of the excitation due to stimulation of the suprasylvian gyrus (Fig. 5 .1A; Purpura et al., 1957), was well established. To what extent then could it be claimed that GABA was an inhibitory transmitter in the brain? It has already been noted that there was not much support for the idea at the end of the 1950s, with Curtis et al., (1959a) concluding from their iontophoresis studies that ‘All evidence indicates that these substances have a nonspecific depressant action’. Kuffler and Edwards (1958; as well as Edwards & Kuffler, 1959) introduced a systematic study of the effects of amino acids on relatively simple preparations, such as the crayfish and lobster stretch receptors, which allowed them to show a correlation between the structure of the neutral amino acids such as GABA and their relative depressant action. They were able to show that GABA had a depressant effect both on the crayfish receptor that possesses an inhibitory innervation as well as on the lobster that does not, which led them to the conclusion that GABA was unlikely to be the inhibitory transmitter, but rather to exert an action on receptors unrelated to the synapse type. In their words, Although GABA is found in the nervous system, the role of an inhibitory transmitter cannot be assigned to it on the basis of available evidence (Kuffler & Edwards, 1958). It was probably the failure to obtain a specific antagonist of GABA which held back the claim that it might be an inhibitory transmitter, despite the fact it was known that the iontophoresis of GABA onto neurones in Deiter’s nucleus and the cerebral cortex produced membrane potential changes similar to those that occur during synaptic inhibition to these cells (Obata et al.,
82
HISTORY OF THE SYNAPSE
THE DISCOVERY OF AMINO ACID TRANSMISSION AT SYNAPSES IN THE CENTRAL NERVOUS SYSTEM
83
Fig. 5.4. Glycine. (A) The effects of glycine on membrane potential, action potential, excitatory postsynaptic potential (EPSP), inhibitory postsynaptic potential (IPSP) and membrane conductance in L7 motoneurones. a1, Antidromic spike, top trace is iontophoretic current level before application of glycine; a2, two superimposed traces, 1 and 2 sec after the onset of 31 n.amp glycine; a3, 2 sec after onset of 90 n.amp glycine; a4, 12 sec after turning off glycine, b1, control EPSP; b2, during 155 n.amp of glycine; voltage calibration: 2 mV for both B and C; C1 control IPSP; c2, during 69 n.amp glycine, d1, Control conductance measure with 35 n.amp current pulse and antidromic spike; d2, during 189 n.amp glycine; current calibration refers to glycine iontophoretic current. Top trace in 1 and middle trace in 2. (From Fig. 1 in Werman et al., 1967). (B) Relationships of IPSP amplitude and membrane hyperpolarization to glycine dose in an L7 motoneurone. , IPSP; , polarization. (From Fig. 2 in Werman et al., 1967). (C) Potentials recorded intracellularly from an anterior tibial motoneurone (resting potential 63–68 mV) in response to the extracellular electrophoretic administration of glycine (30 n.amp) and GABA (60 n.amp); the lower traces indicate the appropriate electrophoretic current. The brief downward deflexions are IPSPs evoked by stimulating the ipsilateral sural nerve. a, b, Control. c, d. During the electrophoretic ejection of strychnine from another barrel of the five-barrel extracellular micropipette (20 n.amp, 2 molar solution of strychnine hydrochloride in 165 molar NaCl). e, f, 4 min after the termination of the strychnine ejection. Calibrations, 4 mV for the intracellular recording, negativity down; 200 n.amp for amino-acid currents. Time, 5 sec. (From Fig. 2 in Curtis et al., 1967b). (D) Comparison of the effect of strychnine on the postsynaptic inhibition and glycine depression of a Renshaw cell. Ordinate: The number of spikes elicited by a constant submaximal ventral root stimulus, the stimuli being repeated at a rate of approximately 50/min. Abscissae: time in minutes, the series a and b are continuous. Pressure was applied mechanically to the ipsilateral hind paw (arrow and middle bar) and glycine was administered electrophoretically (2.5 nA, lower bar) before, during and after the administration of strychnine (40, 60nA, upper bar) from a micropipette barrel which contained 2 mM strychnine hydrochloride in 165 mM NaCl. (From Fig. 2 in Curtis et al., 1968).
1967; Krnjevic & Schwartz, 1967). Furthermore, evidence from neurochemical experiments strongly indicated that the GABA synthesising enzyme GAD was associated with synaptic structures (Salganicoff & De Robertis, 1965; Fonnum, 1968) and several laboratories had already shown that GABA was released from the brain during the activity of inhibitory neurones (Obata & Takeda, 1969; Mitchell & Srinivasan, 1969).
84
HISTORY OF THE SYNAPSE
In the early 1970s Johnston and his colleagues made a study of several convulsant isoquinoline alkaloids which might combine with GABA receptors in much the same way as the convulsant strychnine combines with the glycine receptor. The consequence of this search was the identification of bicuculline as a specific antagonist of GABA. Bicuculline is a phthalide isoquinoline alkaloid found in Dicentra cucullaria and several Corydalis species. Its convulsant activity was described by Welch and Henderson (1934) and most of the chemical studies on bicuculline necessary for determining its molecular structure were performed by Manske (1933; stereochemistry: Blaha et al., 1964), studies which singled out bicuculline as a possible GABAantagonist. Intravenous administration of bicuculline blocked the excitation of Purkinje cells in the cerebellum (Fig. 5.1C) while the effects of GABA iontophoretically applied to spinal cord Renshaw cells was also blocked by bicuculline, leaving the effects of iontophoretically applied glycine on the same cells unaffected (Fig. 5.4D; Curtis et al., 1970a, b). In the conclusion to this pathfinding study, the authors commented that ‘These observations that bicuculline blocks the inhibitory effect of microelectrophoretically administered GABA on central neurones and reduces strychnine-resistant inhibition of cortical pyramidal and cerebellar Purkinje cells, provide for the first time decisive pharmacological evidence that, of the amino acids present in the mammalian nervous system, GABA is most likely to be the actual transmitter at these inhibitory synapses.’ A nice addendum to this story is that Hugh McLennan, who had worked since the early 1950’s in an attempt to identify the inhibitory transmitter in the brain (Florey & McLennan, 1955), and was aware of the possibility that GABA might be this transmitter as well as the transmitter at inhibitory junctions on the crayfish stretch receptor (McLennan, 1957), showed in the laboratory of David Curtis in 1970 that bicuculline blocked both the effects of applied GABA to the stretch receptors as well as the inhibitory effects of stimulating the inhibitory nerves to these cells (McLennan, 1970). Bicuculline then provided the definitive evidence for identifying GABA as the principal inhibitory transmitter. 5.5 L-Glutamate as a neurotransmitter: synaptic excitation, ion fluxes and neurotransmitter transporters It was agreed by 1970, with a few caveats (see Godfraind et al., 1970) that two amino acid inhibitory transmitters had been isolated in the central nervous system, namely glycine and GABA. Since these had been recognised in the 1950s as the principal depressant amino acids on central neurones, it was natural to suppose that the principal excitant amino acids identified in the 1950s, namely L-glutamate and L-aspartate, would prove to be the principal excitatory amino acid transmitters in the brain. L-Glutamate had been known to exist in brain at very high levels, a fact, however, which did not make it easier for L-glutamate to be accepted as a synaptic transmitter. There were other hurdles yet to be overcome. Thus Lucas et al., (1957) showed that application of L-glutamate onto the mammalian retina could cause lesions and this ‘neurotoxic’ nature of L-glutamate (as well as of several other acidic amino acids) received much publicity in late 1960s and early 1970s mainly through the work of Olney (Olney et al., 1971; 1972). Perhaps, it was felt that nature would not design excitatory synapses which would release large amounts of potentially neurotoxic substance into the extracellular space from where it could irreversibly damage the central neurons. There were additional problems, namely the ionic basis of amino acid action. In the case of GABA and glycine it was quickly accepted that the receptors were positively linked to chloride channels but a very different mechanism of action was proposed for the excitatory amino acids. Krnjevic (1970) suggested that the Lglutamate-evoked deplarisation could be triggered off merely by electrogenic transport of L-glutamate. This did not seem such a radical proposal at the time since Stern et al. (1949) had demonstrated that slices of mammalian brain in vitro could avidly take up glutamate probably via an active transport system, and amino acid transport was known to be electrogenic. Furthermore, Logan and Snyder (1971) demonstrated a very efficient Na+-dependent glutamate uptake in brain tissue. Almost simultaneously, however, similar Na+-dependent uptake systems specific, respectively, for glycine and GABA were reported from Iversen’s laboratory (Iversen & Neal, 1968; Iversen & Johnston, 1971; Johnston & Iversen, 1971). It was thought more probable that such uptake systems were ‘mopping up’ excess of released transmitter thus contributing to its ‘recycling’ and limiting its action in time and space. Later on, when specific substrates/inhibitors for GABA uptake had become available this hypothesis could be tested and backed by experimental evidence (Johnston and Krogsgaard-Larsen, 1976; for a review see Borden, 1996). In fact, continuation of this work eventually resulted not only in the first cloning of a neurotransmitter-specific transporter (GAT-1, Guastella et al., 1990; for a review see Borden, 1976) but lead to the development of at least one powerful inhibitor of GABA transport with a therapeutic potential (tiagabine, Leach & Brodie, 1998). The story of glutamate transport was, however, more complex. It was not difficult—by showing that several strongly excitatory amino acids were not taken up by the Na+-dependent ‘high affinity’ uptake system specific for acidic amino acids (Balcar & Johnston 1972)—to put in serious doubt the hypothesis of glutamate-evoked depolarisation via an electrogenic transport. In 1970s, however, several laboratories demonstrated the presence of the Na+-dependent ‘high affinity’ uptake system in glial cells (for reviews see Fagg & Lane, 1979; Schousboe, 1981; Fonnum, 1984). This helped to revive earlier suggestions about a glutamate-glutamine metabolic cycle (Berl et al., 1961): potentially toxic L-glutamate would be rapidly transported away from the synaptic cleft into glial cells, converted to harmless L-glutamine that would be passed, either passively or via a
THE DISCOVERY OF AMINO ACID TRANSMISSION AT SYNAPSES IN THE CENTRAL NERVOUS SYSTEM
85
Fig. 5.5. Excitatory amino-acids. (A) Response to iontophoretic application of ibotenic acid. Lower traces, depolarizations to ibotenic acid superimposed on hyperpolarizations of muscle-fibre. Ejection current illustrated in upper traces (a-d) and current for altering membrane potential illustrated in upper trace of a. a, Response of a Nl-fibre. b-d, Successive responses of a second Nl-fibre to repetitive application of ibotenic acid: ejection rate 1 pulse per 5 seconds. Note fall in amplitude of transient depolarization or ibotenic acid potential with repetition. This fibre showed some hyperpolarizating rectification. Calibrations same for b-d. (From Fig. 11 in Lea & Usherwood, 1973). (B) Effects on amino acid and synaptically evoked excitations. A neurone in VPL was excited by the electrophoretic application of DLH (20 nA) and glutamate (GL; 100 nA) and by electrical stimulation of the contralateral tibial nerve (0.1 msec, 3 V). The figure shows ratemeter records of the amino acid induced firings on the left, and on the right computed histograms show the summed evoked responses to 50 stimuli delivered at a frequency of 1/4 sec, which in this case usually consisted of two spikes. The stimulus artifacts are indicated by arrows, a, before; b, during application of GDEE, 160 nA for 6 min; c, 6 min after cessation of the GDEE current. (From Fig. 3 in Haldeman et al., 1972). (C) Rate-meter records showing selective effects of D- -aminoadipate (DaAA) on amino acid induced excitation of dorsal horn cells. In the upper record, responses to L-aspartate (Asp 30 nA) were reduced more than responses to L-glutamate (Glut 60 nA) by D AA 5 nA. In the lower record, responses to N-methyl-D-aspartate (NMDA 34 nA) were markedly reduced by D AA 10 nA while responses to kainate (KA 35 nA) were unaffected by this agent. (From Fig. 1 in Biscoe et al., 1977a). (D) Effects of , -diaminopimelic acid on DR-VRPs and ventral root depolarizations induced by different excitatory amino acids in isolated frog spinal cord, a: ventral root response to dorsal root stimulation (2 min) and to 80 sec applications of L-glutamate (1.3 mM) and Laspartate (1.5 mM). The bar over the record indicates the period of superfusion with 500 mM , -diaminopimelic acid ( , -DAP). The gap in the record represents 25 min. b: TTX-blocked spinal cord. Record shows ventral root depolarisations induced by 80 sec applications of kainate (KA, 10 mM) and NMDA (25 mM), before, during and 40 min after superfusion with 500 mM, , -diaminoopimelic acid (DAP). Calibration (both records): 1 mV, 10 min. (From Fig. 1 in Evans et al., 1978). (E) Order of sensitivity of excitatory amino acid induced responses to various antagonists, and relative effects of these antagonists on synaptic excitation. Antagonism was assessed as follows: in vivo, by the degree of depression of spike discharge frequency for cat spinal neurones excited by the various agonists or by electrical stimulation of a dorsal root (iontophoretic administration of both agonists and antagonists); in vitro, by the depression of agonist-induced or dorsal root-evoked motoneuronal depolarization, recorded from ventral roots of the frog isolated spinal cord (bath administration of agonists and antagonists). The difference in antagonist sensitivity of responses to different agonists increases with the distance between the agonists in the Table. The agonists in large type have been the most extensively studied. See Biscoe et al. 1978. (From Fig. 1. in Davies et al., 1979).
specific uptake system (Balcar & Johnston, 1975) back to glutamatergic neurons where it would be converted to L-glutamate and taken up by synaptic vesicles ready to be released upon stimulation (Naito & Ueda 1985; Roseth et al., 1995). It is interesting to note that, almost 20 years later, when the glutamate transporters were cloned and antibodies developed, at least the most abundant of them was shown to be present predominantly, if not exclusively in glial cells (Levy et al., 1993; Rothstein et al, 1994; Lehre et al., 1995; review: Robinson & Dowd, 1997, see also Mennerick et al, 1998; Meany et al., 1998). The story of electrogenic transport and glutamate action also had a happy ending: Takahashi et al., (1996) and Otis et al., (1997) demonstrated that a significant portion of currents associated with the glutamate action on the cerebellar Purkinje cells
86
HISTORY OF THE SYNAPSE
may be carried by a glutamate transporter (probably by EAAT 4, found in significant quantities only on Purkinje cells: it is non-thermodynamically coupled to Cl− flux, effectively acting as a Cl−-permeable channel, Fairman et al., 1995). 5.6 NMDA receptors: the first amino acid receptor identified at central excitatory synapses It was tacitly assumed, at least until early 1970s, that the excitatory amino acids acted on the same receptor (Davies et al., 1979). However, work on locust muscle pointed to the existence of at least two different types of excitatory amino acid receptors there (Lea & Usherwood, 1973): ibotenic acid was shown to act on glutamate receptors that are located in nonsynaptic regions whereas glutamic acid appeared to act on both these receptors as well as on those at the synaptic region (Fig. 5.5A). The subsequent emergence of selective antagonists for the action of excitant amino acids in the vertebrate nervous system also indicated that more than one receptor for these existed but at the time only very few compounds had been reported to antagonise the postsynaptic action of L-glutamate and none of them were either specific or very potent. Experiments using two of these compounds produced, however, interesting data leading the way towards a new, rapidly expanding area of studies—the nature and characteristics of glutamate receptors. It was McLennan and his colleagues who showed that two derivatives of glutamic acid, DL- -methylglutamate and Lglutamic acid diethyl ester, which do not themselves possess excitatory action on central neurones, block both the excitation of thalamic neurones produced by exogenous glutamate as well as the excitatory transmission in the thalamus due to electrical stimulation of hindlimb nerves (Fig. 5.5B; Haldeman et al., 1972). In 1977 this group went further and suggested that D- aminoadipate may be an excitatory amino acid antagonist (Hall et al., 1977), an idea that was soon tested in detail by Watkins and his colleagues. They showed that this compound indeed depressed both synaptically-evoked excitation in dorsal horn interneurones and Renshaw cells as well as the effects of iontophoresis of NMDA and L-aspartate on these neurones (Fig. 5.5C; Biscoe et al., 1977a, b). The observations pointed strongly to the idea that an excitatory amino acid, possibly Laspartate, was the transmitter. Subsequent study in frogs of the effects of , -diaminopimelic acid on the ventral root potentials arising from stimulation of the corresponding dorsal root, as well as on the potentials recorded in the ventral root in response to application of L-glutamate and L-aspartate showed that all of these were reduced by the agent (Fig. 5.5D; Evans et al., 1978). However , -diaminopimelic acid had a much greater blocking effect on the ventral root depolarisation’s produced by NMDA than by kainate, indicating the existence of two types of amino acid receptors (Fig. 5.5B; Davies et al., 1979). It was therefore realised at this time that agents such as D- -aminoadipate and , -diaminopimelic acid depress or abolish responses to NMDA without affecting the responses to kainate or quisqualate. The other compound found by Haldeman et al. (1972) to have antagonist affects on the excitation due to application of Lglutamate, namely L-glutamic acid diethyl ester (GDEE), was shown at this time to exert almost the reverse spectrum of antagonist actions on cat spinal neurones to that of the NMDA antagonists (McLennan and Lodge, 1979), so that while responses to L-glutamate were blocked by GDEE the NMDA-induced responses were unaffected. A surprising observation made at this time was that kainate-induced responses were unaffected by either GDEE or the NMDA antagonists. Taken together, the above results provided evidence for two or more excitant amino acid receptors. This proposition was given further credence by the discovery that DL-2-amino-4-phosphonobutyrate (2-APB) depresses responses to both kainate and NMDA to a large extent, whilst having a much lesser affect on the responses to the other excitant amino acids, such as Laspartate, L-glutamate and quisqualate. At the end of the 1970s Watkins was able to summarise the situation as far as the relative effects of antagonists on the responses of central neurones to excitatory amino acids by means of the diagram shown in Fig. 5E, which synthesizes the data discussed above. Davies et al. (1979) comment that From the evidence reported in recent publications (Biscoe et al, 1977, 1978; Evans et al., 1978; Davies et al., 1979; Lodge et al., 1978), it seems likely that these NMDA receptors are located at excitatory synapses operated by an amino acid transmitter. They go on to say, however, that The importance of non-NMDA receptors in synaptic function is less clear. The possibility that at least some of these are also synaptic receptors but of a different type from NMDA receptors cannot be excluded. Thus by the end of the 1970s NMDA receptors were taken to mediate excitation at some synapses in the central nervous system.
THE DISCOVERY OF AMINO ACID TRANSMISSION AT SYNAPSES IN THE CENTRAL NERVOUS SYSTEM
87
5.7 Non-NMDA receptors at excitatory synapses That more than one amino acid receptor type exists at synapses in the central nervous system became evident in the 1980s with the discovery of more specific antagonists. The NMDA receptor was established as a major amino acid receptor at excitatory synapses in the central nervous system. This came about with the recognition that magnesium ions powerfully antagonise the effects of applied NMDA but not that of kainate, L-glutamate or quisqualate (Fig. 5.6A; Ault et al., 1980). Furthermore, the Dform of 2-amino-5-phosphonovalerate (D-APV) was shown to be a highly potent antagonist of amino acid-induced excitation in the cat spinal cord through an action specifically on NMDA receptors (Fig. 5.6A; Davies & Watkins, 1982) and of synaptic action involving these receptors (Fig. 5.6B). In late 1970s and early 1980s studies of glutamate receptors received a powerful boost from the introduction of techniques using radiolabelled ligands. Initial studies using [14C]L-glutamate (Roberts, 1974) proved difficult but, eventually, experiments with [3H]L-glutamate and synthetic 3H-radiolabelled ligands made it possible to identify the major classes of receptors both in membrane-containing homogenates and in thaw-mounted sections of frozen brain (reviews: Foster & Fagg 1984; Cotman et al., 1987; Hansen & Krogsgaard-Larsen, 1990). Using this approach, the highest levels of NMDA-type of binding sites were found in the hippocampus and cortex (homogenates: Fig. 5.6C; Olverman et al., 1984, confirmed by autoradiography: Cotman et al., 1987; Beaton et al., 1992; Buller et al., 1994; Balcar et al., 1995a; 1995b), a discovery that was to have profound repercussions for our understanding of memory and of the development of synaptic connections (for a review see Collingridge & Lester, 1989). The identification and classification of non-NMDA receptors turned out to be even more complex than the characterisation of NMDA sites. Firstly, introduction of a more specific non-NMDA agonist (Krogsgaard-Larsen et al., 1980) made it possible to differentiate more clearly between ‘kainate’ and ‘AMPA’ receptors (for a review see Bettler & Mulle, 1995). Secondly, quisqualate, used earlier to describe a subclass of ionotropic non-NMDA receptors was shown to activate also glutamate ‘metabotropic’ receptors (Sladeczek et al., 1985), an entirely new class of G-protein linked sites of L-glutamate action (review: Pin & Duvoisin, 1995). The existence of non-NMDA receptors at excitatory synapses in the central nervous system, that Watkins and his colleagues had hesitated in claiming at the end of the 1970’s, was finally clearly established by the discovery by Drejer and Honoré in 1987 that the quinoxalinediones DNQX (6, 7-dinitro-quinoxaline-2, 3-dion=FG9041) and CNQX (6-nitro-7-cyanoquinoxaline-2, 3-dion=FG9065) are very potent antagonists of kainate and quisqualate induced responses but not of NMDA induced responses (Fig. 5.6D). Cloning of ionotropic glutamate receptors at the end of 1980s and in early 1990s (for a review see Hollmann & Heinemann, 1994; Lodge, 1997) opened new ways of testing hypotheses about specific roles of glutamate in central nervous function but it did not significantly change the system of receptor classification and has not so far lead to a major revision of the involvement of glutamate receptors in the mechanism of synaptic excitation as it was established by a combination of neurochemical and physiological approaches in earlier years. 5.8 GABA receptors Evolution of the field of GABA receptors followed much the same course as that of glutamate receptors. At least two major classes of ionotropic receptors have emerged: ‘classical’ bicuculline-sensitive GABAA and bicuculline-insensitive or GABAC, receptors (review: Johnston, 1996) The term ‘GABAB receptors’ is reserved for bicuculline-insensitive G-protein linked receptors responding to the synthetic GABA analogue baclofen (Bettler et al., 1998). Again, cloning of GABA receptors (Barnard et al., 1998; Bettler et al., 1998) confirmed the classification based on earlier experiments. 5.9 Conclusion The forty years of research in amino acid transmission in the central nervous system reviewed here, occurred largely before the revolution in receptor studies brought about by the introduction of molecular biology techniques. The very fortunate start to this story was the systematic study by Curtis and Watkins of the actions of different excitant and depressant amino acids, subsequent to the introduction of the iontophoresis technique in the late 1950s. This set a solid foundation in place for all subsequent studies. Next, the careful comparison by Werman and his colleagues in the late 1960s of the effects of the interaction of glycine with inhibitory postsynaptic potentials, made it very likely that glycine acted as an inhibitory transmitter. The fact that this action of glycine was blocked by strychnine, as is inhibitory transmission to spinal cord interneurones, then sealed the argument. Indeed, the acceptance of any substance as the putative transmitter at a particular
88
HISTORY OF THE SYNAPSE
Fig. 5.6. NMDA. (A) Selective depression by Mg2+ of frog motoneurone depolarizations induced by excitant amino acids. Records show ventral root potentials induced by upper series, 12 mM-kainate (KA) and 40 mM-NMDA (MA); lower series, 1 mM L-glutamate (G), 100 mM Lhomocysteate (H), 1–2 mM L-aspartate (A) and 5 mM-quisqualate (Q). Mg2+ (1 mM) was present in the medium during the periods represented by the bars above the records. The break in the lower record represents 20 min. Calibration: horizontal, 10 min; vertical 1 mV. (From Fig. 1 in Ault et al., 1980). (B) Differential potencies of APV optical isomers as antagonists of amino acid-induced and synaptic excitation of spinal neurones. Records A and B are from two different neurones. A: comparative effects of D-APV (5 and 10 nA) and L-APV (100 and 150 nA) on responses of the same dorsal horn neurone induced by quisqualate (Q, 40 nA), N-methyl-D-aspartate (NMDA, 55 nA) and kainate (KA, 40 nA). Note DAPV was approximately 15–30 times more effective than L-APV as a NMDA receptor antagonist on this cell. B: the effects of L- and DAPV on chemically and synaptically evoked excitation of the same Renshaw cell. The upper ratemeter record shows the more marked depression by 10 nA D-APV of responses to L-aspartate (Asp 50 nA) compared with those to L-glutamate (G1 40 nA) and acetylcholine (ACh 0 nA) and the lack of effect of 100 nA L-APV on these responses. The lower two records (a-d) are representative oscilloscope sweeps of the synaptic responses of the same cell evoked by a constant submaximal dorsal root (DR) and ventral root (VR) stimulus; before (a) 3 min following commencement (b); 1 min after termination of the ejection of D-APV 10 nA (c), and 3 min after commencing the ejection of L-APV 100 nA (d). The numbers below each record indicate the mean+S.E.M. number of spikes per response for 20 consecutive stimuli. (From Fig. 1 in Davies & Watkins, 1982). (C) Regional distribution of [3H]-D-AP5 binding sites. (From Fig. 2 in Olverman et al., 1984) (D) Time course of a dose-response experiment showing DNQX inhibition of kainate induced [3H]-GABA release from the cultured neurones. Cells were stimulated every 4 min with kainate in the absence or presence of DNQX. Open bars: mean basal release in the 4 min washout periods. Hatched bars: stimulated release during 30 s superfusions with 8 mM kainate. Concentrations of DNQX in the stimulation media are given in the figure. (From Fig. 1 in Drejer & Honoré, 1988).
synapse came to depend quite critically on the discovery of a potent antagonist that must work firstly on the synapse in question and then on the effects of the exogenous application of the substance in question. It was not then till the early 1970s,
THE DISCOVERY OF AMINO ACID TRANSMISSION AT SYNAPSES IN THE CENTRAL NERVOUS SYSTEM
89
with the discovery of bicuculline as a very potent inhibitor of GABA and of inhibitory transmission in the cerebellum and the cortex by Curtis, Duggan, Felix and Johnston, that GABA became accepted as another central transmitter. The introduction of a series of derivatives of L-glutamic acid that antagonise excitatory transmission in the thalamus as well as the actions of exogenous glutamate by Haldemann, Huffman, Marshall and McLennan in the early 1970s was a major discovery in the effort to obtain antagonists of central excitatory transmission. The use of these and similar compounds led to the definitive identification of the NMDA receptor at excitatory synapses and of the realisation that there are at least two excitatory amino acid receptors at these synapses by Davies et al. (1979). Finally, in 1987 Drejer and Honoré identified the quinoxalinediones as very potent inhibitors of this non-NMDA receptor. From the present perspective, forty years on from the discovery that amino acids can produce excitant and depressant effects on central neurones, it is clear that the dominant factor in accepting a substance as a transmitter is the development of suitable antagonists. This is, as it was in the 1950s, a formidable task even given the advances that have been made in molecular biology since that time.
6 Monoaminergic Synapses and Schizophrenia: the Discovery of Neuroleptics
6.1 Introduction This chapter reviews the history of neuroleptics, especially in the context of their effects in monoaminergic transmission. It is offered in the hope of adding some insights into how far we have progressed with the neuroleptics and to what extent there are plausible hypotheses to guide future developments. 6.2 Chlorpromazine In 1944 Paul Charpentier, at the Rhone-Poulenc Laboratories of Specia in Paris, first synthesised the amine derivatives of the phenothiazines (see Charpentier, 1947). Three of these were parent analogs of chlorpromazine. One was the compound R.P. 3276, which because of its very poor antihistamine activity was not proceeded with further. A second was promethazine, which was subsequently used extensively because of its very good antihistamine activity. The third was dietnazine, subsequently used in cases of Parkinson’s disease. In 1947 the French surgeon Laborit suggested that the peripheral and central autonomic nervous system had to be inhibited if the results of shock were to be controlled. To this end Laborit began in 1949 to use promethazine, which to his surprise both calmed and relaxed his postoperative patients, a phenomenon that had been noted by others who regarded it as a disturbing side effect to the action of the drug as an antihistamine. It is to Laborit that the accolades must go for first identifying in 1949 that the so called disturbing sedative side effects of promethazine were really due to its primary central action as a tranquilizer, allowing him to use it in this context for clinical purposes. Laborits interest in using a compound that might be even more effective than promethazine in alleviating the conditions of postoperative shock lead to his testing the phenothiazine chlorpromazine, another antihistamine-like compound, which was synthesized by Charpentier in 1950 (Deniker, 1970). The extraordinary antipsychotic power of chlorpromazine was then quickly revealed in fifteen publications from February to October, 1952 (Deniker, 1989). In the first publication on chlorpromazine by Laborit and his colleagues in February of that year, the effect of the drug to produce ‘disinterest’ is mentioned together with the possibility that this property might make it of psychiatric use (Laborit et al., 1952). The first use of chlorpromazine for specifically psychiatric purposes was by Hamon and his colleagues later in February of 1952, in which they reported the results of treating a manic patient with several different doses, although the results were difficult to interpret, because of the use of multiple drug treatments simultaneously. It was not until Delay and Deniker began a systematic study of the effects of chlorpromazine on 38 patients suffering from mania and acute psychosis in the period from May to July, as reported in six publications, that the extraordinary powers of the drug became evident (see Delay & Deniker, 1952; Delay et al., 1952). In these publications reference is made to the antipsychotic effects of chlorpromazine together with its failure both to treat cases of depression and to relieve the negative symptoms of schizophrenia. There have been substantial controversies concerning who should receive the accolades for the discovery of the antipsychotic properties of chlorpromazine. However there seems little doubt that Delay and Deniker should receive the major recognition in this regard (Piochet, 1996; Hippius, 1996; Garattini, 1996). In 1931 Sen and Bose first used Rauwolfia alkaloids in Indian psychiatric medicine. In 1933 Chopra wrote in Indigenous Drugs of India that ‘Rauwolfia recently attained prominence as a remedy for insomnia, hypochondria etc, the hypnotic action appears to have been known to the poorer classes in Bihar [who use it to put] children to sleep…the drug is sold as ‘pagal-kadawa [insanity specific]’. In 1949 Vakil alerted Western medicine to reserpine with an article in the British Heart Journal. In 1952 reserpine was isolated from the root of Rauwolfia serpentina by Muller, Schlitter and Bein (1952). The similarities between the effects of the alkaloid of Rauwolfia serpentina (reserpine) on mental patients and that of the phenothiazine chlorpromazine was evident to Deniker and Delay (Delay et al., 1954). That such drugs as different as a phenothiazine and an
MONOAMINERGIC SYNAPSES AND SCHIZOPHRENIA: THE DISCOVERY OF NEUROLEPTICS
91
alkaloid should have similar effects on the mental life of patients suggested to them that they must have something in common, which they named ‘neuroleptic’ (from the Greek, ‘which takes the nerve’; Delay & Deniker, 1957; Deniker, 1990). 6.3 Haloperidol By 1957 only chlorpromazine and some of the phenothiazine derivatives related to it together with reserpine were known to be neuroleptics. In that year Paul Janssen discovered a propiophenone with a high morphine-like potency. He then set out to make systematic structural changes in the molecule in order to enhance its potency. The lengthening of the two-carbon chains linking the basic nitrogen to the ketonic carbon atom produced a butyrophenone which when injected into mice produced behavioral changes symptomatic of both morphine and chlorpromazine. Thus the mice at first showed morphine-like excitement, dilation of the pupils and insensitivity to noxious stimuli; they then became progressively calm, sedated and slightly catatonic (Janssen, 1970). Related molecules were then synthesised with the intention of removing the morphine-like effect and increasing the neuroleptic effect. To this end the ester moiety was replaced with a tertiary alcohol group, producing a butyrophenone devoid of morphine-like effects but maintaining all its chlorpromazine-like properties. In 1958 certain aromatic substituents on the two aromatic rings were made that produced a compound which was many times more active than chlorpromazine as well as longer lasting and devoid of antiadrenergic effects. It was by far the most powerful neuroleptic, and was called haloperidol (Janssen, 1970). The clinical trials carried out in 1958 in the psychiatric clinic of Liege University (by Divry et al., 1959) showed that haloperidol was the neuroleptic to be used in cases of hallucinations, manic syndromes, paranoia and delusions. As Divry et al. (1 959) reported: Following the experimentation of a new drug, R1625 or haloperidol (which is not a penothiazine: fluoro-chloro-phenylhydroxy-piperidmo-butyrphenone), the authors have administered it orally to fifty patients. The medium dose is only 7. 5–15 mg, given uninterruptedly or not. Its neuroleptic properties are powerful: the typical patient shows a marked Parkinsonism with akinesia, hypertonia, asthenia-aboulia, emotional inhibition of great therapeutic interest. Twenty eight patients showed the syndrome, eleven of which were in an extremely marked state. Eight others had tremor, one of which was especially spectacular. Nine other subjects showed neuroleptic effects without Parkinsonism. When the treatment is stopped, all the effects of the drug regress spontaneously in delays between one to thirty days according to the intensity. Neuroleptic properties are indicated for states of agitation of any etiology. The results are excellent in manic psychoses, good in melancholia (sometimes even superior to electro-shock treatment), contradictory in paranoia, favorable in organic diseases (agitated and thymopathic patients) excellent in psychopathic impulsions and aggressiveness. It easily blocks, by Parkinsonism, choreic hypercinesia.’ 6.4 The dopamine hypothesis for neuroleptics This hypothesis developed as a consequence of the study of how reserpine might act as a neuroleptic. Twarog and Page (1953) as well as Gaddum and his colleagues (Amin et al., 1954) first detected serotonin in nerve tissue and the brain. The possibility that serotonin might have a role in the generation of mental phenomena was indicated by its capacity to block the hallucinogenic effects of LSD (Gaddum, 1954). The discovery was then made in 1955 by Brodie that if you give reserpine to an animal serotonin disappears from both the peripheral nervous system as well as from the brain. In 1954 Marthe Vogt described the distribution of noradrenaline and adrenaline in the brain (Vogt, 1954). Shortly after that Carlsson and his colleagues showed that reserpine could deplete monoamines from both peripheral organs and the brain (Carlsson & Hillarp, 1956), leading to a failure of aminergic transmission (Bertler et al, 1956). The possibility was then entertained that the antipsychotic action of reserpine involved its capacity to deplete monoamines. This hypothesis seemed particularly attractive as delivery of the precursor L-Dopa, reversed the behavioral effects of reserpine. However delivery of a precursor for serotonin, namely 5-hydroxytryptophan, did not (Carlsson et al., 1957). This indicated that reserpine depletion of the catecholamines was important for the behavioral effects. The fact that after such treatment there was no increase in adrenaline or noradrenaline but there was in the precursor dopamine pointed to this substance as being an agonist in the brain. Techniques were therefore developed for obtaining quantitative measurements of dopamine in the brain (Carlsson et al., 1958). It was soon shown that the correlation with the return of the normal behavioral response following reserpine treatment and the introduction of the precursors was the reappearance of dopamine in the brain rather than that of adrenaline or noradrenaline. With the discovery that dopamine might be a transmitter in the brain it was natural to turn to the possibility that chlorpromazine and haloperidol might act on dopamine transmission, given that these together with reserpine, were the main
92
HISTORY OF THE SYNAPSE
neuroleptics at this time. To test this possibility, Carlsson and Lindqvist (1963) as well as Anden et al. (1964) showed that both the penothiazine and the butyrophenone accelerated the turnover of dopamine and noradrenaline as indicated by the accumulation of the O-methylated catecholamine metabolites normetanephrine and 3-methoxytyramine (Figs. 6.1 A and 6.1B). This accelerated turnover of both dopamine and noradrenaline was not accompanied by any changes in their steady-state levels. As it was also known that both chlorpromazine and haloperidol antagonised the behavioral effects of L-Dopa, it was proposed that these neuroleptics block the postsynaptic monoamine receptors and by this means alter a negative feedback mechanism which controls the synthesis, release and metabolism of the monoamines (Carlsson & Lindqvist, 1963). Direct evidence for the blockade of dopamine receptors by the neuroleptics was obtained by Rossum (1966) who showed that the fall in blood pressure in the cat due to infusion of dopamine in the presence of an alpha-receptor blocker such as yohimbine was abolished by the neuroleptic spiramide (Fig. 6.1C). Subsequent studies showed that whilst chlorpromazine and haloperidol had actions on both dopaminergic and adrenergic receptors, many other neuroleptics did not block the adrenergic receptors at all, although they did the dopaminergic receptors. The blockade of dopaminergic receptors was then established as an essential mode of action of the neuroleptics in general (Anden et al., 1970; Nyback & Sedvall, 1970; Carlsson, 1978). 6.5 Dopaminergic projections in the brain The establishment of dopamine blocking effects of neuroleptics encouraged research programs for determining the distribution of the dopaminergic projections in the brain. This was greatly aided by the introduction of fluorescent techniques for the identification of catecholamine containing nerve terminals (Anden et al., 1966). Two main dopaminergic systems were shown to be present. One included projection from the ventral tegmentum in the mesencephalon to the frontal lobes and cingulate cortex (referred to as the mesocortical system), as well as to the hippocampus and amygdala (referred to as the mesolimbic system; Fig. 6.2A and 6.2B). The other dopaminergic system was detected projecting from the substantia nigra to the striatum (referred to as the nigrostriatal system; Fig. 6.2C). Malfunction of the nigrostriatal system gives rise to Parkinson’s syndrome, so the extrapyramidal symptomology of the neuroleptics used at that time could be explained by their blockade of this projection. It was further suggested in the 1960s that blocking the mesolimbic system, involving the dopaminergic projections to the amygdala, could explain the changes in mood and behavour following the administration of neuroleptics. Such changes involve inappropriate facial expressions associated with different emotions. Blocking the mesocortical dopaminergic system was then taken to be responsible for the therapeutic effects of the drug on the mental life of the patient. 6.6 Identification of the D1-like and D2-like dopamine receptors With the establishment of the dopamine hypothesis in the 1960s, following discovery of the neuroleptics in the 1950s, more specific tests of the action of the neuroleptics on dopamine receptors were devised. These were mostly formulated in the 1970s as appropriate techniques became available. Given that the quantitative methods for measuring catecholamines devised in the 1960s showed that the striatum received a rich dopaminergic projection from the substantia nigra (Fig. 6.2C), attempts were made to use homogenates and membranes of striatial tissue for the study of dopaminergic receptors. Libet had already established that dopamine was a transmitter in sympathetic ganglia (Libet & Tosaka, 1970) and soon afterwards it was shown that the action of dopamine in these ganglia was to elevate adenylate cyclase (Kebabian & Greengard, 1971). Having established the technique in the peripheral nervous system, Greengard and his colleagues took homogenates of the caudate nucleus of the striatum and showed that these also have their adenosine 3 :5 -cyclic monophosphate (cAMP) elevated by dopamine (Fig. 6.3D; Kebabian et al., 1972). Such stimulatory effects of dopamine were blocked by low concentrations of either haloperidol (Fig. 6.3D) or chlorpromazine (Fig. 6.3E). It was therefore established that dopamine acts to elevate cAMP. However haloperidol, which is about 100 times more effective clinically than chlorpromazine, is weaker than chlorpromazine when it comes to blocking the dopamine-sensitive elevation of adenylate cyclase (Clement-Cormier et al., 1974). Some other explanation was therefore sought for the action of these neuroleptics on the dopamine synapses in the striatum. The main difficulty in obtaining direct evidence for the blocking effects of neuroleptics on dopaminergic receptors arose because these drugs are highly-surface reactive and fat soluble. This gives them a high nonspecific solubility in membranes and so prevents the identification of specific binding sites for neuroleptics. The problem was first overcome with the discovery of a neuroleptic called butaclamol that possessed both an active (+) and an inactive (−) stereoisomer (Bruderlein & Humber, 1975). In 1975 the groups of Seeman on the one hand and that of Snyder on the other took advantage of this discovery to report observations on the direct effects of dopamine and its neuroleptic antagonists on isolated membranes from
MONOAMINERGIC SYNAPSES AND SCHIZOPHRENIA: THE DISCOVERY OF NEUROLEPTICS
93
Fig. 6.1. Discovery of the monoaminergic hypothesis of action of chlorpromazine and haloperidol. (A) Table showing that both chlorpromazine and haloperidol enhance the accumulation of the O-methylated catecholamine metabolites normetanephrine and 3-methoxytyramine in the mouse brain after treatment with a monoamine oxidaze inhibitor (the values are means ± standard errors of the mean, expressed in ug/g brain; figures in brackets indicate the number of experiments; from Table 1 in Carlsson and Lindqvist, 1963). (B) graphs showing the effects of chlorpromazine (5mg/kg i.v.) and haloperidol (0.5 mg/kg i.v.) on the levels of dopamine metabolites 3,4-dihydroxyphenyl-acetic acid (DOPAC; dotted line) and homovanillic acid (HVA; dashed line) at different times after administration of the drugs in the rabbit corpus striatum; the continuous line is the cumulative effect (from Figs. 1 and 2 in Anden, Roos and Werdinus, 1964). (C) the effect of dopamine on cat blood pressure and heart rate: this causes a depressor effect, which is blocked in the presence of the neuroleptic spiramide (from Fig. 1 in Rossum, 1966).
the striatum. In these experiments they showed that the binding of [3H] dopamine to the striatial membranes was inhibited by the active stereoisomer of butaclamol but not by the inactive stereoisomer (Fig. 6.3C) and that haloperidol had much the same effect as the active stereoisomer (Fig. 6.3A). This was also the case for the penothiazine neuroleptics (Fig. 6.3B). In these studies by the Seeman and Snyder groups the penothiazines displayed an affinity for the dopamine receptors that could be correlated with their clinical potencies, unlike that of their affinities for the dopamine-sensitive adenylate cyclase system. In a very revealing graph, Seeman et al. (1975) plotted the IC50 for the ability of the various neuroleptics to block the binding of [3H] haloperidol or of [3H] dopamine to rat striatal membranes against the average clinical doses used to control
94
HISTORY OF THE SYNAPSE
Fig. 6.2. The distribution of dopaminergic nerves and their terminals in the brain. (A) Diagram of the three-dimensional relationship between the parts of the brain that contain dopaminergic neurons and their terminals. The ventral tegmentum and the substantia nigra in the mid-brain contain the cell bodies of these neurons. Those in the ventral tegmentum make a dense projection to the frontal lobes of the cortex as well as to the cingulate cortex (not shown here as it is a deep structure shown best in B and C below) and to the hippocampus (another deep structure, see B and C). The neurons in the substantia nigra project exclusively to the basal ganglia where they terminate in the different sets of neurons that comprise this structure, namely the caudate nucleus, the putamen, the globus pallidus and the lateral medial neurons. (B) A midline view of the distribution of dopaminergic nerve terminals from the ventral tegmentum. Two main nerve pathways are shown, one of which possesses terminals in the frontal lobes as well as to a lesser extent over the rest of the cortex. The other pathway possesses terminals in the cingulate cortex, above the corpus collusum group of axons that join the two hemispheres; this pathway extends around the cingulum to eventually reach the hippocampus. (C) Similar to B, but showing in this case the projection from the substantia nigra to the basal ganglia. PET scanning can be used to examine the regulatory role of dopamine on cortical function. There is an impaired cognitive activation of the cingulate cortex in schizophrenics performing a verbal task involving word repetition. PET scanning shows that an agonist for dopamine receptors such as apormophine substantially improves the function of the cingulate cortex. The results suggest that in schizophrenic patients there is a failure of a normal dopaminergic modulation of neurons in the cingulate cortex. Because of the widespread connections of the cingulate cortex this could have repercussions for the integration of activity for relatively remote regions of the cortex that are connected via the cingulate cortex (from Bennett, 1997).
schizophrenia (Fig. 6.4A). The results revealed a close correlation between the clinical dose of either the penothiazines or the butyrophenones and that of their ability to block receptors. This is an observation that will be returned to again below.
MONOAMINERGIC SYNAPSES AND SCHIZOPHRENIA: THE DISCOVERY OF NEUROLEPTICS
95
6.7 Determination of different classes of dopamine receptors Kebabian and his colleagues first noticed, in their assays involving homogenates of the caudate nucleus, that the effects of neuroleptics on the dopamine-enhanced increase in adenylate cyclase could be divided into two classes according to their receptor-effector coupling; these two classes were designated D1 and D2 (Figs. 6.5 and 6.6; Kebabian & Calne, 1979; Stoof & Kebabian, 1981; see also Roufogalis et al., 1976; Spano et al., 1978). Subsequent pharmacological studies showed that the D1 receptors responded to dopaminergic agonists and antagonists of the benzazepine class (such as SCH-23390 and SKF-38526; Fig. 6.5B; Seeman & Niznik, 1990) and the D2 receptors to the penothiazines (chlorpromazine), the butyrophenones (haloperidol, spiperone), and the subsitituted benzamides (raclopride, sulpiride; see Seeman, 1980; Niznik, 1987; Niznik et al., 1989). In the late 1980s techniques in molecular biology led to the cloning of the dopamine receptors from rat thus giving substance to the claim that distinct classes of dopamine receptors exist (Bunzow et al., 1988). Using a beta2-adrenergic receptor gene as a hybridization probe, related genes including a cDNA for the D2 receptor were isolated; the latter was called RGB-2. Its amino acid sequence shows that the receptor belongs to the G-protein-coupled receptor family. The distribution of its mRNA parallels that of the D2 receptor and it has a similar pharmacological profile when its cDNA is transfected into mouse fibroblast cells (Seeman, 1981; 1992). Thus the antagonist affinities were found to be: Dopamine D1 receptor antagonists such as SCH23390 and selective serotonin receptor antagonists like ketanserin showed little antagonism of [3H] spiperone binding to the cloned D2 receptor. Subsequent cloning and expression of a D2 receptor showed that its effector coupling is mediated through a Gi-protein or Go-protein (Neve et al., 1989). The gene for the D2 receptor has a sequence identity of 48% with the human alpha 2 receptor, 39% with the hamster beta2 adrenergic receptor, and 27% with the porcine M1 receptor. The binding of the D2 ligand [3H] spiperone to membranes prepared from cells transfected with the cDNA of RGB-2 and that of rat striatal membranes was found to be comparable. This binding to the membranes of the transfected cells was inhibited by a number of D2 receptor antagonists including haloperidol and (+) butaclamol as well as sulpiride. This initial cloning effort led in the early 1990s to the discovery of three new receptor types, namely D3 (Giros et al, 1989; Sokoloff et al. 1990), D4 (O’Malley et al., 1992; Van Tol et al., 1991) and D5 (Grandy et al., 1991; Sunahara et al., 1991; Weinshank et al., 1991). In addition gene variants of these five receptor subtypes have been observed. The dopamine D2 receptor exists in alternate splice forms in rat (Dal Toso et al., 1989; Giros et al., 1989) and humans (Dal Toso et al., 1989; Grandy et al., 1989; Selbie et al., 1989). These five dopamine receptor subtypes fall into one of the two receptor classes originally defined by Kebabian and his colleagues. These will be referred to as D1-like and D2-like receptors. Dopamine D5 receptors have similar characteristics to the dopamine D1 receptors and so belong to the D1-like class while dopamine D3 and dopamine D4 receptors are similar to the dopamine D2 receptors and so belong to the D2-like class (Fig. 6.6). Recently, mice have been produced that do not possess the D2 receptor, using homologous recombination (Baik et al., 1995). The striatum and nucleus accumbens is devoid of the D2 receptor, as shown in autoradiographic studies. These mice show greatly reduced spontaneous movements as well as ability to maintain their balance without falling from a slowly rotating rod; this is the case for both heterozygous and homozygous animals. The phenotype of these D2 deficient mice is like that of the lack of spontaneous movement and abnormal gait in the extrapyramidal symptoms of Parkinsons’s disease. It will be of great interest to see if behavioral syndromes associated with psychotic disorders will be capable of analysis in D2 and other dopamine receptor deficient animals. 6.8 Clozapine In 1958 Eichenberger and his colleagues began a research program involving determination of the pharma-cological and antipsychotic properties of newly synthesised tricyclic compounds. Clozapine was amongst the 1900 compounds synthesised, for which the following pharmacological description was offered in 1961 (see Schmutz & Eichenberger, 1982 and Fig. 6.7): Compound HF-1854, like chlorpromazine, has a central sedative action and strong antagonistic effects against adrenaline, noradrenaline, acetylcholine and histamine. Like chlorpromazine, it inhibits the effects of sympathetic stimulation without, however, producing ganglioplegic effects. HF-1854 differs from chlorpromazine in particular on account of its very strong inhibition of the pain reaction in mouse and rabbit. Moreover, no real catalepsy such as that produced by chlorpromazine was seen with HF-1854. The substance may be classified as a neuroleptic agent with
96
HISTORY OF THE SYNAPSE
Fig. 6.3. The effect of neuroleptics on the binding of dopamine to receptors and on adenosine 3 :5 -cyclic monophosphate (cAMP). (A) The effects of the neuroleptic butaclamol, which has both an active (+) and an inactive (−) stereoisomer, on the binding of [3H] dopamine (left) and [3H] haloperidol (right) to a fraction from the rat striatum; the stereospecific difference gives the difference in the amount of ligand bound in the presence of (−)-butaclamol minus that in the presence of (+)-butaclamol (from Fig. 2 in Seeman et al., 1975). (B) and (C) Competition of the penothiazines (B) and stereoisomers of butaclamol and of dopamine (C) for [3H] dopamine binding sites on the membranes of rat corpus striatum cells (from Figs. 3 and 5 in Burt et al., 1975). (D) and (E) effect of haloperidol (D) and chlorpromazine (E) in the absence (− dopamine) and presence (+ dopamine) of 40 uM dopamine on adenylate cyclase activity in homogenates of rat caudate nucleus; in the absence of added drugs the level of cAMP activity was 26.43 pmol; the cAMP accumulation is expressed as the increase above this basal level (from Fig. 3 and 4 in Kebabian et al., 1972).
strong analgesic, parasympatholytic and sympatholytic actions. In man, the central neuroleptic effects may be expected to appear with lower doses than those affecting autonomic functions, but autonomie effects should be borne in mind as possible side effects. In the normal cat, only the higher dose levels produced autonomic effects alongside the tranquilizing action. We propose that the compound be tested in man. In 1966 Gross and Langner in Vienna carried out clinical trials with clozapine on 28 schizophrenic patients. The antipsychotic effects of the neuroleptic were judged as very good or good in 21 and questionable in 2 patients. In particular the absence of tardive dyskinesia and Parkinsonian-like side effects was noted, making this an atypical antipsychotic. Clozapine was not
MONOAMINERGIC SYNAPSES AND SCHIZOPHRENIA: THE DISCOVERY OF NEUROLEPTICS
97
Fig. 6.4. Relationship between the binding of neuroleptics to dopamine receptors and their clinically effective free drug concentration. (A) Relationship between the IC50 values for the different neuroleptics shown in blocking the binding of [3H] haloperidol and [3H] dopamine (as indicated by the filled circles and filled diamonds respectively) to rat striatal fractions and the average clinical dose (in mg/ day) of the neuroleptics in controlling schizophrenia; the continuous line gives the one to one relationship between the blocking concentration and the clinical dose; the dashed line indicates the presynaptic blocking action of the neuroleptics on stimulated release of [3H] dopamine from rat striatal slices (from Fig. 5 in Seeman et al., 1975). (B) Relationship between the clinically effective dose of different neuroleptics and their dissociation constant (KD) for inhibiting the binding of [3H] spiperone to the striatum or anterior pituitary tissues; the arrow is for clozapine, when the dissociation constant of D4 is used. The individual points in the ellipses indicates the results from different laboratories (from Seeman, 1992). (C), (D) and (E) Relationship between the dissociation constants (Ki) of the different neuroleptics shown for the cloned D2, D3 and D4 receptors and the therapeutic concentrations measured in the water phase of plasma of the spinal fluid of schizophrenic patients; the lines show the percentage of receptors (50% or 75%) that are occupied by the drugs under therapeutic conditions; the points in the ellipses give the range of observations; such drugs as the banzamides and molindone do not bind to D4 receptors at all as is indicated by their being encircled at the top of the figure for the D4 receptor binding (from Fig. 1 in Seeman & van Tol, 1994).
continued in the remaining 5 patients either because of lack of therapeutic effect or because of side effects. Subsequently Skarsfeldt (1988) showed that clozapine preferentially inhibits the mesolimbic dopamine neurons after 21 days of treatment. This is in contrast to the effects of a typical neuroleptic such as haloperidol, which inhibits both mesolimbic and nigro-striatal dopamine neurons to the same extent. The lack of Parkinsonian-like side effects of the atypical neuroleptics, probably arises because they do not block dopamine receptors in the striatum, in which failure of dopaminergic transmission is known to give rise to the Parkinsonian syndrome, as noted earlier. There is now no doubt that clozapine does have distinct advantages in the
98
HISTORY OF THE SYNAPSE
Fig. 6.5. The first distinction between D1-like and D2-like receptors by Kebabian and Calne (1979). (A) Different types of dopamine receptors in the nigro-striatal pathway. Autoreceptors are found on the terminals of dopaminergic neurons that control the rate of action potential firing in the substantia nigra (site 1) or tyrosine hydrozylase activity (site 2). The cortical neurons that project to the striatum possess sites that bind radiolabelled haloperidol and these sites have been identified as dopamine receptors (site 3). Neurons within the caudate of the striatum possess dopamine-sensitive adenylyl cyclase activity (site 4). Terminals of the striatal neurons that project to the substantia nigra also contain a dopamine-sensitive adenylyl cyclase (site 5). It should be noted that the dopamine receptors at sites 1 to 3 are not associated with adenylyl cyclase activity. The receptors at sites 4 and 5 are the only ones that control adenylyl cyclase activity. (B) Criteria for the classification of dopamine receptors according to Kebabian and Calne (1979). The two categories of receptors were originally called alpha-dopaminergic and beta-dopaminergic and later changed to D1 and D2 to avoid confusion with the subtypes of adrenergic receptor. Note that flupenthixol has been used as a ligand specific for the D1 receptor.
treatment of at least 30% of schizophrenic patients that are not helped by the older neuroleptics (Gross, Langner & Pfolz, 1974; Hippius, 1989). The D4 dopamine receptor was cloned by Van Tol and his colleagues in 1991. This receptor has similar pharmacological properties to the D2 and D3 receptors but with a very high affinity for clozapine (Van Tol et al., 1991). Subsequently six allelic variant forms of the D4 receptor were found in humans (Van Tol et al., 1992), differing as a consequence of the number of times a specific 48-bp sequence found in cytoplasmic loop 3 repeats. These variants were given the names D4.2 to D4.8, with D4.2 and D4.4 showing a decrease in their affinity for clozapine when NaCl is left out of the assay whereas D4.7 does not. Unfortunately no correlation has been found in the response of clozapine responding and non-responding schizophrenic patients with D4 allele frequency (Nanko et al., 1993; Shaikh et al., 1993; see also above). 6.9 Distribution of D1 and D2 receptors in the striatum of schizophrenics In 1976 it was shown that neuroleptics bind to the D2-like class of dopamine receptors and that the clinical potency of neuroleptics correlates well with their activity on these receptors (Creese et al., 1976; Seeman et al., 1976). Since that time positron emission tomography (PET) has shown that there is a normally high level of D2-like receptors in the human striatum (Fig. 6.8). This discovery centred the search for differences between normal and schizophrenic brains to the distribution of D2 receptors. Increases in the density of D2 receptors in the brains of schizophrenics were subsequently claimed for the striatum, particularly the caudate nucleus, putamen and nucleus accumbens, regions of the brain that are normally high in D2 receptors (Wong et al., 1986; Seeman et al., 1987; see however Farde et al., 1990). The striatum can be also divided into striosomes or patches as well as matrices according to the distribution of transmitter systems. For example, in the case of the cholinergic system within the matrix compartment, there is a substantial level of the
MONOAMINERGIC SYNAPSES AND SCHIZOPHRENIA: THE DISCOVERY OF NEUROLEPTICS
99
Fig. 6.6. Action of dopamine receptors at the cellular level. The subclasses of D1-like and D2-like dopamine receptors, defined in terms of the ability of the latter to bind haloperidol, are shown, together with their ability to either increase or decrease adenylate cyclase (AC) or increase potassium channel (K+) activity in neurons. The species column indicates the species from which the various receptors have been cloned (from Table 1 in Gingrich & Caron, 1993).
enzyme acetylcholinesterase, of high-affinity uptake sites for the precursor choline and of muscarinic M2 receptors but not of muscarinic M1 receptors. In the case of the dopaminergic system, the matrix is high in dopamine containing afferents and in D2 receptors, with the striosomes exhibiting high levels of D1 receptors, particularly in the dorsal striatum (Joyce et al., 1988). With the serotonergic and adrenergic systems, the distribution of serotonin uptake sites and of beta-adrenergic receptors is high in the matrix, with the highest levels in the ventral striatum. In addition to the basal ganglia, high levels of dopamine D2 receptors occur in the cingulate and temporal cortex (Bouthenet et al., 1987), and in the entorhinal cortex (Charuchinda et al., 1987; Fig. 6.9). There is a change in the balance of these transmitter systems in the brains of schizophrenics compared with controls. In the dopaminergic system, there is a reduction in dopamine afferents and D1 receptors in the caudate nucleus whereas there is a substantial increase in the number of D2 receptors in the nucleus accumbens and ventral putamen. In the case of serotonergic and adrenergic systems, the number of serotonin uptake sites and beta-adrenergic receptors is increased in the ventral striatum as well as in the hippocampus and the temporal cortex (Fig. 6.9). There is also an increase in the number of serotonin type 2 receptors in the nucleus accumbens; this increase amounts to about 200% (Joyce et al., 1988). As five different neuronal dopamine receptor genes have now been identified (that is D1 to D5; Fig. 6.6), with the D2-like family including D2, D3 and D4 (Fig. 6.6), it was of great interest to see if the elevation in the D2 receptors in the striatum was specific to one of the subclasses of receptors. The typical neuroleptics antagonise D2 and D3 receptors but not D4 receptors, the reciprocal being the case for atypical neuroleptics such as clozapine (Sokoloff et al., 1990; Van tol et al., 1991). Interest therefore centred on whether the D4 receptor is differentially affected in schizophrenics. To this end Seeman, Guan and Van Tol (1993) measured the D4 receptor numbers in schizophrenics compared with controls. In the absence of a specific D4 antagonist, this was carried out by subtracting the binding of 3H-raclopride, which labels D2 and D3 receptors, from that of 3HYM 09151– 2 which labels D2, D3 and D4 receptors collectively. This procedure showed that up to a six-fold increase in D4 receptors occurs in the putamen of the striatum in schizophrenics. Thus the increase in D2 receptors in schizophrenia reported above might really be due to an increase in D4 receptors, as the radioligands used were 3H spiperone or 3H-apomorphine or 3Hflupenthixol, all of which bind to D2, D3 and D4 receptors. Murray et al (1995) have shown that the D4 receptor can be found throughout the striatum and nucleus accumbens, with a two fold greater number in both areas in the brains of schizophrenics. As the mRNA for D4 receptors is in negligible amounts in the striatum (Meador-Woodruff et al., 1994), then it is unlikely that the D4 receptors are synthesised in the striatum but rather are localised on presynaptic nerve terminals in the striatum. A recent criticism suggesting that the D4 receptors may not be elevated in schizophrenia (Reynolds and Mason, 1994), is probably based on failure of raclopride competition against [3H] nemonapride to distinguish between D2 and D4 receptors (Seeman & Van Tol, 1995a, b). Some neurons contain mRNA for more than one subtype of dopamine receptor. For example 20% of striatal neurons contain mRNA for both D1 and D2 receptors (Weiner et al., 1991). Both synergistic and antagonistic interactions may occur between the two classes of receptors (Seeman et al., 1989), with radioligand binding showing pharmacological coupling between the D1-like and D2-like receptors (Seeman et al., 1989). The elevation of D2 receptors in schizophrenia may then occur because the D1 receptor fails to modulate the D2 receptor (Seeman & Niznik, 1990).
100
HISTORY OF THE SYNAPSE
Table 1 First Characterization of Clozapine in Comparison with Chlorpromazine (CPZ). Summary of Early Pharmacological Data (January 1961) Test Inhibition of locomotion and behavior Mouse, jiggle cage Cat, indifference, passivity Somnolence, mydriasis, salivation, prolapse of nictitating membrane Antinociceptive effects Mouse, phenylbenzoquinone syndrome Rabbit, tooth pulp stimulation threshold Cardiovascular and autonomic effects Anaesthetized cat: Decrease in blood pressure Decrease in heart rate NA-induced increase in blood pressure Nictitating membrane contraction after sympathetic stimulation Blood press, decrease after vagus stimulation Guinea pig, isolated ileum Acetylcholine-induced contraction Histamine-induced contraction Guinea pig, histamine-induced bronchospasm Effects on rabbit EEG Slow waves, high voltage, synchron. Electrically stimulated arousal Anticonvulsive effects Mouse, pentetrazol (120 mg/kg s.c.) Mouse, strychnine nitrate (1.3 mg/kg s.c.)
Clozapine (mg/kg)
CPZ (mg/kg)
ID90 p.o. EDmin p.o. EDmin p.o.
2.5 2 >4
3.5 2 4
ID90 p.o. ED200% i.v.
3 ~0.25
5 ~1.5
EDmin i.v. EDmin i.v. IDmin i.v. IDmin i.v. IDmin i.v.
0.1 0.5 0.2 0.1 0.1
0.2 0 >2 0.5 0.2
IC>90% IC>90% ID50 p.o.
0.16a 0.01a 3.6
0.25a 0.02a >5
EDmin i.v. IDblock s.c.
0.02 0.63
0.2 1.4
No protection with 50 mg/kg p.o.
Table 2 Comparison of the Pharmacological Effects of Clozapine, Chlorpromazine (CPZ), Haloperidol, Chlordiazepoxide (CDP), and Barbiturates (December 1961) Pharmacological effects
Clozapine
CPZ
Haloperidol
CDP
Barbiturates
Adrenolytic effects Noradrenolytic effects Sympatholytic effects (peripheral) Parasympatholytic effects (peripheral) Antihistaminic/antianaphylactic action Antiserotonin effects (peripheral) Antinociceptic effects Decrease in spontaneous motility Cataleptic effects Inhibition of decerebration rigidity (incompl.) Anticonvulsive action Pentobarbital potentiation EEG arousal inhibition Inhibition of conditioned suppression
+++ ++ +++ ++ ++ ++ +++ Wide dose range 0 +++
+++ + +++ + + +++ + Wide dose range + +++
0 0 (+) +
0 0 0 0 0
+ Wide dose range +++ +
0 Wide dose range 0 +
0 0 0 0 0 0 0 Narrow dose range 0 +
0 +++ +++ 0
0 +++ ++ 0
+++ 0
+++ + 0 +++
++ Additive ++ +
Fig. 6.7. The synthesis and characterization of clozapine (see Schmutz & Eichenberger, 1982). Table 1 shows the first comparison between clozapine and chlorpromazine. Table 2 shows a comparison between the pharmacological effects of clozapine, chlorpromazine, haloperidol, chlordiazepoxide and barbiturates. For the publications that support the contents of these Tables, see Hunziker et al., 1963; Stille et al., 1965; Stille et al., 1971.
MONOAMINERGIC SYNAPSES AND SCHIZOPHRENIA: THE DISCOVERY OF NEUROLEPTICS
101
Fig. 6.8. PET images show the distribution of radioactivity in horizontal brain sections at the level of the striatum. The control subject is shown in the two brain images on the left and a schizophrenic subject that had been treated with risperidone in the two brain images on the right. The [11C] raclopride was used to determine the binding of risperidone to the D2 receptors (upper row of images) and the specific 5HT2a receptor blocker NMSP (made into the radioligand [11C] NMSP) used to determine the binding of risperidone to the 5HT2a receptors (lower row of images). Note the high level of D2 receptors in the striatum (from Fig. 4 in Farde, 1996).
Fig. 6.9. A diagram of the altered expression of monoamine receptors in the temporal cortex and hippocampus of schizophrenics. The temporal cortex and hippocampus show an increase in 5-HT2 receptors in schizophrenics with also an increase in the beta 2 subtype of adrenergic receptor in the hippocampus. The position of relatively high levels of D2 receptors in the normal brain is indicated by * (from Fig. 5 in Joyce, 1993).
We have seen that dopamine receptors were first discovered as a consequence of the antagonistic effects of antipsychotics on the receptors. The subsequent discovery that neuroleptics block D2-like receptors in proportion to their clinical antipsychotic potencies (Fig. 6.4A) was a powerful argument in favour of the idea that antagonistic effects of neuroleptics on this class of dopamine receptor is responsible for their clinical potency. The cloning of dopamine receptors has apparently refined the process of developing new neuroleptics. The relative antagonistic effects of various antipsychotic drugs on the cloned D2, D3 and D4 receptors can now be compared with their therapeutic concentrations (Fig. 6.4). Considering the principal neuroleptics, the correlation between their clinical potency and their blocking power is best for the D2 receptors (Fig. 6.4B and 6.4C). It is clear, however, that the correlation for clozapine and olanzapine is not good. In the case of the D4 receptor the correlation between the clinical potency and the blocking concentration is very good for clozapine and olanzapine (Fig. 6.4B and E). Most neuroleptics therefore act on the D2 and D3 receptors with the exception of clozapine and olanzapine which act on the D4 receptors. It should be noted however that clozapine is a powerful blocker at a large number of receptors other than those involving dopamine, such as adrenergic alpha 1 and alpha2 receptors as well as serotonin receptors. This will be discussed in more detail below. D1 and D2 receptor mRNAs are found in relatively large amounts in the medial prefronatal insular and cingulate cortices but in low amounts in the motor and parietal cortices. D2 mRNA containing neurons are restricted to cortico-cortical, corticothalamic and corticostriatal pyramidal neurons. D1 mRNA containing neurons are present amongst these neurons. There is then evidence that the receptor genes for these two receptors are differentially expressed in different pyramidal neurons (Gaspar et al., 1995). Furthermore, there is no guarantee that dopaminergic nerve terminals will form synaptic connections on these D1 and D2 mRNA containing neurons as dopamine can diffuse for large distances after its release from nerve terminals (Vahabzadeh & Fillenz, 1992). In this case the dopamine receptors do not have to be matched to the sites of dopamine nerve terminals at all (Goldsmith & Joyce, 1994).
102
HISTORY OF THE SYNAPSE
6.10 Mixed aminergic actions of the neuroleptics: serotonin and dopamine receptor blockade In 1974 Janseen and his colleagues pointed out that although the neuroleptics act as dopamine antagonists they in general possesses a wide variety of other pharmacological activities (Carlos et al., 1979; Niemegeers et al., 1979). Using the typical behavioral response of the animals to apomorphine, tryptamine, compound 48/80 and norepinephrine, they investigated the affects on these of dopamine antagonists and found that these were often very considerable. Antagonism of tryptamine induced effects is a mark of serotonin antagonists, that of norepinephrine induced effects a mark of alpha adrenergic blockers, and compound 48/80 is an antihistamine. It therefore seemed that the neuroleptics may be having effects on other Monoaminergic receptors than just dopamine. This idea has focussed particular attention on the possibility that at least some neuroleptics are antagonistic to serotonin receptors as well as dopamine receptors. This is clearly the case for clozapine, which in addition to its high affinity for D4 receptors also has antagonistic affects at serotonin receptors and muscarinic receptors (Strange, 1994). It seems clear that in cases involving positive psychotic symptoms, as occur in acute schizophrenia and mania, dopamine receptor blockade is a necessary requirement of any neuroleptic. However serotonin type 2 receptor (5HT2 receptors) antagonism may also be important especially in relation to the reduction of extrapyramidal side effects. Indeed neuroleptics that specifically possess dopamine and serotonin antagonism, such as spiperone and pipamperone, may have useful applications at particular periods in the development of psychotic disorders. It seems very unlikely that the principal role of antipsychotics is as a consequence of their blocking serotonin receptors. Blockade of 5-HT receptors with the specific antagonist ritanserin does not prevent hallucinations and delusions in schizophrenic patients. However it is still possible that the principal action of antipsychotics, particularly of the atypical kind, arises from their blockade of 5-HT receptors with some additional blockade of D2 dopamine receptors (Meltzer & Nash, 1991). The typical neuroleptics block D2 receptors at a lower concentration than they do 5-HT2 receptors (except for chlorpromazine), whereas some atypical neuroleptics block 5-HT2 receptors at a lower concentration than they do the D2 receptors whilst others have the opposite antagonist potencies. Antagonism of 5HT2 receptors may be involved in the beneficial effects of neuroleptics that act to relieve the negative symptoms of schizophrenia as well as relieve some extrapyramidal side effects due to blocking D2 receptors; such a neuroleptic is risperidone (Leysen et al., 1993). It seems likely that in the design of more powerful neuroleptics consideration will have to be given to the action of these at several different types of receptors rather than at just one receptor. This point is emphasised by consideration of the actions of clozapine, the development of which has been considered the most significant advance in antipsychotic pharmacology since the discovery of chlorpromazine (Schmutz & Eichenberger, 1982). As already mentioned, clozapine has potent antagonistic effects on receptors for serotonin, acetylcholine, norepinephrine and histamine, as well as for the D4 receptor. The relatively low extrapyramidal side effects of clozapine and other atypical neuroleptics may then be related to their serotonin receptor blockade, specifically their action on 5-HT2 receptors (for recent reviews see Busatto & Kerwin, 1997; Reynolds, 1997). Certainly the most popular atypical antipsychotic at the present time, risperidone, has powerful antagonistic effects on both D2 and 5-HT2 receptors. 5-HT2 receptor antagonists can modulate the pattern of firing in mesolimbocortical dopamine neurons (Ugedo et al., 1989) and this has led to the hypothesis that neuroleptics such as clozapine are atypical because of this property (Svensson et al., 1989). It would seem that the best way of ascertaining the pharmacological properties of a potential neuroleptic, is to proceed to first determine the dopamine receptor blocking ability of the neuroleptic. Next its action on other receptors such as the 5HT2 serotonin receptor needs to be measured. Finally the toxic properties of the drug must be evaluated in order to form some judgement as to the safety of the potential antipsychotic. 6.11 Cellular and molecular mechanisms of action of dopamine receptors Dopamine D1 receptors are positively coupled to adenylate cyclase through the Gs-class and Goff-class of G-proteins (Sibley and Monsma, 1992; Fig. 6.10). Activation of adenylate cyclase by these G-proteins increases cystolic cAMP which leads to the release of the regulatory subunit from cAMP-dependent protein kinase (PKA; Fig. 6.10). PKA may then phosphorylate a number of different proteins such as the alpha 1 subunits of N-type and P-type calcium channels to reduce channel activity or in the case of L-type channels to increase their activity (Hell et al., 1993). Such changes may be mediated by an action of PKA to enhance protein phosphatases (PP1). These phosphatases may also be regulated by cytosolic inhibitors such as cAMP-regulated phosophoprotein (DARPP-32; Hemmings et al., 1990). The complexity of these pathways for controlling the different classes of calcium channels gives some indication of the diversity of effects on synaptic activity expected when D1 receptors are activated. The effect of agonists on dopamine D2 receptors in tissues produces, via G-protein activation, a whole range of responses. In the case of neurons these include inhibition of adenyl cyclase activity, potentiation of calcium evoked arachidonic acid
MONOAMINERGIC SYNAPSES AND SCHIZOPHRENIA: THE DISCOVERY OF NEUROLEPTICS
103
Fig. 6.10. Shown is a summary of the effects of activation of D1-like and D2-like receptors on cells.
release and modulation of potassium channels (Fig. 6.10). Stimulation of protein kinase C directs preferential coupling of transfected D2 receptors, via Gi proteins, from inhibition of adenyl cyclase to potentiation of arachidonic release. This probably occurs as a consequence of changes of the GTP sensitivities of these two pathways, perhaps due to phosphorylation of components of the receptor Gi protein complex. Gi mediated responses may be modulated in an activity dependent manner, as protein kinase C is regulated by transmitters and by neuronal depolarization, so that the signalling pathways of the D2 receptor may be determined by the relative activity of the neuron (DiMarzo et al, 1993), Activation of the D2 receptor in rat melanotrophs leads to a decrease in the secretion of the peptide hormones alphamelanocyte stimulating hormone and beta-endorphin. This occurs in part as a consequence of changes in the intracellular calcium concentration in the melanotrophs (Thomas et al., 1990; Taraskevich & Douglas, 1990) due to changes in calcium currents (Keja et al., 1992). D2 agonists such as quinpirole decrease the intracellular calcium concentration by decreasing the calcium influx across the plasmalemma, mainly through inhibiting L-type calcium channels; this is consequent on hyperpolarization of the membrane, probably through potassium channels, although there may also be some direct effect on the calcium channels (Stack & Surprenant, 1991). Activation of D2 receptors in rat lactotrophs, reduces calcium currents through L-type and T-type channels (Lledo et al., 1990). In frog melanotrophs (Valentijn et al., 1991) and cultured mesencephalic dopamine neurons (Liu et al., 1992) D2 receptor activation leads to a decrease in calcium currents through Ltype and N-type channels. Dopamine and quinpirole acting on D2 415 receptors transfected into NG 108–15 cells reduce both L-type and T-type calcium currents (Castellano et al., 1993). Dopamine and quinpirole acting on D2 415 receptors transfected into AtT20 cells decreases an L-type voltage dependent calcium channel (Snyder et al., 1992). Either Go-protein or Gi-protein mediate this closing of the voltage dependent channels. In summary, neuroleptics were used to first discover dopamine receptors and how dopamine acts on these receptors to modulate the adenylate cyclase system. Subsequently receptors were divided into two different classes according to whether they decreased or increased adenylate cyclase activity, namely the D1-like and D2-like receptors, with the latter specifically blocked by the butyrophenones. The subsequent cloning of the dopamine receptors revealed at least five different classes (D1 to D5) with considerable polymorphism in the human population of D4 receptors. The dopamine receptors are coupled to Gproteins to modulate a large number of cellular pathways via second messenger systems. 6.12 The time course of action of neuroleptics on dopamine receptors and the emergence of antipsychotic effects Neuroleptics do not produce their antipsychotic effects until some time after they produce maximum blockade of dopamine receptors. The possibility that changes in dopamine receptor activation leads to modification of cellular factors that take some time to establish must then be entertained. Immediate early genes might couple some of the acute actions of these drugs to long-term alterations in neural activity. It has been shown that changes occur in immediate early gene expression as well as that of neuropeptide genes in the striatum following activation of dopamine D1 receptors (Graybiel et al., 1990; Nguyen et al., 1992). The rapid time course of receptor blockade produced upon administration of neuroleptics compared with the slow emergence of clinical actions might also be explained by the slow development of a depolarization block in dopaminergic
104
HISTORY OF THE SYNAPSE
neurons. Repeated administration of neuroleptic drugs to rats gives rise to supression of dopamine-neurone firing in the midbrain. This occurs over a period of time as a consequence of a depolarization block. In humans the therapeutic usefulness of antipsychotic drugs is related to their ability to induce depolarization block of mesolimbic dopamine neurons. On the other hand their potential to produce extrapyramidal side effects is related to their ability to produce depolarization block in the nigrostriatal dopamine system. The usefulness of an antipsychotic drug may then be predicted by the extent to which it can produce such depolarization blocks (Grace et al., 1997). 6.13 Conclusion This chapter has considered the emergence of the monoaminergic hypothesis for schizophrenia, especially as related to dopamine. However, although it seems clear that the dopamine D2-like receptor class are involved in this psychosis there is no agreed model of how changes in the former give rise to the latter. There seem to be several main ways forward at this time. One of these is to increase the level of sophistication and technology involved in the genetic linkage studies on the D2-like receptors in schizophrenia as is being persued by van Tol and others. Another involves the recent discovery that neuroleptics can produce a depolarization block of dopamine neurons, and the extent of this block is proportional to the relative efficacy of the drugs as antipsychotics (Grace et al., 1997). Attempts to place this effect of the neuroleptics into an explanation for their antipsychotic activity at the level of synaptic connectivity is especially interesting. In addition, there is the use of new ligands of high specificity for D2 receptors, such as epidepride, for determining the distribution of extrastriatal D2 receptors (Hall et al., 1996). Finally, there is the development of new hypotheses to explain the multiple actions of such drugs as clozapine, as for example, the recent ideas of Arvid Carlsson implicating changes at other receptor types than the dopamine receptors in schizophrenia (Carlsson & Carlsson, 1990; Carlsson, 1996). Let us hope that the range of technologies now available will allow us to both determine before too long what receptors are to be targeted by neuroleptics and why it is that these particular receptors are implicated in the psychosis of schizophrenia.
7 The Discovery of Transmitters Other than Noradrenaline and Acetylcholine at Synapses in the Peripheral Nervous System
7.1 Introduction: J.N.Langley, H.H.Dale and non-adrenergic non-cholinergic (NANC) transmission Research on the synaptic connections of the autonomic nervous system was lead for over 40 years by J.N.Langley (1852– 1925) of Cambridge (Fig. 7.1A). His genius was both as an experimentalist (with an unbroken record of publications in the Journal of Physiology during each year from 1878 to 1925) as well as in a critical ability that allowed for generalizations so firmly embedded in fact that most of them have withstood the tests of time. They range from the teasing out of the enormous complexity of peripheral nerve connections to the idea of the chemical synapse and the concept of secretion of transmitter from nerve terminals onto receptors.
Fig. 7.1. Langley and the Autonomic Nervous System. (A) J.N.Langley (1852–1925) (B) Langley’s drawing of the origin of the preganglionic fibres from the spinal cord (from Fig. 1 in Langley, 1921) (C) Langley’s definition of the peripheral nerves, excluding the olfactory, optic and auditory nerves (from Langley, 1921).
106
HISTORY OF THE SYNAPSE
Fig. 7.2. The emergence of the classical autonomic nervous system paradigm. (A) Dale’s diagram of the peripheral nervous system. He says that at points marked C there is evidence of a cholinergic transmission, at those marked A of an adrenergic transmission. Doubtful cases marked? (from Dale, 1934b). (B) The classical autonomic paradigm as it still exists to this day (1997). The innervation of the gastrointestinal tract only is shown as abstracted from Goodman and Gilman’s The Pharmacological Basis of Therapeutics (from Fig. 4.1 in Gilman et al., 1985). Note the thoracolumbar outflow of preganglionic nerves to the sympathetic ganglia and from there to the gastrointestinal tract with the parasympathetic outflow from the vagus and pelvic nerves to this organ. Compare with A.
A subsequent generation attempted to build on Langley’s plan of the autonomic nervous system following his death in 1925, principally following the lead of H.H.Dale. Two seductive generalizations concerning the autonomic nervous system, which Langley held back, were established. The first of these ideas was that the sympathetic and parasympathetic nerves have antagonistic influences on most autonomic effectors. The second was that sympathetic neuroeffector junctions use adrenaline as transmitter, whereas parasympathetic neuroeffector junctions use acetylcholine as transmitter (Fig. 7.2A). The tremendous influence of Dale and Loewi, on the one hand in establishing acetylcholine as one transmitter at autonomic nerve endings, and that of Euler and Axelrod on the other, in determining that noradrenaline is the other transmitter, has consolidated these generalizations. Such ideas have held sway in popular texts up to the present (Fig. 7.2B). Yet in the 1960s a definitive series of experiments established that the muscles of the gastrointestinal tract, taken classically to receive an antagonistic innervation from the sympathetic and parasympathetic nerves, did not do so. Furthermore, a major component of transmission to these muscles was shown to be mediated by neither adrenaline or acetylcholine. This gave rise to the concept of non-adrenergic noncholinergic (NANC) transmission in autonomic neuromuscular control. In the subsequent 35 years NANC transmission has been discovered to mediate nervous control throughout the autonomic nervous system, with nitric oxide, adenosine triphosphate, and neuropeptides being identified as neurotransmitters. The present work gives an historical background to the discovery of NANC transmission, together with a summary of the contemporary view of the identity and function of this form of neuroeffector transmission at the site in which it was first discovered, namely in the smooth muscle of the gastrointestinal tract.
DISCOVERY OF OTHER TRANSMITTERS IN THE PERIPHERAL NERVOUS SYSTEM
107
7.1.1 J.N.Langley: defining the autonomic nervous system The great paper by J.N.Langley on the union of cranial autonomic (visceral) fibres with the nerve cells of the superior cervical ganglion (Langley, 1898a) led to the present definition of the autonomic nervous system. In this paper he wrote: I propose the term ‘autonomic nervous system’, for the sympathetic system and the allied nervous system of the cranial and sacral nerves, and for the local nervous system of the gut’. Noting that Gaskell (1886) had shown conclusively that the very great majority of the nerve fibres passing to the sympathetic arose from a limited region of the spinal cord, forming a ‘thoracic outflow’, Langley (1921) went on to divide the autonomic nervous system into tectal, bulbo-sacral, and sympathetic systems, and considered that each had a different developmental history (Fig. 7.1B). He concluded (Langley, 1898a) that: …there is no fundamental difference between the preganglionic fibres of the body, whether they belong to the cranial, the sympathetic or the sacral autonomic systems. The allied nervous systems of the cranial and sacral nerves were grouped in part by his discovery (Langley, 1901) that: The effects produced by supra-renal extract are almost all such as are produced by stimulation of some one or other sympathetic nerve…the effect of supra-renal extract in no case corresponds to that which is produced by stimulating normal conditions of cranial autonomic or a sacral autonomic nerve. Langley (1905, 1921) commented further: …that there is some fundamental difference between the sympathetic and the rest of the autonomic system was much strengthened by the discovery that the effects produced by adrenaline were apparently confined to effects caused by stimulating sympathetic nerves. Since other drugs caused effects more or less confined to those produced by stimulating tectal and bulbo sacral nerves, it was convenient to have a common name for these nerves, and I placed them together in the parasympathetic system… I use the word parasympathetic for the cranial and sacral autonomic systems. The drugs which caused effects ‘more or less confined to those produced by stimulating tectal and bulbo sacral nerves’ were nicotine and eserine. The qualification ‘more or less’ came from the observation (Langley, 1905) that: …atropine readily paralyses most postganglionic cranial autonomic fibres, but it has a comparatively slight effect on the cranial viscero-motor fibres, and little if any on the cranial vaso-dilator fibres. This is a difficulty which Langley would not submerge in a general theory of the autonomic outflow, and more or less restricted him from giving a simplified diagram of the autonomic nervous system. This was later provided by Dale (1934b; Fig. 7.2A), on the basis that it helped to clarify the nervous supply to the viscera and vasculature. Rather, Langley confined himself completely to the facts in presenting both the sympathetic and parasympathetic divisions of the autonomic nervous system (Fig. 7.1B). Langley considered the cells of Auerbach’s and Meissner’s plexuses as being in a class by themselves. He completed his subdivision of the peripheral nerves, defining them in Schafer’s Text Book of Physio-logy in 1900 as belonging to the enteric nervous system (Fig. 7.1C). He noted (Langley, 1921) that: …we should expect the cells of Auerbach’s and Meissner’s plexuses to be on the course of the bulbar and sacral nerves, but as there was no clear proof of their central connection, and as their obvious histological characters differed from those of any other peripheral nerve cell, I placed them in a class by themselves as the enteric nervous system. The subsequent discovery of the course of connections between the sympathetic and the enteric nervous system (see below), indicated that the note of caution was correct. This completed Langley’s great and lasting definitions that have withstood the tests of experimentation over nearly one hundred years. The emphasis I have placed on Langley’s achievements here and elsewhere (Bennett, 1994) is not generally accepted (see for example, Davenport, 1989). The traditional view concerning the discovery of chemical transmission is to highlight Eliott (1904a)’s ideas concerning the role of adrenaline at sympathetic nerve terminals (or even earlier suggestions of du Bois Reymond), make some reference to the experiments of Barger and Dale (1910) on the sympathomimetic activity of amines,
108
HISTORY OF THE SYNAPSE
and then to move on to the famous experiments after the Great War by Loewi (1921). It should be noted that Loewi had visited Langley and Eliott nearly twenty years earlier, at a time when they were carrying out their experiments with adrenaline (Dale, 1962). I have argued that it was Langley who initially produced the substantial experimental and conceptual framework for the idea of chemical transmission and indeed for transmitter receptors (Bennett, 1994). By 1906 Langley was able to summarise his ideas on neuromuscular transmission with the comment that …the stimuli passing the nerve can only affect the contractile molecule by the radical which combines with nicotine and curare. And this seems in its turn to require that the nervous impulse should not pass from nerve to muscle by an electrical discharge, but by the secretion of a special substance at the end of the nerve (Langley, 1906). Eliott left Cambridge and physiology at an early age, not because of any hostility from his mentor Langley but for reasons associated with career opportunities (Dale, 1961). One wonders why Langley’s greatness is not more generally recognised, especially amongst his British colleagues. Bacq (1974), in his history of chemical transmission asks …why is there relative neglect of this part of Langley’s work? Is it deliberate? Had Langley annoyed his contemporaries by his excessively severe control of the Journal of Physiology? Perhaps the comparative neglect of Langley can be traced to the great influence which Dale had over British physiology. Bernard Katz (1981) comments that ‘Dale’s criticism of Langley’s idea seems always a little too hard and somewhat idiosyncratic. (One may wonder whether J.N.Langley, who not only owned, but very thoroughly edited the Journal of Physiology, had in the course of his duties produced powerful antibodies among some of the authors who submitted their work to his Journal, though it should be noted that others—like A.V.Hill and W.M.Fletcher—have recorded their deep gratitude to Langley for his editorial severity)’. 7.1.2. H.H.Dale: emergence of the classical autonomic paradigm Dale’s great prestige was established by the late 1920s as a consequence of his discoveries concerning acetylcholine. The first of these was the identification of acetylcholine in ergot and of its ability to reproduce the various effects of parasympathetic nerves when it was injected into the circulation. As he put it (Dale, 1914): I have been familiar for sometime with a pronounced inhibitor effect on the heart, shown by some specimens of ergot, and always associated with an intense stimulant action on intestinal muscle. Both actions were blocked by atropine. His other great discovery was the identification of acetylcholine as a natural constituent of the body specifically in the horse spleen (Dale & Dudley, 1929). Otto Loewi had discovered in 1921 that the vagus nerve inhibited the heart by secreting an inhibitory substance which was found to correspond, in every test, to that of acetylcholine; atropine annulled the action of the transmitter and eserine (physostigmine) inhibited the action of an esterase in the heart that rapidly destroyed the transmitter (Loewi, 1921). Together, the laboratories of Dale and Loewi had shown unequivocally that the effect of the vagus on the heart was mediated by acetylcholine, found in the body and normally hydrolyzed by an esterase after producing its inhibitory effect on secretion. The origins of the classical autonomic paradigm (Fig. 7.2A) can be found in Dale’s (1934a) note in the Journal of Physiology, in which he says: We can then say that postganglionic parasympathetic fibres are predominantly, and perhaps entirely, ‘cholinergic’, and that postganglionic sympathetic fibres are predominantly, though not entirely, ‘adrenergic’, while some, and probably all of the preganglionic fibres of the whole autonomic system are ‘cholinergic. This work of Loewi and Dale on the chemical transmission of inhibitory effects of the vagus to the heart was followed by similar experiments to establish the chemical nature of inhibition and excitation to the stomach. Firstly Finkleman (1930; Fig. 7.3A), using the technique of Loewi, showed that when movements of a piece of intestinal muscle are inhibited by stimulating the periarterial nerve, a substance appears in the fluid passing over the surface, which has the power of inhibiting a second piece of gut. Hence it is argued that the inhibitory nerves to plain muscle act by liberating peripherally an inhibitory substance. This substance was taken to be adrenaline secreted from sympathetic nerve terminals. As for the vagus, Dale and Feldberg (1934; Fig. 7.3B) soon showed that acetylcholine was released within the muscle of the stomach on stimulating the vagus, providing identification of the excitatory transmitter. Yet this interpretation was not straightforward, for as Dale (1934b) subsequently exclaimed:
DISCOVERY OF OTHER TRANSMITTERS IN THE PERIPHERAL NERVOUS SYSTEM
109
…there are some parasympathetic effects, such as the action of the vagus on the intestine, and the vaso-dilator actions of parasympathetic nerves in general, which are resistant to atropine, though the otherwise similar actions of injecting or applying acetylcholine are readily abolished by it. He suggests (Dale 1934b) that: It seems to be very much more probable that the actions of the vagus on the small intestine, and of the pelvic nerve on the large intestine, are similarly transmitted by the release of acetylcholine, but in such proximity to the reactive structures that atropine interferes but little with its action. Dale felt this was the supposition he needed in order to make the generalizations embodied in the classical paradigm (Fig. 7.2A). Although he still made the cautionary note (Dale and Feldberg, 1934) that: …it must be admitted that an assumption that acetylcholine acts as the general parasympathetic transmitter requires additional suppositions to explain their anomalous difference. What then was this difficulty with the action of atropine, which was later mostly ignored, so leading to the acceptance of the erroneous generalizations given in the classical autonomic paradigm (Figs. 7.2A and 7.2B). 7.2 Parasympathetic neuromuscular junctions in the gastrointestinal tract: mechanical studies 7.2.1 Parasympathetic excitatory nerves in the gastrointestinal tract resistant to atropine In 1898, Langley described the inhibitory fibres in the vagus to the stomach and commented (Langley, 1898b) that: The after-contraction is perhaps the most constant visible effect of vagus stimulation; it may be conspicuous although the previous relaxation has been barely visible, and even if there has been previous contraction or rhythmic contraction. This contraction was resistant to atropine (Langley, 1905): Atropine readily paralyses most post-ganglionic cranial autonomic fibres, but it has a comparatively slight effect on the cranial viscero-motor fibres. In their original description of the movements and innervation of the small intestine, Bayliss and Starling (1899) made two striking discoveries. The first came to be known as the Law of the Intestine and was that: Local stimulation of the gut produces excitation above and inhibition below the excited spot.These effects are dependent on the activity of the local nervous mechanism. In this way, inhibition in front of a bolus of food was identified as a necessary concomitant of peristalsis. Their other discovery (Fig. 7.4A) was that: …the vagus nerve contains two sets of fibres, inhibitory and augmenter. The inhibitory fibres have a short latent period, the augmenter fibres a long latent period…. The action of the vagus on the intestines is unaltered by atropine, but is permanently abolished by a small dose of nicotine. So 35 years prior to Dale's (1934) generalisation the discoverers of the role of inhibition and excitation in peristalsis had shown that the vagus to the intestines was both inhibitory and excitatory, with each of the synapses mediating these effects being resistant to atropine. It was clearly recognised by Bayliss and Starling (1901) that the sympathetic nerves in the splanchnic were inhibitory to the intestine. However they hesitated over the claim that the vagus contained inhibitory nerves to the intestine in addition to excitatory nerves, commenting (Bayliss & Starling, 1901) that:
110
HISTORY OF THE SYNAPSE
Fig. 7.3. Evidence for chemical transmission in the gastrointestinal tract. (A) When the periarterial nerves to the intestine are stimulated, a substance appears in the bathing fluid which can inhibit a second piece of intestine. The lower trace is from the innervated intestine. The upper trace is from the lower piece of intestine, over which drips fluid from the upper piece. On stimulating the nerve (shown by arrows) inhibition of the innervated preparation takes place almost immediately. After about 30 sec. there is relaxation of the lower piece of intestine (from Fig. 6 in Finkleman, 1930). (B) Vagal excitatory control of the stomach is likely to be mediated by acetylcholine. Shown are tests of a perfusate of Locke’s solution containing physostigmine passed through the vessels of the stomach of a dog during vagal stimulation; the samples collected before stimulation were largely inactive. Tests are shown of the concentrated vagus effluent on: a, blood-pressure of a cat under chloralose; b, frog’s heart; c, frog’s rectus abdominis; d, leech muscle. In each series, B shows the effect of the perfusate collected during vagal stimulation; A and C correspond to two strengths of acetylcholine (from Fig. 5 in Dale & Feldberg, 1934).
…we are inclined to regard the preliminary inhibition occurring in the dog (Fig. 7.4C), not as the direct result of stimulating inhibitory fibres in the vagus, but as the consequence of a state of excitation occurring higher up in the alimentary canal. However there was a dilemma as:
DISCOVERY OF OTHER TRANSMITTERS IN THE PERIPHERAL NERVOUS SYSTEM
111
Fig. 7.4. Contraction of smooth muscles of the gastrointestinal tract following stimulation of inhibitory nerves. (A) Tracing of contractions of the first part of the duodenum showing the inhibitory effect of stimulating the vagus nerve at the time indicated by the square wave pulses. Relaxation occurs throughout the period of stimulation and this is followed by a very pronounced motor after-effect. The inhibitory fibres have a short latent period, the augmenter fibres a long latent period. This action of the vagus on the intestines is unaltered by atropine. (Time marker is 6 seconds; from Fig. 26 in Bayliss & Starling, 1899). (B) Tracing of contractions of the colon to stimulation of the pelvic nerve. During the stimulation period, indicated by the square pulse, there is a slight relaxation and this is followed by a vigorous contraction at the end of stimulation. This trace led to the conclusion that ‘the pelvic visceral nerve is motor to both coats’. (Time marker is 6 seconds; from Fig. 6 in Bayliss & Starling, 1900). (C) Tracing of contractions (upper line) of the small intestine due to stimulation of the vagus and splanchinc nerves at the times indicated by the square waves. Note that the vagus gives a relaxation followed by enhanced contractions during the period of stimulation; the splanchnic only gives relaxation during stimulation. The conclusion was that ‘in all animals investigated the vagus acts as the motor nerve and the splanchnic as the inhibitory nerve to the small intestine’ (Time marker is in seconds; from Fig. 11 in Bayliss & Starling, 1901). (D) Tracings of the contractions of muscle strips from the fundus of the stomach due to the addition of first adrenaline, and then relaxation of the muscle to adrenaline after pituitrin. This led to the conclusion that ‘the addition of adrenaline to strips from the fundus and body of the stomach may have an excitatory or inhibitory effect’ (Time marker not clear; from Fig. 1 in McSwiney & Brown, 1926).
the result of stimulating the vagus (after administration of atropine) is shown in Fig. 7.4C. It will be noticed that the initial inhibition is as well marked in the cat as in the dog. The fact that it was impossible to obtain evidence of local reflexes or descending inhibition, in the experiment from which this tracing was taken, presents a serious objection to the view put forward in the last section (quoted above) as to the causation of the initial vagal inhibition. This difficulty we are not at present in a position to overcome. The situation was not improved when Bayliss and Starling (1900) turned their attention to the innervation of the large intestine. They found that the pelvic nerves acted on the large intestine in the same way as the vagus on the small intestine, namely they found that the effect of stimulating the pelvic nerves on the colon in dogs was a relaxation followed after some interval by contraction (Fig. 7.4B) and: The sympathetic supply to the colon (colonic splanchnic s) have a purely inhibitory effect on the muscular coats of the bowel.
112
HISTORY OF THE SYNAPSE
However, again, they were uncertain as to whether to ascribe a synaptic origin to the inhibition mediated by the pelvic nerves or not: …although in many cases the response to stimulation of this nerve (‘the pelvic visceral nerve’) is directly augmenter, in the majority of cases, especially if the intestine be in a state of tone, the first effect is an initial inhibition which is followed by the augmentation. This inhibition is exactly analogous to that which is found in this animal in the small intestine as the result of stimulation of the vagus, and, as in this case, we are inclined to ascribe it to a descending inhibition, resulting from contraction of the gut some distance above the ballon, and regard it therefore, though initial, as only a secondary or mediate effect of the stimulation of the nerve. Thus Bayliss and Starling used their discovery of the Law of the Intestine to explain the initial relaxation following stimulation of either the vagus or the pelvic nerves. However, the impossibility in some experiments of observing local reflexes involving descending inhibition whilst at the same time observing relaxation in response to stimulation of either the vagus or the pelvic nerves, really made this argument untenable. The problem of explaining away relaxation of the gastrointestinal tract in response to stimulation of the parasympathetic nerves, and in particular the atropine resistant contraction, continued to plague the classical paradigm for the autonomic innervation of the viscera (Fig. 7.2) for over half a century. For example, Paton and Vane (1963) obtained relaxation followed by vigorous contractions of the stomach to stimulation of the vagus nerve in the presence of the muscarinic antagonist hyoscine. As adrenaline was thought to be the sole inhibitory transmitter to the gastrointestinal tract, Carlson et al. (1922) were surprised when: …the effects obtained with the drug (adrenaline) on the cardiac were similar to those previously described on stimulation of the nerve fibres. Adrenaline has a motor action when the stomach was hypotonic and an inhibitory action when hypertonic. Thus, tone seemed to hold the key to the augmentation phenomenon. The augmentation led McSwiney (1931) in his review to explain that: …it is obvious that the vagus cannot be considered as the motor, nor the sympathetic as the inhibitor nerve, since stimulation of either nerve may cause contraction or relaxation of the organ (see Fig. 7.4D). The relationship between tone and the atropine-resistant augmentation was studied in the taenia coli of the caecum by Campbell (1966b). He showed that whilst atropine abolished the contractions occurring during the period of stimulation, the contractions following stimulation were often augmented. In low tone preparations these after-contractions increased following an increase in the frequency of stimulation in the presence of atropine. In summary then, an atropine-resistant contraction of the muscles of the gastrointestinal tract to stimulation of parasympathetic nerves had been observed for over 60 years by the early 1960s. It had been ‘put aside’, probably because of the seductive idea that the sympathetic and parasympathetic systems secreted two different transmitters only (Fig. 7.2A). Any contradictions to this idea were ascribed to anomalies involving, e.g. an atropine-resistance arising because (Dale & Gaddum, 1930; Dale, 1934b): …the nerve impulses liberate acetylcholine so close to the reactive structures that atropine cannot intervene, whereas it can prevent its access to them when it is artificially applied from without. 7.2.2 Parasympathetic inhibitory nerves exist in the gastrointestinal tract In 1889, Openchowski showed that relaxation and contraction of the cardia muscle of the stomach could be obtained on stimulation of strands of the vagus nerve, referring to the inhibitory component in these strands of the vagus as ‘the dilator nerve of the cardia’ (Openchowski, 1889). This observation was confirmed by Doyon (1894) and Kelling (1903). Langley (1898b) went on to show that the vagus could produce relaxation of the muscle in both the cardia and the fundic region of the stomach in the presence of atropine, especially if the tone of the muscle was high. The discovery of vagal inhibition of the stomach musculature soon was extended to the cardiac and pyloric sphincters by May (1904; Fig. 7.5A), who commented that:
DISCOVERY OF OTHER TRANSMITTERS IN THE PERIPHERAL NERVOUS SYSTEM
113
Fig. 7.5. Evidence for vagal and sympathetic inhibition of the stomach. (A) Stimulating the vagus to the stomach and sphincters in the presence of atropine. The upper line gives the height of fluid help up by the cardiac sphincter; the middle line gives the period of excitation (E) of the vagus nerve; the lowest line gives the time in tens of seconds. A pronounced relaxation accompanies each period of stimulation (from Fig. 5 in May, 1904). (B) If the gastric wall is in tonic contraction, then it relaxes following an act of deglutition. A gives gastric relaxation after deglutition with the right vagus cut; B, at the moment of cutting the left vagus nerve as well; C, the complete failure of relaxation to occur when both vagi are cut (from Fig. 3 in Cannon & Lieb, 1911). (C) The response of an isolated vagus nerve-smooth muscle preparation to stimulation of the nerve before and after atropine. A shows the normal contractile response to a tetanus; B gives the time at which atropine was added; C and D give the responses to a tetanus in the presence of atropine. The square waves indicate the period of stimulation. (Time interval is 10 sec, from Fig. 3 in McSwiney & Robson, 1929). (D) The effects of stimulating the thoracic sympathetic chain on the contractile response of the stomach. The tracings show the augmentation and inhibition of responses after eserine and atropine respectively. At 6–18 pm, the right thoracic sympathetic chain was stimulated; at 6–46pm the stimulation was repeated after injecting eserine; at 8–46 pm, the stimulation was repeated again after injecting atropine. Note that the frequency of stimulation was 60 Hz, so that the contractions are probably due to rebound firing (from Fig. 2 in Harrison & McSwiney, 1936). (E) The vagal inhibitory effect on the stomach by stimulating the medulla oblongata is shown in A. In this experiment, the spinal cord was severed between the cervical and thoracic cord to prevent the sympathetics affecting the stomach. The vagus nerves were intact. In B the vagal inhibitory effect was abolished after blocking ganglion cells with hexamethonium (from Fig. 2 in Semba et al., 1964). (F) The sites of stimulation in the medulla oblongata that give rise to vagal inhibition of the stomach. The diagram shows a coronal section through the level of the rostral side of the alae cinerea. Solid circles show the inhibitory areas through the vagus nerve (from Fig. 3 in Semba et al., 1964).
…the effects of vagus stimulation on the cardiac and pyloric sphincters is practically the same as on the stomach so that for the cardiac sphincter after atropine was given the vagus showed an inhibitive action, followed by an increased tone
114
HISTORY OF THE SYNAPSE
(Fig. 7.5A). It had been realised by Moral that the vagal fibres to the stomach may serve as the efferent path of an inhibitory reflex as early as 1893 (Moral, 1893). The role of this inhibition of the stomach in swallowing was investigated by Cannon and Lieb (1911) who found (Fig. 7.5B) that: …if the gastric wall is in tonic contraction, it relaxes after an act of deglutition. Intragastric pressure then falls nearly to zero, and the capacity of the stomach is readily enlarged. Following sections of both vagi this relaxation fails to occur after deglutition (Fig. 7.5B), indicating that inhibition is mediated by the vagus. The identification of relaxatory vagal fibres to the stomach that constitute an efferent link in a vagal-vagal reflex associated with deglutition has been confirmed repeatedly in more recent times (Abrahamsson & Jansson, 1969, 1973; Jansson, 1969; Ohga et al., 1970). Further investigations in the 1920s, of the effects of the vagus on the stomach supported the existence of both excitatory and inhibitory nerves in the vagus (Carlson, et al., 1922; McCrea et al., 1925; Veach, 1925). Emphasis was given to the idea that the inhibitory effect was best observed when the muscle was in a high degree of tonus (McCrea et al., 1925). The next technical advance on this subject was the introduction by McSwiney and Robson (1929) of an isolated vagusnerve smooth muscle preparation, in which (Fig. 7.5C): the contraction evoked by stimulation of the nerve was converted into relaxation by the addition of atropine. The definitive evidence for the existence of separate populations of inhibitory and excitatory nerves in the vagus to the gastrointestinal tract was obtained by the identification of different neurones in the medulla oblongata initiating vagal inhibition and excitation. Inhibition of the stomach could be obtained through the vagus nerve by stimulating the medulla oblongala unilateral to the dorsal median sulcus and at a level between the central alae cinerea and its oral side whereas stimulation of the alae cinerea itself gave contraction (Fig. 7.5E & F; Semba et al., 1964). The existence of vagal inhibitory nerve fibres to the muscles of the stomach and small intestine has been confirmed repeatedly, even though largely ignored in the classical autonomic paradigm, at least by European physiologists. However, the question of whether inhibitory nerves exist in the sacral pelvic innervation of colonic muscle is still not entirely clear. Langley and Anderson (1895) only observed motor responses to stimulation of the pelvic nerve in the rabbit colon, a ‘brief’ inhibition being a ‘rarity’. However, Bayliss and Starling (1900) (Fig. 7.4B) noted that in the dog: …although in many cases the response to stimulation of this nerve (the pelvic visceral nerve) is directly augmenter, in the majority of cases, especially if the intestine be in a state of tone, the first effect is an initial inhibition which is followed by augmentation. The question of which of these great experimenters is correct, or whether both are as there may be species differences, has remained a contentious issue to this day. Garry and Gillespie (1955), Furness (1969), as well as Davidson and Pearson (1979) sided with Langley and Anderson (1895), the latter commenting that: …stimulation of the pelvic nerves to the colon of the rabbit in vitro always causes contraction… The initial state of the preparation, high tone or low tone, does not affect the result. On the other hand, Hulten and his colleagues (Hulten, 1969; Fasth el al., 1980) agree with Bayliss and Starling, noting in the cat that: …the colonic contraction on pelvic nerve stimulation remained unchanged after atropine but was delayed in onset. Moreover, in the transverse and distal colon it was preceded by a relaxation which was most pronounced in the distal part. It is not possible then to conclude that a parasympathetic inhibitory innervation occurs for the musculature of the entire gastrointestinal tract. 7.2.3 Parasympathetic inhibitory nerves in the gastrointestinal tract and the release of adrenaline We have seen that the atropine-resistant excitation of the muscles of the gastrointestinal tract was explained away by a special relationship between ‘excitatory’ nerve endings and smooth muscle, which did not allow atropine access to the receptors at
DISCOVERY OF OTHER TRANSMITTERS IN THE PERIPHERAL NERVOUS SYSTEM
115
these endings. The existence of a vagal inhibition was more difficult to put aside in order to sustain the classical paradigm of the autonomic nervous system (Fig. 7.2). Bayliss and Starling (1899, 1900, 1901) had suggested that the inhibition might actually be part of the peristaltic relaxation, triggered by an excitatory effect of the vagus on the muscle. We have seen, however, that they were aware of difficultes with this interpretation. Another possibility, that seemed to be largely accepted by those that recognised the existence of a vagal inhibition, was that it was mediated by adrenergic neurones in the vagus, due to the existence of sympathetic axons in the vagus nerve (Ranson et al., 1933). Thus, Harrison and McSwiney (1936), having clearly shown that an atropine-resistant vagal inhibition of the stomach was present (Fig. 7.5D), entertained the idea that: The relaxation of the stomach obtained on stimulation of the vagus nerve is partly or entirely due to the presence of adrenergic fibres. This idea persisted for many years (Greeff et al., 1962). Furthermore, Paton and Vane (1963) argued that they could reduce the relaxation due to vagal stimulation of the stomach by pretreatment of the animal with reserpine, a catecholamine depleting agent, although Martinson and Muren (1963) in the same year were unable to block this relaxation with sympalholytic drugs. Paton and Vane (1963) concluded that the vagus must innervate both adrenergic inhibitory neurones in the muscle as well as cholinergic excitatory neurones that were part of the accepted classical paradigm. They concluded that: No evidence has been obtained for chemical transmitters other than acetylcholine or sympathen. On the other hand, Martinson and Muren (1963) concluded that: The excitatory fibres (in the vagus) appear to be cholinergic, whereas the inhibitory effect seems to be mediated by some other transmitter substance. What then was the origin of these adrenergic neurones? Semba, Fujii and Kimura (1964), on the basis of the stimulation experiments on the nuclei of the vagus in the medulla oblongata, concluded that since there were both inhibitory and excitatory nuclei, and that inhibition to the gastrointestinal tract was mediated by adrenaline, then: It may be concluded from this experiment that the adrenergic nerve fibres which inhibit the stomach movements are contained in the vagus nerve. The conclusion that inhibitory actions are exerted by parasympathetic nerves to the gastrointestinal tract, although now established by many experts over 70 years, was not generally acceptable by European physiologists, probably because it was in contradiction to the classical autonomic nervous system paradigm (Fig. 7.2). However, since these nerves were taken as adrenergic by many researchers, this part of the paradigm remained in place. 7.2.4 Doubts arise concerning the possibility that parasympathetic inhibitory nerve terminals release adrenaline Selective sympathetic blocking drugs had become available by the early 1960s which might be expected to help test the idea that inhibitory axons in the parasympathetic nervous system secrete noradrenaline at neuromuscular junctions. Paton and Vane (1963) were amongst the first to use TM10 (choline 2, 6-xylyl ether) or bretylium, which inhibits responses to adrenergic nerve stimulation without impairing responses to exogenous catecholamines. They suggested that the relaxant response to vagal stimulation could be reduced by TM10, potentiated by cocaine in low doses, and somewhat reduced by antiadrenalines. This was taken to indicate that the vagus innervated adrenergic neurones. In contrast to this result, Burnstock et al. (1964) showed that drugs which inhibit the release of noradrenaline (bretylium and guanethidine) could block the sympathetic relaxation of the longitudinal muscle of the guinea-pig caecum (called the taenia coli) without affecting the relaxation of the caecum to stimulation of the enteric inhibitory neurones (Fig. 7.6B; this figure is essentially the same as that of Figure 5 in Burnstock et al., 1966). They concluded that these enteric inhibitory neurones were distinct from adrenergic neurones. Martinson (1965) at that time reached similar conclusions concerning the vagal and sympathetic inhibitory supply to the stomach (Fig. 7.6C), namely: …that the relaxation of the stomach on excitation of ‘high-threshold’ efferent vagal nerve fibres is mediated via preganglionic vagal fibres, which do not exert their effect by any adrenergic mechanism.
116
HISTORY OF THE SYNAPSE
Campbell (1966a; Fig. 7.6D) obtained similar experimental results to those of Martinson (1965). However by 1966, Burnstock et al. were equivocating. On the basis of the same experimental results as those shown in Fig. 7.6B they commented that: It is apparent that the intramural inhibitory neurones of the guinea pig taenia differ someway from the perivascular sympathetic nerves but we are uncertain about the transmitter substance released from them. It is possible that they are adrenergic but attempts to demonstrate (or disprove) this by depletion of catecholamines with reserpine, blockade by or -adrenoreceptors with appropriate blocking drugs or enhancement of responses to released catecholamines with cocaine have given equivocal results (Burnstock et al., unpublished observations). There was still uncertainty amongst those pharmacologists who were willing to admit that a parasympathetic inhibitory innervation of the gastrointestinal tract existed, as to whether this used the transmitter noradrenaline or not (see also Holman & Hughes, 1965). 7.3 Parasympathetic neuromuscular junctions in the gastrointestinal tract: electrophysiological studies 7.3.1 Discovery of the electrical signs of inhibitory transmission: the inhibitory junction potential Two techniques were instrumental in the discovery of inhibitory junction potentials in the muscles of the gastrointestinal tract. One of these was due to Gaskell (1886) who introduced a method for stimulating intramural nerves of a visceral tissue by passing current across the muscle at sufficiently low strength so that the nerves but not the muscle cells are directly excited. In this way he showed that: …the diminution of the contractions of the isolated muscular strip when a weak interrupted current passes through it is evidence of the existence of inhibitory nerve fibres in the strip; consequently the inhibitory nerves pass as such into the muscular tissue itself and act upon that tissue directly without need of any intermediate apparatus. This technique was brought to its modern form by Paton (1954). The other method used in the discovery of the inhibitory junction potential was due to Huxley and Stampfli (1951). They invented a new technique for recording the action potential at nodes of Ranvier of single myelinated axons with external electrodes. This involved exposing a node to Ringers’ solution and depolarizing an adjacent node with a high KCl solution; a high-resistance substance was then introduced into the internodal region by exposing it to oil or sucrose. This technique could also be used to record action potentials in smooth muscles as the muscle cells form a continuous electrical syncytium (Burnstock & Straub, 1958; Bennett & Burnstock, 1966). In 1962 this method, by now called the sucrose-gap technique, was established in the laboratories of Geoff Burnstock in the Zoology Department at Melbourne University and was used to determine the ionic basis of the action potential in smooth muscle. The ureter smooth muscle was chosen, as we thought it did not contain nerves, except at the proximal end. Hence stimulation with currents across the walls of the ureter can be used to initiate action potentials in the muscle without the complexities associated with exciting an intramural nerve supply. The surprising results of this investigation were that the rising phase of the action potential and its form depends on calcium ions (Bennett et al., 1962a; Bennett et al., 1962b). The existence of calcium action potentials was later confirmed for other visceral smooth muscles (Bennett, 1967). Although it was known that calcium action potentials could be obtained in invertebrate muscle if these were first treated with ammonium ions or their internal calcium concentration buffered (Fatt & Gibsborg, 1958; Hagiwara & Naka, 1964), these observations on smooth muscle were the first to establish naturally occurring calcium action potentials in excitable tissue. In the Christmas of 1962 an attempt was made to see if such calcium action potentials occurred in the gastrointestinal tract. To that end a taenia coli muscle was mounted in the sucrose-gap apparatus. The surprising result obtained was that stimulating currents across the wall of the taenia coli gave hyperpolarizing responses that blocked spontaneous muscle action potentials (Fig. 7.7A). These hyperpolarizations in response to single stimuli lasted for about one second and were several millivolts in amplitude. It seemed likely that these might be akin to the inhibitory post-synaptic potentials that Eccles and his colleagues in Canberra had for some years been describing, both at meetings of the Australian Physiological and Pharmacological Society (Eccles, 1962), in Nature (Andersen & Eccles, 1962) and in his recent review (Eccles, 1961). This review contains a very full description of inhibitory synaptic potentials, including reproductions of the records of del Castillo and Katz (1957), of vagal inhibitory junction potentials in the arrested frog heart (Fig. 7.7C). On the other hand the hyperpolarizations in response to transmural stimulation of the taenia coli might have been due to contraction of the muscle at
DISCOVERY OF OTHER TRANSMITTERS IN THE PERIPHERAL NERVOUS SYSTEM
117
Fig. 7.6. The first unequivocal demonstrations of relaxations of the gastrointestinal tract due to nerve stimulation that are resistant to adrenergic blocking drugs. (A) Shown in A are contractions of the stomach to vagal stimulation in three different stimulation series with variations in the stimulation duration and voltage as indicated by the square waves. B, relaxations of the stomach using the same stimulus parameters as in A, 10 minutes after atropine. This relaxation was not affected by the sympathetolytic dihydroergotamine, leading to the possibility that the relaxation was not due to adrenergic inhibitory fibres (from Fig. 3 in Martinson & Muren, 1963). (B) Frequency-relaxation curves for the taenia coli following stimulation of different nerves, measured by G. Campbell. The curves show the maximum relaxation of the muscle at the stimulus frequency indicated, a, shows the results for stimulation of the sympathetic periarterial nerve supplying the taenia coli both before ( ) and after guanethidine (10−6 g/ml; ). b, shows frequency relaxation curve for stimulation of the intramural nerves before ( ) and after ( ) guanethidine (10−6g/ml). Note that the peak at the lower frequency of 10 Hz is not abolished by guanethidine. This led to the conclusion that
…the inhibitory responses to transmural stimulation which persist in the presence of these drugs are mediated by intrinsic nerves which are distinct from the sympathetic or parasympathetic systems (from Fig. 1 in Burnstock et al., 1964). (C) The effect of adrenergic blocking agents on the relaxations of the stomach to stimulation of the splanchnic (S) or vagal (V) nerves at the frequencies indicated at the horizontal bars. A, gastric and blood pressure responses to sympathetic and vagal stimulation before and after guanethidine; the sympathetic relaxation is abolished but the relaxation due to the vagus is unaffected. B, gastric response to intraarterial infusion of adrenaline (indicated by the horizontal bars) and to stimulation of the vagus (V), before and after the adrenergic blocking drug, methalide; the adrenergic response is blocked except at very high concentrations whilst the vagal response is unaffected (from Fig. 3 in Martinson, 1965). (D) The effect of adrenergic blocking drugs on the relaxation of the stomach to vagal and sympathetic stimulation. Shown are the effects of a low concentration of bretyluim on responses of the isolated atropinised stomach to stimulation of the vagus (V) or the perivascular (P) nerves. The vagus was stimulated at 10 Hz for 10 sec and the perivascular nerves at 25 Hz for 10 sec. a, Shows control responses, b, shows the responses 46 min after the addition of bretylium (from Fig. 4 in Campbell, 1966a).
the site of stimulation. This would then bring muscle previously in sucrose at the sucrose-ringer interface into the ringer solution, so changing both the liquid junction potential at the interface as well as the properties of the muscle recorded from at this interface (Bennett & Burnstock, 1966). Concerns about the possibility that the hyperpolarization could be an artefact of the recording technique were only alleviated when Mollie Holman, then in the Department of Physiology, was asked to join the research program on the innervation of the taenia coli muscle. Her experience with intracellular recording from smooth muscle cells, both in the taenia coli (Bulbring et al., 1958) and in the vas deferens (Burnstock & Holman, 1961), allowed the unequivocal demonstration, in one experiment, that the hyperpolarizations in response to stimulation of the intramural nerves were not peculiar to the sucrose-gap recording technique (Fig. 7.7B). The claim was now made at a meeting of the Australian Physiological and Pharmacological Society in February 1963 in Canberra that the hyperpolarization was an inhibitory junction potential or UP (Burnstock et al., 1963b; see also 1963c). The sucrose gap was next used that year to determine that
118
HISTORY OF THE SYNAPSE
Fig. 7.7. The electrical signs of inhibition in the gastrointestinal tract: the inhibitory junction potential. (A) The first identified recording of the inhibitory junction potential in smooth muscle. Sucrose-gap records from the smooth muscle of the guinea-pig taenia coli showing inhibitory junction potentials in response to stimulation of the intramural nerves. Frequencies of stimulation: A, 1 Hz; B, 10 Hz; C, 30 Hz. Upper trace, tension; lower trace, membrane potential. Note the phase of increased firing of action potentials following the hyperpolarisations in B and C, already recognised as a ‘rebound firing’ (from Fig. 1 in Burnstock et al., 1963a). (B) The first identified intracellular recording of inhibitory junction potentials in smooth muscle. Intracellular records from the smooth muscle of the guinea-pig taenia coli showing the inhibitory junction potentials in response to stimulation of the intramural nerves. Frequencies of stimulation: A, 1 Hz; B, 2 Hz; C, 4 Hz. Upper trace, membrane potential; lower trace, tension. Note downwards deflexion represents an increase in tension (from Fig. 2 in Burnstock et al., 1963a). (C) Intracellular records of muscle fibres of the arrested frog heart showing the effects of inhibitory stimulation. A, shows hyperpolarisations due to two single vagal volleys and B the hyperpolarisation during repetitive stimuli indicated by the vertical dashes. C, shows the hyperpolarisation due to a single vagal volley at higher amplification. D, shows the hyperpolarisation due to external application of acetylcholine by a brief current pulse at the arrow (from del Castillo & Katz, 1957; reproduced in Eccles, 1961). (D) Intracellular recordings of the inhibitory junction potential in the smooth muscle of the taenia coli. The intramural nerves were stimulated a frequencies of 60, 4, 2 and 1 Hz in a, b, c and d respectively. The increased rate of firing of the action potentials (‘rebound’) was noted at the time of recording (from Fig. 8 in Bennett et al., 1966a). (E) Lack of effect of adrenergic blocking drugs on the inhibitory junction potential of the gastrointestinal tract. Shown is the lack of effect of guanethidine (10−6 gm/ml) on the inhibitory junction potential recorded in response to single pulses to the intramural nerves. Part (a) shows the junction potential before guanethidine added and (b) the junction potential 50 min after adding guanethidine. As noted at the time ‘there was no changes in the characteristics of the UP (inhibitory junction potential) in the presence of guanethidine or bretyluim’. As atropine was present in these experiments, this was the first published record of a failure to block a synaptic potential with either adrenergic or cholinergic blocking drugs (from Fig. 12 in Bennett et al., 1966a). (F) First recordings of the inhibitory junction potential in the smooth muscle of the stomach due to stimulation of the vagus nerve. Shown are inhibitory junction potentials due to single stimuli to the vagal nerve, followed by stimulation at high frequency during the period indicated by the bar. The voltage scale is 50 mV and the time marker 1 sec (from Fig. 6B in Beani et al., 1971). (G) First recordings of the UP in the smooth muscle of the distal colon to stimulation of the intramural nerves. Stimulation of the pelvic nerves only gave an excitatory junction potential (EJP). The stimulation of intrinsic neurones is indicated by I and of the pelvic nerve by P. In (a), the stimulation of the pelvic nerve preceded the stimulation of intrinsic nerves by 1050 msec. There was no interaction between EJP and IJP. In (b), pelvic stimulation preceded transmural stimulation by 140 msec. The initial stage of the EJP was normal but its subsequent time course was cut short by the UP. In (c), pelvic nerve stimulation followed transmural stimulation by 170 msec. The EJP was blocked completely by the IJP. In (d), the pelvic nerves were stimulated as the IJP reached peak amplitude. The EJP again was blocked completely. In (e), pelvic nerve stimulated 1100 msec after transmural stimulating in the last stages of the IJP. The EJP was partly blocked by the IJP. In (f), the pelvic nerve was stimulated 1300 msec after the intrinsic fibres. There was no blockade of the EJP which occurred after the decay of the IJP. All records are from the same cell using constant stimulus parameters (from Fig. 5 in Furness, 1969).
the IJP was due to an increase in the potassium permeability of the muscle cells subsequent to the secretion of transmitter; this
DISCOVERY OF OTHER TRANSMITTERS IN THE PERIPHERAL NERVOUS SYSTEM
119
was reported to the British Physiological Society by Geoff Burnstock in July (Bennett et al., 1963). The potassium dependence of inhibitory transmission was eventually described in a Nature article (Bennett, 1966b) and later confirmed by Tomita (1972). Intracellular recording techniques were established in Geoff Burnstock’s laboratory at the end of 1963. This allowed intracellular recordings of the IJP to be made (Fig. 7.7D), which were eventually written up (see below) and published (Bennett et al., 1966a). The interpretation of the hyperpolarization in response to transmural stimulation as an IJP was still met with considerable scepticism in Great Britain. This was especially the case in the Pharmacology Department at Oxford, which at that time was the leading laboratory on autonomic nervous system pharmacology. In 1965 Bulbring et al. (1965) commented that the: …‘inhibitory potential’ arising from intramural nerve stimulation: …is usually not abolished by local anaesthetics nor by cold storage for 4 days. More evidence is required before it will be possible to know whether the inhibitory potential is the result of nerve or of muscle stimulation, or both. The specific sodium action potential blocker tetrodotoxin became available in 1967, and it was soon shown that it blocked the ‘inhibitory potential’. This allowed Bulbring and Tomita (1967) to say that: All the above evidence supports the view that the inhibitory potential is not the result of muscle stimulation but that it is a junction potential produced by intrinsic inhibitory nerves as postulated by Bennett et al. (1963, 1966b) and Burnstock et al. (1963a, b). It is ironical in this regard that early records of the electrophysiological response of intestinal tissue to stimulation of intramural nerve, carried out in Bulbring’s laboratory using the technique described by Gaskell (1886), clearly show the UP (see Figures 7 & 12 in Bulbring et al., 1958). 7.3.2 Rebound excitation following inhibitory transmission: atropine-resistant excitatory junctions One of the most striking results of stimulating the intramural nerves to the gastrointestinal smooth muscle observed in the original experiments was the heightened frequency of action potential firing in the muscle when it recovered from hyperpolarization (Fig. 7.7A). Even a single IJP was often associated with muscle action potential firing (Fig. 7.8A; Bennett, 1966a). Furthermore, if the rate of spontaneous firing in the muscle was very low, then stimulating the intramural nerves at short intervals gave rise to a few inhibitory junction potentials as well as greatly accelerating the rate of impulse firing following the hyperpolarization. This accelerated rate of impulse firing was not blocked by atropine (Fig. 7.8B). Andersen and Eccles (1962; Fig. 7.8C) in Canberra had recently described accelerated firing of action potentials following the hyperpolarization due to inhibitory postsynaptic potentials in thalamic-cortical relay cells. They suggested that this was due to a kind of anodal-break excitation. Rebound had actually been described many years earlier by Kuffler and Eyzaguirre (1955) (Fig. 7.8D). We thought that on seeing the accelerated firing in the smooth muscle that this was either due to an unknown excitatory transmitter or to rebound firing following anodal-break excitation caused by the preceding UP (Bennett, 1965). Identifying the basis of the atropine-resistant excitatory response of the smooth muscle after intramural nerve stimulation is still not resolved. Further consideration of the identity of this NANC excitation will be given in section 8. Substance P was discovered by Euler and Gaddum (1931) as a consequence of their efforts to find other sources of acetylcholine than in the spleen (Dale & Dudley, 1929). Euler and Gaddum (1931) found that suitably treated extracts of intestine and brain gave rise to contraction of the rabbit’s isolated intestine in the presence of atropine. However these authors did not relate this substance to a putative role in a non-cholinergic excitatory transmission to the intestine. Later, Gaddum and Schild (1934), together with Euler (1934), investigated the chemical properties of this material that was then named substance P. However, even 20 years later, a primary interest in substance P was the possibility that it acted as a central transmitter (Amin et al., 1954) rather than it might provide an explanation for atropine-resistant excitation of the gastrointestinal tract. By that time a number of other pharmacologically active substances were known to exert affects on the gastrointestinal tract, such as histamine and serotonin, but there was no coherent attempt to use these in order to provide an explantation for the anomalies concerned with the classical autonomic paradigm as described here. Rebound firing explains the difficulties that so many investigators have had, since the time of Bayliss and Starling (Bayliss & Starling, 1899; 1900; 1901), in blocking contractions of the gastrointestinal tracts in response to parasympathetic nerve stimulation, with atropine. Inhibitory junction potentials can produce very large increases in action potential frequency (Fig. 7.8B) and consequently in contraction (Campbell, 1966a); this is especially the case with low-tone situations in which there is little spontaneous firing of action potentials. Such rebound firing is very prominent during vagal stimulation of inhibitory neurones in atonic stomach (Fig. 7.8E; Beani et al., 1971). This is the most likely reason for atropine-resistant contractions in
120
HISTORY OF THE SYNAPSE
response to parasympathetic nerve stimulation in the gastrointestinal tract, although it is clear that nonadrenergic and noncholinergic excitatory transmitters exist in the tract, as is discussed in section 7.8. Atropine-resistant contractions do not occur because (Dale 1934b): nerve impulses liberate acetylcholine so close to the reactive structures that atropine cannot intervene, whereas it can prevent its access to them when it is artificially applied from without. 7.3.3 Discovery that inhibitory junctional transmission involves non-adrenergic non-cholinergic (NANC) transmitters A serious problem arose in our laboratory in early 1963, with the discovery that the IJP could not be blocked in the taenia coli muscle with guanethidine or bretylium, the adrenergic-nerve blocking drugs that had become available at the end of the 1950s (Fig. 7.7E). More uncertainty arose with the publication on the one hand of Martinson and Muren (1963)s paper showing that vagal inhibition of the stomach could not be blocked with sympatholytics and on the other hand the paper of Paton and Vane (1963) suggesting that vagal inhibition of the stomach could be reduced by TM10. Despite this we decided to write up our observations on the IJP for a note to Nature, which was published in November (Burnstock et al., 1963a). No mention is made in this note of the difficulties we had in affecting either the relaxation or the UP with adrenergic blocking drugs. By 1965 we had proved to our satisfaction that an IJP existed in the gastrointestinal tract which was not blocked by bretylium or guanethidine (Fig. 7.7E; Bennett et al., 1966a). This inhibition clearly traced its origins to the relaxation first observed by Openchowski (1889), Gaskell (1886) and Langley (1898b). The existence of these synapses neither supported the idea that the gastrointestinal tract receives antagonist innervation only from sympathetic and parasympathetic nerves, nor that synaptic transmission is mediated solely by noradrenaline and acetylcholine, as in the classical autonomic paradigm of Dale (1934b) (Fig. 7.2A). The doubts have been noted concerning the existence of a relaxation of the gastrointestinal tract to parasympathetic nerve stimulation (see for example, Bayliss & Starling, 1900, mentioned above) and whether such relaxations were mediated by the secretion of noradrenaline (compare for example, Paton & Vane (1963) with Martinson (1965)). However the recording of single inhibitory junction potentials in the muscle of the intestine, together with their resistance to adrenergic blockade, established unequivocally for the first time the existence of neurones in the autonomic nervous system that used transmitters which were neither adrenaline nor acetylcholine. In the laboratory these junctions were referred to as non-adrenergic non-cholinergic (NANC). The electrical signs of NANC transmission were soon obtained for vagal inhibition of the stomach (Beani et al., 1971; Fig. 7.7F) as well as for intrinsic inhibitory neurones in the distal colon (Furness, 1969; Fig. 7.7G). Rand and Mitchelson (1986) provide a different explanation for the emergence of the concept of NANC neurones in the autonomic nervous system. They comment that this occurred: …when Burnstock, Campbell and Rand (1966) showed that their were inhibitory neurones in the taenia of the guinea pig caecum that were clearly distinguishable from the noradrenergic sympathetic neurones. However, as detailed above, there were many published and conflicting claims on this issue based on the same kind of mechanical studies of gastrointestinal tissue as that used by Burnstock et al. (1966) (viz Martinson, 1965; Paton & Vane, 1963). Why, then, should this have amounted to the radical department from orthodoxy that Rand and Mitchelson (1986) suggest? This is particularly the case when the extent to which Burnstock et al. (1966) qualified the interpretation that they were dealing with a new class of synapse is considered: It is possible that they are adrenergic but attempts to demonstrate (or disprove) this by depletion of catecholamines with reserpine, blockade of a, or b-adrenoceptors with appropriate blocking drugs or enhancement of responses to released catecholamines with cocaine have given equivocal results. It seems more likely that three discoveries turned over the classical autonomic paradigm: first, the electrophysiological description of inhibition in the gastrointestinal tract; second, the failure to block this with bretylium, and third the discovery of rebound excitation as an explanation for atropine resistant excitation. This seems to have provided the radical departure from orthodoxy.
DISCOVERY OF OTHER TRANSMITTERS IN THE PERIPHERAL NERVOUS SYSTEM
121
Fig. 7.8. The discovery of rebound increases in the frequency of action potential firing following hyperpolarisations due to stimulation of nerves in the peripheral and central nervous systems. (A) The effect of stimulating the intramural nerves of the taenia coli with a single impulse during a high rate of spontaneous action potential firing. Note the decrease in the intervals between the action potentials immediately following the inhibitory junction potential (from Fig. 1 in Bennett, 1966a). (B) The effect of stimulating intramural nerves of the taenia coli with repetitive impulses during a low rate of spontaneous action potential firing. There is a very clear increase in the rate of action potential firing after inhibition. Frequency of stimulation in a, 1 Hz; b, 10 Hz (from Fig. 4 in Bennett, 1966a). (C) The effect of inhibitory post-synaptic potentials in thalamic cortical relay cells on the firing of action potentials following the inhibition. Parts A and B show intracellular records of neurons in the ventrobasal complex of the thalamus recorded with a potassium citrate-filled microelectrode. The inhibitory synaptic potentials were evoked by stimulation of the cortical terminals of the thalamocortical relay cells. Note the depolarisation phase immediately following each hyperpolarisation; the initial phases of several truncated action potentials are shown on top of this depolarisation (from Fig. 1 in Andersen & Eccles, 1962). (D) The initiation of action potentials by stimulation of the inhibitory nerves to the crustacean stretch receptor. In a, the inhibitory nerve was stimulated in a quiescent receptor at 23 Hz for the period between the arrows; this was followed by a series of action potentials giving ‘post inhibitory excitation’. In b the inhibitory nerve was stimulated with 6 impulses between the arrows, giving a single post inhibitory action potential (from Fig. 9 in Kuffler & Eyzaguirre, 1955). (E) Rebound firing of action potentials in smooth muscle cells of the stomach to stimulation of either the vagus or sympathetic nerves. Part a shows vagal stimulation at 20 Hz for 3 sec (horizontal bar) gives rise to a hyperpolarisation of about 20 mV; this is followed in this quiescent cell by rebound firing of action potentials within 4 sec. Part b shows sympathetic nerve stimulation at 20 Hz giving rise to a large hyperpolarisation in a quiescent cell, followed by a rebound firing of high-frequency action potentials (from Figs. 6C and 8C in Beani et al., 1971).
7.4 NANC transmission: the new autonomic paradigm The enteric nervous system of the gastrointestinal tract Mall (1896) described the relaxation of the circular muscle on the anal side of a bolus moving down the intestine, together
122
HISTORY OF THE SYNAPSE
with the contraction of muscle oral to it. Bayliss and Starling (1899) then gave a very comprehensive description of this peristalsis, after severing the sympathetic and vagal nerves. This showed that the phenomenon arose from an intrinsic nervous mechanism, presumably that identified as the enteric nervous system by Auerbach (1862; 1864). The following classical description of peristalsis, quoted here in full because of its importance to the arguments developed in this chapter, was given by Bayliss and Starling (1899; Fig. 7.9A): In the experiment from which the curves Fig. 7.9Aa were obtained two enterographs were placed at right angles to one another at a point 130 cm from the pyorus. The position of the levers is shown in Fig. 7.9A. Parts a and b being the levers of the longitudinal enterograph (a being the movable lever), c and d the levers of the enterograph recording the contractions of the circular muscle. At the beginning of the observation the intestinal wall was contracting rhythmically, the contractions affecting both coats, and being synchronous in both. At (A), a bolus made of cotton-wool coated with Vaseline was inserted by an opening into the intestine 4.5 in above the enterographs. It will be seen that the contractions of the circular coat cease instantly, and this inhibition is accompanied by a gradually increasing relaxation. There is some relaxation of the longitudinal coat, but the rhythmic contractions do not altogether cease. On inspecting the intestine it was seen that the introduction of the bolus caused the appearance of a strong constriction above it. This constriction passed downwards, driving the bolus in front of it. The numbers above the tracing of the circular fibres indicate the distance of the bolus in inches from the uppermost enterograph lever. At B, the bolus had arrived at the upper longitudinal lever and at C had passed this and was directly under the transverse enterograph, or a little below it. At this point a strong tonic contraction of both coats occurs, expelling the bolus beyond the levers. This strong contraction passes off to be succeeded by another, which like the first is moving down the intestine. This description of peristalsis, and the workings of an enteric nervous system employing inhibitory neuromuscular synapses, was not improved upon until 1917. At that time, Trendelenburg (1917) determined the conditions in which reflexes of the longitudinal and circular muscle layers could be elicited from isolated segments of small intestine. He showed that the longitudinal muscle contracts first when the lumen of the intestine is distended, causing a shortening of the intestinal segment; this was termed the ‘preparatory phase’ of peristalsis (Fig. 7.9B). A contraction wave of the circular muscle then follows while the longitudinal muscle relaxes and this proceeds in an anal direction expelling the contents; Trendelenburg termed this the ‘emptying phase’ of peristalsis (Fig. 7.9B). The cycle is repeated when the lumen fills again, and the cycle continues until pressure in the lumen is restored to its original low level; at this time both longitudinal and circular muscles relax (Fig. 7.9B) (see Kosterlitz, 1967). The contraction of the longitudinal muscle during peristalsis involves a graded-reflex contraction with superimposed rapid-rhythmic contractions (Fig. 7.9C) (Kosterlitz, 1967). At very low pressures the longitudinal muscle contracts and the circular muscle remains relaxed (Fig. 7.9C). If a preparation is used which requires greater than normal luminal pressures to elicit peristalsis, the faster contractions of the longitudinal muscle of the peristaltic reflex are then superimposed on the slower graded reflex contraction (Fig. 7.9C). The relaxations of the longitudinal muscle are due to an active inhibitory synapse; luminal distention relaxes the muscle if it has been previously contracted by acetylcholine (Watt quoted by Kosterlitz, 1967). The observation that this relaxation is not blocked by the adrenergic blocking drugs, guanethidine and bretyluim, indicated an intrinsic inhibitory mechanism which is not mediated by noradrenaline. The IJP due to stimulation of the intramural nerves, discovered in the taenia coli (Bennett et al., 1966a), is very likely to mediate the relaxation phase of peristalsis in both the circular and longitudinal muscle coats. Another possibility is that the inhibitory junction potential only mediates the inhibition due to extrinsic nerves, like the vagus, which impinge on inhibitory neurones. However, Furness (1969) showed that inhibitory junction potentials could be recorded in response to stimulation of the intramural nerves of the distal colon, whereas stimulation of the pelvic nerves only gave excitatory junction potentials (EJPs) (Fig. 7.10A). The inhibitory junction potentials recorded by Furness (1969) are very likely to be part of an intramural inhibition that mediates peristalsis. The inhibitory neurones of the enteric nervous system receive a synaptic input from enteric sensory neurones, with terminals in the mucosa. Hirst and McKirdy (1974) showed that distension of the isolated small intestine gives rise to an inhibitory junction potential on the anal side of the stimulated region, and that this can be blocked with tubocurarine (Fig. 7.10B). This experiment then provided electrophysiological evidence that enteric neurones synapsing on muscle receive at least one synapse from other enteric neurones. Kuntz (1922) as well as Feldberg and Lin (1949), had suggested this earlier on the basis of mechanical measurements. This led Hirst and McKirdy (1974) to suggest that the descending inhibitory pathway from the oral to the anal side of the circular muscle involved in the peristaltic reflex receives inputs from interneurones; these in turn are innervated by sensory neurones with receptor terminals for distension (Fig. 7.10C). This idea calls forth a complementary pathway for excitation of the circular muscle during peristalsis, only in this case we are dealing with an ascending pathway from anal to oral as shown in Fig. 7.10D. The changes in the contraction of the longitudinal circular muscles during peristalsis first described adequately by Bayliss and Starling (1899), and involving an active inhibition of the
DISCOVERY OF OTHER TRANSMITTERS IN THE PERIPHERAL NERVOUS SYSTEM
123
Fig. 7.9. The peristaltic reflex: descending inhibition and ascending excitation of the gastrointestinal tract. (A) The original record of Bayliss and Starling (1899) showing the peristaltic movements as a bolus passes down the intestine (for their own description of this record, see text). The upper trace is for contraction of the longitudinal muscle (L) and the lower for the circular muscle (C). It should be noted that relaxation is indicated by an upward deflection. At the time indicated by A, a bolus was inserted 4.5 in oral to the point at which the recording is made. Both the longitudinal and circular muscles relax in response to this. The bolus moves towards the recording point, with the letters indicating times at which the bolus was situated at 3, 2 and 1 in oral to the recording point. Parts B and C indicate the times between which the bolus was beneath the recorder. When the bolus was anal to the recording site, that is, beyond C, there was contraction of the muscle. (B) Examples of the Trendelenburg method for recording peristaltic waves in the small intestine. Upper trace shows isotonic recording from the longitudinal muscle, with contraction upwards; lower trace shows recording of the filling of the lumen, with increased filling downwards. Three periods of distension of the lumen are shown, each to a pressure of 2.5 cm H2O for 45 secs. Note the repeated peristaltic waves of the longitudinal and circular muscles in each episode (from Fig. 1 in Kosterlitz, 1967). (C) Examples of the tonic increase in the longitudinal muscle with peristaltic contractions superimposed during filling of the lumen of the small intestine, using the Trendelenburg method. At 1, a 1-cm H2O increase in the intraluminal pressure was produced, giving only a graded reflex contraction of the longitudinal muscle. At 1.5 and 3.0 cm of H2O increase in intraluminal pressure, fast contractions of the longitudinal muscle are superimposed on the graded reflex contraction. These fast contractions only occur when the circular muscle contracts and expels fluid from the lumen. In this preparation peristaltic reflexes are elicited only intermittently compared with those in the preparation in B (from Fig. 2 in Kosterlitz, 1967).
muscle ahead of a bolus (Fig. 7.6A), can be realised by the enteric nervous pathway involving inhibitory neurones, given in Fig. 7.10C & 7.10D, Fig. 7.11A & 7.11B.
124
HISTORY OF THE SYNAPSE
Fig. 7.10. Inhibitory junction potentials mediating the relaxation component of the peristaltic reflex. (A) In the distal colon, inhibitory junction potentials are not elicited by stimulating extrinsic nerves, only the intramural nerves. These are likely to produce the relaxation of the colon in peristalsis. An inhibitory junction potential is shown in the circular coat to stimulation of intramural nerves, but not to stimulation of the pelvic or sympathetic nerves. This potential is not blocked by adrenergic blocking drugs (from Fig. 1b in Furness, 1969). (B) Inhibitory junction potentials in the small intestine due to distension. Distension of the muscle occurred between the arrows (i) before and (ii) after blocking the enteric ganglia with tubocurarine. Distension was oral to the recording site. Ganglion blockade reduces the hyperpolarisation considerably, indicating that at least one cholinergic synapse intervenes between the point of distension and the analy located inhibitory neurone (from Fig. 7b in Hirst & McKirdy, 1974). (C) The first neuronal model of the descending inhibitory pathway responsible for relaxation during peristalsis. Afferent processes of the sensory neurone (cell body, stippled circle) are stimulated by distension. Efferent processes synapse on the cell body of an interneurone (open circle) or may synapse directly on the cell body of an inhibitory neurone (large filled circle). Inhibitory nerves pass in an anal direction. The inhibitory transmitter is assumed to be released from the varicosities, indicated by small filled circles (from Fig. 8 in Hirst & McKirdy, 1974). (D) Schematic drawing illustrating basic circuits in motility control. The sensory neurons project 2–3 mm around the intestinal muscle, make contact with excitatory interneurons that project along the intestine to innervate circular muscle motoneurons, longitudinal muscle motor neurons, and other longitudinally projecting interneurons. The circular muscle motoneurons have a substantial projection around the intestine, perhaps up one-half of the circumference. All the cell bodies lie in the myenteric plexus (after Fig. 14 in Bomstein et al., 1991).
The role of sympathetic synapses in the gastrointestinal tract If the parasympathetic nerves are capable of supplying both a cholinergic excitation of the gastrointestinal tract and a nonadrenergic inhibition, what then is the role of the sympathetic nerves? Early on, Van Braam-Houckgeest (1874) had claimed that the sympathetic nerves provided a direct inhibitory innervation of the muscle, an idea which had become the accepted
DISCOVERY OF OTHER TRANSMITTERS IN THE PERIPHERAL NERVOUS SYSTEM
125
Fig. 7.11. The extrinsic nerve synapses on inhibitory neurones in the gastrointestinal tract. (A) Vagal nerve synapses on inhibitory neurones in the stomach. Shown is a schematic representation of reflex mechanisms producing gastric receptive relaxation through non-adrenergic neurones to the stomach that are innervated by the vagus. Sensory terminals are indicated by filled circles. The enteric connection to inhibitory neurones mediating peristalsis are not shown. Abbreviations: VC=vomiting centre; TZ=trigger zone for centrally acting emetic drugs (from Fig. 3 in Abrahamsson, 1973). (B) Vagal nerve synapses on inhibitory neurones in the proximal colon. Shown is a schematic representation of neural pathways involved in the descending inhibitory component of peristalsis. The cell body of the cholinergic excitatory neurone is shown by an open circle. The nonadrenergic non-cholinergic inhibitory neurone by a filled circle. A non-adrenergic inhibitory neurone that is sensitive to blockade of 5hydroxytryptamine synthesis is shown by a crossed circle. Note that in this representation the same inhibitory neurone (filled circle) is shown to receive both intrinsic sensory input from mechanoreceptors involved in the peristaltic reflex as well as a modulatory extrinsic input from the vagus nerve (from Fig. 9 in Jule, 1980).
paradigm. Garry (1934), in his influential review on the movements of the large intestine, had stressed the extent to which sympathetic nerves provide a tonic inhibitory drive to the gastrointestinal tract. The answer to the question concerning the role of the sympathetic nerves came from two different sources in the early 1960s. The great Swedish School of neurohistologists, beginning with Hillarp, Falck, Fuxe and Dahlstrom, had introduced techniques for identifying catecholamines in nerve fibres. Amongst their first discoveries was that the sympathetic nerves containing noradrenaline do not enter the smooth muscle bundles of the gastrointestinal tract. Rather, they innervate the blood vessels in the musculature as well as neurones in the enteric nervous system (see e.g., Norberg, 1964). This also seemed to be true for the taenia coli (Fig. 7.12B) (Bennett & Rogers, 1967), in which the IJP had been discovered. In addition, when the distribution of a and (adrenoceptors was determined for the gastrointestinal tract, the -receptors could only be found on neurones in the enteric plexus whereas receptors could only be found in the muscle (Kosterlitz & Watt, 1964). Furthermore, Gillespie (1962) had observed that stimulation of sympathetic nerves to the gastrointestinal tract did not give rise to a discrete IJP. Membrane hyperpolarization and the depression of muscle action potential firing could only be detected at relatively high frequencies of nerve stimulation, in excess of 5 Hz (Fig. 7.12 A). This was also found to be the case for the taenia coli (Bennett et al., 1966b). The picture which emerged from these studies (and recently challenged by Andrews & Lawes, 1984) is that adrenergic nerves do not innervate the muscle of either the stomach (Abrahamsson, 1973) (Fig. 7.12C) or the non-sphincter parts of the intestine (for a review, see Furness & Costa, 1987). Any hyperpolarisation of these muscles due to high frequency nerve stimulation occurs as a consequence of the overflow of noradrenaline from junctions on blood vessels and synapses on enteric neurones. The sympathetic innervation of the enteric nervous system is principally a presynaptic inhibitory innervation of cholinergic nerve terminals; these arise from sensory pathways to cholinergic excitatory motor neurones within the enteric nervous system or directly from cholinergic excitatory nerves to the stomach (Fig. 7.12C). The new autonomic paradigm for the gastrointestinal tract Campbell and Burnstock wrote in a review (Campbell & Burnstock, 1968) that: This appraisal has been attempted because of two recent findings made on the mammalian gut.
126
HISTORY OF THE SYNAPSE
Fig. 7.12. The mechanism by which sympathetic nerves inhibit the gastrointestinal tract. (A) The effect of stimulating sympathetic nerves to the smooth muscle of the intestine at different frequencies (for the period indicated by the horizontal line) on the membrane potential and spontaneous action potential firing in the muscle. The upper line in each record is tension, the lower is membrane potential. Part A shows the spontaneous activity in the absence of nerve stimulation; B, C, D and E show the effect of the frequencies of nerve stimulation shown to the left of each trace. Downward deflection of the contraction trace indicates an increase in tension. Note that there is no hyperpolarisation of the membrane and decrease in spontaneous action potential firing until frequencies of 10 Hz are reached (from Fig. 8 in Gillespie, 1962). (B) Distribution of noradrenergic nerve fibres in the muscle and enteric nervous system of the guinea-pig taenia coli. Fluorescent structures following formaldehyde condensation method for monamines. The nerve fibres (n) on the surface of the ganglion cells (G) are intensely fluorescent. Fluorescent nerve fibres (n) are sparsely scattered throughout the muscle bundles and in the connective tissue bands between the bundles; these nerve fibres are often associated with blood vessels. Part E indicates the serosal surface of the taenia coli (from Fig. 20 in Bennett & Rogers, 1967). (C) Diagram showing the innervation of the enteric nervous system by the sympathetic (splanchnic) and vagus nerve, derived on the basis of studies on the inhibitory control of gastric motility. Note that the sympathetic nerves do not inhibit the muscle directly, through neuromuscular inhibitory junctions. Rather, the sympathetic nerve terminals inhibit transmission from the vagal excitatory preganglionic nerve terminals, probably through a form of presynaptic inhibition (from Fig. 1 in Abrahamsson, 1973).
These two recent findings necessitated a new paradigm. First, it has been shown that electrical stimulation of inhibitory nerves in the longitudinal muscle of the guinea-pig caecum can cause a rebound excitation of the muscle and this reexamination is made more pertinent by a second
DISCOVERY OF OTHER TRANSMITTERS IN THE PERIPHERAL NERVOUS SYSTEM
127
discovery made on mammalian gut, namely, that there are intramural inhibitory neurones in Auerbach’s plexus that differ pharmacologically from adrenergic nerves and that may not even be adrenergic. We have seen the extent to which the discovery of the electrical signs of inhibition to the smooth muscle of the gastrointestinal tract illuminated its function. Firstly, it identified the IJP involved in the relaxation phase of peristalsis, controlled by the enteric nervous system; this became the focus of attempts to provide a detailed circuitry for this nervous system (Sections 7.3 and 7.4; Figs 7.10C and 7.12C). Secondly, the discovery of the IJP settled once and for all the question concerning the existence of a vagal inhibition to the gastrointestinal tract extending from the stomach to the proximal colon (Sections 7.2 and 7.3; Figs. 7.13A and B). Thirdly, the recognition that an atropine-resistant rebound firing of action potentials following intramural nerve stimulation and the IJP provided a description of the ‘atropine-resistant’ excitation of the gastrointestinal tract (Sections 7.2 & 7.3). Finally, the IJP was recognised unequivocally as the first synaptic potential in the peripheral nervous system to be unaffected by adrenergic or cholinergic blocking drugs (Sections 7.2 & 7.3). It then gave rise to the concept of NANC transmission and all that has arisen for neuropharmacology from that idea. The identity of NANC transmitters in the smooth muscle of the gastrointestinal tract has been the subject of a number of recent reviews (Hoyle & Burnstock, 1989a; Makhlouf & Grider, 1993; Sanders & Ward, 1992; Stark & Szurszewski, 1992) and a monograph (Furness & Costa, 1987) and is considered in more detail in Sections 7.5–7.8. It was Gaskell who emphasised the antagonistic innervation of effector tissues by the autonomic nervous system and Dale who introduced the terms adrenergic and cholinergic for the description of this innervation (Dale, 1934a). The seductive symmetry of these concepts was emphasised by yet another dualism, namely the division of the autonomic nervous system into sympathetic and parasympathetic components by Langley. The temptation was then irresistible to place these three concepts together into what became the classical autonomic paradigm (Fig. 7.2). A new paradigm for the gastrointestinal tract is presented in Fig. 7.13. One hopes that it escapes the criticism already leveled at the old paradigm, namely that sufficient evidence existed at the time of its composition to make it invalid. The discovery of NANC transmission is the basis for this paradigm. 7.5 Contemporary views on the identity of NANC inhibitory transmitters This review is primarily concerned with historical aspects of the discovery that there are more transmitters than acetylcholine and noradrenaline mediating autonomic neuroeffector transmission. However it would be incomplete without consideration of modern views as to the identity of these transmitters and their function, at least as far as inhibitory transmission is concerned. Gillespie (1962) gives an interesting account of some of the literature that emerged in the 1970s concerning the identity of NANC inhibitory transmitters. The contemporary view of the problem is considered below. 7.5.1 Identification of a fast and a slow IJP Since the discovery of the IJP 35 years ago (Bennett et al., 1963; Burnstock et al., 1963a) several putative transmitters have been claimed to mediate this form of NANC transmission. These include adenosine triphosphate (ATP; Burnstock et al., 1970; Burnstock, 1972), nitric oxide (NO; Sanders & Ward, 1992), vasoactive intestinal peptide (VIP; Makhlouf & Grider, 1993), carbon monoxide (CO; Farrugia et al., 1993) and pituitary adenylyl cyclase-activating peptide (PACAP; Jin et al., 1994). As already noted, in the first publication on the electrical signs of NANC transmission Bennett et al. (1993) reported that the amplitude of the IJP acted as a potassium electrode in response to changes in the extracellular potassium concentration, an observation that later was confirmed by Tomita (1972). A discovery that greatly clarifies the possible contributions of putative transmitters to the IJP is that apamin, a neurotoxic polypeptide from bee venom, blocks a fast component of the IJP (IJPf) to reveal a slower IJP (IJPs; e.g., in stomach muscle; He & Goyal, 1993). This occurs as a consequence of apamin blocking a small conductance calcium-activated potassium channel that mediates the hyperpolarizing effects of some transmitters on smooth muscle (Banks et al., 1979; Maas et al., 1980; Capiod & Ogden, 1989). A more recent observation is that IJPf is blocked by -conotoxin GVIA whereas IJPS is not (Bridgewater et al., 1995). The use of apamin to distinguish between IJPf and IJPs, with perhaps the additional use of -conotoxin GVIA, greatly assists in the problem of identifying the transmitters responsible for the IJP. Thus, apamin has been shown to reduce or block IJPf in the guinea-pig colon (Ganitkevich et al., 1983), canine ileocolonic sphincter and guinea-pig taenia coli (Maas & den Hertog, 1979) although it blocks completely the UP in other smooth muscles, such as the circular smooth muscle of the guinea-pig ileum (Niel et al., 1983a; He & Goyal, 1993). These results then point to at least two transmitters acting at inhibitory junctions in the guinea-pig taenia coli and colon as well as the canine ileocolonic sphincter but perhaps only one in the circular muscle of the guinea-pig ileum. Indeed in some
128
HISTORY OF THE SYNAPSE
Fig. 7.13. The new autonomic paradigm for innervation of the gastrointestinal tract. This arose from the discovery of inhibitory junction potentials resistant to adrenergic and cholinergic nerve blockade (at NANC synapses), as well as NANC excitation in 1963, together with determination of the distribution of the NANC synapses in the succeeding 10 years. Part A shows the parasympathetic and sympathetic innervation of the enteric neurones of the gastrointestinal tract; in the vagus nerve the continuous line is excitatory to the inhibitory NANC motoneurones ( ) whereas the dot-dash line is excitatory to the excitatory motoneurones ( ). In the sympathetic supply the dashed lines eventually synapse on the excitatory terminals of vagal and pelvic nerve endings on the motoneurones. Note that intrinsic synapses on the motoneurones are not shown. Part B shows in greater detail the synapses on the inhibitory and excitatory motoneurones in the enteric nervous system (intrinsic connection to the motoneurones not shown). The sympathetic axons terminate in white inhibitory varicosities on the black vagal excitatory varicosities on the motoneurons. These in turn project to the circular muscle layer to form varicose junctions with smooth muscle cells. Abbreviations: S=seroa; LM=longitudinal muscle; MP=myenteric plexus; CM=circular muscle; SP=submucosa plexus (details not shown); MM=muscularis mucosa; M=mucosa.
cases it is possible that one transmitter is responsible for both IJPf and IJPs, as seems to be the case for NO in the canine pylorus and ileocolonic sphincter (Ward et al., 1992c; Bayginov & Sanders, unpublished observations). The importance of this distinction between IJPf and IJPs has recently been highlighted by experiments which show that peristalsis in the guinea-pig small intestine is disrupted completely in the presence of both apamin and the NO synthase blocker L-NAME (Waterman & Costa, 1994). This occurs as a consequence of IJPf being required for the propagation of circular muscle contraction and IJPs (for which NO is the putative transmitter) for setting the threshold for peristaltic emptying. It is likely that in many parts of the gastrointestinal tract, such as the circular muscle of the guinea-pig ileum, IJPs is
DISCOVERY OF OTHER TRANSMITTERS IN THE PERIPHERAL NERVOUS SYSTEM
129
mediated primarily by NO (Lyster et al., 1992). There are a number of conjectures for the role of NO and the other putative inhibitory neuroeffector transmitters such as VIP, ATP, CO and PACAP in the process of inhibitory transmission to smooth muscle (Furness et al., 1995). These will now be reviewed in turn. 7.5.2 Contributions of nitric oxide to the IJPs Nitric oxide synthase (NOS)-containing neurones are found throughout the gastrointestinal tract of all vertebrate species (reviewed in Sanders & Ward, 1992; Furness et al., 1994). In addition, antibodies to neuronal and constitutive NOS label the interstitial cells of Cajal (ICC) found between the circular and longitudinal muscle layers, but do not label the smooth muscle cells themselves (Xue et al., 1994; Young et al., 1993). Relaxation of gastric smooth muscle in response to stimulation of the NANC nerves in the vagus is partly blocked by L-NAME (NG-L-arginine-methyl-ester), which blocks NOS (Desai et al., 1991a; Allescher et al., 1992). Relaxation due to stimulation of NANC nerves is blocked partly by NOS inhibitors in the canine terminal ileum (Boeckxstaens et al., 1990; 1991a), canine colon (Huizinga et al., 1992), rat stomach (Lefebvre et al., 1992), rat gastric fundus (Boeckxstaens et al., 1992), canine lower oesphageal sphincter (De Man et al., 1993), canine jejunum (Stark et al., 1991), dog duodenal longitudinal muscle (Toda et al., 1990), canine proximal colon (Ward et al., 1992a), canine ileocolonic sphincter (Ward et al., 1992c), and the opossum oesophageal sphincter (Tottrup et al., 1991). The adaptive relaxation of the stomach, mediated by intrinsic reflexes, is blocked by L-NAME (Desai et al., 1991b). Stimulation of NANC transmitting nerve terminals in the smooth muscle of the rat colon leads to the release of NO (Grider, 1993) as does stimulation of these terminals to the rat gastric fundus (Boeckxstaens et al., 199 1a), and canine ileocolonic junction (Boeckxstaens et al., 1991b). Such NO releasing nerves have been termed nitrergic (Li & Rand, 1990). A tonic level of NO may be released continually from nitrergic nerve terminals as oxyhaemoglobin depolarises smooth muscle of the canine ileocolonic sphincter (Ward et al., 1992c). The IJP in response to NANC stimulation is reduced by L-NAME in the canine jejunum (Stark et al., 1991), canine ileoco-Ionic sphincter (Ward et al., 1992c), canine proximal colon (Dalziel et al, 1991), and opossum esophagus (Du et al., 1991). Exogenous NO (supplied by sodium nitroprusside), increases cGMP levels in guinea-pig taenia coli and gastric smooth muscle cells (Jin, et al., 1993a), dog proximal colon (Ward et al., 1992a) and opossum lower esophageal sphincter (Torphy et al., 1986). It is likely that NO contributes to IJPs as apamin has no effect on the exogenous NO-induced relaxation in the circular muscle of the human colon (Boeckxstaens et al., 1993), guinea-pig proximal colon (Briejer et al., 1995), rabbit distal colon (Ciccociioppo et al., 1994), canine ileoco-Ionic junction (De Man et al., 1993) and the guinea-pig trachea (Ellis & Conanan, 1994). Furthermore, the apamin-insensitive IJPs in the circular muscle of the guinea-pig ileum (He & Goyal, 1993) and the circular muscle of the human colon (Keef et al., 1993) is significantly reduced by NOS blockers. One caveat to the generalization that NO is likely to contribute to IJPs is that the hyperpolarization produced by exogenous NO in the rat proximal colon (Serio et al., 1995), canine pyloric sphincter (Bayguinov & Sanders, 1993) and canine proximal duodenum (Bayguinov et al., 1992) is reduced by apamin. Whereas the identity of NO as at least one of the principal inhibitory transmitters to the rest of the gastrointestinal tract seems well established for most vertebrate species, this is not the case for the taenia coli. The NOS containing nerves are present in the taenia coli (Furness et al., 1994). However blocking NOS with NW-nitro-L-arginine or removing NO with oxyhaemoglobin does not affect the IJP in response to intramural nerve stimulation in the presence of atropine (Bridgewater et al., 1995). The lack of effect of NOS inhibitors on the IJP seems to be conditional on the presence of atropine. As the overflow of acetylcholine from the taenia on nerve stimulation is unaffected by NO donors or by the inhibition of NOS it seems unlikely that NO acts to inhibit the release of acetylcholine, so assisting in increasing the size of the IJP. Acetylcholine probably acts to allow the release of endogenous NO or to inhibit the release of an excitatory substance such as a tachykinin that usually masks the effects of endogenous NO. In either scenario, NO is a principal inhibitory transmitter to the taenia. However, this leaves open the question of the contribution of NO to the generation of IJPf and IJPs in the taenia (Bridgewater et al., 1995). The most comprehensive model of the role of NO in generating the IJP has been offered by Sanders and his colleagues (Sanders et al., 1992). They argue that release of NO from the nitrergic nerve terminals which innervate both the circular smooth muscle and the ICC in the canine colon, leads to an elevation of cGMP and the opening of calcium channels in the ICC. The consequent elevation of NO in the ICC greatly amplifies the amount of NO available to act on the smooth muscle cells. The cGMP in the smooth muscle cells is therefore is elevated greatly with the effect of increasing protein kinase G which in turn acts to open potassium channels and close calcium channels in the smooth muscle cell membrane, so leading to relaxation of the muscle (Fig. 7.14A; Publicover et al., 1993; Shuttleworth & Sanders, 1996). Evidence for this model comes from a number of different experiments. These include the observed increase in cGMP immunohistochemically in both ICC cells and smooth muscle cells following their exposure to NO or stimulation of intramural nerves (Shuttleworth et al., 1993). Elevation of calcium in ICC’s can also be induced by application of an L-type calcium channel agonist which leads to activation of ryanodine receptors and the release of calcium from internal stores; this in turn leads to generation of NO that
130
HISTORY OF THE SYNAPSE
Fig. 7.14. Models of the action of NANC inhibitory transmission. Upper diagram illustrating a proposed mechanism for the release of NO producing IJPs. NO released from nitrergic nerves causes an increase of intracellular calcium in the interstitial cells of Cajal (ICC) which induces synthesis of NO in these cells. The release of this NO causes elevation of calcium in adjacent ICCs, leading to a positive feedback generation of NO release. This NO acts on the smooth muscle cells to decrease their calcium concentration, so producing relaxation (from Publicover et al., 1993). Lower diagram showing a proposed mechanism for the release of VIP producing an IJP. VIP released from NANC terminals interacts with VIP receptors on smooth muscle cells to generate NO in the muscle cells. This NO diffuses to NANC nerve terminals where it facilitates further VIP release. The VIP also interacts with other VIP receptors on the smooth muscle cells to activate adenylate cyclase and so generate adenosine 3 , 5 -cyclic monophasphate (cAMP). It will be noted that NO itself is released by these NANC nerve terminals to facilitate the release of NO from muscle cells as well as to modulate VIP release (from Makhlouf & Grider, 1993).
diffuses to adjacent smooth muscle cells and lowers their calcium content. Elevation of calcium in smooth muscle cells using the same approach does not modulate the calcium level in adjacent smooth muscle cells (Publicover et al, 1993). The model that emerges from these studies is that the ICC act to produce a powerful amplification of the initial release of NO from nitrergic nerves and so greatly increase the changes in adjacent smooth muscle cells that lead to their relaxation. The only source of elevation of cGMP in the muscle according to this scheme comes from the interaction of NO released by nitrergic nerve terminals and NO released by ICC cells with the guanylate cyclase in the smooth muscle cells. There is then no constitutive NOS in the smooth muscle cells. 7.5.3 Contributions of vasoactive intestinal peptide (VIP) to IJP VIP-containing neurones are found extensively throughout the gastrointestinal tract where they also contain NOS (Furness et al., 1992). For example about one-third of the nerves innervating the taenia coli contain vasoactive intestinal peptide (VIP; Furness et al., 1992). These nerves are known to have projections that are the same as those of the inhibitory motoneurones to the taenia (Furness et al., 1981). Anti-serum to VIP reduces relaxation due to NANC nerve stimulation and completely abolishes relaxation to exogenous VIP in the smooth muscles of the rabbit internal anal sphincter, opossum lower oesophageal sphincter, canine esophageal muscle, and rat gastric fundus (Biancani et al, 1985; Goyal et al., 1980; Grider & Rivier, 1990; Behar et al., 1989; Li & Rand, 1990). The VIP is released from the rat colon, guinea-pig stomach and taenia coli in response to NANC stimulation (Grider & Jin, 1993; Grider et al., 1985, 1992). Both VIP and NO may be secreted in stoichiometrically equivalent amounts on stimulation of NANC nerves (Grider & Jin, 1993). Exogenous VIP increases cAMP in isolated smooth muscle cells of the guinea-pig taenia-coli and stomach as well as the cat lower esophageal sphincter without altering levels of cGMP (Jin et al., 1993b; Szewczak et al., 1990). Blocking cAMP with a protein kinase A inhibitor prevents most of the relaxing effects of
DISCOVERY OF OTHER TRANSMITTERS IN THE PERIPHERAL NERVOUS SYSTEM
131
exogenous VIP in the rat colon (Grider, 1993). How this increase in cAMP is linked to membrane hyperpolarisation and relaxation is not known. Nicotine evokes the release of VIP in the taenia coli in a calcium-dependent way, which is accompanied by relaxation of the muscle (Iselin et al, 1988). Exogenous VIP relaxes the taenia coli, an effect that is blocked by the VIP antagonist alpha-chymotrypsin, which does not affect the relaxation due to intramural nerve stimulation (Mackenzie & Burnstock, 1980). The VIP is a candidate transmitter for IJPs in the circular muscle of the guinea-pig ileum (Crist et al., 1992), rat duodenum (Mule et al., 1992), human colon (Schworer et al., 1993) and the taenia coli (Schworer et al., 1992) as in all these cases apamin fails to block the relaxing and hyperpolarising effects of VIP. Makhlouf, Grider and their colleagues have highlighted a model of the action of VIP that involves its interaction with NO. It is proposed that the release of VIP from nerves is regulated by NO and that this VIP then acts on smooth muscle cells to both promote the production of more NO as well as of cAMP (Fig. 7.14B; Grider et al., 1992; Makhlouf & Grider, 1993). One way in which VIP could excite the release of NO from smooth muscles involves the claim that there is a calcium-calmodulin NOS in the membranes of gastric smooth muscle cells that is activated by a VIP receptor which in turn is coupled to a pertussis toxin sensitive G-protein (Murthy & Makhlouf, 1994; Jin et al., 1994); this process then leads to relaxation of the muscle (Murthy et al., 1993). These observations suggest that the effects of the peptides are mediated by NO, a claim that has recently been challenged in the case of circular smooth muscle cells of the canine proximal colon: here the electrical and mechanical effects of VIP do not depend on the synthesis of NO; furthermore VIP does not cause the release of NO from these muscle cells. These observations indicate that VIP and NO are inhibitory co-transmitters rather than working in series (Keef et al., 1994). 7.5.4 Contribution of adenosine triphosphate to the IJP The first claim as to the identity of a transmitter for the IJP was made for ATP (Burnstock et al., 1970; Cocks & Burnstock, 1979; Burnstock, 1972; 1981). Many of the neurones of the rat ileum, colon and anococcygeus muscles contain ATP, as evidenced by their quinacrine-fluorescence (Belai & Burnstock, 1994). These neurones also contain NOS, indicating that some nitrergic nerves may use ATP as a cotransmitter (Hoyle & Burnstock, 1989b, c; Hoyle et al., 1990), as seems to be the case for the putative transmitters VIP and PACAP. Inhibition of NOS with L-NAME in the rat pyloric sphincter reduces the nerve mediated relaxation by 40% whereas blocking the action of endogenous ATP release with reactive blue reduces it by 50%, suggesting that different components of the relaxation are independently affected by NO and ATP (Soediono & Burnstock, 1994). Apamin, while partly reducing the relaxation due to NANC stimulation in the taenia coli and the circular muscle of the human colon, completely blocks the relaxation due to exogenous , -meythlene ATP (Boeckxstaeus et al., 1993; Maas & den Hertog, 1979), leaving the relaxation due to exogenous VIP unaffected (Costa et al., 1986). This observation suggests that the component of the IJP blocked by apamin (namely IJPf) is due to the secretion of ATP as a transmitter. Apamin greatly decreases the UP due to stimulation of NANC nerves in guinea-pig taenia coli and stomach as well as the actions of exogenously applied ATP on these tissues (Shuba & Vladimirova, 1980; Maas et al., 1980). The IJPf in the guineapig internal anal sphincter is blocked by apamin as well as by antagonists to ATP whereas IJPs is blocked by NO antagonists. IJPf in the circular muscle of the guinea-pig colon is blocked by ATP, antagonists whereas IJPs is blocked by NO antagonists (Zagorodnyuk & Maggi, 1994). These observations suggest that ATP and NO are jointly responsible for the IJP in these organs, with IJPf due to ATP and IJPs to NO (Rae & Muir, 1996). 7.5.5 Contribution of carbon monoxide to the IJP Haeme oxygenase 2 is the enzyme, together with NADPH-cytochrome c (P 450) reductase, for converting haeme and oxygen to CO (Rae & Muir, 1996). The CO is formed from NADPH-dependent oxidative haeme destruction catalyzed by haeme oxygenase 2. Cleavage of haeme by haeme oxygenase 2 results in the release of biliverdin and CO. Haeme oxygenase 2 is present in smooth muscles of the intestine and certain neurones (Maines, 1988). In the hippocampus, CO appears to act as a retrograde messenger in long-term potentiation, as blocking the production of CO by inhibiting haeme oxygenase 2 with zinc protoporphyrin 9 or zinc deutero-porphyrin, blocks long-term potentiation from Schaeffer collaterals to CA1 pyramidal neurones (Stevens & Wang, 1993; Zhuo et al., 1993). CO inhibits the calcium-activated potassium currents in urinary bladder monocytes and leads to relaxation, as does NO supplied by sodium nitroprusside (Trischmann et al., 1991). This probably occurs as a consequence of elevation of cGMP decreasing the open-probability of voltage-dependent calcium channels (Trischmann & Isenberg, 1989; Utz & Ullrich, 1991). The CO increases voltage-dependent potassium currents in smooth muscle cells of the human jejunum, leading to membrane hyperpolarisation (Farrugia et al., 1993). It is possible, then, that CO is a NANC transmitter mediating inhibitory transmission in the gastrointestinal tract.
132
HISTORY OF THE SYNAPSE
7.5.6 Contribution of pituitary adenylyl cyclase-activating peptide to the IJP The recent discovery of PACAP in most VIP-containing nerves within the taenia coli (McConalogue et al., 1995), together with the observation that the peptide has a powerful relaxing affect on the taenia that is blocked by apamin (Schworer et al., 1992) and by antibodies to PACAP (Jin et al., 1994) makes this a candidate as an inhibitory transmitter. Furthermore, as IJPf is blocked by apamin (Bridgewater et al., 1995), and apamin blocks the hyperpolarisation due to PACAP (McConalogue et al., 1995), then PACAP is a likely candidate transmitter for IJPf. 7.5.7 Identity of the transmitters for the fast and slow IJP At the present, there is some indication of a generalisation concerning the identities of the transmitters that contribute to IJPf and IJPs. The IJPf seems to involve ATP and PACAP in different stoichiometric amounts depending on the region of the tract, whereas IJPs involves NO and VIP, again in different stoichiometric amounts. Some caveats to this generalisation have been mentioned already with respect to the apparent apamin sensitivity of the effects of exogenously applied NO of some parts of the gastrointestinal tract. Of course these caveats to the generalisation may, in the end, bring it down in the same way as the caveats to the existence of atropine-resistant excitation helped to bring down the classical autonomic paradigm. The claim that a particular substance has been found to be the transmitter at a particular synapse or junction seems no longer to be appropriate since the discovery of NANC transmission. This points naturally to the concept that many different substances may be involved in the process of transmission at a nerve terminal, with some predominating as transmitters at particular frequencies and others playing a modulatory role on these. There is no better example of the complexity of these processes than the recent work describing the functional roles of different neuropeptide transmitters together with acetylcholine at a single nerve terminal in controlling the radula closer muscle in Aplysia (Brezina et al., 1996). 7.6 Ionic mechanisms involved in generating the IJP 7.6.1 Modulation of potassium channels by NANC transmitters Nitric oxide activates multiple potassium channels in canine colonic smooth muscle through the elevation of cGMP (Koh et al., 1995). These include the large conductance calcium-activated potassium channel (270 pS; the Bk channel), and two other potassium channels, one with a conductance of 80 pS and the other with a conductance of less than 4 pS. The potassium channel type that is activated depends on the source of the nitrergic input, at least in the guinea-pig proximal colon. For example, exogenous NO and S-nitroso-L-cysteine may activate tetraethylammonium-resistant potassium channels, one of which is blocked by apamin, whereas sodium nitroprusside-activated potassium channels are resistant to the effects of tetraethylammonium ions but may be activated by 8-bromo-cyclic GMP (Watson et al., 1996). The ATP can modulate Ik(ca) channels through an action in some cases on smooth muscle purinergic receptors, either up (Gelband et al., 1990) or down (Giovannardi et al., 1992). The VIP appears to decrease calcium influx in rabbit cavernosal smooth muscle through activation of cGMP. The effects of transmitters on potassium channels that are mediated by second messengers can involve quite complex pathways. cGMP and cGMP-dependent protein kinases increase the activity of Ik(ca) channels over eight-fold in cerebral artery smooth muscle, due to a direct effect of the kinase on the channels (Robertson et al., 1993). cGMP can also act by increasing cAMP phosphodiesterase, thus decreasing cAMP and so decreasing the phosphorylation of a class of Ik(ca) channels which then are turned off. The regulation of Ik(ca) channels by cAMP and its protein kinase catalytic subunit is well established. The cAMP analogues activate Ik(ca) channels in pituitary tumour cells (Levitan & Kramer, 1990), in the basolateral membrane of crypt cells in the colon (131 pS; Loo & Kaunitz, 1989) and in airway smooth muscles (Savaria et al., 1992). The high-conductance Ik(ca) channels may be controlled by both phosphorylation and dephosphorylation. For example, Ik(ca) channels in the smooth muscle cells of the canine proximal colon have their open probability increased by protein kinase A and further increased by okadaic acid (Carl et al., 1991). As the latter inhibits protein phosphatases, the observations indicate that the channels’ open-probability is regulated up by phosphorylation through kinase A and down by dephosphorylation through a protein phosphatase, which is probably regulated by protein kinase G. Apamin blocks the small conductance calcium-activated potassium channel that mediates the hyperpolarizing effects of ATP and PACAP, indicating that the action of these transmitters is to open potassium channels (Banks et al., 1979; Maas et al., 1980; Capoid & Ogden, 1989; Schworer et al., 1992).
DISCOVERY OF OTHER TRANSMITTERS IN THE PERIPHERAL NERVOUS SYSTEM
133
7.6.2 Modulation of intracellular calcium levels by NANC transmitters Nitric oxide decreases the calcium levels in canine gastric smooth muscles (Ozaki et al., 1992), canine colonic smooth muscles (Publicover et al., 1993) and rat anococcygeus muscle (Ramagopal & Leighton, 1989; Li & Rand, 1989a, b). This could be due to a decrease in the activity of voltage-activated calcium channels consequent on membrane hyperpolarisation or due to their phosphorylation by cGMP. It may even be due to channels possessing thiol groups, as these may complex with NO and so modulate the channels that possess them (as do NMDA receptors, Lei et al., 1992). Hyperpolarisations due to NO increasing the open probability of large conductance (263 pS) Ik(ca) channels could lead to a decrease in open-probability of voltage-sensitive calcium channels and therefore to a decrease in intracellular calcium levels (Thornbury et al., 1991). 7.7 The secretion of NANC transmitters responsible for the IJP 7.7.1 Identification of NANC inhibitory motorneurone types Dogiel type 1 neurones in the myenteric plexus possess a single long process and a large number of short lamellar processes. This class of neurone includes the motoneurones to the smooth muscle of the gastrointestinal tract as well as some interneurones (Bornstein, et al., 1991). Dogiel type 2 neurones possess smooth cell bodies with several long processes and a few short processes; these are probably sensory neurones (Bornstein et al., 1991). The motor Dogiel type 1 neurones can be subdivided further into those that contain substance P, and are therefore probably excitatory motoneurones, and those that contain VIP, and are therefore probably inhibitory motoneurones (Furness et al., 1992; Llewellyn-Smith et al., 1988). When NOS occurs in neurones it is found in the VIP-containing Dogiel type 1 cells (Costa et al., 1992; Furness et al., 1992). Enteric neurones may be dissociated from the rat and guinea-pig myenteric plexus into culture either with smooth muscle cells (Nishi & Willard, 1985; Song et al., 1995) or without (Saffrey et al., 1991). Both Dogiel type 1 neurones as well as Dogiel type 2 neurones can be identified in these cultures. Using immunostaining it has been shown that about 20 to 30% of the neurones in culture contain VIP and about the same proportion contain substance P (lessen et al., 1980; 1983; Nishi & Willard, 1985; Buckley et al., 1988). This observation indicates that about one-half of all the neurones present are likely to be either inhibitory or excitatory motoneurones. The recent discovery that about 30% of the neurones in cultured myenteric plexus taken from beneath the guinea-pig taenia coli contain NOS is consistent with the idea that VIP-containing inhibitory motoneurones in culture also contain NOS (Saffrey et al., 1992). 7.7.2 Storage and secretion of NANC transmitters from inhibitory motoneurones There are contradictory accounts concerning the storage of NOS in nerve terminals visualised using peroxidaze conjugated with the antibody to NOS. Peroxidase-labelled antibodies against NOS are localised on the membranes of the endoplasmic reticulum and the outer membranes of mitochondria in the calyx of chick ciliary ganglia (Nichol et al., 1995). This result is in contrast to that obtained using the same antibody on the enteric nervous system. Here NOS has not been identified with any particular organelle in the neurone somas or their nerve terminals (Llewellyn-Smith et al., 1992). It has been shown recently, however, that NOS is a predominantly membrane-bound enzyme (Kemp et al, 1988; Wolf et al., 1992). It is found at the ultrastructural level deposited on the membranes of the endoplasmic reticulum of a wide variety of cell types as well as occasionally on the outer membrane of mitochondria and on organelles in some nerve terminals. Biochemical studies of the localisation of NADPH-diaphorase have shown that the highest enzyme activity occurs in the microsomal fraction indicating that the enzyme is membrane bound (Kuonen et al., 1988). However, it has been suggested that the apparent NADPHdiaphorase reactivity seen in the particulate fractions is not a correct indication of NOS activity (Dun et al., 1992) and that in some tissues, high NADPH-diaphorase reactivity does not correlate with high NOS activity (Matsumoto et al., 1993). One suggestion is that a significant proportion of apparent NADPH-diaphorase activity is due to the presence of cytochrome P-450 reductase (Kemp et al., 1988). Certainly there is a close similarity of amino-acid sequences between the two enzymes (Bredt et al., 1991). The NO may be stored in nerve terminals in the form of S-nitrosothiols (Ignarro, 1990), such as complexes of dinitrosyliron2 with protein thiol groups (Mulsch et al., 1991). Support for this idea comes from studies with the NO carrier Snitrosocysteine, which breaks down fast enough when added to the canine proximal colon to mimic the IJP due to NANC stimulation; both of these effects are blocked by oxyhemoglobin (Thornbury et al., 1991). If NO is stored in synaptic vesicles in this way then NOS will have to be localised at the outer or inner membranes of the vesicles for the NO to be complexed
134
HISTORY OF THE SYNAPSE
with the thiol-containing compounds before it diffuses away. In addition, the vesicles probably will require uptake systems for substrates such as amino acids like cysteine, which contain thiol groups. Two pieces of physiological evidence support the idea of secretion of NO from synaptic vesicles. One is that the secretion of transmitter giving rise to the NO-component of the IJP is blocked in high magnesium and low calcium solutions, conditions for decreasing quantal secretion. However the same conditions will lead to a decrease in the synthesis of NO due to the calcium-calmodulin dependence of NOS (Stark et al., 1993). The other is that an accelerated secretion of transmitter is observed in the lower oesophageal sphincter smooth muscle when it is exposed to scorpion venom; this depletes NANC neurones of their NO secreting ability, implying that NO is stored in synaptic vesicles (Tottrup et al., 1991). One physiological experiment which does not seem to support the vesicle concept for NO storage is that the time for block of the NO component of NANC neurotransmission by L-NMMA (about 10 min) is independent of stimulation during development of the block (Stark et al., 1993). 7.8 NANC excitatory transmission in the gastrointestinal tract Although this review has been concerned with NANC inhibitory transmission to smooth muscle, some mention should be made of recent developments that involve the identification of NANC excitatory transmitters in the gastrointestinal tract. 7.8.1 The EJPs: identity of the excitatory NANC transmitters The EJP in some smooth muscle cells of the taenia coli is cholinergic whereas the EJP revealed in the presence of apamin in other smooth muscle cells of this preparation is due to an as yet unidentified NANC transmitter (Bennett, 1966c; Bridgewater et al., 1995). The tachykinins, such as substance P, neurokinin A and neurokinin B are main candidates for excitatory transmitters (Itoh et al, 1995). Often the hyperpolarisation due to inhibitory NANC transmitters is masked by the excitatory effects due to the concomitant release of tachykinins, such as substance P (He & Goyal, 1993). Substance P induces contraction of the guinea-pig proximal colon and rat gastrointestinal tract through stimulation of tachykinin NK1 receptors (Itoh et al, 1995; Sterini et al., 1995). In the case of the human lower esophageal sphincter, tachykinins contract the muscle through NK2 receptors (Huber et al., 1993). Substance P, neurokinin A and neurokinin B, together with senktide, act on NK2 and NK3 receptors in the rat gastric fundus to cause contraction (Smits & Lefebvre, 1994). It has been argued that the action of the tachykinins is to activate nonselective cation conductance which result in depolarisation of the smooth muscle cells (Lee et al., 1995). 7.8.2 Rebound excitation; identity of NANC excitatory transmitters The IJP observed after stimulation of the intramural nerves to the gastrointestinal tract is often followed by an atropineresistant depolarisation and increased excitability of the smooth muscle cells, as already has been described in section 7.3 (Bennett, 1966). This NANC excitability traditionally had been interpreted as due to the action of atropine-resistant excitatory nerves when just the contraction response of the muscle was recorded (see Section 7.2; Campbell, 1966b; Murray et al., 1991). One candidate NANC transmitter for this rebound effect is prostaglandin, as rebound is reduced greatly by the cyclooxygenase inhibitor indomethacin in the guinea-pig taenia coli (Burnstock et al., 1975), the rat duodenum (Maggi et al, 1984), mouse colon (Fontaine et al., 1984), and the canine proximal colon (Ward et al., 1992b). The other candidate transmitter for the rebound effect is NO, as blockers of NOS reduce or abolish the rebound excitatory response under conditions in which an increased anodal break excitation is unlikely, as in the neck muscle of the urinary bladder of sheep (Thornbury et al., 1995) and in the canine colon (Ward et al., 1992b). One mechanism by which this could occur is that NO releases another NANC transmitter that is excitatory, such as a prostaglandin, causing depolarisation and contraction of the muscle (Thornbury et al., 1995). In guinea-pig ileal longitudinal smooth muscle, NO induces a relaxation followed by a rebound contraction that is blocked by tetrodotoxin. This rebound is reduced partly by atropine and completely blocked by antagonists to substance P, suggesting that it might be due to the excitatory actions of both acetylcholine and substance P (Niel et al., 1983b; Bartho & Lefebvre, 1995).
DISCOVERY OF OTHER TRANSMITTERS IN THE PERIPHERAL NERVOUS SYSTEM
135
7.9 Conclusion This chapter has been concerned primarily with providing an historical account of the discovery of NANC transmission in the peripheral nervous system. It shows the extent to which aspects of the physiology and pharmacology of autonomic neuromuscular transmission that did not fit in with the classical paradigm were largely ignored. In some cases, they were explained away by recourse to explanations concerning the lack of accessibility of receptor antagonists such as atropine to the neuromuscular junctions. The introduction of electrophysiological techniques to the study of autonomic neuromuscular transmission transformed the situation by showing that junction potentials in a whole range of cases could not be blocked by classical antagonists to noradrenaline or acetylcholine, hence the discovery of NANC transmission. The concept of NANC transmission produced a focused effort to identify the transmitters involved. This chapter has considered the range of NANC inhibitory transmitters now under consideration, with NO, CO, ATP, neuropeptides such as VIP, and PACAP as candidates. It has also commented on the problem of identifying the NANC excitatory transmitters, with the tachykinins (including substance P) as well as the prostaglandins as possibilities. The great enthusiasm in the field of autonomic synaptic and neuromuscular transmission at present may be traced to the realisation that the techniques are now in hand to unravel the complexity of autonomic transmission.
8 Development of the Concept of a Calcium Sensor in Transmitter Release at Synapses
8.1 Introduction It is now more than 45 years since a series of papers from the laboratory of Bernard Katz established, using electrophysiological techniques, that calcium controls the amount of transmitter released at individual nerve terminals. First, del Castillo and Stark (1952) showed that changing the external calcium concentration changed the size of the endplate potential recorded with an extracellular electrode without changing the response of the endplate to exogenously applied acetylcholine. Second, the changes in amplitude of the endplate potential with changes in calcium concentration were then confirmed using the recently invented intracellular microelectrode by Fatt and Katz (1952a). Finally, Fatt and Katz (1952b) discovered the spontaneous miniature endplate potentials and showed that while changes in the calcium concentration altered the amplitude of the endplate potential it did not affect the miniature endplate potentials. This lead to the concept that the endplate potential consists of multiples of miniature endplate potentials, each termed a quantum, the actual number of which is determined by the calcium ion concentration. In this historical essay, the development of ideas concerning the role of calcium in transmitter release is described, together with the efforts to identify the sensor for calcium which triggers transmitter release. 8.2 Calcium is necessary for the release of transmitter Locke (1894) reported that (Fig. 8.1 A): Selzt man jetzt 0.02 Procent Ca C12 der Kochsalzlosung hinzu, und pruft man in 3 bis 5 Minuten nachher wieder die indirecte Erregbarkeit, so zeigt sich diese wieder vorhanden. In this way calcium ions were found necessary for successful transmission of the nerve impulse well before the concept of chemical transmission was conjectured in 1904 by Elliott working in Langley’s laboratory (Elliott, 1904a). However, the actual way in which calcium ions mediate the transmission process had to await the establishment of the chemical nature of transmission, which may be taken as at the time of the award of the Nobel Prize to Henry Dale and Otto Loewi in 1936 for their discovery of acetylcholine as a transmitter substance. That year Brown and Feldberg introduced the technique of liberating acetylcholine from the superior cervical ganglion perfused with an eserinised Locke solution using high KCl concentrations in the perfusion fluid (Brown & Feldberg, 1936a, b) (Fig. 8.1B). Interestingly enough they emphasised in their paper that increasing the CaC12 concentration had the effect of reducing the release of acetylcholine by the elevated KCl (Fig. 8.1B), and they only mentioned in an aside that: It may be noted that the injection of a Ca2+ free solution containing the normal concentration of K+ was without any detectable effect. This was to be expected, given that the acetylcholine release only occurred in an elevated potassium. This paper does not establish that calcium ions are necessary for transmitter release. The observation that elevated potassium concentrations could liberate acetylcholine from ganglia was extended to brain slices by Mann et al. (1939). Again they emphasised that high calcium concentrations in the medium could antagonise the release of acetylcholine from the brain slices by elevated potassium concentrations. However, they did note in passing that:
DEVELOPMENT OF THE CONCEPT OF A CALCIUM SENSOR IN TRANSMITTER RELEASE AT SYNAPSES
137
Moreover, the effects of adding 0.027 M KCl to a saline medium (i.e. free from Ca++) on the liberation of free acetylcholine by brain slices is far less marked than when this concentration of K+ is added to a Locke medium Figure 8.1(D) shows their table of results in which comparison of Experiments B and C indicate that in the Locke solution with 0.027 M KCl there is about a 30% increase in release of acetylcholine over that in the NaCl solution with the same concentration of KCl. It was not until four years after the observations of Brown and Feldberg that Harvey and Macintosh (1940) used the perfused superior cervical ganglion to show that in a calcium-free solution neither stimulation of the preganglionic nerves nor elevated potassium in the perfusate could liberate acetylcholine from the ganglion (Fig. 8.1(c)). They state that: The omission of calcium ions from the perfusion fluid, other cations being present in normal concentrations, has the following effects: synaptic transmission fails, owing to the failure of preganglionic nerves to liberate acetylcholine. This paper, entitled ‘Calcium and synaptic transmission in a sympathetic ganglion’, is entirely devoted to the necessity of calcium ions for secretion of acetylcholine, and constitutes the first occasion in which the role of calcium is explicitly analysed. 8.3 Electrophysiological evidence that calcium is necessary for the release of transmitter: the concept of a calcium sensor for secretion In 1942, whilst carrying out research at the Kanematsu Institute in Sydney during the Second World War, Bernard Katz and his colleagues showed that the amplitude of the endplate potential recorded with an extracellular electrode from the frog neuromuscular junction is affected by calcium ions (Katz & Kuffler, 1942). Feng and his colleagues, working in China during the Sino-Japanese war, had already suggested that calcium may exert its effect on transmission by altering the amount of transmitter released. This conjecture was based on the observations that an increase in calcium concentration leads to failure of neuromuscular transmission in response to a high-frequency train of impulses but that such an increase in calcium could restore transmission by a single impulse in partly curarised muscles (Feng, 1940, Feng & Li, 1941). Both of these phenomena could be explained by an increase in calcium leading to an increase in transmitter release (Feng, 1940). However, the first direct electrophysiological evidence that an increase in calcium does increase transmitter release was provided by del Castillo and Stark (1952). They recorded the changes in amplitude of the endplate potential with an extracellular electrode in different calcium concentrations and showed that a linear relationship exists between the amplitude of the endplate potential and the logarithm of the calcium concentration (in the range of concentrations from 0.45 mM to 7.2 mM; Fig. 8.2A). Next they determined that the amplitude of the depolarisation of the muscle caused by exogenous acetylcholine was not affected by changing the calcium concentration, indicating that the sensitivity of the endplate potential to changes in calcium was most likely due to changes in the amount of transmitter released rather than in the effect of this transmitter on the muscle (Fig. 8.2A). These results were confirmed in the same year with intracellular electrodes by Fatt and Katz (1952b), who showed that the endplate potential decreases along a gradually steepening gradient as the calcium is decreased below about 0. 45 mM (Fig. 8.2(B)). The advent of the microelectrode allowed the discovery of spontaneous miniature endplate potentials by Fatt and Katz in the spring of 1950 (see Katz, 1996). The result of reducing the calcium concentration on these in relation to its effects on the endplate potential was then reported in 1952. It was first noted that, in reduced calcium concentrations, the amplitude of the endplate potential during a train of impulses varies in a step wise manner (Fig. 8.2CB), a phenomenon that is not observed in normal calcium at curarised endplates (Fig. 8.2(C)D). Next it was noted that the amplitude of the spontaneous miniature endplate potentials was similar to that of the size of the fluctuations in the endplate potential (Fig. 8.2 (C)C). Indeed: A curious effect was observed when reducing the calcium concentration. This causes the size of the epp to diminish without affecting the size of the miniature potentials. With a sufficiently low calcium level, the response to a motor nerve impulse may be no larger than a single miniature epp. In this condition, successive nerve impulses evoke a random display of minute epp’s whose sizes vary in a step-like fashion, seemingly corresponding to multiples of a miniature discharge (cf Fig. 8.2BB, bottom record, where steps of 0, 1, 2 and 3 can be recognised). They went on to note that: Calcium deficiency is known to reduce the size of the endplate potential (epp); this effect appears to take place in ‘steps’, involving an all-or-none blockade of a variable number of miniature components. The size of individual miniature potentials in contrast to that of the epp is not affected by calcium lack.
138
HISTORY OF THE SYNAPSE
Fig. 8.1. The discovery that calcium is required for transmitter release. (A) Reproduction of the original paper by Locke in 1894 reporting the results of removing calcium from the solution bathing a neuromuscular preparation on blocking synaptic transmission (from Locke, 1894). (B) Introduction by Brown and Feldberg (1936a) of the cat superior cervical ganglion perfused with high potassium concentrations for the purpose of releasing acetylcholine from nerve terminals. Upper panel shows the contraction of the nictitating membrane following injections into the perfused ganglion of elevated potassium and calcium concentrations. Lower panel gives the blood pressure of a cat after collection of the venous effluent from the ganglion and injection into the cat. C gives the results for an eight-fold increase in the normal KCl concentration, which produces a greater loss of blood pressure in the cat than does the injection of 0.01 mM of acetylcholine as shown in B, indicating that the output of acetylcholine from the ganglion is in excess of this concentration. In A and D are shown the reduction in contraction of the nicitating membrane and the changes in blood pressure for perfusion of the ganglion with 4.5 times and 8 times the normal CaCl2 concentration in the presence of 8 times the KCl concentration. Note that both of these physiological responses are reduced compared with the control in C. If the CaCl2 concentration is increased to 16-fold normal in the presence of 8 times the normal KCl then E shows that there is no response, so that all of the acetylcholine release appears to be inhibited. F shows that the reduction of acetylcholine release due to the elevated CaCl2 persists for some time, as subsequent elevation of KCl alone still gives a less than normal release of acetylcholine (compare F with C). (from Fig. 4 in Brown & Feldberg, 1936a). (C) The first definite evidence that calcium is necessary for the release of a transmitter, by Harvey and Macintosh (1940). The experimental design is the same as that of Brown and Feldberg above, so that the upper panel is for the contraction of the nictitating membrane and the lower panel that for the cat’s blood pressure. A–C give the results for perfusion of the cat superior cervical ganglion with normal Locke’s solution containing eserine. A, with no stimulation; B, maximal preganglionic stimulation for 10 sec; C, injection of 2 mg of KCl. D–F give the results for perfusion of the ganglion with calcium-free Locke’s solution containing eserine. D, With no stimulation; E, maximal preganglioninc stimulation for 10 sec gives no further contraction of the nicitating membrane; F, after injection of 2 mg of KCl. G, Gives the time signal and the effects of 0.005 µg of acetylcholine (from Fig. 2 in Harvey & Macintosh, 1940). (D) The effects of elevated KCl on the release of acetylcholine from brain slices bathed with an eserine-Locke’s solution (Exp B) for comparison with an eserine-NaCl solution (Exp C). Note that the release of free acetylcholine in the former case amounted to 9. 6 µg/g−1 whereas in the latter case, in the absence of calcium, it only amounted to 6.5 µg/g−1 (from Table X in Mann et al., 1939).
DEVELOPMENT OF THE CONCEPT OF A CALCIUM SENSOR IN TRANSMITTER RELEASE AT SYNAPSES
139
Fig. 8.2. The first electrophysiological evidence that calcium is required for transmitter release, all carried out on frog muscle endplates. (A) The effects of different concentrations of calcium (logarithmic abscissae) on the amplitude of the endplate potential (A, ordinate) or the depolarisation due to applied acetylcholine (B, ordinate), both recorded with an extracellular electrode; the amplitude of the endplate potential and the depolarisation due to acetylcholine have been normalised to their sizes in normal Ringer’s solution (from Fig. 3 in del Castillo & Stark, 1952). (B) The effects of different calcium concentrations (abscissae, concentrations normalised to 1.8 mM) on the percentage change in amplitude of the endplate potential (ordinate, amplitude normalised to that in 1.8 mM) recorded with an intracellular electrode (from Fig. 8 in Fatt & Katz, 1952b). (C) The effect of lowering the calcium concentration on the amplitude of the evoked endplate potential and on spontaneous miniature endplate potentials. In A to C the muscles were soaked in reduced (1/4) calcium concentrations with prostigmine bromide present in B. The records in A and the three top records in C were single sweep records; all other records were obtained with multiple sweeps repeated at ca 1 per sec-1, during each of which the nerve was stimulated at the instant marked by an arrow, though there was not always an endplate response. The top record in A and the three top records in C show spontaneous discharges only. All other records show endplate potential responses to nerve stimulation. These vary in a step-like manner between zero and a few millivolts. In some records (in A and C) spontaneous discharges are seen on the same sweep, immediately before or after the endplate potential response. For comparison with the effect of a lack of calcium, the relative constancy of the endplate potential response in a curarised fibre is shown in D. The voltage scale is a millivolt and the time scale shows 50 Hz (from Fig. 9 in Fatt & Katz, 1952a). (D) The effect of calcium and magnesium ions on the increase in discharge of spontaneous miniature endplate potentials during depolarisation of a motor nerve terminal with an extracellular cathode electrode. The ordinate shows the increments in discharge rate and the abscissae the relative current intensities used to stimulate. Five successive periods of stimulation were used as follows: first, normal Ringer ( ); second, 10 mM magnesium ( ); third, 10 mM magnesium with 6 mM of extra calcium ( ); fourth, 10 mM ( ); fifth, 10mM magnesium with 6mM of calcium added (X) (from Fig. 6 in del Castillo & Katz, 1954c). (E) Table summarizing the effects of polarisation of motor nerve terminals with an external electrode on the frequency of miniature epps and the number of such miniature epps composing the evoked epp (that is the quantal content). Shown for comparison are the effects of other agents on this frequency and quantal content (from Table 1 in del Castillo & Katz, 1954c). (F) The first description of the calcium-sensor hypothesis for transmitter release. For a description of this hypothesis, see the Text (from del Castillo & Katz, 1954c).
Given that the effects of varying calcium on the size of the epp was due to changes in the release of transmitter as del Castillo and Stark (1952) had found, and that calcium did not change the size of the miniature epp, it was natural to propose that the effect of decreasing calcium was to decrease the number of miniature epps that compose the epp, by decreasing the probability that such miniature components are released (del Castillo & Katz, 1954a). The possibility that the probability of release of such miniature epps might be increased on arrival of the nerve impulse was next examined (del Castillo & Katz, 1954b, c). Motor-nerve terminals were depolarised with an external electrode and a greatly accelerated discharge of miniature epps noted. Unlike the amplitude of the miniatures, their frequency of discharge during the depolarisation was dependent on the calcium and magnesium concentration, dropping to nearly zero when the
140
HISTORY OF THE SYNAPSE
magnesium was increased and being restored to normal when calcium was added (Fig. 8.2D). The results of depolarisation under a cathode and of hyperpolarisation under an anode on the frequency of miniature epps are summarised in Fig. 8.2E. These observations gave rise to the concept that calcium controls the probability of discharge of miniature epps under a depolarisation as a consequence of its interacting with a carrier molecule in the terminal, leading to the production of a miniature epp. The scheme is spelt out in Fig. 8.2F and is described in the words of del Castillo and Katz (1954c) as follows: On this hypothesis, the common step in ‘spontaneous as well as evoked activity is the release of an active ‘carrier molecule’ (X’) which transports, or allows the passage of, a large number of ACh ions and leads to the production of a miniature e.p.p. There are different ways in which X’ can be formed: (i) from a CaX compound which is specifically acted on by the nerve impulse and transformed to Ca+X’, (ii) from other inactive precursors (X) which may change to X’ spontaneously, due to thermal activity. Only the first of these resources is blocked by Mg, or by Ca deficiency. This paragraph gives the first statement concerning a calcium sensor in the nerve terminal for transmitter release. Several points might be noted. First, the formation of the active molecule X’ involves the prior formation of the complex CaX; second, that this complex is formed under a depolarisation; and third, that X’ may form spontaneously as a consequence of some thermal activity in the terminal. It is not made explicit where X’ is, on the outside or the inside of the terminal, or is in the membrane and acts as a carrier? The model provided in Fig. 8.2F has proved to be one of the most prescient in the history of neuroscience, as subsequent sections of this review will point out. 8.4 The calcium action potential In 1952 Hodgkin and Huxley published their monumental work on the ionic basis of the action potential in the giant axon of the squid Loligo in a series of five papers in the Journal of Physiology. This study introduced a new level of analysis into neuroscience, both in its quantitative rigour and elegance of presentation. However it was something of a disappointment to the authors themselves. This was because their introduction of the notion of voltage-dependent ion channels with different degrees of ionic selectivity, was still treated at the phenomenological level of analysis, they being unable to provide direct evidence that such channels actually existed (Hodgkin & Huxley, 1952). Nevertheless, this work was generally taken to establish the existence of voltage-dependent ion channels, and indeed of the generalisation that the rising phase of action potentials per se was due to the influx of sodium ions through voltage-dependent sodium channels. Such an emphasis was reinforced by the subsequent observations by Hodgkin and Keynes (1957), using tracer fluxes in squid axons, that an amount of calcium amounting to only about 0.006 pmole/cm2 entered the giant axon per impulse, a negligible amount compared with the influx of sodium of 4 pmole/cm2. However in the late 1950s and early 1960s the notion gradually emerged that calcium could be a charge carrier for the action potential in some tissues under some circumstances, and perhaps even that tissues existed for which calcium was normally the only inward charge carrier. This then gave rise to the concept of voltagedependent calcium channels, that now figure so largely in our concept of transmitter release. The way in which the concept of such calcium channels arose is now detailed. 8.4.1 Induced calcium action potentials in invertebrate muscles Fatt and Katz in 1953, shortly after their first quantitative description of the spontaneous miniature endplate potentials in frog muscle (Fig. 8.2(C)), observed that most crab muscle fibres did not give rise to an all-or-none action potential when depolarised with an intracellular electrode, unless the sodium had been replaced with quaternary ammonium ions such as tetraethylammonium (TEA) or tetrabutylammonium (TBA). This was a very interesting observation, since only the year before Hodgkin and Huxley (1952) had given what appeared to be the definitive description of the action potential in terms of an influx of sodium ions. Admittedly this was for a cephalopod axon from the giant axon of the squid, but all indications were that the description for the squid would hold for other species, although Lorente de No (1949) had reported that some nerve fibres in frog could conduct action potentials if the sodium were replaced with TEA ions. As both calcium and magnesium ions were present in the quaternary anion solutions bathing the crab muscle fibres in the experiments of Fatt and Katz (1953), they suggested that: …there may be two ways of accounting for this observation: (i) it may be that TB A remains adsorbed to the fibre surface, but is mobilised during excitation and temporarily transferred into the cell interior; (ii) alternatively, influx of calcium or magnesium, or outflux of some internal anion may be responsible for the transport of charge.
DEVELOPMENT OF THE CONCEPT OF A CALCIUM SENSOR IN TRANSMITTER RELEASE AT SYNAPSES
141
Fig. 8.3. Crustacean muscle fibres are capable of generating calcium action potentials if exposed to quaternary ammonium ions or if the internal calcium concentration is lowered with calcium buffers. (A) Action potentials recorded from a crayfish muscle fibre in solutions of increasing calcium concentration (left to right, 4, 8 and 16 mM) after treatment for 20 hr with replacement of sodium with TEA ions. Stimulating current pulse was of 10 msec duration. The record on the right includes potential changes produced by pulses which failed to stimulate (from Fig. 3 in Fatt & Ginsborg, 1958). (B) Action potentials recorded from a barnacle muscle fibre in solutions of increasing calcium concentration (left to right, 20, 43, 85, 169 and 338 mM). Two series of records are shown (A and B) for which the intracellular calcium concentration was buffered with K2SO4 and EDTA. The period of current injection is shown in the lower part of the bottom panel and the zero reference potential by the horizontal line (from Fig. 5 in Hagiwara & Naka, 1964). (C) The effects of changes in the external calcium concentration (logarithmic abscissae) on the overshoot of the action potential in three different barnacle muscle fibres (upper graph) and on the threshold for firing the action potential in these fibres (lower graph). The broken line in the upper graph has a gradient of 29 mV per tenfold increase in calcium concentration and is the gradient expected if the overshoot of the action potential is mostly due to calcium ions (from Fig. 6 in Hagiwara & Naka, 1964). (D) Evidence that the inward current in the neuron somas of both marine and terrestrial molluscs is carried by both calcium and sodium ions. Current records taken during voltage clamp of the snail neuron soma of the marine mollusc Aplysia in different test salines. The bathing solutions were (left to right columns): normal saline; calcium-free and sodium-containing; sodium-free and calcium-containing; normal saline with tetrodotoxin. The holding potentials were from left to right: −36; −45; −43; and −40 mV. The horizontal calibration is 20 msec. The vertical calibration for left to right columns in µA are 4, 4, 1, 2. The potential of the command step is indicated (in mV) next to each record (from Fig. 3 in Geduldig & Gruener, 1970).
The possibility was therefore entertained that calcium ions might act as the charge carrier for the action potential. It was left to Fatt and Ginsborg (1958) to show that calcium ions were indeed the charge carrier in crustacean muscle bathed in TEA, as increasing the calcium concentration increased the size of the action potential in muscle treated with TEA (Fig. 8.3(A)), whereas removal of calcium ions led to failure of the action potential, so that: The action potential occurring in these circumstances depended on the presence of calcium and was independent of Na and Mg. They also made the very interesting additional observation that strontium or barium ions could not only substitute for calcium ions but that action potentials with large overshoots could be obtained in these circumstances even in the absence of quaternary anions. Furthermore, the amplitude of this overshoot changed by between 6mV and 13 mV for each doubling of the strontium concentration, an amount which they pointed out was to be expected on the basis of the Nernst potential if the rising phase of the action potential was due to an influx of strontium ions in these circumstances. They were then led to the conclusion that:
142
HISTORY OF THE SYNAPSE
…the action potential occurring in solutions of strontium or barium resulted from the muscle fibre membrane becoming highly permeable to these ions. The irreversible action of tetraethylammonium and other substituted ammonium compounds is attributed to their having an action on the mechanism of permeability change, whereby the ability of calcium to cross the membrane is prolonged. So Fatt and Ginsborg (1958) not only defined the conditions under which these divalent cations could act as charge carriers in crustacean muscle but also introduced the idea that strontium and barium could pass across the membranes of cells in a more efficacious way than calcium in such circumstances, an observation that was to be used to considerable advantage in studying calcium channels in the future. Finally a direct test of the ability of calcium ions to carry the charge during an action potential in crustacean muscle was made by Hagiwara and Naka (1964). They showed that if the intracellular calcium concentration in these muscles was lowered by introducing a calcium buffer such as K2SO4 then action potentials could be obtained that had large overshoots which increased with an increase in the calcium concentration even in the absence of quaternary ions (Fig. 8.3B). Quantification of this result showed that the overshoot increased with a slope of 29 mV per tenfold change in the calcium concentration (Fig. 8.3C), almost exactly that expected according to the Nernst equation for a calcium electrode. 8.4.2 Naturally occurring sodium and calcium action potentials in vertebrate cardiac muscle Draper and Weidmann (1951) showed that the rapid depolarisation phase of the action potential in Purkinje fibres depends on the external sodium concentration, establishing that the major charge carrier for this component of the action potential was likely to be due to sodium ions. After publication of the Hodgkin and Huxley (1952) analysis of the sodium action potential in the squid giant axon it seemed likely that a similar approach could be applied to the cardiac action potential, with similar kinetics for the sodium inward current. However it soon emerged that a slow phase in the upper part of the upstroke of the action potential, including the plateau phase, was much less sensitive to changes in sodium. Huffman and Suckling (1956) noted that there was an increase in the rate of rise of the action potential in cardiac tissue with an increase in the external calcium concentration. Application of the voltage-clamp technique to sheep Purkinje fibres showed that there was indeed a slow calcium-sensitive inward current (Fig. 8.4A), which persists in the absence of sodium ions (Reuter, 1967). Reuter stated that: In my opinion the sensitivity of the current voltage curve to alterations in external calcium concentration indicated that calcium ions carry electric charge. Further study of these Purkinje fibres suggested that this inward calcium current was probably responsible for the increase in the rate of depolarisation of the active response to current injection in the fibres (Fig. 8.4B; Reuter & Scholz, 1968), an observation that was soon confirmed for ventricular fibres as well (Scholz, 1969). The conclusion that part of the late rising phase and plateau phase of the cardiac action potential could be attributed to an influx of calcium ions was not straightforward. It was argued that there was still a high concentration of sodium immediately outside the individual fibres constituting the complex array of fibres that make up this tissue in sodium-free solutions; in this case sodium could still carry the inward current in sodium-free solutions and changes in calcium concentration might act simply by altering this sodium current (Johnson & Lieberman, 1971). Keenan and Niedergerke (1967) showed that the total sodium in the frog heart ventricle does not decrease after incubation for 15 min in a sodium free solution, suggesting that any remaining sodium is tightly bound or distributed in an inacessible space within the tissue. In order to sustain the claim that sodium acts as the only inward carrier it therefore had to be argued that calcium had access to this space in order to modulate an inward flux of sodium in a sodium-free solution, although sodium could not diffuse from the space. Furthermore, as tetrodotoxin blocks the cardiac action potential, due to its inhibiting sodium influx, then the toxin must have access to the so-called inacessible space (Scholz, 1969). These arguments made untenable the claim that sodium carries all of the inward current in the cardiac action potential, thus establishing that a calcium current must exist as well as the major sodium current in generating the cardiac action potential (Reuter, 1973). 8.4.3 Naturally occuring sodium-calcium action potentials in invertebrate neurons The discovery of sodium-calcium action potentials in cardiac muscle in the late 1960s was paralleled by similar discoveries with respect to the action potentials in the neurons of molluscs. In 1967 Junge observed that the action potential in the marine mollusc Aplysia could persist when sodium was replaced by sucrose (Junge, 1967). This led Scholz (1969) to investigate
DEVELOPMENT OF THE CONCEPT OF A CALCIUM SENSOR IN TRANSMITTER RELEASE AT SYNAPSES
143
whether calcium ions carried part of the inward current during the action potential. The recent discovery of tetrodotoxin enabled them to show that this blocked neither the action potential nor the spike that persisted in a sodium-free solution that still contained calcium (Scholz & Reuter, 1969). In the absence of either ion the action potential overshoot behaved as a sodium or a calcium electrode according to the Nernst equation for that particular ion species that remained, indicating that both ions contributed to the rising phase of the action potential. These authors went on to carry out the then difficult experiment of voltage-clamping the soma of these Aplysia neurons, which allowed them to show that the inward current was composed of both a calcium component and a sodium component (Geduldig & Gruener, 1970; Fig. 8.3D). 8.4.4 Naturally occurring calcium action potentials in vertebrate smooth muscle In 1958 Mollie Holman noted that the spontaneous action potentials in the smooth muscle cells of the mammalian gastrointestinal tract (the longitudinal muscle of the guinea-pig caecum) were largely unaffected if the sodium concentration was reduced to 20 µM (Holman, 1958). The obvious interpretation of this observation, namely that the smooth muscle action potential is not due to an influx of sodium ions, was complicated by the fact that further reduction of the sodium led to failure of the action potential. Given the possibility that the pacemaker activity in spontaneously active smooth muscle cells might be due in part to sodium, the complexity of determining the ionic basis of the action potential in such muscles is obvious. Nevertheless the maximum rate of rise of the action potential was found to be proportional to the extracellular calcium concentration in this muscle (Bulbring & Kuriyama, 1963). With the introduction of tetrodotoxin to block voltage-activated sodium channels in the late 1950s (Furukawa et al., 1959; Narahashi et al., 1960), it was possible to test again if these spontaneous action potentials in the gastrointestinal tract were due to sodium ions: the results showed that tetrodotoxin had no effect on the action potentials, indicating that sodium was unlikely to be the charge carrier for the action potential in smooth muscle cells (Nonomura et al, 1966). In order to identify the charge carrier for the smooth muscle action potential, Bennett (1967) selected a preparation for which there is no spontaneous activity (the guinea-pig vas deferens), and for which there was a known sensitivity of the action potential to calcium ions (Kuriyama, 1964). He determined the extent to which the overshoot of the action potential is sensitive to calcium ions versus sodium ions. The results were unequivocal, the overshoot of the action potential and its maximum rate of rise were unaffected by reducing the sodium to 25 µM (Fig. 8.4C) whereas the maximum rate of rise was very sensitive to changes in calcium and the overshoot changed at the rate of 22 mV per ten-fold change in the calcium concentration as expected for a calcium electrode according to the Nernst equation (Fig. 8.4D). Furthermore this calcium effect was not dependent on the constancy of sodium ions since it was obtained when calcium was altered by leaving the ratio of calcium concentration to sodium concentration (squared) constant. This provided the first direct evidence for a naturally occuring calcium action potential in any excitable cell. 8.5 Are calcium movements across the nerve terminal necessary for evoked secretion? The concept of the voltage-dependent calcium channel arose as a consequence of the discovery in the late 1950s that calcium could act as a principal inward current carrier for action potentials in crustacean muscle fibres under special conditions and in the early 1960s that calcium was likely to be a current carrier for cardiac muscle action potentials and the only current carrier for smooth muscle action potentials. This provided a background to the possibility that such channels might be present in nerve terminals and so mediate the known calcium sensitivity of the transmitter release process. del Castillo and Katz, in their 1954 model of the calcium-sensor for transmitter release, had not made explicit where the calcium-sensor was to be found. Indeed their specification of the molecule, X’, that was ultimately responsible for secretion as a carrier molecule, implied that it carried the unit of transmitter across the nerve terminal membrane in the release process. The fact that X’ was formed from CaX did not make clear where the CaX itself was formed (Fig. 8.2F). Experiments performed in the mid-1960s showed that either Ca2+ or Ca2+X had to pass into the nerve terminal from the outside. With the discovery of tetrodotoxin it was now possible to depolarise nerve terminals with a focal electrode placed at a site on the nerve terminal without generating action potentials, and record the consequent evoked release of units of transmitter with an intracellular electrode. When this was done at the frog motor endplate it was noted that the minimal latency at which a transmitter unit was released increased with the duration of the depolarizing pulse (Fig. 8.5A; Katz & Miledi, 1967c, d, e). This was regarded by Katz and Miledi as: …evidence suggesting that entry into the axon membrane of a positively charged substance (external Ca2+ ions or a calcium compound CaR+) is the first step leading to the release of acetylcholine packets from the terminal.
144
HISTORY OF THE SYNAPSE
Fig. 8.4. Vertebrate cardiac and smooth muscle cells use calcium as a principal charge carrier for the action potential. (A) and (B) Cardiac muscle fibres normally use both sodium and calcium ions to generate the action potential. (A) Voltage-clamp records of the membrane currents in a short Purkinje fibre from sheep in Tyrode solution with 1.8 mM of calcium on the left and 0 mM calcium on the right. The upper part of each figure shows that the voltage clamp was from −84 to −8mV. The lower part of the figure shows that there is a slow inward current at the beginning of the trace in the presence of calcium ions (left panel) and this is not present when the calcium ions are removed (right panel); the inward current is therefore likely to be carried by calcium ions (from Fig. 5 in Reuter, 1967). (B) The characteristics of action potentials due to electrotonic depolarisation of isolated sheep trabeculae in different calcium concentrations. Shown are the resting membrane potential ( ), the threshold for activation of the active response ( ) and the height of the active response (X). Note that the active response increases in amplitude with an increase in the calcium concentration (log scale; from Fig. 8 in Reuter & Scholz, 1968). (C) and (D) Smooth muscle fibres normally generate calcium action potentials. (C) Action potentials recorded in the guinea-pig vas deferens in solutions of different sodium concentrations. Sodium in A is 118 mM, in B, 86 mM and in C, 25 mM. Note that the rate of rise and overshoot of the action potential remains approximately the same at each sodium concentration (from Fig. 3 in Bennett, 1967). (D) The effects of changes in calcium concentration on the characteristics of the action potential. Abscissa, calcium concentration, log scale. Ordinate, membrane potential, negative inside with respect to outside. , overshoot of the action potential. , threshold for initiation of the action potential. , resting potential. The overshoot changes at 22 mV per tenfold change in calcium concentration in the range from 0.2 mM to 2.5 mM. Vertical bars give twice the standard error of the mean, n=7 (from Fig. 9 in Bennett, 1967).
More direct evidence for this idea was obtained by direct depolarisation of the presynaptic membrane with an intracellular electrode in the presynaptic axon of the stellate ganglion of the squid and another intracellular electrode in the postsynaptic axon (Katz & Miledi, 1967b). In the presence of tetrodotoxin, short depolarizing pulses of the presynaptic membrane lead to a postsynaptic response that did not occur until the depolarisation was ended in the case of strong depolarizing pulses (Fig. 8.5B). This was the result to be expected if the strong presynaptic depolarisation prevented the influx of the positively charged Ca2+ or the CaR+. The experiment did not distinguish however between the proposition that it was the entry of Ca2+ that was necessary for transmitter release or of the calcium compound. Indeed, subsequent experiments that year on the frog motor endplate by Dodge and Rahamimoff (1967), showing the quantitative relation between the size of the e.p.p. and calcium at much lower calcium concentrations than that analysed quantitatively by del Castillo and Stark nearly 25 years earlier (Fig. 8.2A), were still interpreted in terms of the formation of CaX on the nerve terminal without distinguishing whether this was on the inside or the outside. Their analysis indicated that at calcium concentrations lower than ca 0.4 µM, in a normal magnesium concentration of 2 µM, a cooperative mechanism involving the formation of CaX and transmitter release was revealed. This lead to the suggestion that (Fig. 8.5D): If we assume that a single Ca ion combines with the receptor to give CaX and that CaX molecules or sites are randomly distributed on the membrane of the nerve terminal, then this model requires that at least 4 CaX are present in a certain small area (A) of the membrane (for example, if the quantum of ACh can be identified with the contents of a vesicle, then the critical small area might be the area of contact between the vesicle and the membrane). The present work suggests that the probability of release depends on a membrane process in which about four calcium ions must be
DEVELOPMENT OF THE CONCEPT OF A CALCIUM SENSOR IN TRANSMITTER RELEASE AT SYNAPSES
145
Fig. 8.5. Evidence that Ca2+ or a Ca2+X compound must enter the nerve terminal during the action potential for transmitter release. (A) Sample records in each of the panels of endplate potentials at high amplification and speed and low amplification and speed due to brief depolarisation pulses applied focally to a motor nerve terminal of the frog. Action potentials blocked with tetrodotoxin. The left hand column of panels is for a short pulse that still elicits the all-or-none appearance of a discrete unit of transmitter release with variable latency. The right hand column is for longer pulses. Note that the latency from the beginning of the pulse to the appearance of a unit increases with the longer pulses (from Fig. 11 in Katz & Miledi, 1967e). (B) Suppression of transmitter release during a large ‘positive voltage step’ of the presynaptic membrane potential of the giant synapse of the squid stellate ganglion, treated with tetrodotoxin to block action potentials. The presynaptic terminal was loaded with TEA ions. In each record the bottom trace shows the presynaptic voltage step, the middle trace shows postsynaptic responses and the top trace monitors the current pulse. The main postsynaptic deflexion is postponed until after the pulse is complete as the pulse gets longer (from Fig. 12 in Katz & Miledi, 1967b). (C) The effect of ionophoretic pulses of calcium (Ca) applied to a motor nerve terminal of the frog endplate on endplate responses evoked by a depolarizing pulse applied at different times with respect to the time of iontophoresis. Tetrodotoxin present to block the action potential. Depolarizing pulses and calcium were applied from a twin-barrel micropipette to a small part of the nerve-muscle junction. Intracellular recordings from the endplate region of a muscle fibre are shown. Bottom traces show current pulses through the pipette. Column A is the depolarizing pulse alone. Column B is when the calcium pulse precedes the current pulse. Column C is when the depolarizing pulse precedes the calcium pulse. Note that it is only when the calcium pulse just precedes the current pulse that there is transmitter release (from Fig. 1 in Katz & Miledi, 1967a). (D) Left-hand graph shows a plot of the reciprocal of the fourth root of epp at the frog neuromuscular junction against the reciprocal of the calcium concentration (Lineweaver-Burk plot). Right-hand graph shows the same plot but for three different magnesium concentrations of 0.5 mM ( ), 2 mM (X) and 4 mM ( ). Note that the straight lines have approximately the same intercept (from Fig. 4 in Dodge & Rahamimoff, 1967).
146
HISTORY OF THE SYNAPSE
Fig. 8.6. Evidence for calcium channels in nerve terminals. (A) Evidence that a regenerative calcium action potential can be generated in a nerve terminal. The presynaptic nerve terminal in the stellate ganglion of L. vulgaris was recorded from using intracellular electrodes during current pulses in the presence of tetrodotoxin to block sodium influx and TEA ions to block potassium efflux, in 45 mM calcium. Shown are two successive superimposed traces in A with the upper trace giving the current pulse and the lower the presynaptic potential response. Note the threshold for the presynaptic response. B, From another synapse in the presence of tetrodotoxin and TEA ions but with a lower calcium concentration of 11 mM, on a slow time base. In this case the upper trace gives the current pulse, the middle trace the post-synaptic potential change and the lower trace the presynaptic potential change. Calibration bar in A is for 1.14 µA and 50 mV and the upper vertical scale bar in B gives 1.83 mA (from Fig. 3 in Katz & Miledi, 1969). (B) Light emission recorded during stimulation of the presynaptic axon in the stellate ganglion of Loligo pealii previously injected with aequorin. During repetitive stimulation, indicated by the dark bar in A, there is a slow increase in the luminescence to about three times the background which then slowly decreases to the dark current line. In B–D the time constant of the amplifier was increased to filter out high frequency photomultiplier noise. In B, threshold repetitive activation of presynaptic axon generated an increased luminescence. In C, after a small reduction of the stimulus intensity to subthreshold level, there was no increase in luminescence. In D, the stimulus intensity was raised again and luminescence was produced as before (note the change in gain between records B and D). Luminescence was measured with a photo multiplier in nanoamperes (from Fig. 2 in Llinas et al., 1972). (C) The relationship between the external calcium concentration and the amplitude of the postsynaptic response, at various presynaptic input potentials (shown in mV by the numbers against each curve) in the stellate ganglion of L. vulgaris The maximum slope of the log response/ log external calcium curves is ca 2.5 (from Fig. 7B in Katz & Miledi, 1970a). (D) The relationship between the external calcium concentration and the amplitude of the endplate potential in different magnesium concentrations ( , 0.5 mM; X, 2.0 mM; , 4.0 mM). Each value represents the average amplitude of 128 or 256 endplate potentials. Note the log/log scales (from Fig. 3B in Dodge & Rahamimoff, 1967). (E) The relationship between presynaptic depolarisation and transmitter release (curve A) and the calcium permeability coefficient (curve B). The abscissa is the presynaptic potential change and the ordinate is in arbitrary units (from Fig. 9 in Katz & Miledi, 1970a). (F) The postsynaptic depolarisation caused by the leakage of calcium ions into the presynaptic terminal of a squid giant synapse from an intracellular electrode containing 0.5 mM calcium chloride. The retention bias on the calcium pipette was switched off at the first arrow and back on at the second, with tetrodotoxin present (from Fig.1 in Miledi, 1973) (G) The postsynaptic depolarisation caused by injecting a pulse of calcium into a presynaptic terminal of a squid giant synapse after replacing calcium with manganese in the extracellular fluid bathing the ganglion. The top trace monitors the current flowing through the calcium pipette inside the nerve terminal, the middle trace the presynaptic membrane potential and the lower trace gives the postsynaptic potential (from Fig. 3 in Miledi, 1973).
present simultaneously in a critical position. Such co-operation is necessary for release of a unit of transmitter by the nerve impulse. Given what we now know about the strategic organisation of the vesicle-associated protein calcium-sensor synaptotagmin and the presynaptic proteins syntaxin and neurexin, this sentence of Dodge and Rahamimoff seems extraordinarily prescient. However it is not clear that they are referring to the position of CaX at the inside or the outside of the membrane, but it is
DEVELOPMENT OF THE CONCEPT OF A CALCIUM SENSOR IN TRANSMITTER RELEASE AT SYNAPSES
147
clear that they are considering that it is on the nerve terminal. One point that was made clear at this time however was that calcium ions had to be present before the terminal was depolarised in order for transmitter to be released (Fig. 8.5C). If calcium was present immediately after depolarisation then there was no release (Fig. 8.5C). This led to the comment by Katz and Miledi (1967a) that: It may be that the opening of the external membrane ‘gates’ to Ca2+ or CaR+ is a transient event much briefer than the subesquent rise and fall of the probability of release. Or some calcium ‘carrier’ or ‘receptor’ only appears for a brief initial interval of time on the external surface of the axon membrane. Again it is clear that there is no way of distinguishing at this time between the two hypotheses of calcium movement through the nerve terminal or of that of the preformed CaX (of which there must be four together within a critical area) into the terminal for transmitter release. The issue was crystallised in terms of testing what came to be known as the ‘calcium hypothesis’, namely that …transmitter release is brought about by influx of external calcium ions through special membrane channels which are ‘opened’ by the depolarizing pulse (Katz & Miledi, 1970). At the end of the 1960s it was not clear if the calcium hypothesis was correct or whether CaR+ had to enter the terminal. An important experiment that showed at least that calcium did enter the terminal was performed at this time by Katz and Miledi (1969), who showed that a regenerative response could be obtained from the presynaptic nerve terminal in the squid stellate ganglion in the presence of tetrodotoxin to block sodium influx and TEA to block potassium efflux. Under these circumstances there was a long-lasting depolarisation with a duration of several hundred milliseconds after the depolarising stimulus (Fig. 8.6A). This response depended on the presence of calcium ions in the medium and was counteracted by magnesium and especially by manganese. It would seem in hindsight that the evidence was clear at this time that the regenerative response observed was due to calcium entering the nerve terminal. However, there was a complicating factor, namely that on removal of sodium ions the response eventually disappeared, leaving open the possibility that these ions contributed to the response. In a further investigation of the role of calcium in transmitter release at the giant synapse in the stellate ganglion at this time, Katz and Miledi (1970a) measured the size of the postsynaptic response for a given presynaptic depolarisation in the presence of tetrodotoxin to block the generation of impulses. The postsynaptic response increased maximally with an increase in the calcium concentration, on log/log coordinates, with a power of ca 2.7 at a given level of presynaptic depolarisation (Fig. 8.6C). This may be compared with the changes in size of the postsynaptic response with changes in calcium at the motor endplate, during impulse evoked transmitter release, in which the slope between the log of the response and the log of the calcium concentration is ca 4 (Fig. 8.6D). The implication is that some difference might exist between the number of calcium ions that must simultaneously act at the calcium sensor in the terminal of the stellate ganglion vs that in the endplate. The curves in Fig. 8.6C from this work show that at constant external calcium concentration and varying voltage a ‘permeability coefficient’ for calcium movement across the terminal might be derived, that changes with voltage in the way shown in Fig. 8.6E (curve B). But this of course presupposes the calcium hypothesis without proving it. 8.6 Direct evidence for calcium entry across the nerve terminal membrane during an impulse The introduction of the bioluminescent protein aequorin from Aequorea forskalea for use as a calcium indicator by Shimomura et al. (1962) allowed Ashley and Ridgway (1968, 1970) to make direct tests of the influx of calcium ions into cells. Baker et al. (1971) used this indicator in the giant axon of Loligo forbesi, and found that calcium entry associated with a depolarising pulse could be divided into an early component which was abolished by tetrodotoxin and a late component which was unaffected by the toxin (Fig. 8.7A and B). The time relationships of the early calcium entry were consistent with its being a small leak of calcium ions through the sodium channel, amounting to < 1% of the sodium current (Fig. 8.7B). The late entry of calcium ions on the other hand was greatly reduced by manganese ions and unaffected by tetrodotoxin or TEA ions (Fig. 8.7B). This entry increased steeply between 35 mV and 50 mV, reached a maximum at a depolarisation of 70 mV and declined to a low value with a further increase from 80 mV to 120 mV (Fig. 8.7A). These results are quantitated in Fig. 8.7C, which shows the amplitude of the response plotted against the applied voltage for three different calcium/ magnesium concentrations. The lower part of the figure shows the square root of the light intensity which is a better measure of ionised calcium than the first power of the response. Of particular interest is that this voltage-response curve for the late entry of calcium ions has a well defined maximum and a similar shape to the curve relating calcium entry to depolarisation at the
148
HISTORY OF THE SYNAPSE
presynaptic ending in the stellate ganglion investigated by Katz and Miledi (1970a) and shown in Fig. 8.6E. In the words of Baker et al. (1971): In several respects the properties of the late component resemble those of the channel which is responsible for release of transmitter at the presynaptic nerve terminals in the stellate ganglion of Loligo (Katz & Miledi, 1967c, 1969). In both cases, manganese and magnesium reduce calcium entry, and the curves relating aequorin and transmitter response to pulse amplitude both have a maximum at 70– 80 mV. Between 35 and 45 mV the square root of the aequorin response increased e-fold in 6 mV, and a similar figure is obtained from Figure 9 of Katz and Miledi (1970a) which gives calcium permeability as a function of membrane potential. These similarities raise the possibility that calcium channels of the kind described by Katz and Miledi may be present all the way along the axon, though probably at greater density in the presynaptic terminal. To a large extent this provided compelling evidence that the regenerative potential changes in the nerve terminal of the stellate ganglion that Katz and Miledi (1969) had observed (Fig. 8.6A), in the presence of tetrodotoxin and TEA ions, were indeed due to the inward movement of calcium ions, despite their caveats concerning the slow run down of the response in sodiumfree solutions. Llinas et al. (1972) were able to show, using aequorin injected into the presynaptic terminal of the squid stellate ganglion, that there was indeed a substantial calcium entry during impulse firing (Fig. 8.6B), although even at this stage they were unable to resolve calcium entry into the presynaptic terminal down to the level of single impulses. With the establishment of one arm of the calcium hypothesis, namely that there is an entry of calcium ions into the terminal due to the nerve impulse, it remained to establish that it was these ions that brought about the release of transmitter. Miledi (1973) carried out a key experiment to test the calcium hypothesis, namely to induce transmitter release by simply injecting calcium ions into the nerve terminal. If a high calcium containing electrode was placed in the presynaptic terminal in the giant synapse of squid, then the leakage of calcium from the electrode gave rise to to a postsynaptic depolarisation (Fig. 8.6F). Fourier analysis of this depolarisation showed that it was made up of the additions of units of about 25 mV amplitude which decayed with time constants of ca 2 ms, approximately the temporal characteristics of the spontaneous miniature postsynaptic potentials. Pulses of calcium into the terminal, in the presence of zero external calcium and high manganese, gave rise to discrete postsynaptic transients (Fig. 8.6G). These transients must have been due to the action of the released calcium within the terminal, and not to the leakage of calcium out of the terminal to act on extracellular receptors, as these would have been blocked by the manganese which is known to prevent the action of external calcium in promoting transmitter release. Furthermore, changes in the presynaptic membrane potential due to the calcium current were not responsible for the transmitter release, as similar changes produced with a potassium chloride filled electrode did not affect transmitter release. Taken together the results indicated that the response to an increase in calcium in the terminal consisted of a greatly accelerated release of units of transmitter equivalent to the units responsible for the spontaneous miniature postsynaptic potentials. This experiment established that the transient elevation of intraterminal calcium, which then acted on some receptor within the terminal, could accelerate the release of transmitter units so that it was no longer necessary to propose that a compound Ca+R might be formed on the outside of the terminal and transported in for transmitter release to occur. The calcium hypothesis that ‘transmitter release is brought about by influx of external calcium ions through special membrane channels which are ‘opened’ by the depolarizing pulse’ was now strongly supported by the experiments enumerated in this section. First, it was clear from the observations of Baker et al. (1971) that calcium ions move through the nerve membrane, probably using specific channels that can be blocked by the same agents that block transmitter release such as manganese. Second, these channels seem to have similar phenomenological characteristics to those observed in the nerve terminal in an indirect way by Baker et al. (1971). Third, calcium release directly into the nerve terminal from an electrode releases transmitter, as shown by Miledi (1973). However the experiments performed at the beginning of the 1970s were unable to show that the action potential releases transmitter in the same way as an intraterminal pulse of calcium ions, especially given the very different time courses of transmitter release by the former compared with the latter (see Fig. 8.6G). For example, the possibility had not been eliminated that changes in the membrane potential itself might be necessary to change the properties of the receptor that intracellular calcium binds to in order to trigger the fast transient transmitter release which occurs in response to an impulse. Furthermore there was no direct evidence for the existence of any kind of ionic channel at this time; these were simply inferred from the kind of phenomenological experiments that Hodgkin and Huxley had carried out in the early 1950s. In the next section consideration is given to the discovery of the calcium channels and their role in transmitter release.
DEVELOPMENT OF THE CONCEPT OF A CALCIUM SENSOR IN TRANSMITTER RELEASE AT SYNAPSES
149
Fig. 8.7. Evidence for calcium channels in neurons. (A) Calcium indicator measurements of calcium fluxes into neurons. Changes in ionised calcium in the giant axon of the coelenterate L. forbesi measured voltage clamp to the voltages shown by recording the light produced by injected aequorin. Aequorin responses are shown for 100 msec. Measurements made in artificial sea water containing tetrodotoxin to block voltage-dependent sodium channels, TEA ions to block voltage-dependent potassium channels, 22 mM calcium and 400 mM of sodium (from Fig. 20 in Baker et al., 1971). (B) The effect of pulse amplitude on the aequorin response to single pulses of duration 100 msec in the giant axon of L forbesi. The abscissa is the amplitude of the pulse and the ordinate is the peak of the aequorin response A, or the square root of the peak response B. The composition of the external solution was: curve 1, 112 mM Ca, 0 mM Mg; curve 2, 22 mM Ca and 0 mM Mg; curve 3, 22 mM Ca and 90 mM Mg. (from Fig. 21 in Baker et al., 1971). (C) Evidence that calcium passes into the giant axon of L. peroni through both the voltage-dependent sodium channel and a separate calcium channel using the calcium indicator aequorin. For each graph the upper curves give the relation between the pulse duration (abscissa) and the increment in light intensity due to aequorin per pulse. A,a before tetrotodotoxin (TTX); B, b in TTX; C,c after removal of TTX; A, B, C 80 mV; a–c 120 mV; the ordinate was measured as the initial rate of rise of light intensity at 200 pulses per sec divided by the initial rate of rise of light intensity at 200 action potentials per sec. For the middle curves the ordinate is the TTX-sensitive component of the calcium entry obtained as D=A−B or d=a−b before TTX and E=C−B or e=c−b after TTX. The bottom curves give the sodium conductance as a function of time before (F, f) and after (H, h) treatment with TTX; (O) by subtraction; (X) from amplitude of tails. Curves G and g give the time course of the potassium conductances estimated from the records in TTX. F, H, G 80 mV; f, h, g 120 mV (from Fig. 17 in Baker et al., 1971).
8.7 Calcium channels in the nerve terminal 8.7.1 Calcium currents and evidence for calcium channels It was not until the late 1970s that the neurons of terrestial molluscs such as the snails Helix aspersa and pomatia were
150
HISTORY OF THE SYNAPSE
Fig. 8.8. Calcium channels in neurons. (A) Clear separation of the sodium and calcium inward currents in ganglionic neurons from the molluscs A Limnea stagnalis and B Helix pomatia. The current traces for the different voltage-clamp levels are shown in the presence of normal sodium concentrations as well as in the absence of sodium. The holding potential was −50 mV. Note that the calcium current is substantially larger in Helix compared with Limnea, that is the current that remains after the relatively fast transient sodium current at the beginning of the trace disappears in the sodium-free solution (from Fig. 3 in Kostyuk et al., 1977b). (B) Membrane currents before, during and after prolonged depolarisation of a voltage-clamped neuron from the mollusc Helix aspersa. The voltage was stepped to +40 mV from a holding potential of −60 mV. The left-hand panel shows the current noise traces and the large transients due to the high-pass filter; right-hand panels show the corresponding low gain recordings of membrane currents Im after leakage correction. Current noise was analysed during the 15 sec periods indicated by the horizontal bars while Im was steady. Note the different current calibrations for the noise and Im panels, b, After suppression of 1k with caesium. Addition of tetrodotoxin or substitution of Tris for sodium had no further effect, and the current noise in (b) is attributed to fluctuations in Ica and leakage current; Im was corrected for leakage and has transient and steady inward components, c, After addition of nickel to the external solution. The difference in noise between (b) and (c) is attributed to suppression of Ica by nickel. Im is outward because of leakage (from Fig. 1 in Akaike et al., 1978).
examined under a voltage-clamp to find out if, as for Aplysia neurons, large inward calcium currents exist besides that of the usual inward sodium current (Standen, 1975). In that year Krishtal and Pidoplichko (1975) introduced the technique of internally dialysing the neuron somas of molluscs such as Limnea in conjunction with the voltage-clamp, a technique that was then perfected by Kostyuk et al. (1977a), Kostyuk & Krishtal (1977) as well as by Lee et al. (1977). This allowed adequate spatial and temporal control of the clamp voltages making possible, amongst other things, for a clear separation of the calcium and sodium currents. For instance, Kostyuk et al. (1977b) were able to show that replacement of potassium ions in these
DEVELOPMENT OF THE CONCEPT OF A CALCIUM SENSOR IN TRANSMITTER RELEASE AT SYNAPSES
151
neurons with Tris entirely eliminates the outward potassium current, revealing an inward current with two components, one carried by sodium ions and the other by calcium ions, with the latter blocked by cadmium (Fig. 8.8A). Interestingly, the inactivation of the calcium current included two first-order kinetic processes, indicating the possibility that more than one calcium channel type mediates the current. However the major breakthrough with these new internal perfusion and voltageclamp techniques was that they allowed the noise generated by the stochastic opening of individual calcium channels to be clearly detected (Fig. 8.8B; Akaike et al, 1978). After blocking both sodium and potassium currents with appropriate ion substitutions, Akaike et al. (1978) found that the depolarizing voltage step initiated a large increase in the total membrane current noise due to calcium, as shown in Fig. 8.8(Bb) compared with the control in the absence of a calcium current, shown in Fig. 8.8(Bc). Analysis of these current fluctuations in terms of power density spectra showed that they consisted of units with conductances of about 10−13 S and with activation time constants of ca 1 ms and two inactivation time constants of 10– 15 ms and 250–350 ms. The 1970s then ended with the first estimates of the properties of individual calcium channels in the nerve membrane. But it was not until the 1980s that techniques became available for direct observation of individual calcium channel activity. 8.7.2 Individual calcium channels identified The introduction of the patch clamp technique by Neher and Sakmann (1976) allowed the currents in single ionic channels to be observed for the first time. Electrical seal resistances of ca 50 M could be obtained using a small heat polished glass pipette pressed against a cell membrane (Neher et al, 1978), ensuring that most of the current originating in the patch of membrane at the tip of the pipette flowed into the pipette and thence to the recording apparatus. Neher went on to show in 1981 that if the pipette surface is cleaned and suction is applied to the pipette interior, then seal resistances of the order of 10 to 100 G . could be obtained. These greatly decreased the background noise of the recordings (Sigworth & Neher, 1980), providing for a much improved patch-clamp technique for high-resolution current recording from cells and cell-free membrane patches (Hamill et al., 1981). The technique was first applied to bovine chromaffin cells for the purposes of recording from single calcium channels by Fenwick et al. (1982a) (Fig. 8.9B) and the neurons of Helix pomatia by Lux and Nagy (1981) (Fig. 8.9A). In the latter case channel openings were observed with a mean open time of 2.5 msec and a current of about 0.8 pA in isotonic barium chloride (Fig. 8.9A (a, b)). The way in which the macroscopic calcium currents that are observed under a conventional whole voltage-clamp of these neurons are generated from the channel currents was made clear by simply computing the average patch current from the samples of the single patch currents under a particular voltage (Fig. 8.9A (c, d)). In the case of the chromaffin cells, the calcium channels had a current amplitude at −5 mV in isotonic barium of 0.9 pA (Fig. 8.9BA) and a mean open time of 0.8 ms (Fig. 8.9BB) and closed time of time constants 1.05 ms and 25.5 ms (Fig. 8.9BC). The very good agreement between these studies on different cell types as to the channel current may be contrasted by the large differences in the open time estimates, which may be attributed to the greater temporal resolution in the study on the chromaffin cells, showing up many very small duration open times <1 ms (Fig. 8.9BB). However, Brown et al. (1982) examined the single calcium currents in a number of different neuron types, including Helix pomatia again (Fig. 8.9C) and chick dorsal root ganglion cells. They found that while the current amplitudes were similar for single calcium channels in these neurons (ca 0.5 pA at −20 mV or −10 mV), the open times were not, with ca 2.8 msec for the Helix neurons and 0.5 msec for the chick neurons, indicating some genuine heterogeneity in channel types. An examination of calcium channels in single rat cardiac cells in culture by Reuter et al. (1982) gave single calcium currents of ca 1 pA and average open times of 1 msec (Fig. 8.9D), close to those of the majority of recordings made on different cell types to that date. 8.7.3 Identification of different calcium channel types Although there had been a number of investigations as to the properties of single calcium channels between 1982 and 1984 (see for example Carbone & Lux, 1984), a very considerable clarification was brought about by Nowycky et al. (1985 a, b) with their systematic analysis of the kinetics and pharmacology of the calcium channel types to be found in chick dorsal root ganglion cells. They showed that three different channel types could be distinguished on the basis of their kinetics and also their sensitivity to dihydropyridines (Fig. 8.10A): large conductance channels that contribute long-lasting current at strong depolarisations with an open probability greatly increased with dihydropyridine agonists (the L-type channel with a current of 2.1 pA at −20 mV; Fig. 8.10A see L); a relatively small conductance responsible for the transient current activated at weak depolarisations (the T-type channel with a current of 0.6 pA; Fig. 8.10A see T); and a third channel that requires strongly negative potentials for complete removal of inactivation and strong depolarisation for activation (the N-type channel with a current of 1.2 pA; Fig. 8.10A see N). In the conclusion to their paper Nowycky et al. (1985b) comment that:
152
HISTORY OF THE SYNAPSE
Fig. 8.9. Calcium channels in neurons and secretory cells. (A) Calcium channel currents in a membrane patch of the neuron soma of the mollusc Helix pomatia during depolarizing voltage-clamp steps, a, Shows the inward and outward current fluctuations during a series of clamp steps to −10 mV. Patch pipette contained 60 mM of CaCl2, 0 NaCl. b, Patch currents with internal and external (patch) TEA. Pipette contained 50 mM CaCl2, 0 Na Cl and 2.5 M.M of tetrodotoxin and 20 mM TEA chloride. The first three records at indicated membrane potentials are shown, holding potential was −50 mV. c, Computed average patch current (thick trace) with superimposed square root of the variance of 32 samples under conditions similar to those in (b). Variance of preceding background activity was subtracted from current variance during pulses, d, Processing as in (c) of 22 records with BaCl2 replacing CaCl2 in the pipette (from Fig. 1 in Lux & Nagy, 1981). (B) Calcium channel currents in a membrane patch of a bovine chromaffin cell. Single current recordings are shown from calcium channels. Cell-attached recordings were obtained with a barium-filled pipette. The patch contained normal saline. A, responses to 70 mV (left) and 90 mV (right) depolarisations. The pipette holding potential was the bath potential. The test potential was ca— 5 mV in the left panel and +15 mV in the right panel. Background currents have been subtracted. In the right panel, the zero current level during pulses is indicated by horizontal lines. Opening rates and mean channel durations were larger at +15 mV than at— 5 mV. At +15 mV, simultaneous openings of several channels were frequently observed (second and third traces on the right). The unit sizes of the single channel events were 0.9 pA and −5 mV and 0.6 pA at +15 mV. 1 kHz low pass filter. B, Open time histogram for the data at −5 mV illustrated in the left part of A. This histogram was approximated by a single exponential with the time constant of t0=0.81 ms. C, Closed time histogram for the same data at −5 mV. The histogram was approximated by a double-exponential curve with time constants tc=1.05 ms and ts=25.5 ms. tc is the mean duration of the short gaps occasionally seen during the single events of the left panel in A. ts represents the mean time between independent single channel events. The mean number of gaps per burst, v, is equal to the ratio of the integral of the fast component to that of the slow component. Its numerical value is 0.57 (from Fig. 16 in Fenwick et al., 1982b). (C) Calcium channel currents from a membrane patch of the soma of a neuron in the mollusc H. pomatia. The entire neuron was voltageclamped. The top traces show the patch calcium currents and the bottom traces the whole-cell membrane calcium currents during steps to the indicated membrane potentials from a holding potential of −50 mV. The calcium concentration in the patch pipette and bath solution were identical (40 mM). …, Average unit current amplitude;—, baseline values. As the potential was increased, the frequency of channel openings increased, simultaneous openings from 2 units appeared and unit amplitudes decreased. The data were filtered at 1 kHz (from Fig. 1 in Brown et al., 1982). (D) Calcium channel currents from a membrane patch of a rat cardiac myocyte. Voltage dependence of single calcium channel currents in a patch of a cardiac myocyte. Calcium channel currents at several voltages. Membrane potential was stepped by the indicated step size ( V) from a holding potential of −10 mV (resting potential of ca −60 mV). The dotted lines indicate the measured open channel current. Beside each record is the average probability (p) that the channel was open, as determined from the individual record shown. This patch had only a single active calcium channel. Data were filtered at 1 kHz (from Fig. 2a in Reuter et al., 1982).
N- or L-type channels may be important for dendritic spikes or neurotransmitter release.
DEVELOPMENT OF THE CONCEPT OF A CALCIUM SENSOR IN TRANSMITTER RELEASE AT SYNAPSES
153
This comment proved to be most prescient, given the subsequent research showing the importance of the N-type channel for transmitter release at many synapses (see below). Although there was some initial resistance to the clarity of the analysis leading to the definitions of L-, T- and N-type channels and to this nomenclature (for as Bernard Katz has remarked, many scientists would rather use the tooth brush of a colleague than accept his nomenclature), it has now been almost universally adopted, particularly with the discovery of toxins and drugs that act as agonists or block one channel type but not the others. For example the dihydropyridines act exclusively on the L-type whereas omega-conotoxin GVIA, from the venom of the fish-hunting cone snail Conus geographus, acts exclusively to block the N-type (Kerr & Yoshikami, 1984; Olivera et al., 1994). Other cell types were soon shown to possess different combinations of these channel types, with for instance T-type as well as L-type occurring in the membranes of guinea-pig ventricular myocytes, in which the pharmacological sensitivity of the L-type channel to the dihydropyridine agonists (such as Bay K 8644) allowed for its easy distinction from the T-type channel (Fig. 8.10B; Nilius et al., 1985). A full kinetic analysis of these three calcium channel types in the membrane of chick dorsal root ganglion cells was carried out by Fox et al., (1987). They showed that the L-type channel has a unitary slope conductance of 25 pS, with brief openings of ca 1. 2 msec and could be blocked with the dihydropyridine antagonist nifedipine (Fig. 8.11A). On the other hand the N-type channel has a smaller unitary slope conductance of 13 pS with an open time of 0.8 ms and inactivated over a very broad range of holding potentials (Fig. 8.11B). In contrast the T-type channel possesses a very small unitary slope conductance of 8 pS, and an open time which is very short and about 0.5 msec (Fig. 8.11C). By the end of the 1980s evidence for a further calcium channel type was forthcoming from studies on brain neurons, in particular those in the cerebellum. Funnel web spider venom was used by Llinas et al. (1989) in affinity gels to obtain a protein from the guinea-pig cerebellum which when reconstituted in lipid vesicles and fused with lipid bilayers possessed the characteristices in patch-clamp recordings of a new calcium channel type (Fig. 8.11D). This was referred to as a P-type channel. It possesses a conductance of ca 10 pS, an open time of 1–3 msec, is unaffected by omega-conotoxin GVIA or dihydropyridines and is blocked by a low molecular fraction from the funnel web spider Agelenopsis aperta as well as to the peptide omega-Aga IVA (at IC50 of 0.3 nM) that has been isolated from the venom of this spider. This toxin has been useful in the delineation of a further current that is carried by a channel referred to as the Q-type with a conductance of about 16 pS (with an IC50 of about 20 nM; Wheeler et al., 1994a). The different IC50 values for the block of P- and Q-type channels by omegaAga IVA indicate that it is possible with care to discriminate between the two channel types using different concentrations of the toxin. Another toxin termed omega-CTX-MVIIC blocks N-type, P- and Q-type channels (Wheeler et al., 1994b). Finally a calcium channel for which no selective antagonist has been found, and which is distinct from the T-type channel, has been termed the R-type channel; as yet the kinetics of this channel have been poorly characterised (Wheeler et al., 1994). In the 1990s molecular cloning has shown the diversity of calcium channel types derived from multiple genes for calcium channel subunits involving alternative splicing (Snutch & Reiner, 1992). The voltage-gated calcium channels are made up of a comparatively large transmembrane protein of 200–260 kDa named the alpha 1 subunit, together with two other subunits called alpha 2-delta and beta. The voltage sensor, gating machinery and channel pore are all found in the alpha 1 subunit, with each of these capable of being modified by the other subunits (Mori et al., 1991). The multiple forms of the alpha 1 subunit give rise to a considerable variety of calcium channels. The mammalian brain contains genes encoding five different forms of calcium channel alpha 1 subunit (Snutch et al., 1990). These are termed A, B, C, D and E (Fig. 8.11E; Llinas et al., 1992b). The sequence homology of the alpha 1 subunit allows it to be organised into two subfamilies. One subfamily possesses alpha 1 cDNA’s which code for calcium channels that are of the L-type, and include class C, D and S clones (Fig. 8.1 1E; Mikami et al., 1989). The other alpha 1 subfamily consists of clones A, B and E with cDNA’s that are derived from nervous tissue (Fig. 8.11E; Williams et al., 1992). One member of this subfamily is alpha1B; when this unit is expressed it gives rise to Ntype calcium channels (Fig. 8.11E; Williams et al., 1992). The functional properties of the class A and E subunits is less clear, but it seems likely that class A encodes the P- and Q-type channels and class E the R-type channel (Fig. 8.11E; Wheeler et al., 1994b). 8.7.4 Identification of the different calcium channel types involved in transmitter release The contribution of different calcium channel types to the influx of calcium that triggers transmitter release at different terminal types has been delineated with the use of the different drugs and toxins that are specific to a particular channel type as described above. It is therefore helpful at this stage to summarise the properties of these agents, before considering their use to determine the channel types involved in synaptic transmission. The 1, 4-dihydropyridines (DHPs) are synthetic organic compounds that can be used to identify a group of calcium channels which produce macroscopic currents that are long lasting and hence the channels that mediate these currents are referred to as L-type calcium channels. The single channels currents have a conductance of ca 25 pS. These L-type channels are not in large numbers in neurons, and do not seem to play a role in evoked transmitter release due to an action potential at any synapse so far investigated. The omega peptide toxins can
154
HISTORY OF THE SYNAPSE
Fig. 8.10. Determination of different classes of calcium channels. (A) Three types of unitary calcium channel activity seen in cell-attached patch recordings with barium as the charge carrier in chick dorsal root ganglion cells, a, T-type channel; b, an N-type channel; c, an L-type channel. The patch pipettes contained (in mM): 110 BaCl2, 10 HEPES and 200 nM TTX, pH 7.4. To put patch membrane potential on an absolute scale, cell resting potential was zeroed with an external solution containing (mM): 140 K-aspartate, 10 K-EGTA, 10 HEPES, 1 MgCl2, pH 7.4. Current signals were filtered at 1 kHz and sampled at 5 kHz. Within each panel traces are consecutive. The unitary current amplitude of the T-type channel at −20 mV was given by the average of amplitudes from three different patches exhibiting long, well-resolved openings. We have seen each of the three channel types in isolation and in all possible combinations. The number of channels under the patch pipette varied considerably from one to many (from Fig. 2 in Nowycky et al., 1985a). (B) Differential effects of the dihydropyridine calcium channel agonist Bay K 8644 on the two channel types with barium or calcium as the charge carrier, a, b, Current traces with 110mM Ba and 33 mM tetrodotoxin in the pipette from a patch containing both channel types, before (a) and after (b) exposure of the cell to 5 mM Bay K 8644. Sweeps with no detectable openings are not shown. Lowest traces in each column are averaged currents from all sweeps (234 in (a) and 274 in (b)). Column b also includes a smaller average currrent signal (just above lowest trace), obtained from 194 selected sweeps that showed no detectable activity of the large conductance channel. C and d, 110 mM Ca in the pipette. Current traces from a patch with only L-type Ca channel activity before (c) and after (d) exposure of the cell to 5 mM Bay K 8644. Sweeps with no detectable openings not shown. Mean currents (below individual traces) are averages of 151 (c) and 518 (d) sweeps (from Fig. 3 in Nilius et al., 1985).
discriminate at least two other channel types which are active in mediating the secretion of transmitter at synapses. Omegaconotoxin GVIA (omega-CTX GVIA) is from the venom of the fish-hunting cone snail, Conus geographus, and blocks a component of the DHP resistant current in neurons, that is identical to the N-type calcium current through channels of ca 12 to 18 pS (Kerr & Yoshikami, 1984). Calcium currents also exist that are unaffected by either DHPs or by omega-CTX GVIA. The venom of the funnel web spider Agelenopsis aperta (omega-Aga IVA) has assisted in the delineation of these currents
DEVELOPMENT OF THE CONCEPT OF A CALCIUM SENSOR IN TRANSMITTER RELEASE AT SYNAPSES
155
that are carried by the P-type calcium channel with conductances of between 10 and 20 pS (with an IC50 of about 0.3 nM; Llinas et al., 1989); these channels are also blocked by a low molecular weight fraction of the funnel web spider toxin. A fourth kind of calcium channel, the T-type, is resistant to all of the above agents but is easily characterised by its fast inactivation kinetics, low activation voltage, and small single-channel conductance. A fifth type of calcium channel for which no selective antagonist has been found, and which is distinct from the T-type channel, has been termed the R-type channel; as yet the kinetics of this channel have been poorly characterised (Mintz et al., 1992 a, b). Finally, a sixth type of channel exists termed the Q-type which is blocked by relatively high concentrations of Aga IVA (with an IC50 of about 20 nM; Wheeler et al., 1994). This channel is also blocked by a toxin termed omega-CTx-MVIIC which is an antagonist for N-type channels and so cannot be used to distinguish the Q-type channel definitively (Wheeler et al., 1994a). The toxins that have helped delineate the different classes of calcium channels enumerated above have been used to determine the relative contributions of these channels to the process of transmitter release at different synapse types. In the central nervous system omega-CTX GVIA antagonises between 10 and 20% of the potassium depolarisation-induced release of glutamate, GABA and noradrenaline from synaptosomes or tissue slices of the hippocampus (Burke et al., 1993; Dooley et al., 1988), acetylcholine from the neocortex (Wessler et al., 1990) and dopamine from the striatum (Woodward et al., 1988). N-type calcium channels therefore contribute to the potassium depolarisation induced transmitter release at these sites, but are not the only calcium channel that do so. Omega-Aga IVA toxin blocks 50– 70% of the calcium dependent release of glutamate from synaptosomes (Turner & Dunlap, 1995) indicating that P- and/or Q-type calcium channels also mediate transmitter release in the central nervous system. These studies with synaptosomes have been confirmed by examining the effects of the toxins on synaptic currents (Takahashi & Momiyama, 1993; Luebke et al., 1993; Dunlap et al., 1995). The synaptic current generated at synapses between CA3 and CA1 pyramidal cells (through the Schaffer collaterals) appears to be also due to a mixture of N-type and Q-type calcium channels (Wheeler et al., 1994b). The possibility that a combination of the N-type channel together with the P-/Q-type channels occurs at the terminals of central synapses has been entertained (Luebke et al., 1993) and greatly strengthened by the discovery that the synaptic current in cerebellar Purkinje cells due to stimulation of the single climbing fibre is reduced 29% by omega-CTX GVIA and 77% by omega-Aga IVA (Regehr & Mintz, 1994). Indeed recent measurements of the sensitivity of transmitter release from the boutons of cultured hippocampal neurons to different toxins suggest that individual terminals have different mixtures of N-type and P-/Q-type channels (Reid et al., 1997). In the peripheral nervous system it is now well established that although omega-CTX GVIA blocks the endplate potential at somatic neuromuscular junction in amphibia (Sano et al., 1987), and is therefore mediated by N-type channels, this is not the case at mammalian junctions, where the endplate potential is completely blocked by low concentrations of funnel web spider toxin in relatively low concentrations and is therefore mediated by P-type channels (Protti et al., 1993). Amongst autonomic ganglia, transmission at the mammalian superior cervical ganglion by single impulses, as measured with extracellular electrodes, is blocked by the low molecular weight fraction of funnel web spider toxin which seems to be specific for P-type channels and is slightly blocked by high concentrations of omega-Aga IVA that acts on Q-type calcium channels; however this transmission is unaffected by omega-CTX GVIA, nifedipine or low concentrations of omega-Aga IVA, indicating minimum contributions from N- and L-type calcium channels (Gonzalez Burgos et al., 1995). On the other hand direct measurement of the synaptic currents in this preparation reveal that besides the effect of omega-Aga IVA there is also substantial block using omega-CTX GVIA, indicating that N-type channels probably do contribute significantly to the secretion of transmitter at this synapse (Ireland et al., 1997). At sympathetic nerve terminals the excitatory junction potential in the guinea-pig and mouse vas deferens in response to low frequencies of stimulation is blocked by omega-CTX GVIA (Brock et al., 1989; Smith & Cunnane, 1996; Waterman, 1997) as is transmission to rat small mesenteric arteries (Waterman, 1997), indicating that N-type channels are engaged in transmitter release at these frequencies. At high frequencies and during long trains of impulses to the sympathetic nerves of the guinea-pig and mouse vas deferens an omega-CTX GVIA resistant component of transmitter release appears. In the case of the guinea-pig vas deferens, this is also resistant to blockade by the other known toxins and may be attributed to the R-type channel (Smith & Cunnane, 1996); in the case of the mouse it is blocked by omega-conotoxin MVIIC, indicating a role for P-/ Q-type channels in transmitter release at high frequencies (Waterman, 1997; de Luca et al, 1990; Wright & Angus, 1996). Confirmation of these observations that sympathetic nerve terminals release transmitter predominantly through N-type channels is that the release of noradrenaline from these terminals is blocked by omega-CTX GVIA in both the rat tail artery (Clasbrummel et al., 1989) and from the terminals of isolated sympathetic neurons in tissue culture (Koh & Hille, 1997). 8.7.5 Calcium channels and the transfer function of the nerve terminal At the beginning of the 1970s the calcium hypothesis had been enunciated, namely
156
HISTORY OF THE SYNAPSE
Fig. 8.11. Kinetic characteristics of L-, N-, T- and P- type calcium channels. (A) Kinetic characteristics of a single L-type calcium channel in a chick dorsal root ganglion cell. A, six consecutive current records evoked by a test pulse from a holding potential of −40 mV to a test potential of −20 mV (upper trace) and the mean current record obtained by averaging 152 current records (lower trace). B, cumulative distribution of latency to first opening. C, distribution of opening times. The smooth curve is a single exponential with a time constant=1.19 msec. D, Distribution of closed times fitted by the sum of two exponentials. Time constants of these exponential components are 0.36 ms and 0.22 ms. (B) Kinetic characteristics of the N-type calcium channels in a two-channel patch from a chick dorsal root ganglion cell. Out of a total of 120 sweeps, six sweeps showed double openings and were discarded. A, consecutive current records evoked from a holding potential of −60 mV to a test potential of +20 mV. The mean current trace obtained by averaging individual records. B, the cumulative distribution of latency to first opening, corrected for the participation of two channels. C, Distribution of open times; the distribution is fitted by a single exponential with a time constant of 0.77 ms. D, closed-time distribution, fitted by the sum of two exponentials. Time constants and amplitudes of the exponential components are fast time constant=0.66 msec, Nf=273 and slow time constant =4.87 ms, Ns=42.
…transmitter release is brought about by influx of external calcium ions through special membrane channels which are ‘opened’ by the depolarizing pulse (Katz & Miledi, 1970). By the beginning of the 1980s several of these channels had been identified at the single unit level and the problem presented itself of determining the quantitative relationship between the influx of calcium ions through these channels in the nerve terminal and the release of transmitter, often called the transfer function of the terminal. This was no easy task as it required successful clamping of the nerve terminal voltage to a given value and simultaneous determination of the quantity of transmitter released in terms of the size of the postsynaptic potential generated. Several attempts were made to measure this
DEVELOPMENT OF THE CONCEPT OF A CALCIUM SENSOR IN TRANSMITTER RELEASE AT SYNAPSES
157
(C) Kinetic characteristics of a single T-type calcium channel from a chick dorsal root ganglion cell. A, six consecutive current records evoked by depolarizing pulses from a holding potential of −80 mV to a test potential of −30 mV. Mean current trace obtained by averaging 195 individual records. B, cumulative distribution of latency to first opening. C, distribution of open times. Smooth curve is a single exponential with time constant=0.53 msec, fitted using the the least-squares analysis. D, closed-time distribution. The smooth curve is the sum of two exponentials. Time constants and amplitudes of the exponential components are fast time constant =0.9 msec, Nf=81, and slow time constant=6.25 msec, Ns=35. (A, B, and C from Figs. 2, 11 and 8 consecutively in Fox et al., 1987). (D) Kinetic characteristics of P-type calcium channels in a neuron from a guinea-pig cerebellum. The channels were isolated using a funnelweb spider toxin affinity gel and then reconstituted in lipid vesicles which were fused with a lipid bilayer. 80 mM barium was in the external solution and the holding potential was −45 mV in A, −30 mV in B, and −15 mV in C. D, A single channel recording with 80 mM barium on the cytoplasmic face of the channel. Holding potentials of −45, −30 and −15 mV are shown. In this particular recording’, three identical channels had fused with the bilayer. The long duration of the open time seems to be an effect of the high amount of internal barium, as it can be converted into the short-duration channel (A–C) when barium is removed from the cytoplasmic side. E, the I/V curves in 80 mM barium for the ‘fast’ channel of A to C constructed from the single channel events obtained from six different experiments. Each point represents the mean current from 200 or more events at each voltage. F, Plot of open probability versus applied voltage showing the voltage dependence of the channel, which reaches a maximum value of 0.6–0.65 above 0 mV. G, The values from E to F multiplied to give an approximation of the macroscopic current (from Fig. 2 in Llinas et al., 1989). (E) The cDNA sequence identity of alpha 1 subunits of voltage-gated calcium channels used to determine their structural relationships. The correspondence between the functional calcium channel types (L, P, Q, N and R), described above, and the corresponding alpha 1 subunit class (A, B, C, D, E and S) is shown together with those tissues that express high levels of a particular class of subunits (from Fig. 1 in Wheeler et al., 1994 adapted from Zhang et al., 1993).
transfer function in the giant synapse in the squid stellate ganglion, but the problem was essentially that of obtaining an adequate spatial voltage-clamp over the presynaptic nerve terminal for, without this, gradients in the spatial voltage distort the time course and voltage dependence of the measured calcium currents and produce heterogeneous entry of calcium ions and transmitter release (Llinas et al., 1981). As a consequence of this difficulty, different values were obtained for the relationship between the logarithm of the calcium entry and the logarithm of transmitter release, which on the basis of studies at the neuromuscular junction was expected to have a gradient of about 3–4 (Dodge & Rahamimoff, 1967), reflecting the number of sites on the calcium sensor molecule that must be occupied by calcium ions in order for transmitter release to be triggered.
158
HISTORY OF THE SYNAPSE
Some claims were that this gradient was ca 1 at the squid giant synapse, indicating that the mechanism of transmitter release at this synapse might be different to that at the neuromuscular junction. The problem of eventually obtaining an accurate transfer function was eventually solved by Augustine et al. (1985a, b). These authors used the voltage-clamp technique to measure the voltage dependence of calcium ion influx into the squid presynaptic terminals, using a three electrode voltageclamp originally designed by Adrian et al. (1970), combined with local application of calcium to the terminal as shown in Fig. 8.12A. The quality of the clamp method for measuring the calcium current was checked using the calcium indicator dye Arsenazo III. Comparison between the peak Arsenazo III signal and the integral of the calcium current measured with the clamp showed that they corresponded very closely (Fig. 8.12B), indicating the accuracy of the voltage-clamp procedure. Using this technique, it was possible to show that the gradient relating the logarithm ot the calcium entry to that of transmitter release was ca 3 (range 2.4–3.5; Fig. 8.12C). This work then showed quantitatively for the first time that the calcium sensor for transmitter release must involve multiple calcium binding sites to initiate the release of transmitter, as had been conjectured by Dodge and Rahamimoff (1967) some 17 years earlier. 8.8 Identification of the calcium sensor molecule 8.8.1 Synaptotagmin Matthew et al. (1981) obtained a monoclonal antibody against a 65 kDa molecule (P65) of synaptic vesicles which was highly conserved in vertebrate phylogeny and present in many different types of nerve terminals. It was soon speculated that that it might play a role in the process of exocytosis of the synaptic vesicle given the ubiquitous distribution of P65 in nerve terminals of both vertebrates and invertebrates (Trifaro et al., 1989). This possibility was greatly strengthened by the discovery that the cytoplasmic domain of P65 contains an internally repeated sequence that is homologous to the regulatory C2-region of protein kinase C, and that this domain binds phospholipids as well as proteins such as calmodulin in a calciumdependent manner (Trifaro et al., 1989). P65 was shown to contain two copies of an internal repeat that is homologous to the regulatory region of protein kinase C. The protein altogether contains five domains (Littleton & Bellen, 1995; Fig. 8.13A): a single transmembrane at the N-terminus that spans the membrane of synaptic vesicles; a sequence separating the transmembrane region from the two repeats (C2a and C2b) that are homologous to protein kinase C; the two repeats homologous to protein kinase C with C2a closest to this transmembrane region; and a carboxyl-terminal sequence following these two repeats (Perin et al., 1991). By 1992 the relationship between phospholipid binding by synaptotagmin, as P65 had now been named, and calcium became clearer. This occurred as a consequence of the description of the cytoplasmic domain of synaptotagmin binding calcium at physiological concentrations in a complex with negatively charged phospholipids, indicating that the molecule might act as the cooperative calcium receptor in exocytosis of vesicles (Brose et al., 1992). Synaptotagmin is anchored at its C-terminal end to the pre-synaptic membrane-bound protein, syntaxin (35 kD; Bennett et al., 1992; Fig. 8.13C). This protein is concentrated at synaptic sites and blocking it with a monoclonal or polyclonal antibody leads to blockade of evoked transmitter release without affecting the influx of calcium ions at the pre-synaptic membrane (Mochida et al, 1995). 8.8.2 Physiological evidence that synaptotagmin is the calcium-sensor molecule Mutations in synaptotagmin of Drosophila (Littleton et al., 1993; 1994; Mochida et al., 1995) and of C. elegans (Nonet et al., 1993) decreases evoked transmitter release by over an order of magnitude. The fact that the animals survive is due to the remaining low level of evoked asynchronous transmitter release. Removing the second calcium binding domain from synaptotagmin in Drosophila mutants decreases the fourth order calcium dependence of transmitter release to two (Littleton et al., 1994). (Note that the first of the C2 domains in synaptotagmins 1 and 3 exhibit similiar calcium affinities, whereas that in synaptotagmin 4 has none; Fig. 8.13B). In wild-type embryos that lack synaptotagmin there is a 1.8-fold dependence of quantal release at the neuromuscular junction on calcium although the affinity of calcium for the release process is high, as the mean quantal content in normal calcium is high (Littleton et al., 1994; Littleton & Bellen, 1995). By the time the third instar stage is reached in the wild-types the mature synaptotagmin is present with its two C2 domains and there is a 3.6-fold dependence of quantal release on calcium and a very high mean quantal release in normal calcium. Null mutations lacking a complete C2 domain are mostly lethal but those that reach an embryonic stage show a 1.8-fold dependence of quantal secretion on calcium and a very low quantal secretion in normal calcium in which most nerve impulses do not secrete any quanta at all. In the case of mutations that only lack the second C2 domain there is still a 1.8-fold dependence of secretion on
DEVELOPMENT OF THE CONCEPT OF A CALCIUM SENSOR IN TRANSMITTER RELEASE AT SYNAPSES
159
Fig. 8.12. Calcium currents and transmitter release at the giant presynaptic terminal in the stellate ganglion of squid. (A) Introduction of a technique to obtain a successful voltage-clamp of the presynaptic terminal in the stellate ganglion of squid. Local application of calcium ions to a voltage-clamped presynaptic terminal is shown. A, Diagram of the arrangement. Presynaptic membrane potential is measured with micro-electordes V1 and V2 and current-passing electrode I, which is connected to V1 in voltage-clamp configuration to control presynaptic membrane potential. The preparation is bathed in a saline containing low (1 mM) calcium and 7 mM manganese. Extracellular pipette was filled with saline containing 11 mM of calcium and no manganese. B, Time course of transmission changes resulting from local calcium application. Presynaptic terminal was bathed in low calcium saline and depolarised with 6 msec-long voltage-clamp steps to 0 mV. Lowering a calcium-containing pipette over the synapse, during the time indicated by the bar (+Ca) caused an abrupt restoration of transmission. This increase in postsynaptic potential amplitude was maintained as long as calcium flow was continued. Gaps between 22 and 35 min represent periods where the presynaptic terminal was depolarised to different potentials to measure a transfer curve (from Fig. 1 in Augustine et al., 1985a). (B) Determination of the adequacy of the voltage-clamp method applied to the presynaptic terminal for ascertaining the calcium currents. Shown is a comparison between Arsenazo III calcium indicator signals and the calcium current measured under a voltage-clamp. A, Simultaneous recordings of Arsenazo III (ArIII) signals and calcium current (lca) resulting from 25-msec-long presynaptic depolarisation (Vpre). Traces represent averages of six sweeps. B, Voltage dependence of calcium current integrals (continuous line) and peak Arsenazo III signal amplitude (open squares) elicited by 6-msec-long depolarisations closely correspond when scaled to the same peak value (from Fig. 12 in Augustine et al., 1985a). (C) Transfer curves relating the magnitudes of the calcium current under a voltage-clamp (measured 3.5 ms after pulse onset) and the postsynaptic current (measured at 5.5 msec after pulse onset). Calcium currents are normalised to peak calcium current, which occcurred at −8 mV. Post-synaptic currents are normalised to peak post-synaptic current which occurred at −3 mV. The same curve is plotted on linear coordinates in the left panel and on double-logarithmic coordinates in the right panel. , Currents from depolarisations to −8 mV and below; , Currents from larger depolarisations. Continuous lines representing third power functions are drawn in the right hand panel; the continuous line in the left-hand panel just joins the points (from Fig. 6 in Augustine et al., 1985b).
calcium and a very low mean quantal content by the third instar stage, but survival to this stage presumably occurs because they release quanta in response to each impulse. Finally, there is a mutation in which only a single amino-acid is changed in the second C2 domain; this does not change the 3.6-fold dependence of secretion on calcium but does greatly lower the
160
HISTORY OF THE SYNAPSE
affinity of the release process for calcium so that the mean quantal content is lowered an order of magnitude below that of the mature wild type. It is interesting in this regard that the first C2 domain is thought to be the only one that can mediate calcium phospholipid binding. This first C2 domain binds calcium in a cooperative manner with a Hill coefficient of 2–3 (Littleton & Bellen, 1995; Fig. 8.13B), and as synaptotagmin is a tetramer the complex has eight potential binding sites for calcium (Ullrich et al., 1994). There is other physiological evidence implicating the C2 domains of synaptotagmin in transmitter release at synapses. If peptides from the C2a domain of synaptotagmin or antibodies to this domain are injected into the presynaptic axon in the stellate ganglion of squid, then the synaptic potential is greatly diminished in size, without any concomitant changes in the calcium influx that accompanies the nerve impulse but with a great increase in docked synaptic vesicles (Bommert et al., 1993; Mikoshiba et al., 1995). On the other hand, injection of inositol polyphosphates into the nerve terminal of the stellate ganglion, so as to bind the C2b domain of synaptotagmin, blocks both the spontaneous and evoked release of transmitter (Llinas et al., 1994) whereas injection of antibodies to the C2b domain only blocks transmitter release if the terminal is repetitively stimulated (Fukuda et al., 1995). Finally, injection of syntaxin into PC 12 cells blocks secretion from these cells (Bennett et al., 1993). An impulse in nerve terminals at the neuromuscular junction or in the hippocampus gives rise to a transient increase in release probability for quanta that decays with two different time constants in the hippocampus of < 5 and 200 msec (Goda & Stevens, 1994). These observations indicate that two different calcium-receptive proteins with different calcium affinities may mediate transmitter release: one with low affinity may mediate the fast synchronous release of transmitter and the other with high affinity may mediate the slow asynchronous transmitter release. Synaptotagmin 1 is the candidate molecule to mediate the fast synchronous release. Evoked transmitter release is greatly decreased in the hippocampus of mice carrying a mutation in the synaptotagmin 1 gene; the fast and synchronous component of evoked transmitter release is greatly decreased but the slow asynchronous evoked release process, as well as spontaneous release triggered by hypotonic solutions and alpha-latrotoxin, is not affected (Geppert et al., 1994). This argues for synaptotagmin 1 being the low affinity calcium receptor for fast transmitter release. It is interesting to note that besides synaptotagmin 1 there are seven other isoforms of synaptotagmin that have distinct distributions over brain synapses (for synaptotagmin 1 and 2 see Perin et al., 1990 and Geppert et al., 1991; for synaptotagmin 3 and 4 see Mizuta et al., 1994 and Hilbush & Morgan, 1994). Synaptotagmin 1 and 2 may be alternative calcium sensors and all synaptotagmins act as receptors for the clathrin protein A2 and therefore are probably involved in endocytosis (Ullrich et al., 1994). Most neurons in the hippocampus express synaptotagmin 3 together with either synaptotagmin 1 or 2. Given that synaptotagmin 1 has been implicated by mutation as mediating the fast and low-affinity calcium release, the results suggest that synaptotagmin 3 may mediate the high-affinity asychronous release; recent studies of calcium binding to synaptotagmin 3 show that this is the case (Ullrich et al., 1994). A developmental analysis of the appearance of the C2a domain of synaptotagmin 1 in a synapse in relation to the appearance of the mature calcium dependence of transmitter release has been carried out (Lin et al., 1996). The C2a domain is detectable with both immunohistochemistry and Western blots within hours of synapses forming. Electrophysiology shows that the mature fourfold dependence of transmitter release on external calcium is already established at this very early time. Are the known affinities of synaptotagmin for calcium consistent with arguments based on theoretical and physiological grounds for the properties of these binding sites (Bennett et al., 1997)? The cytoplasmic domain of synaptotagmin 1 binds phospholipids in a calcium dependent manner, with an EC50 of ca 4–10 µM; calcium is more effective in this regard than strontium or barium which in turn are far more effective than magnesium (Brose et al., 1992; Davletov & Sudhof, 1994; Chapman & Jahn, 1994). The C2a domain of synaptotagmins 1 and 2 possess an EC50 for divalent ion-dependent phospholipid binding of about 3 µM for calcium (and a Hill coefficient of between 2 and 3.1) with a much lower affinity binding for strontium (EC50 of 133 µM); synaptotagmin 3 however has a much higher affinity for strontium of 23 µM with the same affinity for calcium-mediated binding as synaptotagmins 1 and 2 (Li et al., 1995a). These results are consistent with the idea that synaptotagmin is the calcium sensor, as magnesium is a weak activator of quantal release whereas strontium can partly substitute for calcium in the release process (Andreu & Barrett, 1980). However, the relatively high affinity of the C2 domains for calcium-mediated phospholipid binding (in the range of 2.6–10 µM) does not fit with the theoretical estimate that a calcium concentration of ca 100 µM is reached at the calcium channel in the secretosome and that a low affinity calcium binding site is then required to be uniquely activated by this high concentration. This difficulty is possibly resolved by the recent discovery that although the C2a domain of synaptotagmins 1, 2 and 3 binds phospholipids in a calcium dependent manner with an EC50 of ca 2.6– 10 µM this is not the case for the calcium-mediated binding of syntaxin by the C2a domain of synaptotagmins 1 and 2, which have an EC50 in excess of 200 µM; synaptotagmin 3 binds syntaxin in a calcium-dependent manner with a much higher affinity of ca 10 µM (Li et al., 1995b). These results point to the possibility that it is the calciumdependent binding of syntaxin to the C2a domain of synaptotagmin that is the important calcium-dependent event in the secretosome for quantal release. A quantitative analysis of secretion from chromaffin cells indicates that this can be described by a process involving three calcium non-cooperative binding steps which have a Kd of 7–13 µM, that is about the Kd for synaptotagmin 1 calcium-
DEVELOPMENT OF THE CONCEPT OF A CALCIUM SENSOR IN TRANSMITTER RELEASE AT SYNAPSES
161
Fig. 8.13. Calcium binding properties of the vesicle-associated protein synaptotagmin and the secretosome hypothesis. (A) The domain structure of synaptotagmin (from Fig. 4A in Littleton & Bellen, 1995.) (B) A ribbon diagram of the C2A domain of synaptotagmin I. The domain is composed of a compact beta-sandwich formed by two beta sheets with three flexible loops on the top and the bottom; each of the beta sheets is composed of four beta strands that are labelled here with roman numerals. The loops have a variable sequence and may confer binding specificities to the C2 domain. Green shows the position of two calcium ions; orange shows the regions of the C2A domain that exhibit calcium-dependent shift changes; blue indicates the regions of the domain that show no calcium-dependent shift changes. The amino and carboxyl terminus are indicated by N and C, respectively (from Fig. 3 in Sudhof & Rizo, 1996). (C) The relationship between different secretosomes organised in an active zone and the sources of calcium for transmitter release. Shown are secretosomes composed of synaptic vesicles (SV), synaptotagmin, syntaxin 1A and an N-type calcium channel (alpha1 B). Calcium may reach the vesicle-associated proteins through the N-type channels during a nerve impulse; other sources of calcium are through ryanodine (RyR) channels or through inositol triphosphate (IP3R) channels in endoplasmic reticulum closely apposed to the vesicle-associated proteins, b, A dispersed secretosome, not located at an active zone, and fragmented into an L-type calcium channel (alpha1 C) that is not attached to the rest of the secretosome complex, which just consists of a synaptic vesicle (SV) and the vesicle-associated proteins synaptotagmin and syntaxin (a and b from Fig. 11 in Robinson et al., 1996). (D) The distribution of calcium in a three-dimensional projection (after image integration) of a region of the presynaptic terminal before A and during stimulation B, illustrating the steepness of the calcium gradients at transmitter release sites. The image intensities were segmented into an 8-bit range with the maximum intensity corresponding to the video intensity level of 255 (white) and the lowest to a level of 0 (black), with intermediate levels ranging from high (red) through intermediate (yellow and green) to low (blue) probabilities that a particular microdomain would be active (from Fig. 4 in Llinas et al., 1992c).
dependent phospholipid binding mentioned above of 7 µM (Brose et al., 1992; Bazbek et al., 1995). Calcium-dependent phospholipid binding using the C2a domain of synaptotagmins 1, 2 or 3 could mediate this effect. On the other hand it could be restricted to the calcium-dependent syntaxin binding using the C2a domain of synaptotagmin 3. In any case, the secretion
162
HISTORY OF THE SYNAPSE
process involves a relatively high affinity binding site that may well suit the relatively slow secretion process in these cells compared with nerve terminals that require a fast process that terminates quickly. Quantal release from the terminals of goldfish retinal bipolar cells can also be described by a scheme analogous to that used to describe secretion from chromaffin cells, with in this case three calcium cooperative binding steps, only the dissociation constant for one of the calcium-mediated bindings must have a Kd of ca 143 µM rather than the much lower Kd of 7–13 µM used for chromaffin secretion (Heidelberger et al., 1994). The inclusion of a low affinity binding step (143 µM), together with high affinity steps (Kd of 9 µM), is required for the quantitative description of release so that the kinetic scheme provides a fast turn-on, fast turn-off and fast decay of quantal release. Previous theoretical studies have anticipated these experimental results. Thus Heidelberger et al., (1994) showed the necessity of including a rapidly equilibrating low-affinity calcium binding site on the calcium sensor protein (with a Kd of 200 µM) as well as slowly equilibrating high affinity site (with a Kd of 1 µM). Similar ranges of Kd values have been used for multiple binding sites on the calcium sensor when modelling calcium mediated quantal release in the stellate ganglion of the squid (Bertram et al., 1996). 8.8.3 The secretosome hypothesis Two studies in 1992 showed that a close relationship exists between synaptotagmin and syntaxin on the one hand and the presynaptic calcium channels that mediate transmitter release on the other (Fig. 8.13C(a)). Antibodies that are involved in the Lambert-Eaton myasthenic syndrome, an autoimmune disease of neuromuscular transmission, immunoprecipitate Ntype calcium channels together with synaptotagmin from the rat brain (Leveque et al., 1992). Antibodies to syntaxin also immunoprecipitate N-type calcium channels from rat brain (Bennett et al., 1992), and immunoprecipitation experiments of some cell lines show that syntaxin, synaptotagmin and N-type calcium channels form a complex (David et al., 1993; Fig. 8.13C (a)). In addition, there is now known to be a direct interaction between the cytoplasmic domains of both syntaxin and the Ntype channel (Sheng et al., 1994), with syntaxin possessing the ability to gate the calcium channels (Bezprozvanny et al., 1995). All of these observations indicate that there is a tight structural association between synaptotagmin, syntaxin and the Ntype calcium channel. This combination has been termed the ‘synaptosecretosome’ by O’Connor et al. (1993) and will be abbreviated here to ‘secretosome’. As discussed above, other calcium channel types than N-type, such as the P-/Q-type, are involved in transmitter release at different nerve terminal types. The concept of the secretosome is likely to be applicable to these nerve terminals also as P-/Q-type calcium channels in rat cerebellar synaptosomes are immunoprecipitated in a complex with synaptotagmin and syntaxin by antibodies against syntaxin (Martin-Moutot et al., 1996). As none of these experiments have implicated the L-type calcium channel in a complex with synaptotagmin and syntaxin, a secretosome that uses this channel does not seem to exist (Fig. 8.13C(b)). 8.9 Conclusion This chapter began with the discovery in the 1940s that calcium is required for transmitter release, which led on to the discovery in the 1960s that this calcium acts on the inside of the nerve terminal, and in the late 1970s to the discovery of the different types of calcium channel that mediate this influx of calcium ions at nerve terminals. Next the discovery of synaptotagmin in the 1980s was described and the facts presented which suggest that synaptotagmin is the principal calcium sensor acted upon by the influx of calcium ions (Sudhof, 1995). Finally, these elements of the story came together in the 1990s with the discovery that a complex exists which consists of calcium channels and synaptic vesicles anchored to the presynaptic membrane through synaptotagmin to form the secretosome. Research on the mechanism of secretion now requires techniques to be refined by which we may visualise in real time the events that occur at a single secretosome on arrival of the nerve impulse. One of these, the measurement of calcium influx through single calcium channels in the secretosomes, is at its furthest stages of development in the hands of Llinas and his colleagues. Funnel web spider venom blocks transmission presynaptically in the squid stellate ganglion by blocking the inward influx of calcium ions (Llinas et al., 1989). In an ingenious use of low-affinity calcium indicators, Llinas et al. (1992a, c) have gone on to determine the calcium flux through these channels during transmission. Computer modelling of calcium near the mouth of a calcium channel in the nerve terminal has predicted, as noted above, that the transient calcium concentration might reach several hundred micromolars near the channel, with the concentration falling off steeply near the cytoplasmic mouth of the channel (see for example Llinas et al., 1981; Chad & Eckert, 1986; Simon et al., 1984; Simon & Llinas, 1985; Zucker & Fogleson, 1986). The degree of resolution that Llinas and his colleagues could obtain with their low affinity calcium indicator, n-aequorin-J, was down to that of a single active zone. This enabled them to observe the elevated calcium level at these active zones (with dimensions of ca 0.6 µm×0.6 µm) during high frequency trains of impulses (Fig. 8.13D). Although this work clearly does not give the calcium transient localised to a single calcium channel within an active
DEVELOPMENT OF THE CONCEPT OF A CALCIUM SENSOR IN TRANSMITTER RELEASE AT SYNAPSES
163
zone, the spatial resolution of which is at present beyond existing technology, it did give estimates for the peak calcium levels of ca 200 µM, approximately that expected on theoretical grounds as mentioned above. The future holds out the possibility that calcium influx through single channels of the secretosomes for interaction with the calcium sensor will soon be within detection.
9 The Discovery of Quantal Secretion and the Statistics of Transmitter Release at Synapses
9.1 Introduction 9.1.1 Discovery of quantal transmitter release In 1950 Fatt and Katz reported the existence of ‘biological noise’ when recording the endplate potential (epps) at the amphibian neuromuscular junction with an intracellular microelectrode. Subsequently they gave a full account of these ‘miniature potentials’, showing that they had many of the physiological characteristics of the epp, and that the amplitudes of these miniature end-plate potentials (mepps) followed a Gaussian distribution (Fig. 1A; Fatt and Katz, 1952a). They also noted that ‘lack of calcium apparently reduces the epp in definite “quanta”, as though it blocks individual nerve terminals, or active patches within them, in an all-or-none manner’ and that under these conditions ‘the normal epp can be seen…, to break down into individual miniature units’. Thus the epp was seen as composed of mepps. This gave rise to the ‘quantal hypothesis’ namely that the mepp represented a quantum of transmitter release and the epp multiples of this quantum. Furthermore, ‘suppose…that there are within the terminal area some 100 discrete “patches” concerned with the release of ACh. If these terminal structures have a special tendency to spontaneous excitation, then our observations would be easily understood”. So the concept of discrete zones for the secretion of transmitter had a physiological basis. These ideas were to be given substantial impetus with introduction of the electron microscope for the examination of end-plate structure. This revealed synaptic vesicles in the nerve terminal, giving rise to the ‘vesicle hypothesis’ that the miniature units or quanta are due to the prepackaging of transmitter in vesicles (Birks et al., 1960). Furthermore, the ultrastructural observations of these authors that ‘one often sees the vesicles concentrated in certain well-defined areas focussed on a dense zone of the axon membrane directly opposite a post-synaptic fold’ which they referred to as ‘special zones of the axon membrane’ introduced the idea that such zones were the ‘discrete zones’ from which quanta are released. This gave rise to the ‘active zone hypothesis’, namely that these zones constitute the sites of quantal transmitter release. 9.1.2. Statistical methods introduced to describe quantal release The quantal hypothesis, that transmitter is released in quanta, allowed a statistical description to be applied for the first time to the process of synaptic transmission. In 1954, del Castillo and Katz gave an explicit statement of the relationship between the epp and the mepp, namely that ‘it has been suggested that the end-plate potential (epp) at a single nerve-muscle junction is built up statistically of small all-or-none units which are identical in size with the spontaneous “miniature epps”. The latter, therefore could be regarded as the least unit, or the “quantum” of end-plate response’. This became known as the ‘quantal hypothesis’. In which ‘transmission at a nerve muscle junction takes place in all-or-none “quanta” whose sizes are indicated by the spontaneously occurring miniature discharges’ (Fig. 1B; del Castillo and Katz, 1954). Curiously enough, the possibility was not entertained that the quantum might arise from the saturation of a ‘unit of receptors’ by the release. They go on to say that ‘statistical analysis indicates that the epp is built up of small all-or-none quanta which are identical in size and shape with the spontaneously occurring miniature potentials’. This statistical analysis is considered in the next section.
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
165
Fig. 9.1 Quantal transmission: a binomial or Poisson variate? A. Distribution of amplitudes of spontaneous miniature end-plate potentials at the amphibian end-plate. Muscle treated with prostigmine. The amplitude distribution of some 800 miniature potentials are shown (from Fatt and Katz, 1952a). B. Histogram showing distribution of amplitudes of spontaneous miniature potentials and end-plate responses at a calcium deficient junction. In the lower part, the continuous curve has been calculated on the hypothesis that the responses are built up statistically of units whose mean size and amplitude distribution are identical with those of the spontaneous potentials. Expected number of failures shown by arrows. Abscissae: scale units is the mean amplitude of spontaneous potentials (0.875 mV; from del Castillo and Katz, 1954a). C. Relation between the coefficient of variation and mean amplitude of end-plate potentials (epps) recorded at the amphibian endplate in low calcium or high magnesium solution. Logarithmic scales. Abscissa: mean epp, divided by mean spontaneous potential (ie. nominal value of m). Ordinate: standard deviation of epp, divided by mean (ie. ‘coefficient of variation’ of epp). Epp amplitudes had been grouped for this purpose in ‘unit classes’. Bars have been placed at ±2 S.E. of the ‘coefficient of variation’. Full line shows theoretical relation for Poisson-distributions (from del Castillo and Katz, 1954). D. Histogram from an experiment with large epps at the amphibian end-plate. Nominal value of m is 32. Dotted curve: expected distribution of epp’s (modified Gaussian curve allowing for scattered unit size: mean=20.4 mV), =3.7 mV. Note large discrepancy between observed and expected distribution (from del Castillo and Katz, 1954).
9.2 Evoked quantal release as a binomial or Poisson variate 9.2.1. Binomial and Poisson theory 9.2.1.1 Binomial theory del Castillo and Katz (1954) suppose there are n units at each nerve-muscle junction with an average probability of responding of p. We define a random variable Xi for i=1, 2,…, n to be Xi=0 if the ith unit does not respond (a ‘failure’) and Xi=1 if the ith unit does respond (a ‘success’). Then Pr (Xi=0)=1−p=q and Pr (Xi=1)=p for i=1, 2,…, n. We assume the n units respond independently of one another (so that the Xis are mutually stochastically independent). Then the sum is the number of units responding to a stimulus. Then Y is a random variable with the binomial distribution. Let Y={y|y= 0, 1, 2,…, n} then Y=y if and only if exactly y of the variables X1, X2,…, Xn have the value 1 and the remaining n−y variables have the value 0. There are 1 ways in which y ones can be assigned to the variables X1,…, Xn and the probabilities of each of these ways is .
1
The choose operator
is read ‘n choose y’ and defined by where (n factorial).
166
HISTORY OF THE SYNAPSE
Therefore for y=0, 1, 2,…, n and this is the probability density function (pdf) of the binomial distribution. Using the moment generating function (mgf) to determine the mean µ y and variance2 we have Then
Hence
Therefore (1) In summary, the binomial distribution for the random variable Y representing the number of units responding to a stimulus is given by: for with mean and variance given by Eq. (1). An estimate of a parameter may be obtained by calculating the corresponding statistic from the sample. Here we rearrange Eq. (1) to estimate the population parameters indicated by the notation for the sample mean and sample variance . (2)
9.2.1.2 Binomial to Poisson Theory The pdf of the Poisson distribution is given in the Appendix Section A7.2 as The mgf is So the mean µ=M (0)=m and the variance
=. 9.2.2 The compound Poisson 9.2.2.1 Theory
Spontaneous end-plate potentials are assumed to occur when one quantum of transmitter is released. They are normally distributed random variables, Zi. Assume each has mean µ 1 and SD .3 Then when n quanta are released we have a random variable Z which is the sum of the n normally distributed random variables (all with mean µ 1 and SD 1). Let . Then the mean and variance of Z are given by 4 (3)
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
167
Fig. 9.2. Quantal release as a binomial variate at crayfish neuromuscular junctions. A. Extracellular records from a single junctional area of a crayfish motor-nerve terminal during nerve stimulation. A. Second potential is an excitatory junction potential (ejp) set up by the nerve impulse; first is a spontaneous miniature potential. B. ejp set up by motor nerve impulse as in A. Note similar size and time course of ejps and of the spontaneous potential (from Dudel and Kuffler, 1961). B. Histogram of size distribution of extracellularly recorded ejps from single junctional area of a crayfish motor-nerve terminal. Ordinate scaled to 1000 counts. There are 270 zero potentials, i.e. failures of transmission. Broken line is drawn according to Poisson’s theorem for a quantum size of E1=70 µV with a standard deviation =12 µV. The quantum content per stimulus is m=1.3. Small arrows indicate the multiples of the unit size. Big arrow gives the average size of the spontaneous miniature potentials (from Dudel and Kuffler, 1961). C. Extracellularly recorded junction potential (erjp) amplitude densities ((a), (b) and (c)) from a superficial distal fibre at the crayfish neuromuscular junction. Ordinate: number of occurrences; abscissa; erjp amplitude (µV). Thirty-seven spontaneous potentials combined to calculated the unit size, Es. Bar histograms, observed data; continuous line, theoretical curve, m calculated from . Arrows at top of graph indicate integral multiples of Es (from Bittner and Harrison, 1970). D. Histograms of quantal distributions at two different frequencies of stimulation ((a) and (b)) of crayfish motor nerves. nx=the number of ejps containing x quanta. Black bar, observed distribution. Stippled, theoretical Poisson distribution with same m. White, theoretical binomial distribution with same m and p=0.044 (a) and 0.49 (b) (from Johnson and Wernig, 1971).
If we assume the quanta are released according to a Poisson process with parameter m; that the effect of the individual quanta is normally distributed with mean µ 1 and SD 1; and that the effects of simultaneously released quanta sum linearly; then we obtain a compound Poisson distribution. Then for each y, the probability that an epp consisting of y quanta has amplitude lying between a and b is given by
and the probability that a stimulus causes a response with amplitude between a and b is
This distribution has three parameters which need to be determined, µ 1,
1
and m (= np).
9.2.2.2 Experiments on amphibian neuromuscular junctions del Castillo and Katz (1954a) used this equation for the compound Poisson to fit their data on the distribution of epps at a Cadeficient junction with mean quantal content m (Fig. 9.1B, lower histogram) for which they had the mean and standard deviation of the spontaneous miniature potential quanta (Fig. 9.1B, upper histogram). They noted, however, that as m
2
See Appendix, Section A7.1 for the derivation. We write f(zi) is n(µ 1, 1) to mean Zi has a Gaussian or normal distribution. The pdf is given by for —
168
HISTORY OF THE SYNAPSE
increased in size the coefficient of variation departed from that for a Poisson distribution (see Fig. 9.1C), so that the ‘observed fluctuation of epp amplitudes covered a smaller range than expected’ (Fig. 9.1D), del Castillo and Katz (1954a) commented that this discrepancy between the observations and that predicted by the compound Poisson equation could be caused by the fact that ‘the superposition theorem may lead to serious error if the epp response exceeds a small fraction (5%) of the resting potential’ a problem now referred to as ‘non-linear summation’. On the other hand they commented that another ‘factor which may be involved is that different members of the populations may not have the same chances of success, and that for large values of m some individual units have a high probability and respond almost every time, while others have a low probability and contribute to the epp only occasionally’. In this latter case, deviations of the kind shown in Figs. 9.1C & 9.1D could not be accounted for by the smaller coefficient of variation of the binomial compared with that of the Poisson, but could by using the coefficient of variation of the compound non-uniform binomial. In this case each of the ‘population of n units’ capable of responding to a nerve impulse may possess a different p. Although del Castillo and Katz (1954a) did not carry out further tests as to whether the deviations from the Poisson predictions given in Figs. 9.1C and 9. 1D were due to non-linear summation or to the existence of release units with a non-uniform probability, evidence was later presented that the former explanation was likely to be correct. However, the analysis of del Castillo and Katz (1954a) introduced the idea of large numbers of release sites at the amphibian motor-nerve terminal, so that Poisson statistics might be expected to apply. Furthermore, the possibility was raised that in some conditions the number of release sites that can respond to an impulse might be low, so that binomial statistics might apply. Finally, that in some cases these release sites might not possess a uniform probability for release. 9.2.3 The binomial 9.2.3.1 Experiments on crayfish and crab neuromuscular junctions In 1961 Dudel and Kuffler showed that recordings of the spontaneous excitatory junction potentials (sejps) from extracellular sites at the crayfish neuromuscular junction resembled in amplitude and time course the size of many of the evoked excitatory junction potentials (ejps) found at the same site (Fig. 9.2A). The histogram of ejp amplitudes was then fitted by a Poisson distribution, with guessed values of the mean and standard deviation of the sejps (Fig. 9.2B). Although the value of these parameters was comparable to that of 10 observed sejps, the Poisson distribution fit to the ejp amplitude histogram was not very impressive (Fig. 9.2B). The number of release sites recorded from by the extracellular electrode was not large (n=10), so that binomial rather than Poisson statistics might have been expected to give a better description for p>0.2. Atwood and Johnston (1968) later attempted to fit a Poisson statistic to their extracellular recordings of sejps and ejps at a crab neuromuscular junction. They also obtained poor fits, for example their Table 1 shows the observed failures (5) was quite different to that predicted (13) according to a Poisson analysis and so they concluded that a compound Poisson hypothesis did not hold. Bittner and Harrison (1970) returned to the crayfish neuromuscular junction, and noted that the amplitude-frequency histograms of extracellularly recorded ejps was reasonably well predicted by the compound Poisson equation at certain frequencies of stimulation but not at others (Fig. 9.2C). In all, 28 out of 34 amplitude-frequency histograms were not well fitted by the compound Poisson equation. They noted that this departure from Poisson statistics might occur because ‘the ending may contain a limited number of transmitter quanta, each having a very high release probability, the number of quanta available increasing at higher stimulus frequencies’. Johnson and Wernig (1971) returned to this problem concerning the histograms of ejps recorded with extracellular electrodes at crayfish and crab neuromuscular junctions. It is likely that only a relatively small number of release sites were recorded from at these invertebrate junctions (10) compared with those at the amphibian neuromuscular junction (200 or so). Furthermore, in the amphibian junction high magnesium solutions were used by del Castillo and Katz (1954b) so that the probability of release was low in the amphibian compared with that in the invertebrates. The invertebrate junction might then be expected to follow a compound binomial while the vertebrate junction a compound Poisson. Johnson and Wernig (1971) tested if extracellular recordings of the ejp from the crayfish neuromuscular junction followed the compound binomial better than the compound Poisson. They found this to be the case (Fig. 9.2D), obtaining values for n and p of about 3 and 0.4, respectively, during recordings from a motor-terminal branch stimulated at 1Hz. This raised for the first time the question of what n and p might refer to; their suggestion was that they ‘may represent the release parameters for such a branch’. Given that Dudel and Kuffler (1961) had obtained an estimate of sixty synaptic contacts per muscle cell it was difficult to relate the parameter n to the number of synaptic contacts, so it was suggested that a subset of these might be an appropriate choice. It was uncertain then what n represents. Another difficulty related to the arguments concerning whether evoked quantal release was distributed as a Poisson or binomial variable in the absence of any error analysis. Histograms of the kind presented in
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
169
Fig. 9.2D comparing the predictions of these statistics with experimental data were not persuasive in the absence of an error analysis. Attempts to see if the quantal content of the intracellulary recorded epps could be described by binomial statistics, as originally suggested by del Castillo and Katz (1954a), were made by Christensen and Martin (1970) after correcting for nonlinear summation, but their observations were not subjected to an error analysis. 9.2.3.2 Standard errors of the estimates of the parameters in the Poisson and the Binomial Zucker (1973) attempted to provide an error analysis for the application of binomial statistics to quantal release at the crayfish motor-nerve terminal, as shown in the following section. 9.2.4 Determination of parameters 9.2.4.1 Zucker’s determination of parameters Zucker (1973) wished to estimate the parameters m, p and n which are required if we assume x, the number of quanta released has either a Poisson or a binomial distribution. As mentioned previously, an estimate of a parameter can be obtained by calculating the corresponding statistic from the sample. Zucker repeatedly measures the response to nerve stimulation in his experiments. The best estimate of the average number of quanta released per simulation is given by the sample mean, where N is the total number of impulses and nx is the number of impulses resulting in x quanta released. (Zucker assumes five as an upper limit to the number of quanta released on any impulse, ) With defined as the average number of quanta released per impulse, we calculate the standard error by Theorem 1,
Zeucker’s Eq. (2). We now use our estimate of the mean, , to estimate p. If x is a binomial random variable then by Eq. (1). So or We use this equation to estimate p. The standard error of is derived in the Appendix in Section A7.3. The equations may be summarised by:
the correct form of Zucker’s (1973) result (7)
as in Robinson’s (1976) Eq. (2)
which is Whitakers result, Eq. (vi). Similarly, for we derive Eq. (37) in the Appendix, Section A7.3 as
which is Robinson (1976) Eq. (2), or Whitaker (1914) Eq. (v) with p and q interchanged, rather than the incorrect result of Zucker (1973). Although Zucker’s (1973) estimates of the standard errors of n and p are not correct his general approach was, namely that without these estimates, claims as to the relative changes in the n and p following different experimental procedures cannot be accepted without an error analysis. His comment that ‘they estimate what an extracellular electrode would usually record from events occurring at two to four release sites. The similarity of this estimate to n makes the identification of n with the number of discrete release sites at a synaptic spot an attractive possibility,’ continues the tradition of attempting to give n a physical representation.
170
HISTORY OF THE SYNAPSE
9.2.5 The compound binomial 9.2.5.1 Theory 9.2.5.1.1 Using a normal variate. We assume that at each of n sites, the probability of release of a quantum of transmitter is p. Further, we assume the quantal size of the (mepps) or spontaneous epps from each of the n sites are identical, independent random variables, Zi, normally distributed with mean µ 1 and standard deviation 1. Thus Z=Z1+Z2+···+ Zy has and since each zi is n(µ 1, ). Let X be a random variable of the amplitude of the evoked synaptic potentials. Then X has a compound binomial distribution and the pdf or the frequency function of X is, for (4) and it is clear that (Robinson, 1976, Eq. (3)). Our sample consists of M observations from the distribution of mepps (or spontaneous potentials) with mean and variance and N observations from the distribution of evoked potentials which we assume occur according to a binomial distribution with mean and variance . Our estimators for the parameters of each Zi are E[Zi]= µ 1 and .E[X] and E[ ] are derived in the Appendix, Section A8.1 using the mgf method, giving: (5) (6) which is Robinson’s (1976) Eq. (4). The moment estimator of p is given by rearranging Eq. (6)
(7) and for any n (8) which is Robinson’s (1976) Eq. (5), and Bennett and Florin’s (1974) Eq. (2). The standard errors of these estimators are derived in the Appendix, Section A8.4. 9.2.5.1.2 With a gamma variate. Robinson (1976) noted that in many cases the histogram of spontaneous potentials was better fitted by a gamma distribution than a normal distribution (Fig. 9.3C). He then introduced the compound binomial hypothesis using a gamma variate as the unit size rather than a normal variate as previously, in the following way. The gamma distribution with parameters ( , k) is given in Appendix Section A8.2 by: which is Robinson’s (1976) Eq. (6). The gamma distribution has mean k/ and variance k/ 2. We use a compound distribution. The pdf of the compound gamma, for x>0, is (9) and P(X=0)=qn (Robinson, 1976, Eq. (7)). We have a normal distribution of mepps with mean and variance In Appendix, Section A8.3 the mgf is used, for to derive the mean and standard deviation of the compound gamma distribution. The mean is given by the first moment about zero as and the variance is which is identical to the compound binomial case. The standard errors of these estimators vary slightly from the case of the normal variate, and again are set out in the Appendix, Section A8.4.
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
171
Fig. 9.3. Quantal release as a binomial variate at amphibian neuromuscular junctions. A. Amplitude-frequency histograms of epps (open histogram) and spontaneous miniature end plate potentials (mepps; filled histogram) at newly formed junctions with different values of p. (a) Histogram for a junction in which p is small; the binomial prediction (continuous line) gives values of p and n of 0.22 and 11.6, respectively; the Poisson prediction (interrupted line) fits the distribution nearly as well as the binomial at these low values of p; arrows give the predicted number of failures, (b) Histogram for a junction in which p is large the binomial prediction (continuous line) gives values of p and n of 1.0 and 11.9 respectively; the Poisson prediction (interrupted line) does not fit the distribution nearly as well as the binomial at these high values of p. Number of epps 90 for each histogram (from Bennett and Florin, 1974). B. Comparison of observed (dotted line) and expected quantal distributions at the amphibian neuromuscular junction. The theoretical distributions are shown for the binomial (B) and the Poisson (P). nx=the number of epps containing x quanta (from Wernig, 1975) C. Frequencies of 159 observed spontaneous potentials recorded in a sympathetic neuron (vertical bars) and fitted with normal and gamma frequency curves (from Robinson, 1976).
9.2.5.2 Experiments on newly formed amphibian neuromuscular junctions Bennett and Florin (1974) investigated the application of binomial statistics to the evoked quantal release at early reinnervating amphibian motor-nerve terminals, when the number of active zones is relatively small compared with the number in mature terminals and the quantal release is subthreshold for initiation of the muscle action potentials in normal calcium and magnesium solutions. Even qualitatively, Poisson statistics could not describe evoked release, whereas binomial statistics could (Fig. 9.3A(b)). Furthermore, facilitation or depression was accompanied by significant changes in p but not in n, which varied in size from 6 to 31 in different muscles, a variation that was attributed without evidence to the number of release sites at these relatively small motor-nerve terminals. Values of p and n were determined in this analysis using a compound binomial hypothesis as the quantal size given by the mepp amplitude-frequency histograms is distributed as a Gaussian or gamma variate. This is in contrast to work carried out with extracellular electrodes on the crayfish motor-nerve terminals described
172
HISTORY OF THE SYNAPSE
above, for which the quanta could be counted so that only the binomial equation need be used. The error analysis for the values of m, n and p obtained from the compound binomial hypothesis were devised by Robinson (1976), and are given in the Appendix (Section A8.4). Use of these expressions for the standard errors gave reliable comparisons for significance in the changes in p accompanying facilitation and depression given in Bennett and Florin (1974), suggesting that these changes in efficacy were due to changes in the probability of secretion rather than is the number of release sites participating in secretion if n is taken, without evidence, as a measure of the number of these. Subsequently Wernig (1975) attempted to show that evoked quantal release at mature amphibian neuromuscular junction in high magnesium concentrations could be best described by binomial rather than Poisson statistics (Fig. 9.3B). However, the error analysis of n and p in this work followed that of Zucker (1973) and so is not correct. 9.2.6 The non-uniform binomial 9.2.6.1 Theory In 1977 Bennett and Fisher consider the effects of variations in the parameters n and p, on the estimates and their interpretation. Even with n and p stationary, it is difficult to distinguish a binomial distribution from a Poisson distribution. Brown et al. (1976) used large values of n and small values of p in their simulations but Bennett and Fisher (1977) and Bennett et al. (1977) did not observe large values of n with small values of p in experimental data. When n is small (n < 10) and p is not small (p > 0.3) different results are obtained. Bennett and Fisher (1977) suggest good estimates are obtained by taking (10) where ri is the number of releases on the ith trial. The choice of estimates depends on the probability that all of the ri are less than n. That is, on the values of (1−pn)N, the probability that in none of the N trials do all n release sites actually release a quantum. If (1−pn)N<0.5 choose the first estimate, otherwise the second. (Another simpler criterion is to choose the former if n<4 and the latter if 4
is fixed and p takes n different values p1,…, pn for the n quanta; Then E[m]=n and with = Then Eq. (10) will give good estimates provided all the p, are moderately large, the number of trials is large and n is not too large. If some pi are very small, it may be reasonable to assume the corresponding quanta are not available quanta. Thus these methods estimate the number of ‘effective’ available quanta. 3.2. P is independently and identically distributed for each quantum and with the same distribution in every trial. If P has mean p, than the distribution of the number of releases will be binomial (n, p) and the methods will be appropriate with
Bennett and Fisher (1977) conclude the erroneous inferences can be avoided by using methods only for appropriate values of n and p and provided standard errors are always given. Bennett and Lavidis (1979) considered Bennett and Fisher (1977) Case 3.1 in which quantal secretion is a binomial variable and the probability of secretion varies between the sites. 9.2.6.1.1 Non-uniform p. We now consider the effect of non-uniformity of probability of release, p on the distribution. Let p, be the probability of release for site i, Then, if n=2 the pdf is given by:
Similarly, for n=3, the pdf is
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
173
For any n, the pdf is calculated similarly. 9.2.6.1.2 Estimation of non-uniform, stationary p. Consider the non-uniformity of p to be given by the random variable P. Assume P varies among release sites (but remains constant in time). Let be the mean or expected value of P, and = be the variance. We can find the first and second moments about zero for the compound binomial using the mgf
Using logarithmic differentiation5:
So and
which is Kendall et al. (1987) relation (5.26). This matches Bennett and Fisher (1977, p. 696) and Bennett and Lavidis (1979, Eq. (1)) with and where
and S2 are the mean and variance of the observed quantal secretion. Then our estimate of p is given by Eq.(2)by
(12)
Bennett and Lavidis (1979), Eq. (4) (13) where is the coefficient of variation. Thus Eq. (2) overestimates (by a factor of 1+ ) the mean release probability and therefore underestimates the number of sites. 9.6.1.3 Estimation of non-stationary p. The simplest model for another generalisation of the binomial and the Poisson is given in Kendall et al. (1987, pp. 164– 165) in which each of n sets is drawn from one of k different populations with proportions p1, p2,…, pk and each population has the same chance of selection. This corresponds to n samples taken over time with probability of responding varying over time. In this case, the array of relative frequencies is and
174
HISTORY OF THE SYNAPSE
and the moments are given by Kendall et al. (1987)
Eq. (5.29) as where is the mean of the p’s and
is the variance. Our estimator for p, from Eq. (2) is
with Thus the temporal variation causes to be underestimated and to be overestimated in contrast to the case of nonuniform p described above. Moreover such estimates are much more sensitive to temporal rather than spatial variation (Brown et al., 1976, p. 2915). 9.2.6.1.4 p follows a beta distribution. Bennett and Lavidis (1979) assume the random variable P has a beta pdf, which is of the form: (14) with mean (15) and variance Since the form of the density distribution of the random variable P is known, estimates of the mean and variance of the density distribution can be made by expectation Eq. (11) and density distribution parameters Eq. (15). If p is non-uniform but stationary, by Eq. (12) we have:
(16)
which are Bennett and Lavidis (1979), Eqs. (7) and (8). Thus and can be determined as functions of
which substituted into Eq. (16) gives
Therefore
5Take
the logarithm of both sides and then differentiate each side.
and
by
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
175
and Bennett and Lavidis (1979), Eqs. (9) and (10). If n then and so that and Also would be small so that . Thus For small 6 and the beta pdf can be approximated by Bennett and Lavidis (1979), Eq. (13)
which is the correct form of Bennett and Lavidis (1979) Eq. (14). 9.2.6.2 Experiments on the crayfish and amphibian neuromuscular junctions as well as the synapses formed by group 1 muscle afferents on dorsal spinocerebellar tract neurons The pdf approach was used to estimate probability at the crayfish neuromuscular junction by Hatt and Smith (1976, see their Method B). These authors also estimated probabilities by assuming only two possible values of the probability of release, say a sites with probability of release pa and b sites with probability of release pb. Then a+b=n and the observed mean is and the observed variance is = apa (1−pa)+bpb (1−pb). This is their Method A for which n=3 gave good fits to the data, according to a 2 criteria, as shown in the Table of Fig. 9.4A. The first evidence for a non-uniformity in the probability of quantal secretion at different release sites of a vertebrate synapse was made for the amphibian motor-nerve terminal (Bennett and Lavidis, 1979). Extracellular recording of quantal release at different sites along terminal branches gave quite different probabilities for the release of a quantum, with frequency distributions of these probabilities shown in Fig. 9.4B. These authors developed the complete theory for the compound nonuniform binomial, which included a flexible distribution for the variate p, namely a beta distribution. Using the values for p and n in a particular calcium concentration (see for example Fig. 9.4B) it is then possible to obtain values for and in the equations in Section 9.2.6.1 above and so determine the frequency distribution of j over the N sites of a terminal, as shown in Fig. 9.4C. This can be done for different calcium concentrations, and Fig. 9.4C gives the pj frequency for a terminal for a range of calcium concentrations from 0.3 mM to 1.0 mM. Walmsley et al. (1988) applied the non-uniform binomial model, using the Hatt and Smith (1976) approach to quantal release at a mammalian central excitatory synapse, namely that formed by single group 1 muscle afferents on dorsal spinocerebellar tract neurones. They obtained much better fits with the non-uniform binomial to their observed distributions of the probabilities of the amplitudes of the multiple quantal components of the excitatory postsynaptic potentials (epsps) they observed, compared with the uniform binomial (Fig. 9.4D). The frequency-histogram of the non-uniform binomial probabilities (the pjs) obtained in this way for the release sites of 12 single group 1 muscle afferents, is shown Fig. 9.4E. This histogram was obtained from dorsal spinocerebellar tract neurones in the intact cat spinal cord, so that the calcium concentration was greater than 1 mM. The histogram may be compared with the density distributions of the pjs in the 1 mM calcium obtained for the amphibian motor-nerve terminal shown in Fig. 9.4C. Each terminal of a spinocerebellar tract neurone possess a large number of relatively high probability release sites compared with the amphibian motor-nerve terminal, even allowing for the higher calcium concentration in which the former were determined. The non-uniform binomial provided the means of analysing quantal release at a terminal using two different approaches. One was to solve for a set of equations giving the pdf and fitting these to the data as introduced by Hatt and Smith (1976) for determining the pjs. The other was to use the compound non-uniform binomial with the very general beta distribution of the pjs, to determine the frequency histogram from the values of n and p in equation as introduced by Bennett and Lavidis (1979).
6
See for example p. 173 of Lindgren (1993) since Therefore and
176
HISTORY OF THE SYNAPSE
Fig. 9.4. Statistical descriptions of non-uniform probabilities over the release sites of a terminal. A. Probability distributions and parameter estimates for quantal release at the crayfish neuromuscular junction assuming uniform and non-uniform p. The non-uniform estimates for each experiment were first calculated assuming two populations of p; this is denoted method A (for description of this see text). The second estimates were chosen by trial and error until the predicted distribution agreed with the data sufficiently well to give X2<4.0 (P>0.5); for site no. V, a successful fit could not be achieved. This is denoted method B (for description of this see text). The average (p) and the variance (var p) of the chosen values of p are also given. Estimation methods marked * fit the data according to a test of X2 (Johnson & Wernig, 1971) with P>0.1 (from Hatt and Smith, 1976) B. The frequency-distribution of the mean quantal content recorded at different extracellular sites (me) at an amphibian neuromuscular junction for which me=pe; that is, the binomial parameter ne was 1.0. Each estimate was made during 100–200 impulses in the different external calcium concentration indicated. The relative frequencies for pe<0.10 were not determined because of sampling problems; the hatched bar indicates pe <0.10. The relative frequency of the large me values increases with an increase in external calcium concentration. It is argued in the Theory of this paper (see Bennett and Lavidis, 1979) that if pe=me, then the recording is from a single release site (j) with probability pj, so pj=pe=me. The curves of fit to the histograms are drawn according to the probability density for the pj, namely f(pj), (see Theory in this paper), in which the tails of the f(pj) distribution have been fitted to the appropriate frequency distributions by setting the value of f(pj) at pj=0.15 equal to that for the number of observations in the 0.1 to 0.2 bin; the remaining bin heights are then reasonably predicted by the f(pj) curves for each external calcium concentration (from Bennett and Lavidis, 1979). C. The density-distributions f(pj) of the probability of secretion of a quantum at release sites (pj) at an amphibian motornerve terminal with n release sites, based on the assumption that pj is a beta variable. The curves are drawn according to Eq. 13 in the Theory of the paper by Bennett and Lavidis (1979) for the different extracellular calcium concentrations, [Ca]0, in millimolar indicated; and had values determined by those of m and p (see Eqs. 11 and 12 in Theory) in different [Ca]0, which are given in Fig. 9.5A in Bennett and Lavidis (1979) and in Figs. 3 and 4 of Bennett et al. (1977), namely for N, , m, and pj respectively; [Ca]0 0.3 mM, 2.1, 2.6, 0.8 and 0.28; [Ca]0 0.4 mM 3.9, 1.5, 2.6, and 0.40; [Ca]0 0.6 mM, 8,1.1, 7.3, and 0.50; [Ca]0 0.7 mM, 8, 0.68, 11 and 0.60; [Ca]0 1.0 mM, 11, 0.44, 22, and 0.70. Note that the fraction of release sites with pj=1.0 increases with an increase in [Ca]0. As each curve is a density-distribution the area under each of the curves must be the same; this does not appear to be the case in this figure as the points of intersection of the curves with the ordinate are not shown on the scale used (from Bennett and Lavidis, 1979). D. Fits obtained for different values of n (the number of release sites at terminals formed by single mammalian group 1 muscle afferents to dorsal spinocerebellar tract neurons) using binomial statistics for uniform pj ((a) and (b)), or binomial statistics with non-uniform pj (otherwise called the compound binomial model; (c) and (d)). Solid curves; measured excitatory postsynaptic potential (EPSP)+ noise distribution. Dotted curves are the proposed fits. Solid bars: noisefree distributions in each case. Probability calibrations refer to these bars (from Walmsley et al., 1988). E. Histogram of underlying quantal probabilities at terminals formed by single mammalian group 1 muscle afferents to dorsal spinocerebellar tract neurons (p ) obtained for 12
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
177
Fig. 9.5. Stochastic descriptions of the kinetics of quantal release at the amphibian neuromuscular junction. A. The average quantal release rate at a small number of release site at an amphibian motor-nerve terminal as a function of time after a brief terminal depolarization. This curve, derived from Fig. 9.4 of Katz and Miledi (1965), describes the estimated release rate for a neuromuscular junction at which the calcium and magnesium concentrations had been adjusted to give an average unit content of about 0.65. The equation giving the theoretical prediction is given in the text (see section 3.1.1; from Stevens, 1968). B. Shaded histogram (a) indicates the probability of a first quantal release as a function of time after motor-nerve stimulation for a series of 482 first quantal latencies recorded in 712 total trials (230 response failures) at 1°C. Continuous lines plot, (t) calculated for these data (see Section 3.1.1 for the equation). Ordinate is release probability per 0.1 msec. Stimulation rate 0.25 per s., (b), semilogarithmic plot of (t) from (a). Line was fitted by eye to points after peak release, giving an estimated time constant of 3.5 msec for the early falling phase (from Barrett and Stevens, 1972). C.Stochastic analysis of the release of quanta by a nerve impulse recorded with an extracellular electrode at an amphibian motor-nerve terminal at high temperature (18°C). The abscissa is the time after passage of the nerve impulse, (a) Is the latency-frequency distribution together with the predicted curve, (b) Gives the value of p (t), the probability density of the time an available quantum waits before it is released, and therefore, gives the time course of the release reaction after the passage of a nerve impulse, (c) Gives the value of p (t), the probability density of the time an available quantum waits before it is made unavailable for release and therefore gives the time course of the reaction which makes quanta unavailable for release, (d) Gives the value of p(t), the probability that a quantum available for release at the time of the nerve impulse will be released by the time indicated in the abscissa, (e) Gives the value of n(t), the number of available quanta which are still available for release at various times following the nerve impulse. The values of the stochastic parameters for this synapse were p=0.49 ± 0.02; n=2.0±0.04, =0.85 ms−1; =0.32 ms−1; k=2.3 and =0. [Ca]0=0.4 mM and [Mg]0=1.2 mM. Histogram in (a) determined from 164 quantal releases. For definition of symbols and derivation of equations giving the quantities in (a) to (e), see Section 3.2 in the text (from Bennett et al., 1977). D. Comparison of the predictions of the stochastic analysis with the observed latency-frequency distribution of quantal releases at an amphibian motornerve terminal in different [Ca]0. The [Ca]0 and [Mg]0 were in A, 0.4 mM and 1.2 mM, and in B, 1.8 mM and 10.8 mM respectively. The latency-frequency distribution of evoked quanta were determined at 7.5°C during continual stimulation at 0.2 Hz during which at least fifty quantal latencies were observed at these two [Ca]0/[Mg]0 ratios. The curves drawn through the distributions are according to the equations given in the text (Section 3.2) in which m, p, n, , , k and the latency were in (a) 2.06±0.21, 0.62±0.11, 3.35±0.69,0.41 ms−1, 0.10 ms−1, 3.10, 1.11 ms and in (b), 3.77±0.33, 0.65±0.13, 5.83±0.43 ms−1, 0.10 ms−1, 3.38, 1.29 ms; ± gives the S.E. of the mean. Intracellular recording. The abscissa is the time after the earliest evoked quantal release was observed (from Bennett et al., 1977).
9.3 Kinetics of release of a quantum In 1965 Katz and Miledi showed that the secretion of quanta from a small number of release sites following an impulse in an amphibian motor-terminal branch occurred with a delay after the arrival of the impulse and that the extent of this delay was stochastic. Stevens (1968) introduced a probabilistic approach for the analysis of the stochastic properties of this delay as follows:
178
HISTORY OF THE SYNAPSE
9.3.1 Stochastic analysis introduced 9.3.1.1 Theory We now assume that the probability of release varies with time and can be described as a function of time by p(t), the average rate of transmitter release (Steven, 1968, uses (t)). Let y be the number of units released per nerve impulse (Steven, 1968, uses k), and let P(y, t) be the probability that y units are released in time t. p(t) t is the probability a unit is released in an interval of time t. Assume that releases do not occur together. To have y units released at time t+ t, we have y by time t then none, or y−1 by time t and then one. Therefore
Take t 0 then: The differential equation is solved in the Appendix (Section A9.1). The solution is where p0 is a constant, p0=p(c) for some c (0, t). P(y, t) is the Poisson distribution with parameter p0t. It has coefficient of variation . Then ln In m=−(1/2) ln p0t, which is a linear relationship between ln V and ln m=ln p0t. Consider depolarisations of short duration, so t is small. Let S(t) be the probability that a synaptic delay is greater than t seconds. Then and Substitution of y=0 into the differential equation gives since Thus where s(t) is the p(t) resulting from short depolarisations. Hence
which is Eq. (5) of Stevens or Eq. (1) of Barren and Stevens (1972), and
may be estimated from Fig. 9.3 in Stevens.
9.3.1.2 Experimental results on the amphibian neuromuscular junction This equation for s(t) has been fitted to the observations of Katz and Miledi (1965) in Fig. 9.5A after obtaining the probability density for synaptic delays from the histogram of the number of synaptic delays that occur in each small time interval following the action potential and S(t) from the cumulative number of the synaptic delays that have occurred up to time t. The peak release rate of quanta is at least three orders-of-magnitude greater than the resting spontaneous rate of quantal release and occurs after an absolute delay following the action potential. Barrett and Stevens (1972) subsequently analysed the kinetics of transmitter release at the frog neuromuscular junction. Using just the delays up to the first quantal release following an impulse they obtained a probability of release function (Fig. 9.5B(a)). When this probability was plotted on log-linear coordinates it was found that the falling phase was approximately linear over two orders-of-magnitude of release rate, indicating a single exponential process was involved (Fig. 9.5B(b)): this was followed by the quantal release probability falling to resting levels over a much longer time course. 9.3.2 A kinetic model of quantal secretion at synapses with uniform p The frequency-histogram of synaptic delays, together with the number of quanta secreted at a site, gives information which provides parameter values for simple models of the release process. The simplest model, introduced here deterministically and
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
179
later stochastically, involves a vesicle with its associated proteins of the SNARE complex, docked at the active zone in a secretosome (Rothman, 1994; Bennett, 1996). The influx of calcium ions through the voltage-dependent calcium channel of the secretosome (as well as contribution of calcium ions from surrounding secretosomes, see Bennett et al. (1999) and their binding to the calcium sensor associated with the secretosome (probably synaptotagmin), triggers a series of changes in molecules that constitute the core of the SNARE complex (syntaxin, SNAP-25; synaptobrevin, etc; their being k of these; see Sudhof, 1995) which in the absence of knowledge to the contrary can be taken as occurring sequentially at rate . A vesicle in the secretosome may then go to exocytosis if these reactions proceed unimpeded, or cease to participate in the release process at say rate . The kinetics of this process were described first deterministically and then stochastically by Bennett et al. (1977) in the manner described in the next sections. 9.3.2.1 Theory 9.3.2.1.1 The reaction equation. Assume a large number of quanta, F are triggered to participate in a series of k reactions leading to the release of quanta according to the scheme:
where F is the number of quanta in the reaction, Qk is the state of the quanta immediately before participation in the intermediate states leading to Q1, which is a released quantum, and are rate constants and Fi(t) are the number of quanta in each state at time t after the reactions have commenced, so that F1(0)=F, F2(0)=F3(0)=…Fk(0)=0. This reaction equation leads to the following set of differential equations
where Solve the first equation to give: so that and the initial condition F1(0)=F implies A0=F. Substitute into the second equation, using integrating factor so that where A1 is a constant and the initial conditon F2(0)=0 gives A1=0. Thus . Continuing in this way
once again,
Bennett et al. (1977) Eq. (9) where t represents elapsed time since t0. Therefore . This deterministic expression gives the average number of vesicles released over the n release sites of a terminal immediately following an impulse for the case of a uniform probability secretion over all sites (ie. pj=p). Now consider the same model stochastically (see Bennett et al., 1997). 9.3.2.2 The differential equation. Assume a quantum available at time t may be released in interval (t, t+ t) with probability (t) t or ceases to be available with probabilty (t) t. Let be be the total probabilty a quantum becomes unavailable in time interval (t, t+ t), n(t)=k is the number of quanta available at time t after impulse, m(t)=l is the number of quanta released by time t after impulse, and P(k, l; t) is the probability that k quanta are available and l have been released, at time t after an impulse. Therefore
since (1−k (t)dt− dt) is the probability that nothing changes; (k+1) (t) dt is the probabilty that one ceases to be available when k+1 are available (k+1) (t) dt is the probabilty that one is released when k+1 are available, and dt is the probability of spontaneous releases ( is the parameter of the Poisson background noise).
180
HISTORY OF THE SYNAPSE
Then
(17)
That is solved using a generating function G(u, u; t)= in the Appendix, Section A 10.1. It is found that the differential equation of the generating function is given by with initial conditions n(0)=n, m(0)=0 and G(u, v; 0)=un. The solution of the partial differential equation with these initial conditions is given by Bennett et al. (1977) as
(18)
where and . This is shown in the Appendix (Section A 10.1). 9.3.2.1.3 Using the solution to calculate probabilities The solution can be used to calculate the probability that a given quantum available for release at time t=0 is still available at time t. Assume no spontaneous release, =0 and n=1, and use the probability it is released, Eq. (18) to obtain
Bennett et al. 1977, Eq. (2.2). Then the average number available for release at time t is [Bennett et al., 1977, Eq. (2.3); see Fig. 9.5C(d)]. Since a quantum is released in interval (t, t+dt) with probability (t)dt and is available for release with probability , we sum the product from 0 to t to find the probability that a quantum available for release at time 0 is released during the time period up to t: (19) which is Bennett et al. (1977), Eq. (2.4). Assume n quanta are available for release at time 0. To obtain the probability of m quanta being released up to time t, assume r quanta are released with probability p(t) and the remaining m−r quanta are released spontaneously following an independent Poisson distribution (with parameter t) so that: (20) which is Bennett et al. (1977), Eqs. (1) and (3.0). This is the distribution of a sum of a binomial (n, p(t)) variable and an independent Poisson ( t) variable. If the quanta are not of uniform size, and cannot be counted separately during release, but sum linearly, then the frequency of the eep being x mV high is
see Eq. (9) here or Bennett et al. (1977), Eq. (2). is a gamma variate derived from the Poisson distribution by considering the waiting time for k changes. It has parameters kr= where µ 1 and are the mean and variance of the gamma distribution (i.e. the mean and variance of the quantal size). 9.3.2.1.4 Determination of the average number of releases. To obtain the probable number of quanta released during time interval (t, t+dt) we find the expectation of the distribution of Eq. (20), assuming uniform size for quanta. Thus, we find the
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
181
expectation of a binomial (n, p(t)) plus a Poisson ( t), E[m(t)]= np(t)+ t(Bennett et al., 1977, p. 669). Differentiating with respect to t gives the rate of change and the average number of releases during the time interval (t, t+dt), which is Bennett et al. (1977), Eqs. (3) and (4.0). We calculate h(t)dt in the Appendix (Section A10.2). It is given by (21) where T is large. (This is Eq. (9.0) of Bennett et al., 1977.) This equation gives the average number of quanta released from all n release sites recorded from in the interval (t, t+dt) for uniform probabilities of release (pj=p). In order to obtain estimates for the parameters n, , and k, maximum likelihood methods have been used which are described below in the Appendix, Section A 12.2. 9.3.2.3 Experimental results on the amphibian neuromuscular junction Bennett et al. (1977) showed that was very dependent on the calcium ion concentration, a result which suggests that not only the initiation of the cascade of vesicle-associated protein changes is dependent on calcium but also those changes themselves. The parameter was not calcium dependent, so that the mechanisms which blocks a vesicle participating in the release process is not calcium dependent. Fig. 9.5D shows the results of the stochastic analysis for predicting the frequency histogram of synaptic delays at two different calcium/magnesium concentrations, k remain at about 3 to 4, whilst and have values about 0.4 ms and 0.10 ms−1, respectively at 7.5°C. 9.3.3 A kinetic theory of quantal release at synapses, non-uniform p If a terminal possess release sites with a non-uniform probability of release, then according to the kinetic model considered above there will be different values for at each release site of the terminal (Bennett and Robinson, 1990). 9.3.3.1 Theory Consider n points x1, x2,…,xn along a branch of a nerve terminal, assumed to be located between 0 (proximal end) and 1 (distal end). When an impulse arrives at a release site, a single quantum is released after a random time T, or there is no release. We assume, for site xi, the probability distribution for time to release Ti is These are independent and 9.3.3.1.1 Probability of secretion from a single release site. We now calculate pn(xi, t). A synaptic vesicle is made available for release by the activation of k phosphorylation steps, which occur with rate (xi). At any time during the phosphorylaton process, the vesicle may become unavailable at rate . The former time to completion is a gamma random variable V with parameters (xi) and k, and the time until the vesicle becomes unavailable for secretion is a random variable S with a distribution that is gamma, with parameters , k=1. A quantum can only release if the vesicle remains available, which has distribution
Therefore
182
HISTORY OF THE SYNAPSE
(21)
where and G( (xi)+ , k; t) is the distribution function of a gamma variable with parameters (xi) + , k After the successful completion of the k phosphorylation steps, the vesicle forms a fusion pore, with rate (an exponential random variable), independent of release site. Thus the time for exocytosis, given that the secretion of a quantum occurs, is a random variable W=T+U where U has distribution G( (xi)+ , k; u) and T is an exponential variate with rate (the pdf of the exponential is 0 < t < w). Thus we make a transformation of random variables T, U to W=T+U, U=U. The distribution of W is calculated in the standard way (see, for instance, Hogg and Craig (1978), p. 138). This leads to the standard convolution formula. The inverse function is U=U, T=W−U. The Jacobian is The original joint pdf is
The change of variables W=T+U and U =U changes our area of integration from to By direct substitution into the original pdf, the distribution function is
which is the pdf for W as given by Bennett and Robinson (1990, p.333), with = (xi) and J=1. The mean and standard deviation are calculated using the method of moments in the Appendix (Section A11.1). Thus, for the random variable W we have
This gives them the expected value for the time to secretion and its standard error at release site xi. 9.3.3.1.2 Probability of secretion from n release sites. Let Nt be the number of secretions from n sites up until time t, and Nt=k means secretions from sites xi1, xi2,…, xik where [i1, i2,…, ik] is a subset of [1, 2,…, n]. The probability of secretion at these sites is pn(xi1)pn(xi2)…pn(xik)(1−pn(xik+1)),…, (1− pn(xin)) and the probability of release at exactly k sites is (23) where
is the sum over all such subsets. We put
where is the total number of quantal secretions and is the beta function. Assume n is large, x1, x2,…, xn evenly spaced or close together so that max(xi+1−xi) 0 as n . For any fixed x in (0, 1) there are a sequence of grid points tending to x, say xi(n). Assuming np(xi(n), t) (x, t). Let and then (t) as n (since integration can be treated as a sum). Therefore (t) is the expected total number of secretions up to time t for the limiting process, (t) as t . In the Appendix (Section A7.2) it is shown that the binomial distribution tends to the Poisson distribution for p small, n large and np fixed. Here we have (t) fixed, and equal to and the same proof applies to show that Eq. (23) becomes a Poisson distribution with parameter (t):
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
183
Eq. (4) p. 335 of Bennett and Robinson (1990), where k is a fixed constant. The mean probability will increase if (t) increases, as demonstrated in Bennett and Robinson (1990). In their model of the kinetics of quantal release, Bennett et al. (1 997) introduced the methods of maximum likelihood estimation as described in the next section. 9.3.3.1.3 Estimation of parameters in the theory. The method of moments is used to estimate the mean and variance for the pdf given by Eq. (20). (See the Appendix (Section A12.2).) The mean of the distribution, µ m=M (0)= t+np, and variance, =M (0)− (M (0)2=np(1−p)+ t. The sample mean and variance are and where fm is the observed frequency of m quanta being released. Equating these equations with the first two moments of the distribution: and Dividing the second equation, by the first gives
our estimate of p is, The estimate for n, from the first equation is This is Eq. (13) of Bennett et al. (1977). Thus, we are able to estimate n and p from the sample mean and variance with determined from the background spontaneous release. Alternatively, if n is small and p is not small (p>0.2), it is better to use either (see Section 9.2.6.1) and and Alternatively, use Robinson’s (1976) Eq. (5) (see p. 15 of this paper):
where is the mean of the spontaneous epps ( ), 2 is the variance of the spontaneous epps ( ), m is the mean of the epp amplitudes ( ) and S2 is the variance of the epp amplitudes ( ). 9.3.3.1.4 The introduction of Maximum Likelihood Estimation (MLE) Methods. We wish to estimate parameters , k and t0 using a MLE method as described in the Appendix (Section 12.2). We have M observations consisting of N nerve impulses with mj quantal releases T seconds after the jth nerve impulse. mj=M and T is set large enough to include all quantal releases following an impulse. We take our interval (t0, T) Thus and divide it into g equal subintervals. The value of t in the analysis is taken as ti, the centre of the ith time interval. The number of quanta falling into the interval ti, on the jth impulse is called lij. Let which represents the number of quanta over all nerve impulses with latency falling in the interval ti. Then We need to determine the joint pdf for the parameters , k and t0 using a random sample, T= t1, t2,…,tg. Then
since there are li observations with value ti. To determine our pdf we use Eq. (21) with =0 We take the logarithm of the function L
184
HISTORY OF THE SYNAPSE
Fig. 9.6. Stochastic descriptions of the kinetics of quantal secretion at release sites of amphibian motor-nerve terminal with non-uniform probabilities. A. The effect of increasing the average number of quanta secreted at release sites of an amphibian motor-nerve terminal (increasing ; see Section 3.3.1 of the text) on the time course of secretion after an impulse, (a) The results of simulations giving the number of quanta secreted at different times following a nerve impulse over 643 trials when the average quantal release from 100 sites was increased by increasing from 0.3 (− − −) to 0.5 (…) to 1.5 (−..−..−) to 3 (−.−.−) to 4 (−). (b) The cumulative number of quanta secreted up to any time during the release process for the results given in (a): (open circles), the results from Van der Kloot (1988) (his Fig. 9.3a, third lowest quantal release normalised to highest quantal release) in a low [Ca2+]0 compared with the results for a higher [Ca2+]0 given by (filled circles) (Fig. 3a in Van der Kloot, 1988, highest quantal release), (c) The cumulative number of quanta secreted up to any time during the release process normalized to the total number of quanta released, for a =4 (−) and for a =0.3 (−). Note that (a)–(c) show that for a thirteenfold increase in m there is only a small change in the timecourse of secretion following an impulse as observed experimentally, or in the simulations. Values of parameters in the simulations that generated these results were U=0, number of trials=643, =0.3 to 4, v=0.02, a=1, b=1, k=2, number of sites=100 sites recorded from 1 to 100, =0.5, maximum autoinhibitory propagation distance=± 50 sites. For definition of these symbols see Section 3.3.1 in the text and Bennett and Robinson, 1990). B. The effects of autoinhibition on the estimates of me, pe and ne determined from extracellular measurements of quantal release at release sites of an amphibian motor-nerve terminal. Log-log coordinates for me, pe and ne against [Ca2+]0 are shown. The continuous line gives the gradients of me, pe and ne changes with [Ca2+]0 from Bennett and Lavidis (1979). Estimates of me were obtained from 100 trials for a given and these placed on the me line that fixed the [Ca2 +] for that ; the estimates of p and n were then plotted at that [Ca2+] . (a) A uniform distribution of p (x ,) over 100 release sites with an 0 e e 0 n i imposed uniform random variability, in the absence of autoinhibition; the results are shown for extracellular recording from sites 21–32 ((open circles), =15 to 30) and for separate extracellular recording from sites 31 and 32 ((filled circles); =10 to 30); note that the pe and ne results for recording from sites 21 to 32 do not lie on the gradient lines but those for sites 31 and 32 do. (b) Results for a uniform distribution of nn (xi, ) over 100 release sites with an imposed normal random variability in the presence of autoinhibition (v=0.02); the results are shown for extracellular recording from sites 21 to 32 ((open circles), ( =5 to 30); note that the pe and ne results almost lie on the gradient lines for the recording from sites 31 and 32. Values of parameters in the simulations that generated these results were U=1, number of trials=100, =2.5 to 30, v= 0.03, a=1, b=1, k= 2, number of sites 100, sites recorded from equals 21 to 32 or 31 and 32, =0.5, maximum autoinhibitory propagation distance ±50 sites. Vertical lines associated with the symbols give the sem (from Bennett and Robinson, 1990; for definition of symbols see this paper and Section 3.3.1 of the text). C and D. Probability functions calculated from localised release. C, distribution of first quantal latencies arranged in 0.5 ms time bins, n1(t), divided by the total number of stimuli (continuous line) and the resulting 1(t) function (dashed line). D, the actual time course of (t), determined by plotting all quantal latencies in 0.5 ms time bins and scaling each bin so (t) integrated to the Poisson-determined mean quantal output, m0 (continuous line) and the 1(t) function shown in C (dashed line). The differences were not significant (P= 1) (from Baldo et al., 1986).
To find the values of the parameters we differentiate l by each parameter in turn and set the derivatives equal to zero.
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
185
Identical to Eqns. (15) of Bennett et al. (1977) except they use mi instead of li. One mechanism that might provide for an inhibitory interaction between release sites during secretion is autoinhibition. In this process, the secretion of a quantum is accompanied by the secretion of inhibitory substances that act on receptors located on the presynaptic membrane. Bennett and Robinson (1990) assume that the first point to secrete releases a quantum of transmitter as well as an autoinhibitory modulator whose influence spreads out along the branch at a constant rate. Points enclosed in the region that reach their secretion time after the release site was inhibited are prevented from secreting. An analytic treatment of a stationary Poisson process with rate . and constant rate of inhibition v concludes that the expected value and variance of N, the total number of releases are given by
(Bennett and Robinson, 1990). In this case, the variance remains a constant low percentage of the mean and a binomial-like result is obtained in the presence of such autoinhibition. 9.3.3.2 Experimental results on the amphibian neuromuscular junction Examples of the predictions of this stochastic analysis compared with the observed latency-frequency histogram of quanta at the amphibian neuromuscular junction are given in Fig. 9.5C. Fig. 9.6C shows the distribution of first quantal latencies of quantal releases and Fig. 9.6D of all quantal latencies recorded from a part of a terminal branch at the amphibian neuromuscular junction (from Baldo et al., 1986) for which the theory of the previous section has been applied (Fig. 9.6A). This gives good predictions for the cumulative number of quanta secreted up to any time during the release process in a high calcium concentration for which secretion was high (filled circles in Fig. 9.6A(b)). The effects of an inhibitory interaction between adjacent release sites in recapitulating experimental results is shown in Fig. 9.6B. Here a comparison is made between binomial predictions in the absence (Fig. 9.6B(a)) and in the presence (Fig. 9.6B(b)) of inhibitory interactions, both for recordings from just two sites along a terminal branch (filled circles) and for recordings from eleven sites (open circles). It will be noted that when relatively large numbers of sites are recorded from (open circles) the departure from the binomial case (given by the straight lines) is larger in the absence of inhibitory interactions between release sites (Fig. 9.6B(a)) than when such interactions are accounted for (Fig. 9.6B(b)). For recordings from small numbers of sites (filled circles) there is little difference. That is for n small we have binomial statistics but for n large we must introduce an inhibitory interaction to maintain the applicability of binomial statistics to describe the evoked frequency and distributions of evoked quantal release.
186
HISTORY OF THE SYNAPSE
9.4 Maximum likelihood estimation of parameters in statistical models of quantal release 9.4.1 Multimodal distributions In 1981 Matteson et al., questioned the unity of the quantal hypothesis, namely that the histogram of mepps at an end-plate is distributed as a normal variate (del Castillo and Katz, 1954a) or as a gamma variate (Robinson, 1976). They suggested that the histogram of the mepps could be discerned to be composed of subunits (for a history of this proposal, see Bennett, 1995). Matteson et al. (1981) attempted to fit multimodal distribution (consisting of multiples of normals, each with the mean and variance of the subunit) to the mepp histograms. In order to obtain values for the large number of parameters in these sums of normals distributions these authors used maximum likelihood estimates procedures, which will now be considered. Three assumptions are used to derive a model of a pdf of mepp amplitudes. 1. Small mode mepps (s-mepps) result from release of a subunit of transmitter. Normally distributed with mean µ, variance () where is the variance due to subunit and is the variance due to noise (measurement error). 2. Larger mepps result from release of two or more subunits which sum independently and linearly. Therefore release of j subunits will be normally distributed with mean jµ, and variance (). 3. Each subpopulation of j subunits is weighted by a factor Wj representing the proportion of releases of j subunits. Therefore the pdf is
(24)
so Nj (x) is normal (j µ, j). We use the MLE for parameters (W1,…, Wk−1, µ, s,). We can ignore Wk by noting that The maximum likelihood method has been described in the Appendix (Section A 12.2). The likelihood function is
is the parameter vector
Therefore
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
187
These were solved using an iterative numerical algorithm. The multimodal distributions with parameters estimated in this way gave very good fits to the histograms of epps (Fig. 9.7A (a)) compared with that of other distributions (see for example Fig. 9.7A(b)). 9.4.2 Binomial distributions and deconvolution Korn et al. (1982) were the first to use MLEs of the parameters of distributions, such as the binomial, to histograms of evoked synaptic potentials (inhibitory postsynaptic potentials, or ipsps, from inhibitory interneurones to Mauthner cells). They first carried out a deconvolution of the noise from the ipsps in their records using the following approach. It is assumed noise and response are independent and additive. The formula for transmembrane potential change following an impulse, v given by: where vn is the background potential ‘noise’, E is the driving potential, g is the conductance change, Gm is the resting membrane conductivity and We assume f, fr, fn are the probability density functions of random variables V, and Vn, and that G and Vn are independent. Let the random variable G have pdf r, and the random variable The derivative is the inverse function is and has derivative [u−1 (y)] = (E/(E−y)]2. A theorem from analysis then relates r to fr as (25) which is the correct form of Korn et al., 1992, Eq. (4). The proof in this case is easy. We use the fact that is a one-to-one function with inverse Then
on substituting Thus the pdf is
as Eq. (25) above. To introduce a quantal hypothesis we assume two new parameters gq, the incremental conductance per quantum, and q, the quantal size (as a potential). Then
188
HISTORY OF THE SYNAPSE
Fig. 9.7. Maximum likelihood estimation of parameters in statistical models of quantal release. A. Fits of different models to the distribution of mepps at the amphibian motor end-plate, (a) The fit produced by the multimodal model (see Section 4.1 of text). The number of subpopulations in this case was 16. For each histogram this number was simply determined by the amplitude range of the histogram, under the assumption that the weighting factors of any larger amplitude subpopulations are zero. The maximum likelihood estimates of the parameters (defined in the text) are as follows: W1=4.52×10−1, W2=2.85×10−2, W3=3.19×10−2, W4=2.67 ×10−2, W5=9.78×10−2, W6=1. 97×10−2, W7=2.46×1-1, W8=1.81×10–1, W9=8.49×10−2, W10=7.13×10−3, W11= 6.83×10−3, W12=2.54×10−2, W13=9.57×10−3, W14=8. 36×10−3, W15=1.16×10−2, W16=3.12×10−3, µ=0.60 mV s =0.065 mV, M=0.071 mV. (b) The fit produced by the bimodal model. Maximum likelihood estimates of the parameters are as follows: W=3.96×10−2, µ= 0.61 V, 1=0.094, µ=4.2 mV, 2=1.4 mV. Using the likelihood ratio test criterion the fit in (a) is significantly better (P < 0.001) (from Matteson, Kriebel and Llados, 1981). B. Evidence that the binomial parameter, n, is equivalent to the number of presynaptic endings at terminals with small numbers of release sites (see section 4.2 of text), (a) Samples of potentials intracellularly recorded in the M-cell. Fluctuating postsynaptic responses (a1, arrows) produced by single impulses in a presynaptic inhibitory interneuron (not shown) and control collateral inhibitory postsynaptic potential (IPSP; following an antidromically evoked action potential), (b) Schematic representation of terminals established by peroxidase-filled interneuron, which was a second-order commissural vestibular cell (comm neuron). All four synaptic boutons (histological n) were within the axon cap. (c) and (d) Comparison of the observed IPSP amplitude variation (stepwise plot for 133 responses) with the best fits obtained assuming Poisson ((c), dashed line) and binomial ((d), continuous curve) equations. None of the curves passed the Kolmogorov test (PB<0.02 against PP<0.01); test for possible equivalence showing that the latter was more satisfactory. Note that the binomial term n had a value of 4 (from Korn et al., 1982). C. Fluctuation analysis of a single group 1 fiber EPSP using the measured EPSP (+ noise) and probability density distributions shown in (a) (see section 4.2 of text). Probability density calibration in (a) applies only to EPSP+noise distribution: calibration is 0.006µV1 for noise distribution. Solid curve in (b)–(d): EPSP+noise probability density distribution. Dotted curves: proposed fits to this distribution. Solid bars: proposed noise-free distribution obtained by the unconstrained quantal model (b), the simple binomial model (c), and the nonuniform (compound) binomial model (d). Probability calibrations in (b)–(d) refer to these bars, dv is the quantal interval and N is the number of underlying quantal components for each distribution (Walmsley et al., 1988). D. The use of maximum likelihood estimates for fitting distributions to histograms of EPSPs due to la afferents on lumbosacral motoneurones. (a) Contains two histograms: the shaded histogram was made up from 2000 measurements of intracellular noise voltage with zero mean and N=130 µV. The unshaded histogram was made up from 2000 samples of a mixture of three distributions. Fifty per cent of the samples were obtained by adding 195 µV to further measurements of intracellular noise; 25% were obtained by adding 390 µV to noise measurements and the remaining 25% were simply drawn from the noise distribution. This made the ratio = 1.50, where V1=195µV and represents the ‘quantal’ increment. Thirty pairs of histograms were constructed in this manner, and the maximum likelihood estimates of the noise and mixture parameters (assuming perfect sampling) is shown by the filled bars in (b), where the height of each bar indicates its probability of occurrence, and its position on the abscissa indicates its amplitude. The maximum likelihood estimates for each pair of histograms have been superimposed with each filled circle corresponding to a probability and an amplitude. The results in (c) and (d) were obtained similarly, except that in (c) the quantal amplitude was reduced to make and in D it was increased to make (from Kullmann et al., 1989). E. Maximum likelihood estimates for the non-uniform (compound) binomial model (see Section 4.2 of text). Models I–VI represent, respectively, the cases were p1=p2=p3=p4; p1=p2=p3
1 (from Smith et al., 1991).
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
189
and (g/gq)=i is an integer number of quanta, so the potential change due to response is
The standard convolution formula (see p. 143 of Hogg and Craig, 1978, or p. 31 above for the method of proof) calculates the pdf for the random variable V= [E G/( G+E)]+Vn or We put so that [dz/d(iq))]=[E2/(iq+E)2], and by Eq. (25) above.
Hence
This result differs from Korn et al. (1982) Eq. (3 ) which is,
whereas the result here gives
where r (i) is the pdf that i units are released after an impulse. We place a characterisation (binomial or Poisson) and a normal pdf on fn and then use the MLE criterion. Using the MLE criterion to obtain the parameters for a uniform binomial or a Poisson distribution Korn et al. (1982) were able to fit their amplitude-frequency histograms of ipsps as shown in Fig. 9.7B(c) and 7B(d). Good fits were obtained for the binomial distribution compared with that for the Poisson, although no consideration was given to the non-uniform binomial distribution. MLE procedures were next used by Smith et al. (1991) to examine the applicability of a non-uniform binomial. They used a numerical method to calculate the MLE, and used moment estimates as starting points for the iterative method. Using this MLE they obtained the estimates for their parameters for a different combination of non-uniformity in the probabilities for secretion for four different release sites at the crayfish motor-nerve terminal, shown in the Table of Fig. 9.7E. Walmsley et al. (1988) used deconvolution to remove the noise from their frequency-histograms of epsp amplitudes and then fitted a non-uniform binomial using MLE criteria. Again a similar technique is used—deconvolution and then MLE. They included an unconstrained model (that is, r (i) is not made to conform to a particular function on i ). For instance in the binomial case, their procedure is: 1. Estimate the pis where the pi is the probability of release for the ith population. 2. Convolve with the measured noise distribution. 3. Compare with the measured EPSP and noise. 4. Correct the pis.
190
HISTORY OF THE SYNAPSE
Fig. 9.7C(d) shows that the compound non-uniform binomial gave a very good fit to the histogram of probabilities of release for different number of quanta after the deconvolution from the noisy histogram (Fig. 9.7C(a)) had been carried out. This may be compared with Fig. 9.7C(c) for the uniform binomial which is not good. This analysis led to the important conclusion that there are many release sites at a single terminal which have low probabilities for secretion and may, therefore be considered to be reserve sites, in much the same way as had been discovered at the neuromuscular junction. The MLE technique was used again by Kullman et al. (1989) and applied in two stages. The noise histogram was approximated by the sum of two normal curves. Then the MLE was applied to the evoked EPSP, treating noise as the contaminating distribution. Their procedure gave them histograms of the probabilities of quantal release shown in Fig. 9.7D. 9.5 Autocorrelation function used to detect quantal release Magleby and Miller (1981) introduced an autocorrelation (AC) procedure into synaptic physiology in a search for objective criteria as to whether the peaks in the amplitude-frequency histogram of mepps observed by Kriebel and his colleagues (see Fig. 9.7A) could be ascribed to a subunit of transmission. This is whether the quantal unit was actually composed of subunits. To this end they obtained autocorrelations of a histogram of simulated mepp amplitudes constructed assuming the subunit hypothesis, as shown in Fig. 9.8A, in a search for regularity of the form of regularly paced multiple peaks, rather than from just random variations in the data. Autocorrelation is usually proposed to analyse the relationship between the values of a process at distinct times s and t. Amplitude histograms do not have time as the independent variable (along the x-axis) but rather they have amplitude. The correlation is carried out on the same process at different amplitudes. For instance, one AC function (Acf) given by Aseltine (1958), is where H(i) is the height of the ith bin, centred at the amplitude value given by the i×bin width.7 The Acf defined here correlates the random variable H at amplitudes i and i+j. When it is applied to amplitude histograms it tests for the occurrence of the regular peaks which would lend support to the subunit hypothesis. This procedure depends on the researcher determining a suitable bin width, usually by ‘guessing’ based on visual scanning of the peaks and troughs of the amplitude histogram. Using this approach Magleby and Miller (1981) ascribed the peaks in the histogram of mepp’s as due to random variations, although Vautrin (1986; see Fig. 9.8B) did not. Edwards et al. (1990) applied the AC procedure to histograms of evoked ipsps at synapses between interneurones and hippocampal granule cells. After applying the Acf, Edwards et al. (1990) introduced the technique of smoothing the obtained function using a five-point rolling average. The smoothed Acf and the original Acf were then superimposed (see Fig. 9.8C(b)) and if the obtained function showed enough consecutive equally spaced peaks and troughs the original amplitudes were judged to be quantally distributed. Again, this depends on an arbitrary decision-making process, for instance, Edwards et al. (1990) required two such consecutive peaks and dips. Application of this approach to simulated data in Fig. 9.8C, indicates the power of the Acf to distinguish peaks and whether they are equidistant apart as required by the quantal hypothesis. Edwards et al. (1990) were able to show that histograms of evoked ipsp amplitudes often showed several peaks which were equidistant, therefore providing evidence for quantal transmission at this synapse. Stratford et al. (1997) extended the AC approach described in Edwards et al. (1990) in three main ways. First, they subtracted a smoothed distribution from the original data distribution to obtain a difference function with enhanced peaks. Second, they quantified the peakiness to obtain a single number ‘AC score’. Finally, they automated the procedure. They also attempted to make the AC scoring procedure objective. Only three decisions need to be made by the researcher: the choice of bin width (which is not critical provided it is ‘small’); the value of the filter to obtain the smoothed curve; and, the value of the filter to smooth the difference function. The method they use is described in detail in the following. 9.5.1 AC Scoring Stratford et al. (1997) binned the amplitude histograms very finely (usually 5µV) to minimise distortions produced by coarse binning. A Fast Fourier Transform (FFT) was then used to perform a low-pass filter of the data points.8 The filter strength
7
(i×bin width) represents a shift along the x-axis (in picoAmps) which would be expected to be a function of an integer multiple of a subunit of transmitter.
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
191
(that is, the frequency cut-off,) was adjusted so there were no discernible ‘bumps’ or ‘peaks’. The smoothed function was then subtracted from the original peaky histogram to obtain adifference function. The difference function itself was lightly smoothed to remove high frequencies while keeping the main peaks. Then its AC function was calculated using a Fourier technique, (see Appendix, Section A13). To obtain a single number measure to characterise this AC function, they took the height of the first peak measured from the base of the preceding trough, and then calculated the absolute difference in amplitude between these two locations as the autocorrelation (AC) score for the histogram. In this method, histograms with a regular succession of evenly spaced peaks and valleys give high AC scores, whereas one has to know the periodicity to obtain a high score using Aseltine’s (1958) function. Stratford et al. (1997) tested the effectiveness of the AC score to perform the task of assessing whether a particular peaky data histogram resulting in a high AC score represents a true succession of equally spaced peaks and thus provides evidence of the subunit hypothesis, or could have arisen by sampling artifact. Both ‘quantal’ and ‘non-quantal’ data distributions were simulated and the results of the AC scoring procedure were compared with those obtained by conventional statistical tests (the 2 test and the Kolmogorov-Smirnov test).
8A
low-pass filter aims to eliminate noise at high frequencies.
192
HISTORY OF THE SYNAPSE
Fig. 9.8. The use of autocorrelation functions to detect quantal release. A. Use of the autocorrelation functions to detect subunits of spontaneous potentials (mepps) at the motor end-plate (see Section 5 of the text), (a) Histogram of 60,000 simulated mepp amplitudes constructed assuming that the mepps were composed of 1 to 14 subunits each 0.1 mV in amplitude. The SD of the subunit was 12% of its mean amplitude. The number of mepps composed of 2, 3, 4,…, 14 subunits were determined from a Gaussian function with a mean of 7 subunits and a SD of 2 subunits. The mean and variance of the distributions of mepp amplitudes for each of the integral numbers of subunits were determined by multiplying the mean and variance of the subunits by the number of subunits for that distribution. Thus the variance of the mepp amplitudes for each distribution increases with the number of subunits, obscuring the peaks of the individual distributions that contain 10 or more subunits. (b) Autocorrelation of the first half of the Histogram in (a). There is a regular repeating pattern with intervals which are integral multiples of 0.1 mV, the subunit amplitude, (c) Histogram of 1000 mepp amplitudes drawn randomly from the distribution in (a). Peaks due to random variation are apparent, mixed in with those due to subunits. (d) Autocorrelation of the first half of the histogram in (c). Despite the presence of random variation in the data, the autocorrelation readily detects the subunit pattern, (e) Histogram of 60, 000 simulated mepp amplitudes constructed assuming that the mepps were not composed of subunits but instead formed two Gaussian distributions with the means of 0.75 mV and 0.05 mV and SD of 0.20 mV and 0.10 mV respectively. The small mode distribution is assumed to extend into the base-line noise, (f), autocorrelation of the first half of the histogram in (e). The autocorrelation is smooth but shows the strong correlation between the two modes of the histogram, (g), the histogram of 1000. mepp amplitudes drawn randomly from the distribution in (e). Note the appearance of multiple peaks, many of which appear to be regularly spaced, (h) Autocorrelation of the first half of the histogram in (g). There is no regular pattern, indicating that the peaks in (g) are irregularly spaced and that these mepps are not composed of subunits. As in (f) the two modes of the histogram correlate with each other, (from Magleby and Miller, 1981). B. Frequency analysis showing stationarity of peak intervals throughout sequential plots of m.e.p.p amplitude distributions, according to autocorrelation analysis, (a) Sequential plots of successive series of 612 (1), 734 (2), 725 (3), 634 (4) and 591 (5) mepp amplitudes. The total of 3297 mepps is plotted in (6). (b) Respective autocorrelograms of the histograms in (a); the oscillation period is constant throughout all the histograms of this experiment. As the tenth wave of the autocorrelograms corresponds to an interval of 59.5 bins (arrow head), the calculated interpeak is 5.95 bins, (c) Respective PDS of the histograms in (a). An increase in power in all the PDS between 20 and 23 cycles for 128 bins (arrow heads) indicates a constant frequency in the sequential plots, (a) The interpeak observed was 5.56 to 6.40 bins or 0.28 to 0.32 mV. The size of the subunit noted was 0.30 mV. The vertical lines appearing over the histograms in (a) represent 21.5 cycles/128 bins, as calculated, and correspond to the peaks in all the sequential plots (from Vautrin, 1986). C. Autocorrelation criterion for equidistant peaks in the amplitude-frequency histograms of inhibitory postsynaptic currents on granule cells in the hippocampus, (a) An example of a simulated distribution with equidistant peaks. Thick line represents the sum of the underlying Gaussians (thin lines), (b) Symbols connected
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
193
The Monte Carlo simulations used samples from two forms of unimodal distribution—the Gaussian function (see Eq. (26) in the Appendix) or the Weibull function:
where ag is the location parameter, bg is the scale parameter and cg is the shape parameter, subject to x ag, bg>0 and cg>0. The advantage of the Weibull function is that it can show skew in either direction. Quantal processes were also simulated using binomial models of the form:
which is a quantal model with nc components, allowing for noise with mean 0 and standard deviation of n, quanta with amplitude Q and standard deviation Q, and where Fi is the frequency of the ith component. This is identical to Eq. (24) described above. To calculate the AC score for the simulated distributions, random numbers were generated and transformed to the required distributions. For the Gaussian distribution the Box-Muller method (see Press et al., 1986, Section 7.2) was used to transform the uniform deviates into Gaussian form. For the Weibull function, uniform deviates U1 were transformed using the expression as in Stratford et al. (1997) Eq. (5), where the ag, bg and cg were calculated by applying a simplex optimisation to the data histograms under consideration, (see Press et al. 1986, Section 10.4.) The simple binomial was simulated using parameters N, for the number of release sites and Pr, the probability that each site releases with a quantal response of amplitude Q. A random number was generated for each of the N sites and if less than Pr, the quantal amplitude Q was added to the response. The amplitude value was modified by adding a Gaussian distributed random variable with mean 0 and standard deviation n to model the recording noise. The procedure was repeated for each of the amplitudes required in the simulated data set. After calibrating the AC scoring method with the simulated data sets, the AC score test was compared with conventional statistical procedures designed to determine whether observed differences between two distributions is significant or due to random sampling. Stratford et al. (1997) performed simulations in which the 2 and Kolmogorov-Smirnov goodness of fit tests were used as well as AC scoring. Samples from a quantal generator (the simple binomial) were compared to unimodal Weibull or Gaussian distributions of similar overall shape. They concluded that the AC scoring method was more specific in extracting quantized responses. It also gives information about the mean spacing of the peaks in the histogram and hence of quantal size. Fig. 9.8D shows an example of this approach. 9.6 Model discrimination: Statistical methods for discrimination between different statistical models of transmitter release Horn (1987) introduced techniques for the discrimination amongst different statistical models as to their applicability to best describe a stochastic biological phenomena. The sections below describe these procedures. A proposed model’s usefulness in data analysis can be characterised by two main properties: firstly, the model’s fit to empirical data, and secondly, the model stability. Model stability is the dependence of the model on the particular observed data set, and is relevant in the case of models which contain unknown parameters that need to be estimated. Two main approaches are recognised, either one model is assumed to be correct and evidence must be found to make the selection, alternatively it is assumed that no one model is correct and therefore the modeler searches for a parsimonious model. The problem addressed here is how to choose between models, based on available data and using statistical criteria. This difficulty was first considered by Matteson et al. (1981) in their analysis of possible multimodal histograms of mepps. There are two general classes of model comparison: nested and non-nested. The former is nested in the sense that one model is a subhypothesis of the other. A standard hypothesis test, where one chooses between accepting or rejecting the null hypothesis (H0 true or H0 not true) is used. H0 is that the sub-hypothesis is true. The level of the test (probability of rejecting H0 when in fact H0 is true) is set, usually at p=0.05 or p=0.01. The power of the test (probability of rejecting H0 when H0 is false) usually increases with sample size. Thus the models are considered asymmetrically. With non-nested or ‘separate’ models, the two models may be considered symmetrically or asymmetrically. Asymmetric model choice is generally referred to as hypothesis testing, symmetric model choice is sometimes called model discrimination.
194
HISTORY OF THE SYNAPSE
9.6.1 Nested Models For composite hypotheses, maximum likelihood methods can be used when the distribution for each model is known, and we wish to estimate parameters. The basis for these methods is described in the Appendix, Section 9.6.1. Let the pdfs of the models be f(x, ) and g(x, ) where x is the data vector, and and are the unknown parameter vectors, of dimensions kf and kg, respectively. Since model F is assumed to be a sub-hypothesis of model G, kg>kf. The natural logarithm of the likelihood ratio (LLR) is defined as
where sup g(x, ) denotes the supremum of g(x, ) and and are the parameter value that maximise the likelihood for each pdf, that is, the maximum likelihood estimates of the and . Under certain regularity conditions, when model F is true, 2×LLR has a central 2 distribution asymptotically with (kg−kf) degrees-of-freedom. Thus this test can be used to test whether model G fits the data better than model F at an -level of significance. 9.6.2 Non-nested models The theory for non-nested models is not as well established. There are two approaches: 1. Prediction error approach (Akaike, 1974, see list of references in Horn, 1987). This ranks non-nested models treating them symmetrically. Unfortunately the significance levels for model discrimination are not known and so one model is always rated better or worse in this method. 2. Monte Carlo methods in which significance levels can be set. In both cases the LLR defined above is used. In the former the critical value is taken as kg−kf. Thus, model F is rejected in favour of model G if LLR> kg−kf. (The critical value here, kg−kf is referred to as the asymptotic information criteria (AIC)). This approach rewards a model for parsimony. 9.6.3 Estimation of the distribution by the log likelihood ratio Under the hypothesis that model F is true, the LLR has a normal distribution asymptotically. see Horn (1987), p. 256, where LLFf denotes the LLR under the assumption model f is true (Cox and Hinkley, 1974, Chap. 9). However, µ f and are not readily calculated for the complicated kinetic models used, for example to describe ionic channels, and so must be estimated using Monte Carlo methods or using a resampling technique known as bootstrapping. In this method, the data is assumed to be identical and independently distributed samples, and are treated as a subpopulation that is resampled with replacement to create artificial data sets. Each data point is chosen with equal probability so that each data set has the same size but with some values omitted and some repeated. A large number of data sets may be generated and the LLR is calculated. An empirical distribution of LLRs is obtained by continuing in this way. Hypothesis tests can be formulated about this distribution. If parsimony is ignored, a critical value of zero is used, and the null hypothesis is that F and G are indistinguishable as models to fit the data. If parsimony is to be rewarded the AIC can be used. This model treats the models symmetrically. These procedures when applied to the fit of various models to the gating kinetics and permeation of the acetylcholine receptors channel are shown in Fig. 9.9A (Horn, 1987). The LLR was first applied in quantal analysis by Korn et al. (1982) to obtain objective descrimination for the application of binomial versus Poisson statistics to the histograms of ipsps at synapses on Mauthner cells, which showed that the binomial distribution was superior (Fig. 9.9B). Jonas et al. (1993) compared the fits of different functions to the histograms of epsps at mossy fibre synapses on CA3 pyramidal cells in the hippocampus and used the LLR to distinguish between these (Fig. 9.9C). Stricker et al. (1996) introduced bootstrap statistical methods and the use of the Wilks test to provide objective criteria as to whether the probability of the LLR for one set of models was superior to that of another (Fig. 9.9D). Model comparison using the Wilks test is treated in the Appendix, Section A14.
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
195
Appendix A.1 Introduction to the theory of random variables We consider an experiment in which the outcome cannot be predicted with certainty, but for which all outcomes can be described prior to performing the experiment. If this kind of experiment can be repeated under the same conditions we call it a random experiment and the collection of every possible outcome is called the experimental space or sample space. If an experiment is repeated N times, and N is large then we can predict approximately how often the various outcomes will occur —the probability of various events. Given a random experiment with a sample space, C, then a function X which assigns to each element c of C a real number x, is called a random variable. Sometimes C is a set of real numbers, and then we would let X (c)=c for c C. Let be the set of real numbers which X maps to. We wish to consider the probability of an event A, a subset of this space. We write Pr(X A) and consider how the probability is distributed over the various subsets A of . A.2 Probability density functions (pdf) Sometimes the distribution of probability for a random variable X can be described by a functon moment we describe the simple case of distributions of one random variable.
[0, 1].9 For the
A.2.1 The discrete type of random variable If a random variable X maps to a finite number of points, and if a function f(x) is defined such that for x random variable of the discrete type and f(x) is a pdf of the discrete type.
and ,10 then X is a
A.2.2 The continuous type of random variable Let be such that where for all x and f(x) has at most a finite number of discontinuities in every finite interval. If is the space of a random variable X and if then X is a random variable of the continuous type and f(x) is a pdf of the continuous type. A.2.3 Functions of two random variables The notion of a pdf of one random variable can be extended to a pdf of two or more random variables. For instance, if we have or . A.3 The distribution function If we consider sets A, from − to x (including the point x), then is called the distribution function F(x). For random variables of the discrete type, and for the continuous type, A.3.1 The normal distribution The normal of Gaussian distribution with mean µ and variance
2
is given by (26)
9 10
[0, 1] is a function f(x) which takes a real number x and assigns to it a number in the interval [0, 1]. means that f(x) summed over all values of x is equal to 1.
196
HISTORY OF THE SYNAPSE
Fig. 9.9. Statistical methods for discriminating between different statistical models of transmitter release. A (a). Applications to gating kinetics and permeation of the acetylcholine receptor (AchR) channel. Histogram of open,times for single acetylcholine receptor channels. Best fit curves are shown for two different models B and C (for details of these see Horn, 1987, and Section 6 of the text), (b). Histogram of the logarithm of the likelihood ratios (LLRs) obtained from 200 bootstrap samples of the original data. The LLR of the original data is shown by an arrow. The curve is the best fit Gaussian to the LLRs. (c). Histograms of LLRs from data simulated under either model C (left) or model B (right). Best fit Gaussian curves are displayed. The LLR from the original data set (arrow) is consistent with model B but not model C (from Horn, 1987). B. Probability density function of inhibitory postsynaptic potentials (IPSPs) in Mauthner cells. Fluctuations and computer-modeled fits with Poisson and binomial predictions. (a1) Sample unitary IPSP (upper trace) evoked by a presynaptic impulse (lower trace). For this and the following figure, recording from and direct stimulation of inhibitory interneurons were achieved using peroxidase-filled microelectrodes. (a2) Evoked IPSP of larger amplitude recorded at a slower sweep with a late-occurring spontaneous PSP (arrow, upper trace) and on-line computed average (n=8) of several responses (lower trace), (b) Comparisons of the observed IPSP amplitude variations (stepwise distributions for 182 counts) with the best fits obtained assuming a Poisson (upper) or binomial (lower) relationship. Abscissas, amplitude of the evoked responses: ordinates, density of observations expressed as the number of occurrences per millivolt (mV). The likelihood criterion was better for the binomial, which yielded a value of 6 for parameter n, while p and q were 0.47 and 300 µV respectively. The binomial prediction satisfied the Kolmogorov test (P>0.05) but the Poisson did not (P<0.01), and the test for possible equivalence between models indicated the binomial was more satisfactory (for details see Section 6.3 of text; from Korn et al., 1982). C. Comparison of quantal and non-quantal models fitted to experimental peak excitatory postsynaptic current (EPSC) amplitude distributions to obtain a quantitative estimate of the ‘quantalness’ of these currents at mossy fibre synapses on CA3 pyramidal cells in the hippocampus. Unbinned data > 2 pA were fitted, and the function with the parameters obtained from maximum likelihood fitting was superimposed on the amplitude histogram after appropriate scaling. For further details see the text of Jonas et al. (1993). (a) Fit of EPSC amplitude distribution with a continuous function, (b) Fit of the same distribution with a quantal function. Natural logarithm of the likelihood ratio (LLR)=5.1. The LLR values when fitting the data with the model pairs (3a)–(4a) and (3c)–(4c) were higher (41.8 and 35.7, respectively). Thus, model pair (3b)–(4b) may provide an estimate of the maximum p of rejecting the non-quantal model when true in the worst case. (For details of the model pairs considered, see Jonas et al., 1993). The value of p can be estimated from the respective LRR value as 1.7%. Therefore, the histogram was judged to be quantally distributed, (c) Fit of another amplitude distribution with a continuous function, (d), fit of the same distribution with a quantal function. LLR=1.2 LLR values for models (a) and (c) were higher (1.5 and 10.5 respectively); maximum P=49.4%. Therefore, the histogram could not be judged to be quantally distributed (from Jonas et al., 1993). D.
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
197
A.4 Mathematical expectation Let X be a random variable with pdf f(x) and let u(x) be a function of X such that u(x)f(x) or u(x) f(x) dx exists for random variables of discrete and continuous type, respectively. This sum or integral is called the mathematical expectation or expected value of u(x) and is denoted by E[u(x)]. Once again the notion can easily be extended to functions of more than one variable. Expectation is a linear operator (a property which follows easily from the definition). This means, that for k1, k2 constants and u1, u2 functions, we have A.4.1 Special mathematical expectations (a) The mean value, µ of a random variable X is defined (when it exists) to be µ=E[X]. (b) The variance, 2 of X may be defined by
since E is a linear operator The standard deviation of X, equals a. It is a measure of the dispersion of the points of a space relative to the mean value µ. (c) The mgf is given by E[etX] and is denoted by M(t). If an mgf exists it is unique and hence, completely determines the distribution of the random variable. By repeatedly differentiating M(t) with respect to t and setting t=0 we obtain
In other words we can generate the moments of the distribution. (E[Xm] is called the mth moment. A.5. Coefficient of variation Pearson’s Coefficient of Variation is defined by A.5. Stochastic independence If X1 and X2 are random variables with joint pdf f(x1, x2), and if X1 and X2 have marginal pdfs f(x1), f(x2) then X1 and X2 are stochastically independent if and only if f(x1, x2)=f(x1)f(x2). A.6 Sampling theory Definition. A function of one or more random variables that does not depend on any unknown parameter is called a statistic. Although a statistic does not depend upon an unknown parameter, the distribution of the statistic may do so. As an example we may know the distribution of a random variable of X except for the value of an unknown parameter. To obtain more information about this distribution (or the unknown parameter) we repeat a random experiment n times. Let the random variable Xi be a function of the ith outcome, i=1, 2,…,n, then X1, X2,…, Xn are items of a random sample from the distribution under consideration. A. 6.1 Two important statistics Let X1, X2,…, Xn denote a random sample of size n from a given distribution. The statistic
198
HISTORY OF THE SYNAPSE
is called the mean of the random sample (or sample mean), and the statistic
is called the variance of the random sample (or sample variance).11 It is also written as var . Theorem 1. Let X1, X2,…Xn be mutually stochastically independent random variables with identical distributions with mean µ and 2 the variance, then the mean of the Xi, the random variable has mean µ and variance 2/n. Proof. Recall that . and The expected value of is
For each Xi, µ is the mean and 2 is the variance, so that . Since the Xis are mutually stochastically independent, E[Xi Xj]=E [Xi] E[Xj]= µ 2 for . There are terms of this form. Therefore
Thus the variance of
is A.6.2 Standard error
The standard error is the standard deviation of the sampling distribution of a statistic (that is the square root of the sampling variance). The standard error is important when the sampling distribution of the statistic under consideration tends to normality and if n is large enough for this tendency to be effective. Thus the standard error is just the standard deviation calculated using the parameters obtained from the sample, rather than from the population parameters. Standard errors are used to obtain an approximate measure of precision in estimating statistical parameters from large samples. A.6.3 Standard errors of functions of random variables To find the standard error of functions of random variables, we take the square root of the sampling variance. Kendall et al. (1987, Section 10.5) defines the sampling variance of a function of two random variables, by12 (29) where the sampling variances and covariances can be determined from the standard error of the moment statistics. It is not intended to fully examine the theory of moments here, but a brief explanation of how to evaluate Eq. (29) is presented in the following.
11
Some authors define
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
199
Kendall et al. (1987) uses to represent the ith moment about a constant (usually zero), so that M(i)(0) as calculated using the mgf defined by Eq. (27). Further, µ i (without the prime) is the ith moment about the mean. In Section 9.3.2, Eq. (3.9) Kendall et al. (1987) presents the following relationships:
(30)
To distinguish population values and statistics, Kendall et al. (1987) adopts the convention of using a Greek letter for the former and Roman letters for the latter. Thus, the rth moment corresponds to the rth moment of the population about zero. The expected value of the rth moment-statistic is given by: Similarly the rth moment-statistic about the mean mr has expected value The latter is not an exact result, but as approximation to order n−1/2 (see Kendall et al. (1987), Section 10.4 for a full derivation). The sampling variance of is given by Kendall et al. (1987) Eq. (10.9) as
For r=1 and r=2, we obtain the variance of the 1st and 2nd moments about the mean. That is, the mean and variance of the sampling distribution. The following results are derived from Kendall et al. (1987) Eq. (30) and (10.10) and our Eq. (30). (31) since (32)
(33) A.6.4 Estimation of parameters An estimate of a parameter may be obtained by calculating the corresponding statistic from the sample. We use â to represent the estimate of a parameter a. The standard error gives an approximate measure of the precision of this estimate. A.7. Binomial and Poisson theory A.7.1 The moments of the binomial distribution The pdf of the binomial distribution is given by for y=0, 1, 2,…, n Using the mgf to determine the mean and variance13 we have
12This
is derived from Eq. (10.12), p.324 of Kendall et al., 1987.
200
HISTORY OF THE SYNAPSE
by the binomial theorem. Then
Hence
(34)
Therefore (35)
A.7.2 Derivation of Poisson from the binomial Consider what happens if p is small and n is large. Let np bea constant, say m=np. Then the binomial term
Let n
and p 0 with m fixed. Then the first y factors
1, and since ,14 the last factor thus
which is the Poisson distribution. The mgf is 13 14
See Section A.4.1. can be derived from the series expansion, ex=1+x+x2/2!+x3+3!+···.
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
201
by the series expansion of e
So the mean µ=M (0)=m and the variance variation as given in Section A.4.2 as
2=
M (0)−µ 2=m2+m−m2. For the Poisson distribution, Pearson’s coefficient of
A.7.3. Determination of standard errors for estimates in the binomial case To obtain a measure of the reliability of the estimate of p, we calculate the standard error of using a result from Kendall et al. (1987) (given here by Eq. (29)) for the variance of a function of two random variables. In our case by Eq.(2)
and by Eqs. (31)–(33) with N the number of random experiments, we have
We now calculate µ 3 and µ 4 for the binomial case by Eq. (30)
by Eq. (34)
since by Eq. (30)
by Eq. (34)
202
HISTORY OF THE SYNAPSE
Then by Eq. (29)
(36)
since
Therefore the standard error is
the correct form of Zucker’s (1973) result (7)
and Robinson’s (1976) result (2)
since and
as in Robinson’s (1976) Eq. (2)
with
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
203
which is Whitakers result, Eq. (vi), since Similarly, if we use to estimate n we have Using Kendall’s formula for sampling variance and noting
and We have
(37)
since
since
which is Robinson’s (1976) Eq. (2), or Whitakers Eq. (v) with p and q interchanged, rather than the incorrect result of Zucker (1973). A.8. The compound binomial A.8.1 The moments of the compound binomial with a normal variate The pdf is, for (38) and it is clear that We use the mgf to derive E[X] and E[
]
204
HISTORY OF THE SYNAPSE
since the rest is n(yµ 1+y t, y )
Therefore
Now
(39)
Also
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
205
(40)
by Eq. (39) above We now calculate the expected value of X and
by
A.8.2. Derivation of gamma from Poisson The gamma distribution can be derived from the Poisson distribution, by considering the waiting time for k changes, where k is a fixed positive integer. The distribution function is
For the random variable W to be greater than w, we have less than k changes in time interval w. Therefore which is the Poisson distribution with parameter m= w. Consider the following integral which we solve using integration by parts
Therefore
We make the change of variable z= y, then
206
HISTORY OF THE SYNAPSE
which is the Gamma distribution with parameters ( , k). The gamma distribution has mean k/ and variance k/ derived from the mgf as in Hogg and Craig, 1978, pp 105–106).
2
(which can be
A.8.3 The moments of the Compound binomial with a gamma variate The pdf of the compound gamma, for x > 0, is (41) and P(X=0)=qn. Our mgf for t
is
by integration by parts. Hence
The mean as given by the first moment about zero is
using the result of Eq. (39) above. The second moment is
by Eqs. (39) and (40). Therefore the variance is
which is identical to the compound binomial case. A.8.4 Estimating the standard errors of and To derive me standard error of our estimates for n and p as given by Eqs. (7) and (8) we use an extension of the sampling variance Eq. (29). For
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
which is Eq. (7) above.
and for which is Eq. (8) above
where
In the case of the normal variate (Robinson, 1976, case (b)), we have
and
and in the case of the gamma variate (Robinson, 1976, case (c)), we have
and
207
208
HISTORY OF THE SYNAPSE
A.9 Stochastic analysis of transmitter release A.9.1 Solving the differential equation (42) This can be solved using an integrating factor Then
Therefore
Also, by Eq. (42), with y=0 therefore, the probability that no units are released up to time t is given by Then
and
Make the substitution . Then
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
and continue to obtain
For a particular t, the mean value theorem shows where the constant p0=p(c) for some c (0, t). Then which is the Poisson distribution. A.10 Kinetic model of quantal secretion at synapses, uniform p A.10.1 Solving the differential equation We wish to solve (43) A.10.1.1 Solution by generating function. The generating function is defined to be (44) (45) (46) (47) Differentiating each line we obtain (48) (49) since k+1=0 at k=−1 (50) (51) and (52) Substitute for using Eq. (43)
209
210
HISTORY OF THE SYNAPSE
(53)
and note that by Eq. (48)
by Eq. (49)
Eq. (51) and P(k, −1; t)=0 by Eq. (46). Substitute into Eq. (53) to obtain
(54)
which is Bennett et al. (1977) Eq. (1.3). At time t assume n quanta are available for release, n(0)=n, and none have been released, m(0)=0 and G(u, v; 0)=un. The solution of the partial differential equation with these initial conditions is given by Bennett et al. (1977) as
where and We show this by noting that and
as required.
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
211
A.10.2 Determining the average number of releases We calculate where p(t) is given by Eq. (19). Assume that for the n quanta available for release at time t015 the waiting times for release or to become unavailable are independent gamma variables. Then (t)= (t)=0, 0 t t0.
These denominators serve to scale the pdfs appropriately, as the gamma function is usually scaled by obtained by substituting of To calculate the expected time for a single quantum to become unavailable, put l=1 to yield:
Calculating *(t) for t > t0 since (t)=0 for 0 t t0
by putting
as Eq. (5.2) of Bennett et al. (1977). Thus the average value of n(t) is
Bennett et al. (1977) Eq. (6.0). Hence the probability a unit is released at time t is given by Eq. (6.0) as
212
HISTORY OF THE SYNAPSE
by substituting for −t0. This is Eq. (7.0) of Bennett et al. (1977). Also since (t)=0 for 0 t thus for T . Hence
t0 P(t)=0 in this region. Integration by parts results in
Thus
which is Eq. (9.0) of Bennett et al. (1977) and used as Eq. (21) above. A.11 Kinetic theory of quantal release, non-uniform p A.11.1 Secretion from a single site We calculate the mean and variance of the distribution function:
We use the method of moments. To find the mgf, we first change the order of integration.
since the latter integral is gamma ( + −t, k)
15
t0 is sometimes assumed to be time of nerve impulse and sometimes the time of increased release probability following a nerve impulse.
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
Therefore the variance is
Thus for the random variable W we have
A.11.2 Estimation of parameters The mgf is
The sums Hence
and
are equivalent.
213
214
HISTORY OF THE SYNAPSE
and
Then
Hence the mean of the distribution, µ m=M (0) = t+np, and variance,
=M (0)−(M (0))2= np(1−p)+ t.
A.12 Maximum likelihood estimation methods A.12.1 Point estimation Let a random variable X have a pdf that is a known function but which depends on an unknown parameter which has any value in a set . We write the pdf as f(x; ); . is called the parameter space. Clearly, this is not a single pdf but a family of distributions {f(x; ) . We wish to determine a good point estimate of , given a set of observed experimental values x1, x2,…, xn from a random sample X1, X2,…, Xn of the distribution. That is, we wish to define a statistic Y1= u(X1,…, Xn) so that the value y1=u1(x1,…, xn) will be a good point estimate of . A.12.2 The likelihood function The function we use is the likelihood function. L( ) which is proportional to the joint pdf for the parameter we are trying to determine. Thus
We ask for the value of which maximises the probability L( ). This value of is the same as would maximise it’s logarithm In L( ). The maximum is found by setting the derivative of the function equal to 0, and solving for . Thus we determine the value of for which
QUANTAL SECRETION AND THE STATISTICS OF TRANSMITTER RELEASE
215
A.13 Correlation using the fast Fourier transform technique In general, the correlation of two functions is defined by Press et al, (1986), Eq. (12.0.10) as which gives the ‘Correlation Theorem’, (Press, et al., 1986, Eq. (12.0.11)) where G(f) is the Fourier transform of g(t), and H*(f) is the complex congugate of the Fourier transform of h(t), ad the ‘ ’symbol indicates transform pairs. The correlation of a function with itself is autocorrelation. The above equation becomes which is the ‘Wiener-Kinchin Theorem’. Thus to compute correlations using the Fast Fourier transform (FFT), the procedure is to 1. FFT the two data sets. 2. Multiply one resulting transform by the complex conjugate of the other. 3. Inverse transform the product. The result will be a vector of real numbers representing the correlation at different time lags. A.14 Model comparison using the Wilks test Stricker et al. (1994) describe the Wilks test as used to decide if a model provides a better fit to observed data than another model, providing the models are nested. In this procedure, let data set X be a family of random variables, yielding the likelihood function
where i are parameters lying in a specified region. Call this model B, and let model A be nested in model B. Then model A for the same data set X has a likelihood where the primed parameters have a specified value lying in the range of parameter values for model B. Furthermore, assume that the likelihood function is asymptotically normal. Let be the likelihood for model A evaluated at its maximum, that is
where
is the maximum likelihood estimator. Then the Wilks statistic
is asymptotically distributed as a X2 random variable with v degrees of freedom. If the two models are not nested, the Wilks statistic W can still be calculated, but now we do not know the distribution of W (as it no longer has a 2 distribution (Striker & Redman, 1994)). Stricker et al. (1994) estimated the distribution of W by choosing one model as the null hypothesis (H0) and the other as the alternative hypothesis (H1). The expectationmaximisation algorithm is used to maximise the log-likelihoods for the best fit to the data, producing L0 and L1. The model yielding the best fit under H0 becomes the parent probability density from which Monte Carlo sampling occurs. Each sample is fitted assuming H0 and H1, and the Wilks statistic (W= −2(L0−L1)) for the best fit to each model is calculated. Stricker et al. (1994) repeated this procedure 250 times and rank-ordered W to produce a cumulative distribution, thus allowing significance testing at an =0.05 level. If H0 cannot be rejected at <0.05, then H0 is accepted. In this case, model selection is based on the principle of parsimony.
10 The Discovery of Long-term Potentiation of Transmission at Synapses
10.1 Introduction: the hippocampus, memory and long-term potentiation (LTP) In 1957 Scoville and Milner reported that bilateral medial temporal-lobe resection in man, sufficient to remove or damage the anterior hippocampus and hippocampal gyrus (Fig. 10.1 A), causes a persistent impairment of recent memory. Furthermore this impairment depends on the extent of hippocampal removal, with no effects occurring if only unilateral removal is performed. The bilateral resections left early memories, technical skill, personality and general intelligence unimpaired (Fig. 10.1B). Scoville and Milner concluded that ‘the anterior hippocampus and hippocampal gyrus, either separately or together, are critically concerned in the retention of current experience.’ This conclusion was subsequently quantified by Prisko (1963). She sampled five different sets of stimuli, each set constituting a separate task, with three of the tasks visual and two auditory; the stimuli used were clicks, tones, shades of red, light flashes and nonsense patterns. Normal subjects averaged only one error in twelve trials even with an intratrial interval of 60 sec whereas a subject with bilateral medial temporal-lobe resection has only one error if one stimulus immediately follows the other, but if the stimuli were separated by 60 sec the score approached the chance level of six errors in 12 trials (Fig. 10.1C). This realisation in the late 1950s, early 1960s that the hippocampus might be the main site in the brain for the processing of recent memories was paralleled by the first intracellular recordings of synaptic transmission to the pyramidal cells in the hippocampus by Kandel et al. (1961) as well as by Andersen et al. (1963). However, these early investigations on transmission did not connect the possibility that long-term changes might be sought in synaptic efficacy at these synapses given the recent discoveries by Scoville and Milner (1957). This was left to Terry Lomo, who as a consequence of experiments performed in Per Anderson’s laboratory in Oslo, reported in 1966 that: Extracellular responses of dentate granule cells, evoked by repetitive stimulation of the entorhinal area or perforant path fibres, were recorded simultaneously with two microelectrodes. One electrode recording from the layer of perforant path synapses on the granule cell dendrites, the other from the layer of granule cell bodies. After an initial depression, lasting for a few seconds, repetitive stimulation led to a large potentiated response, compared to the response evoked by a single volley. This effect, frequency potentiation, was seen as an increase of the amplitude and a decrease of the latency of the population spike and as an increase of the rate of rise and amplitude of the extracellular excitatory synaptic potentials. This represents an example of a plastic change in a neuronal chain, expressing itself as a long-lasting increase of the synaptic efficiency. The effect, which may last for hours, is dependent upon repeated use of the system (Lomo, 1966). Lomo’s research on the hippocampus, first described in 1966, was repeated in 1969 in a collaborative effort with Tim Bliss who joined Lomo in Oslo. They found that on stimulating the perforant path in anaesthetised rabbits and recording the synaptic responses in the granule cells, that ‘one or more brief episodes of tetanic stimulation (15/sec−1 for 10–15 sec) produces a potentiation of the monosynaptic response evoked by single shocks which may last for several hours (Bliss & Lomo, 1970)’. The full report of this research, which appeared in 1973, constitutes the first quantitative description of LTP. It begins a long series of experiments by Bliss to elucidate the mechanism of LTP that continue to the present. The essential components of this report involve measurements of the amplitude of the population post-synaptic potential (EPSP), signalling the depolarisation of the granule cells, and the amplitude and latency of the population action potential (Fig. 10.2A(b)) following stimulation of the perforant path fibres to the dentate area of the rabbit hippocampus (Fig. 10.2A(a–c)). The population EPSP and action potential were elevated, following an initial period of depression immediately after stimulation at about 15 Hz for 15 sec, for periods up to 10 hours (Fig. 10.2B, C). The extent of potentiation following different numbers of
THE DISCOVERY OF LONG-TERM POTENTIATION OF TRANSMISSION AT SYNAPSES
217
Fig. 10.1. Memory in the hippocampus. (A) Diagrammatic cross-sections of human brain illustrating extent of attempted bilateral medial temporal lobe resection in the radical operation (for diagrammatic purposes the resection has been shown on one side only; from Fig. 2 in Scoville & Milner, 1957). (B) Table shows the histories and individual test results for 10 cases following bilateral hippocampal lesions. These cases have been divided into three groups representing different degrees of memory impairment (from Table in Scoville & Milner, 1957). (C) Effect of bilateral medial temporal lobe resection on the ability to compare stimuli that are separated by a short time interval. The graph shows the mean error scores of patient H.M., for five tasks, as a function of the intratrial interval. Six errors would be chance performance. (From L.Prisko (1963) and Figure 25.2 in Milner, 1972).
conditioning trains is shown in Fig. 10.2C(c): the amplitude of the population EPSP is elevated 300% by four episodes of stimulation, with the last two separated by about one hour. In a subsequent paper Bliss and Gardner-Medwin (1973) repeated these experiments, only on anaesthetised rabbits in which they were able to study the time course of LTP over days rather than just hours. These experiments showed that the amplitude of the population action potential in the granule cell layer of the dentate gyrus after appropriate conditioning stimulation of the pyriform pathway axons (Fig. 10.2E) was still elevated after 16 weeks compared with controls (Fig. 10.2D). Douglas and Goddard (1975) obtained similar results in the dentate gyrus following stimulation of the entorhinal cortex, with LTP being followed for two months. Furthermore, at this time Schwartzkroin and Wester (1975) showed that LTP could also be induced in synapses on CA1 neurones in the hippocampus,
218
HISTORY OF THE SYNAPSE
Fig. 10.2. The discovery of LTP. (A) Diagrammatic parasagittal section through the hippocampal formation, showing a stimulating electrode placed beneath the angular bundle (ab) to activate perforant path fibres (pp), and a recording microelectrode in the molecular layer of the dentate area (AD), (b) The region enclosed in the rectangle in (a) enlarged to show the apical dendritic field of the granule cells, with the perforant path fibres confined to the central one third of the field. The population responses evoked by a strong perforant path volley in the synaptic layer (upper trace) and in the cell body layer (lower trace) are displayed on the right. The spots (lower trace) mark the peaks between which the amplitude of the population spike was measured, (c) Arrangement of electrodes for experiments in which the control pathway was situated in the contralateral hippocampus, (d) Electrode arrangement for experiments in which both test and control pathways were on the same side. Abbreviations: ab, angular bundle: AD, dentate area; CA1, CA3, pyramidal fields CA1 and CA3; Fim, fimbria; hipp fiss, hippocampal fissure; inf, mossy fibres; pp, perforant path (from Fig. 1. in Bliss & Lomo, 1973). (B) Frequency potentiation and the immediate after-effects of a conditioning train. The graphs show the changes produced in three parameters of the evoked response by increasing the rate at which the perforant path was stimulated from 0.5 to 15–1 sec for 15 sec. The points obtained during the conditioning train and the following 1 min 45 sec are shown on an expanded time scale. Note immediately after the train the brief period of spike potentiation followed by depression, and the subsequent maintained potentiation. The value of the population EPSP during the train could not be accurately measured from the film record and is not plotted (from Fig. 2 in Bliss & Lomo, 1973). (C) An experiment in which all three standard parameters of the evoked response were potentiated. Three superimposed responses obtained in the synaptic layer for both the experimental and control pathways are shown in a. (before conditioning) and in b. (2– 5 hours after the fourth conditioning train), c. Graph showing the amplitude of the population EPSP for the experimental pathway ( ) and the ipsilateral control pathway ( ) as a function of time. Each point was obtained from the computed average of 30 responses by measuring the amplitude of the negative wave 1 msec after its onset. The values are plotted as percentages of the mean pre-conditioning value. Conditioning trains (15–1 sec for 10 sec) were given through a medially placed conditioning electrode at the times indicated by the arrows (from Fig. 4 in Bliss & Lomo, 1973). (D) Measurements of spike amplitude (from averages of 16 responses to 30 V stimuli) plotted as several trains of stimuli (15–1 sec, 15 sec) were given at the indicated strengths, and at various indicated times afterwards. The approximate mean spike amplitude before the conditioning trains is shown by the dotted lines. Average responses at the times marked a, b and c are shown beneath (from Fig. 7 in Bliss & Gardner-Medwin, 1973). (E) Atypical average response recorded using a computer, with the derived parameters labelled on it. Average of 16 responses. (From Fig. 1 in Bliss & Gardner-Medwin, 1973).
following stimulation of the stratum radiatum, and that this lasted for at least several hours. Thus at the end of the 1970s both the pyriform synapses on dentate granule cells and the Schaffer collateral synapses on CA1 hippocampal neurones were
THE DISCOVERY OF LONG-TERM POTENTIATION OF TRANSMISSION AT SYNAPSES
219
identified as sites for the generation of LTP. At the end of the papers by Bliss and his colleagues the possibility that LTP may be related to the capacity of the hippocampus to process the short term memory first described by Milner and her associates is enunciated: The perforant path is one of the main extrinsic inputs to the hippocampal formation, a region of the brain which has been much discussed in connection with learning and memory. Our experiments show that there exists at least one group of synapses in the hippocampus whose efficiency is influenced by activity which may have occurred several hours previously—a time scale long enough to be potentially useful for information storage. Whether or not the intact animal makes use in real life of a property which has been revealed by synchronous, repetitive volleys to a population of fibres the normal rate and pattern of activity along which are unknown, is another matter (Bliss & Lomo, 1973). and Since the phenomenon is present in healthy anaesthetised animals it is at least possible that its mechanism could underlay some form of plasticity under normal conditions in the hippocampus (Bliss & Gardner-Medwin, 1973). The problem posed by these prescient comments, namely is LTP involved in a memory process in the hippocampus, is still not resolved to the present time. However, as we will see below, much has been revealed concerning the distribution of synapses that posses LTP in the brain, and how this is induced and maintained. 10.2 LTP at synapses in the brain 10.2.1 LTP in different parts of the brain The phenomenon of LTP is not restricted to synapses in the hippocampus. In the early 1980s LTP lasting for about 15 hr was discovered in the kittens’ visual striate cortex at synapses formed by geniculate cells (Komatsu et al., 1981). Soon after, Gerren and Weinberger (1983) showed that synapses formed by the brachium of the inferior colliculus in the medial geniculate nucleus of the auditory system also possessed LTP that was sustained for at least 1 hr. Racine et al. (1983) made a systematic study of the distribution of synapses that possess the ability to undergo LTP in the limbic system. They carried out a comprehensive study in vitro of the affects of stimulating various input pathways to particular parts of the limbic system (amygdala, entorhinal cortex, dentate gyrus, subiculum, septum and hippocampus CA1 and CA3 fields), on the size of the response amplitude in these areas (termed the output). LTP was determined for each of these areas 24 hours after a three-day control run or 24 hours after three days of administration of potentiating trains to the stimulation site giving the experimental run. Input/output curves were then constructed relating the size of the stimulating pulse applied to the input and the response amplitude in the target area. Fig. 10.3A shows the results for the control ( ) and the experimental ( ) cases. It will be noted that the largest increase in the experimental response over that of the controls occurs at about 50% maximum amplitude in the controls. These observations show that the stimulating sites, in order of increasing effectiveness, are as follows: the lateral olfactory tract (LOT) shows no LTP, the pyriform cortex shows little LTP, followed by the amygdala, entorhinal cortex, perforant path, CA1, septal area and fornix/fimbria; stimulation of the last three structures produces the largest and most reliable LTP effects. The threshold for producing LTP effects was highest in the pyriform cortex and became lower as the stimulation site moved caudally along the pyriform lobe. Note that hippocampal pathways show significantly larger LTP effects than do the non-hippocampal pathways. The time course of maintenance of the LTP in each of these limbic structures was extensive as originally described by Bliss and Gardner-Medwin 10 years earlier for the pyriform pathway input to the dentate gyrus granule cells: the LTP in each area in which it was shown to be significant decayed with a long time constant of about five days (Fig. 10.3B, compare with Fig. 10.2D(b)). Repeated stimulation of the inputs failed to extend the period of maintenance of LTP (de Jonge & Racine, 1985). Racine et al. (1983) completed their study with the comment that: ‘With few exceptions, LTP could be produced throughout the limbic forebrain’. 10.2.2 LTP and senescence If the hippocampus is involved in the consolidation of new memories and this involves the mechanism of LTP, then it is natural to enquire as to whether any changes occur in LTP with ageing, as clearly memory deteriorates with longevity.
220
HISTORY OF THE SYNAPSE
Neurones in the human hippocampus and dentate gyrus as well as the adjacent entorhinal cortex undergo substantial changes with senescence, with a loss of dendritic spines and eventually loss of whole dendrites: thus a large number of synaptic connections are lost with senescence. In 1979 Barnes carried out an enquiry in which she first implanted electrodes into the fascia dentata and perforant path of both mature (10–16 months) and senescent (28–34 months) rats. Senescent rats did show a poorer memory for the rewarded places when tested on a circular platform that favoured the use of spatial clues. She then recorded the granule cell responses to perforant path high-frequency stimulation during four separate sessions. The mature rats maintained the increased synaptic strength associated with LTP for at least 14 days following the fourth session (Fig. 10.4A), whereas LTP declined rapidly over 24 hours following the fourth stimulation session in the senescent rats (Fig. 10.4A). These observations of Barnes showed that the amount of synaptic enhancement during LTP was statistically correlated with the ability to perform the circular platform task both within and between the groups. The expected correlation between LTP and senescence based on the idea that spatial memory depends on LTP in the hippocampus was therefore realised, although of course no causal relationship between memory and LTP was established by these experiments. 10.2.3 Development of LTP Although the expected correlation between senescence and LTP was found at an early stage of hippocampal research, in the 1970s, it is not so clear as to what may be expected regarding the development of LTP. Recognition memory that is dependent on the hippocampus develops early in humans, reaching a high level of maturation in the first 21 months (Diamond, 1990). This correlates with the observation that the human hippocampus is 40% mature at birth and fully mature at 15 months (Kretschmann et al., 1986). By comparison the rat hippocampus is only mature by three weeks, equivalent to about three years in a human (Diamond, 1990). It is therefore of considerable interest to examine the time course of LTP development in rats in order to see if the maturation of the memory process is paralleled with that of the mechanisms responsible for LTP in this relatively late maturing hippocampus. Neonatal rats are capable of simple learning tasks (Caldwell & Werboff, 1962; Johanson & Hall, 1982; Thoman et al., 1968), so that such rats should possess some form of LTP at this stage of development. However, although post-tetanic potentiation is present at one week in rats, LTP does not appear until the second week, either in the dentate gyrus (Fig. 10.4B; Wilson, 1984; see also Wilson & Racine, 1983) or in area CA1 (Fig. 10.4D; Harris & Teyler, 1984) of the hippocampus. Before 10 days of age the maintenance of LTP is relatively transitory, lasting for only 3– 4 days, whereas by 15 days the fully matured LTP is in place, showing maintenance without decay over a 4-day period (Fig. 10.4C). It is interesting that Harris & Teyler (1984) noted that the size of the LTP in two-month-old rats was less than that in two-week-old rats, so that the decrease in LTP associated with senescence observed by Barnes (1979; see Fig. 10.4A) which occurs between rats aged about 16 and 34 months is an underestimate if the comparison is made with twoweek-old rats. 10.3 The induction of associative LTP in the brain In 1978, McNaughton et al. showed that LTP of perforant path synapses on granule cells following high-frequency stimulation is a cooperative process requiring coactivity of a large number of axons. A threshold stimulus intensity during the high-frequency stimulation was required in order to obtain any LTP, considerably higher than that required to show a synaptic response at all, as is shown in Fig. 10.5a. Furthermore, the magnitude of this LTP was shown to increase as the intensity of the high-frequency stimulation increased. In addition, concurrent high-frequency stimulation of axons of the medial and lateral perforant pathway produced LTP at intensities that were identical, whereas independent stimulation of the two pathways either failed or was relatively ineffective. These authors came to the important conclusion that ‘the results of these two sets of studies separately and jointly indicate that enhancement (LTP) is a cooperative phenomenon’. This work went some way towards the discovery of ‘a minimal logical requirement for a physical theory of associative memory’, namely that ‘there be some change in the potential for interaction between input and output elements of a system, resulting from use. This change must be such that a pervious output may be reactivated by some subset of the corresponding previous input’. Furthermore, the observations of McNaughton et al., pointed to the correlation between high-frequency synaptic activity and postsynaptic depolarisation as a factor for the induction of associative LTP. The search in the early 1980s was then to identify how a postsynaptic depolarisation might bring this about.
THE DISCOVERY OF LONG-TERM POTENTIATION OF TRANSMISSION AT SYNAPSES
221
Fig. 10.3. LTP in different parts of the brain. (A) Representative input (pulse intensity)/(output (O, control; experimental) curves, before and after LTP for 16 different pathways involving limbic system structures. (B) (a) The population EPSP and the population spike of the dentate response evoked by perforant path stimulation, (b) Septal responses evoked by subiculum and CA1 stimulation, (c) CA1 responses evoked by subiculum and septal stimulation. Twenty trains were delivered each day for three days, starting on day 0 (see abscissa). The baseline and day 3 EPSP traces are also shown. The exponential curves and associated statistics are shown. For the inserts: vertical calibration, 1500µV (a) and 1000µV (b and c); horizontal calibration, 5 msec (from Figs. 1 and 4 in Racine et al., 1983).
10.3.1 Associative LTP is blocked by N-methyl-D-aspartate (NMDA) antagonists The discovery by both Watkins and McLennan and their colleagues of amino-acid antagonists in the early 1980s (see Watkins & Evans, 1981; McLennan & Liu, 1982) precipitated enquires into the identification of the excitatory transmitters in the hippocampus. Collingridge et al. (1983) made the momentous discovery that the selective NMDA antagonist DL-2-amino-5phosphononalerate (APV) in doses that abolished responses of pyramidal neurons in CA1 to N-methyl-DL-aspartate (NMA) had no effect on the synaptic potentials due to stimulation of the Schaffer collaterals, but blocked induction of LTP (Fig. 10.6A). On the other hand, -D-glutamylglycine (DGG) could antagonise the action of the agonists quisqualate, kainate
222
HISTORY OF THE SYNAPSE
Fig. 10.4. LTP during senescence and development. (A) Mean fractional change of the slope of the EPSP of granule cells following high-frequency stimulation of the perforant pathway over its value at V0 ,T1, T2, T3, and T4 represent the first through fourth times, respectively, that these rats were given high-frequency stimulation. The time points shown after high-frequency stimulation [2 min, 10 min, 1 hr, etc.] represent the time elapsed between the end of the stimulation episode and the beginning of the test pulse measurements for that given time point (from Fig. 12 in Barnes, 1979). (B) Percentage of animals per age group demonstrating steady tetanic potentiation (___, STP; ---, LTP) of the EPSP recorded in the dentate hilus (from Wilson, 1984, Fig. 1). (C) The maturation of LTP in the hippocampus. The longevity of LTP was monitored beyond the 20 min post-tetanus test in several experiments. Four examples are presented here. The ten pretetanus population spike amplitudes are plotted before the vertical line. The vertical line indicates the time at which the tetanus was delivered to the nerve, and then a scatter plot of the post-tetanus response amplitudes is shown for 1-, 5-, 8-, 10- and 15-day old rats. The dashed lines at the 1 mV population spike amplitude is shown to illustrate the magnitude of the change following the tetanus. Gaps in the records from the 8-day-old (at 25–30 min.) 10-day-old (at 63–68 min) and 15-day-old animals (at 22–27 min. o) are where the field EPSP threshold was measured, and the monitor of LTP was discontinued briefly. At 8 and 10 days the response declined back to base line at about 1 h and 1 ½h, respectively. By 15 days the response was stable at a potentiated level for at least 72 min, the duration of the period monitored. An unusual response pattern is present at 5 days (O): temporary increases in the response amplitude, followed by response depression, occurred at 24, 36 and 48 min after tetanus (from Harris & Teyler, 1984, Fig. 4). (D) The magnitude of LTP at different ages in the CA1 region of the hippocampus, (a) Pre- and post-tetanus wave forms from representative experiments on rats of different ages, (b) Two estimates of the magnitude of LTP produced at several ages. On the left, the amount of LTP was calculated as a percentage change in the amplitude of the post-tetanus population spike relative to the pretetanus population of the ([post-pre/post]×100%). On the right, the amount of LTP was calculated as the absolute increase in the amplitude of the population spike following tetanus (post-pre). The dashed line at zero indicates no change in the response. For the sake of clarity the means are plotted with standard error bars in one direction only (from Harris & Teyler, 1984, Fig. 3).
and NMA and blocked both synaptic transmission and LTP induction (Fig. 10.6A). Their conclusion was that ‘the present study has shown that the NMA receptor plays no role in mediation of synaptic excitation but may be involved in the generation of LTP in the Schaffer collateral-commissural projection to the CA1 region of the rat hippocampus’. These
THE DISCOVERY OF LONG-TERM POTENTIATION OF TRANSMISSION AT SYNAPSES
223
Fig. 10.5. Cooperative processes involved in LTP of perforant path synapses on granule cells. Height of the extracellular EPSP in granule cells to constant low intensity (30 µsec) test impulses to the perforant path plotted for four preparations. Pulse width during each 100 pulse trains at 100 is shown on the abscissa. Vertical calibration bars are all 0.5 mV. (a) One of 15 preparations in which pulse width was increased progressively during high-frequency trains. Note that <110 µsec, only short-lasting changes were observed. Above this value, the amount of long-lasting enhancement increased with increasing stimulus intensity, (b) A second example similar to a, but in which some enhancement was observed at 90 usec. The stimulus train marked with an asterisk shows that once some enhancement was generated at an intermediate intensity (130 µsec), the low-intensity train (30 msec) still had no long-lasting effect, (c) An example of an experiment in which repeated low-intensity trains produced no enhancement, while a subsequent high-intensity train did. (d) Repetition of a moderate intensity train (50 µsec) produced enhancement which saturated. Subsequent higher intensity stimulation (225 µsec) produced further enhancement (from Fig. 2 in McNaughton et al., 1978).
observations were confirmed by Harris et al. (1984) using 5- and 7-phosphono compounds as NMDA receptor antagonists (Fig. 10.6B) as well as by others (Wigstrom et al., 1986a; Reymann et al., 1989). 10.3.2 Associative LTP requires postsynaptic calcium increase In 1983 Lynch and his colleagues showed that intracellular injection of the calcium chelator EGTA into pyramidal cells of the CA1 region in the hippocampus blocked the induction of LTP due to stimulation of the Schaffer collaterals to the neurones (Fig. 10.7A). These results pointed to the dependence of induction on the level of free calcium in the postsynaptic cell, which then brings about some modification of the cell. The simple suggestion that depolarisation of the cell by a tetanus to the axons forming synapses on it opens postsynaptic voltage-dependent calcium channels that give rise to the influx of calcium ions that triggers induction of LTP will, of course, not suffice in the case of associative LTP of the kind demonstrated by McNaughton et al. (1978): if this occurred LTP would not be associative as all the synapses on a cell would be potentiated, not just those that had been stimulated. The reciprocal of the Lynch et al. (1983) experiment, namely increasing the calcium level in the postsynaptic cell and seeing if this induces LTP, was not performed until 1988 (Malenka et al., 1988). These authors showed that if a photolabile nitrobenzhydrol tetracarboxylate calcium chelator, nitr-5, is used to release calcium in CA1 pyramidal cells in response to ultraviolet light, then LTP was induced in the synapses on these neurones for a period of about one hour (Fig. 10.7B). 10.3.3 Magnesium gates NMDA receptor channels: the missing link in the induction of associative LTD At the end of 1983 three facts had been established concerning the induction of associative LTP, which at this time did not fit into a coherent scheme of LTP induction. They were that a threshold depolarisation seemed to be necessary, that NMDA receptors must be occupied by the transmitter glutamate and that an increase in calcium in the postsynaptic cell was necessary. All of these observations were brought together by the discovery due to Nowak et al. (1984) that magnesium gates
224
HISTORY OF THE SYNAPSE
Fig. 10.6. NMDA receptors implicated in associative LTP. (A) Effects on the population spike and on the generation of LTP of NMDA receptor antagonists. The amplitude of the population spike was recorded in stratum pyramidal in response to 0.1 Hz stimulation of the Schaffer collateral-commissural projection and antagonists were administered ionophoretically in stratum radiatum for the durations indicated by the bars. The pathway was stimulated at 100 Hz at the times shown by the arrows below the trace. DGG reduced the amplitude of the population spike and prevented LTP in response to high frequency stimulation. Following recovery from the effects of DGG, LTP was produced using a period of identical high frequency stimulation. After 17 min, APV was applied and had no effect on the population spike, 2 min later the stimulus intensity was reduced (lower record) and 3 min later the effects of high frequency stimulation were again tested. Although some short lasting post-tetanic potentiation resulted, LTP was prevented. After the APV injection had been terminated for 8 min, 100 Hz stimulation for 1 sec was again able to produce LTP which did not recover completely over the time that responses were measured (45 min) (from Fig. 5 in Collingridge et al., 1983). (B) Effects of NMDA receptor antagonist ± AP5 on Schaffer-collateral responses LTP. Constant current biphasic pulses (40–100 µsec, 50– 200 µA) were delivered to a bipolar electrode placed on the Schaffer-collateral fibers in stratum radiatum of CA1 at one every 20 sec. A series of three high-frequency trains (50 or 100 Hz for 1 sec, one every 20 sec) was delivered at the times indicated by S1 and S2. Antagonist was present in the medium bathing the slice during the time indicated by the bar. (a) Upper trace, responses recorded by an electrode placed in stratum radiatum. Representative responses from the following conditions: before and during drug application. Ten minutes after the first high-frequency stimulation (S1), following drug washout, and 10 min after the second high-frequency stimulation (S2). Following the high-frequency stimulation in the absence of drug (S2), the synaptic response is increased in amplitude, and a population spike appears (fifth trace, asterisk). Lower trace, frequency potentiation during the high frequency stimulation. The first five responses recorded during high frequency stimulation in the presence of ± AP5 (SO and after washout (S2) are superimposed (dotted and solid traces respectively), (b) Extracellular evoked responses recorded by an electrode placed near the CA1 pyramidal cell layer. Experiment as in part a, except that the (−) isomer of AP5 (final concentration 50 µM) was applied before the first high-frequency stimulation (S1). Calibration bars, 1 mV, 5 ms. (From Harris et al., 1984, Fig. 2). (C) Comparison of L- and D- isomers of AP4 and AP5 on the induction of LTP. L-AP4 or D-AP5 were present during the intervals indicated by bars. High frequency stimulation as delivered at the times indicated by arrowheads. The commissural/associational response (a) is insensitive to L-AP4, but the induction of commissural/associational LTP is completely blocked by D-AP5. In contrast, mossy fibre responses (b) are profoundly reduced by L-AP4, and although there is a slight effect of D-AP5 on mossy fibres responses, induction of UP is not blocked by D-AP5. (Calibration: top, 2 min, 0.5 mV; bottom, 2 min, 0.3 mV) (from Harris & Cotman, 1986, Fig. 2). (D) Magnesium gates NMDA receptor channels. Glutamate-induced single-channel currents in outside-out patches. In Mg2+-free solution (a), the single-channel currents appear as simple events on both sides of the reversal potential (ca −5mV in this experiment). After addition of Mg2+ (10 µM) (b) the currents recorded at +40 mV are unchanged whereas the currents recorded at −60 mV appear as bursts. The size of the single-channel current is not altered by the addition of 10µM Mg2+. In this experiment, MgCl2 (2 mM) had been added to the internal solution. Note that the single-channel conductance is 51 pS at −60 mV, and only 33 pS at +40 mV. This discrepancy was observed in all experiments where Mg2+ (2 mM) was present in the internal solution but was not seen when Mg2+ was absent. Thus, when Mg2+ was absent both the internal and external solutions, the I-V relationship of the glutamate-induced current was linear (between −60 and +60mV) at the single-channel level as well as at the whole-cell level (from Nowak et al., 1984, Fig. 2).
glutamate-activated channels. In 1980 Watkins (the discoverer of the NMDA receptor) and his colleagues (Ault et al., 1980) had shown that the response of these receptors was greatly potentiated by reducing the extracellular magnesium concentration to below the physiological level of 1 mM. Subsequently, MacDonald et al. (1982) showed that the NMDA receptor possesses
THE DISCOVERY OF LONG-TERM POTENTIATION OF TRANSMISSION AT SYNAPSES
225
Fig. 10.7. Associative LTP requires postsynaptic calcium increase. (A) Effects of the calcium chelator EGTA in CA1 neurones on LTP due to stimulation of Schaffer collaterals. Intracellular recordings of an average of six successive responses 10 sec apart collected immediately before and 15 min after five bursts of 300sec−1 stimulation, each lasting 35 msec, delivered through a bipolar electrode in the trajectory of the Schaffer-commissural fibres. Calibration bars: 5 mV and 5 msec for this and subsequent panels, (b) Intracellular recording with an EGTA-filled electrode from a CA1 pyramidal neurone. Averaged responses to six single-pulse stimulations collected immediately before and 15 min after four high frequency stimulation trains are superimposed, (c) Percentage change in amplitude of the EPSP after high-frequency stimulation (arrow) for the cell illustrated in a. The x axis is time in minutes. Each line represents a single response and the average amplitude for the control period is expressed as 100%. The high amplitude responses are spikes that have been chopped in preparation of the figure, (d) Amplitude of individual EPSPs expressed as percentage of the averaged responses collected before high-frequency stimulation for the cell described in b. EGTA was injected before collecting control data, (e) Intracellular recording from a neurone with an EGTA-filled electrode. Calibration as in a. (f) Recording as in e, but after withdrawal from the cell (from Lynch et al., 1982, Fig. 1). (B) The release of calcium in CA1 pyramidal neurones of hippocampus induces LTP. This figure shows a comparison of the photolysis of calcium loaded and unloaded nitr-5 in CA1 neurones on stimulation by flash photolysis. , Mean EPSP slope recorded with electrodes filled with the calcium loaded nitr-5. , Mean EPSP slope recorded with electrodes filled with unloaded nitr-5. The duration of the flash varied from 10 to 45 sec. The bars represent ± SEM. Each point was normalised to the average value of 20 EPSP slope measurements immediately prior to the flash (from Malenka et al., 1988, Fig. 2).
an unusual voltage sensitivity: above the resting potential the current induced by a particular concentration of glutamate increases with depolarisation of the cell. This observation was explained in terms of the discovery of the magnesium dependence of the NMDA receptor when Mayer et al. (1984) designed an experiment which showed that the voltagesensitivity of the NMDA receptor is dependent on the presence of Mg ions. The work of Nowak et al. (1984) provided a complete explanation for both the voltage-dependent properties of the NMDA receptor and the criteria that had to be met for the induction of associative LTP. Their work indicated that in magnesium-free solutions glutamate opens voltage-independent cation channels, but that in the presence of physiological levels of magnesium the single-channel currents measured at the resting potential are chopped into bursts and the probability of the channels opening is greatly reduced (Fig. 10.6D). The voltage-dependence of the NMDA receptor-linked conductance increase to cations is therefore due to the voltage dependence of the magnesium block. In the case of the induction of LTP, the fact that a threshold number of axons that form synapses on NMDA receptors must be stimulated follows from the requirement that in order for these receptors to allow cations to pass in any quantities, it is first necessary to remove the magnesium block of the channels by depolarisation (Wigstrom & Gustafsson, 1985a, b). As these cations include calcium ions, the open NMDA channel provides a major pathway for increasing the calcium concentration in the postsynaptic cell, thus fulfilling the intracellular calcium requirement for the induction of LTP.
226
HISTORY OF THE SYNAPSE
Elaboration of this scheme also provides an explanation for how the NMDA receptor mediated LTP is associative. Consistent with the above observations, Wigstrom and his colleagues (Wigstrom et al., 1986b) as well as Malinow and Miller (1986) showed that LTP could be induced by pairing single afferent volleys with intracellularly injected depolarising current pulses in the postsynaptic cell (Fig. 10.8A), and that LTP could be blocked by intracellular injection of hyperpolarising currents in the cell during conditioning high-frequency volleys (Fig. 10.8B). These results explain why a strong conditioning input coincident with a weak input leads to potentiation of the response of the weak input: postsynaptic depolarisation induced by the strong input is electrotonically propagated to the site of the weak input so that the summed depolarisation at the weak input is sufficient to remove magnesium ions from the NMDA receptors that have been activated by glutamate there, thus allowing an influx of calcium ions to occur at the weak input leading to an induction of LTP at the previously weak synapses. The question next arises as to why this associative LTP should be confined to just the excited synapses if a rise in intracellular calcium in the postsynaptic cell is to be taken as the necessary and sufficient condition for induction. Why do not all the terminals on the cell, once calcium increases, have LTP induced? One possible explanation for this is that only synapses on the spinous processes of pyramidal cells possess NMDA receptors. In this case the appropriate rise in calcium ion concentration necessary for induction of LTP, which is contingent on the opening of NMDA receptors, could be confined to the spines on which the active terminals form synapses. Evidence that spines are involved in LTP was provided as early as 1977 by Fifkova and van Harreveld who showed that stimulation of the perforant path could induce an increase in the area of the dendritic spines of granules cells in the distal third of the dentate molecular layer. Within a period of one hour after a tetanus, the area of these spines had increased by 38% (Fig. 10.9A), whereas the density of vesicles in the nerve terminals on the spines had decreased by 19% (Fig. 10.9A). There were no changes in the spines or terminals on them for these granule cells in the proximal third of the dentate molecular layer, which does not receive synaptic input from the pyriform pathway. These observations pointed to the possibility that the calcium increase does not occur throughout the granule cell on stimulation of the perforant path but is restricted to just those spines that receive input from the terminals of the perforant axons. This offered an explanation for why associative LTP is confined to just excited synapses whilst being dependent on an increase in the calcium concentration of the postsynaptic cell at the same time. 10.4 The maintenance of associative LTP 10.4.1 Increase in protein synthesis accompanies LTP In 1973 Bliss and Gardner-Medwin showed that LTP lasted for at least several days in the dentate gyrus following suitable stimulation of the pyriform pathway input (see Fig. 10.2D). It was clear that such long periods of maintenance of LTP must involve new protein synthesis initiated by the stimulation protocol used to induce the LTP. However, it was not until 1981 that this was shown to be the case. At that time Duffy et al. (1981) had gathered considerable experience in the use of techniques for the study of protein synthesis in the goldfish brain after training. This had enabled them to show that two specific proteins are rapidly labelled and released into the extracellular fluid after the goldfish acquire a new pattern of behaviour. They used this approach, which involved a double-labelling technique in which the pattern of protein synthesis in a single potentiated slice of rat hippocampus is compared with that of an unpotentiated control slice. This method allowed them to show that LTP of the hippocampal slice in either the dentate gyrus following perforant path stimulation or area CA1 of the hippocampus following Schaffer collateral stimulation, resulted in an increase in the incorporation of labelled valine into proteins destined for secretion into the extracellular medium from just those regions that had been stimulated (Fig. 10.10A). The expected link between LTP in a particular region and metabolic processes that lead to protein synthesis in that region was then established. However, whilst the story concerning the requirements for the induction of associative LTP unfolded in a strikingly exciting way, as did that for the correlation between LTP and protein synthesis, this was not and is still not the case for the mechanisms that are responsible for the maintenance of LTP. These touch on the question of what changes are being induced to provide the increase in synaptic efficacy at synapses that gives rise to the LTP. During the late 1970s and early 1980s all the possible changes that could enhance synaptic efficacy were claimed by one researcher or another to provide the basis for LTP.
THE DISCOVERY OF LONG-TERM POTENTIATION OF TRANSMISSION AT SYNAPSES
227
Fig. 10.8. The effects of polarisation of neurones during stimulation on the generation of LTP. (A) Effects of depolarising currents, (a) Diagram showing the arrangement of stimulating (STIM) and recording (REC) electrodes, (b) Averaged records (n=10) of intracellularly recorded postsynaptic potentials following single volleys. Potentials obtained before conjunction are shown as dotted lines, those obtained 5 min after the end of conjunction as solid lines, (c) Measurements of the initial slopes of the intracellularly recorded EPSPs resulting from activation of STIM I (input 1) and STIM 2 (input 2) are shown for a series of test responses. During the whole recording period shown (except that indicated by a heavy bar) a 6 nA, 100 msec current pulse was given through the intracellular electrode 400 msec after the stimulus to input 1. During the period indicated by the heavy bar, the current pulse commenced 7 msec after the stimulus to input I. Resting membrane potential was −62 mV and input resistance 22 m (from Malinow & Miller, 1986, Fig. 1). (B) Effects of hyperpolarising currents, (a) Diagram of a hippocampal slice showing the location of the bipolar extracellular stimulating electrode (s), the intracellular electrode in the control cell (c), and the intracellular electrode in the test cell (t) (adapted from Ramon y Cajal). (b) The upper traces show an intracellular recording from a control cell (c), which did not receive any current passage during the conditioning high frequency stimulation (HFS). Each trace is the average of five successive responses 4 sec apart recorded immediately before and 10 minutes after the conditioning HFS. Lower traces in b show a simultaneous intracellular recording from a nearby test cell (t) which received passage of a 3-nA hyperpolarizing current during the HFS. Averaged responses were recorded immediately before and 10 min after the HFS. The control cell’s response was potentiated by the HFS whereas the EPSP of the test cell was not (from Wigstrom et al., 1986a, b, Fig. 1).
10.4.2 Increase in the size of dendritic spines One possibility has already been touched upon, namely that there is an increase in the size of dendritic spine (Fig. 10.9A). Fifkova and van Harreveld (1977) conjectured that:
228
HISTORY OF THE SYNAPSE
The swelling of spines will widen the passage from the synaptic region to the dendritic branch. This can have a profound effect on the spread of synaptic current, which is governed by the ratio of spine stalk resistance to the input resistance of the dendritic branch. In instances in which this ratio is close to 1.0 (i.e. when the width and thus the resistance of the spine stalk is matched with that of the dendritic tree), even a small increase in the width of the spine stalk will decrease considerably its resistance, and enhance the electrotonic spread of synaptic current. Consequently the chance that the neuron will discharge becomes greater.’ Subsequently this stimulation-induced spine enlargement was shown to be suppressed by protein synthesis blockers (Fifkova et al. 1982). 10.4.3 Increase in the amount of transmitter released Dolphin et al. (1982) developed a push-pull cannula technique that enabled them to infuse [3H]-glutamate into the dentate gyrus where perforant path axons terminate, and to measure the subsequent release of newly synthesised [3H]-glutamate. At the same time they were able to monitor the field potentials evoked in the dentate gyrus by stimulation of the perforant pathway (Fig. 10.9B). They showed that there was a prolonged increase in the release of glutamate following induction of LTP in this pathway (Fig. 10.9C). This work supported and extended an observation made in 1981 by Skrede and MaltheSorenssen that the release of previously accumulated [3H]-D-aspartate in area CA1 of the hippocampus is increased following tetanic stimulation of afferent fibres to the pyramidal cells in this region (Skrede & Malthe-Sorenssen, 1981a, b). Subsequent work by Bliss and his colleagues established that stimulations that produced LTP enhance the potassium induced release of radiolabelled glutamate from the dentate gyrus of rats (Lynch et al., 1985; Lynch & Bliss, 1986a), and that an increased overflow of both aspartate and glutamate from the dentate gyrus occurs following the induction of LTP (Bliss et al., 1986). Given that the induction of LTP requires an increase in calcium concentration in the postsynaptic cell, any claims that the maintenance phase of LTP involves an increase in transmitter release requires that a retrograde message from the postsynaptic cell to the presynaptic terminal occurs in order to trigger an increase in transmitter release. Bliss and his colleagues set out to identify this message, and soon showed that one likely candidate was arachidonic acid. Weak activation of the perforant pathway, insufficient in itself to induce LTP, was able to do so in the presence of exogenous arachidonic acid (Williams et al., 1989). At the end of the 1980s Malinow and Tsien (1990) supported this conceptual framework due to Bliss and his colleagues, by showing that although LTP seemed to involve an enhanced transmitter release, dialysis of the postsynaptic neurone with a whole cell voltage clamp electrode blocked the induction of LTP, pointing to the requirement that a postsynapticaly generated signal from the cell to the synaptic terminal is necessary before the enhanced transmitter release can occur. 10.4.4 Increase in the density of glutamate receptors Lynch et al. (1982) carried out an experiment in which they stimulated a large number of contiguous groups of axons with currents that produced typical postsynaptic responses in slices of hippocampus CA1 subfield and then measured [3H]glutamate binding in this subfield following high frequency stimulation. They found that an increase in [3H]-glutamate binding occurred that was correlated with the induction of LTP (Fig. 10.11A). This binding was blocked by conditions that inhibited the development of LTP and was absent in slices that did not exhibit LTP (Fig. 10.11A). Scratchard analyses showed that the increased binding was due to an increase in the number and not the affinity of glutamate receptors. The question that next arises is which type of amino-acid receptor is involved in this increase in [3H]-glutamate binding. Davies and his colleagues showed, using iontophoretically applied quisqualate receptor ligands on pyramidal neurones in area CA1 of the hippocampus, that the sensitivity of these neurones to the ligands slowly increased following induction of LTP, being first detected at ca 15 min and increasing for a further hour or so thereafter (Fig. 10.11B; Davies et al., 1989). As LTP is induced within 30 sec of tetanic stimulation these observations pointed to a presynaptic change generating the early maintenance phase of LTP with postsynaptic changes in the quisqualate receptors perhaps implicated in a later stage in the maintenance of LTP.
THE DISCOVERY OF LONG-TERM POTENTIATION OF TRANSMISSION AT SYNAPSES
229
Fig. 10.9. Evidence for long-lasting changes in dendritic spines of granule cells and in the release of glutamate from pyriform pathway axons onto granule cells in LTP. (A) Long-lasting morphological changes in dendritic spines of dentate granule cells following stimulation of the entorhinal area (from Table 2 in Fifkova & van Harreveld, 1977). (B) Observations indicating an increase in the release of glutamate during LTP. Shown is the method of perfusion of the dentate gyrus through a push-pull cannula during stimulation of the perforant pathway axons, (a) Two recording electrodes were glued to the outer (‘pull’) tube of the concentric push-pull cannula. The upper electrode protruded 0.2 mm and the lower electrode 0.5–0.7 mm below the outer cannula. The electrode/cannula assembly was lowered stereotaxically into the dentate gyrus of the hippocampus. The outer cannula was fixed to the skull with dental cement, and the inner (‘push’) cannula was then lowered so that its tip protruded 0.3 mm below the outer cannula. The perforant path (PP) was stimulated through a concentric bipolar stimulating electrode with constant voltage pulses (0.05 msec, range 10–50 V) at a strength near threshold for a population spike (the negative-going spike superimposed on the synaptic wave and reflecting the discharge of granule cells). Throughout the experiment, evoked potentials were recorded every 90 sec on a chart recorder, and on disc for subsequent computer analysis (from Fig. 1 in Dolphin et al., 1982). (C) In vivo release of [3H]-glutamate from the dentate gyrus perfused through a push-pull cannula described in (B). In all experiments 3Hglutamate (20 µCl, 0.75 nmol) was infused in 3 oxygenated (95% O2/5% Co2) perfusion medium (composition in mM: Na+. 152; K+ 3.7; Mg2+, 1.2; Ca2+, 1.8; Cl−, 140; H2PO4, 1.2; SO4−2, 1.2; HCO3, 16; glucose, 10) containing a trace of Phenol Red, at a rate of 0.1 ml min−1. Perfusion with continuously oxygenated medium at 10 ml min−1 was begun 1 h later, using a dual-channel peristaltic pump. Sequential 7.5min fractions of perfusate were collected on dry ice. The phenol red served to mark the entry of the perfusate into the fraction collector, and thus the lag time of the apparatus (8 min). [3H]-glutamate was separated from [3H]− -aminobutyric acid and from their precursor 3Hglutamine using anion-exchange columns. The [3H]-glutamate released in a single experiment was expressed for each fraction as a percentage of the release in the fraction ending at 120 min. Values for each group of animals are given as mean ± 1 SEM for all fractions between 90 and 195 min following the start of perfusion, (a) Control perfusion, PP stimulated continuously at 0.1 Hz (n=7). Values for control [3H]-glutamate release are repeated in b, c and d (dashed lines) where * indicates significant increase in [3H]-glutamate release compared with control (p < 0.5, Student’s t-test). (b) PP stimulated at 0.1 Hz until 120 min after start of perfusion when the rate was increased to 3 Hz and the stimulus strength was doubled for the remainder of the experiment as indicated by the solid bar (n=7). (c) and (d) PP stimulated continuously at 0.1 Hz, and at 120 min after the start of perfusion (arrow) a conditioning train was given, (c) [3H]-Glutamate release in animals not showing LTP (n=6). (d) 3H-glutamate release in animals showing LTP (n=11) (from Fig. 2 in Dolphin et al., 1982).
230
HISTORY OF THE SYNAPSE
10.5 Biochemical pathways implicated in the maintenance phase of associative LTP Protein kinases in the maintenance of LTP In 1980 Bar and his colleagues showed that after stimulation of the perforant pathway in hippocampal slices sufficient to initiate LTP there was an enhanced incorporation of [32P] phosphate into a protein band with a molecular weight of about 50K (Fig. 10.12A; Bar et al., 1980); if the stimulation parameters used were such that LTP was not induced then no enhanced incorporation of [32P] phosphate was observed. These authors suggested that ‘electrical stimulation enhances the activity of the protein kinase that phosphorylates the 50K protein’. However, it was not until 1987 that associative LTP in the hippocampus was shown to last for only one or two hours following extracellular application of a protein kinase inhibitor (Lovinger et al., 1987; see also Reymann et al., 1988; Malinow et al., 1988). As this time is similar to that of the quisqualateinduced response described above, the scheme that suggested itself to these authors was that protein kinases are involved in the initiation of the slowly developing postsynaptic change in the quisqualate receptors responsible for a late phase of LTP maintenance. The early phase in the first hour or so after induction might then be due to an increase in transmitter release that is not dependent on protein kinases, as described by Dolphin et al. (1982). However it was shown in 1989 that intracellular injection into CA1 pyramidal cells of the protein kinase inhibitor H-7 blocks the induction of LTP (Fig. 10.13A; Malinow et al., 1989; see also Malenka et al., 1989). As H-7 when injected intracellularly did not block the maintenance of LTP after its induction, but as noted above extracellular H-7 does, then a presynaptic protein kinase seems to be involved in the maintenance phase whilst a postsynaptic protein kinase is involved in the induction phase. 10.5.2 Protein kinase C and calcium/ calmodulin-dependent protein kinase II in the induction and maintenance of LTP The question that next arises is what particular protein kinases are involved in LTP? Lynch and Bliss (1986b) had shown that calmodulin could greatly increase the release of glutamate from synaptosomes that had been prepared from hippocampus CA3 regions previously subjected to LTP. Routtenberg et al., (1986) as well as Malenka et al. (1986) showed that phorbol esters that are know to activate protein kinase C (PKC) enhance synaptic transmission at synapses in the hippocampus in a manner that is similar to that of LTP. Furthermore the Routtenberg group were able to show in the same year that PKC activity was elevated two fold in membranes of neurones from regions of the hippocampus that had been subjected to LTP (Akers et al., 1986). Hu et al., (1986) showed that direct injection of PKC into pyramidal cells in area CA1 gives rise to an LTP-like potentiation of synaptic transmission onto the injected cells. It seems likely that both protein kinase C (PKC) and calcium/calmodulin-dependent protein kinase II (CaMKII) in the postsynaptic cell are necessary for the induction of LTP. The induction of LTP is blocked by injection of specific inhibitors of PKC (such as PKC-(19–31), a conserved region of the regulatory domain of the PKC family which is a pseudo substrate antagonist of PKC activity; Malinow et al., (1989), or injection of a synthetic peptide containing the autoinhibitory domain of CaMKII and CBP, a synthetic CaM-binding peptide (the sequence of which is based on the CaM-binding domain of CaMKII; see Fig. 10.13 (B); Malenka et al., 1989). The involvement of postsynaptic CaMKII lends itself to the simplest hypothesis, namely that the influx of calcium ions through NMDA receptor channels activates CaMKII, which then modifies quisqualte receptors so that there is an enhancement of the postsynaptic response to glutamate, thus increasing the size of the excitatory postsynaptic potentials. 10.6 Evidence that associative LTP is involved in memory The question arises as to whether all of this research on the mechanism of associative LTP in the hippocampus can be regarded as elucidating an essential component in the establishment of associative memory in animals? Although correlations had been found as early as 1979 between the degree of senescence of an animal and the extent to which LTP could be generated in the hippocampus of these animals (see Fig. 10.4A), these observations did not establish a causal relationship between the two. The first attempt to test whether LTP was required for learning was made by Morris et al., (1986). They took advantage of the then recent discovery that amino-phosphonovaleric acid (AP5) could block associative LTP by blocking NMDA receptors (see Fig. 10.6). In their experiment a chronic intraventricular infusion of D, L-AP5 caused a selective impairment of place learning, a process thought to be dependent on the hippocampus (O’Keefe & Nagel, 1978), whereas the L-isomer of AP5 did not (Fig. 10.14A). As there was no impairment in visual discrimination tests of the rats, a process that is not dependent on LTP, the results went some way to establishing a link between LTP and learning. However at this time the
THE DISCOVERY OF LONG-TERM POTENTIATION OF TRANSMISSION AT SYNAPSES
231
Fig. 10.10. Enhanced protein synthesis accompanies LTP. Table A. Enhanced secretion of labelled proteins in the hippocampus and dentate gyrus accompanies LTP. After potentiation, each slice was incubated for 2 h before being labelled for 3 h. In all experiments except experiment 6, [3H]valine (sp. act., 5 Cl/ mmol-1) was used for the potentiated slice, and [14C]valine (sp. act., 280 mCi/mmol-1) was used for controls. In experiment 6, the labels were reversed. RECF is the ratio of labelled proteins in the extracellular fluid of a potentiated slice to that in a control unstimulated slice; Rcyto is the ratio of labelled proteins in he cytoplasmic fraction of a potentiated slice to that in a control slice after removal of the extracellular fluid fractions; H and D refer to the hippocampal and dentate regions, respectively. Comparisons with a t-test of experimental vs control data give p< 0.0005. Table B. Labelling of cytoplasmic fractions of potentiated vs unpotentiated regions of the same slice. RH is the 3H/14C ratio of the labelled cytoplasmic proteins for the hippocampal fragment of a potentiated (3H) to a control (14C) slice. RD is the ratio of he labelled cytoplasmic protein for the remaining dentate segments of the same two slices. H and D denote the potentiation locus at Schaffer collaterals to CA1 and perforant path to dentate, respectively. Comparisons of the data with the controls in Table A give p<0.005. Rp and Rc are the ratios for the potentiated and control regions of the slices, respectively (from Tables 1 and 2 in Duffy et al., 1981).
possibility was not tested that the NMDA receptor antagonist might be blocking some part of the motor pathway that is uniquely utilised in the learning process. Some complexity in the interpretation of these kind of experiments was illustrated by the work of Mondadori et al. (1989). They showed that NMDA receptor antagonists modify learning in a task-dependent manner: such drugs administered prior to training resulted in impaired learning performance in place-navigation and darkavoidance paradigms, but improved performance in step-down passive avoidance tasks. Alternative approaches to that of drug intervention in the search for evidence that LTP is required for learning and memory involved stimulation of the pyriform pathway to the dentate gyrus in vivo. This was done in such a way as to saturate the inducible LTP in that pathway, with subsequent testing of the animals on a spatial learning task. This experiment was carried
232
HISTORY OF THE SYNAPSE
Fig. 10.11. Evidence that an increase in the density of glutamate receptors accompanies LTP. (A) Changes in 3H-labelled glutamate binding to hippocampal membranes elicited by high-frequency stimulation in CA1 ‘mini slices’. Following 15 min of high frequency stimulation to 24–40 sites on a CA1 mini slice, crude synaptic membranes were prepared and [3H] glutamate binding was measured at a glutamate concentration of 100 nM. A control minislice was removed at the same time as the stimulated minislice. Data were collected from 38 stimulated and 38 control slices. Robust and poor LTP refer to the extent of synaptic facilitation produced in two subgroups (n=13 each) of stimulated minislices. The control values are for the groups (n=13 each) of slices removed and processed at the same time as the slices exhibiting robust or poor LTP. The asterisk indicates p < 0.02 two-tailed t-test (from Fig. 1 in Lynch & Baudry, 1984). (B) LTP is associated with a delayed increase in sensitivity of single CA1 neurones to quisqualate. a. Current (top) and voltage (bottom) records from an intracellularly recorded CA1 pyramidal neurone. Depolarizations are in response to ionophoretic ejection of quisqualate (130 nA, 2 sec) into stratum radiatum. These are preceded by hyperpolarizations in response to constant current pulses (0.5 nA, 400 msec), used to monitor resting input resistance. Panels illustrate from left to right, control records, sections taken immediately after, and sections taken 30 min after the tetanus (100 Hz, 1 sec, test intensity). Two of the last four quisqualate-induced responses depolarize the cell sufficiently to fire action potentials, b. Plot showing the amplitude of quisqualate-induced depolarizations for 15 min before and 30 min following the tetanus (arrow). Thereafter (for a further 45 min) quisqualate-induced depolarizations excited the cell sufficiently to elicit firing, c. Synaptic responses taken immediately before and 30 min after the tetanus. In these experiments intracellular recordings were obtained from the cellbody layer using electrodes containing 3M potassium acetate and the Schaffer collateral-commissural pathway was stimulated at 30 sec intervals by a single electrode placed in stratum radiatum. An ionophoretic pipette was positioned in stratum radiatum at the level of the stimulated fibres, adjacent to the recording site. Quisqualate (20 mM in 150 mM NaCl) was applied ionophoretically (46–163 nA, 1–2 sec) 5 sec after each synaptic response was elicited. The position of the ionophoretic pipette was adjusted to achieve a compromise between short latency response (indicating that the pipette was close to a dendritic branch of the recorded cell) and the maintenance of stable recordings. Only those cells where passive membrane properties and both synaptic and agonist-induced responses remained stable for 15 min before tetanic stimulation (100 Hz 1 sec, test intensity) and where healthy impalements could be maintained for at least a further 30 min were included in the analysis (from Fig. 2 in Davies et al., 1989).
out by McNaughton et al. (1986): spatial learning was severally affected by this procedure. In a follow-up experiment Castro et al. (1989) showed that the learning capacity recovered at about the same rate as it took for the induced and saturated LTP to decay (Fig. 10.14C). The results indicate that spatial memory involves at least temporary storage in the fascia dentata through the mechanism of LTP.
THE DISCOVERY OF LONG-TERM POTENTIATION OF TRANSMISSION AT SYNAPSES
233
Fig. 10.12. Protein kinases are involved in the maintenance of LTP. The effect of tetanic stimulation of rat hippocampal slices on post hoc endogenous phosphorylation. After the phosphorylation assay, proteins are separated according to their molecular weight and autoradiography is performed to visualise radiophosphate incorporation into protein bands. Absorption patterns (500 nm) are shown, one of an untreated slice (---) and one of a stimulated slice (___, 15 pulses/sec-1, 15 sec). Arrows indicate the increased absorption of the band with MW=50,000, as a result of an enhanced [32P] phosphate incorporation into that band. On the x axis Mw are indicated of marker proteins, the y axis gives the absorbance in arbitrary units (from Fig. 2 in Bar et al. 1980).
Another approach, that does not involve the administration of drugs, is to see if a particular learning procedure is correlated with the development of LTP in that part of the brain in which the learning is carried out. Adopting this procedure, Roman et al. (1987) examined changes in synaptic efficacy between the LOT and layer I of the piriform cortex following training of rats on two-odor discriminations over a period of several days. Comparison between the monosynaptic responses in the piriform cortex due to stimulation with single pulses of the LOT showed that the population synaptic responses were greatly potentiated following training, for a period of at least 24 hours (Fig. 10.14B). A correlation was then discerned between learning in this pathway and the establishment of LTP, although evidence that the former is dependent on the latter is of course not revealed by this type of experiment. By the end of the 1980s it could be said that circum-stantial evidence existed for a role of associative LTP in learning. 10.7 The induction of non-associative LTP Although properties of NMDA receptors provided the conceptual framework for understanding the induction of associative LTP, it was soon shown that several different synapses at which LTP could be induced did not seem to possess many if any NMDA receptors. This was the case, for example, with the synapses formed by mossy fibres on the pyramidal cells in area CA3 of the hippocampus (Monaghan et al., 1983). NMDA receptor antagonists have no affect on the induction of LTP at these synapses (Harris & Cotman, 1986; Fig. 10.6C). The same is the case for LTP at synapses on sympathetic ganglion cells in the superior cervical ganglion (Fig. 10.15 A; Brown & McAfee, 1982) as well as at the crayfish neuromuscular junction (Fig. 10.16; Baxter et al., 1985). The fact that there is no associative LTP at these synapses fits neatly with the proposition that this requires an influx of calcium ions into the postsynaptic cell, which is mediated by receptors with the properties of the NMDA receptor. Such an influx of calcium ions is not required for LTP induction at these non-associative synapses although they all require an elevation of calcium in the nerve terminal for an elevated transmitter release to occur (Briggs et al., 1985). 10.8 Summary and Conclusion The way in which the mechanism of induction of associative LTP was elucidated following the discovery of LTP in the 1960s makes a very satisfactory story. It was most fortunate that work involving recognition that the NMDA receptor mediates excitatory amino-acid transmission together with discovery of the role of magnesium ions in the function of this receptor should have occurred just at the time when attempts were being made to understand the induction of associative LTP (see Chapter 5). It is unfortunate, however, that another happy juxtaposition of events did not also occur in the 1990s to help clarify whether this induction leads to presynaptic or postsynaptic changes at the synapse, or both. Nor is it clear that LTP is involved in the establishment of spatial memory systems in the hippocampus, the study of which has been pioneered in the beautiful work of
234
HISTORY OF THE SYNAPSE
Fig. 10.13. PKC and CAKII are involved in LTP. (A) Protein kinase inhibitor H-7 in the postsynaptic cell prevents LTP. a–e. Maximal slope of rising phase of EPSPs as a function of time, given as ensemble averages of time-registered data from several slices. Tetanic stimulation is indicated by an upward arrow. Error bars indicate SEMs for representative individual time points. (Inserts) Representative records of EPSPs (average of ten consecutive synaptic potentials) obtained at the time denoted by lower-case letter in graph. Simultaneous (a) extracellular and (b) intracellular recording of EPSPs show comparable percentage increases in synaptic transmission after tetanic transmission (average of 11 slices), c–e. As in a. and b., but with H-7 in the intracellular electrode (n=11). Intracellular recording with H-7-containing electrode shows no LTP d., despite LTP measured extracellularly in nearby cells c. e. Synaptic transmission in a nontetanized pathway, independent of the tetanized pathway, monitored with the same H-7-containing intracellular electrode (diagram). EPSP slope shows no significant change during course of experiment, indicating that intracellular H-7 does not cause a general rundown of transmission, and that the quality of recording is maintained (from Fig. 1 in Malinow et al., 1989). (B) Effect of intracellular application of the calcium/calmodulin (CaM)-binding peptide (CBP) on LTP. a. The magnitude of the initial EPSP slope in populations of cells recorded with microelectrodes containing either ( ) CBP (190 µM: n=11) or ( ) the control peptide CTP2 (190 H.M; n=8). b. The initial field EPSP slope recorded in the two populations of slices demonstrating that the LTP was essentially identical in the two populations, c. The peptide, CBP−3, which inhibits CaM activity but does not block Ca/CaM-independent CaM-KII substrate phosphorylation, has the same ability to block LTP ( ) as does CBP ( ). Microelectrodes contained 190 µM CBP−3 (n=6). The data for the CBP-filled cells are the same as shown in a. d. LTP generated in the field potential is very similar for the two populations of slices (from Fig. 2 in Malenka et al., 1989).
O’Keefe and his colleagues (O’Keefe, 1976; O’Keefe & Nagel, 1978). Nevertheless, it should not be long before these problems are also resolved, given the continuing vigour with which the study of the hippocampus is at present being persued.
THE DISCOVERY OF LONG-TERM POTENTIATION OF TRANSMISSION AT SYNAPSES
235
Fig. 10.14. Evidence that associative LTP is involved in memory. (A) Blocking NMDA receptors blocks spatial learning in a Morris water tank. Swimming path taken in a Morris water tank by the rat closest to the group mean (calculated according to distribution of time spent per quadrant) is shown in the Control, D.L-AP5 (D, L isomer of the NMDa receptor antagonist AP5) and L-AP5 (L isomer of AP5 which is not an antagonist of NMDA receptors) groups, respectively. For each animal the computerized tracking system calculated the times spent in the four ordinal quadrants of the pool during the 60 sec transfer test. The tank is seen in plan view in the diagrams. The data were organized into time spent in the original training quadrant (Train), the opposite (Opp) and the two adjacent quadrants (Adj/L, Adj/R). In the original training, an escape platform is placed in the training quadrant (Train) and rats develop a spatial map of the location of the platform. The swimming path is then measured in the absence of the platform in subsequent trials. An analysis of the group scores, plotted beneath each representative animal, revealed a highly significant greater increase of time spent by the control and L-AP5 treated rats than for the D. L-AP5 treated animals. The mean times spent ± one SE in the training quadrant were control=35.4 ± 2.2 sec; (59% of the 60 sec test where chance=25%, n=20); D. L-AP5=18.2 ± 2.0 s (30%); and L-AP5=37.3 ± 4.8 (62%) (from Fig. 2 in Morris et al., 1986). (B) Evidence that there is a change in synaptic efficacy between the LOT and layer I of the pyriform cortex following training of rats in twoodour discriminations over 24 hours. Changes in the slope of the evoked potential immediately after and 24 hr after a period of patterned stimulation. Rats were implanted bilaterally with stimulating and recording electrodes. One day after obtaining baseline values for the slope of evoked potentials, patterned stimulation was delivered unilaterally (positive cue) and the slope of the evoked potential was measured 0.5 and 24 hr later. The following day, patterned stimulation was also applied to the contralateral electrode (negative cue) and the slope of the evoked potential was measured 0.5 and 24 hr later. The percent change was calculated for each rat using its own baseline value and averaged for the group of rats (vertical bars represent SEM **p<0.02; *p<0.05; Mann-Whitney U-test comparing percent changes from baseline values for the positive and negative electrodes with those for the control electrodes (n=5) (from Fig. 4 in Roman et al., 1987). (C) Correlation between recovery from saturating LTP in the dentate gyrus and the behavoural return of spatial memory. A typical search pattern during the ‘probe’ trial of a rat that had previously learned the spatial location of the escape platform in the Morris water tank. During the probe trial, the platform was removed from the pool, and the animal was placed into the water and allowed to search the pool for 60 sec. For purposes of analysis, the apparatus was divided into four quadrants. T, Training quadrant that contained the escape platform; A, adjacent quadrant; and O, opposite quadrant. The amount of time each animal spent in each quadrant was recorded. b. and c. Mean percentage of time spent in each of the four quadrants for the low frequency stimulation of the perforant pathway (LF; not sufficient to give LTP) and high frequency stimulation (HF; sufficine to give LTP) groups on day 19 following experimentally induced saturation of LTP in the fascia dentate of the HF group. A one-way analysis of variance of quadrant search time for the LF group was significant, (F(3, 9)=13.3 p<0.01). Post hoc analysis (Newman-Keuls, p<0.01) showed that the animals in the LF group spent significantly more time in the training quadrant compared with any of the other quadrants, d-f. The mean percentage of the time spent in the four quadrants for the HF behavioural naive group, the LF control group, and the HF group that was previously trained on day 19. On this day, the platform was moved to the location diagonally opposite its day 19 location. These data show that the behavioural impairment associated with LTP saturation recovers with the decay of LTP. Separate one-way analyses of variance on these data were significant, all F values ((3, 9)>8.3, p values <0.01), and post hoc analyses (Newman-Keuls, p<0.01) showed that for all groups all the rats spent more time in the training quadrant than in any of the other quadrants (from Fig. 2 in Castro et al., 1989).
236
HISTORY OF THE SYNAPSE
Fig. 10.15. Non-associative LTP in autonomic ganglia. (A) The stimulating and recording arrangement. The internal carotid nerve (right) is in a recording suction electrode, and the preganglionic nerve (left) to the superior cervical ganglion is in the stimulating suction electrode. The arrows indicate direction of suction applied to the stimulating or recording electrode). (B) Time course of the decay of post-tetanic enhanced transmission after 10–20 sec of 20-Hz stimulation. The four plots represent data from different ganglia. Note that in ganglion 1 and ganglion 2 there is <50% diminution of the initial response over one hour after stimulation. This indicates a form of LTP as the potentiation has lasted more than an order of magnitude longer than LTP. For convenience, data points taken at times >60 minutes are not illustrated. In general, the data analysis was discontinued when the responses declined to about 10% above control levels [l(t)=0.1] because of the increased coefficient of variation. (Compare the variability in the data points associated with curves 1 and 4 at t=50 min). Inset, representative postganglionic recordings taken from curve 3 at the times in minutes, (relative to the end of tetanic stimulation) indicated (from Fig. 1 in Brown & McAfee, 1982).
THE DISCOVERY OF LONG-TERM POTENTIATION OF TRANSMISSION AT SYNAPSES
Fig. 10.16. LTP in the crayfish opener nerve-muscle preparation. (A) Averages of 32 successive EPSP waveforms obtained at the indicated portions of the experiment. (B) Plot of data from the same cell. Each datum point represents the average of 32 EPSP waveforms. Note that the increased efficacy of transmission over the control lasts for >4 hours after initial stimulation (from Fig. 1 in Baxter et al., 1985).
237
11 Emergence of the Concept of Synapse Formation Molecules
11.1 Introduction At the beginning of the twentieth century J.N.Langley and F.Tello made a series of studies on the reinnervation of denervated muscle and autonomic ganglia that were to establish principles of synapse formation that were not capitalised on for over 50 years. Tello (1907) showed that following muscle denervation by cutting a nerve, motor axons often grow down the empty Schwann cell tubes to enter the general region of the denervated synaptic sites of the muscle (Fig. 11.1A). Here the axons frequently branched profusely out of the enveloping Schwann tube to innervate the synaptic sites delineated by their characteristic endplate structure of a raised cytoplasm and concentration of nuclei (Fig. 11.1B). This indicated for the first time that although the branching axon had access to a large amount of muscle membrane it favoured terminal differentiation at the original synaptic site, suggesting that this site has a particular capacity for triggering the formation of terminals. Subsequently, Elsberg (1917) showed that adult muscle cells that possess an intact innervation resist additional innervation from another axon, emphasising that terminal formation appears to be restricted to synaptic sites that are not already occupied by synaptic terminals. The possibility that reinnervating axons may at least in some cases favour forming terminals on particular cells according to the segmental origins of the axons was raised by the experiments of Langley (1895). He showed that if the superior cervical ganglion was denervated by cutting the cervical sympathetic nerve trunk, subsequent reinnervation of the ganglion by each of the thoracic segmental nerves that contribute to the trunk is normal. In this case each axon comes to form synapses on the same ganglion cells that it previously innervated (Fig. 11.1C), providing evidence for the specificity of synapse formation, and leading Langley to conjecture that ‘at bottom then the phenomenon is a chemiotactic one’. At the beginning of the twentieth century, then, experiments had already been carried out indicating that the postsynaptic site contains information that can trigger terminal formation and that such sites may even distinguish between axons of different segmental origin. In the subsequent 50 years the tremendous stimulus to further research that might have been expected from the conceptual framework put in place by these experiments was not forthcoming. This chapter traces the historical development of how the work of Tello and Langley was taken up again in the latter half of the century, and of the progress made in identifying the molecules involved in the ‘chemiotatic’ process of terminal and synaptic site formation. 11.2 Synapse formation in muscle 11.2.1 Site of synapse formation in reinnervated adult muscle Following Tello’s work in 1907, the proposition that reinnervating motor-nerve axons form terminals at the vacated synaptic sites on the muscle was lost as a consequence of confusion arising from the way in which the muscle nerve was lesioned and then placed in proximity to the muscle. Gutmann and Young (1944) agreed that terminals do form at the vacated synaptic sites if the reinnervating nerve fibres are led back by the Schwann cell tubes but in the absence of this guidance ‘fibres may be led into contact with the sarcoplasm at a muscle fibre, where they cause the production of a new endplate’ (Fig. 11.2A). Thus, if these nerve fibres ‘come into contact with the surface of the sarcoplasm they can cause differentiation of the latter to produce an end-plate’. Likewise, Aitken (1950) comments that ‘as soon as a small branch came into very close contact with the muscle fibre, the nerve divided into numerous fine branches forming an endplate’ (Fig. 11.2B). The conclusion here is that there is no particular information peculiar to the vacated endplate for triggering terminal formation, as it occurs at random sites over the surface of muscle fibres.
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
239
Fig. 11.1. The discovery that synaptic sites on muscle cells are a region of preferred reinnervation and that autonomic ganglion cells show a selectivity for the terminals of particular preganglionic neurones during reinnervation. (A) and (B) are experiments of Tello establishing that the synaptic site on a muscle cell laid down during development is preferentially reinnervated following a lesion to the muscle nerve. (A) Muscle cells at different stages of reinnervation. B; Nerve fibre innervating several muscle cells; C; complicated terminal on a muscle cell which has conserved the original endplate accumulation of nuclei; D and H; muscle cells with nerve endings of different degrees of complexity; E; muscle cell innervated by a bouton; F; a collateral of a nerve fibre innervates a muscle cell and then passes on as shown (I). (B) Three reinnervated muscle cells are shown. A and B show muscle cells innervated by the same nerve fibre. C is a muscle cell that is reinnervated by the nerve fibre. D also produces a collateral that grow in the retrograde direction; H; a collateral nerve branch which grows into an empty nerve sheath (I). (A) and (B) from Figs. 14 and 15 in Tello, 1907). (C) Experiments by Langley showing that there is information in autonomic ganglia that establishes correct synapse formation after lesion of a nerve. The table shows the extent of specific synapse formation by preganglionic nerves reinnervating the superior cervical ganglion in the cat after cutting the cervical sympathetic. The effects of stimulating different spinal nerves with therefore different preganglionics on the response of the nictating membrane, dilation of the pupil, contraction of ear arteries, erection of hairs and secretion for the reinnervated ganglion (that on the right) and the normal contralateral control are shown. It will be noted that after reinnervation the effects of the different spinal cord outflows on the different end organs is almost normal. (From Table in Langley, 1895).
In the early 1970s experiments were carried out to see if definitive evidence could be obtained one way or the other on the issue as to whether synaptic sites contain information for triggering motor-synaptic terminal formation. A combination of techniques was used to ascer-tain the sites of synaptic terminal formation. These included cholinesterase staining of the endplates (Fig. 11.3A, B), silver staining of the nerves (Fig. 11.3E) and ultrastructural identification of the synaptic terminals (Fig. 11.3F). With this approach it was shown that if the axons of motoneurones to either mammalian or amphibian fast or slow (twitch) myofibres, which normally possess a single ‘en plaque’ terminal, are cut just outside the muscle and allowed to spontaneously reinnervate the muscle, the axons eventually penetrate the epimysium of the muscle. They then grow through the muscle along the surfaces of myofibres and blood vessels, both in the direction of the old synaptic sites as well as towards the tendinous insertions of the myofibres (Fig. 11.3CB). These axons only form synaptic terminals at the old synaptic sites (11.3CB), that is, over the still remaining postsynaptic folds (Bennett et al., 1973a, b). Furthermore the synaptic sites receive a polyneuronal innervation at this time (Fig. 11.4D see also Fig. 11.4C; McArdle, 1975). Observations of this kind also hold for the reinnervation of muscle fibres that normally possess several synaptic sites. For example, if the axons of motoneurones to amphibian fast or slow (twitch) myofibres, each of which normally possess two or three ‘en plaque’ synapses, are cut and allowed to spontaneously reinnervate the muscle, they grow into the muscle and only form synapses at the two or three original synaptic sites (Miledi, 1960; Landmesser, 1972; Bennett & Pettigrew, 1975). In addition, if the axons of motoneurones to avian slow (graded) myofibres, which normally possess ‘en grappe’ synaptic terminals located at intervals of about 1000 mm along their length, are cut and allowed to spontaneously reinnervate the muscle, the normal pattern of synaptic terminals on the myofibres is reconstituted (Bennett et al., 1973c). Thus it was found that during spontaneous reinnervation of myofibres synaptic terminals form uniquely at the old synaptic sites. It appears, then, that the original synaptic sites are regions of preferred innervation for the reinnervating motoneurones. Another test of this proposition was made by using the axons of motoneurones to amphibian fast or slow (twitch) myofibres that normally possess a single ‘en plaque’ synapse and crossing these with the axons of motoneurones to twitch myofibres that normally possess two or three ‘en plaque’ synapses. If this cross-reinnervation is allowed to proceed spontaneously, then
240
HISTORY OF THE SYNAPSE
Fig. 11.2. Confusion on the issue that the synaptic site on mature muscle cells is a region of preferred reinnervation following a lesion to the muscle nerve. (A) Nerve fibres 30 days after their arrival in a muscle after suture. Three Schwann tubes are shown. That on the left contains a large and a small fibre, but the latter runs past the endplate. The motor fibre sends branches into the plate and then passes as an ultraterminal fibre along the muscle and back up along another tube to end in complex branches on another muscle fibre (from Text—Fig. 7 in Gutmann & Young, 1944). (B) Drawing of a group of large irregular motor endings which are derived from one large nerve trunk. Four muscle fibres (a, b, c, d) are each innervated by two or more endplates (1, 2, 3 etc.). cap., capillary (from Aitken, 1950). (C) Table summarising the extent to which ectopic synapses are formed outside of the original synaptic site on muscle cells following reinnervation that involves spontaneous growth of the nerve into the muscle versus that due to implanting the severed nerve on the muscle. It will be noted that this summary of the experiments performed over more than 20 years shows that ectopics do not form if implanting the nerve is avoided (from Table 2 in Bennett & Pettigrew, 1976).
synaptic terminals only form at the original synaptic sites on the myofibres (Bennett & Pettigrew, 1975). Furthermore, if the axons of amphibian or mammalian preganglionic neurones are used to cross-reinnervate fast or slow (twitch) myofibres, synaptic terminals only form at the old synaptic sites (Landmesser 1971; Bennett et al., 1973b). Finally, if the axons of motoneurones to avian fast (twitch) myofibres that normally possess a single ‘en plaque’ synapse are crossed with those to slow (graded) myofibres that possess an extensive distribution of ‘en grappe’ synapses, then the myofibres that are crossreinnervated possess their normal innervation pattern (Bennett et al., 1973c). Taken together, all these experiments showed unequivocally that the original synaptic sites contain information for the triggering of synaptic nerve terminal formation.
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
241
Fig. 11.3. Definitive evidence that the synaptic site on mature muscle cells is a position of preferred reinnervation, using electrophysiological, histological and ultrastructural techniques. (A) Map of a spontaneously reinnervated rabbit hemidiaphragm seven weeks after section of the phrenic nerve indicating the extent of reinnervation. The dashed line indicates the position of the original nerve, the continuous line the visible branching of the regenerated nerve. The stippled areas are the sections of the muscle removed for sampling for electron microscopy and therefore not examined electrophysiologically, The numbered regions indicate the stages of reinnervation of the muscle. ‘1’ represents late reinnervation; most of the fibres contracted in response to nerve stimulation, a normal population of mepps was found in the middle of the muscle fibres, and nerves associated with ChE deposits were demonstrated histologically. ‘2’ represents early reinnervation; no contraction was seen, subthreshold epsps and mepps of both normal and prolonged time course were recorded, and nerves were present associated with deposits of ChE. ‘3’ represents preservation, no evoked responses were recorded but in some cells low frequency mepps of prolonged time course were seen, and nerves were present although only some small deposits of ChE are stained. ‘4’ represents denervated muscle without electrical evidence of synapses or histological demonstration of nerves or ChE (from Text—Fig.1 in Bennett et al.,1973). (B) The position of endplates in control and reinnervated diaphragms. Histogram of numbers of endplates in sections of control (upper) and spontaneously reinnervated (lower) rabbit hemidiaphragms, measured at different positions along the length of the muscle fibres between the central tendon and the insertion on the ribs. Following spontaneous reinnervation, the endplates are found in the same position on the muscle fibres as they are in control muscles (From Text—Fig. 2 in Bennett et al., 1973a, b). (C) Diagrammatic representation of the pattern of innervation by fast (twitch) motoneurones of control muscles in A and of operated muscles where there is good matching in B or poor matching in D between the innervating axons and the muscle (after combined Ag-ChE staining of muscle sections used in this study). In each diagram, the myofibres are orientated vertically, and the small round dark spots represent the sites of the synapses in the muscle. Dashed lines represent portions of small blood vessels within the muscles. A, Control muscle. The nerve enters the muscle near the end-plate zone and, after reaching the end plates, divides into two branches (each of 30–40 axons), which traverse the muscle at right angles to the longitudinal axis of the myofibres; these branches then give rise to small groups of axons (2–5 axons), each of which directly innervates the synaptic sites at the end-plate zone. B, Reinnervated muscle or cross-reinnervated muscle where there is good matching between the motoneurone type and the muscle. When the nerve enters the muscle away from and proximal to the normal point of nerve entry, it divides into nerve trunks (<15 axons), which eventually reach the original end-plate zone by following blood vessels and the surface of myofibres down the long axis of the muscle. The trunks then divide into small bundles (2–3 axons), which can be observed throughout the end-plate zone, and which finally give rise to single axons that innervate the denervated synaptic sites ‘en passage’ or by collateral sprouting. Single axons can sometimes be observed coursing along blood vessels and myofibres, and occasionally ectopic synapses are formed near the point of nerve entry if the nerve is implanted into the muscle. No axons can be observed in the distal half of the muscle. C, Attempted hyperinnervation of muscle. When the additional nerve enters the muscle, it breaks up into small nerve bundles (3–5 axons) and single axons, which grow along blood vessels, connective tissue structures and sometimes along myofibres; the axons can be observed throughout the muscles, but no synapses are formed by these axons except occasionally near the point of nerve implantation. The innervation of the endplate zone is as described for the control muscle. (D) A cross-reinnervated muscle where there is poor matching between motoneurones and the muscle. As in (C), the foreign nerve breaks up into small nerve bundles and single axons at its point of entry into the muscle, and these spread throughout the muscle without any preferred orientation by following blood vessels and connective tissue elements. Denervated synaptic sites in the original end-plate (small dots) are only rarely innervated, and very occasionally synapses are formed at the site of nerve implantation (from Fig 1 in Bennett & Pettigrew, 1976). (D) Number and distribution of synapses in Ag-ChE-stained sections of a fast (twitch) muscle (extensor digitorum longus) after different experimental procedures. A, control; B, spontaneous reinnervation by the extensor digitorum longus nerve; C, innervation by the implanted tibialis anterior nerve; D, attempted hyperinnervation by the implanted tibialis anterior nerve; E, spontaneous competitive reinnervation by the tibialis anterior and extensor digitorum longus nerves. Maps of the distribution of ChE deposits in representative muscle sections (upper portion of the figure) were drawn from projections through an inverted microscope (arrows in C and D indicate the point of tibialis anterior nerve entry), and the longitudinal axis of the myofibres runs at an angle between the tendinous insertions on the left- and right-hand sides of each map. The vertical line of dots, representing ChE deposits in the middle of the control muscle section, indicates the location of the synaptic sites in the normal end-plate zone of the muscle. The histograms below each map show the pooled data for the number of synapses normalised to the peak of the control distribution along the normalised length of the myofibres for each class of experiment. In experimental situations B, C and E, the end-plate zone has been extensively reinnervated, and in all experiments very few, if any, ectopic synapses are formed, these usually being confined to the vicinity of nerve implants (from Fig. 2 in Bennett & Pettigrew, 1976). (E) The arrangement of nerves with respect to endplates during reinnervation. A spontaneously reinnervated rat tibialis anterior muscle is shown in which groups of endplates are innervated by the collateral sprouts of a few axons. Combined silver and cholinesterase staining, with calibration of 50 µm (from Plate 2b in Bennett et al., 1973a).
242
HISTORY OF THE SYNAPSE
The question arises then as to how the confusion arose during the 60 or so years following Tello’s experiments as to the uniqueness of the denervated synaptic site for triggering synaptic terminal formation. This can be elucidated by reference to the Table in Fig. 11.2C. It will be noted that if axons are cut and then implanted in the muscle, instead of being allowed to spontaneously reinnervate the muscle, they terminate both at the site of the implant (ectopic synapses) as well as at the old synaptic sites. These ectopic synapses form as a result of injury at the site of the implant, as it is known that this acts as a stimulus to synapse formation (Miledi, 1963). The difficulty in reaching a conclusion as to whether a denervated synaptic site uniquely contains information for the triggering of synaptic terminal formation arose as a consequence of the different methods used in carrying out the reinnervation experiment (Fig. 11.2C). 11.2.2 Site of synapse formation in developing muscle The elongation of mammalian muscles that occurs during development seems to be affected by the addition of sarcomeres at each end of the myofibres (Kitiyakara & Angevine 1963; Williams & Goldspink 1971, 1973) rather than by the intercalation of new sarcomeres between old ones. Passive tension determines the rate of sarcomere formation (Goldspink et al., 1974) which is the same at both ends of developing mammalian muscles (Williams & Goldspink, 1971, 1973; Schattenberg, 1973). During early development of muscle cells that are destined to receive a focal innervation from fast and slow (twitch) motoneurones, exploratory axons of the motoneurones form an initial synaptic contact at random along the length of the elongating myotubes (Fig. 11.5C and D). These myotubes subsequently receive a transient hyperinnervation from other exploring axons, but this additional innervation is constrained to the point of the initial synaptic site (Fig. 11.5A and 5B) (Tello, 1917; Redfern, 1970; Bennett & Pettigrew, 1974a). It follows that as myotubes continue to increase in length and become myofibres, no additional synaptic formation occurs on the new membrane laid down at the growing ends of the muscle cells (Fig. 11.5E). The initial synaptic contact is not established at any particular point along the length of the short myotubes (300 mm long), so that the entire myotube membrane is equally receptive for initial synapse formation (Bennett & Pettigrew, 1974a). Furthermore, as the transient hyperinnervation of the muscle cells is constrained to the initial synaptic sites, the first synapse must have made the myotube membrane refractory to synapse formation elsewhere. These observations support the early suggestion of Harrison (1910) that the entire myotube membrane is likely to be receptive for the initial synaptic contact, and then subsequently become refractory after the initial synaptic contact is established, in a way that is analogous to that of the spermatozoa and the ovum. In his own words There is nothing in the present work which throws any light upon the process by which the final connection between the nerve fibre and its end organ is established. The foregoing facts suggest that there may be a certain analogy here with the union of egg and sperm cell. The nerve fiber during its growth comes into contact with a cell of the proper kind. Assuming the latter to be in a condition of ripeness, a more intimate contact or perhaps even actual fusion may take place between nerve twig and end cell. A connection of this kind once established would terminate the susceptibility of the cell to further innervation, and nerve fibres growing subsequently in the same path would pass along to other end cells which were mature but not yet innervated. The nerve fibre itself, however, apparently retains its power of growth and ramification, for it usually becomes connected finally with a large number of end cells, as is plainly the case with muscle and ordinary cutaneous endings. When the exploring axons of avian motoneurones grow into a group of muscle cells that receive a distribution innervation from slow (graded) motoneurones in maturity they form ‘en grappe’ synapses along the length of each of the myotubes at intervals of ca 170 µm; each of these synaptic sites subsequently receives a transient hyperinnervation from further exploring axons (Bennett & Pettigrew, 1974a). No synapses form at the growing ends of the muscle cells until the length of new membrane layed down is > 170 µm. These observations suggest that slow (graded) motoneurones can only maintain a length of about ±170 µm of muscle cell membrane refractory to synapse formation. It is likely that some of the multiple innervation of synaptic sites observed electrophysiologically arises as a consequence of electrical coupling between adjacent myotubes (Fischbach et al., 1971), each of which possesses at least a single synapse. Nevertheless, each site on an individual muscle cell probably receives a transient hyperinnervation, as muscle cells are multiply innervated up to about 3 weeks postnatal, and during this time they have already matured to myofibres that are not electrically coupled together (Bennett & Pettigrew, 1974a). These experiments in the mid 1970s were summarised in terms of the existence of a specific receptor molecule associated with the muscle cell membrane and required for synapse formation. It was argued that the receptor molecule occurs over the surface of the myotube membrane before the initial contact with an axon terminal is made, but, after this is made, the receptor molecule is restricted to the site of the axon terminal contact with the muscle cell. Thereafter incorporation of the receptors into developing myofibre membrane is restricted for some distance from the synapse, this distance being dependent on the
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
243
Fig. 11.4. First evidence of polyneuronal innervation of synaptic sites during development and following reinnervation of muscle. (A) Motor endplates in the tongue of the human foetus. Magnification×1200. a, b and c: endplates innervated by a single nerve, d and f: endplates receiving a polyneuronal innervation, g and h: endplates innervated by the same nerve (from Fig. 35 in Tello, 1917). (B) Endplate potentials (epps) from the curarised diaphragm muscle of a 10 day old, 16 g rat. The resting potential was 70 mV on insertion of the electrode and 50 mV during the recording. The endplate potential consists of two units which are added to the epp by progressively increasing the stimulus strength, and subtracted from the epp in reverse order, by reducing the stimulus strength (from Fig.3 in Redfern, 1970). (C) Diagram illustrating the reinnervation of the synaptic site in muscle following partial denervation and reinnervation, a and b are the nerve fibres; en, endoneurial sheath; col, collateral branch; rb, regenerative buds at the tip of the axon; s, sole plate; ue, ultraterminal endplate; ul, ultraterminal process. Following partial denervation, in B, the denervated endplate becomes innervated by a collateral sprout from the a node of Ranvier of an axon innervating a nearby endplate, as in C, or by an sprout directly from an innervated endplate for which the details are given in the insert. On return of the previously lesioned nerve supply, the endplate innervated by a collateral sprout becomes transiently polyneuronally innervated as shown in D and insert (from Fig. 5 in Hoffman, 1951). (D) Endplate potentials, in response to increasing the intensity of nerve stimulation, recorded from an extensor digitorum longus muscle at 18 days after crushing the peroneal nerve. Muscle paralysis was produced by increasing the concentration of MgCl2 in the bathing solution to 8 mM A or by adding 1 µM of tubocurarine B. Note that with an increase in the stimulus intensity, indicated by the arrow, there appears a second component to the endplate potential, showing that the synaptic site is polyneuronally innervated (from Fig.2 in McArdle, 1975).
motoneurone type. The synaptic site(s) then receive a transient polyneuronal innervation. The formation of synaptic terminals in reinnervated mature muscle is triggered by the axons coming into contact with this receptor molecule at the established synaptic site. These sites then also receive a transient polyneuronal innervation. One basis for the transient nature of this polyneuronal innervation both during development as well as in maturity is that the several synaptic terminals at a site compete for the receptor molecule, with one of these terminals depriving the others of the molecule to the extent that they are no longer able to form a synaptic terminal. In the 20 years since this hypothesis was enunciated many different experiments have been performed on the formation of synaptic terminals in developing and mature muscles. The possibility that these experiments might give insights into the properties of the synapse formation molecules, both the receptor molecule associated with the muscle as well as the donor molecule proffered by the axon terminal to the receptor, requires that the hypothesis be put in formal mathematical terms. This was done by Bennett and Robinson (1989), and subsequently termed the dualconstraint hypothesis by Rasmussen and Willshaw (1993). The next section is concerned with applying the dual-constraint hypothesis to the body of research concerned with synapse formation, in the expectation that this will define the necessary minimal properties of synapse-formation molecules. 11.3 Synapse-formation molecules in muscle and the elimination of polyneuronal innervation 11.3.1 Elimination of polyneuronal innervation during development of muscles During the formation of limb muscles, myoblasts first fuse to form primary generations of three to ten myotubes at specific locations in the limb. The muscle cells at each of these locations are destined to form distinct muscles (Noakes et al., 1986). Motor axons leave the main limb nerves opposite the specific locations in the limb at which individual muscles are forming from the primary myotubes. The growth cones of these axons are observed throughout the length of primary myotubes (Bennett & Pettigrew, 1974a; Ontell & Dunn, 1978; Dennis et al., 1981), The initial site of synapse formation occurs at
244
HISTORY OF THE SYNAPSE
Fig. 11.5. Development of the first synaptic contact between muscle cells and synaptic terminals. (A) Vertebral muscle of the chick embryo at 7 days incubation. Magnification, X 740. a, nerve entering the muscle; b, section through a myotube, passing down the longitudinal axis; c, superficial view of a myotube, with the nuclear accumulation on which the nerves end shown in d; e, intramuscular connective tissue; f, nerve ramifications passing along the length of a myotube; m, exploratory nerve fibres in a fasicle of fibres; n, fascicle of nerve fibres penetrating the muscle (from Fig. 9 in Tello, 1917). (B) Transverse section through a leg muscle of a chick embryo at 9 days’ incubation. Magnification, X 740. a, cluster of myotubes; b, isolated myotubes; c, primary myotube at the level of a single nucleus; d, primary myotube with two nuclei; e, primary myotube at the level where there are no nuclei; f, o, p, s, exploratory nerve fibres touching the clusters of myotubes (from Fig. 11 in Tello, 1917). (C and D) Distribution of the axons of (twitch) motoneurones in a mammalian twitch muscle (rat hemidiaphragm) at 15 days gestation. The primary (phrenic) nerve enters near the centre of the hemidiaphragm and divides into two secondary nerve branches, which traverse the myotubes (my) at right angles and extend to the pools of myoblasts (mb) at the ends of the hemidiaphragm. Smaller, tertiary nerve bundles leave the secondary nerve trunks and run, in general parallel to the myotubes and obliquely into the pool of myoblasts. The tertiary nerve branches extend almost the entire length of those myotubes adjacent to the pool of myoblasts. (C) Tertiary nerve bundles at the end of the secondary nerve trunk radiating into the pool of myoblasts (mb) at the end of one hemidiaphragm. Calibration 50 µm. D, Single axons leaving the tertiary nerve branches at a position 1 mm from the point of nerve entry into the hemidiaphragm and growing out between and along the myotubes (my) in this region. The tertiary nerve bundles arise from the secondary nerve trunk(s) at the bottom of the plate. Some axons have been retouched since they pass out of the plane of focus of the three photomicrographs used to make up the plate. Calibration 30 µm. (from Plate 1 in Bennett & Pettigrew, 1974b). (E) Diagram of the pattern of innervation in one-half of a mammalian twitch muscle (rat hemidiaphragm) by the axons of (twitch) motoneurones over the last 6 days gestation. In A the primary (phrenic) nerve (p) enters the hemidiaphragm and divides into two secondary nerve trunks which traverse the hemidiaphragm along its long axis. Smaller tertiary nerve bundles leave the secondary branches and extend along the myotubes for a short distance, corresponding to the length of the myotubes at the time of their differentiation. This low-power diagram also shows the large increase in dimensions of the muscle during this period and the small increase in width of the area occupied by nerves. Vertical dashed lines show the point of phrenic nerve entry into the muscle and the position 1 mm lateral to this point. B Shows a high-power diagram indicating the distribution of nerves in the area of the muscle 1 mm from the point of nerve entry (bounded by dashed lines). The vertical lines represent the orientation of the myotubes and the small dots at and after 16 days gestation represent cholinesterase deposits (c). The secondary nerve trunk(s) transverse the diaphragm at right angles to the myotubes and give rise to tertiary nerve bundles (t), which extend along the myotubes for a short distance corresponding to the length of the myotubes at the time of their differentiation. At 15 days gestation, the nerves at 1 mm from the point of nerve entry occupy almost the entire length of the myotubes and extend into the adjacent pool of myoblasts at the end of the muscle. From electrophysiological evidence, some of the cells at this stage are innervated at a single point by a single axon. At 16 and 17 days gestation, cholinesterase deposits appear in the middle of the myotubes and these are innervated by at first one, and then later by a number of silver-impregnated axons. At this stage, evoked end-plate potentials and spontaneous miniature end-plate potentials of the fastest rise time are recorded in the middle of the muscle cells. From electrophysiological criteria, the muscle cells are innervated at one site by up to four synaptic terminals. By 21 days gestation, there is a complex network of axons extending between the synapses and no newly forming synapses are observed. (From Text Fig. 1 in Bennett & Pettigrew, 1974a).
random along the length of individual primary myotubes (Bennett & Pettigrew, 1974a; Ziskind-Conhaim & Dennis, 1981; Chow & Cohen, 1983; Phillips et al., 1985; Phillips & Bennett, 1987). Each of these sites is innervated within about two days by the terminals of a number of different motoneurones (Bennett & Pettigrew, 1974a; Letinsky & Morrison-Graham, 1980).
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
245
By the time the maximum polyneuronal innervation of the synaptic sites is established, the plasma membrane at the site is covered with basal lamina (Chiu & Sanes, 1984; Kullberg et al., 1977; Hirokawa & Heuser, 1982; Sanes & Chiu, 1983). This polyneuronal innervation is transient as it is removed from most muscle cells in the subsequent few weeks (Frog: Bennett & Pettigrew, 1975; Letinsky & Morrison-Graham, 1980; Morrison-Graham, 1983; Bennett & Lavidis, 1986; Malik & Bennett, 1987. Birds: Bennett & Pettigrew, 1974a; Atsumi, 1977; Pettigrew et al., 1979; Pockett, 1981. Rats: Redfern, 1970; Bennett & Pettigrew, 1974; Benoit & Changeux, 1975; Brown et al.,, 1976; Korneliussen & Jansen, 1976; Rosenthal & Taraskevich, 1977; Riley, 1977; Benoit & Changeux, 1978; O’Brien et al., 1978; Betz et al., 1979; Miyata & Yoshioka, 1980; Riley, 1981; Denniss et al., 1981; Taxt et al, 1983; Caldwell & Ridge, 1983; Bennett et al., 1986a. Rabbit: Bixby & Van Essen, 1979; Bixby, 1981. Cat: Bagust et al., 1973. Human: Tello, 1917). Before this removal about half the motoneurones in the spinal cord degenerate (Frog: Prestige, 1967. Bird: Hamburger, 1975; Oppenheim & Chu-Wang, 1983. Rat: Harris & McCaig, 1984). The eventual decline of polyneuronal innervation involves competition between motoneurones for synaptic sites. For example, the motoneurone pool to the rat lumbrical muscles gives rise to axons that enter the muscle via the lateral plantar and sural nerves. If the lateral plantar nerve is cut at birth, the sural nerve continues to innervate a much larger number of myofibres than it would if the lateral plantar axons were present (Betz et al., 1980). Similarly, if one of the two segmental nerves to the rat medial gastrocnemius is cut at birth the remaining segmental nerve continues to innervate a much larger number of muscle cells than it would if both segmental nerves were intact (Bennett et al., 1986). The growth of axons depends on the elongation of microtubules that form their principal longitudinal structural element; tubulin subunits are transported to the growth cone where they are assembled onto the free ends of microtubules (Bamburg et al., 1986). The axon ceases to elongate when a degradation process is set in train at the end of the axon whereby the microtubules are denatured at the same rate as they are assembled; this degradation is possibly due to the activation of a protease in the terminal by an influx of calcium ions (Schlaepfer & Hasler, 1979; Roots, 1983). One model of synapse formation involves competition of the postsynaptic cell for a presynaptic source (T) provided by synaptic terminals and competition of synaptic terminals at synaptic sites for a postsynaptic source (R) provided by the cell. The complex formed by these synapse-formation molecules then triggers the accelerated synthesis of microtubule and neurofilament proteins and therefore an increase in size of synaptic terminal (Campenot, 1982). The terminal continues to grow depending on the amount of synapse formation molecules made available at the synaptic site and synaptic terminal membrane respectively; equilibrium is reached when the assembly of microtubules just balances their degradation. According to this scheme, the elimination of polyneuronal innervation involves competition between terminals and the postsynaptic cell for synapse-formation molecules (Fig. 11.6). Whether this competition involves a trophic molecule, as has been suggested (Changeux & Danchin, 1976; Purves, 1977; Purves & Lichtman, 1980; Bennett, 1983; Gouze et al., 1983; Herrera & Grinnell, 1985; Purves, 1986), or an agrin-agrin receptor complex (McMahan, 1990) will be considered in a later section. 11.3.2 Elimination of polyneuronal innervation during reinnervation of mature muscles If the axons of motoneurones to mammalian, avian or amphibian muscles are cut just outside the muscle and allowed to spontaneously reinnervate the muscle, the axons eventually penetrate the epimysium of the muscle. They then grow through the muscle along the surfaces of myofibres and blood vessels, but only form synapses at the old synaptic sites, that is, over the remaining postsynaptic folds (Bennett et al., 1973a; Van Essen & Jansen, 1974; Sanes & Hall, 1979; for a review see Bennett & Pettigrew, 1976 and Bennett 1983). In this way the normal innervation pattern of the myofibre is re-established as we have seen, whether this involves a focal or distributed innervation, because the nerves terminate at their old synaptic sites. The plasma membrane at these synaptic sites contains high concentrations of a neural cell adhesion molecule (N-CAM; Covault & Sanes, 1985) and is covered in a basal lamina rich in heparan sulphate proteoglycans (Anderson & Fambrough, 1983; Anderson, 1986). Each of the reinnervated synaptic sites receives a polyneuronal innervation (McArdle, 1975; Thompson, 1978). At least in the case of mammalian muscle polyneuronal innervation is subsequently eliminated or reduced to the normal low levels observed in control muscles; this occurs over a period of about three weeks, that is, about the same time as that for the elimination of polyneuronal innervation in developing muscles (Taxt, 1983; for comparison with amphibia see Rotshenker & McMahan, 1976). The similarities between this elimination process during reinnervation compared with development suggests that terminals may be competing for synapse-formation molecules during reinnervation as well as during development.
246
HISTORY OF THE SYNAPSE
Fig. 11.6. Diagram illustrating the dual-constraint theory of terminal elimination at a synaptic site. (a) A motor growth cone (size proportional to Snm(o)), containing synapse formation molecules Tnm(0) makes initial synaptic contact (t=0) with a pool of synapse formation molecules in the muscle (Rm(0); at some time later (t) the growth cone has developed to from a terminal as a consequence of the reaction between the synapse formation molecules Rm(t) and Tnm(t). (b) The terminals of two neurones (1 and 2) have established themselves at a synaptic site on muscle cell 1 by time t; each terminal receives synapse-formation molecules (T1 (t) and T2 (t)) which are incorporated into the presynaptic membrane (T11(t) and T21(t)) where they react with the postsynaptic synapse-formation molecules (R1(t)); the size of the terminals is then proportional to the products of this reaction (S11(t) and S21(t)). (c) The connections between two motoneurones with synapse-formation molecules T1 (t) and T2(t) and three muscle cells with synapse-formation molecules R1 (t), R2(t) and R3(t) at time t. (From Fig. 1 in Bennett & Robinson, 1989)
11.3.3 A dual-constraint theory for the role of synapse-formation molecules in the elimination of polyneuronal innervation In this section a model is presented of how synapse-formation molecules on the several synaptic terminals at a polyneuronally innervated site compete for synapse-formation molecules made available by the post-synaptic site. The model is offered in an attempt to bring together in a coherent framework the large number of observations that have been made on the elimination of polyneuronal innervation. It is assumed that a motor axon, through its collaterals, either has equal chance to come into contact with all the muscle cells present at the time of innervation or that it has equal chance to come into contact with a subset of the muscle cells. Each muscle cell, m, has synapse-formation molecules (Rm) made available at a region of the site of initial synaptic contact between the muscle cell and a synaptic terminal (Fig. 11.6a). Each motoneurone, n, has synapse-formation molecules (Tn) that are transported to its terminals and incorporated into the membrane (Fig. 11.6a). The reaction between the Tn of a terminal on muscle cell ‘m’ namely (Tnm) and Rm gives rise to the formation of new synaptic terminal which is made in amounts directly proportional to the product (Snm) of the reaction (Fig. 11.6a). Several collaterals, each from a different motoneurone, may initially form terminals at the same synaptic site on a muscle cell (Fig. 11.6b and c). At time ‘t’ the reaction between R & T for a particular collateral is in which the area of terminal is directly proportional to Snm(t); knm and K are then the forward and reverse rate constants. At time zero, when the synaptic terminal first forms on the muscle, the Tnm molecules are free to move in the presynaptic membrane and Rm molecules in the synaptic site area of the postsynaptic membrane. The probability of collision between a Tnm molecule and a Rm molecule is taken as proportional to the product of the amounts of these, so the forward reaction is proportional to Rm Tnm; the back reaction is proportional to K Snm(t). If there is an autocatalytic effect, whereby the amount of Snm(t) produced accelerates the availability of Rm(t), then n gives an index of the autocatalytic effect so that:
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
247
Fig. 11.7. (A) The size of synaptic terminals are shown in A and B possessed by 20 ± 8 (±s.d.) collaterals of each of 10 motoneurones (identified by their ‘Nerve number’) distributed randomly over 50 muscle cells (identified by their ‘Muscle number’). The length of each horizontal bar is proportional to the size of the terminal (Snm) at 2.7 days in A and 37 days in B after initial synaptic contacts formed. C and D. The time course of changes in size of neve terminals on muscle cell number 14 (Sn14) and muscle cell number 36 (Sn36) given in A and B. The constants in the equations generating Snm(t) in this figure and all subsequent figures in sections 2, 3 and 4 are Snm(o), 0.4; K, 0.01; Rm(o), 1.0; Tnm(o), 5. (from Fig. 2 in Bennett & Robinson, 1989).
The factor Cn allows for changes in the affinity for nerve n and is generally put equal to 1. The equations derived for this model and their numerical solutions for the terminal sizes over time are described in (Bennett & Robinson, 1989). Rasmussen & Willshaw (1993) have made a complete analysis of this model of synapse formation and elimination. They termed it the ‘dual-constraint’ theory as there are two kinds of competition, namely competition for a presynaptic resource (T molecules in the terminal) and for a postsynaptic resource (R molecules in the muscle). Postsynaptic competition is required to remove polyneuronal innervation and presynaptic competition explains many other phenomena detailed below. This dual-constraint theory provides a quantitative procedure for considering most of the phenomenological experiments on the formation and elimination of nerve terminals as is now described in Sections 11.4–6. Subsequent sections present an attempt to identify the T and R molecules in the model with the molecules that have recently been isolated at developing muscle synapses (section 11.7). 11.4 Elimination of polyneuronal innervation during development described by dual-constraint theory The mechanisms regulating the loss of polyneuronal innervation of muscle fibres have been discussed in a number of reviews incorporating different perspectives. Competitive mechanisms of elimination are discussed by Jansen and Fladby (1990), Betz et al. (1990) and by Herrera and Werle (1990) with a role of competition for trophic molecules being stressed by Thompson (1985) as well as by van Essen et al. (1990). However, in none of these is there an attempt to derive a minimal molecular model that gives a quantitative prediction of all the phenomenological experiments involving the formation and elimination of synaptic terminals in development and following injury to the nerve supply in adults. Here the dual-constraint model is used to supply this description.
248
HISTORY OF THE SYNAPSE
Fig. 11.7. (B) Comparison between the decline in motor-unit size observed experimentally and that according to the model of Fig.11.6. The ordinate gives the size of motor units expressed as their percentage of the maximal twitch to direct stimulation; the abscissae is days postnatal. , Mean for 12–25 motor units in the developing rat lumbrical muscle; each vertical line gives ±S.D. (data taken from Betz et al. (1979), (Fig. 5)). , Mean size of motor units in the developing rat soleus muscle; each vertical line gives the range of observation on one animal (data taken from Brown et al., (1976), Text-Fig. 6). , Motor-unit sizes determined for the model illustrated in Fig. 11.6. As the mature motor units in the soleus are twice the size of those in the lumbrical, the data points for the soleus have been doubled for the sake of comparing the relative time course of decrease in motor unit sizes between the two muscles and the predictions of the model. (C) Comparison between the decline in polyneuronal innervation observed experimentally and that according to the model of Fig. 11.6. (a), , Mean numbers of terminals per muscle cell of the rat hemidiaphragm determined electrophysiologically at different times postnatal. The error bars represent the S.E. of the mean; data from Bennett & Pettigrew, 1974a, Text Fig. 4). , Values for the model illustrated in Fig. 11.6 (b), each filled circle gives the percentage of muscle cells within the rat lumbrical muscle which received a synaptic input, as determined electrophysiologically at the postnatal time given in the abscissa (data from Betz et al., 1980, Fig. 10); , Values for the model illustrated in Fig. 11.6. Terminals were considered detectable by electrophysiological means if they had a Snm > 0.05. (D) Comparison between the decline in polyneuronal innervation determined using different techniques and that according to the model of Fig. 11.6. (a). A, each filled circle gives the mean number (±S.E. of mean) of polyneuronally innervated fibres in the rat soleus muscle determined electrophysiologically (continuous line) expressed as a percentage of the total number of fibres examined at different times postnatal; , Polyneuronal innervation determined histologically (broken line) (data from O’Brien et al., 1978; Text Fig. 2); , Values from the model in Fig. 11.6 in which terminals were considered detectable by histological means if they had a Snm > 0.04. B, , Percentage of rat soleus muscle fibres innervated by more than one axon at different days postnatal (data from Taxt et al., 1983, Fig. 2); , Values for the ‘cut muscle’ preparation which allows detection of terminals generating very small synaptic potentials; , , , Values for the model in Fig. 11.6 in which terminals were detectable if they had a Snm>0.04, 0.05 or 0.07 respectively.
11.4.1 Loss of polyneuronal innervation During normal impulse traffic A decrease in polyneuronal innervation occurs in rodents the first during 2–3 weeks after birth. A synapse formation matrix, used to model this process, is shown in Fig. 11.7AA: it consists of 10 motor nerve and 50 muscle cells shortly (2.7 days) after terminals have contacted synaptic sites. Each motoneurone has been allowed to form 20 ± 8 (± S.D.) collaterals at random over the 50 muscle cells; as a consequence some motor nerves have only a few synapses (e.g. motor nerve 8 has eight synapses) whereas others have a great many (e.g. motor nerve 1 has 27 synapses). Of the 50 muscle cells, four are innervated at this early time and 44 receive a polyneuronal innervation. The three muscle cells that do not receive any innervation are 4, 5 and 7; these primary myotubes would be expected to eventually degenerate. The three muscle cells that receive only a single innervation are located on muscle cells 1, 2 and 8.
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
249
The time course of events during the subsequent elimination of polyneuronal innervation are determined by ascertaining the number of days taken to reach different levels of multiple innervation of synaptic sites in the rat diaphragm (Bennett & Pettigrew, 1974a) and fitting this to the number of iteration steps required to reach the same level of multiple innervation in the model. At 2.7 days after the terminals are first established, the size of these terminals (Snm) has almost reached the maximum value of 1.0, as a consequence of lack of competition from other terminals. At 37 days after synapses first form, only 4 of the muscle cells now receive a polyneuronal innervation; these are muscle cells 16, 20 and 22 and 34 (Fig. 11.7AB). It will be noticed that the synapses of motoneurones with very few terminals (namely numbers 5 and 8) at 2.7 days have retained nearly all their synapses at 37 days (Figs. 11.7AA, AB). This occurs as the amount of T molecule is the same for each motoneurone so that the amount of T molecule per terminal is relatively high for motoneurones with relatively few terminals; such a large amount of T molecule confers a competitive advantage on these terminals in the competition for synaptic sites. Each motoneurone at equilibrium has about five mature-size terminals; this occurs because each motoneurone has five-times as much T molecule as there is R molecule in a muscle; it follows that after polyneuronal innervation is eliminated, each motoneurone can totally occupy the R molecules of five muscle cells. The time course of elimination of excess terminals at synaptic sites which receive a polyneuronal innervation occurs over a period of 3–4 weeks (Figs. 11.7AC, D). All terminals grow in size over the first four days, at which time the reduction in size of terminals destined to be removed begins. It is not possible to predict which terminals are destined to be eliminated by inspection of the synapse formation matrix at very early times: although most of the terminals of motoneurones with a large number of collaterals are removed, whereas most of those with few collaterals are not, some collaterals of each kind of motoneurone are eliminated (Figs. 11.7AA, B). It will be noted that there is a residual 5–10% polyneuronal innervation that is not eliminated. The model gives a good prediction of the loss of polyneuronal innervation during the postnatal period (Taxt et al., 1983; Betz et al., 1980; Fig. 11.4), although there is a tendency for the predicted loss to occur slightly earlier than that observed and to continue slightly later. However, the time course of loss of polyneuronal innervation reported by various investigators depends on the sensitivity of the methods used (O’Brien et al., 1978): intracellular recording from muscle cells requires that the synaptic potential be made subthreshold for the initiation of the muscle impulse, a procedure which often suppresses the size of small synaptic potentials to the extent that they can no longer be detected. An alternative and more sensitive method has been used: this involves cutting the muscle cells at sites removed from their regions of innervation so that they become partly depolarised, a condition in which impulse initiation is largely suppressed. Using the former technique on the rat soleus muscle, polyneuronal innervation is eliminated by 3 weeks postnatal (Brown et al. 1976; Betz et al. 1980; O’Brien et al, 1978); using the cut muscle technique the elimination is not complete at 4 weeks postnatal (Taxt et al., 1983). In the case of the amphibia, up to ca 20% of fibres may retain their polyneuronal innervation into maturity (Ridge & Thomson, 1980; Herrera, 1984; Malik & Bennett, 1987). The size of motor units, measured as the extent of muscle force produced by a unit expressed as a percentage of the force produced by the whole muscle, declines as a consequence of the loss of polyneuronal innervation. Motor unit sizes can be calculated from a synapse formation matrix by simply expressing the number of muscle cells innervated by a motoneurone as a percentage of the total number of innervated fibres in the matrix. If 10 motoneurones each have 20 ± 8 terminals spread at random over 50 muscle cells (as in Fig. 11.7AA), then the range of motor unit sizes at different times is given in Fig. 11.7B: the 10 motor units have sizes that range from 10% to 70% of the whole muscle at 2 days; this then declines with the loss of collateral sprouts that accompanies the loss of polyneuronal innervation, until by ca 4 weeks all the motor units have been reduced to the size of the smallest unit, namely about 10%. This result illustrates again that smaller units retain their terminals over those of large units, because the former have more T receptors per terminal than do the latter. The theoretical results are in good agreement with the observations of Betz et al., (1979) on the lumbrical muscle and of Brown et al., (1976) on the soleus muscle (Fig. 11.7B). Indeed the observations of Bennett and Pettigrew (1974a, b) as well as those of Betz et al., (1980) on the loss of polyneuronal innervation are well described by the model (Fig. 11.7C) as are those of O’Brien et al., (1978) as well as Taxt et al., (1983), as is shown in Fig. 11.7D. During a decrease in impulse traffic If electrical conduction in the sciatic nerve of rats is blocked with tetrodotoxin at 9 days postnatal, the elimination of polyneuronal innervation is arrested some 2– 4 days later (Thompson et al., 1979). This is not due to the formation of motor nerve sprouts making new synaptic contacts (Brown et al., 1981). Injection of botulinum toxin into 10-day-old rat muscles (soleus, extensor digitorum longus or peroneus tertius), with consequent blockade of synaptic transmission, is also followed by a delay in the loss of polyneuronal innervation (Brown et al., 1977). The delay involves the return of the electrical signs of multiple innervation, perhaps due to the reactivation of inputs still in close contact with the synaptic site; it is not due to the formation of nerve sprouts.
250
HISTORY OF THE SYNAPSE
Studies of the number of terminals at neuromuscular sites in developing muscles indicate that a wide variety of procedures that block neuromuscular transmission will also block the loss of polyneuronal innervation: the use of either -bungarotoxin, suxemethonium or curare on avian embryos increases the number of nerve profiles observed at synaptic sites compared with controls (Srihari & Vrbova, 1978; Sohal et al., 1979; Sohal, 1981; Ding et al., 1983). In rat soleus muscles, treatment with bungarotoxin at 10 days blocks the loss of axon profiles at synaptic sites (Duxson, 1982). Reduction of neuromuscular transmission by decreasing the afferent innervation of motoneurones with spinal cord isolation at 7 days postnatal in rat also greatly prolongs the period of polyneuronal innervation which is not eliminated until 30 days (Caldwell & Ridge, 1983). A decrease in impulse traffic during the normal period of loss of polyneuronal innervation, either by tetrodotoxin paralysis (Brown et al., 1981; Thompson, 1985; Fig. 11.8BA) or by severing the spinal cord rostral to the origins of the motor nerves (Caldwell & Ridge, 1983; Fig. 11.8BB), greatly delays the process of elimination. This process can be predicted by the models given in Fig. 11.8 A), if the autocatalytic effect (namely in the equations) is removed. Under these circumstances, as shown in Fig. 11.8B, very few terminals are removed over a 27– 40 day period after synapses are first formed. The loss of polyneuronal innervation is therefore very slight during the normal period of elimination (Fig. 11.8BA and 11.8BB). One interpretation of the removal of autocatalysis from the model is as follows: impulse traffic in a terminal normally gives rise to synaptic potentials whose electrical field may preferentially attract R molecules to the terminal (for membrane bound molecules acting in this way see Poo, 1981; 1985; Fraser, 1985; Fraser & Poo, 1982); the attraction of R molecules to terminals increases the rate of formation of S and hence the increase in size of the terminal which in turn increases the amount of transmitter released and the size of the electrical field. The resulting autocatalysis greatly accelerates the process of competition for R molecules. In the absence of this, the rate of elimination of polyneuronal innervation is decreased (Fig. 11.8B). One observation that does not fit with this interpretation is that if a fraction of the nerve supply to the neonatal soleus muscle is made inactive, then the motor units of these blocked nerves become greater than those of the unblocked nerve (Callaway et al., 1987; 1989). During an increase in impulse traffic There is an accelerated loss of terminals from synaptic sites in the soleus muscle if synaptic potentials are enhanced by treating the muscle with acetylcholinesterase inhibitors at 10 days (Duxson & Vrbova, 1985). In addition, an increase in the impulse traffic of motor nerves in neonatal mammals increases the rate of loss of polyneuronal innervation: this was first observed for the rat soleus muscle following stimulation of the sciatic nerve at 6 days postnatal for 4 h per day at 8 Hz for 2–4 days (O’Brien et al, 1978). Subsequently, Thompson (1983a, b) showed that the optimum stimulation for accelerating the loss of polyneuronal innervation in the soleus was 100 Hz for 1 sec repeated every 100 sec for 3–4 days. This process can be predicted if the autocatalytic reaction is accelerated: in this case the model of Fig. 11.9A shows that at 10 days postnatal the extent of removal of the polyneuronal innervation is almost complete if the reaction is speeded up (see Fig. 11.9AB) compared with the normal reaction rate (see Fig. 11.9AA). These results are quantitated in Fig. 11.9B, which shows that the accelerated loss of polyneuronal innervation according to the model gives a reasonable prediction of the observations by Thompson (1983a, b). Following reinnervation with normal impulse traffic Following crush of the nerve to the soleus muscle in neonatal rats, the reinnervating axons establish a polyneuronal innervation that is as extensive as that found during normal development; this subsequently disappears during the second postnatal week as in normal animals (Brown et al., 1976). Different results have been obtained for neonatal rat intercostal muscles in which transmission is not even restored for three weeks following denervation at birth (Dennis & Harris, 1980). If the nerves to the neonatal soleus muscle are crushed at birth, then polyneuronal innervation is reconstituted on reinnervation, and subsequently removed at about the same time as occurs normally (Brown et al., 1976; Fig. 11.10). This process can be predicted by a model in which each of ten reinnervating motoneurones forms much fewer collaterals than occurs during normal devel- opment and the terminals of these collaterals form synapses on a group of adjacent muscle cells rather than being distributed over the whole muscle. The predictions for the case in which each motoneurone formed 10±4 collaterals, rather than the usual 20±8 collaterals (as in Fig. 11.7A, Fig. 11.8A, Fig. 11.9A), is given in Fig. 11.10: there is a reasonably good fit between the theoretical and experimental data. However, this was not the case if motoneurones formed a greater number of collaterals: in this case the time for elimination of polyneuronal innervation is longer than that observed. The process of elimination of polyneuronal innervation is accelerated in the model with only 10±4 collaterals per motoneurone grouped together, as many muscle cells only receive two competing terminals under these circumstances; in addition, the relatively large amount of T receptor in the terminals of motoneurones with such relatively few collaterals accelerates the rate of terminal removal. The model suggests that motoneurones form many more collaterals during development than during reinnervation of developing muscles.
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
251
Fig. 11.8. (A) The size of synaptic terminals possessed by the collaterals of each of 10 motoneurones over 50 muscle cells when the synaptic terminal impulse traffic is blocked (interpreted as removing the autocatalytic reaction so alpha=0 and Snm=1). (A) and (B): the distribution and size of terminals; A gives the size of terminals reached by 20±8 collaterals of each motoneurone at 2.7 days; B gives the size of these terminals at 40 days. C and D The distribution and size of terminals in the presence of synapse-formation receptors with topographical information; C gives the size of terminals reached by 20±8 collaterals of each motoneurone at 16 days; D gives the size of these terminals at 27 days. Note that polyneuronal innervation is not eliminated in the absence of impulse traffic (or the autocatalytic reaction). (B) Comparison between the effects of a decrease in neuromuscular transmission on the loss of polyneuronal innervation observed experimentally and that determined by a model in which blocking nerve impulse traffic has the effect of removing the autocatalytic reaction. A, Frequency of polyneuronally innervated rat soleus muscle fibres during normal development and following paralysis with tetrodotoxin; each symbol represents a single muscle in which 20 or more fibres were examined by intracellular recording for incremental changes in the amplitude of the endplate potential in response to graded intensity stimulation of the muscle nerve; fibres were considered polyneuronally innervated if more than one increment was seen (curve drawn by eye fits the points taken from Brown et al., 1976, Text-Fig. 6); , muscles continuously paralysed from day 9 by tetrodotoxin application to the right sciatic nerve; , muscles paralysed with tetrodotoxin from day 10 (all data is from Thompson, 1985; Fig. 1); O, results from the model shown in (A). B, , Frequency of polyneuronal innervation determined as in A but for the rat lumbrical muscle following spinal cord isolation by severing the cord at T12 or T13 at 7 days postnatal; (the envelope encloses the normal data from Betz et al.,1979); the data is from Caldwell & Ridge, 1983, Fig. 3); , give the results from the model shown in (A). Note that there is only a very small loss of polyneuronal innervation in the absence of impulse traffic.
11.4.2 Emergence of mature motor units In the presence of muscle growth The formation of secondary myotubes in rat muscles continues into the first week postnatal (Harris, 1981). It is then possible
252
HISTORY OF THE SYNAPSE
Fig. 11.9. (A) The size of synaptic terminals possessed by 20±8 collaterals of each of 10 motoneurones over 50 muscle cells when the average impulse traffic is increased during the early loss of polyneuronal innervation. A Gives the distribution of terminals sizes at 10 days postnatal in the absence of enhanced impulse activity. B Gives the distribution of terminal sizes at 10 days postnatal when impulse activity had been enhanced preceding this time by increasing the autocatalytic reaction. Note the accelerated loss of polyneuronal innervation in B compared with A. (B) Comparison between the experimentally determined frequency of polyneuronally innervated muscle fibres in 10 to 11-day-old rat soleus muscles following increased activity evoked by stimulation with the predictions of a model in which the autocatalytic reaction has been increased; the horizontal hatched region gives the level (mean ± S.E.M.) of polyneuronal innervation present in all muscles of normal, untreated animals reared by their mothers; the pair of unhatched bars to the left represents the polyneuronal innervation in right soleus muscles stimulated with a 100 Hz stimulus pattern from day 7 (stim) and in the left, contralateral muscle (unstim) from eight animals. The accompanying hatched bars give the predictions of the model in (A)A. The pair of unhatched bars to the right shows the polyneuronal innervation present in muscles from three animals whose right soleus muscles received a 1 Hz stimulation pattern from day 7 (stim) and whose left soleus muscles served as controls (unstim); bars give the S.E.M. (data from Thompson, 1983a.b).
that the motor unit size does not change much in the first week postnatal as the loss of terminals from primary myotubes is offset by the innervation of the new secondary myotubes that also receive a polyneuronal innervation (Betz et al., 1979). The rat lumbrical muscle grows in the postnatal period by the addition of secondary myotubes: this involves a doubling of the number of muscle cells in the first two weeks postnatal (Betz et al. 1979; Fig. 11.11BA), that is, during the period of loss of polyneuronal innervation. Thus as polyneuronal innervation is eliminated from more mature muscle cells, later developing cells receive a polyneuronal innervation for the first time. The result is that motoneurones maintain an approximately constant number of terminals but reallocate them to the new muscle cells as they arise: the relative motor unit tensions (expressed as a percentage of the total tension of the muscle) then decline during the loss of polyneuronal innervation (Fig. 11.11BB) whereas the motor unit size (expressed as the number of muscle cells innervated by a motoneurone) remains relatively constant (Fig. 11.11BC). This process can be modelled as shown in Fig. 11.11(A): the proportional increase in the number of muscle cells in the lumbrical muscle is from 5 to 7 and from 7 to 9 on days 2 and 10 respectively (Fig. 11.11BA) and the synapseformation matrix has been increased in size by the same proportions at the same times (namely 25–35 muscle cells and 35–45 muscle cells on days 2 and 10, respectively; see Fig. 11.11BA); the elimination of terminals is well underway on the first 25 muscle cells by 8 days (2 days postnatal) when the next 10 muscle cells are added (Fig. 11.11Ab); most polyneuronal innervation is eliminated on the first 25 muscle cells by the time the last 10 muscle cells are added at 16 days (10 days
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
253
Fig. 11.10. Comparison of the experimental determination of the polyneuronal innervation of the reinnervated neonatal soleus muscle with a model in which 50 muscle cells are reinnervated by 10 motoneurones each of which can form 10±4 collaterals grouped on no more than 18 muscle cells in the presence of topographical information. The soleus nerve was crushed at birth at the point of its entry into the muscle and , Experimental results (from Brown et al., (1976), Text-Fig. 10). , Predictions of the model, in which the beginning of reinnervation is taken one day after the nerve crush.
postnatal; Fig. 11.11Ac); polyneuronal innervation has been eliminated from most cells (90%) by 47 days postnatal (Fig. 11.11Ad). The range of motor unit tensions may be ascertained for each motoneurone during this postnatal period by determining the number of muscle cells inner- vated by each motoneurone and expressing this as a percentage of the total number of muscle cells present. Calculated in this way, the range of motor unit tensions declines during the 3 weeks postnatal in a way which gives a good fit to the experimental results for the lumbrical muscle (Fig. 11.11BB). If motor unit size is simply defined as the number of muscle cells innervated by a motoneurone then motor-unit size does not decline as much over the postnatal period as does the percentage motor unit tensions (Fig. 11.11BC): this occurs because motoneurones with displaced terminals relocate them on newly formed muscle cells (see Fig. 11.11A). 11.4.3 Intrinsic withdrawal of motor-synaptic terminals Extensions of the model required Rasmussen and Willshaw (1993) have pointed out that the dual-constraint model considered here (Bennett & Robinson, 1989) has an unrealistic requirement that the ratio of T0 to R0 (that is the ration of the amounts of T and R initially present in each motoneurone and muscle system) equals the ratio of M to N (that is the ration of the total number of motoneurones and muscle cells). This is improbable, as information about R0, M and N is unlikely to be made available to each motor neurone, especially in situations such as partial denervation experiments where the value of N is changed by the experimenter. Rasmussen and Willshaw (1993) dealt with this by introducing explicit equations for the rate of orthograde transport of T to growing terminals and of anterograde transport of T from terminals that are shrinking. Incorporation of their equations (see their Equation 21) into the existing model then gives explicit evaluation of the ratio of T0 to R0 without the arbitrary assumptions of Bennett and Robinson (1989). This extended dual-constraint model then allows predictions to be made of the extent to which the remaining motor units in a partially denervated neonatal muscle will grow in size, given that values of the ratio T0 to R0 can be calculated. The theoretical results show that in the case of the mouse soleus muscle partially denervated at birth, there is an approximately proportional relationship between the number of motor units remaining after the partial denervation and the number of fibres that are innervated (Fig. 11.12A), and this is the same as the experimental findings of Fladby and Jansen (1987). Apparent contradictory experimental results concerning whether the loss of terminals can be ascribed solely to competition or if some other mechanism of intrinsic withdrawal is involved can be accomodated by the modified dual-constraint theory, if appropriate changes are made to the ratio of T0 to R0. For instance, Betz et al., (1980) found no evidence for intrinsic withdrawal following partial denervation of the lumbrical muscle whereas Fladby and Jansen (1987) found that a reduction in motor-unit size does occur in partially denerated soleus muscle. These results can be predicted, as shown in Fig. 11.12B, if the ratio of T0 to R0 is much greater in the lumbrical muscle than the soleus muscle. That is the number of terminals that can be supported by a motoneurone in the lumbrical muscle is much greater than the number that can be supported to the soleus muscle, so that there is an intrinsic withdrawal of synaptic terminals in the latter muscle, that is a
254
HISTORY OF THE SYNAPSE
Fig. 11.11. (A) The size of synaptic terminals possessed by the collaterals of each of 10 motoneurones over a muscle that grew during the loss of polyneuronal innervation. Each of the 10 motoneurones had 20 ± 8 collaterals distributed at random over 45 muscle cells, but only the terminals on 25 muscle cells were allowed to participate in synapse formation and elimination up to 8 days (two days postnatal) when the terminals on an additional 10 muscle cells were activated. The terminals on all 35 muscle cells then competed for sites up till 16 days (10 days postnatal) when the terminals on the final 10 muscle cells were allowed to participate in synapse formation, a, gives the results for 25 muscle cells at eight days; b, for 35 muscle cells at 16 days; c, for 45 muscle cells at 24 days and d for 45 muscle cells at 53 days. The parameters used in the model were the same as those given in Fig. 11.7(A) (from Fig. 3 in Bennett & Robinson, 1989) (B) The loss of polyneuronal innervation in muscle showing substantial growth in the postnatal period observed experimentally and compared with the model given in (A). A, Gives the time course of development of the total number of muscle cells in the rat lumbrical muscle; each filled circle shows the mean ± S.D. for five to ten muscles (data from Fig. 6 in Betz et al., 1979). The broken lines give the approximations used in determining that postnatal muscle growth increases the muscle size from 500 to 700 and from 700 to 900 at 2 and 10 days postnatal respectively; this formed the basis of increasing the size of the muscle in the model in (A) from 25 to 35 and from 35 to 45 at 2 and 10 days postnatal, respectively. B, Gives the time course of reduction in motor unit tension (as % of total tension) during postnatal development; each filled circle shows the mean ± S.D. for 12–25 lowest threshold motor units (data from Fig. 5 in Betz et al., 1979). O, Mean ± S.D. of motor unit tensions from the model in A, in which only terminals with an Snm >0.05 were considered capable of generating a twitch. C, Estimates of average motor unit (mean ± S.E.) in lumbrical muscles of rats of different ages; the motor unit size remained approximately constant at ca 120 muscle cells for the first 10 days after birth, and thereafter declined to the adult level (of about eighty-seven muscle cells; data from Fig. 8 in Betz et al., 1979). O, Results in which only terminals with an Snm >0.05 were considered capable of generating a twitch; these results from the model have been normalised to the observed values at 0 to 2 days.
withdrawal that is independent of competition (Fig. 11.12B). There is then a component of terminal withdrawal that is associated with the fact that different muscles have different ratios of the initial number of terminals to the value of T0/R0.
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
255
11.5 Elimination of polyneuronal innervation during reinnervation of muscles described by dual-constraint theory 11.5.1 Loss of polyneuronal innervation During normal impulse traffic Synaptic sites on mature muscle cells are innervated by different numbers of synaptic terminals, and this varies between muscles of a given species as well as between species. Considerable care must be taken in the estimation of the number of terminals at sites using physiological criteria, as often the terminals release so few quanta of transmitter that they are difficult to observe; such terminals certainly go undetected if tubocurarine is used. Taxt et al. (1983) have shown that, besides the normal large endplate potentials (e.p.p.s), very small e.p.p.’s are present in the rat soleus muscle; these are only detectable if the e.p.p. is made subthreshold using a cut muscle preparation which exaggerates the small e.p.p.s in relation to the larger e.p.p.s. Herrera (1984) has used a different approach to avoid the suppression of small e.p.p.s with tubocurarine: formamide destroys excitation-contraction coupling and allows the unsuppressed e.p.p.s to be observed; using this method he showed that the frog sartorius muscle receives a much higher level of polyneuronal innervation (20%) than that determined in the presence of tubocurarine (8%). With these qualifications in mind, most mammalian skeletal muscles have been shown to possess a single synaptic site innervated by a single axon (Brown & Matthews, 1960; Bennett & Pettigrew, 1974a; Brown et al., 1976). The major exception discovered so far is the rat lumbrical muscle: 30% of the synaptic sites in this muscle receive a polyneuronal innervation (Taxt, 1983). Mature amphibian muscles have variable numbers of polyneuronally innervated synaptic sites: 20% of fibres in the frog iliofibularis muscle receive a polyneuronal innervation (Luff & Proske, 1979); 47% in the frog piriformis muscle (Elizalde et al., 1983); 23% in the toad glutaeus muscle (Malik & Bennett, 1987) and about the same in the toad extensor digitorum longus IV (Ridge & Thomson, 1980). Direct measurement of the extent of polyneuronal innervation of a muscle with intracellular impalements at synaptic sites during the reinnervation process reveals that about half the muscle cells initially receive a polyneuronal innervation (McArdle, 1975; Taxt, 1983). As noted above, the rat lumbrical muscle normally possesses a very high polyneuronal innervation of about 30%; after reinnervation the polyneuronal innervation reaches ca 60% and then declines to normal within 25 days of the muscle nerve crush (Fig. 11.13AA). This effect can be modelled if each of the motoneurones forms 10±4 collaterals over a group of 18 adjacent muscle cells 3 days after the muscle nerve is crushed (Fig. 11.13AA). The number of collaterals formed by motoneurones in this model is less than that used in the developmental model (namely 20±8, see for example Fig. 11.7AA); this has the effect of decreasing the competitive differences which occur between terminals at synaptic sites during reinnervation compared with development. The terminals of motoneurones in the present model of reinnervation tend more nearly to have the same amount of T molecules and this has the effect that many synaptic sites retain their polyneuronal innervation (Fig. 11.13AA). Similar results have been obtained for reinnervated synaptic sites in the frog sartorius muscle (Werle & Herrera, 1988). The rat lumbrical muscle is, however, exceptional in possessing such a high level of polyneuronal innervation (~30%), as most mature mammalian muscles probably have <5%. During reinnervation of this latter kind of muscle, such as the rat extensor digitorum longus, polyneuronal innervation is very high early during the reinnervation process and then declines over the succeeding 4 weeks >5% (McArdle, 1975; Fig. 11.13AB). This has been modelled by allowing motoneurones to form only very few collaterals (namely 5) and to give the trophic molecules and their receptors topographical information. Without this information the polyneuronal innervation remains high at equilibrium instead of being mostly eliminated; with more than a small number of collaterals the maximum polyneuronal innervation reaches values between 60% (as in Fig. 11.13AA) and 100% rather than the smaller value of 30% (see Fig. 11.13AB). Whether motoneurones that reinnervate mature muscles have far fewer collateral sprouts than do motoneurones during development, and whether some mature mammalian muscles possess fibres that have different kinds of R molecules allowing for topographical projections to be established remains to be investigated. During a decrease in impulse traffic Polyneuronal innervation increases to high values in reinnervated muscles that have been paralysed with tetrodotoxin applied to their nerves; this excess innervation then fails to retract. In the rat lumbrical muscles, application of the toxin to the sciatic nerve at the time of peak polyneuronal innervation (10 days after reinnervation commences) increases the extent of
256
HISTORY OF THE SYNAPSE
Fig. 11.12. (A) Degree of innervation in 6 week old muscles that were partially denervated at birth. The graph shows that there is an approximately proportional relationship between the number of motor units remaining after partial denervation and the number of fibres receiving innervation. This suggest that all motor units reach the same size. Results are for the mouse soleus that has 600 fibres and before partial denervation 20 motor units. X, Theoretical results and , Findings of Fladby & Jansen (1987). (B) Simulation of the development of motor unit size in the absence of competition from other motoneurones. Muscles were partially denervated at birth, so that only a single motor unit survived in each case. The effect of intrinsic withdrawal is seen for the rat soleus muscle only. At 6 weeks the size of the soleus motor unit is close to its final value. (From Fig. 4(b) and 5 in Rasmussen & Willshaw, 1993).
polyneuronal innervation from 50% to 80% (Taxt, 1983); there is then no loss of polyneuronal innervation back to the control levels of 25%. Neuromuscular blockade by botulinum toxin in mature rat soleus, extensor digitorum longus and peroneus tertius muscles, leads to the formation of terminal sprouts but does not induce polyneuronal innervation in these muscles (Brown et al., 1981); imposed muscle stimulation can prevent this sprouting (Brown et al. 1977). Similarly, if part of the nerve supply to the rat hind-foot muscles is paralysed by application of tetrodotoxin, motor nerve sprouting occurs from the terminals of the unblocked axons, but these sprouts fail to form synapses so that the extent of polyneuronal innervation in the muscles remains very low (Betz et al., 1980).
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
257
Blocking impulse traffic in motor nerves, following a muscle nerve crush, prevents the elimination of polyneuronal innervation that is established early during reinnervation (Taxt, 1983, Fig. 11.13B). This can be modelled by removing the autocatalytic reaction. As has been shown previously, this has the effect of greatly slowing the reaction and hence the competitive elimination of terminals; polyneuronal innervation then remains for very long times (Fig. 11.13B). Following partial denervation Partial denervation of neonatal muscles does not lead to the formation of nerve sprouts from the axons of the remaining innervated muscles in rats (Betz et al., 1980) although it does in frogs (Morrison-Graham, 1983). However, such partial denervation can arrest the loss of polyneuronal innervation by the remaining axons (Betz et al., 1980). In partially denervated rodent soleus, peroneus tertius and lumbrical muscles, the remaining motor units form collateral sprouts and innervate the denervated muscle fibres (Thompson, 1978; Brown & Ironton, 1978; Ribchester & Taxt, 1983); leaving only three out of 11 motor units intact gives complete innervation of the muscles in 12 days. If reinnervation occurs at ca 14 days there is a decrease in the size of the sprouted motor-units, with previously denervated synaptic sites transiently occupied by the terminals of both sprouts and the reinnervating axons (Guth, 1962); most of this polyneuronal innervation is eliminated in the following few weeks, although 10% or so might remain (Brown & Ironton, 1978). If, however, the crushed axons return much later than 14 days then the sprouted units can maintain their synaptic connections (Thompson, 1978). We have seen that blocking impulses in motor nerves leads to sprouting in the muscle without synapse formation. Blocking impulses in reinnervating nerves and muscle leads to an excess polyneuronal innervation which is not removed. Blocking impulses in the remaining axons with tetrodotoxin following a partial denervation some 10 day earlier leads to sprouting from the reinnervating axons in the rat lumbrical muscle with polyneuronal innervation of sites by both blocked and unblocked axons (Ribchester & Taxt, 1983); there is then a decrease in polyneuronal innervation as the blocked terminals are removed in favour of the unblocked terminals; eventually ca 6.5% of the muscle is innervated by each motor unit of the unblocked nerve. (This may be contrasted with the results of cutting the previously intact nerve supply rather than blocking its impulse traffic with tetrodotoxin; in this case the reinnervating nerve innervates ca 30% of the muscle per motor unit; this indicates that the absence of an intact terminal is a stronger stimulus for sprouting than its inactivity (Ribchester & Taxt, 1983)). After partial denervation of a muscle, collateral sprouts form which innervate many of the synaptic sites on the denervated muscle cells; there is then progressive innervation of these sites by the original nerves on their return (Brown & Ironton, 1978; Fig. 11.14BA) with a consequent removal of many of the collateral sprouts from the synaptic sites (Fig. 11.14BB). This process can be predicted by a model in which 20±8 collaterals of each of only 5 motoneurones first form terminals; in this way many muscle cells are innervated by only small terminals (Fig. 11.14AA) and on the formation of new terminals by an additional 20±8 collaterals from each of a further five motoneurones (Fig. 11.14AB), many of the small terminals formed by the first set of five motoneurones are eliminated (Fig. 11.14AC); of course, any muscle cells not innervated at all by the first set of five motoneurones are innervated by the second set of five motoneurones (Fig. 11.14AC). The time course of these events is shown in Figs. 11.14AD and AE: adding to the synaptic terminals of a further 5 motoneurones to the muscle, leads over a period of about four weeks to the eventual elimination of many of the terminals formed by the first set of five motoneurones; this occurs because the terminals of the first set of motoneurones are at a relative disadvantage in that they belong to ‘over extended’ motor units; in these the amount of T molecules at the smaller terminals is relatively low compared to the amount which can be distributed to appropriate terminals from the second set of motoneurones; this is a consequence of the large terminals formed by the first set of motoneurones taking up most of their T molecules, and leaving only sufficient for a few small terminals to be formed, namely those that are eliminated on addition of the terminals from the second set of motoneurones. This model predicts the time course and magnitude of the extent to which partially denervated muscles are reinnervated by the returning nerves ( Fig. 11.14BA) and the removal of the excess innervation (Fig. 11.14BB). The soleus muscle in some species is innervated both by the usual soleus muscle nerve as well as by an aberrant nerve. On cutting the soleus muscle nerve, the aberrant nerve to this muscle produces collateral sprouts that innervate many of the denervated synaptic sites; on return of the soleus nerve, many of these sites receive a dual innervation from both nerves which is eliminated over 4–5 weeks (Thompson, 1978; Fig. 11.14C). This process can be predicted by a model similar to that used in Fig. 11.14A except that each motoneurone has eight collaterals. The results show that the maximum convergence of the two nerves at synaptic sites is ca 40% compared with an observed range of between 10 and 30%; this convergence decreases to about zero by 5 weeks after the operation in both the experiment and the model (Fig. 11.14C). The extension by Rasmussen and Willshaw (1993) of the dual constraint model, to allow for differences in the ratio of T0 to R0 for different muscles as determined by the results of neonatal partial denervation experiments, predicts the observations of Ribchester (1988a, b) on partially denervated lumbrical muscle. In these experiments, Ribchester (1988a, b) removed one of the two nerves innervating the muscle, the lateral plantar nerve, and observed that the remaining three or so motoneurones sprouted and innervated the denervated muscle fibres. When the lateral plantar axons returned a few weeks later they were able to successively compete with the expanded motor units and innervate the muscle. These observations on the expansion
258
HISTORY OF THE SYNAPSE
Fig. 11.13. (A) A, Comparison between the polyneuronal innervation of mature muscles following their reinnervation and predictions of a model in which 50 muscle cells are reinnervated by 10 motoneurones each of which can form 10±4 collaterals collected together in groups on no more than 18 muscle cells. , Experimental results for reinnervation of the rat lumbrical muscle; , Controls; and , Predictions of the model; the ordinate was determined from % S.N.=100%×(L+S−LS)/S where L is the tetanic tension the muscle produces by supramaximal stimulation of the lateral plantar nerve alone, S is the tetanic tension produced by supramaximal stimulation of the sural nerve (S.N.) alone and LS is the maximum tetanic tension produced by supramaximal stimulation of both nerves simultaneously (data from Taxt (1983), Fig. 5A). The time of reinnervation in the model was taken as 3 days after the muscle nerve crush. B, Comparison between the polyneuronal innervation of the rat extensor digitorum longus muscle at various times after crushing the deep peroneal nerve ( ) with the predictions of a model in which 50 muscle cells are reinnervated by 10 motoneurons each of which can form 5 collaterals collected together in groups on no more than 8 muscle cells in the presence of topographical information; the time of complete reinnervation of all synaptic sites in the model was taken as 10 days after nerve crush (data from McArdle (1975), Table 1). (B) Comparison between the polyneuronal innervation of mature muscles following their reinnervation in the absence of impulse activity and the predictions of a model in which 50 muscle cells are reinnervated by 10 motoneurones each of which can form 10±4 collaterals collected together in groups on no more than 18 muscles cells in the absence of the autocatalytic effect. , Experimental results for reinnervation of the rat lumbrical muscle and , predictions of the model. (Data from Taxt (1983), Fig. 5B).
and retraction of the intact nerve supply on the removal and return of the lateral plantar nerve are well predicted by the modified model, as shown in Fig. 11.15.
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
259
Fig. 11.14. (A) The size of synaptic terminals during the reinnervation of partially denervated muscle. (A) Distribution and size of terminals at 18.7 days formed by five of 10 motoneurones (numbers 1–5) following the removal of innervation by the remaining 5 motoneurones (numbers 6–10) at 0 days; each of motoneurones 1–5 possessed 20 ± 8 (±S.D.) collaterals; note that each motoneurone forms only large terminals (as for neurones 2 and 5 taken as no collateral sprouting following denervation) or a mixture of large terminals and small terminals (as for neurones 1, 3 and 4 taken as forming collateral sprouts). (B) Results 1 day after the synaptic terminals of reinnervating motoneurones (numbers 6 to 10) have been established at synaptic sites; each of these motoneurones was allowed to form 20 ± 8 collaterals on the muscle cells and as a consequence 5 of the muscle cells receive a polyneuronal innervation at this time, (c) Shows that 33.3 days after the reinnervation of the partially denervated muscle only 1 cell receives a polyneuronal innervation and 20 of the 45 muscle cells now uniquely possess a terminal from a reinnervating motoneuron. (D) and (E), the time course of changes in size of synaptic terminals on muscle cell number 14 (Sn14) and muscle cell number 28 (Sn28). (D) gives the decrease in size of collateral terminals on muscle cell 14 belonging to motoneurones 1 and 4 after the reinnervation of the synaptic site by motoneurone 9 at a time taken as t=0. (E) Decrease in size of a collateral terminal on muscle cell 28 belonging to motoneurone 2 following reinnervation of the synaptic site by motoneurones 6, and 9; note that eventually all the reinnervating terminals except that from motoneurone 9 are eliminated.
11.5.2 Reestablishment of mature motor-unit sizes During normal impulse traffic Reinnervation of mature muscles, after crushing the muscle nerve, leads to a period of enhanced polyneuronal innervation in both amphibia and muscles. This is followed by a period in mammals during which much of the polyneuronal innervation is
260
HISTORY OF THE SYNAPSE
Fig. 11.14. (B) Comparison between the innervation of the mouse peroneus tertius muscle following its partial denervation observed experimentally and that determined by the model in A. A, , Percentage of the total muscle twitch tension generated by reinnervating axons in the peroneus tertius muscle at various times after crushing either spinal ramus L4 or L5 at their point of exit from the spinal column (data from Brown & Ironton, 1978; Text-Fig 8B); , predictions according to the model in (A), in which motoneurones 6–10 reinnervating a muscle are in competition with motoneurones 1–5; in this model each motoneurone had 20±8 collaterals for innervation. B, , Percentage of the total muscle twitch tension generated by the intact sprouting axons in peroneus tertius muscles at various times since partial denervation; , results for muscles in which re-innervation by the severed root had not occurred (data from Brown & Ironton, 1978; TextFig. 3B); , predictions according to the model in (A), in which motoneurones 1–5 are at first allowed to innervate a muscle in the absence of motoneurones 6–10, and then withdraw some of their processes on the return of motoneurones 6–10. In both A and B, the commencement of reinnervation in the model has been taken as 23.5 days after the operation, when reinnevation is established experimentally. (C) The innervation of the rat soleus muscle, which receives an innervation from both an aberrant nerve as well as the soleus nerve, following its partial denervation by cutting the soleus nerve. Experimental results are compared with a model in which each motoneurone had 8 collaterals for innervation and these were grouped together on 10 muscle cells. , Experimental results (data from Thompson (1978), Table II) and , results from the model in which soleus nerve contained the axons of motoneurones 4– 10 and the aberrant nerves axons of motoneurones 1–3. The model was similar to that given in A except for the above qualifications. The percentage convergence is defined as the number of muscle cells innervated by both the soleus and aberrant nerve expressed as a percentage of muscle cells innervated by the soleus nerve. The time of reinnervation in the model was taken as 6.7 days after the return of the soleus axons (4–10) and this is equivalent to 10 days in the figure.
eliminated until the normal low level is reached (Taxt, 1983; Fig. 11.16B). In the rat lumbrical muscle, which as noted above is innervated by both the sural and lateral plantar nerve, there is an increase in the range of motor-unit sizes early during the reinnervation process until by ca 4 weeks when the mature motor-unit sizes are reestablished (Fig. 11.16B). These changes can be predicted by a model in which each motoneurone forms only 10±4 collaterals with terminals grouped together in the muscle (Fig. 11.16A). In this case the return of the normal range of motor-unit sizes over three weeks to between 5 and 10% is well predicted by the model (Fig. 11.16B). However, these results are not predicted if each motoneurone gives rise to a
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
261
Fig. 11.15. Plasticity of the pattern of connections in the simulation of adult rat lumbrical muscle. The adult muscle was partially denervated so that only four motor units remained of the N=12 that are normally present. These four motor units are assumed to have sprouted to innervate the entire muscle. Motor unit sizes are shown for the four intact units. When the injured motoneurones grew back, they recovered control over some muscle fibres, at the expense of the undamaged neurones. In the graph it is seen that all motoneurones have terminals, and motor units 1–4 have shrunk (from Fig. 6 in Rasmussen & Willshaw, 1993).
relatively large number of collaterals, such as the 20±8 used in the modelling of terminal formation during normal development: in this case the motor unit sizes decrease to much smaller values. In the absence of impulse traffic If the rat lumbrical muscle is reinnervated by its muscle nerve, after impulse traffic in the nerve has been blocked with tetrodotoxin, then the range of motor-unit sizes is very large compared with normal and remains so for over 3 weeks (Taxt, 1983; Fig. 11.17). This can be predicted by the same model used in the reinnervation case above except that the autocatalytic reaction is removed. In this case, as during development (11.1) there is very little decrease in the number of terminals at synaptic sites and the motor units do not decrease much in size (Fig. 11.17). In truncated muscles The reinnervation of muscles that have been truncated by removing half of the muscle cells leads to the formation of smaller motor units than normal. The terminals of these smaller motor units probably release larger amounts of transmitter than in controls, as the force produced by the units does not decrease on a decrease in the calcium concentration nearly as much as do motor units in reinnervated whole muscles (Herrera & Grinnell, 1985; Fig. 11.18B). This is interpreted as showing that most of the larger releasing terminals in the truncated muscle maintain a suprathreshold synaptic potential when the calcium concentration is lowered, whereas those in the whole muscle do not. If larger releasing terminals are physically larger (see Bennett & Raftos, 1977; d’Alonzo & Grinnell, 1985; Bennett et al., 1986b) then it is of interest to compare the size of terminals generated in a model for which a truncated muscle is reinnervated with the size of terminals generated in a model for which a whole muscle is reinnervated (Fig. 11.18A). If each of 10 motoneurones form 20 ± 8 collaterals during reinnervation of 50 muscle cells, then the final terminal size is on average marginally less than that when 10 motoneurones form 10 ± 4 collaterals during reinnervation of 25 muscle cells (Fig. 11.18A). This results because the smaller motor units have more T receptor per terminal and are therefore able to participate in the reaction producing S more effectively. However, the model fails to quantitatively predict the extent of increase in the mean percentage twitch tension in the truncated muscle, consequent on an increase in transmitter release at terminals in the muscle (Fig. 11.18B); this probably occurs because other factors besides synaptic terminal size determine the extent of transmitter release (see Bennett et al., 1986b).
262
HISTORY OF THE SYNAPSE
Fig. 11.16. (A) Size of synaptic terminals possessed by 10 motoneurones, each possessing 10±4 collaterals collected together into two groups of no more than 18, during reinnervation of 50 muscle cells. (A) Distribution and size of terminals at 8 days after the first synaptic contact is made and (B) Distribution 67 days after the first synaptic contact. These results may be compared with those in Fig. 7AA and 7AB in which each motoneurone possessed 20±8 collaterals during the initial reinnervation process. (B) Comparison between motor unit sizes in the sural nerve (s.n.) of mature lumbrical muscles following their reinnervation and predictions of the model given in (A). , Experimental results for reinnervation of the rat lumbrical muscle; , controls; and , predictions of the model; the ordinate was determined from s.n. (motor unit size %)=(S×100)/(Ns×Max tetanic tension) where Ns is the number of s.n. motor units, and S is the tetanic tension to stimulation of the s.n. nerve (data from Taxt (1983), Fig. 7A). In the model, three days has been allowed for the terminals to first reach the synaptic sites after the nerve crush, and the sural nerve contained motoneurones 1–4 and the lateral plantar nerve motoneurones 5–10.
11.6 Elimination of polyneuronal innervation and establishment of topographical maps in muscle described by dual-constraint theory 11.6.1 Muscle cell type specification as a mechanism for map formation There are at least three muscle cell types in birds and mammals, distinguishable on the basis of their myosin isozyme content
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
263
Fig. 11.17. Comparison between the motor unit sizes in the sural neve (s.n.) of mature muscles following their reinnervation in the presence of tetrodotoxin and predictions of a model in which 50 muscle cells are reinnervated by 10 motoneurones each of which can form 10±4 collaterals collected together in groups of no more than 18 muscle cells. , Experimental results for reinnervation of the rat lumbrical muscle in the presence of tetrodotoxin and , give the predictions of the model; the ordinate was determined as given in the legend to Fig. 11.14B (Data from Taxt, (1983), Fig. 7B).
or myosin ATPase (Butler et al., 1982; McLennan, 1983; Phillips & Bennett, 1984): these are type I (slow-twitch), type II (fast-twitch) and type III (slow-graded). In mature muscles, the different muscle fibre types are innervated by different classes of motoneurones. The question arises as to how the appropriate connections are formed between different types of motoneurones and muscle cells during development. As early as 5.5 days in ovo future type III avian muscles possess homogeneous populations of primary myotubes with a myosin isozyme content indicating that they will differentiate into type III myofibres; these primary myotubes are later surrounded by embryonic type III secondary myotubes so as to produce a homogeneous myofibre population in the muscle (Butler et al. 1982; Phillips & Bennett, 1984). An analogous sequence of events occurs during the formation of type I and type II muscles. However, some muscles consist of a heterogeneous mixture of myofibre types in which, for instance, type II myofibres occur in one compartment of the muscle and a mixture of type I and type II myofibres in another compartment. In this case the embryonic muscle will have type lemb primary myotubes in one compartment of the muscle and type IIemb primary myotubes in the other compartment. Most of the secondary myotubes which are subsequently generated are type IIemb, giving a mature muscle with a mixture of type II and type I myofibres in one compartment and only type II in the remaining compartment. These various spatial distributions of primary myotube types arise independently of the nervous system as they still occur following early removal of the neural tube (Butler et al. 1982; Phillips & Bennett, 1984; Fredette & Landmesser, 1991). If motoneurones are of primary types (namely I, II and III) at the time of primary myotube formation, then the problem arises as to the matching of connections between the appropriate motoneurones and myotubes in muscle with a heterogeneous mixture of muscle cell types. There is evidence that the synaptic sites on some mature muscle types favour connections with their appropriate motoneurone type. For example, motoneurones that normally innervate avian fast-twitch muscles (type II) form very few terminals at denervated synaptic sites on slow-graded (type III) myofibres (Van Essen & Jansen, 1974; Zelena & Jirmanova, 1973). Furthermore, amphibian motoneurones that normally innervate slow-graded muscles (type VI) can displace the terminals of motoneurones that normally innervate fast-twitch muscles (type III), if these are first allowed to innervate slow-graded muscles (Elizalde et al., 1983). One possible model to explain these observations is that muscle cells of different myosin isozyme types possess different synapse-formation molecule types. Although there is as yet no evidence that this is the case, it is interesting to explore the ramifications of this idea. A model in which synaptic-formation molecules and their receptors can possess different affinities has been derived. In this case the T molecules in the synaptic terminals have different domains depending on the type of motoneurone; the R molecules in the muscle fibres likewise have different domains depending on the myosin isozyme type of the fibre. Both the different motoneurone types and fibre types may be arranged in such a way as to give a topographical map. Thompson et al., (1984) have shown that motoneurones predominantly innervate either darkly-ATPase staining or lightlyATPase staining myofibres in the 8 day postnatal rat soleus muscle. As most myofibres at this time still receive a polyneuronal
264
HISTORY OF THE SYNAPSE
Fig. 11.18. (A) The size of synaptic terminals possesses by the collaterals of 10 motoneurones when these are used to reinnervate a muscle of 50 muscle cells as in A or after the muscle has been reduced to 25 muscle cells as in B. A: The distribution and size of terminals 26.6 days after the 10 motoneurones had distributed each of their 20±8 collaterals over 50 muscle cells; 26 of the muscle cells still receive a polyneuronal innervation. B: The distribution of terminals 26.6 days after 10 motoneurones had distributed each of their 10±4 collaterals over 25 muscle cells; 20% of the muscle cells still receive a polyneuronal innervation. (B) Comparison between the experimental results for the synapse safety margins in reinnervated half sartorius muscles with those according to the predictions of the model given in A. The histogram bars give the synaptic safety margins in reinnervated half sartorius muscles (experimental E), reinnervated whole sartorius muscles (control C) and unoperated sartorius muscles (normal, N) examined 7–18 weeks postoperatively. Safety margin is defined as the mean nerve stimulus-evoked twitch tension remaining after (Ca)0 is lowered to 1.0 or 0.6 mM, relative to tension generated in 1.8 mM; bars give ± S.E.M. and number of observations is in parenthesis (data from Fig. 1 in Herrera & Grinnell (1985)). The predictions are given by the broken line histograms for (E) and (C) in a (Ca)0 of 1 mM.
innervation it follows that most of the polyneuronal innervation of dark (or light) fibres must come from motoneurones which form motor units of the dark (or light) type. Similar conclusions have been reached by Gordon & Van Essen (1985) from a determination of the contractile properties of motor units in neonatal rabbits. Fibre type recognition seems to play a part in the loss of polyneuronal innervation in the 4th deep lumbrical muscle (Betz et al., 1990; Gates & Ridge, 1992). Reinnervation of either the neonatal or adult soleus muscle gives selective innervation of fast and slow fibres indicating a specificity for synapse formation in this muscle (Fladby & Jansen, 1988; Soileau et al., 1987; see however Soha et al., 1989). However in the soleus muscle any sorting out of motor-synaptic terminals according to fibre type during the neonatal period does not seem
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
265
to contribute to a net loss of polyneuronal innervation during this period (Thompson et al., 1987; Fladby & Jansen, 1990). Since the myosin-ATPase type at this stage is determined autonomously of nervous influences, these observations suggest that motoneurones of a particular type consolidate their terminals on muscle cells of the appropriate type very early during development, before polyneuronal innervation is eliminated from most muscle cells. The model in Fig. 11.19A shows that the sorting out of motoneurone collaterals can occur so that near homogeneous type motor units are present in a muscle of heterogeneous type shortly after synapse formation commences, even though most of the muscle cells still receive a polyneuronal innervation at this time. In this model 20 ± 8 collaterals of each of the 10 motoneurones initially form synapses at random over 50 muscle cells; 5 of the motoneurones (regarded as type II) are given receptors with twice the affinity for alternate sets of 5 muscle cells (namely type II) in the synapse formation matrix than the remaining 5 motoneurones (regarded as type I; Fig. 11.19AA). This results, after 11 days, in a distribution of synaptic terminals for which most motoneurones of a given type synapse exclusively with muscle cells of the same type (Fig. 11.19AB); at this time over two-thirds of the muscle cells still receive a polyneuronal innervation. It is not until some 26.6 days after initial synapse formation that a very low level of polyneuronal innervation is reached (Fig. 11.19AC). Thus the establishment of homogenous units can occur as a consequence of trophic molecules and their receptors containing motoneurone and muscle cell type information, and these units may be established before most polyneuronal innervation is eliminated. This model can be used to predict the observed homogeneity of motor units which emerges during postnatal development of the soleus muscle (Thompson et al., 1984). As Fig. 11.19B shows, there is good agreement between the predictions of the model and the observed homogeneity of both type (11.19BA) and type II (Fig. 11.19BB) motor units at 6– 8 days when most polyneuronal innervation is still present and at 16–20 days when most of it has been eliminated. Synapse formation molecules may have affinities such that the number and type of domain on the T molecules changes in an almost continuous way across the motor synaptic terminals according to the position of their cell bodies in the spinal cord; similarly, the number and type of domain on the R molecules may change in an almost continuous way across the muscle cells according to their position in the muscle. Such a change in the R molecules across a muscle may be associated with an almost continuous change in the myosin isozyme type of fibres across the muscle. If a very gradual change in the domains is made for both the T & R molecules across their respective motor nerves and muscle cells, so that a shallow affinity gradient is developed, then a fine-grained topographical map will be established during synapse formation. Fig. 11.20(A) shows a synapse-formation matrix 2.7 days after terminals have first contacted synaptic sites in which this topographical information is present. Already, at 2.7 days, the emergence of a topographical map can be detected (Fig. 11.20A(a)): most of the largest terminals lie along a right diagonal joining the top right hand corner to the bottom left hand corner. The only large terminals which clearly do not conform to this distribution are in the upper left hand corner of the matrix on muscle cells 1 & 2; these terminals have been allowed to grow, even though they are in a topographically inappropriate pattern, as a consequence of a lack of competition from any other terminals. At 33 days after synapses first form only two muscle cells receive a dual innervation, each motoneurone has mature-size terminals on about 5 muscle cells, and a clear topographical map has emerged between the motoneurone pool and the muscle (Fig. 11.20A(b)). This map is imprecise to the extent that synaptic terminals fail to lie on the right diagonal of the synapse formation matrix. Such a failure arises primarily as a consequence of the random allocation of synaptic terminals of each of the 10 motoneurones: some motoneurones, such as numbers 8 and 10, have nearly all their terminals allocated to inappropriate parts of the muscle, that is in areas of the muscle for which there is a topographical mismatch between their T molecules and the R molecules on the muscle cells; some of these terminals will have sufficient T molecules to allow their synapses to grow in competition with better matched terminals of other motoneurones that do not have many T molecules; this may occur as a consequence of these motoneurones possessing relatively large number of terminals. The time course of elimination of terminals at polyneuronally innervated synaptic sites is the same whether the synapse formation molecules have the small amount of topographical information used to generate the map of 11.20A(b) or no topographical information as for the map of Fig. 11.7AB. In both cases all terminals reach a maximum size about 4 days after initial innervation, and elimination of polyneuronal innervation then occurs over a subsequent three-week period (compare Figs. 11.20A(c, d) with Figs. 11.7AC and 11.7AD respectively). This should not be taken to imply, however, that the introduction of topographical information into the synapse formation molecules does not affect the time course of terminal development and elimination: if a steep affinity gradient is introduced for the T & R molecules, rather than the shallow one used in Fig. 11.20A, then inappropriately matched terminals decrease in size on first making synaptic contact, and the most appropriately matched terminal grows to mature size in < 1 day. In the soleus and extensor digitorum longus muscles there does not seem to be any preferential loss of terminals of motor units acccording to their segmental origins in the spinal cord (Fladby, 1987; Callaway et al, 1989; Thompson, 1983a, b; Gordon & van Essen, 1983; Balice-Gordon & Thompson, 1988). On the other hand the topographical innervation of the rat diaphragm, in which the rostrocaudal axis of the motoneurone pool systematically maps onto the rostrocaudal axis of the muscle surface of the diaphragm, is reinstated after the loss of polyneuronal innervation following a reinnervation experiment (Laskowski & Sanes, 1987; 1988). Furthermore, the topographical map formed between motoneurone pools and the rat intercostal muscles is reinstated following the loss of polyneuronal innervation in reinnervation experiments (Hardman &
266
HISTORY OF THE SYNAPSE
Fig. 11.19. (A) Size of synaptic terminals possessed by the 20±8 collaterals of each of 10 motoneurones over 50 muscle cells when the motoneurones and muscles are matched according to their type. A: Distribution and size of terminals with type information 2.6 days after initial synapse formation; in this case motoneurones 1, 3, 5, 7 and 9 were classified as type II and muscle cells 1 to 5, 11 to 15, 21 to 25, 31 to 35 and 41 to 45 as type II also, with all other motoneurones and muscle cells type I; the affinity of type II motoneurones for type II muscle cells was made twice that of type II motoneurones for type I muscle cells and the reciprocal case for type I motoneurones; at this time 94% of the muscle cells receive a polyneuronal innervation. B: Distribution and size of terminals as in A but 11 days after initial synapse formation (equivalent to about 5 days postnatal); note that at this time a clear selectivity of motoneurones of a given type for their appropriate muscle type has almost completely emerged, with type II motoneurones mostly innervating type II muscle cells indicated as lying within each bracket to the side of the individual matrices; over 64% of the muscle cells still receive a polyneuronal innervation at this time. C: Distribution and size of terminals as in A & B but 26.6 days after initial synapse formation (equivalent to about 21 days postnatal); <13% of the muscle cells now receive a polyneuronal innervation. (B) Comparison between the fibre type composition of motor units in the soleus muscle, at 8 and 16–17 days postnatal and that determined by the model given in (A). The experimental results for numbers of muscle fibres of type II (dark myosin ATPase; hatched bars) and type I (light myosin ATPase; open bars) are shown for motor units of each age (data from Thompson & Sutton, 1984), after normalisation to the number of fibres predominantly at 8 days. (A) Units composed predominantly of type I fibres. (B) Units composed predominantly of type II fibres. The predicted results given by the model are indicated by cross-hatched histograms. These were determined as follows: the number of fibres not predominantly in one kind of unit in the model (say type I in B) was normalised to the number that did predominate in the model giving the cross-hatched histograms at eight days in A and B; for the results at 16–17 days, the number of fibres predominantly in one kind of unit in the model was equated to that observed experimentally, and the number not predominating normalised to this value giving the results indicated by the cross-hatched histograms. Note that there is reasonable agreement between the experimental and predicted results and that the muscle still receives over 63% polyneuronal innervation at six days in the model.
Brown, 1987). A very interesting variation of this experiment was carried out by Wigston and Sanes (1985), who showed that if external intercostal muscles from any particular segmental level are transplanted to the neck and the cervical sympathetic nerve is cut, then the preganglionic nerves from a particular segmental level in the cervical sympathetic preferentially innervate the intercostal muscles derived from the same segmental level. It would appear then that the topographical
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
267
organisation of the innervation of some muscles is associated with a preference of the individual muscle cells for motoneurones at particular segmental levels. Experiments have shown that a fine-grained topographical map emerges during the period of loss of polyneuronal innervation in the mouse peroneus tertius muscle, although it is not yet known if this is associated with fibres of many different isozyme types existing in the muscles. The experiments involve teasing single ventral root axons, noting their position along the rostro-caudal extent of the spinal cord, and determining the spatial extent of their motor-unit contraction (Brown & Booth, 1983; Figs. 11.20CA, B). After the loss of polyneuronal innervation, the spatial extent of each motor unit is such that more rostral axons have motor units that are confined to one margin of the muscle whereas the most caudal axons have motor units that are found at the opposite margin of the muscle; axons from intermediate positions along the spinal cord project to muscle cells lying at appropriate intermediate positions in the muscle (Fig. 11.20CB). It follows that a regression line fitted to the middle of the range of each motor units spatial extent passes through 45° on the axis of the graph relating the spread of individual motor units to the spatial origins of their axons (Fig. 11.20CB). Such is not the case before the loss of polyneuronal innervation: at this time individual motor units have a much greater spatial spread and the middle of the spread for individual units does not lie along a regression line at 45° (Fig. 11.20CA). This indicates that the removal of excess collateral sprouts does not simply involve a motor nerve retracting terminals in a symmetrical way around the parent axon; rather terminals are more often removed from the topographically most unsuitable part of the muscle. These experimental results can be partly predicted from the model given in Fig. 11.20B: before the loss of polyneuronal innervation the motor units are spread out over the 50 muscle cells in positions given in Fig. 11.20CC; after the loss of polyneuronal innervation, which involves the retraction of terminals from the least appropriate parts of the muscle, the regression line fitted to the middle of the range of each motor units spatial distribution passes more nearly through 45° on the axes of the graph relating the spread of individual motor units to the spatial origins of their axons (Fig. 11.20CD). Motoneurone death may have the effect of improving the preciseness of topographical maps. It has been argued that motoneurones, at a critical stage of their development, require sufficient amounts of a motoneurone growth factor made available at synaptic sites on muscle cells in order to survive the cell death period. If the capacity of a synaptic terminal to take up growth factor from a muscle cell is proportional to the presynaptic membrane area of the terminal, then uptake at each terminal will be proportional to Snm for that terminal. It follows that the summation of Snm over all terminals of a motoneurone must be greater than a particular value during the cell death period in order for that particular motoneurone to survive. Motoneurones with a large proportion of terminals in inappropriate parts of a muscle will then be at a disadvantage, as these terminals will in general not grow at the same rate as the more appropriate terminals (see for example Fig. 11.20A(c) and A(d); such motoneurones will therefore have a relatively smaller summed Snm at any particular time. If motoneurones are eliminated at ca 2 days after initial synapse formation (Hamburger, 1975; Harris & McCaig, 1984) due to their summed Snm being less than a set amount then a rather precise topographical map emerges. Fig. 11.20BA shows the random distribution of terminals over muscle cells at 2.7 days, as in Figs. 11.7AA and 11.20A(a), with the difference that motoneurones have been eliminated 8 hours earlier that do not have a summed Snm >2.1; this value was chosen as it has the effect of removing 40% of the motoneurones, a number similar to that observed to degenerate experimentally (Oppenheim & ChuWang, 1983; Harris & McCaig, 1984). It will be noted that a more precise topographical map is evident for the larger terminals in the presence of motoneurone death even at these early stages (compare Fig. 11.20BA with Fig. 11.20A(a)). This is confirmed after most of the polyneuronal innervation has been eliminated (Fig. 11.20B(b)): for instance, motoneurone 10 which had some mature terminals in the most inappropriate part of the map (on muscle cells 1 and 2, see Fig. 11.20A(b) has been eliminated, as has motoneurone 8 which had many inappropriate terminal locations (compare Fig. 11.20BB with Fig. 11.20A(b)). However, the preciseness of the topographical map has only been marginally increased by cell death, with the elimination of a few motoneurones with very poor terminal locations (such as motoneurones 8 and 10). One interesting result of motoneurone death is to marginally increase the number of muscle cells with polyneuronal innervation still present at equilibrium from 2 (Fig. 11.20A(b)) to 6 (Fig. 11.20BB) as well as to increase the number of muscle cells that receive no innervation from 3 to 11. This occurs if smaller motor units have most of their terminals in inappropriate parts of the map, as in the present case (Fig. 11.20A(a)): these motoneurones are then eliminated by cell death and this leaves motoneurones with more nearly similar numbers of terminals; if these terminals compete in parts of the muscle in which they have about the same level of topographical matching (for example, motor nerves 4 and 6 on muscle cell 21 in Fig. 11.20BB) then there will be little to confer advantage on one over the other during competition; dual innervation of the muscle cell will then remain in this case. The time course of elimination of polyneuronal innervation is the same in the presence of motoneurone death as in its absence (compare Fig. 11.20BC and 11.20BD) with Fig. 11.20A(c) and 11.20A(d) respectively): in both cases all terminals which are not eliminated by cell death, but which are destined to be eliminated during later competition, grow to reach their maximum size at about 4 days and then decrease in size over the subsequent 3 weeks (Fig. 11.20BC and 11.20BD). This model of cell death has the effect of giving a better prediction to the observations on the emergence of fine-grained topographical maps in some muscles (see Fig. 11.20CC and 11.20CD).
268
HISTORY OF THE SYNAPSE
Fig. 11.20. (A)a and (A)b: the size of synaptic terminals possessed by 20 ± 8 collaterals of each of 10 motoneurones over 50 muscle cells when T and R molecules possess topographical information. The length of each horizontal bar is proportional to the size of the terminal (Snm) at 2.7 days in (a) and 33 days in (b) after initial synaptic contacts form. Note the clustering of terminals in (b) around the right diagonal indicating the emergence of a topographical map. (c) and (d): the time course of changes in size of synaptic terminals on muscle cell number 14 (Sn14) and muscle cell number 36 (Sn36) given in a and b. A shallow affinity gradient was used (from Fig. 5 in Bennett & Robinson, 1989).
11.6.2 The influence of impulse traffic on map formation Competition between nerves for synaptic sites during the loss of polyneuronal innervation can be biased by artificially increasing the impulse traffic in one set of axons to a muscle (Ridge & Betz, 1984): the axons with the increased traffic are favoured in the establishment of a stable innervation at synaptic sites during the loss of polyneuronal innervation. The rat lumbrical muscle receives innervation from both the sural nerve and the lateral plantar nerve, with fewer motor axons in the former than the latter. It is not known whether each of these nerve predominantly innervate one part of the lumbrical muscle; for the purposes of the following argument it will be taken that they do. Stimulation of the sural nerve from postnatal day 6 to 13 increases the size of the sural motor units compared with normal whereas stimulation of the lateral
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
269
Fig. 11.20. (B) A and (B) B: The size of synaptic terminals possessed by 20 ± 8 collaterals of each of 10 motoneurones over 50 muscle cells when T and R molecules possess topographical information and motoneurone death occurs. A motoneurone was eliminated at 2.4 days if the summed size of all its terminals (Sn) was <2.1. Results are given at 2.7 days in A and 26.6 days in B after initial synaptic contacts formed. Note that the clustering of terminals in B around the right diagonal has been refined by cell death. C and D: The time course of changes in size of synaptic terminals on muscle cell number 14 (Sn14) and muscle cell number 36 (Sn36) given in A and B. A shallow affinity gradient was used.
plantar nerve over the same period depresses the size of sural motor units compared with normal (Ridge & Betz, 1984; Fig. 11.21B). These results have been interpreted to mean that the increased impulse traffic in the terminals of sural motor units gives them an advantage in the competition with lateral plantar motor units, so that the former motoneurones are favoured in maintaining their terminals over the latter. This may be modelled by modifying the autocatalytic reaction in the nerve with the raised impulse traffic, namely by changing an in the equations. An alternative model is one in which the increased impulse traffic in nerve n increases the affinity for nerve n; this is equivalent to an increase in Cn in the equations. Fig. 11.21AA shows the results of a model in which the sural nerve (nerves 1–4) was given an advantage by leaving C1 to C4 equal to 1.0 and making the lateral plantar nerve (motor nerves 5 to 6) possess a C5 to C10 each equal to 0.4; these may be contrasted with the results of Fig. 11.21 AB in which the sural nerve (motor nerves 1–4) was disadvantaged with C1 to C4 equal to 0.4 and the lateral plantar nerve (motor nerves 5–10) advantaged with C5 to C10 equal to 1.0. Inspection of
270
HISTORY OF THE SYNAPSE
Fig. 11.20. (C) Comparison between the spatial extent of motor units in developing muscle observed experimentally and according to a model in which T and R molecules have topographical information. A and B show correlations between the site of contraction of motor units in a rat glutaeus muscle (ordinate) and the rostro-caudal position of the motor axon causing the contraction (abscissa); positions in the muscle and root have been normalised; in the muscle, 0 represents the posterior margin of the muscle and 100 the most anterior portion of the glutaeus made to contract by stimulating the whole of the inferior gluteal nerve; in the root, 0 represents the rostral end of L6 and 100 the rostral end of L3; A is for rats 0–1 day and B is 13–17 day old; the regression line is calculated using the midpoint of each unit contraction site; for A the slope is 0.54 with r=0.66 and for B the slope is 0.83 with r=0.83 (results from Brown & Booth, 1983). C and D show correlations between the site of contraction of motor units and the position of the motor axons according to the model given in B; positions along the 1–50 muscle cells and the 1–10 nerve have been normalised; in the muscle, 0 represents muscle cell number 1 and 100 muscle cell number 50; in the nerve, 0 represents nerve number 1 and 100 represents nerve number 10; C is for rats 1 day old and D is for rats 27 day old; the regression lines have been calculated as above and have slope 0.22 with r=0.72 in C and slope 0.80 with r=0.87 in D, E and F shows correlations between the site of contraction of motor units and the position of the motor axons according to a model in which the 20 collaterals of each of 10 motoneurones are distributed over the 22 nearest muscle cells from the random point of axon sprouting (broken vertical lines) and this is accompanied by motoneurone death (continuous vertical lines); positions have been determined as for C and D; E is for rats 3 days before birth and D is for rat 27 days old; the regression lines have been calculated as above and have slope 0.76 with r=0.93 in E and slope 0.91 with r=0.98 in F.
Fig. 11.21A shows that the effect of these changes is to increase the size of sural nerve units relative to controls in sural nerve stimulated preparations and to decrease the size of sural nerve units relative to controls in lateral plantar nerve stimulated preparations. These results are quantitated in Fig. 11.21BB, which may be compared with the observations on the muscles given in Fig. 11.21BA: the predicted relative changes in motor unit sizes are comparable to those observed experimentally although the absolute values are not.
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
271
Fig. 11.21. (A) The size of synaptic terminals possessed by the collaterals of each of 10 motoneurones over 50 muscles cells when the average impulse traffic varies between different motoneurones (so Cn or an take on different values). The T and R molecules in both A and B contain topographical information. A, Distribution of terminal sizes at 18.7 days for C1 to C4 equal to 1.0 and C5 to C6 equal to 0.4. B, Distribution of terminal sizes at 18.7 days for C1 to C4 equal to 0.4 and C5 to C6 equal to 1.0. Note that the result of impulse traffic being higher for motoneurones 1–4 in A than in B is that they innervate more muscles in A than in B. (B) The effects of stimulating one of two nerves innervating a developing rat lumbrical muscle on the size of motor units in the two nerves observed experimentally compared with that predicted by the model given in (A). A, Experimental results and shows: , the mean motor unit twitch tensions for the sural nerve (SN), following stimulation of the nerve from postnatal day 6 to 13; , the mean motor unit twitch tension for the sural nerve in control muscles; , the mean motor unit twitch tension for the sural nerve in lateral plantar nerve (LPN) stimulated muscles between days 6 and 13; s, the mean motor unit twitch tension for the lateral plantar nerve in control muscles; the ordinate gives the tension in millinewtons (all data from Ridge & Betz, (1984), Fig. 3). B, gives the theoretical results from the model in (A) in which C1 to C10 had the value of 0.4 in the ‘control’; nerve numbers 1–4 were taken to represent the SN so that in the simulation for stimulation of the SN, C1 to C4 were given the value 1.0 whilst C5 to C10 were given the value 0.4; nerve numbers 5–10 were taken to represent the LPN so that in the simulation for stimulation of the LPN, C1 to C4 were given the value 0.4 whilst C5–C10 were given the value 1.0; the symbols then give the number of muscle cells (as a percentage of 50) that were innervated (ie. had terminals with a Snm >0. 05) by either nerves 1–4 (SN) or 5–10 (LPN) in the same conditions as in (A); the s.e.m. is given in each case.
11.7 Identification of the synapse-formation molecules 11.7.1 Identification of Agrin as the synaptic terminal (T) synapse formation molecule In 1978 Sanes et al. carried out their classic experiment in which they damaged denervated frog muscles so that the muscle
272
HISTORY OF THE SYNAPSE
Fig. 11.22. The first demonstration that the basal lamina at the synaptic site contains information for triggering synaptic terminal formation. (A) Sketch of the cutaneous pectoris muscles with attached origin and insertion. The nerve enters from the lateral edge. Two rectangular areas have been cut from the muscle on the right, leaving behind a row of muscle fibre segments, a ‘bridge’, between undamaged medial and lateral fibres. In some experiments, a ‘foreign’ nerve (arrow), which normally supplies the forelimb, was implanted near the bridge. Bar, 3 mm (from Fig. 1 in Sanes et al., 1978). (B) Terminal aborisations on myofibres in normal muscle (a) and in a basement membrane sheath in a bridge, four weeks after denervation, muscle damage and X-irradiation (b), camera lucida drawings from whole sections of zinc iodide osmium stained preparations; Axons exit from the perineurium(stippled) and run along the myofibre (a) or its surviving basal lamina (BL) (b). Like terminals in reinnervated undamaged muscle, terminals in the bridge leave the basal lamina and continue to grow, while normal terminals end abruptly on the myofibres surface. Bar, 100 µm (from Fig. 9 in Sanes et al., 1978). (C) Preterminal and terminal proteins of axons reinnervating the irradiated bridge 21 days after the muscle was damaged and denervated; by morphological criteria, terminals have differentitiated. (A), shows axons in the nerve trunk, wrapped by a Schwann cell, contain many neurofilaments and microtubules but few synaptic vesicles, (b), shows a synaptic terminal, apposed to basal lamina of the myofibre sheath, contains numerous vesicles, some of which are focused on an active zone that lies opposite an intersection (arrow) of synaptic cleft and junctional fold basal lamina. Bar is 0.5 mm (from Fig. 15 in Sanes et al., 1978). (D) Synaptic vesicles and cytoplasmic densities of active zones are associated with intersections of synaptic cleft and junctional fold basal lamina in the absence of myofibres. Shown is a tracing of a synaptic terminal from one of 84 electron micrographs to reveal the synaptic basal lamina (indicated by the open arrow) and an active zone (indicated by the filled arrow), together with numerous synaptic vesicles. Note the alignment of the active zone with the secondary synaptic fold of the basal lamina. (From Fig. 17(a) in Sanes et al., 1978).
fibres and synaptic terminals degenerated and the regeneration of the muscle fibres was prevented by irradiating the animals (Fig. 11.22A and B). The regenerating synaptic terminals were observed to contact the basal lamina at the original synaptic sites that could still be identified as a consequence of a number of indicators, including the disposition of cholinesterase. These terminals acquired active zones and synaptic vesicle accumulations that were aligned with the folds of the basal lamina,
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
273
identified as the secondary synaptic folds beneath which AChR would normally be accumulated if the muscle fibres were still present (Fig. 11.22C and D). This experiment radically changed the search for synaptic terminal formation molecules from the plasmalemma of the synaptic site to that of the overlying basal lamina. It seemed to me to be extraordinary that McMahan and his colleagues should have designed an experiment of this kind as there was little indication in the mid 1970’s that the basal lamina should be tested for the presence of synapse formation molecules. I asked McMahan how he conceived such an experiment when he visited Sydney in 1994. He recounted that the frogs had been left accidentally over the holiday period in the basement without adequate food so that their muscles had severely wasted, to the extent that in some cases they were entirely missing. On inspection it was found that the muscles were still ‘innervated’ even though there were no muscle fibres present. Cholinesterase stain showed that the terminals were present at the original synaptic sites, as indicated by the disposition of cholinesterase. As this was known to exist in the basal lamina the latter became a region of major interest in that it was able to sustain for at least a short time a differentiated synaptic terminal. It was then natural to test this supposition out by reinnervating the muscle when synaptic basal lamina was present in the absence of muscle fibres. Hence the Sanes et al., (1978) experiment. McMahan’s laboratory then set out to isolate the synaptic terminal triggering molecule in the synaptic basal lamina. To this end a rich source of such basal lamina was sought in the electric organ of Torpedo. An insoluble proteinaceous fraction from the extracellular matrix caused up to a 20-fold increase in the number of AChR clusters on the surface of cultured myotubes, probably due to the lateral migration of AChR on the surface of the myotubes, as well as generating antibodies that bound the extracellular matrix of frog muscles (Godfrey et al., 1984). Monoclonal antibodies against this proteinaceous fraction from Torpedo were subsequently used to isolate four polypeptides from the electric organ, of which two (150 kD and 95 kD) aggregated both AChR and cholinesterase on myotubes in culture, and were hence given the name ‘agrin’ (Nitkin et al., 1987; Fig. 11.23A and B). The possibility that agrin was factor T, required by the phenomenological experiments described above to be synthesised in motoneurons and transported to their synaptic terminals, was greatly strengthened by the discovery that antiagrin monoclonal antibodies selectively stain the cell bodies of motoneurons which contain molecules that are antigenically similar to agrin. The conclusion reached was that …agrin molecules are most likely synthesised by motoneurons and released from their axon terminals to become incorporated into the synaptic basal lamina where they direct formation of synapses in development and regeneration (Magill-Soc & McMahan, 1988). This conclusion was subsequently strengthened by the further observation that antibodies to agrin prevented the formation of AChR aggregates at the neuromuscular junctions formed by motoneurons in tissue culture (Reist et al., 1992) and that the neurites of embryonic spinal cord neurones deposit agrin-like molecules on embryonic muscle cells in culture (Cohen & Godfrey, 1992), a property that is not shared with dorsal root ganglion cells (Cohen et al., 1994). Wallace (1989), who had worked in the McMahan laboratory, then went on to show that agrin was responsible for the induction of at least six components of the postjunctional membrane, including AChRs, globular forms of acetyl- and butyrlcholinesterase as well as a heparin sulfate proteoglycan in the extracellular matrix. Agrin was shown to concentrate AChR and butyrlcholinesterase by redistribution of existing molecules rather than by changing their synthesis, with the former occurring by lateral migration. On the other hand agrin did not aggregate the heparin sulfate proteoglycan in the extracellular matrix in the absence of protein synthesis, so that the action of agrin is through different mechanisms depending on the molecule being aggregated. Next, an agrin cDNA clone was isolated from electromotor neurones of the marine ray and these used to characterise the corresponding cDNAs from a rat embryonic spinal cord library; these predicted that agrin has 9 domains homologous to protease inhibitors, a region similar to that of domain III of laminin as well as four epidermal growth factor repeats (Rupp et al., 1991). When the cDNA’s from a chick brain library that codes for agrin were isolated a 33 base pair was found in the agrin cDNA which was responsible for its AChR and acetylcholinesterase aggregating activity; the transcripts were found to be present in motoneurones (Tsim et al., 1992). Agrin was also discovered in muscle at about this time, transiently throwing some doubt on the identity of agrin as factor T (Lieth & Fallen, 1993). Isoforms of agrin were then isolated that result from the alternative splicing of the product of the agrin gene and it was soon shown that these muscle (and indeed Schwann cell) isoforms of agrin lacked one (agrin related protein 1) or two (agrin related protein 2) regions in agrin that are required for its AChR aggregating activity (Ruegg et al., 1992). Thus only the agrin isoforms that have AChR aggregating activity are found in neural tissue (Ferns et al., 1993). Agrin isoforms were then shown to exist throughout the central and peripheral nervous systems (Biroc et al., 1993), raising the exciting possibility that agrin isoforms may play a role in the formation of synapses throughout the nervous system (McMahan et al., 1992; Nastuk & Fallen, 1993). This is further supported by the fact that agrin mRNA is expressed by
274
HISTORY OF THE SYNAPSE
Fig. 11.23. Schematic diagrams of agrin-dependent events during neuromuscular junction development. (A) Upper diagram, agrin is synthesised by both motoneurones and myotubes, but only neural agrin has biological activity. Neural a released from motor synaptic terminals and binds to the local basal lamina. Neural agrin interacts with a receptor named MASC on the surface of differentiated skeletal myotubes, the agrin-MASC complex recruits and activates the MuSK tyrosine kinase. Tyrosine phosphorylation initiates clustering of post-synaptic proteins including AChRs, erbB, and rapsyn. Motoneurones also release neuregulins grin is that interact with clustered erbB receptor tyrosine kinases to initiate a second signaling pathway. The two signaling pathways induce changes in sub-synaptic nuclear transcription and initiate retrograde signals from muscle to the presynaptic nerve terminal that stops axon growth and initiates presynaptic differentiation. Lower diagram, schematic representation of one of several possible models of the MuSK receptor complex for agrin, depicting requirements for a MASC (myotube-associated specificity component) and possible interactions to additional components that may be required for signaling or coupling to various effectors or substrates; these couplings may be mediated extracellularly (for example, via agrin binding to the dystroglycan complex) or intracellularly (for example, via interactions of the SH2 domain-containing proteins to phosphorylated tyrosines on MuSK) (from Fig. 1 in Kleiman & Reichardt, 1996). (B) Schematic representation of the role played by the Agrin/ MuSK signaling system. MuSK appears to activate signaling cascades that are responsible for all aspects of neuromuscular junction formation, including postsynaptic organisations, synapse-specific transcription, as well as presynaptic growth and differentiation (perhaps by regulating the elaboration of retrograde signals, or simply by promoting synaptic activity that down-regulates production of sprouting factors). MuSK activates at least two independent pathways, one that is rapsyn dependent and leads to AChR and dystroglycan clustering, and one that appears to be rapsyn independent and results in synapse-specific transcription. In the model depicted, synapse-specific transcription is presumed to involve clustering of ErbB receptors. However, it has recently been shown that synapse-specific transcription occurs deficient mice although they lack clustered ErbB3 receptors, indicating that synapse-specific expression does not require clustering of ErbB3 in particular; it may still involve clustering of other ErbB receptors (e.g. ErbB4), or alternatively it may depend on ErbBindependent pathways (from Fig. 7 in DeChiara et al., 1996). in rapsyn(C) Table showing a comparison of neuromuscular junction phenotypes in animals with mutations in neural agrin, MuSK, S-laminin, Neuregulin, and ErbB-2 (from Table 1 in Kleiman & Reichardt, 1996).
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
275
neurones distributed througout the adult central nervous system, with four alternatively spliced agrin mRNA’s (namely agrin 0, 8, 11 and 19) found in the spinal cord and agrin 11 localised to the forebrain (O’Connor et al., 1994; Hoch et al., 1994). Alternative splicing of chick agrin mRNA at two sites, named A and B, gives eight isoforms of agrin of which four are expressed in motoneurones (Gesemann et al., 1995). That agrin is the principal neurone signal for inducing the formation of the synaptic site was shown rather definitively by an experiment in which agrin expression plasmids were injected into the extrasynaptic region of muscle fibres resulting in the formation of ectopic postsynaptic membrane, with AChR and acetylcholinesterase (Jones et al., 1997). The Agrin Hypothesis (McMahan, 1990) was amply justified when the first mouse muscles of agrin deficient mutants were examined. These possessed markedly reduced AChR aggregates both in number and size as well as density (Gautam et al., 1996; Fig. 11.23C). The additional interesting observation was made that intramuscular nerve branches and presynaptic differentiation were abnormal indicating an impaired retrograde signalling from the defective apparatus. Trophic factors such as transforming growth factor beta and platelet-derived growth factor might provide such retrograde signals in normal muscle, as the synaptic basal lamina can act as a reservoir for such factors (Patthy & Nikolics, 1994). This occurs because the Nterminus of agrin contains nine follistatin-related modules that have the ability to bind these trophic factors, thus providing a matrix-bound concentration of them that could act in the formation and maintenance of appropriate nerve branching and presynaptic differentiation (Patthy & Nikolics, 1993). The other prime candidate as a retrograde signal for terminal differentiation is the glycoprotein s-laminin (laminin beta 2), which is selectively associated with the synaptic basal lamina (Hunter et al., 1989). 11.7.2 Identification of MuSK as the key molecule in the muscle synapse-formation molecule complex (R) In 1991, Wallace et al. (1991) made the very important observation that agrin causes an increase in tyrosine phosphorylation of the beta subunit of AChR ca 30 min after adding agrin to myotube cultures. As blocking the agrin induced phosphorylation of the gamma and delta subunits of AChR had no effect on its ability to aggregate the AChR, it would appear that the aggregation effect is through the agrin-induced tyrosine phosphorylation of the beta subunit. Thus nerve-induced phosphorylation of the AChR is mediated by agrin (Qu & Huganir, 1994). Staurosporine, an antagonist of both protein serine and tyrosine kinases, blocks agrin-induced phosphorylation of the AChR beta subunit as well as AChR aggregation, emphasising the role of tyrosine phosphorylation of the beta subunit in the aggregation process (Wallace, 1994; Ferns et al., 1996) and the action of agrin in activating a protein tyrosine kinase that regulates ‘the formation and stability of AChR aggregates, apparently by strengthening the interaction between AChR and the cytoskeleton’ (Wallace, 1995). This receptor tyrosine kinase (RTK), specific to skeletal muscle, was subsequently isolated by Valenzuela et al., (1995) and termed ‘MuSK’ for muscle specific kinase. It is expressed in developing muscle cells, is then down-regulated as these mature at which time it is localised at the synaptic site, and is greatly increased in mature fibres if they are denervated or if their electrical activity is blocked (Valenzuela et al., 1995; Fig. 11.23A and B). Neuromuscular junctions do not form in mice with targeted disruption of the gene encoding MuSK, with a failure of clustering of AChR and of other postsynaptic proteins, indicating that this receptor is required …to activate the signalling cascade responsible for all aspects of synapse formation, e.g. postsynaptic and presynaptic differentiation (De Chiara et al., 1996; Fig. 11.23C). MuSK is not very abundant compared with other proteins that are clustered at the synaptic site so that it might act catalytically on these or mediate their phosphorylation to lead to clustering. On the other hand MuSK may activate downstream enzymes that are themselves responsible for clustering. 11.7.3 Identification of MASC as the receptor molecule in the muscle synapse formation molecule complex (R) MuSK alone is not sufficient to confer agrin responsiveness on a muscle cell as there is no binding of the extracellular domain of MuSK to agrin (Glass et al., 1996). This suggests that an as yet unidentified myotube-associated specific component (MASC) is required for agrin binding, so that an agrin-MuSK-MASC complex forms (Fig. 11.23A). The concept is that MASC is the agrin receptor, and that it is the agrin-MASC complex that recruits and complexes the MuSK tyrosine kinase that then leads to the cascade of events that form the synaptic site, including the induction of retrograde factors for differentiation of the synaptic terminal. MASC has not yet been identified (Fig. 11.23A). One possibility is that it is related to the dystrophinassociated glycoproteins - and -dystroglycan, which can act as agrin receptors (Sugiyama et al., 1994; Gee et al., 1994; Gesemann et al., 1996; Bowe et al., 1994). -Dystroglycan, which binds agrin, is a constituent of the dystrophin-associated
276
HISTORY OF THE SYNAPSE
complex of proteins that bind the basal lamina to the actin cytoskeleton. As inactive isoforms of agrin as well as active isoforms bind to -dystroglycan, and deletion of the -dystroglycan binding site on agrin does not block the AChR clustering activity of agrin (Gesemann et al., 1996), it is unlikely that -dystroglycan is MASC. Glass et al. (1996) also argue against the idea that alpha-dystroglycan is MASC. They note that rapsyn is a key component of the subsynaptic cytoskeletal complex that anchors the AChR and dystroglycan into clusters (Apel et al., 1997). However, as several aspects of agrin/MuSK-mediated signalling are still intact in mice which lack rapsyn and for which there is no normal dystroglycan clustering but there is MuSK clustering (see Gautam et al., 1995; Apel et al., 1997), it is unlikely that dystroglycan is MASC. The possibility that rapsyn itself is MASC has been entertained. The phenotype of rapsyn mutants is similar to that of agrin and MuSK mutants (Fig. 11.23C). In the rapsyn mutant there is a failure of clustering of AChR and of dystroglycan although AChR gene expression in the muscle nuclei at synapses is nearly normal. There also appears to be normal complements of ACh esterase, agrin and S-laminin in the synaptic basal lamina. It would appear then that rapsyn is an important effector molecule for some but not all targets of the agrin/MuSK activated signalling cascade. Another possibility is that a synapse-specific carbohydrate that is required for the induction properties of agrin through an N-acetylgalactosaminyl-dependent step is the receptor (Martin & Sanes, 1995). It is clearly of great importance to identify MASC in the near future. 11.7.4 Identification of the neuregulin ARIA as an additional T factor for controlling the number of AChRs at the synaptic site The number of AChR at the synaptic site is under a dual control system. Firstly, the subsynaptic muscle nuclei transcribe the AChR subunit genes at higher rates than do the extra-synaptic nuclei. Secondly, once the AChR are inserted into the membrane they form high-density clusters as a consequence of being anchored to the postjunctional cytoskeletal complex (Guatam et al., 1995). The synaptic-site specific transcription of AChRs is most likely due to an acetylcholine receptor inducing activity (ARIA) glycoprotein of 42 kDa that has been isolated from the chick brain on the basis of its ability to increase AChR incorporation in myotubes (Jessel et al., 1979; Usdin & Fischbach, 1986; Fig. 11.23B). ARIA has the effect of increasing the mRNA levels for the AChR subunits -, -, - and -, with the latter being enhanced over ten fold (Martinou et al., 1991). The mRNA for ARIA is found concentrated in motoneurones (Falls et al., 1993; Jo et al., 1995), is expressed in all cholinergic neurones in the central nervous system, with ARIA immunoreactivity localised to the synaptic basal lamina in mature muscle (Goodearl et al., 1995). This localisation assists in the spatially restricted gene expression for the epsilon subunit of AChR in the muscle at the synaptic site (Chu et al., 1995). ARIA and its human homolog heregulin beta 1 are ligands of the ErbB receptor tyrosine kinases, of which ErbB2 and ErbB3 are localised to the synaptic site in adult muscle and ErbB3 to this site in immature muscle (Zhu et al., 1995; Moscoso et al., 1995a). It is the localisation of these ErbB receptor tyrosine kinases at the synaptic site that is likely to be the predominant factor in determining that ARIA acts to increase the epsilon subunit of AChR specifically at the synaptic site. The spatial regulation of AChR numbers is then ultimately determined by the mechanism that concentrates the ErbB receptor tyrosine kinases to the synaptic site. This is very likely to be due to the clustering effects orchestrated by the agrin/MuSK signalling system. 11.7.5 Identification of S-laminin as a retrograde signal for terminal differentiation Agrin and MuSK mutants show major changes in the form of intramuscular nerve branching, with the nerves growing along the whole length of muscles such as those in the diaphragm rather than forming a well-defined branching pattern in the middle of the muscle with a compact band of endplates. In the MuSK mutants synaptic antigens are found throughout the muscle rather than in a discrete band associated with the endplate band and in agrin mutants most of the nerve terminals fail to differentiate with their ususal properties. These observations point to the necessity of the agrin-initiated signalling through MuSK producing a retrograde factor from the muscle that acts on the terminals to promote their growth and differentiation. It appears then that although the action of agrin on the MASC/MUSK complex triggers the cascade of events that leads to the differentiation of the postsynaptic site, there must be a retrograde message(s) from the forming site that leads to the differentiation of the synaptic terminal, promoting its laying down of active zones. Thus as the synaptic site grows it will progressively release retrograde factor from the more recently formed component of the site, so that an overlying terminal is differentiated. A glycoprotein, termed S-laminin (laminin beta 2), a homologue of the beta 1 chain of the glycoprotein laminin, is selectively associated with the synaptic basal lamina at the neuromuscular junction (Hunter et al., 1989) although it is not detectable in synapse-rich areas of the adult brain and spinal cord (Hunter et al, 1972a). The synaptic localisation of S-laminin is due to a 16-amino acid carboxyl-terminal sequence in the beta 2 chain, which also happens to be the site at which S-laminin adheres to motoneurones (Martin et al., 1995). S-laminin is synthesised by muscle cells. The laminin-beta2 gene is preferentially
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
277
expressed by nuclei at the synaptic site, with S-laminin then concentrated in the synaptic basal lamina so that although the RNA encoding S-laminin is detectable extrasynaptically it is more abundant at the synaptic site (Moscoso et al., 1995b). S-Laminin fragments selectively adhere to motoneurones, inhibit neurite outgrowth and appear to act as a stop signal for neurites. In agrin or MuSK mutant mice the intracellular pathway responsible for directing lamininbeta2 to the synaptic basal lamina is interupted. Most importantly, mice with targeted mutations in the S-laminin gene form normal synaptic sites but the synaptic terminal at the sites fails to differentiate properly, with a relative lack of active zones and vesicle accumulations (Noakes et al., 1995). These observations point to S-laminin as a retrograde factor produced by the muscle, concentrated in the synaptic basal lamina, and acting to determine the differentiation of the synaptic terminal. However the mutant phenotypes are not sufficiently disruptive of presynaptic differentiation to suggest that laminin-beta2 is the only retrograde factor required to direct the normal growth and differentiation of the presynaptic terminal. Fig. 11.23C summarises the results for the phenotypes of mice with mutations in genes with products that are found in the neuromuscular junctions. It is clear form the Table that there are two principal anterograde messengers involved in determining the differentiation of the postsynaptic membrane: these are agrin and the neuregulins like ARIA that the nerve deposits in the basal lamina of the muscle cell and which then activate their tyrosine kinases. It is now established that the agrintriggered signal acting through MuSK is the critical pathway for directing the clustering of AChR at the synaptic site, as well as other molecules that characterise this site. 11.7.6 The formation of active zones in the nerve terminal The active zone in mature transmitter release compartments that occur about every 1 µm or so along motor-terminal branches consist of a pattern of two double rows of particles of about 10 nm diameter that are aligned with the junctional folds of the underlying muscle (Ko, 1981); these particles are most likely voltage-dependent calcium channels (see for example, Robitaille et al., 1993). During development the synaptic terminal active zones are formed in a series of steps in bullfrog larvae (Lynch & Ko, 1983; Ko, 1985) as they are in regenerating adult bullfrogs (Ko, 1984). First, individual active zone particles are scattered throughout a release compartment although exocytosis still occurs; second, the particles align themselves into short lines, which become orientated along the axis of the junctional folds as these develop; finally these discontinuous active zones join to form the mature zone. These studies showed for the first time that active zones are not necessary for secretion to occur from release compartments of developing or regenerating neuromuscular junctions. This is also the case in mature release compartments, for in very low calcium concentrations (<0.1 nM) the active zones break up into two or three pieces, which become unaligned with the junctional folds and from which the active zone particles drift free, whilst transmitter release in response to a single impulse remains nearly normal (Meriney et al., 1996). These results, pointing to a lack of correlation between active zone structure and transmitter release by a single impulse, appear to contradict the observations that quantal content per unit length of motor synaptic terminals is positively correlated with amount of active zone per unit length of the terminals (Propst et al., 1986; Propst & Ko, 1987). The alignment of the active zone particles with the underlying junctional folds suggests that the former are anchored to the latter in some way. Sanes et al. (1978) noted that the mature organisation of active zones and the accumulation of synaptic vesicles within release compartments was reconstituted in amphibia when regenerating axons made contact with the synaptic basal lamina after degeneration of the muscle fibres. This indicates that the molecules for directing formation of the active zone are present in the basal lamina. Normal active zones also form at motor synaptic terminals in mutant Drosophilia which lack mesoderm (Prokop et al., 1996), supporting the observations of Sanes et al (1978) on vertebrates. In amphibia, removal of the synaptic basal lamina for as little as one hour using collagenase and protease disrupts the active zone, as it becomes segmented and randomly oriented, with particles drifting out of the zone (Nystrom & Ko, 1988). Taken together, these observations suggest the hypothesis that the active zone consists of calcium channels that are tethered in alignment with the underlying synaptic basal lamina at the junctional folds, by molecules that are provided by the folds; this anchoring is calcium dependent. A good candidate molecule for this anchoring is integrin, which is a transmembrane molecule at the release compartments that is attached to the basal lamina of the synaptic folds (Chen & Grinnell, 1997). Given that normal transmitter release by a single impulse is not affected by the dispersal of the zones (Meriney et al., 1996), the question arises as to the function of the active zones. According to the secretosome hypothesis, exocytosis occurs at sites which consist of a complex of voltage-dependent calcium channel, synaptic vesicle and vesicle associated proteins that are necessary for exocytosis (O’Connor et al., 1993). If the 10 nm particles are the calcium channel within a secretosome, then it may be the secretosome that is normally tethered in the active zone. This being the case, then fragmentation of the active zone amounts to the movement of secretosomes from a state of alignment to being dispersed. This should not make any difference to transmitter release by a single impulse, as it is likely that the secretosomes act independently of each other under an impulse whether they are packaged in an active zone or not (Bennett, 1996; Bennett et al., 1997). The probability of a quantal release is dependent on the total number of secretosomes within a release compartment (Bennett, 1996), rather than whether
278
HISTORY OF THE SYNAPSE
the secretosomes are organised into active zones. This explains away the apparent contradiction, that transmitter release by a single impulse is unaffected by the dispersement of active zone particles, but the average quantal release per unit length of motor synaptic terminal branch is correlated with the amount of active zone per unit length of terminal branch: it is the total number of secretosomes (as measured by total active zone length) that is important in determining the probability of quantal secretion not how the secretosomes are organised. What then is the purpose of active zones? Meriney et al. (1996) found that there was an increase in facilitation and post-tetanic potentiation at release compartments following the dispersal of the active zones. It may be that calcium buffering is organised within intact active zones in such a way as to provide rapid removal of residual calcium after impulses, so removing facilitation and post-tetanic potentiation. Insights into factors that determine the differention of synaptic terminals have been provided by studies on the formation of synapses by motoneurons in Helisoma by Haydon and his colleagues. Two cholinergic neurones, named B5 and B19, show quite distinct requirements for the formation of a secretory apparatus, with B19 only forming synapses and a mature secretory apparatus with appropriate target muscle cells whereas B5 does not discriminate between muscle cells (Zoran et al., 1990). This latter neurone appears to possess an intrinsic program for generating both presynaptic calcium channels and a secretory apparatus before making contact with muscle cells at all, whereas B19 requires contact with the appropriate muscle cell for a few hours before the terminal differentiates in this way (Haydon et al., 1990; Zoran et al., 1991). Following this contact there is also a dramatic increase in the basal calcium level in the neurone, which continues after the connection with the muscle is severed (Zoran et al., 1993). This inductive effect of the muscle on the cytosolic calcium levels in the neurone appears to be due to a cAMP-dependent protein kinase in the muscle, as injection of a peptide inhibitor of cAMP-dependent protein kinase prevents the muscle from enhancing the calcium accumulation in the neurone (Funte & Haydon, 1993). 11.7.7 Acetylcholine receptor redistribution during loss of polyneuronal innervation This section has been concerned with identification of the synapse formation molecules that are involved in the initial laying down of the synaptic site and terminal as well as in the process of transient polyneuronal innervation. The dual-constraint theory provides what its names suggests, namely constraints on the properties of the anterograde and retrograde molecules that are responsible for guiding the transient polyneuronal innervation. The evidence reviewed in Chapter 11 so far points to the agrin-MuSK hypothesis as supplying many of the ingredients necessary for the formation of the mature synaptic site and terminal. These conform to the requirements of the dual-constraint hypothesis as it applies during development. The question now arises as to whether these same molecules can be taken as providing the T and R factors in the dual-constriant hypothesis as it applies to the loss of polyneuronal innervation during the reinnervation of mature muscles. There is evidence that a redistribution occurs in both AChR as well as in synaptic basal lamina during this loss of polyneuronal innervation that is consistent with the idea that agrin and MuSK are still main players in the process of polyneuronal loss in mature muscle. AChR are lost beneath terminals that are being eliminated, with this decrease apparently leading the removal of the terminal, whilst an adjacent stable terminal retains its normal complement of AChR (Rich & Lichtmann, 1989). A causal link between the AChR and a stable terminal is also indicated by experiments in which AChR are removed from a site by perturbing them with anti-AChR antibodies, so that there is a subsequent loss of the overlying terminal (Rich et al., 1994). Given the likelihood that the AChR are under the control of the synaptic basal lamina, then changes in the former indicate alterations in the latter. Indeed it is known that in mature muscle fibres the synaptic expression of AChR genes is probably due to factors that are bound in the synaptic basal lamina (Klarsfield et al., 1989; Fontaine & Changeux, 1989; Brenner et al, 1990; Sanes et al., 1991; Brenner et al., 1992; Jo & Burden, 1992). If this is the case then there are changes in the synaptic basal lamina, containing as it does molecules for triggering the maturation of the nerve terminal, that accompany the loss of polyneuronal innervation in mature muscles. Direct evidence for such changes has been supplied by Ko and his colleagues using peanut agglutinin that recognises the synaptic basal lamina. They showed that there is a change in this basal lamina during remodelling of the nerve terminal in such a way that it is likely that redistribution of the former guides changes in the latter (Chen et al., 1991). Thus during the consolidation of a terminal the synaptic basal lamina grows out several tens of microns, accompanied by the processes of a Schwann cell, beyond the growth of the terminal branches that are following (Chen & Ko, 1994; Ko & Chen, 1996). Although there is still much work to be done, the present observations on changes in the distribution of synaptic basal lamina and AChR at the sites of polyneuronal innervation loss in mature muscle suggest that identifying the T factor in the dual-constraint theory for this loss with agrin and the R factor with MuSK or MASC provides a plausible mechanism for the loss of polyneuronal innervation during the reinnervation of mature muscle.
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
279
11.7.8 Relation between growth factor molecules and synapse formation molecules It is now established that secretory molecules that are expressed in skeletal muscle during embryonic development and which support the survival of motoneurones in culture as well as to some extent in vivo, include some neurotrophin members of the NGF family, namely BDNF, NT-3 and NT-4/5, as well as FGF-5, IGF-I and LIF (Thoenen et al., 1993). Whether any of these are made available to motor synaptic terminals at synaptic sites is not known. However it has recently been shown that BDNF and NT-3 acting on the Trk family of receptor tryrosine kinases on the neurones in Xenopus nerve-muscle cultures promote the formation of Gaussian distributions of spontaneous endplate potentials as well as potentiating the efficacy and reliability of evoked endplate potentials (Wang et al., 1995). In addition, delivery of exogenous BDNF and NT-3 to developing rat muscles causes the transient retention of the polyneuronal innervation of muscle cells in the neonatal period (English & Schwartz, 1995). It seems likely then that growth factors may act as retrograde messengers for guiding the differentiation, growth and and elimination of nerve terminals, as has been previously argued (Bennett & Robinson, 1989). In the context of the present interpretation of the dual-constraint hypothesis, control of the expression of these growth factors is through the agrin-MuSK pathway may occur as has been shown to be the case for the established retrograde factor S-laminin. 11.8 Synapse formation in autonomic ganglia 11.8.1 Site of synapse formation in reinnervated ganglia If synaptic sites on ganglion cells are visualised before and after reinnervation following denervation, then the distribution of synaptic boutons on the cell bodies is generally different after reinnervation (Purves & Lichtman, 1987). These observations suggest that the synaptic site formed during development is not a site of preferred reinnervation, and so does not contain information for synapse formation as does the synaptic site in muscle. 11.8.2 Site of synapse formation in developing ganglia The development of synapses in the superior cervical ganglion of the rat follows a similar time course to that in muscle. Synapses first form between E12 and E13 on these neurone somas as there are no dendrites at this time, and by E14 the segmental innervation of individual neurones resembles that found in the adult (Rubin, 1985). When dendrites do form, beginning at E14, synapse formation is almost restricted to these processes. The number of primary dendrites and the size of their aborisations is matched to the size of the target muscles that they innervate (Voyvodic, 1989), so that neurones innervating larger muscles possess more synapses and are innervated by more axons (Purves & Hume, 1981; Hume & Purves, 1983). The synapses from a particular axon are not randomly allocated over the surface of the ganglion cell dendritic structure, at least not in the rabbit ciliary ganglion, but are limited to a portion of the postsynaptic surface that includes only a subset of dendrites, indicating perhaps that there are restricted postsynaptic domains on ganglion cells for synapse formation by individual axons (Forehand & Purves, 1984). 11.9 Elimination of polyneuronal innervation in autonomic ganglia 11.9.1 Elimination of polyneuronal innervation during development of ganglia There is a loss of innervation of ganglion cells by preganglionic axons during the first few weeks postnatal in the rat as there is of muscle by motoneurones. The submandibular ganglion consists of monopolar neurones that are mostly (75%) innervated by a single axon in mature animals, but by about 5 axons in the neonatal period; this polyneuronal innervation is lost in the first 5 weeks postnatal (Lichtman, 1980). On the other hand the rabbit ciliary ganglion consists of dendritic neurones that receive 2.2 axons on average in the adult but 4.6 axons on average in the early neonatal period, although there is an increase in the number of synapses formed by the diminishing number of axons (Johnson & Purves, 1981). During the neonatal period in hamsters each ganglion cell receives up to about 12 axons, with the subsequent loss of axons reducing the average number of
280
HISTORY OF THE SYNAPSE
segments innervating ganglion cells from 3.7 to 2.8; this might reflect a refinement of the selective innervation of ganglion cells during the period of loss of excess innervation (Lichtman & Purves, 1980). 11.10 Elimination of polyneuronal innervation during reinnervation of ganglia 11.10.1 After complete denervation: establishment of innervation patterns The guinea-pig superior cervical ganglion is innervated by about six axons arising from spinal cord segments C8 to T7. Each ganglion cell is strongly innervated by axons from one or two of these segments, with adjacent segments contributing a synapic influence which diminishes with distance from the dominant one (Nja & Purves, 1977a, b; Fig. 11.24B). If the preganglionic nerve supply to the superior cervial ganglion in mature guinea-pigs is interrupted by freezing the nerve, then the reinnervating axons establish the same pattern of synapse formation on individual ganglion cells, with adjacent segments contributing a synaptic influence which diminishes with distance from the dominant one; furthermore the end organ responses to stimulation of different segmental inputs is appropriately reestablished (Nja & Purves, 1978). If the ganglion undergoes a partial denervation which spares some preganglionic axons arising from each of the spinal cord segments, the pattern of innervation observed in the normal ganglion is partly restored: ganglion cells are predominantly innervated by two strong adjacent inputs and to a successively lesser degree by inputs arising from more distal spinal segments (Henningsen et al., 1985). If, on the other hand, the innervation from spinal cord segments T3 to T7 are cut, then the remaining segments C8, T1 and T2 show a definite pattern of innervation after sprouting has occurred, in which C8 sprouting axons tend to innervate those neurones that are innervated by T1 but not those innervated by T2 (Liestol et al., 1987). Taken together these results suggest that there is a segmental selectivity of neurones for synaptic connections in the ganglia. It has been known for one hundred years that preganglionic axons from different levels of the spinal cord make preferential connections with different classes of ganglion cells, defined in terms of the end organs that they innervate (Langley, 1892, 1900; Fig. 11.24A). However there is no support for the topographical organisation of these neurones in the ganglion according to the end organs that they innervate that would assist in the formation of selective innervation of the neurones as a consequence of the pathways chosen by particular segmental axons as they grow into the ganglion (Lichtman et al, 1979; Purves & Wigston, 1983). Further evidence for such selectivity comes from an ingeneous experiment of Purves et al. (1981). They transplanted thoracic and lumbar sympathetic ganglia from donor guinea-pigs to the bed of an excised superior cervical ganglion in the guinea-pig and determined the segmental innervation pattern from the cervical sympathetic trunk. Mid-thoracic transplanted ganglia were reinnervated more frequently and strongly by axons arising from more caudal thoracic segments than neurones in transplanted superior cervical ganglia, indicating that the neurones retain a specific segmental identity that determines they receive a predominance of axons from the appropriate segmental level. In a variation on this experiment, Wigston and Sanes (1985) transplanted external intercostal muscles from one of several thoracic levels to the neck of adult rats, cut the cervical sympathetic trunk, and then determined the extent of segmental innervation of the muscle sometime later. Muscles from T2 received more innervation from rostral segments whereas muscles from T8 received a predominant innervation from caudal segments. The results suggest that these muscles, like ganglion cells, possess segmental preferences for particular preganglionic neurones, indicating that they possess some topographical information related to their segmental origins. The specificity of segmental innervation by preganglionic nerves of neurones in autonomic ganglia, that appears to be retained in the adult ganglia, is not present between the axons of the ganglia and the end organs they innervate. Thus, if the postganglionic nerve supply from the superior cervical ganglion is crushed the segmental innervation pattern of end organs such as the pupillary dilator in the eye is not restored even after three months (Purves & Thompson, 1979). 11.10.2 After partial denervation After partial denervation of neurones in the guinea-pig superior cervical ganglion, there is sprouting of the residual preganglionic axons to form synapses; on return of the original nerve supply the extent of innervation by the sprouts is reduced as the reinnervating nerves establish synaptic connections (Maehlen & Nja, 1984).
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
281
11.10.3 Elimination of polyneuronal innervation in cross-reinnervated ganglia Cross innervation of the denervated mature guinea-pig superior cervical ganglion with the vagus nerve leads to the formation of synapses on all the ganglion cells although this innervation is weak compared with that of the synapses formed by the reinnervating cervical sympathetic trunk (Purves, 1976). However, if the den-ervated ganglion is reinnervated by both the vagus and the sympathetic trunk at the same time, then both sets of axons innervate at least half of the ganglion cells and these synapses remain stable over at least a period of several months, with the native synapses failing to displace the foreign synapses (Purves, 1976). 11.11 Identification of synapse formation molecules in autonomic ganglia 11.11.1 Agrin as the synaptic terminal (T) synapse formation molecule Agrin is synthesised in the cell bodies of motoneurones and transported from there to motor synaptic terminals where it is secreted into the synaptic basal lamina (Tsim et al., 1992). The question arises as to whether agrin occurs in other neurone types, especially in autonomic ganglia, where it might have an analogous role in forming synaptic sites to that which it carries out in muscle. Several different isoforms of agrin are expressed in neurones of the developing brain before synapse formation, raising the possibility that they participate in the formation of synapses much as they do at the neuromuscular junction (O’Connor et al., 1994; Ma et al., 1994; So et al., 1996; Cohen et al., 1997). Such a possibility is supported by the observation that rat embryonic dorsal horn neurones in culture express the mRNA for the four agrin isoforms B0, B8, B11 and B12 before synapse formation, and that agrin immunoreactive axons are found adjacent to gephyrin, the postsynaptic glycine receptorassociated protein (Escher et al., 1996). Furthermore, punctate anti-agrin immunoreactivity is associated with the cell bodies and processes of retinal neurones, where in most cases they are colocalised with clusters of gephyrin. The agrin isoforms B8 and B11 are found in developing sympathetic neurones (Ma et al., 1995) as well as in the preganglionic neurones of the chick ciliary ganglion, at the time at which AChRs are being organised in the ganglion cells (McAvoy et al., 1996). The ciliary ganglion cells themselves synthesise agrin, the mRNA of which is relatively unaffected by deafferentation of the ganglion although significantly decreased on axotomy, consistent with a role for agrin in organising the synaptic site at the terminals of the ciliary neurones in muscle, rather than that of the synaptic sites on the ganglion cells themselves (Thomas et al., 1995). 11.11.2 Identification of the neural equivalent of MASC in ganglia Dystroglycan is required for the stabilisation of the dystrophin-associated protein complex that links the cytoskeleton to the extracelluar matrix in muscle. However the cDNA for dystroglycan is expressed in very few brain structures and not at all in the cortex (Gorecki et al., 1994). In the cerebellum it appears to provide a linkage between dystrophin-related protein in perivascular endfeet and laminin-2 in the parenchymal basement membrane similar to that described in skeletal muscle (Tian et al., 1996). It is therefore uncertain what role it might play in synapse formation between neurones. 11.11.3 Identification of neuregulins like ARIA in autonomic ganglia Neuregulins, a family of multipotent epidermal growth factor like factors that arise from splice variants of a single gene, may be involved in synapse formation between neurones. For example, neuregulins are expressed in the retina and brain during early development as well as three of the known receptor tyrosine kinases, erbB2/ neu, erbB3 and erbB4/tyro2 (Meyer & Birchmeier, 1994; Chen et al., 1994; Francoeur et al, 1995; BerminghamMcDonogh et al., 1996; Carraway et al., 1997). In the somatic nervous system, ARIA, a neuregulin, is expressed in developing motoneurones, as would be expected given that it is secreted at motor synaptic terminals to act on the expression of AChR in muscle (Loeb & Fischbach, 1997). In the autonomic nervous system, beta neuregulin is expressed in the preganglionic neurones that innervate the ciliary ganglion in the chick, where it is responsible for conferring calcium sensitivity to certain classes of potassium channels in the ganglion (Subramony & Dryer, 1997). This raises the further possibility that the brain neuregulins may act as ARIA does to regulate the level of certain transmitter receptor subunits in the gangion.
282
HISTORY OF THE SYNAPSE
11.11.4 Is S-laminin a retrograde signal for terminal differentiation? S-laminin is not found associated with neurones in the central nervous system of either adult or embryonic brains, although it is present in the pia and capillaries (Hunter et al., 1992a) and may play a role in the differentiation of the neural retina (Hunter et al., 1992b). 11.11.5 Acetylcholine receptors controlled by nerve terminals During development of the ciliary ganglion, AChR levels increase substantially at the time of synapse formation at about embryonic day 6, whereas they are very much reduced if this normal synapse formation is prevented, a modulation that does not occur for several other membrane bound proteins (Arenella et al., 1993). Specifically, it is the alpha 3, alpha 5 and beta 4 AChR subunit transcript levels that appear to be under the control of the innervating preganglionic axons (Levely et al., 1995). When synaptogenesis in the ganglion is complete, the subunit transcript levels cease increasing (Levey & Jacob, 1996). Acetlycholine-sterase activity in the ganglion is also under the control of the innervating preganglionic axons (Chiappinelli et al., 1976). The possibility that the AChR activity is regulated by synaptic terminals in a way that is analogous to that by motor synaptic terminals has been greatly strengthened by the discovery of agrin mRNA in the Edinger-Westphal nucleus that supplies the preganglionic synaptic terminals to the ciliary ganglion; this agrin mRNA appears at the time of AChR organisation at synapses in the ganglion (McAvoy et al, 1996). 11.11.6 Molecules conferring `chemotactic' specificity We have seen that Langley’s discovery of a ‘chemiotatic’ phenomenon in the reinnervation of sympathetic ganglia (Fig. 11.24A) has been confirmed in detail by Purves and his colleagues (Fig. 11.24B). Furthermore, transplantation experiments between muscles found at different rostro-caudal levels and sympathetic ganglia has established that neurones and muscles may share molecules on their surfaces that determine rostro-caudal positions. This has prompted attempts to discover the molecular basis of the mechanism wherebye spinal cord axons use rostro-caudal positional information in their innervation of sympathetic ganglia and intercostal muscles. A monoclonal antibody, ROCA1, has been generated by Patterson and his colleagues (Suzue et al., 1990) that shows a graded rostro-caudal decline in binding in the nerves innervating intercostal muscles (Fig. 11.25A) as well as along the chain of adult rat sympathetic ganglia (Fig. 11.25B). ROCA1 recognises two antigens in the membrane/cytoskeletal fractions of peripheral nerves and ganglia, one of which is a 26 kD protein that possesses an epitope which is predominantly visualised with immunohistochemisty in rostrol nerves and ganglia, indicating that it is probably masked in caudal nerve and ganglia (Kaprielian & Patterson, 1993). Interestingly, the 26 kD protein possesses significant homology with a cell surface protein involved in intercellular signalling in hematopoietic cells called CD9. The mRNA for this protein is expressed in both sympathetic neurones and Schwann cells as well as in hematopoietic cells (Kaprielian et al., 1995). The existence of this protein in Schwann cells is particularly interesting as these cells have been implicated in guiding axons to their target muscles in the peripheral nervous system during development (Noakes & Bennett, 1987; Noakes et al., 1983; Noakes et al., 1988). An antibody generated against CD9, called SMRA1, enhances Schwann cell migration on both axons of cultured dorsal root ganglion cells as well as on those of the sciatic nerve (Fig. 11.25C and D). This regulation of Schwann cell motility appears to be due to CD9 inducing a rise in cytosolic calcium as well as of tyrosine phosphorylation of a number of Schwann cell proteins (Anton et al., 1995). It will be of interest to see if Langley’s idea of ‘chemiotatics’ amounts to the ability of Schwann cells to guide axons of different segemental origins to their appropriate targets. 11.12 Conclusion Tello and Langley erected a conceptual framework at the beginning of this century for the study of the mechanism of synapse formation through their experimental work on the development and reinnervation of mature muscle and autonomic ganglia. Two main ideas came out of this research: one was that transient hyperinnervation occurs at reinnervated synaptic sites; the other was the existence of specificity of reinnervating segmental nerves for particular target sites. It was not until the second half of the century that the challenge of this work was taken up seriously, with the establishment of two further ideas. The first was that the synaptic site in mature muscle contains unique molecular constituents for triggering the differentiation and growth of a motor-nerve terminal. The second was that these molecules are laid down by the exploring motor-nerve itself
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
283
during development. The process of innervation by motor nerves of either mature or developing muscle cells involves a transient hyperinnervation of the cells. It has been shown that the large body of experimentation on the formation and loss of this polyneuronal innervation in developing and mature muscles can be quantitatively accounted for by a theory that incorporates competition of nerves for a postsynaptic R factor and of muscle cells for a presynaptic T factor. This theory is the dual-constraint hypothesis. Identification of the molecules that constitute R and T has been greatly accelerated by the discovery that T factor is lodged in the basal lamina of muscle cells during development. The subsequent identification of T as principally the molecule agrin has opened up the search for its R receptor which has has very recently been narrowed down to the receptor tyrosine kinase MuSK and associated molecules. The attempt to define at the molecular level the ‘chemiotactic’ principal wherebye segmental nerves prefer innervation of specific classes of autonomic ganglion cells as discovered by Langley has not yet been successful. The observation that muscle cells of different segmental origin also show specificity for nerves originating from the same segmental level is important. It opens up the possibility that the study of the innervation of muscle will provide the insights necessary for the elucidation of the specificity of innervation now that a century has passed since Langley’s great discovery.
284
HISTORY OF THE SYNAPSE
Fig. 11.24. Specificity of synapse formation by the regenerating cervical sympathetic nerve. Comparison between the observations of Langley on the cat (Langley, 1897; Table II), shown in (A), and those of a modern investigation eighty years later (Nja & Purves, 1977a.b; on the guinea pig; Table 1), shown in (B). (A) Gives Langley’s results for the control of different spinal nerves of the end organs innervated by the superior cervical ganglion at 2–15 months after cutting the cervical sympathetic. These may be compared with the normal spinal nerve control of these end organs given in (B) as well as in Fig. 11.1(c). (B), the effects of in vivo stimulation of the last cervical and the first seven thoracic ventral roots 3–4 months after section of the cervical sympathetic trunk, compared to ventral root stimulation in normal animals are shown. The results for reinnervation in (B) are much the same as those in (A).
EMERGENCE OF THE CONCEPT OF SYNAPSE FORMATION MOLECULES
285
Fig. 11.25. Evidence for quantitative changes in the distribution of a molecule in sympathetic ganglia and in motor-nerves according to the rostro-caudal origins of the ganglia and nerves in rats. (A), shows longitudinal sections of intercostal nerves from various segmental levels, as indicated, stained with the monoclonal antibody ROCA1. (B), Sections of adult sympathetic ganglia from various segmental levels stained with ROCA1; the levels are rostrocaudal from the top down (namely superior cervical ganglion, stellate ganglion, T5, T9 and L2). (C) and (D) Effects of an antibody that recognises a Schwann cell antigen (SMRA1) on Schwann cell migration. (D) Enhanced Schwann cell migration over adult sciatic nerve substrates in the presence of SMRA1 compared with that in the absence of the antibody shown in (C). (A) and (B) are from Figs. 4 and 2 in Suzue et al., 1990. (C) and (D) are from Fig. 4 in Anton et al., 1995).
Epilogue
It seems appropriate that this work should end with a drawing of the underside of the human brain, made by the third President of the Royal Society, Sir Christopher Wren, in the middle of the seventeenth century. It can be claimed that this is the most significant figure in the neurosciences, even though it was made over three hundred years ago. It was commissioned by Thomas Willis, Professor of Anatomy at Oxford, to illustrate his thesis that it is not the ventricles which give rise to the contents of our consciousness as in seeing, hearing, thinking and remembering, but the cortex. The ventricles had dominated physiological thinking about brain processes for about one and a half thousand years up to the time of Willis. Surely one can argue that a drawing summarising the shift of focus for neuroscience from the ventricles to the cortex holds major significance in the history of neuroscience. Yet each generation, especially it would seem at the beginning of a century, let alone a millennium, believes that it has the priority of insight, has made the greatest contributions to the understanding of how neurons and their synaptic connections function. This view tends to be driven by technological advances and the accelerated accumulation of facts, which this progress stimulates. The present work offers a caution, suggested by the history of our subject, and epito-mised by Wren’s drawing. Often the greatest insights are provided by a combination of sustained logical thought and high skilful experimentation, as in the case of Swammerdam’s experiment disproving the millenniumold concept that substances flow along nerves into muscle to make them swell and shorten, or the experiments of Galvani showing electrical excitability and conduction in nerve and muscle, or Langley’s proof concerning the existence of transmitter receptors. Willis argued cogently in his treatise The Anatomy of the Brain and Nerves that the ventricles could not be the seat of conscious experience as a consequence of his dissections of the cadavers of patients who had suffered serious nervous maladies. In my mind the history of the synapse illustrates that there is no substitute for painstaking elaboration of experiments to which are applied both in their design, execution and interpretation a tenacious adherence to logical arguments that weigh the alternative hypotheses under consideration. John Newport Langley and Bernard Katz are without peer in this regard, and it is to them that this work is dedicated. There are a number of technical breakthroughs that appear in the history of the synapse which led to significant conceptual insights. Among these were the development of the Leyden jar by van Musschenbroek which so helped in establishing the existence of animal electricity; the increase in sensitivity of the galvanometer by Nobili that allowed the negative variation in current of the action potential to be identified; the beautiful device designed by Helmholtz which allowed him to show that the action current travelled at a finite velocity; the Golgi silverstain which when applied to sites of retrograde neuronal degeneration showed that cytoplasmic continuity at the synapse was unlikely; in the twentieth century the invention of the microelectrode by Ling and Gerard which provided the means for showing that transmitter release is quantised and finally the patch-clamp technique by Sakmann and Neher which provided the first analysis of the activity of a single protein in real time, namely that of the transmitter receptor. In some cases, the synaptic experimentalists themselves invented the techniques which allowed a major leap in progress to be made, in other cases not. Galvani did not develop novel instrumentation but rather used a persistent application of logical argument in the interpretation and design of experiments in the face of Volta’s criticisms so that he finally won the day. Is he to be regarded as less than Helmholtz who both designed novel instrumentation and used it to such great effect? Where is the recognition of a Ling and Gerard for their invention of the microelectrode whose application so revolutionised synaptic physiology? The story of the synapse suggests that it is invidious to apportion glory to individuals when our understanding of the nervous system is the collective enterprise of so many. Nevertheless, as much as one wants to contain oneself in a scholarly restraint, the combination of technical inventiveness and experimental design which leads to a paradigm shift in our understanding of nature can but excite a thrill of admiration. Helmholtz produced this in the middle of the nineteenth century when he showed that what had been the psychic pneuma of Aristotle, the corpuscles of Descartes and the electricity of Galvani, conducted at finite velocity along nerves. This beautiful experiment reduced a previously considered mysterious and infinitely fast process, that seemed so appropriate for a nervous system that could give rise to consciousness, to a phenomenon open to further empirical investigation.
287
The arrogance of the scientific enterprise is exemplified in two ways. First there is an inability to realise that the scientific method of enquiry, primarily initiated by Galileo at the end of the sixteenth century and soon thereafter systematised by Locke and Hobbes, is one of but many different ways of understanding ourselves and the world about us, as was emphasised most profoundly by Ludwig Wittgenstein in the first half of the twentieth century. The other is the extent to which great scientists pronounce, especially towards the ends of their careers, that the component of science to which they had contributed is complete. So Macfarlane Burnet, who could be rightly regarded as a foundation figure in both virology and immunology, claimed towards the end of his life that molecular biology, to which he had contributed very significant insights, would never provide any practical benefit to mankind. Perhaps more cynically, one could point to Lord Kelvin at the end of the nineteenth century and his suggestion that the physical explanation of nature was almost complete, this less than ten years before relativity and the quantisation of energy. At the end of the twentieth century we saw much proselytising on behalf of a ‘theory of everything’. However, the history of the synapse over two and a half thousand years points to the never-ending endeavour of human kind to understand how our nervous system works. The caveat is that neuroscience is but one method of probing what it is to be human. Nevertheless, neuroscience offers a seemingly inexhaustible vista of problems which need to be solved in order to assist human beings in their efforts to avoid suffering and lead reasonably happy lives.
References
Numbers in square brackets refer to the pages on which the reference is cited. Abel, J.J., 1899. Ueber den blutdruckerregenden Bestandtheil der Nebenniere, dasEpinephrin. Z.Physiol. Chem. 28, 318– 324. [65] Abrahamsson, H., 1973. Studies on the inhibitory nervous control of gastric motility . Acta. Physiol. Scand. Suppl. 390, 1–38. [123–125] Abrahamsson, H., Jansson, G., 1969. Elicitation of reflex vagal relaxation of the stomach from pharynx and oesophagus in the cat. Acta. Physiol. Scand. 77, 172–178. [113] Abrahamsson, H., Jansson, G. 1973. Vago-vagal gastro-gastric relaxation in the cat. Acta. Physiol. Scand. 88, 289–295. [113] Adrian, E.D., 1924. Some recent work on inhibition. Brain. 47, 399–416. [31] Adrian E.D., Bronk, D.W., 1929. The discharge of impulses in motor nerve fibres. Part II. The frequency of discharge in reflex and voluntary contractions. J Physiol. 67, 119–151. [30] Adrian, R.H., Chandler, W.K., Hodgkin, A.L. 1970. Voltage clamp experiments in striated muscle fibres. J. Physiol. 208, 607–644. [158] Ahlquist, R.P., 1948. A study of the adrenotropic receptors. Am. J. Physiol. 153, 586–600. [66, 67] Aitken, J.T., 1950. Growth of nerve implants in voluntary muscle. J. Anat. 84, 38–49. [233, 235] Akaike, H., 1974. A new look at the statistical model of identification. IEEE Trans. Auto. Con. AC-19, 716–723. [195] Akaike, N., Fishman, H.M., Lee, K.S., Moore, L.E., Brown, A.M., 1978. The units of calcium conduction in Helix neurones. Nature. 274, 379–382. [150, 151] Akers, R.F., Lovinger, D.M., Colley, P.A., Linden, D.J., Routtenberg, A., 1986. Translocation of protein kinase C activity may mediate hippocampal long-term potentiation. Science. 231, 587–589. [228] Aldini, G., 1804. Essal theorique et experimental sur le galvanisme, Vol. 1, Fournier, Paris. Allescher, H.D, Tougas, G., Vergara, P., Lu, S., Daniel, E.E., 1992. Nitric oxide as a putative nonadrenergic noncholinergic inhibitory transmitter in the canine pylorus in vivo. Am. J. Physiol. 262, G695–G702. [127] Allgaier, C., Feuerstein, T.J., Jackisch, R., Hertting, G., 1985. Islet-activating protein (pertussis toxin) diminishes alpha 2-adrenoceptor mediated effects on noradrenaline release. Naunyn-Schmiedebergs Arch. Pharmacol. 331, 235–239. [77] Amin, A.H., Crawford, T.B.B., Gaddum, J.H., 1954. The distribution of substance P and 5–hydroxytryptamine in the central nervous system of the dog. J. Physiol. 126, 596– 618. [92, 118] Anden, N.B., Butcher, S.G., Fuxe, K., 1970. Protection of the neostriatal dopamine stores against reserpine by local treatment with metatyramine. Acta. Pharmacol. Toxicol. 28, 39–48. [92] Anden, N.B., Dahlstrom, A., Fuxe, K., Larsson, K., 1966. Functional role of the nigro-neostriatal dopamine neurons. Acta. Pharmacol. Toxicol. 24, 263–274. [95] Anden, N.B., Roos, B.E., Werdinius, B., 1964. Effects of chlorpromazine, haloperidol and reserpine on the levels of phenolic acids in rabbit corpus striatum. Life Sci. 3, 149– 158. [92, 93] Anderson, M.J., 1986. Nerve-induced remodeling of muscle basal lamina during synaptogenesis. J. Cell Biol. 102, 863– 877. [242] Anderson, M.J., Fambrough, D.M., 1983. Aggregates of acetylcholine receptors are associated with plaques of a basal lamina heparan sulfate proteoglycan on the surface of skeletal muscle fibers. J. Cell Biol. 97, 1396–1411. [242] Andersen, P., Eccles, J.C., 1962. Inhibitory phasing of neuronal discharge. Nature. 196, 645–647. [117–119] Andersen, P., Eccles, J.C., Loyning, Y., 1963. Recurrent inhibition in the hippocampus with identification of the inhibitory cell and its synapses. Nature. 198, 541–542. [40, 41, 211] Andreu, R., Barrett, E.F., 1980. Calcium dependence of evoked transmitter release at very low quantal contents at the frog neuromuscular junction. J. Physiol. 308, 79–97. [162] Andrews, P.L.R., Lawes, I.N.C., 1984. Interactions between splanchnic and vagus nerves in the control of mean intra-gastric pressure in the ferret. J. Physiol. 351, 473–490. [125] Anton, E.S., Hadjiargyrou, M., Patterson, P.H., Matthew, W.D., 1995. CD9 plays a role in Schwann cell migration in vitro. J. Neurosci. 15 (1 Pt 2), 584–595. [282, 283] Apel, E.D., Glass, D.J., Moscoso, L.M., Yancopoulos, G.D., Sanes, J.R., 1997. Rapsyn is required for MuSK signaling and recruits synaptic components to a MuSK-containing scaffold. Neuron. 18(4), 623–635. [274] Aprison, M.H., Werman, R., 1965. The distribution of glycine in cat spinal cord and roots. Life Sci. 4, 2075–2083. [82] Arenella, L.S., Oliva, J.M., Jacob, M.H., 1993. Reduced levels of acetylcholine receptor expression in chick ciliary ganglion neurons developing in the absence of innervation. J. Neurosci. 13(10), 4525–4537. [279] Aristotle., 1984. Generation of Animals. In: J.Barnes (ed.): The Complete Works of Aristotle. Princeton. [1] Aseltine, J.A., 1958. Transform Methods in Linear Systems Analysis. McGraw Hill, New York. [190] Ashley, C.C., Ridgway, E.B., 1968. Simultaneous recording of membrane potential, calcium transient and tension in single muscle fibers. Nature. 219, 1168–1169. [148] Ashley, C.C., Ridgway, E.B., 1970. On the relationships between membrane potential, calcium transient and tension in single barnacle muscle fibres. J. Physiol. 209, 105–130. [148]
REFERENCES
289
Atsumi, S., 1977. Development of neuromuscular junctions of fast and slow muscles in the chick embryo: a light and electron microscopic study. J. Neurocytol. 6, 691–709. [242] Atwood, H.L., Johnston, H.S., 1968. Neuromuscular synapses of a crab motor axon. J. Exp.Zool.. 167, 457–470. [170] Auerbach, L., 1862. Ueber einen plexus myentericus, einen bisher unbekannten ganglio-nervosen Apparat im Darmkanal der Wirbelthiere. E.Morgenstern, Breslau. [120] Auerbach, L., 1864. Fernere vorlaufige mitteilung uber den nervenapparat des Darmes. Arch. Pathol. Anat. Physiol. 30, 457–460. [120] Auerbach, L., 1898. Nervenendigung in der Centralorganen. Neurol. Central bl. 17, 445–454. [17, 18] Augustine, G.J., Charlton, M.P., Smith, S.J., 1985a. Calcium entry into voltage-clamped presynaptic terminals of squid. J. Physiol. 367, 143–162. [158, 159] Augustine, G.J., Charlton, M.P., Smith, S.J., 1985b. Calcium entry and transmitter release at voltage-clamped nerve terminals of squid. J. Physiol. 367, 163–181. [158, 159] Ault, B., Evans, R.H., Francis, A.A., Oakes, D.J., Watkins, J.C., 1980 Selective depression of excitatory amino acid induced depolarizations by magnesium ions in isolated spinal cord preparations. J. Physiol. 307, 413–428. [88, 89, 221] Awapara, J., Landau, A., Fuerst, F., Scale. B.L., 1950. Free gamma-aminobutyric acid in brain. J. Biol. Chem. 187, 35– 39. [79] Bacq, Z.M., 1974. Chemical transmission of nerve impulses. Pergamon Press, Oxford. [67, 108] Bacq, Z.M., 1983. Chemical transmission of nerve impulses. In: ‘Discoveries in Pharmacology. Vol. 1. Psycho- and Neuro-Pharmacology’. M.J.Parnham & J.Bruinvels (eds). pp. 49–103. Elsevier, Amsterdam & New York. [62] Bacq, Z.M., Brown, G.L., 1937. Pharmacological experiments on mammalian voluntary muscle, in relation to the theory of chemical transmission. J. Physiol. 89, 45–60. [59] Bagust, J., Lewis, D.M., Westerman, R.A., 1973. Polyneuronal innervation of kitten skeletal muscle. J. Physiol. 229, 241– 255. [242] Baik, J.H., Picetti, R., Saiardi, A., Thiriet, G., Dierich, A., Depaulis, A., Le Meur, M., Borrelli, E., 1995. Parkinsonian-like locomotor impairment in mice lacking dopamine D2 receptors. Nature. 377, 424–428. [98] Baker, P.P., Hodgkin, A.L., Ridgway, E.B., 1971. Depolarization and calcium entry in squid giant axons. J. Physiol. 218, 709–755. [148, 149, 151] Balcar, V.J., Dias, L.S., Li, Y, Bennett, M.R., 1995a. Inhibition of [3H]CGP39653 binding to NMDA receptors by a P2 antagonist suramin. Neuroreport 7, 69–72. [88] Balcar, V.J., Johnston, G.A.R., 1972. Glutamate uptake by brain slices and its relation to the depolarization of neurones by acidic amino acids. J. Neurobiol. 3, 295–301. [87] Balcar, V.J., Johnston, G.A.R., 1975. High affinity uptake of glutamine in rat brain slices. J. Neurochem. 24, 875–879. [87] Balcar, V.J., Li, Y, Killinger, S., 1995b. Effects of L-trans-Pyrrolidine-2, 4-dicarboxylate and L-threo-3-hydroxyas-partate on the binding of [3H]L-aspartate, [3H]a-amino-3-hydroxy-5-methyl-4-isoxazolepropionate (AMPA), [3H]DL-(E)-2-amino-4-propyl-5-phosphono-3pentenoate (CGP 39653), [3H]6-cyano-7-nitroquinoxaline-2, 3-dione (CNQX) and [3H]kainate in sections of rat forebrain. Neurochem. Inter. 26, 155–164. [88] Baldo, G.J., Cohen, I.S., Van der Kloot, W., 1986. Estimating the time course of evoked quantal release at the frog neuromuscular junction using end-plate current latencies. J. Physiol. 374, 503–513. [185] Balice-Gordon, R.J., Thompson, W.J, 1988. The organization and development of compartmentalised innervation in rat extensor digitorum longus muscle. J. Physiol. 398, 211– 231. [264] Bamburg, J.R., Bray, D., Chapman, K., 1986. Assembly of microtubules at the tip of growing axons . Nature. 321, 788– 790. [242] Banks, B.E., Brown, C., Burgess, G.M., Burnstock, G., Claret, M., Cocks, T.M., Jenkinson, D.H., 1979. Apamin blocks certain neurotransmitter-induced increases in potassium permeability. Nature. 282, 415–417. [127, 131] Bar, PR., Schotman, P., Gispen, W.H., Tielen, A.M., Lopes Da Silva, F.H., 1980. Changes in synaptic membrane phosphorylation after tetanic stimulation in the dentate area of the rat hippocampal slice. Brain Res. 198, 478–484. [225, 228] Barger, G., Dale, H.H., 1910. Chemical structure and sympathomimetic action of amines. J Physiol. 41, 19–59. [65–67] Barnard, E.A., Skolnick, P., Olsen, R.W., Möhler, H., Sieghart, W., Biggio, G., Braestrup, C., Bateson, A.N., Langer, S.Z., 1998. International Union of Pharmacology. XV. Subtypes of gamma-aminobutyric acidA receptors: classification on the basis of subunit structure and receptor function. Pharm. Rev. 50, 291–313. [90] Barnes, C.A., 1979. Memory deficits associated with senescence: a neurophysiological and behavioural study in the rat. J. Comp. Physiol. Psychol. 93, 74–104. [214, 216, 217] Barrett, E.F., Stevens, C.F., 1972. The kinetics of transmitter release at the frog neuromuscular junction. J. Physiol. 227 (3), 691–708. [179, 180] Bartho, L., Lefebvre, R.A., 1995. Nitric oxide-mediated contraction in enteric smooth muscle Archiv. Internat, de Pharmaco. et de Ther. 329, 53–66. [133] Baxter, D.A., Bittner, G.D., Brown, T.H., 1985. Quantal mechanism of long-term synaptic potentiation. Proc. Nat. Acad. Sci. USA. 82, 5978–5982. [230, 232] Bayguinov, O., Sanders, K.M., 1993. Role of nitric oxide as an inhibitory neurotransmitter in the canine pyloric sphincter. Amer. J. Physiol., 264, G975–G983. [128] Bayguinov, O., Vogalis, F., Morris, B., Sanders, K.M., 1992. Patterns of electrical activity and neural responses in canine proximal duodenum. Am. J. Physiol. 263, G887– G894. [128] Bayliss, W.M., Starling, E.H., 1899. The movements and innervation of the small intestine. J. Physiol. XXIV, 99–143. [110, 113, 118, 120, 121, 125]
290
REFERENCES
Bayliss, W.M., Starling, E.H., 1900. The movements and innervation of the large intestine. J. Physiol. XXVI, 107–118. [110, 111, 113, 118, 120] Bayliss, W.M., Starling, E.H., 1901. The movements and innervation of the small intestine. J. Physiol. XXVI, 125–138. [110, 111] Beani, L., Bianchi, C., Crema, A., 1971. Vagal non-adrenergic inhibition of guinea-pig stomach. J. Physiol. 217, 259–279. [117–120] Beaton, J.A., Stemsrud, K., Monaghan, D.T., 1992. Identification of a novel N-methyl-D-aspartate receptor in the rat medial thalamus. J. Neurochem. 59, 754–757. [88] Behar, J., Guenard, V, Walsh, J.H., Biancani, P., 1989. VIP and acetylcholine: neurotransmitters in esophageal circular smooth muscle. Amer. J. Physiol. 257, G380–G385. [129] Belai, A., Burnstock, G., 1994. Evidence for coexistence of ATP and nitric oxide in non-adrenergic, non-cholinergic (NANC) inhibitory neurones in the rat ileum, colon and anococcygeus muscle. Cell. Tiss. Res., 278, 197–200. [129] Bennett, M.K., Calakos, N., Scheller, R.H., 1992. Syntaxin: a synaptic protein implicated in docking of synaptic vesicles at presynaptic active zones. Science. 257, 255–259. [160, 163] Bennett, M.K., Garcia-Arraras, J.E., Elferink, L.A., Peterson, K., Fleming, A.M., Hazuka, C.D., Scheller, R.H., 1993. The syntaxin family of vesicular transport receptors. Cell. 74, 863–873. [160] Bennett, M.R., 1965. A study of transmission from autonomic nerves to smooth muscle. MSc Thesis. Melbourne University. [118] Bennett, M.R., 1966a. Rebound excitation of the smooth muscle cells of the guinea pig taenia coli after stimulation of intramural inhibitory nerves. J. Physiol. 185, 124–131. [118, 119] Bennett, M.R., 1966b. Model of the membrane of smooth muscle cells of the guinea-pig taenia coli muscle during transmission from inhibitory and excitatory nerves. Nature 211, 1149–1152. [117] Bennett, M.R., 1966c. Transmission from intramural excitatory nerves to the smooth muscle cells of the guinea-pig taenia coli. J. Physiol. 185, 132–147. [132] Bennett, M.R., 1967. The effect of cations on the electrical properties of the smooth muscle cells of the guinea-pig vas deferens. J. Physiol. 190, 465–479. [115, 143, 145] Bennett, M.R., 1973. An electrophysiological analysis of the uptake of noradrenaline at sympathetic nerve terminals. J. Physiol . 229, 533–46. [69, 119] Bennett, M.R., 1983. Development of neuromuscular synapses. Physiol. Rev. 63, 915–1048. [242] Bennett, M.R., 1994. The concept of neurotransmitter release. In: Introduction to Molecular and Cellular Mechanisms of Neurotransmitter Release, Werner-Gren International Symposium, Stockholm. Advances in Second Messenger & Phosphoprotein Research L.Stjarne, P.Greengard, S. Grillner, T.Hokfelt & D.Ottson (eds.), 29, 1–29, Raven Press, New York. [24, 59, 107] Bennett, M.R., 1995. The origin of Gaussian distributions of synaptic potentials. Prog, in Neurobiol. 46(4), 331–350. [186] Bennett, M.R., 1996. Neuromuscular transmission at an active zone: the secretosome hypothesis. J. Neurocytol. 25(12), 869–891. [180, 276] Bennett, M.R., 1997. The Idea of Consciousness. Gordon and Breach. Harwood Academic Press, London. [94] Bennett, M.R., Balcar, V., 1999. Forty years of amino acid transmission in the brain. Neurochem. Intn. 35, 269–280. [232] Bennett, M.R., Burnstock, G., 1966. Application of the su-crose-gap method to determine the ionic basis of the membrane potential of smooth muscle. J. Physiol. 183, 637–648. [115, 117] Bennett, M.R., Burnstock, G., Holman, M.E., 1962a. The effects of ions on the electrical activity of the smooth muscle of the ureter. Papers presented at the fourth meeting of the Australian Physiological and Pharmacological Society, 23– 25th May, 1962. Dept. Physiology, University of Melbourne; (first formal publication in Proc. Aust. Physiol. Pharmacol. Soc. 20 (Suppl.) p. 46. [115] Bennett, M.R., Burnstock, G., Holman, M.E., Walker, J.W., 1962b. The effect of calcium ions on plateau-type action potentials in smooth muscle. J. Physiol. 161, 47–48P. [115] Bennett, M.R., Burnstock, G., Holman, M.E., 1963. The effect of potassium and chloride ions on the inhibitory potential recorded in the guinea-pig taenia coli. J. Physiol. 169, 33– 34P. [34, 117, 118, 127] Bennett, M.R., Burnstock, G., Holman, M., 1966a. Transmission from intramural inhibitory nerves to the smooth muscle of the guinea-pig taenia coli. J. Physiol. 182, 541–558. [34, 117, 118, 120, 123] Bennett, M.R., Burnstock, G., Holman, M.E., 1966b. Transmission from perivascular inhibitory nerves to the smooth muscle of the guineapig taenia coli. J. Physiol. 182, 527– 540. [125] Bennett, M.R., Farnell, L., Gibson, W., 1999. Quantal secretion at calcium domains of active zones: a Monte Carlo analysis. Proc. Aust. Neurosci. Soc. 10, 138. [180] Bennett, M.R., Fisher, C., 1977. The effects of calcium ions on the binomial parameters that control acetylcholine release during trains of nerve impulses at amphibian neuromuscular synapses. J. Physiol. 271(3), 673–698. [173, 174] Bennett, M.R., Fisher, C., Florin, T., Quine, M., Robinson, J., 1977. The effects of calcium ions and temperature on the binomial parameters that control acetylcholine release by a nerve impulse at amphibian neuromuscular synapses. J. Physiol. 271(3), 641–672. [179–182] Bennett, M.R., Florin, T., 1974. A statistical analysis of the release of acetylcholine at newly formed synapses in striated muscle. J. Physiol. 238(1), 93–107. [171, 173] Bennett, M.R., Gibson, W.G., Poznanski, R.R., 1993. Extracellular current flow and potential during quantal transmission from varicosities in a smooth muscle syncytium. Phil. Trans. Roy. Soc. Lond, Series B, Biolog. Sci. 342. [117, 160] Bennett, M.R., Gibson, W.G., Robinson, J., 1997. Probabilistic secretion of quanta and the synaptosecretosome hypothesis: evoked release at active zones of varicosities, boutons and endplates. Biophys. J. 73, 1815–1829. [162, 276]
REFERENCES
291
Bennett, M.R., Ho, S., Lavidis, N.A., 1986a. Competition between segmental nerves at endplates in rat gastrocnemius muscle during loss of polyneuronal innervation. J Physiol. 381, 351–376. [242] Bennett, M.R., Jones, P., Lavidis, N.A., 1986b. The probability of quantal secretion along visualised terminal branches at amphibian (bufo marinus) neuromuscular synapses. J Physiol . 379, 257–274. [260] Bennett, M.R., Lavidis, N.A., 1979. The effect of calcium ions on the secretion of quanta evoked by an impulse at nerve terminal release sites. J. Gen. Physiol. 74(4), 429–456. [173–175, 177] Bennett, M.R., Lavidis, N.A., 1986. Topographical projections of segmental nerves to the frog glutaeus muscle during loss of polyneuronal innervation. J. Physiol. 375, 303–325. [242] Bennett, M.R., McLachlan, E.M., Taylor, R.S., 1973a. The formation of synapses in reinnervated mammalian striated muscle. J. Physiol 233, 481–500. [233, 236, 242] Bennett, M.R., McLachlan, E.M., Taylor, R.S., 1973b. The formation of synapses in mammalian striated muscle reinnervated with autonomic preganglionic nerves. J. Physiol. 233, 501–517. [233, 236, 238] Bennett, M.R., Pettigrew, A.G., 1974a. The formation of synapses in striated muscle during development. J. Physiol. 241, 515–545. [238, 240–242, 245, 246, 252] Bennett, M.R., Pettigrew, A.G., 1974b. The formation of synapses in reinnervated and cross-reinnervated striated muscle during development. J. Physiol. 241, 547–573. [240, 247] Bennett, M.R., Pettigrew, A.G., 1975. The formation of synapses in amphibian striated muscle during development J. Physiol. 252, 203–239. [236, 238, 242] Bennett, M.R., Pettigrew, A.G., 1976. The formation of neuromuscular synapses. Cold Spring Harbor Symp. Quant. Biol. 40, 409–424. [235, 236, 242] Bennett, M.R., Pettigrew A.G., Taylor, R.S., 1973c. The formation of synapses in reinnervated and cross-reinnervated adult avian muscle. J. Physiol. 230, 331–357. [236, 238] Bennett, M.R., Raftos, J., 1977. The formation and regression of synapses during the reinnervation of axolotl striated muscles. J. Physiol. 265, 261–295. [260] Bennett, M.R., Robinson, J., 1989. Growth and elimination of nerve terminals at synaptic sites during polyneuronal innervation of muscle cells: a trophic hypothesis. Proc. Roy. Soc. Lond.-Ser. B: Biol. Sci. 235(1281), 299–320. [240, 243, 244, 251, 252, 265] Bennett, M.R., Robinson, J., 1990. Probabilistic secretion of quanta from nerve terminals at synaptic sites on muscle cells: non-uniformity, autoinhibition and the binomial hypothesis. Proc. Roy. Soc. Lond. Ser. B: Biol. Sci. 239(1296), 329–358. [183, 185, 186] Bennett, M.R., Rogers, D.C., 1967. A study of the innervation of the taenia coli. J. Cell Biol. 33, 573–596. [124, 125] Benoit, P., Changeux, J.P., 1975. Consequences of tenotomy on the evolution of multiinnervation in developing soleus muscle. Brain Res. 99, 354–358. [242] Benoit, P., Changeux, J.P., 1978. Consequences of blocking the nerve with a local anaesthetic on the evolution of multiinnervation at the regenerating neuromuscular junctions of the rat. Brain Res. 149, 89–96. [242] Bentivoglio, M, 1996. ‘1898–1996’: The Centennial of the Axon. Brain Res. Bull. 41, 319–325. [5, 11] Berl, S, Lajtha, A., Waelsch, H., 1961. Amino acid and protein metabolism. VI. Cerebral compartments of glutamic acid metabolism. J. Neurochem. 7, 186–197. [87] Bermingham-McDonogh, O., McCabe, K.L., Reh, T.A., 1996. Effects of GGF/neuregulins on neuronal survival and neurite outgrowth correlate with erbB2/neu expression in developing rat retina. Develop. 122(5), 1427–1438. [279] Bernard, C., 1965. (1850–1860) Cahier de Notes. Comments by M.D.Grmek. (ed.) Gallimard, Paris, pp. 314. [46] Bernard, C., 1883. Lecons sur les effets des substances toxiques et medicamenteuses. Bailliere, Paris, 1883 (edn.). [46] Bernstein, J., 1912. Elektrobiologie. Friedr. Vieweg and Sohn, Brawunschweig. [15] Bernstein, J., 1868. Uber den zeitlichen Verlauf der negativen Schwankung des Nervenstroms. Pflugers Arch. f.d. ges. Physiol. 1, 173–207. [15] Bernstein, J., 1871. Untersuchungen über den erregungsvorgang in nerven und muskelsysteme. Carl Winter, Heidelberg. [15] Bertler, A., Carlsson, A., Rosengren, E., 1956. Release by reserpine of catecholamines from rabbits’ hearts. Naturwissenschaften 22, 521. [92] Bertram, R., Sherman, A., Stanley E.F., 1996. Single-domain/ bound calcium hypothesis of transmitter release and facilitation. J. Neurophysiol. 75, 1919–1931. [162] Bettler, B., Kaupma, K., Bowery, N.G., 1998. GABAB receptors: drugs meet clones. Curr. Opin. Neurobiol. 8, 345–350. [90] Bettler, B., Mulle, C., 1995. Review: Neurotransmitter receptors II. AMPA and kainate receptors. Neuropharmacol. 34, 123– 139. [88] Betz, W.J., Caldwell, J.H., Ribchester, R.R., 1979. The size of motor units during postnatal development of rat lumbrical muscle. J. Physiol. 297, 463–478. [242, 245–251] Betz, W.J., Caldwell, J.H., Ribchester, R.R., 1980. Sprouting of active nerve terminals in partially inactive muscles of the rat. J. Physiol. 303, 281–297. [242, 245–247, 252, 254] Betz, W.J., Ribchester, R.R., Ridge, R.M., 1990. Competitive mechanisms underlying synapse elimination in the lumbrical muscle of the rat. J. Neurobiol. 21(1), 1–17. [246, 262] Bezprozvanny, I., Scheller, R.H., Tsien, R.W., 1995. Functional impact of syntaxin on gating of N-type and Q-type calcium channels. Nature. 378, 623–626. [163] Biancani, P., Walsh, J, Behar, J., 1985. Vasoactive intestinal peptide: a neurotransmitter for relaxation of the rabbit internal anal sphincter. Gastroenterol. 89, 867–874. [129] Birks, R., Huxley, H.E., Katz, B., 1960. The fine structure of the neuromuscular junction of the frog. J. Physiol. 150, 145–168. [165]
292
REFERENCES
Biroc, S.L., Payan, D.G., Fisher, J.M., 1993. Isoforms of agrin are widely expressed in the developing rat and may function as protease inhibitors. Develop. Brain. Res. 75, 119–129. [273] Biscoe, T.J., Davies, J., Dray,. A., Evans, R.H, Francis, A.A., Martin, M.R., Watkins, J.C., 1977a. Depression of synaptic excitation and of amino acid-induced excitatory responses of spinal neurones by D- -aminoadipate, alpha, epsilon-diaminopimelic acid and HA-966. Eur. J. Pharmacol. 45, 315–316. [87, 88] Biscoe, T.J., Evans, R.H, Francis, A.A., Martin, M.R., Watkins, J.C., Davies, J., Dray, A., 1977b. A-a-aminoadipate as a selective antagonist of amino acid-induced and synaptic excitation of spinal neurones. Nature. 270, 243–745. [88] Biscoe, T.J., Davies, J., Dray,. A., Evans, R.H, Martin, M.R., Watkins, J.C., 1978. D-alpha-aminodipate, alpha, epsilon-diominopimelic acid and HA-966 as antagonists of amino acid-induced and synaptic excitation of mammalian spinal neurones in vivo. Brain Res. 148, 543–548. [87] Bittner, G.D., Harrison, J., 1970. A reconsideration of the Poisson hypothesis for transmitter release at the crayfish neuromuscular junction. J. Physiol. 206(1), 1–23. [170] Bixby, J.L., 1981. Ultrastructural observations on synapse elimination in neonatal rabbit skeletal muscle. J. Neurocytol. 10, 81–100. [242] Bixby, J.L., van Essen, D.C., 1979. Regional differences in the timing of synapse elimination in skeletal muscles of the neonatal rabbit. Brain Res. 169, 275–286. [242] Blaha, K., Hrbek, J. Jr., Kovar, J., Pijewska, L., Santavy, F., 1964. Configuration of nitrogen-containing compounds. XVIII. Determination of relative and absolute configuration of phthalide isoquinoline alkaloids. Coll. Czech. Chem. Comun. 29, 2328–2340. [85] Blakeley, A.G., Brown, G.L., 1964. Effects of an anticholineste-rase on the transmitter overflow and the response of the spleen to nerve stimulation. J. Physiol. 194, 71–72. [69] Blaschko, H.K.F., 1985. U.S. von Euler. Biograph. Mem. Fell. Roy. Soc. 31, 143–170. [38] Blaschko, H., Welch, A.D., 1953. Localization of adrenaline in cytoplasmic particles of the bovine adrenal medulla. Naunyn-Schmiedebergs Arch. Exp. Path. Pharmakol. 219, 17–22. [40] Blazemore, A., Elliott, K.A., Florey. E., 1956. Factor I and gamma-aminobutyric acid. Nature. 178, 1052–1053. [79] Bliss, T.V., Douglas, R.M., Errington, M.L., Lynch, M.A., 1986. Correlation between long-term potentiation and release of endogenous amino acids from dentate gyrus of anaesthetised rats. J. Physiol. 377, 391–408. [223] Bliss, T.V., Gardner-Medwin, A.R., 1973. Long-lasting potentiation of synaptic transmission in the dentate area of the unanaesthetised rabbit following stimulation of the perforant path. J. Physiol. 232, 357–374. [211, 213, 214, 223] Bliss, T.V., Lomo, T., 1970. Plasticity in a monosynaptic cortical pathway. J. Physiol. 207, 61P. [211] Bliss, T.V, Lomo, T., 1973. Long-lasting potentiation of synaptic transmission in the dentate area of the anaesthetised rabbit following stimulation of the perforant path. J. Physiol. 232, 331–356. [213, 214] Boeckxstaens, G.E., Pelckmans, P.A, Bult, H., Man, J.G., Herman, A.G., Maercke, Y.M., 1990a. Non-adrenergic non-cholinergic relaxation mediated by nitric oxide in the canine ileocolonic junction. Europ. J. Pharmacol. 190, 239–246. [128] Boeckxstaens, G.E., Pelckmans, P.A., Boyers, J.J., Bult, H., Man, JG., Oosterbosch, L., Herman, A.G., Maercke, Y.M., 1991a. Release of nitric oxide upon stimulation of nonadrenergic noncholinergic nerves in the rat gastric fundus. J. Pharmacol Exper. Therap. 262, 441–447. [128] Boeckxstaens, G.E., Pelckmans, P.A., Herman, A.G., Maercke, Y.M., 1993. Involvement of nitric oxide in the inhibitory innervation of the human isolated colon. Gastroenterol. 104, 690–697. [128] Boeckxstaens, G.E., Pelckmans, P.A., Man, J.G., Bult, H., Herman, A.G., Maercke, Y.M., 1992. Evidence for a differential release of nitric oxide and vasoactive intestinal polypeptide by nonadrenergic noncholinergic nerves in the rat gastric fundus. Arch. Intn. Pharmacodynamie et de Therapie. 318, 107–115. [128] Boeckxstaens, G.E., Pelckmans, P.A., Ruytjens, I.F., Bult, H., Man, J.G., Herman, A.G., Maercke, Y.M., 1991b. Bioassay of nitric oxide released upon stimulation of non-adrenergic non-cholinergic nerves in the canine ileocolonic junction. Brit. J. Pharmacol. 103, 1085–1091. [128] Boerhaave, H., 1735. Institutiones Medicae, in Usus Annuae Exertationis Domesticos Digestae ab Hermanno Boerhaave. Leyden. [10, 11] Bommert, K., Charlton, M.P., DeBello, W.M., Chin, G.J., Betz, H., Augustine, G.J., 1993. Inhibition of neurotransmitter release by C2-domain peptides implicates synaptotagmin in exocytosis. Nature. 363, 163–165. [160] Borden, L.A., 1996. GABA transporter heterogeneity: pharmacology and cellular localization. Neurochem. Internat. 29, 335–356. [87] Borelli, G.A., 1670. De Motionibus Naturalibus a Gravitate Pendentibus Reggiodi Calabria, Bologna. [7] Bornstein, J.C., Furness, J.B., Smith, T.K., Trussell, D.C., 1991. Synaptic responses evoked by mechanical stimulation of the mucosa in morphologically characterised myenteric neurons of the guinea-pig ileum. J. Neurosci. 11, 505–518. [122, 131] Bouthenet, M.L., Martres, M.P., Sales, N., Schwartz, J.C., 1987. A detailed mapping of dopamine D-2 receptors in rat central nervous system by autoradiography with [125I]iodosulpride. Neurosci. 20, 117–155. [101] Bowe, M.A., Deyst, K.A., Leszyk, J.D., Fallen, J.R., 1994. Identification and purification of an agrin receptor from Torpedo postsynaptic membranes: a heteromeric complex related to the dystroglycans. Neuron. 12(5), 1173–1180. [274] Bradley, K., Easton, D.M., Eccless, J.C., 1953. An investigation of primary or direct inhibition. J. Physiol. 122, 474–488. [82] Brazier, M.A.B., 1959. The historical development of neurophysiology. Handbook of Physiology, Section 1. Neurophysiology. Vol. 1. J.Field (ed.). Ch. 1, pp. 1–58. Amer. Physiol. Soc. Washington D.C. [5, 6, 13] Brazier, M.A.B., 1984. A History of Neurophysiology in the 17th and 18th Centuries. Raven Press, New York. [2, 6, 9, 11] Bredt, D.S., Hwang, P.M., Glatt, C.E., Lowenstein, C., Reed, R.R., Snyder, S.H., 1991. Cloned and expressed nitric oxide synthase structurally resembles cytochrome P-450 reductase. Nature. 351, 714–718. [132]
REFERENCES
293
Brenner, H.R., Herczeg, A., Slater, C.R., 1992. Synapse-specific expression of acetylcholine receptor genes and their products at original synaptic sites in rat soleus muscle fibres regenerating in the absence of innervation. Develop. 116(1), 41–53. [277] Brenner, H.R., Witzemann, V., Sakmann, B., 1990. Imprinting of acetylcholine receptor messenger RNA accumulation in mammalian neuromuscular synapses. Nature. 344(6266), 544–547. [277] Bresadola, M., 1998. Medicine and science in the life of Luigi Galvani (1737–1798). Brain Res. Bull 46, 367–380. [12] Brezina,V., Orekhova, I., Weiss, K., 1996. Functional uncoupling of linked neurotransmitter affects by combinatorial convergence. Science. 273, 806–810. [130] Bridgewater, M., Cunnane, T.C., Brading, A.F., 1995. Characteristic features of inhibitory junction potentials evoked by single stimuli in the guinea-pig isolated taenia caeci. J. Physiol. 485, 145–155. [127, 129, 130, 132] Briejer, M.R., Akkermans, L.M., Meulemans, A.L., Lefebvre, R.A., Schuurkes, J.A., 1995. 5-HT-induced neurogenic relaxations of the guinea-pig proximal colon investigation into the role of ATP and VIP in addition to nitric oxide. Naunyn-Schmiedebergs Arch. Pharmacol. 351, 126–35. [128] Briggs, C.A., Brown, T.H., McAfee, D.A., 1985. Neurophysiology and pharmacology of long-term potentiation in the rat sympathetic ganglion. J. Physiol. 359, 503–521. [230] Brock, L.G., Coombs, J.S., Eccles, J.C., 1952, The recordings of potentials from motoneurones with an intracellular electrode. J. Physiol. 117, 431–460. [29, 38] Brock, J.A., Cunnane, T.C., Evans, R.J., Ziogas, J., 1989. Inhibition of transmitter release from sympathetic nerve endings by omegaconotoxin. Clin. Exp. Pharmacol. Physiol 16, 333–339. [158] Brodie, T.G., Dixon, W.E., 1904. Contributions to the physiology of the lungs. Part 11. On the innervation of the pulmonary blood vessels; and some observations on the action of suprarenal extract. J. Physiol. 30, 476–502. [27, 50] Brodie, B.C., 1811. Experiments and observations on the different modes in which death is produced by certain vegetable poisons. Phil. Trans. Roy. Soc. Lond. 101, 178–207. [45] Brooks, C.McC., Eccles, J.C., 1947. An electrical hypothesis of central inhibition. Nature. 159, 760–764. [31, 36] Brose, N., Petrenko, A.G., Sudhof, T.C., Jahn, R., 1992. Synaptotagmin: a calcium sensor on the synaptic vesicle surface. Science. 256, 1021–1025. [160, 162] Brown, A.M., Camerer, H., Kunze, D.L., Lux, H.D., 1982. Similarity of unitary Ca2+ currents in three different species. Nature. 299, 156–158. [153] Brown, D.A., Caulfield, M.P., 1979. Hyperpolarizing ‘alpha 2’-adrenoceptors in rat sympathetic ganglia. Brit. J. Pharmacol. 65, 435–445. [77] Brown, G.L., 1937. The actions of acetylcholine on denervated mammalian and frog’s muscle. J. Physiol. 89, 438–461. [60, 62] Brown, G.L., Eccles, J.C. 1934. The action of a single vagal volley on the rhythm of the heart beat . J. Physiol. 82, 211– 241. [28, 30] Brown, G.L., Dale, H.H., Feldberg, W., 1936. Reactions of the normal mammalian muscle to acetylcholine and to eserine. J. Physiol. 87, 394–424. [59] Brown, G.L., Feldberg, W., 1936a. The action of potassium on the superior cervical ganglion of the cat. J. Physiol. 86, 290–305. [135–137] Brown, G.L., Feldberg, W., 1936b. The acetylcholine metabolism of a sympathetic ganglion. J. Physiol. 88, 265–283. [135] Brown, G.L., Gillespie, J.S., 1957. The output of sympathetic transmitter from the spleen of the cat. J. Physiol. 138, 81– 102. [67, 69] Brown, M.C., Booth, C.M., 1983. Postnatal development of the adult pattern of motor axon distribution in rat muscle. Nature. 304, 741–742. [264, 267] Brown, M.C., Goodwin, G.M., Ironton, R., 1977. Prevention of motor nerve sprouting in botulinium poisoned mouse soleus muscle by direct stimulation of the muscle. J. Physiol. 267, 42P–43P. [247, 254] Brown, M.C., Holland, R.L., Hopkins, W.G., 1981. Restoration of focal multiple innervation in rat muscles by transmission block during a critical stage of development. J. Physiol. 318, 355–364. [247, 254] Brown, M.C., Ironton, R., 1978. Sprouting and regression of neuromuscular synapses in partially denervated mammalian muscles. J. Physiol. 278, 325–348. [254, 257, 258] Brown, M.C., Jansen, J.K.S., Van Essen, D.C., 1976. Polyneuronal innervation of skeletal muscle in new-born rats and its elimination during maturation. J. Physiol. 261, 387– 422. [175, 245–248, 250, 252] Brown, M.C., Matthews, P.B.C., 1960. An investigation into the existence of polyneuronal innervation of individual skeletal muscle fibres in certain hindlimb muscles of the cat. J. Physiol. 151, 436–457. [252] Brown, T.H., McAfee, D.A., 1982. Long-term synaptic potentiation in the superior cervical ganglion. Science. 215, 1411– 1413. [230, 232] Brown, T.H., Perkel, D.H., Feldman, M.W., 1976. Evoked neurotransmitter release: a statistical effects of nonuniformity and nonstationarity. Proc. Nat. Acad. Sci. USA, 73(8), 2913– 2917. [173] Brucke, F.Th.v., 1937. The cholinesterase in sympathetic ganglia. J. Physiol. 89, 429–437. [58, 59] Bruderlein, F.T., Humber, L.G., 1975. Neuroleptic agents of the benzocycloheptapyridoisoquinoline series. 1 Syntheses and stereochemical and structural requirements for activity of butaclamol and related compounds. J. Medic. Chem. 18, 185–188. [95] Buckley, N.J., Saffrey, M.J., Hassall, C.J., Burnstock, G., 1988. Localization of muscarinic receptors on peptide-containing neurones of the guinea pig myenteric plexus in tissue culture. Brain Res. 445, 152–156. [131] Bulbring, E., Kuriyama, H., 1963. Effects of changes in the external sodium and calcium concentrations on spontaneous electrical activity in smooth muscle of guinea-pig taenia coli. J. Physiol. 166, 29–58. [142] Bulbring, E., Tomita, T., 1967. Properties of the inhibitory potential of smooth muscle as observed in the response to field stimulation of the guinea-pig taenia coli. J. Physiol. 189, 299–315. [118]
294
REFERENCES
Bulbring, E., Burnstock, G., Holman, M.E., 1958. Excitation and conduction in the smooth muscle of the isolated taenia coli of the guinea-pig. J. Physiol. 142, 420–437. [117, 118] Bulbring, E., Kuriyama, H., Tomita, T., 1965. Discrimination between the ‘inhibitory potential’ and electrotonic hyperpolarization in smooth muscle. J. Physiol. 181, 8P–10P. [118] Buller, A.L., Larson, H.C., Schneider B.E., Beaton, J.A., Morissett, R.A., Monaghan, D.T., 1994. The molecular basis of NMDA receptor subtypes: native receptor diversity is predicted by subunit composition. J. Neurosci. 14, 5471– 5484. [88] Bunzow, J.R., Van Tol, H.H.M., Grandy, D.K., Albert, P., Salon, J., Christie, M., Machida, C.A., Neve, K.A., Civelli, O., 1988. Cloning and expression of a rat D2 dopamine receptor cDNA. Nature. 336, 783–787. [98] Burke, S.P., Adams, M.E., Taylor, C.P., 1993. Inhibition of endogenous glutamate release from hippocampal tissue by Ca2+ channel toxins. Eur. J. Pharmacol. 238, 383–386. [155] Burn, J.H., Rand, M.J., 1959. Sympathetic postganglionic mechanism. Nature. 184, 163–165. [67] Burnstock, G., 1972. Purinergic nerves. Pharmacol. Rev. 24, 509–581. [127, 129] Burnstock, G., 1981. Neurotransmitters and trophic factors in the autonomic nervous system. J. Physiol. 313, 1–35. [129] Burnstock, G., Holman, M.E., 1961. The transmission of excitation from autonomic nerve to smooth muscle. J. Physiol. 155, 115–133. [117] Burnstock, G., Straub, R.W., 1958. A method for studying the effects of ions and drugs on the resting and action potentials in smooth muscle with external electrodes. J. Physiol. 140, 156–167. [115] Burnstock, G., Campbell, G., Bennett, M., Holman, M.E., 1963a. Inhibition of the smooth muscle of the taenia coli. Nature. 200, 581–582. [117, 118, 120, 127] Burnstock, G., Campbell, G., Bennett, M., Holman, M.E., 1963b. The transmission of inhibition from autonomic nerves to the smooth muscle of the guinea-pig taenia coli. Paper presented at the fourth meeting of the Australian Physiological and Pharmacological Society, 13– 15 February 1963; first formally published in Proc. Aust. Physiol. Pharmacol. Soc. 20, (Suppl): 62–63. [117, 118] Burnstock, G., Campbell, G., Bennett, M., Holman, M.E., 1963c. The effects of drugs on the transmission of inhibition from autonomic nerves to the smooth muscle of the guinea-pig taenia coli. Sec. Intn Pharmacol. Meet. Prague. Biochem. Pharmacol. 12 (Suppl): P. 134, Abstract 467. [117] Burnstock, G., Campbell, G., Bennett, M.R., Holman, M.E., 1964. Innervation of the guinea-pig taenia coli: are there intrinsic inhibitory nerves which are distinct from sympathetic nerves? Int. J. Neuropharmacol. 3, 163–166. [114] Burnstock, G., Campbell, G., Satchell, D., Smythe, A., 1970. Evidence that adenosine triphosphate or a related nucleotide is the transmitter substance released by non-adrenergic inhibitory nerves in the gut. Brit. J. Pharmacol. 40, 668–688. [127, 129] Burnstock, G., Campbell, G., Rand, M.J., 1966. The inhibitory innervation of the taenia of the guinea-pig caecum. J. Physiol. 182, 504–526. [114, 120] Burnstock, G., Cocks, T., Paddle, B., Staszewska-Barczak, J., 1975. Evidence that prostaglandin is responsible for the ‘rebound contraction’ following stimulation of non-adrenergic, non-cholinergic (‘purinergic’) inhibitory nerves. Europ. J. Pharmacol. 31, 360–362. [132] Burt, D.R., Enna, S.J., Creese, I., Snyder, S.H., 1975. Dopamine receptor binding in the corpus striatum of mammalian brain . Proc. Nat. Acad. Sci. USA. 72, 4655–4659. [96] Busatto, G.F., Kerwin, R.W., 1997. Perspectives on the role of serotonergic mechanisms in the pharmacology of schizophrenia. J. Psychopharm. 11, 3–12. [103] Butler, J., Cosmos, E., Brierley, J., 1982. Differentiation of muscle fiber types in aneurogenic brachial muscles of the chick embryo. J. Exp. Zool. 224, 65–80. [260, 262] Cahan, D., 1993. Hermann von Helmholtz and the Foundations of Nineteenth Century Science, University of California Press, Berkeley. [13, 14] Cajal, S.Ramon, 1959. Degeneration and regeneration of the nervous system. 1928. (R.M.May, transl.). Hafner, New York. [23] Cajal, S.Ramon, 1995. Histology of the Nervous System of Man and Vertebrates. (1909–1911) Translated from the Spanish by Dr. L.Azoulay. Translated from the French by N.Swanson and L.W.Swanson. Vols. 1 and 2. Oxford University Press. New York. [17–22] Caldwell, D.F., Werboff, J., 1962. Classical conditioning in newborn rats. Science. 136, 1118–1119. [217] Caldwell, J.H., Ridge, R.M.A.P., 1983. The effects of deafferentation and spinal cord transection on synapse elimination in developing rat muscles. J. Physiol. 339, 145– 159. [242, 247, 248] Callaway, E.M., Soha, J.M., Van Essen, D.C., 1987. Competition favouring inactive over active motor neurons during synapse elimination. Nature. 328(6129), 422–426. [247] Callaway, E.M., Soha, J.M., Van Essen, D.C., 1989. Differential loss of neuromuscular connections according to activity level and spinal position of neonatal rabbit soleus motor neurons. J. Neurosci. 9(5), 1806–1824. [247, 264] Campbell, G., 1966a. The inhibitory nerve fibres in the vagal supply to the guinea-pig stomach. J. Physiol. 185, 600– 612. [114, 115, 118] Campbell, G., 1966b. Nerve-mediated excitation of the taenia of the guinea-pig caecum. J. Physiol. 185, 148–159. [111] Campbell, G., Burnstock, G., 1968. The comparative physiology of gastrontestinal motility. In: Handbook of Physiology, Section 6. Alimentary Canal Vol. IX, pp. 2213–2266. C.F. Code (ed.). Amer. Physiol. Soc., Washington D.C. [125] Campenot, R.B., 1982. Development of sympathetic neurones in compartmentalised cultures. I. Local control of neurite growth by nerve growth factor. Develop. Biol. 93, 1–12. [242] Cannon, W.B., Bacq, Z.M., 1931. A hormone produced by sympathetic action on smooth muscle. Amer. J Physiol. 96, 392–99. [67] Cannon, W.B., Uridil, J.E., 1921. Studies on the conditions of activity in endocrine glands. VIII Some effects on the denervated heart of stimulating the nerves of the liver. Amer. J. Physiol. 58:353–364. [65, 66] Cannon, W.B., Lieb, C.W., 1911. The receptive relaxation of the stomach. Amer. J. Physiol XXIX, 267–273. [112, 113]
REFERENCES
295
Capiod, T., Ogden, D.C., 1989. The properties of calcium-activated potassium ion channels in guinea-pig isolated hepatocytes. J. Physiol. 409, 285–295. [127, 131] Carbone, E., Lux, H.D., 1984. A low voltage-activated calcium conductance in embryonic chick sensory neurons. Biophys. J. 46, 413–418. [153] Carl, A., Kenyon, J.L., Uemura, D., Fusetani, N., Sanders, K.M., 1991. Regulation of Ca(2+)-activated K+ channels by protein kinase A and phosphatase inhibitors. Amer. J. Physiol. 261, C387–C392. [131] Carlos, J., Niemegeers, E., Janssen, P.A.J., 1979. A systematic study of the pharmacological activities of dopamine antagonists Life. Sci. 24, 2201–2216. [102] Carlson, A.J., Boyd, T.E., Pearcy, J.F., 1922. The innervation of the cardia and the lower end of the esophagus in mammals. Amer. J. Physiol. 61, 14–41. [111, 113] Carlsson, A., 1978. Mechanism of action of neuroleptic drugs. In: Psychopharmacology. A Generation of Progress. Lipton, M.A., DiMascio, A., Killman, K.F., (eds.) pp. 1057–1070, Raven Press, New York. [92] Carlsson, A., 1996. The rise of Neuropsychopharmacology: impact on basic and clinical science. In: The Psychopharmacologists, Healy, D., (ed.) pp 51–80. Chapman & Hall, New York. [104] Carlsson A., Hillarp, N.A., 1956. Kgl Fysiogr Sallsk Lund Forhandl 26: No. 8. [92] Carlsson, A., Lindqvist, M., 1963. Effect of chlorpromazine or haloperidol on formation of 3–methoxytyramine and normetanephrine in mouse brain. Acta. Pharmacol Toxicol. 20, 140–144. [92, 93] Carlsson, A., Rosengren, E., Bertler, A., Nilsson, J., 1957. In: Psychotropic Drugs. Garattini, S., Ghetti, V, (eds.) pp. 363–372. Elsevier, Amsterdam. [92] Carlsson, A., Waldeck, B., 1958. A fluorimetric method for the determination of dopamine (3-hydroxytyramine). Acta. Physiol. Scand. 44, 293. [92] Carlsson, M., Carlsson, A., 1990. Interactions between glutamatergic and Monoaminergic systems within the basal gan-glia-implications for schizophrenia and Parkinson’s disease. Trends. Neurosci. 13, 272–276. [104] Carraway, K.L. 3rd., Weber, J.L., Unger, M.J., Ledesma, J., Yu, N., Gassmann, M. and Lai, C., 1997. Neuregulin-2, a new ligand of ErbB3/ ErbB4-receptor tyrosine kinases. Nature. 387(6632), 512–516. [279] Castellano, M.A., Liu, L.X., Monsma, F.J. Jr, Sibley, D.R., Kapatos, G., Chiodo, L.A., 1993. Transfected D2 short dopamine receptors inhibit voltage-dependent potassium current in neuroblastoma x glioma hybrid (NG108–15) cells. Mol. Pharmacol. 44, 649–656. [103] Castro, C.A., Silbert, L.H., McNaughton, B.L., Barnes, C.A., 1989. Recovery of spatial learning deficits after decay of electrically induced synaptic enhancement in the hippocampus. Nature. 342, 545–548. [230, 231] Cedarbaum, J.M., Aghajanian, G.K., 1977. Catecholamine receptors on locus coeruleus neurons: pharmacological characterisation. Eur. J. Pharmacol. 44, 375–385. [73, 75] Chad, J.E., Eckert, R., 1984. Calcium domains associated with individual channels can account for anomalous voltage relations of Cadependent responses. Biophys. J. 45, 993– 999. [163] Changeux, J.P., Danchin, A., 1976. Selective stabilization of developing synapses as a mechanism for the specification of neuronal networks. Nature. 264, 705–712. [242] Chapman, E.R., Jahn, R., 1994. Calcium-dependent interaction of the cytoplasmic region of synaptotagmin with membranes. Autonomous function of a single C2-homologous domain. J. Biol. Chem. 269, 5735–5741. [162] Charpentier, P., 1947. Sur la constitution d’un dimethylamino-N-phenothiazine. C.R. Hebd. Seances Acad. Sci. 225, 306– 308. [91] Charuchinda, C., Supavilai, P., Karobath, M., Palacios, J.M., 1987. Dopamine D2 receptors in the rat brain: autoradiographic visualization using a high-affinity selective agonist ligand. J. Neurosci. 7, 1352–1360. [101] Chen, B.M., Grinnell, A.D., 1997. Kinetics, Ca2+ dependence, and biophysical properties of integrin-mediated mechanical modulation of transmitter release from frog motor nerve terminals. J. Neurosci. 17(3), 904–916. [276] Chen, L., Ko, C.P., 1994. Extension of synaptic extracellular matrix during nerve terminal sprouting in living frog neuromuscular junctions. J. Neurosci. 14(2), 796–808. [277] Chen, L.L., Folsom, D.B., Ko, C.P., 1991. The remodeling of synaptic extracellular matrix and its dynamic relationship with nerve terminals at living frog neuromuscular junctions. J. Neurosci. 11(9), 2920–2930. [277] Chen, M.S., Bermingham-McDonogh, O., Danehy, F.T. Jr., Nolan, C., Scherer, S.S., Lucas, J., Gwynne, D., Marchionni, M.A., 1994. Expression of multiple neuregulin transcripts in postnatal rat brains. J. Comp. Neurol. 349(3), 389–400. [279] Chiappinelli, V, Giacobini, E., Pilar, G., Uchimura, H., 1976. Induction of cholinergic enzymes in chick ciliary ganglion and iris muscle cells during synapse formation. J. Physiol. 257(3), 749–766. [279] Chiu, A.Y., Sanes, J.R., 1984. Differentiation of basal lamina in synaptic and extrasynaptic portions of embryonic rat muscle. Develop. Biol. 103, 456–467. [242] Chow, I., Cohen, M.W., 1983. Developmental changes in the distribution of acetylcholine receptors in the myotomes of Xenopus Laevis. J. Physiol. 339, 553–571. [242] Christensen, B.N., Martin, A.R., 1970. Estimates of probability of transmitter release at the mammalian neuromuscular junction. J. Physiol. 210, 933–945. [170] Chu, G.C., Moscoso, L.M., Sliwkowski, M.X., Merlie, J.P., 1995. Regulation of the acetylcholine receptor epsilon subunit gene by recombinant ARIA: an in vitro model for transynaptic gene regulation. Neuron. 14(2), 329–339. [274] Ciccocioppo, R., Onori, L., Messori, E., Candura, S.M., Coccini, T., Tonini M., 1994. Role of nitric oxide-dependent and -independent mechanisms in peristalsis and accommodation in the rabbit distal colon. J. Pharmacol. Exper. Therapeut. 270, 929–937. [128]
296
REFERENCES
Clarke, D.M., 1995. Descartes philosophy of Science and the scientific revolution. In: ‘The Cambridge Companion to Descartes’. Cottingham, J. (ed.), pp. 258–285. Cambridge University Press. [3, 20] Clasbrummel, B., Osswald, H., Illes, P., 1989. Inhibition of noradrenaline release by omega-conotoxin GVIA in the rat tail artery. Br. J. Pharmacol. 96, 101–110. [158] Clement-Cormier, Y.C., Kebabian, J.W., Petzold, G.L., Greengard, P., 1974. Dopamine-sensitive adenylate cyclase in mammalian brain: a possible site of action of antipsychotic drugs. Proc. Nat. Acad. Sci. USA. 71, 1113–1117. [95] Cocks, T., Burnstock, G., 1979. Effects of neuronal polypeptides on intestinal smooth muscle; a comparison with non-adrenergic, noncholinergic nerve stimulation and ATP. Europ. J. Pharmacol. 54, 251–259. [129] Cohen, M.W., Godfrey, E.W., 1992. Early appearance of and neuronal contribution to agrin-like molecules at embryonic frog nerve-muscle synapses formed in culture. J. Neurosci. 12(8), 2982–2992. [271] Cohen, M.W., Moody-Corbett, R, Godfrey, E.W., 1994. Neu-ritic deposition of agrin on culture substrate: implications for nerve-muscle synaptogenesis. J. Neurosci. 14, 3293– 3303. [271] Cohen, N.A., Kaufmann, W.E., Worley, P.P., Rupp, R, 1997. Expression of agrin in the developing and adult rat brain. Neurosci. 76(2), 581–596. [279] Collingridge, G.L., Kehl, S.J., McLennan, H., 1983. Excitatory amino acids in synaptic transmission in the Schaffer col-lateral-commissural pathway of the rat hippocampus. J. Physiol. 334, 33–46. [218] Collingridge, G.L., Lester, R.A.S., 1989. Excitatory amino acid receptors in the vertebrate central nervous system. Pharmacol. Rev. 40, 145–195. [88] Cooper, D.M., 1982. Bimodal regulation of adenylate cyclase. FEBS Letts. 138, 157–163. [77] Cornford, F.M., 1956. Plato’s Cosmology, Routledge & Kegan Paul Ltd. London. [1] Costa, M., Furness, J.B., Humphreys, C.M., 1986. Apamin distinguishes two types of relaxation mediated by enteric nerves in the guineapig gastrointestinal tract. Naunyn Schmiedebergs Arch. Pharmacol. 332, 79–88. [130] Costa, M., Furness, J.B., Pompolo, S., Brookes, S.J.H., Bornstein, J.C., Bredt, D.S., Snyder, S.H., 1992. Projec-tions and chemical coding of neurons with immunoreactivity for nitric oxide synthase in the guinea-pig small intestine. Neurosci. Letts. 148, 121–125. [131] Cotman, C.W., Monaghan, D,T., Ottersen, O.P., Storm-Mathisen, J. , 1987. Anatomical organisation of excitatory amino acid receptors and their pathways. Trends. Neurosci. 10, 273– 279. [88] Cottingham, J., 1995. Cartesian dualism: theology, meta-physics, and science. In: ‘The Cambridge Companion to Descartes’. Cottingham, J. (ed.). pp. 236–254. Cambridge University Press. [1] Couteaux, R., Nachmansohn, D., 1940. Changes of choline esterase at end plates of voluntary muscle following section of sciatic nerve. Proc. Soc. Exper. Biol. Med. 43, 177–181. [60, 62] Covault, J., Sanes, J., 1985. Neural cell adhesion molecule (N-CAM) accumulates in denervated and paralysed skeletal muscle. Proc. Nat. Acad. Sci. USA 82, 4544–4548. [242] Cox, D.R., Hinkley, D.V., 1974. Theoretical Statistics. Chapman and Hall. London. [195] Creed, R.S., Eccles, J.C., 1928. The incidence of central inhibition on restricted fields of motor units. J. Physiol. 66, 109–120. [27] Creed, R.S., Sherrington, C.S., 1926. Observations on concurrent contraction of flexor muscles in the flexor reflex. Proc. Roy. Soc. Lond. B. 100, 258–267. [27] Creese, I., Burt, D.R., Snyder, S.H., 1976. Dopamine receptor binding predicts clinical and pharmacological potencies of antischizophrenic drugs. Science. 192, 481–483. [99] Crist, J.R., He, X.D., Goyal, R.K., 1992. Both ATP and the peptide VIP are inhibitory neurotransmitters in guinea-pig ileum circular muscle. J. Physiol. 447, 119–131. [129] Croone, B.W. 1665. De Ratione Motus Musculorum. (Published anonymously, bound with Thomas Willis’ ‘Cerebri Anatome’). Hayes, London 1665. Amsterdam; 1667. [6, 8] Cubeddu, L., Barnes, E.M., Langer, S.Z., Weiner N., 1974. Release of norepinephrine and dopamine-beta-hydroxylase by nerve stimulation. I. Role of neuronal and extraneuronal uptake and of alpha presynaptic receptors. J. Pharmacol. Exp. Therapeut. 190, 431–450. [71, 72] Curtis, D.R., 1962. The depression of spinal inhibition by electrophoretically administered strychnine. Int. J. Neuropharm. 1, 239–250. [81] Curtis, D.R., Davis, R., 1962. Pharmacological studies upon neurones of the lateral geniculate nucleus of the cat. Br. J. Pharm. 18, 217–246. [81] Curtis, D.R., Duggan, A.W., Felix D., Johnston, G.A.R., 1970a. GABA bicuculline and central inhibition. Nature. 226, 1222– 1224. [81, 85] Curtis, D.R., Duggan, A.W., Felix D., Johnston, G.A.R., 1970b. Bicuculline and central GAB A receptors. Nature. 228, 676– 677. [85] Curtis, D.R., Eccles, R.M., 1958a. The excitation of Renshaw cells by pharmacological agents applied electrophoretically. J. Physiol. 141, 435–445. [79] Curtis, D.R., Eccles, R.M., 1958b. The effect of diffusional barriers upon the pharmacology of cells within the central nervous system. J. Physiol. 141, 446–463. [79] Curtis, D.R., Koizumi, K., 1961. Chemical transmitter substances in the brain stem of the cat. J. Neurophysiol. 24, 80–90. [81] Curtis, D.R., Phillis, J.W., 1958. Gamma amino-n-butyric acid and spinal synaptic transmission. Nature. 182, 323. [79] Curtis, D.R., Phillis, J.W., Watkins, J.C., 1959a. The depression of spinal neurones by gamma-amino-n-butyric acid and -alanine. J. Physiol 146, 185–203. [79, 81, 83] Curtis, D.R., Phillis, J.W., Watkins, J.C., 1959b. Chemical excitation of spinal neurones. Nature. 183, 611–612. [79, 81]
REFERENCES
297
Curtis, D.R., Phillis, J.W., Watkins, J.C., 1960. The chemical excitation of spinal neurones by certain acidic amino acids. J. Physiol. 150, 656–682. [81] Curtis, D.R.,Watkins, J.C., 1960. The excitation and depression of spinal neurones by structurally related amino acids. J. Neurochem. 6, 117–141. [79, 81, 82] Curtis, D.R., Watkins, J.C., 1961. Analogues of glutamic and gamma-amino-n-butyric acids have potent actions on mammalian neurones. Nature. 191, 1010–1011. [81] Curtis, D.R., Watkins, J.C., 1963. Acidic amino acids with strong excitatory actions on mammalian neurones. J. Physiol. 166, 1–14. [81] Curtis, D.R. Hösli, L., Johnston, G.A.R., Johnson, I.H., 1967a. The hyperpolarization of spinal motoneurones by glycine and related amino acids. Exp. Brain Res. 5, 235–258. [82] Curtis, D.R. Hösli, L., Johnston, G.A.R., Johnson, I.H., 1967b. Inhibition of spinal neurones by glycine. Nature 215, 1502– 1503.[82–84] Curtis, D.R. Hösli, L., Johnston, G.A.R., Johnson, I.H., 1968. A pharmacological study of the depression of spinal neurones by glycine and related amino acids Exp. Brain Res. 6, 1–18. [84] Dakin, H.D., 1905. The synthesis of a substance allied to noradrenaline. Proc. Roy. Soc. Lond. Series. B. LXXVI: 491–497. [65] Dal Toso, R., Sommer, B., Ewert, M., Herb, A., Pritchett, D.B., Bach A., Shivers, B.D., Seeburg, P.H., 1989. The dopamine D2 receptor: two molecular forms generated by alternative splicing. E.M.B.O. J. 8, 4025–4034. [98] Dale, H.H., 1910. The active principles of ergot. Proc. Internal. Physiol. Congress. Vienna, 27–30. [27, 65, 107] Dale, H.H., 1914a. The action of certain esters and ethers of choline, and their relation to muscarine. J. Pharm. Exper. Therap. 6, 147–190. [54, 55] Dale, H.H., 1914b. The occurrence in ergot and action of acetylcholine. J. Physiol XLVIII, iii-iv. [27, 54, 57] Dale, H.H., 1934a. Nomenclature of fibres in the autonomic system and their effects. J. Physiol. 80, 10P. [108, 127] Dale, H.H., 1934b. Chemical transmission of the effects of nerve impulses. Brit. Med. J.I. 835–841. [26, 27, 34, 106– 108, 111, 118, 120] Dale, H.H., 1937. Transmission of nervous effects by acetylcholine. 32, 229–245. [34] Dale, H.H., 1939. John Jacob Abel 1857–1938. Obit. Not. Fell Roy. Soc. Lond., Roy. Soc. 2, 577–85. [65] Dale, H.H., 1961. Thomas Renton Elliott 1877–1961. Biograph. Mem. Fellows Roy. Soc. London 7, 53–74. [43, 107] Dale, H.H., 1962. Otto Loewi. Biograph. Mem. of Fellows of Roy. Soc. 8, 67–89. [107] Dale, H.H., Dudley, H.W., 1929. The presence of histamine and acetylcholine in the spleen of the ox and the horse. J. Physiol. LXVIII, 97–123. [56, 57, 108, 118] Dale, H., Feldberg, W., 1934. The chemical transmitter of vagus effects of the stomach. J. Physiol 81, 320–334. [58, 108, 109] Dale H.H., Feldberg, W., Vogt, M., 1936. Release of acetylcholine at voluntary motor nerve endings. J. Physiol 86, 353–380. [59] Dale, H.H., Gaddum, J.H., 1930. Reactions of denervated voluntary muscle, and their bearing on the mode of action of parasympathetic and related nerves. J. Physiol. 70, 109– 144. [32, 111] d’Alonzo, A.J., Grinnell, A.D., 1985. Profiles of evoked release along the length of frog motor synaptic terminals. J. Physiol. 359, 235–258. [260] Dalziel, H.H., K.D., Ward, S.M., Sanders, K.M., 1991. Involvement of nitric oxide synthetic pathway in inhibitory junction potentials in canine proximal colon. Amer. J. Physiol. 260, G789–G792. [128] Davenport, H.W., 1989. Gastrointestinal physiology, 1895–1975: motility, Handbook of Physiology, Section 6, Vol. 1, Part 1. Schultz, S.G., Wood, J.D., Rauner, B.B. (eds.). Am. Physiol Soc. Bethesda, 1–101. [107] David, P., el Far, O., Martin-Moutot, N., Poupon, M.F., Takahashi, M., Seagar, M.J., 1993. Expression of synaptotagmin and syntaxin associated with N-type calcium channels in small cell lung cancer. FEBS Letts. 326, 135–139. [163] Davies, J., Evans, R.H, Francis, A.A., Watkins, J.C., 1979. Excitatory amino acid receptors and synaptic excitation in the mammalian central nervous system. J. Physiol. (Paris) 75, 641–654. [87, 88] Davies, J., Watkins, J.C., 1982. Actions of D and L forms of 2-amino-5-phosphonovalerate and 2-amino-4-phosphono-butyrate in the cat spinal cord. Brain Res. 235, 378–386. [88, 89] Davies, S.N., Lester, R.A., Reymann, K.G., Collingridge, G.L., 1989. Temporally distinct pre- and post-synaptic mechanisms maintain long term potentiation. Nature. 338, 500– 503. [225, 227] Davison, J.S., Pearson, G.T., 1979. The role of intrinsic, nonadrenergic, non-cholinergic, inhibitory nerves in the regulation of distensibiliry of the guinea-pig colon. Pflugers. Arch. 381, 73–77. [113] Davletov, B.A., Sudhof, T.C., 1994. Ca2+-dependent conforma-tional change in synaptotagmin I.J. Biol. Chem. 269, 28547– 28550.[162] De Groat, W.C., Volle, R.L., 1966. The actions of the catecholamines on transmission in the superior cervical ganglion of the cat. J. Pharmacol Exper. Therapeut. 154, 1–13. [75] de Jonge, M., Racine, R.J., 1985. The effects of repeated induction of long-term potentiation in the dentate gyrus. Brain Res. 328, 181–185. [214] De Langen, C.D., Hogenboom, F., Mulder, A.H., 1979. Presynaptic noradrenergic alpha-receptors and modulation of 3H-norarenaline release from rat brain synaptosomes. Eur. J. Pharmacol 60, 79–89. [73, 75] De Luca, A., Li, C.G., Rand, M.J., Reid, J.J., Thaina, P., Wong-Dusting, H.K., 1990. Effects of omega-conotoxin GVIA on autonomic neuroeffector transmission in various tissues. Br. J. Pharmacol 101, 437–447. [158] De Man, J.G., Boeckxstaens, G.E., Pelckmans, P.P., De Winter, B.Y., Herman, A.G. and Van Maercke, Y.M., 1993. Prejunctional modulation of the nitrergic innervation of the canine ileocolonic junction via potassium channels. Brit. J. Pharmacol 110, 559–564. [128] de Robertis, E.D.P., Bennett, H.S., 1955. Some features of the submicroscopic morphology of synapses in frog and earthworm. J. Biophys. Biochem. Cytol 1, 47–58. [36, 40]
298
REFERENCES
de Robertis, E.D.P., Bennett, H.S., 1954. Submicroscopic vesicular component in the synapse. Fed. Proc. 13, 35. [40] DeChiara, T.M., Bowen, D.C., Valenzuela, D.M., Simmons, M.V., Poueymirou, W.T., Thomas, S., Kinetz, E., Compton, D.L., Rojas, E., Park, J.S., Smith, C., Di Stefano, P.S., Glass, D.J., Burden, S.J., and Vancopoulos, G.D., 1996. The receptor tyrosine kinase MuSK is required for neuromuscular junction formation in vivo. Cell. 85, 501–512. [273, 274] Deiters, O., 1865. Untersuchungen überGehrin undRückenmark des Menschen und der Säugethiere (M. Schultze, ed.). F. Braunschweig: Vieweg und Sohn: 18. [17, 18] del Castillo, J., Katz, B., 1956. Biophysical aspects of neuromuscular transmission. Prog. Biophys. Biophys. Chem. 6, 121–170. [36, 40] del Castillo, J., Katz, B., 1954a. Quantal components of the end-plate potential. J. Physiol. 124, 560–573. [36, 167, 169, 170, 186] del Castillo, J., Katz, B., 1954b. The effect of magnesium on the activity of motor nerve endings . J. Physiol. 124, 553– 559. [139, 170] del Castillo, J., Katz, B., 1954c. Changes in end-plate activity produced by presynaptic polarization. J. Physiol. 124, 586– 604. [139] Del Castillo, J., Katz, B., 1955. On the localization of acetylcholine receptors. J. Physiol. 128, 157–181. [79] del Castillo, J., Katz, B., 1957. Modificiations de la membrane produites par des influx nerveux dans la ref on du pacer-maker der coeur. Microphysiologie comparee des elements excitables. Colloq. Int. C.N.R.S. (Paris) 67, 271–279. [28, 30, 117] del Castillo, J., Stark, L., 1952. The effect of calcium ions on the motor end-plate potentials. J. Physiol. 116, 507–515. [135, 157] Delay, J., Deniker, P., 1952. Le traitment des psychoses par une methode neurolytique derivee de l’hibernotherapie. C.R. 50e Congres des alienistes et neurol. de langue Fse. 495– 502. [91] Delay, J., Deniker, P., 1957. Psychotropic Drugs, (Garattini, S., Ghetti, V., (eds.) pp. 485–501, Elsevier, Amsterdam. [91] Delay, J., Deniker, P., Harl, J.M., 1952. Utilisation en therapeutique psychiatrique d’une phenothiazine d’action centrale elective. Ann. Med. Psychol. 110, 112–117. [91] Delay, J., Deniker, P., Tardieu, Y., Lemperiere, Th., 1954. Pre-miers essais on therapeutique psychiatrique de la reserpine, alcalooide nouveau de la Rauwolfia serpentina. CR 52c Congres des alienistes et neurol de Langue Fse. 836–841. [91] Deniker, P., 1970. Introduction of neuroleptic chemotherapy into psychiatry. In: Discoveries in Biological Psychiatry. Ayd, F.J., Blackwell, B., (eds.) Lippincott, Philadelphia, pp. 155–163. [91] Deniker, P., 1989. From chlorpromazine to tardive dyskinesia (Brief history of the Neuroleptics). Psychiatr. J. Univ. Ot-tawa. 14, 253–259. [91] Deniker, P., 1990., The neuroleptics: a historical survey. Acta. Psychiatr. Scand. 82, (Suppl 358), 83–87. [91] Dennis, M.J., Harris, A.J., 1980. A transient inability of neonatal rat motoneurones to reinnervate muscle. Develop. Biol. 74, 173–183. [247] Dennis, M.J., Ziskind-Conhaim, L., Harris, A.J., 1981. Development of neuromuscular junctions in rat embryos. Develop. Biol. 81, 266–279. [242] Desai, K.M., Sessa, W.C., Vane, J.R., 1991a. Involvement of nitric oxide in the reflex relaxation of the stomach to ac-commodate food or fluid. Nature. 351, 477–479. [127] Desai, K.M., Zembowicz, A., Sessa, W.C., Vane, J.R., 1991b. Nitroxergic nerves mediate vagally induced relaxation in the isolated stomach of the guinea pig. Proc. Nat. Acad. Sci. USA 88, 11490–11494. [128] Descartes, R., 1662. De Homine Figuris, et Latinitate Donatus Forentino Schuyl. Moyardum & Leffen, Leyden. [4, 6] Descartes, R., 1972. Traité de l’Homme. (Including La Formation du Foetus). Clerselier, Le Gras, Paris (ed.). Translated by T.S.Hall. Harvard University Press, Cambridge, Mass; 1664. [4, 6] Di Marzo, V, Vial, D., Sokoloff, P., Schwartz, J.C., Piomelli, D., 1993. Selection of alternative G-mediated signaling pathways at the dopamine D2 receptor by protein kinase C. J. Neurosci. 13, 4846–4853. [103] Diamond, A., 1990. Rate of maturation of the hippocampus and the developmental progression of children’s performance on the delayed non-matching to sample and visual paired comparison tasks. Annals. N.Y. Acad. Sci. 608, 394–433. [217] Ding, R., Jansen, J.K.S., Laing, N.G., Tonnesen, H., 1983. The innervation of skeletal muscles in chickens curarised during early development. J. Neurocytol. 12, 87–919. [247] Dismukes, K., de Boer, A.A., Mulder, A.H., 1977. On the mechanism of alpha-receptor mediated modulation of 3H-noradrenaline release from slices of rat brain neocortex. Naunyn-Schmiedebergs. Arch. Pharmacol. 299, 115–122. [73–75] Divry, P., Bobon, J., Collard, J., Pinchard, A., Nols, E., 1959. Etude et experimentation cliniques du R 1625 ou haloperidol, nouveau neuroleptique et ‘neurodystleptique’. Acta. Neurol. Psychol. (Belg.) 59, 337–367. [92] Dixon, W.E., 1906. Vagus inhibition. Brit. Med. J. Dec. 22nd: 1807. [54] Dodge, F.A. Jr., Rahamimoff, R., 1967. Co-operative action of calcium ions in transmitter release at the neuromuscular junction. J. Physiol. 193, 419–432. [39, 40, 145, 147, 158] Dolphin, A.C., Errington, M.L., Bliss, T.V., 1982. Long-term potentiation of the perforant path in vivo is associated with increased glutamate release. Nature. 297, 496–498. [223– 225] Dooley, D.J., Lupp, A., Hertting, G., Osswald, H., 1988. Omega-conotoxin GVIA and pharmacological modulation of hippocampal noradrenaline release. Eur. J. Pharmacol. 148, 261–267. [155] Douglas, R.M., Goddard, G.V., 1975. Long-term potentiation of the perforant path-granule cell synapse in the rat hippocampus. Brain. Res. 86, 205–215. [211] Doyon, M., 1894. De l’action exercee par le systeme nerveux sur l’appareil exreteur de la bile. Arch, de Physiol. Norm et Path. (Paris) VI, 887. [111] Draper, M.H., Weidmann, S., 1951. Cardiac resting and action potentials recorded with an intracellular electrode. J. Physiol. 115, 74–94. [142]
REFERENCES
299
Drejer, J., Honoré, T., 1987. Functional studies of a new non-NMDA antagonist (FG9041) in neuronal cell culture models. Soc. Neurosci. Abst. 13, 209. [88] Drejer, J., Honoré, T., 1988. New quinoxalinediones show potent antagonism of quisqualate responses in cultured mouse cortical neurons. Neurosci. Lett. 87, 104–108. [89] du Bois-Reymond, E., 1841. Untersuchungen über thierische Electricität. 4 vols. Reimer, Berlin. Vol. 1. [14] du Bois-Reymond, E., 1842. Untersuchungen über thierische Electricität . 4 vols. Reimer, Berlin. Vol 2. part 1. [14] du Bois-Reymond, E., 1845. Untersuchungen über thierische Electricität. 4 vols. Reimer, Berlin. Vol. 2, part 2, pp. 385– 579. [14] du Bois-Reymond, E., 1860. Untersuchungen über thierische Elektricität. 4 vols. Reiner, Berlin. Vol. 2, part 2, pp. 1– 384. [14] du Bois-Reymond, E., 1877. Gesammette Abhandlungen zur allgemeinen Muskel-und Nervenphysik. 2 vols. Verlag von Veit & Co. Leipzig. [10, 14] Du, C., Murray, J., Bates, J.N., Conklin, J.L., 1991. Nitric oxide: mediator of NANC hyperpolarization of opossum esophageal smooth muscle. Amer. J. Physiol. 261, G1012-G1016. [128] Dubocovich, M.L., Langer, S.Z., 1974. Negative feed-back regulation of noradrenaline release by nerve stimulation in the perfused cat’s spleen: differences in potency of phenoxybenzamine in blocking the pre- and post-synaptic adrenergic receptors. J. Physiol. 237, 505–591. [71, 72] Dudel, J., Kuffler, S.W., 1961. The quantal nature of transmission and spontaneous miniature potentials at the crayfish neuromuscular junction. J. Physiol. 155, 514–529. [169, 170] Duffy, C., Teyler, T.J., Shashoua, V.E., 1981. Long term potentiation in the hippocampal slice: evidence for stimulated secretion of newly synthesized proteins. Science. 212, 1148–1151. [223, 226] Dun, N.J., Dun, S.L., Forstermann, U., Tseng, L.F., 1992. Nitric oxide synthase immunoreactivity in rat spinal cord. Neurosci. Letts. 147, 217–220. [132] Dunlap, K., Luebke, J.I., Turner, T.J., 1995. Exocytotic Ca2+ channels in mammalian central neurons. Trends NeuroSci. 18, 89–98. [155] Dusser De Barrenne, J.G., 1933. The mode and site of action of strychnine in the nervous system. Physiol. Rev. 13, 325– 335. [79] Duxson, M.J., 1982. The effect of postsynaptic block on development of the neuromuscular junction in postnatal rats. J. Neurocytol 11, 395–408. [247] Duxson, M.J., Vrbova, G., 1985. Inhibition of acetylcholinesterase accelerates axon terminal withdrawal at the developing rat neuromuscular junction. J. Neurocytol. 14, 337–363. [247] Eccles, J.C., 1934. Synaptic transmission through a sympathetic ganglion. J. Physiol 81, 8P–9P. [58] Eccles, J.C., 1936. Synaptic and neuro-muscular transmission. Ergebn. Physiol. 38, 339–444. [32, 35] Eccles, J.C., 1937. The discharge of impulses from ganglion cells. J. Physiol. 91, 1–22. [32] Eccles, J.C., 1943. Synaptic potentials and transmission in sympathetic ganglion. J. Physiol. 101, 465–483. [32, 36] Eccles, R.M., 1955. Intracellular potentials recorded from a mammalian sympathetic ganglion. J. Physiol. 130, 572– 584. [32, 36] Eccles, J.C., 1961. The mechanism of synaptic transmission. Ergebn. Physiol. 51, 299–430. [117] Eccles, J.C., 1962. The organization of presynaptic inhibition in the spinal cord. Papers presented at the fourth meeting of the Australian Physiol. and Pharmacol. Soc. 23rd–25th May, 1962. Dept. of Physiol., Univ. of Melbourne, first formally published in Proc. Aust. Physiol. Pharmacol. Soc. 20 (Suppl.) 41. [117] Eccles, J.C., 1969. The inhibitory pathways of the central nervous system. The Sherrington Lectures IX, Liverpool University Press. [40, 41] Eccles, J.C., Katz, B., Kuffler, S.W., 1941. Nature of the ‘endplate potential’ in curarized muscle. J. Neurophysiol. 4, 362–387. [35, 36, 62] Eccles, J.C., Katz, B., Kuffler, S.W., 1942. Effect of eserine on neuromuscular transmission. J. Neurophysiol. 5, 211–230. [34] Eccles, J.C., Kuffler, S.W., 1941. Initiation of muscle impulses at neuro-muscular junction. J. Neurophysiol. 4, 402–417. [34] Eccles, J.C., Llinas, R., Sasaki, K., 1966. The inhibitory interneurones within the cerebellar cortex. Exp. Brain. Res. 1, 1–16. [40, 41] Eccles, J.C., Magladery, J.W., 1936. Pharmacological investigation on smooth muscle. J. Physiol. 87, 87P. [32] Eccles, J.C., Magladery, J.W., 1937. The excitation and response of smooth muscle. J. Physiol. 90, 31–67. [32] Eccles, J.C., O’Connor, W.J., 1938. Action potentials evoked by indirect stimulation of curarized muscle. J. Physiol. 94, 9P–11P. [34, 35, 62] Eccles, J.C., O’Connor, W.J., 1939. Responses which nerve impulses evoke in mammalian striated muscle. J. Physiol. 97, 44–102. [34] Eccles, J.C., Sherrington, C.S., 1930a. Reflex summation in the ipsilateral spinal flexion reflex. J. Physiol. 69, 1–28. [27, 29, 36] Eccles, J.C., Sherrington, C.S., 1930b. Improved bearing for the torsion myograph. J Physiol. 69, Pi. [27] Eccles, J.C., Sherrington, C.S., 1931. Studies on the flexor reflex VI. Inhibition. Proc. Roy. Soc Lond. B. 109, 91–113. [27] Edwards, F.A., Konnerth, A., Sakmann, B., 1990. Quantal analysis of inhibitory synaptic transmission in the dentate gyrus of rat hippocampal slices: a patch-clamp study. J. Physiol. 430, 213–249. [190, 193] Edwards, C., Kuffler, S.W., 1959. The blocking effect of gamma-aminoaminobutyric acid (GABA) and the action of related compounds on single nerve cells. J. Neurochem. 4, 19–30. [83] Ehrenberg, C.G., 1837. Beobachtungeiner auffallenden bisher unerkannten Strukfurdes Seelenorgans bei Menschen und Thieren. Königlichen Akademie der Wissenschaft en, Berlin. English transl, in Edinburgh Med. Surg. J. 48, P258– P309, 1836. [15] Ehrlich, P., 1885. The requirement of the organism for oxygen: an analytical study with the aid of dyes. In: F.Himmelweit (ed.)., 1956. Vol. 1, 433–496. The Collected Papers of Paul Ehrlich, 4 vols. Pergamon Press, London. [45] Ehrlich, P., 1897. Uber die Constitution des Diphtheriegiftes. In: F.Himmelweit (ed.). (1957) Vol. 2: pp. 126–133. The Collected Papers of Paul Ehrlich, 4 vols. Pergamon Press, London. [45] Ehrlich, P., 1900. Croonian Lecture. On immunity with special reference to cell life. Proc. Roy. Soc. Lond. LXVI, 424– 448. [45, 51]
300
REFERENCES
Ehrlich, P., 1902. The relations existing between chemical constitution, distribution and Pharmacological Action. In: Himmelweit, F. (ed.). The Collected Papers of Paul Ehrlich, Vol. 1 (1956). Pergamon Press, London, pp. 596–618 (4 vols). [47] Ehrlich, P., 1907. Experimental researches on specific therapy. In: F.Himmelweit (ed.). 1960, Vol. 3, pp. 106–134. The Collected Papers of Paul Ehrlich, 4 vols. Pergamon Press, London. [47] Ehrlich, P., 1914. Chemotherapy. In: F.Himmelweit (ed.), 1960. Vol. 3, pp. 505–518. The Collected Papers of Paul Ehrlich, 4 vols. Pergamon Press, London. [54] Ehrlich, P., Morgenroth, J., 1900. On haemolysins: third communication. In: F.Himmelweit (ed.), 1957. Vol. 2, pp. 205– 212. The Collected Papers of Paul Ehrlich, 4 vols. Pergamon Press, London. [47] Elizalde, A., Huerta, M., Stefani, E., 1983. Selective reinnervation of twitch and tonic muscle fibres of the frog. J. Physiol. 340, 513–524. [252, 262] Elliott, P., Marsh, S.J., Brown, D.A., 1989. Inhibition of Ca-spikes in rat preganglionic cervical sympathetic nerves by sympathomimetic amines. Brit. J. Pharmacol. 96, 65– 76. [76] Elliott, T.R., 1904a. On the action of adrenalin. J. Physiol. 34, Pxx–xxi. [27, 50, 65, 135] Elliott, T.R., 1904b. On the innervation of the ileo-colic sphincter. J. Physiol. XXXI, 157–168. [48] Elliott, T.R., 1905. The action of adrenaline. J. Physiol. XXXII, 401–467. [50, 103] Ellis, J.L., Conanan, N.D., 1994. Effect of potassium channel blockers on relaxations to a nitric oxide donor and to nonadrenergic nerve stimulation in guinea pig trachea. J. Pharmacol. Exper, Therapeut. 271, 782–786. [128] Elsberg, C.A., 1917. Experiments on motor nerve regeneration and the direct neurotisation of paralysed muscles by their own and foreign nerves. Science. 45, 318–320. [233] Engelhart E., Loewi, O., 1930. Fermentative Azetylcholin-spaltung im Blut und ihre Hemmung durch Physostigmin. Naunyn-Schmiedeberg’s Archiv. fur Experimentelle. Pathologie, und Pharmakologie. 150, 1–13. [57] English, A.W., Schwartz, G., 1995. Both basic fibroblast growth factor and ciliary neurotrophic factor promote the retention of polyneuronal innervation of developing skeletal muscle fibers. Develop. Biol. 169, 57–64. [277] Escher, G., Bechade, C, Levi, S., Triller, A., 1996. Axonal targeting of agrin in cultured rat dorsal horn neurons. J. Cell Sci. 109, 2959–2966. [279] Euler, U.S.V., 1934. Arch. exp. Path. Pharmak. 175, 78. [118] Euler, U.S.V., 1946. A specific sympathomimetic Ergone in Adrenergic Nerve Fibres (Sympathin) and its relations to Adrenaline and Noradrenaline. Acta. Physiol. Scand. 12, 73–97. [38, 66, 67] Euler, U.S.V., Gaddum, J.H., 1931. An unidentified depressor substance in certain tissue extracts. J. Physiol. 72, 74–87. [118] Evans, R.H, Francis, A.A., Watkins, J.C., 1978. Mg2+-like selective antagonism of excitatory amino acid-induced responses by alpha, epsilon-diaminopimelic acid, D-alpha-aminoadipate and HA-966 in isolated spinal cord of frog and immature rat. Brain Res. 148, 536–542. [87, 88] Everson, S., 1996. Psychology. In: ‘The Cambridge Companion to Aristotle’. Barnes, J. (ed.). pp. 168–184. Cambridge University Press. [1, 3] Ewins, A.J., 1914. Acetylcholine, a new active principle of ergot. Biochem. J. VIII, 44–49. [57] Fagg, G.E., Lane, J.D., 1979. The uptake and release of putative amino acid neurotransmitters. Neurosci. 4, 1015–1036. [87] Fairman, W.A., Vandenberg, R.J., Arriza, J.L., Kavanaugh, M.P., Amara, S.G., 1995. An excitatory amino acid transporter with properties of ligand-gated chloride channel. Nature. 375, 599–603. [87] Falls, D.L., Rosen, K.M., Corfas, G., Lane, W.S., Fischbach, G.D., 1993. ARIA, a protein that stimulates acetylcholine receptor synthesis, is a member of the neu ligand family. Cell. 72(5), 801–815. [274] Farde, L., 1996. The advantage of using positron emission tomography in drug research. Trends Neurosci. 19, 211– 214. [101] Farde, L., Wiesel, F.A., Stone-Elander, S., Halldin, C., Nordstrom, A.L., Hall, H., Sedvall, G., 1990. D2 dopamine receptors in neurolepticnaive schizophrenic patients. A positron emission tomography study with [11C]raclopride. Arch. Gen. Psychiat. 47, 213–219. [99] Farnebo, L.O., Hamberger, B., 1971. Drug-induced changes in the release of (3H)-noradrenaline from field stimulated rat iris. Brit. J. Pharmacol. 43, 97–106. [69–72, 77] Farrugia, G., Irons, W.A., Rae, J.L., Sarr, M.G., Szurszewski, J.H., 1993. Activation of whole cell currents in isolated human jejunal circular smooth muscle cells by carbon monoxide. Amer J. Physiol. 264, G1184–G1189. [127, 130] Fasth, S., Hulten, L., Nordgren, S., 1980. Evidence for a dual pelvic nerve influence on large bowel motility in the cat. J. Physiol. 298, 159–169. [113] Fatt, P., Katz, B., 1950. Some observations on biological noise. Nature. 166, 597–598. [165] Fatt, P., Katz, B., 1951. An analysis of the end-plate potential recorded with an intra-cellular electrode. J. Physiol. 115, 320–370. [35, 36] Fatt, P., Katz, B., 1952a. Spontaneous subthreshold activity at motor nerve endings. J. Physiol. 117, 109–128. [36, 38, 135, 139, 165, 167] Fatt, P., Katz, B., 1952b. The effect of sodium ions on neuromuscular transmission J. Physiol. 118, 73–87. [135, 137, 139] Fatt, P., Katz, B., 1953. The effect of inhibitory nerve impulses on a crustacean muscle fibre. J. Physiol. 121, 374– 389. [141] Fatt, P., Ginsborg, B., 1958. The ionic requirements of the production of action potentials in crustacean muscle fibre. J. Physiol. 142, 516–543. [115, 141] Feldberg,W.S., 1970. Henry Hallett Dale 1875–1968. Biograph. Mem. of Fellows Roy. Soc. 16, 77–174. [43] Feldberg, W., Gaddum, J.H., 1934. The chemical transmitter at synapses in a sympathetic ganglion. J. Physiol, 81, 305– 309. [32, 58] Feldberg, W., Lin, R.C.Y., 1949. The action of local anaesthetics and d-tubocurarine on the isolated intestine of the rabbit and guinea-pig. Br. J. Pharmacol. 4, 33–44. [125] Feldberg, W., Minz, B., Tsudzimura, H., 1934. The mechanism of the nervous discharge of adrenaline. J. Physiol. 80, 15P– 16P. [58]
REFERENCES
301
Feldberg, W., Vartiainen. A., 1934. Further observations on the physiology and pharmacology of a sympathetic ganglion. J. Physiol. 83, 103–128. [58] Feng, T.P., 1940. Studies on the neuromuscular junction. XVIII. The local potentials around N–M junctions induced by single and multiple volleys. Chin. J. Physiol. 15, 367– 404. [137] Feng, T.P., Li, T.H., 1941. Studies on the neuromuscular junction. XXIII. A new aspect of the phenomena of eserine potentiation and posttetanic facilitation in mammalian muscle. Chin. J. Physiol. 16, 37–50. [137] Fenwick, E.M., Marty, A., Neher, E., 1982a. A patch-clamp study of bovine chromaffin cells and of their sensitivity to acetylcholine. J. Physiol. 331, 577–597. [151] Fenwick, E.M., Marty, A., Neher, E., 1982b. Sodium and calcium channels in bovine chromaffin cells. J. Physiol. 331, 599– 635. [152] Ferns, M.J., Campanelli, J.T., Hoch, W., Scheller, R.H., Hall, Z., 1993. The ability of agrin to cluster AChRs depends on alternative splicing and on cell surface proteoglycans. Neuron. 11(3), 491–502. [273] Ferns, M., Deiner, M., Hall, Z., 1996. Agrin-induced acetylcholine receptor clustering in mammalian muscle requires tryosine phosphorylation. J. Cell Biol. 132, 937–944. [273] Fessard, A., 1967. Claude Bernard and The Physiology of Junctional Transmission. In: Claude Bernard and Experimental Medicine pp. 105–124. Grande, F., Visscher, M.B. (eds.). Schenkman Pub. Co. Inc.; Cambridge, Mass. [44, 45] Fex, S., Jirmanova, L, 1969. Innervation by nerve implants of ‘fast’ and ‘slow’ skeletal muscles of the rat. Acta. Physiol. Scand. 76, 257–269. [235] Fifkova, E., Andersen, C.L., Young, S.J., Van Harreveld, A., 1982. Effect of anisomycin on stimulation-induced changes in dendritic spines of the dentate granule cells. J. Neurocytol. 11, 183–210. [223] Fifkova, E., Van Harreveld, A., 1977. Long-lasting morphological changes in dendritic spines of dentate granular cells following stimulation of the entorhinal area. J. Neurocytol. 6, 211–230. [221–224] Filinger, E.J., Langer, S.Z., Perec, C.J., Stefano, F.J., 1978. Evidence for the presynaptic location of the alpha-adrenoceptors which regulate noradrenaline release in the rat submaxillary gland. Naunyn-Schmiedebergs Arch. Pharmacol. 304, 21–26. [71, 72] Finger, S., 1944. Origins of Neuroscience: a history of explo-rations into brain function. Oxford University Press. [1, 18, 43, 65] Finkleman, B., 1930. On the nature of inhibition in the intestine. J. Physiol. 70, 145–157. [108, 109] Fischbach, G.D., Nameroff, M., Nelson, P.G., 1971. Electrical properties of chick skeletal muscles fibers developing in cell culture. J. Cell Physiol. 78, 289–299. [238] Fladby, T., 1987. Postnatal loss of synaptic terminals in the normal mouse soleus muscle. Acta Physiol. Scand. 129(2), 229–238. [264] Fladby, T., Jansen, J.K., 1987. Postnatal loss of synaptic terminals in the partially denervated mouse soleus muscle. Acta Physiol Scand. 129 (2), 239–246. [252, 253] Fladby, T., Jansen, J.K., 1988. Selective innervation of neonatal fast and slow muscle fibres before net loss of synaptic terminals in the mouse soleus muscle. Acta Physiol Scand. 134(4), 561–562. [262] Fladby, T., Jansen, J.K., 1990. Development of homogeneous fast and slow motor units in the neonatal mouse soleus muscle. Develop. 109 (3), 723–732. [262] Florey, E., 1957. Chemical transmission and adaptation. J. Gen. Physiol. 40, 533–545. [79] Florey, E., McLennan, H., 1955. The release of an inhibitory substance from mammalian brain and its action on peripheral synaptic transmission. J. Physiol. 129, 384–392. [85] Fonnum, F., 1968. The distribution of glutamate decarboxylase and aspartate transaminase in subcellular fractions of rat and guinea-pig brain. Biochem. J. 106, 401–412. [85] Fonnum, F., 1984. Glutamate: a neurotransmitter in mammalian brain. J. Neurochem. 42, 1–11. [87] Fontaine, B., Changeux, J.P., 1989. Localization of nicotinic acetylcholine receptor alpha-subunit transcripts during myogenesis and motor endplate development in the chick. J. Cell Biol. 108(3), 1025–1037. [277] Fontaine, J., Grivegnee, A., Reuse, J., 1984. Adrenoceptors and regulation of intestinal tone in the isolated colon of the mouse. Brit. J. Pharmacol. 81, 231–243. [132] Fontana, F., 1760., Dissertation épistolaire adressée au R.P. Urbain Tosetti. In: Haller, A., ed. Mémoires sur les parties sensibles et irritables du corps animal. Vol. III, Lausanne: S.D’Arnay. 3, 157–243. [8] Fontana, F., 1767. De irritabilitatis legibers nunc primus sancitis et de spiritum animalium in movendis musculis inefficacia. Lucca: I.Ricomini. [8] Fontana, F., 1775. Richerche filosofiche sopra la fisica animale. Firenze: Cambia-gi; 1775. [8] Fontana, F., 1781. Traité sur le Venin de la Vipère sur les Poisons Américains, sur le Laurier-Cerise, et Quelque Autres Poisons Vegetaux. 2 Vols. Florence (English translation by J.Skinner. London 1787). [8] Forehand, C.J., Purves, D., 1984. Regional innervation of rabbit ciliary ganglion cells by the terminals of preganglionic axons. J. Neurosci. 4(1), 1–12. [277] Forel, A., 1887. Einige hirnanatomische Betrachtungen und Ergebnisse. Arch. Psychiat. Berlin 18, 162–198. [18] Foster, A.C., Fagg, G.E., 1984. Acidic amino acid binding sites in mammalian membranes: their characteristics and relationship to synaptic receptors. Brain Res. Rev. 7, 103–164. [88] Foster, M., Assisted by Sherrington, C.S., 1897. A textbook of physiology, part three: the central nervous system, 7th ed. London: MacMillan & Co. Ltd. [22, 23] Fox, A.P., Nowycky, M.C., Tsien, R.W., 1987. Single-channel recordings of three types of calcium channels in chick sensory neurones. J. Physiol. 394, 173–200. [153, 157]
302
REFERENCES
Francoeur, J.R., Richardson, P.M., Dunn, R.J., Carbonetto, S., 1995. Distribution of erb-B2, erb-B3, and erb-B4 in the developing avian nervous system. J. Neurosci. Res. 41(6), 836–845. [279] Frank, E., Jansen, J.K.S., Lomo, T., Westgaard, R., 1974. Main-tained function of foreign synapses on hyperinnervated skeletal muscle fibers of the rat. Nature. 247, 375–376. [235] Fraser, S.E., 1985. Cell interactions involved in neuronal patterning: an experimental and theoretical approach. In: Molecular bases of neural development. Edelman, G.M., Gall, E., Cowan, W.M. (eds.). pp. 481–507. Neurosci. Res. Fdn. [247] Fraser, S.E., Poo, M., 1982. Development, maintenance and modulation of patterned membrane topography: models based on the acetylcholine receptor. Curr. Top. Dev. Biol. 17, 77–100. [247] Fredericq, H., 1924. L’action bathmotrope du pneumogastrique cardiaque, appreciee, chez le chien, par la mesure de la chronaxie du myocarde ventriculaire. Bull, de l’Academie Royale de Medecine de Belgiqu. IV, 442–451. [59] Fredette, B.J., Landmesser, L.T., 1991. Relationship of primary and secondary myogenesis to fiber type development in embryonic chick muscle. Develop. Biol. 143(1), 1–18. [262] Fredholm, B.B., 1995. Purinoceptors in the nervous system. Pharmacol. Toxicol. 76, 228–239. [67] Freedman, JE., Aghajanian, G.K., 1984. Idazoxan (RX 781094) selectively antagonizes alpha 2-adrenoceptors on rat central neurones. Eur. J. Pharmacol. 105, 265–272. [73, 75] Fuder, H., Muscholl, E., Spemann, R., 1983. The determination of presynaptic pA2 values of yohimbine and phentolamine on the perfused rat heart under conditions of negligible autoinhibition. Brit. J. Pharmacol. 79, 109–119. [73] Fuhner, H., 1918. Untersuchungen uber den Synergismus von Giften. IV. Die Chemische Erregbarkeitssteigerung glatter Muskulatur. Arch. Exp. Pathol. Pharmacol. 82, 51–85. [58] Fukuda, M., Moreira, J.E., Lewis, F.M., Sugimori, M., Niinobe, M., Mikoshiba, K., Llinas, R., 1995. Role of the C2B domain of synaptotagmin in vesicular release and recycling as determined by specific antibody injection into the squid giant synapse preterminal. Proc. Natn. Acad. Sci. U.S.A. 92, 10708–10712. [160] Funte, L.R., Haydon, P.G., 1993. Synaptic target contact enhances presynaptic calcium influx by activating cAMP-dependent protein kinase during synaptogenesis. Neuron. 10(6), 1069–1078. [276] Furness, J.B., 1969. An electrophysiological study of the innervation of the smooth muscle of the colon. J. Physiol. 205, 549–562. [113, 117, 120, 122, 123] Furness, J.B., Costa, M., 1987. The Enteric Nervous System. Churchill Livingstone, London. [125, 127] Furness, J.B., Costa, M., Walsh, J.H., 1981. Evidence for and significance of the projection of VIP neurons from the myenteric plexus to the taenia coli in the guinea pig. Gastroenterol. 80(6), 1557–1561. [129] Furness, J.B., Li, Z.S., Young, H.M., Forstermann, U., 1994. Nitric oxide synthase in the enteric nervous system of the guinea-pig: a quantitative description. Cell Tiss. Res. 277, 139–149. [127, 128] Furness, J.B., Pompolo, S., Shuttleworth, C.W.R., Burleigh, D.E., 1992. Light- and electron-microscopic immuno-chemical analysis of nerve fibre types innervating the taenia of the guinea pig caecum. Cell Tiss. Res. 270, 125–137. [129, 131] Furness, J.B., Young, H.M., Pompolo, S., Bornstein, J.C., Kunze, W.A., McConalogue, K., 1995. Plurichemical transmission and chemical coding of neurons in the digestive tract. Gastroenterol. 108, 554–563. [127] Furukawa, T., Sasaoka, T., Hosoya, Y., 1959. Effects of tetrodotoxin on the neuromuscular junction. Jap. J. Physiol. 9, 143–152. [145] Gaddum, J.H., 1954. In: Ciba Foundation Symposium on Hypertension: Humoral and Neurogenic Factors, pp. 75– 77, Little Brown, Boston MA. [92] Gaddum, J.H., Schild, H., 1934. Depressor substances in extracts of intestine. J. Physiol 83, 1–14. [118] Galen. De motu musculorum. In: K.G.Kuhn (1821–1833) Medicorum Graecorum Opera Quae Exstant. Vols. 1–20 (Leipzig); 1821–1833. [3] Galen, O., 1962. Anatomical Procedures, The Later Books. C. Singer, transl. Cambridge University Press, Cambridge. [3] Galen, O., 1978. On the Doctrines of Hippocrates and Plato. P.DeLacy, transl. Adkademie-Veikag, Berlin. [3] Galen, O., 1968. On the Usefulness of the Parts of the Body. M.May, transl. Cornell University Press, Ithaca, N.Y. [3] Galvani, L., 1783. De volatilium aure. De Bononiensi Scientiarum et Artuim Institute atque Academia Commentarii 6, 420–424. [10] Galvani, L., 1791. De viribus electricitatis in motu musculari commentarius. De Bononiensi Scientiarum et Artium Instituto atque Academia Commentarii 7:363–418; (for translation see: M.G.Foley (1954). Galvani: Effects of electricity on muscular motion. Norwalk: Burndy Library). [9, 10] Galvani, L., 1794. Dell’uso e dell’attività dell’arco conduttore nelle contrazioni dei muscoli. Bologna, a S Tommaso d’Aquino. [10] Galvani, L., 1797. Memorie ed experimenti inediti. Bologna: Cappelli; 1937. [10] Galvani, L. 1794. Memorie sulla elettricità animale al celebre Abate Lazzaro Spallazani. Aggiunte alcune elettriche esperienze di Gio. Aldini. Bologna; Sassi. [10] Ganitkevich, V.I., Baidan, L.V., Tishkin, S.M., Shuba, M.F., 1983. Properties of synaptic currents during non-adrenergic inhibition of the smooth muscle cells of the large intestine of the guinea pig. Neirofiziologiya 15, 624–631. [127] Garattini, S., 1996. The role of independent science in psychopharmacology. In: The Psychopharmacologists, Healy, D., (ed.) pp 135–158. Chapman & Hall, New York. [91] Garry, R.C., 1934. The movements of the large intestine. Physiol. Rev. 14, 103–132. [125] Garry, R.C., Gillespie, J.S., 1955. The responses of the musculature of the colon of the rabbit to stimulation, in vitro, of the parasympathetic and of the sympathetic outflows. J. Physiol. 128, 557–576. [113] Gaskell, W.H., 1886. On the structure, distribution and function of the nerves which innervate the visceral and vascular systems. J. Physiol. VII, 1–80. [107, 114, 118, 120]
REFERENCES
303
Gaspar, P., Bloch, B., Le Moine, C., 1995. D1 and D2 receptor gene expression in the rat frontal cortex: cellular localization in different classes of efferent neurons. Eur. J. Neurosci. 7, 1050–1063. [102] Gates, H.J., Ridge, R.M., 1992. The importance of competition between motoneurones in developing rat muscle; effects of partial denervation at birth. J. Physiol. 445, 457–472. [262] Gautam, M., Noakes, P.G., Moscoso, L., Rupp, F., Scheller, R.H., Merlie, J.P., 1996. Defective neuromuscular synaptogenesis in agrindeficient mutant mice. Cell. 85(4), 525– 535. [273] Gautam, M., Noakes, P.G., Mudd, J., Nichol, M., Chu, G.C., Sanes, J.R., Merlie, J.P., 1995. Failure of postsynaptic spe-cialization to develop at neuromuscular junctions of rapsyn-deficient mice. Nature. 377(6546), 232–236. [274] Geduldig, D., Gruener, R., 1970. Voltage clamp of the Aplysia giant neurone: early sodium and calcium currents. J. Physiol. 211, 217–244. [141, 142] Gee, S.H., Montanaro, F., Lindenbaum, M.H., Carbonetto, S., 1994. Dystroglycan-alpha, a dystrophin-associated glycoprotein, is a functional agrin receptor. Cell. 77(5), 675–686. [274] Gelband, C.H., Silberberg, S.D., Groschner, K., Breemen, C., 1990. ATP inhibits smooth muscle Ca2(+)-activated K+ channels. Proc. Roy. Soc. Lond. Series B: Biol. Sci. 242, 38. [131] Geppert, M., Archer, B.T., Sudhof, T.C., 1991. Synaptotagmin II. A novel differentially distributed form of synaptotagmin. J. Biol Chem. 266, 13548–13552. [162] Geppert, M., Goda, Y, Hammer, R.E., Li, C., Rosahl, T.W., Stevens, C.F., Sudhof, T.C., 1994. Synaptotagmin I: a major Ca2+ sensor for transmitter release at a central synapse. Cell. 79, 717–727. [162] Gerren, R.A., Weinberger, N.M., 1983. Long term potentiation in the magnocellular medial geniculate nucleus of the anesthetized cat. Brain Res. 265, 138–142. [214] Gesemann, M., Cavalli, V, Denzer, A.J., Brancaccio, A., Schumacher, B., Ruegg, M.A., 1996. Alternative splicing of agrin alters its binding to heparin, dystroglycan, and the putative agrin receptor. Neuron. 16(4), 755–767. [274] Gesemann, M., Denzer, A.J., Ruegg, M.A., 1995. Acetylcholine receptor-aggregating activity of agrin isoforms and mapping of the active site. J. Cell Biol. 128(4), 625–636. [273] Gillespie, J.S., 1962. Spontaneous mechanical and electrical activity of stretched and unstretched intestinal smooth muscle cells and their response to sympathetic-nerve stimulation. J. Physiol. 162, 54–75. [124–127] Gilman, A.G., Goodman, L.S., Rall, T.W., Murad, F., (ed.) 1985. ‘Goodman and Gilman’s The Pharmacological Basis of Therapeutics’ 7th Edition. Macmillan Publishing Company, New York. [106] Gingrich, J.A., Caron, M.G., 1993. Recent advances in the molecular biology of dopamine receptors. Ann. Rev. Neurosci. 16, 299–321. [99] Giovannardi, S., Racca, C., Bertollini, L., Sturani, E., Peres, A., 1992. P2Y purinoceptors in normal NIH 3T3 and in NIH 3T3 overexpressing c-ras. Exper. Cell Res. 202, 398–404. [131] Giros, B., Sokoloff, P., Martres, M.P., Riou, J.P., Emorine, L.J., Schwartz, J.C., 1989. Alternative splicing directs the expression of two D2 dopamine receptor isoforms. Nature. 342, 923–926. [98] Glass D.J., Bowen, D.C., Stitt, T.N., Radziejewski, C., Bruno, J., Ryan, T.E., Gies, D.R., Shah, S., Mattsson, K., Burden, S.J., DiStefano, P.S., Valenzuela, D.M., DeChiara, T.M., Yancopoulos, G.D., 1996. Agrin acts via a MuSK receptor complex. Cell. 85(4), 513–523. [274] Goda, Y, Stevens, C.F., 1994. Two components of transmitter release at a central synapse . Proc. Natn. Acad. Sci. U.S.A. 91, 12942–12946. [160] Godfraind, J.M., Krnjevic, K., Pumain, R., 1970. Doubtful value of bicuculline as a specific antagonist of GABA. Nature 228, 675–676. [85] Godfrey, E.W., Nitkin, R.M., Wallace, E.G., Rubin, L.L., McMahan, U.J., 1984. Components of Torpedo electric organ and muscle that cause aggregation of acetylcholine receptors on cultured muscle cells. J. Cell Biol. 99, 615–627. [271] Goldsmith, S.K., Joyce, J.N., 1994. Dopamine D2 receptor expression in hippocampus and parahippocampal cortex of rat, cat and human in relation to tyrosine hydroxylase-immunoreactive fibers. Hippocamp. 4, 354–373. [102] Goldspink, G., Tabary, C., Tabary, J.C., Tardieu, C., Tardieu, G., 1974. Effect of denervation on the adaptation of sarcomere number and muscle extensibility to the functional length of the muscle. J. Physiol. 236, 733–742. [238] Golgi, C., 1886a. Sulla fina anatomia degli organi centrali del sistema nervosa. Milano: Hoepli. [17, 18] Golgi, C., 1886b. Sur l’anatomie microscopique des organes centraux du système nerveux. Arch. Ital. Biol. 7, 15–47. [18] Gonzalez Burgos, G.R., Biali, F.I., Cherksey, B.D., Sugimori, M., Llinas, R.R., Uchitel, O.D., 1995. Different calcium channels mediate transmitter release evoked by transient or sustained depolarization at mammalian sympathetic ganglia. Neurosci. 64, 117–123. [158] Goodearl, A.D., Yee, A.G., Sandrock, A.W. Jr., Corfas, G., Fischbach, G.D., 1995. ARIA is concentrated in the synaptic basal lamina of the developing chick neuromuscular junction. J. Cell Biol. 130(6), 1423–1434. [274] Gopfert, H., Schaefer, H., 1938. Uber den direkt und indirekt erreglen Aktionssfrom und die Funktion der motorischen endplatt. Pflugers. Arch. Ges. Physiol. 239, 597–619. [34, 35] Gopfert, H., Schaefer, H., 1937. Uber den direkt und indirekt erregten Aktionsstrom und die Funktion der motorischen Endplatte. Pflugers Archiv fur die Gesamte Physiologie des Menschen und der Tiere. 239, 597–619. [62] Gordon, H., van Essen, D.C., 1983. The relation of neuromuscular synapse elimination to spinal position of rabbit and rat soleus motoneurones. J. Physiol. 339, 591–597. [264] Gordon, H., Van Essen, D.C., 1985. Specific innervation of muscle fibre types in a developmentally polyinnervated muscle. Develop. Biol. 111, 42–50. [262]
304
REFERENCES
Gorecki, D.C., Derry, J.M., Barnard, E.A., 1994. Dystroglycan: brain localisation and chromosome mapping in the mouse. Human Molec. Genet. 3(9), 1589–1597. [279] Gothert, M., Pohl, I.M., Wehking, E., 1979. Effects of presynaptic modulators on Ca2+ induced noradrenaline release from central noradrenergic neurones. Noradrenaline and enkephalin inhibit release by decreasing depolarisation-induced Ca2+ influx. NaunynSchmiedebergs Arch. Pharmacol. 307, 21–27. [75] Gouze, J.L., Lasry, J.M., Changeux, J.P., 1983. Selective stabilization of muscle innervation during development: a mathematical model. Biol Cybern. 46, 207–215. [242] Goyal, R.K., Rattan, S., Said, S.I., 1980. VIP as a possible neurotransmitter of non-cholinergic non-adrenergic inhibitory neurones. Nature. 288, 378–380. [129] Gozenbach, H.R., Waser, P.G., 1973. Electron microscopic studies of degeneration and regeneration of rat neuromuscular junctions. Brain Res. 63, 167–172. [235] Grace, A.A., Bunney, B.S., Moore, H., Todd, C.L., 1997. Dopamine-cell depolarization block as a model for the therapeutic actions of antipsychotic drugs. Trends. Neurosci. 20, 31–37. [104] Graham, L.T., Shank, R.P., Werman, R., Aprison, M.H., 1967. Distribution of some synaptic transmitter suspects in cat spinal cord, glutamic acid, aspartate acid, gamma-aminobutyric acid, glycine and glutamine. J. Neurochem. 14, 465–472. [82] Grande, F., Visscher, M.B., 1967. Claude Bernard and Experimental Medicine. Schenkman Publishing Co. Inc., Cambridge, Mass. [46] Grandy, D.K., Litt, M., Allen, L., Bunzow, J.R., Marchionni, M., Makam, H., Reed, L., Magenis, R.E., Civelli, O., 1989. The human dopamine D2 receptor gene is located on chromosome 11 at q22–q23 and identifies a TaqI RFLP. Amer. J. Hum. Gen. 45, 778–785. [98] Grandy, D.K., Zhang, Y.A., Bouvier, C., Zhou, Q.Y., Johnson, R.A., Allen, L., Buck, K., Bunzow, J.R., Salon, J., Civelli, O., 1991. Multiple human D5 dopamine receptor genes: a functional receptor and two pseudogenes. Proc. Nat. Acad. Sci. USA. 88, 9175–9179. [98] Granit, R., 1966. Charles Scott Sherrington: An Appraisal. Thomas Nelson, London. [21] Graybiel, A.M., Moratalla, R., Robertson, H.A., 1990. Am-phetamine and cocaine induce drug-specific activation of the c-fos gene in striosoma-matrix compartments and limbic subdi vidions of the striatum. Proc. Nat. Acad. Sci. USA. 87, 6912–6916. [104] Greeff, K., Kasperat, H., Osswald, W., 1962. Paradoxe Wirkungen der elektrischen Vagusreizung am isolierten Magen- und Herzvorhofpraparat des Meerschweinchens sowie deren Beeinflussung durch Ganglienblocker, Sympathicolytica, Reserpin und Cocain. Naunyn-Schmiedeberg’s Arch. Pharmacol. Exp. Path. Pharmakol. 243, 528–545. [113] Grider, J.R., Jin, J.G., 1993. Vasoactive intestinal peptide release and L-citrulline production from isolated ganglia of the myenteric plexus: evidence for regulation of vasoactive intestinal peptide release by nitric oxide. Neurosci. 54, 521–526. [128, 129] Grider, J.R., Rivier, J.R., 1990. Vasoactive intestinal peptide (VIP) as transmitter of inhibitory motor neurons of the gut: evidence from the use of selective VIP antagonists and VIP antiserum. J. Pharmacol. Exper. Therapeut. 253, 738–742. [129] Grider, J.R., Cable, M.B., Bitar, K.N., Said, S.I., Makhlouf, G.M., 1985. Vasoactive intestinal peptide. Relaxant transmitter in tenia coli of the guinea pig. Gastroenterol. 89, 36– 42. [129] Grider, J.R., Murthy, K.S., Jin, J.G., Makhlouf, G.M., 1992. Stimulation of nitric oxide from muscle cells by VIP: prejunctional enhancement of VIP release. Amer. J. Physiol. 262, G774–G778. [129] Gross, H., Langner, E., Pfolz, H., 1974. Clozapin in long-term therapy of chronic schizophrenia. Arzneimittel-Forschung. 24, 987–989. [99] Gross, C.G., 1998. Brain, Vision, Memory: Tales in the History of Neuroscience, pp. 10–12, Bradford Book, The MIT Press, Cambridge. [1] Guastella, J., Nelson., N., Nelson, H., Czyzyk, L., Keynan, S., Miedel, M.C., Davidson, N., Lester, H.A., Kanner, B.I., 1990. Cloning and expression of rat brain GAB A transporter. Science 249, 1303–1306. [87] Guth, L., 1962. Neuromuscular function after regeneration of interrupted nerve fibers into partially denervated muscle. Exp. Neurol. 6, 129–141. [254] Guth, L., Zalewski, A.A., 1963. Disposition of cholinesterase following implantation of nerve into innervated and denervated muscle. Exp. Neurol. 7, 316–322. [235] Gutmann, E., Hanzlikova, V., 1967. Effects of accessory nerve supply to muscle achieved by implantation into muscle during regeneration of its nerve. Physiol. Bohemoslov. 16, 244–250. [235] Gutmann, E., Young, J.Z., 1944. The reinnervation of muscles after various period of atrophy. J.Anat. 78, 15–43. [233, 235] Gwyn, D.G., Aitken, J.T., 1966. The formation of new motor endplates in mammalian skeletal muscle. J. Anat. 100, 111– 126. [235] Haefely, W.E., 1969. Effects of catecholamines in the cat superior cervical ganglion and their postulated role as physiological modulators of ganglionic transmission. Prog. Brain. Res. 31, 61–72. [75] Hagiwara, S., Naka, K., 1964. The initiation of spike potential in barnacle muscle fibers under low intracellular calcium. J. Gen. Physiol. 48, 141–162. [115, 141] Haldeman, S., Huffman, R.D., Marshall, K.C., McLennan, H., 1972. The antagonism of the glutamate-induced and synaptic excitations of thalamic neurones. Brain Res. 39, 419–425. [87, 88] Hall, H., Farde, L., Halldin, C., Hurd, Y.L., Pauli, S., Sedvall, G., 1996. Autoradiographic localization of extrastriatal D2-dopamine receptors in the human brain using [125I] epidepride. Synapse. 23, 115–123. [104] Hall, J.G., McLennan, H., Wheal, H.V., 1977. The actions of certain amino acids as neuronal excitants. J. Physiol. 272, 52–53. [88] Hall, T.S., 1975. A History of General Physiology: 600 BC to AD 1900. Vol. 1. From Pre-Socratic times to the Enlight-ment. Univ. of Chicago Press; Chicago & London. [1, 3, 4] Hamburger, V., 1975. Cell death in the development of the lateral motor column of the chick embryo . J. Comp. Neurol. 160, 535–546. [242, 268]
REFERENCES
305
Hamill, O.P., Marty, A., Neher, E., Sakmann, B., Sigworth, F.J., 1981. Improved patch-clamp techniques for high-resolu-tion current recording from cells and cell-free membrane patches. Pflugers Arch.-Eur. J. Physiol 391, 85–100. [151] Hamon, M., 1989. Dopamine and serotonin. Involvement in the central effects of psychotropic agents. Encephale 15 Spec No. 119–125. [91] Hansen, J.J., Krogsgaard-Larsen, P., 1990. Structural, con-formational, and stereochemical requirements of central excitatory amino acid receptors. Med. Res. Rev. 10, 55–94.5 [88] Hardman, V.J., Brown, M.C., 1987. Accuracy of reinnervation of rat internal intercostal muscles by their own segmental nerves. J. Neurosci. 7(4), 1031–1036. [264] Harris, E.W., Cotman, C.W., 1986. Long term potentiation of guinea pig mossy fiber responses is not blocked by N-methyl D-aspartate antagonists. Neurosci. Letts. 70, 132– 137 . [218, 230] Harris, E.W., Ganong, A.H., Cotman, C.W., 1984. Long-term potentiation in the hippocampus involves activation of N-methyl-D-aspartate receptors. Brain Res. 323, 132–137. [218] Harris, K.M., Teyler, T.J., 1984. Developmental onset of long-term potentiation in area CA1 of the rat hippocampus. J. Physiol. 346, 27–48. [216, 217] Harris, A.J., 1981. Embryonic growth and innervation of rat skeletal muscles. I. Neural regulation of muscle fire numbers. Phil. Trans. R. Soc. Lond. Ser. B. 293, 257–277. [250] Harris, A.J., McCaig, C.D., 1984. Motoneurone death and motor unit size during embryonic development of the rat. J. Neurosci. 4, 13–24. [242, 268] Harrison, J.S., McSwiney, B.A., 1936. The chemical transmitter of motor impulses to the stomach. J. Physiol. 87, 79– 86. [112, 113] Harrison, R.G., 1910. The outgrowth of the nerve fibre as a mode of protoplasmic movement . J. Exp. Zoo. 9, 787–848. [238] Harvey, A.M., Macintosh, F.C., 1940. Calcium and synaptic transmission in a sympathetic ganglion. J. Physiol. 97, 408–416. [39, 40, 135, 137] Hatt, H., Smith, D.O., 1976. Non-uniform probabilities of quantal release at the crayfish neuromuscular junction. J. Physiol. 259(2), 395–404. [175, 177] Hayashi, T, 1952. A physiological study of epileptic seizures following cortical stimulation in animals and its application to human clinics. Jap. J. Physiol. 3, 46–64. [79] Hayashi, T., 1956. Chemical Physiology of Excitation in Muscle and Nerve. Tokyo, Nakayama-Shoten. [79] Hayashi, T., 1959. The inhibition action of beta-hydroxy-gamma-aminobutyric acid upon the seizure following stimulation of the motor cortex of the dog. J. Physiol. 145, 570–578. [83] Hayashi, T., Nagai, K., 1952. Action of omega-amino acids on the motor cortex of higher animals especially gamma-amino-betahydroxybutyric acid as the real inhibitory principle in the brain. Abstr. XXth Int. Physiol. Congr. 410. [79] Haydon, P.G., Zoran, M.J., Man-Son-Hing, H.J., Sievers, E.M., Doyle, R.T., 1990. A relation between synaptic specificity and the acquisition of presynaptic properties. J. de Physiologie. 84(1), 111–120. [276] He, X.D., Goyal, R.K., 1993. Nitric oxide involvement in the peptide VIP-associated inhibitory junction potential in the guinea-pig ileum. J. Physiol. 461, 485–499. [127, 128, 132] Heidelberger, R., Heinemann, C., Neher, E., Matthews, G., 1994. Calcium dependence of the rate of exocytosis in a synaptic terminal. Nature. 371, 513–515. [162] Hell, J.W., Yokoyama, C.T., Wong, S.T., Warner, C., Shutch, T.P., Catterall, W.A., 1993. Differential phosphorylation of two size forms of the neuronal class C L-type calcium channel alpha 1 subunit. J. Biol. Chem. 268, 19451–19457. [103] Helmholtz, H., 1850a. Messungen uber den zeitlichen Verlauf der Zuchung animalischer Muskeln und die Fortpflanzun-gsgeschwindigkeit der Reizung in den Nerven . Arch. Anat. Physiol. 277–364. [13, 15] Helmholtz, H., 1850b. Note sur la vitesse de propagation de l’agent nerveux dans les nerfs rachidiens. C.R. Acad. Sci. (Paris) 30, 204–206. [15] Helmholtz, H., 1850c. Uber die Fortpflanzungsgeshwindikeit der Nervenreizung. Arch. Anat. Physiol. Wiss. Med. 71–73. [15] Helmholtz, H., 1851. Deuxième note sur la vitesse de propagation de l’agent nerveux. C.R. Acad. Sci. (Paris) 32, 262– 265. [15] Helmholtz, H., 1852. Messunger über fortpflanzungsgesch-windigkeit der reizung in den nerven-zweite reihe. Arch. Anat. Physiol. Wiss. Med. 199–216. [15] Hemmings, H.C. Jr, Nairn, A.C., Elliott, J.I., Greengard, P., 1990. Synthetic peptide analogs of DARPP-32 (Mr 32,000 dopamine- and cAMPregulated phosphoprotein), an inhibitor of protein phosphatase-1. Phosphorylation, dephosphorylation and inhibitory activity. J. Biol. Chem. 265, 20369–20376. [103] Henderson, V.E., Roepke, M.H., 1934. The role of acetylcholine in bladder contractile mechanisms and in parasympathetic ganglia J. Pharmacol. 51, 97–110. [32] Henningsen, I., Liestol, K., Maehlen, J., Nja, A., 1985. The selective innervation of guinea-pig superior cervical ganglion cells by sprouts from intact preganglionic axons. J. Physiol. 358, 239–253. [278] Herrera, A.A., 1984. Polyneuronal innervation and quantal transmitter release in formamide-treated frog sartorius muscles. J. Physiol. 355, 267–280. [246, 252] Herrera, A.A., Grinnell, A.D., 1985. Effects of changes in motor unit size on transmitter release at the frog neuromuscular junction. J. Neurosci. 5, 1896–1900. [242, 260, 261] Herrera, A.A., Werle, M.J., 1990. Mechanisms of elimination, remodeling, and competition at frog neuromuscular junctions. J. Neurobiol. 21(1), 73–98. [246] Hilbush, B.S., Morgan, J.I., 1994. A third synaptotagmin gene, Syt3, in the mouse. Proc. Natn. Acad. Sci. U.S.A. 91, 8195– 8199. [162]
306
REFERENCES
Hill, A.V., 1933. Excitation and accommodation in time. Proc. Roy. Soc. Lond. B. 119, 305–355. [36] Hill, A.V., Katz, B., Solandt, D.Y., 1936. Nerve excitation by alternating current. Proc. Roy. Soc. Lond. B. 121, 74–133. [34] Hillarp, N.A., Lagerstedt, S., Nilsson, B., 1953. The isolation of a granular fraction from the suprarenal medulla, containing the sympathomimetic catecholamines. Acta Physiol. Scand. 29, 251. [40] Hippius, H., 1989. The history of clozapine. Psychopharmacol. 99, S3–5. [99] Hippius, H., 1996. The founding of the CINP and the discovery of clozapine. In: The Psychopharmacologists, Healy, D., (ed.) pp 187–214. Chapman & Hall, New York. [91] Hirokawa, N., Heuser, J.E., 1982. Internal and external differ-entiations of the postsynaptic membrane at the neuromuscular junction. J. Neurocytol. 11, 487–510. [242] Hirst, G.D.S., McKirdy, H.C., 1974. A nervous mechanism for descending inhibition in guinea-pig small intestine. J. Physiol. 238, 129–143. [122–125] His, W., 1886. Zur Gechichte des menschlichen Ruckenmarkes und der Nervenwurzeln. Abhandl. Math.—Phys. Class. Konigl. Sach. Gesellsch. Wiss., Leipzig 13, 147–209. [18] His, W., 1889. Die Neuroblasten und deren Entstehung im embryonalen Mark; Abh. kgl. Sächs. Ges. Wissensch. math, phys. Kl. 15, 311–372. [18] Hoch, W., Campanelli, J.T., Harrison S., Scheller, R.H., 1994. Structural domains of agrin required for clustering of nicotinic acetylcholine receptors. EMBO J. 13(12), 2814–2821. [273] Hodgkin, A.L., Adrian, E.D., 1978. Baron Adrian of Cambridge. Biograph. Mem. Fell. Roy. Soc. 25, 1–73. [30, 34, 139– 142, 151] Hodgkin, A.L., Huxley, A.F., 1952. A quantitative description of membrane current and its application to conduction and excitation in nerve. J. Physiol. 117, 500–544. [139, 141, 142] Hodgkin, A.L., Keynes, R.D., 1957. Movements of labelled calcium in squid giant axons. J. Physiol. 138, 253–281. [139] Hoffman, B.F., Suckling, E.E., 1956. Effect of several cations on transmembrane potentials of cardiac muscle. Am. J. Physiol. 186, 317–324. [142] Huffman, H., 1951. Fate of interrupted nerve fibres regenerating into partially denervated muscles. Aust. J. Exp. Biol. Med. 29, 211–219. [239] Hogg, R.V., Craig, A.T., 1978. Introduction to Mathematical Statistics Macmillan. New York. [182, 189, 203] Hollmann, M., Heinemann, S., 1994. Cloned glutamate receptors. Annu. Rev. Neurosci. 17, 31–108. [90] Holman, M.E., 1958. Membrane potentials recorded with high-resistance micro-electrodes; and the effects of changes in ionic environment on the electrical and mechanical activity of the smooth muscle of the taenia coli of the guinea pig. J. Physiol. 141, 464–488. [142] Holman, M.E., Hughes, J.R., 1965. Inhibition of intestinal smooth muscle. Aust. J. Exp. Biol. Med. Sci. 43, 277–290. [114] Holzbauer, M., Vogt, M., 1956. Depression by reserpine of the noradrenaline concentration in the hypothalamus of the cat. J. Neurochem. 1, 8–11. [67] Horn, R., 1987. Statistical methods for model discrimination. Applications to gating kinetics and permeation of the acetylcholine receptor channel. Biophys. J. 51 (2), 25 5–263. [194– 197] Hoyle, C.H., Burnstock, G., 1989a. Galanin-like immunoreactivity in enteric neurons of the human colon. J. Anat. 166, 23–33. [127] Hoyle, C.H., Burnstock, G., 1989b. Neuronal populations in the submucous plexus of the human colon. J. Anat. 166, 7–22. [129] Hoyle, C.H.V., Burnstock, G., 1989c. Neuromuscular transmission in the gastrointestinal tract. In: Handbook of Physiology. The Gastrointestinal System. Motility and Circulation. Bethesda,MD: Am. Physiol. Soc.Vol 1, pt. 1, Ch.13, p.435– 464. [129] Hoyle, C.H., Knight, G.E., Burnstock, G., 1990. Suramin antagonizes responses to P2-purinoceptor agonists and purinergic nerve stimulation in the guinea-pig urinary bladder and taenia coli. Brit. J. Pharmacol. 99, 617–621. [129] Hu, G.Y., Hvalby, O., Walaas, S.I., Albert, K.A., Skjeflo, P., Andersen, P., Greengard, P., 1987. Protein kinase C injection into hippocampus pyramidal cells elicits features of long term potentiation. Nature. 328, 420–429. [228] Huber, O., Bertrand, C., Bunnett, N.W., Pellegrini, C.A., Nadel, J.A., Debas, H.T., Geppetti, P., 1993. Tachykinins contract the circular muscle of the human esophageal body in vitro via NK2 receptors. Gastroenterol. 105, 981–987. [132] Huizinga, J.D., Tomlinson, J., Pintin-Quezada, J., 1992. Involvement of nitric oxide in nerve-mediated inhibition and action of vasoactive intestinal peptide in colonic smooth muscle. J. Pharmacol Exper. Therapeut. 260, 803–808. [128] Hulten, L., 1969. Extrinsic nervous control of colonic motility and blood flow. An experimental study in the cat. Acta. Physiol Scand. Suppl. 335. [113] Hume, R.I., Purves, D., 1983. Apportionment of the terminals from single preganglionic axons to target neurones in the rabbit ciliary ganglion. J. Physiol. 338, 259–275. [277] Hunt, R., Taveau, R.M., 1906. On the physiological action of certain cholin derivatives and new methods for detecting cholin. Brit. Med. J. II, Dec 22, 1788–1791. [54, 57] Hunter, D.D., Llinas, R., Ard, M., Merlie, J.P., Sanes, J.R., 1992a. Expression of s-laminin and laminin in the developing rat central nervous system. J. Comp. Neurol. 323(2), 238–251. [279] Hunter, D.D., Murphy, M.D., Olsson, C.V., Brunken, W.J., 1992b. S-laminin expression in adult and developing retinae: a potential cue for photoreceptor morphogenesis. Neuron. 8(3), 399–413. [279] Hunter, D.D., Shah, V, Merlie, J.P., Sanes, J.R., 1989. Alaminin-like adhesive protein concentrated in the synaptic cleft of the neuromuscular junction. Nature. 338(6212), 229–234. [273, 275] Hunziker, F., Kunzle, F., Schmutz, J., 1963. Helv. Chim. Acta 46: 2337. [100] Huxley, A.F., Stampfli, R., 1951. Direct determination of membrane resting potential and action potential in single myelinated nerve fibres. J. Physiol 112, 476–495. [114]
REFERENCES
307
Ignarro, L.J., 1990. Nitric oxide. A novel signal transduction mechanism for transcellular communication. Hyperten. 16, 477–483. [132] Ip, M.C., Vrbova, B., 1973. Motor and sensory reinnervation of fast and slow mammalian muscles. Z. Zellforsch. Mikrosk. Anat. 146, 261–279. [235] Ireland, D., Davies, P., McLachlan, E.M., 1997. Voltage-activated Ca2+ channels mediating acetylcholine release in mammalian sympathetic ganglia. Proc. Aust. Neurosci. Soc. 8, 70. [158] Iselin, C.E., Martin, J.L., Magistretti, P.J., Ferrero, J.D., 1988. Stimulation by nicotine of enteric inhibitory nerves and release of vasoactive intestinal peptide in the taenia of the guinea-pig caecum. Europ. J. Pharmacol. 148, 179–186. [129] Itoh, H., Sakai, J., Imoto, A. and Creed, K.E., 1995. The control of smooth muscle tissues by nonadrenergic noncholinergic (NANC) nerve fibres in the autonomic nervous system. J. Smooth Mus. Res. 31, 67–78.7 [132] Ivers3n, L.L., 1967. The uptake and storage of noradrenaline in sympathetic nerves. Cambridge of the University Press. [69] Iversen, L.L., Johnston, G.A.R., 1971. GAB A uptake in central nervous system: Comparison of uptake in slices and homogenates and the effect of some inhibitors. J. Neurochem. 18, 1939–1950. [85] Iversen, L.L., Neal, M.J., 1968. The uptake of 3H-GABA by slices of rat cerebral cortex. J. Neurochem. 15, 1141–1149. [85] Iwayama, T., 1969. Relation of regenerating nerve terminals to original endplates. Nature. 224, 81–82. [235] Jacobson, M., 1993. Foundations of Neuroscience. Plenum Press. New York. [17] Jang, C.S., 1940. Interaction of sympathomimetic substances on adrenergic transmission. J. Pharmacol. 70, 347–361. [67] Jansen, J.K., Fladby, T., 1990. The perinatal reorganization of the innervation of skeletal muscle in mammals. Prog. Neurobiol. 34(1), 39–90. [246] Janssen, P., 1970. The Butyrophenone Story. In: Discoveries in Biological Psychiatry (Ayd, F.J., Blackwell, B. (ed.) Lippincott, Philadelphia, pp 165–179. [92] Jansson, G., 1969. Vago-vagal reflex relaxation of the stomach in the cat Acta . Physiol. Scand. 75, 245–252. [113] Jessell, T.M., Siegel, R.E., Fischbach, G.D., 1979. Induction of acetylcholine receptors on cultured skeletal muscle by a factor extracted from brain and spinal cord. Proc. Nat. Acad. Sci. USA. 76(10), 5397–4501. [274] Jessen, K.R., Saffrey, M.E., Burnstock, G., 1983. The enteric nervous system in tissue culture. I. Cell types and their interactions in explants of the myenteric and submucous plexus from guinea pig, rabbit and rat. Brain Res. 262, 17– 35. [131] Jessen, K.R., Saffrey, M.J., Van Noorden, S., Bloom, S.R., Polak, J.M., Burnstock, G., 1980. Immunohistochemical studies of the enteric nervous system in tissues culture and in situ: localization of vasoactive intestinal polypeptide (VIP), substance P and enkephalin immunoreactive nerves in the guinea-pig gut. Neurosci. 5, 1717–1735. [131] Jin, J.G., Katsoulis, S., Schmidt, W.E., Grider, J.R., 1994. Inhibitory transmission in tenia coli mediated by distinct vasoactive intestinal peptide and apamin-sensitive pituitary adenylate cyclase activating peptide receptors. J. Pharmacol. Exper. Therapeut. 270, 433–439. [127–130] Jin, J.G., Misra, S., Grider, J.R., Makhlouf, G.M., 1993a. Functional difference between SP and NKA: relaxation of gastric muscle by SP is mediated by VIP and NO. Amer. J. Physiol. 264, G678–G685. [128] Jin, J.G., Murthy, K.S., Grider, J.R., Makhlouf, G.M., 1993b. Activation of distinct cAMP- and cGMP-dependent pathways by relaxant agents in isolated gastric muscle cells. Amer. J. Physiol. 264, G470–G477. [129] Jo, S.A., Burden, S.J., 1992. Synaptic basal lamina contains a signal for synapse-specific transcription. Develop. 115(3), 673–680. [277] Jo, S.A., Zhu, X., Marchionni, M.A., Burden, S.J., 1995. Neuregulins are concentrated at nerve-muscle synapses and activate ACh-receptor gene expression. Nature. 373 (6510), 158–161. [274] Johanson, I.B., Hall, W.G., 1982. Appetitive conditioning in neonatal rats: conditioned orientation to a novel odour. Dev. Psychobiol. 15, 379–397. [217] Johnson, E.A., Lieberman, M., 1971. Heart: excitation and contraction. Annu. Rev. Physiol. 33, 479–532. [142] Johnson, D.A., Purves, D., 1981. Post-natal reduction of neural unit size in the rabbit ciliary ganglion. J. Physiol. 318, 143– 159. [278] Johnson, E.W., Wernig, A., 1971. The binomial nature of transmitter release at the crayfish neuromuscular junction. J. Physiol. 218(3), 757–767. [169, 170] Johnston, G.A., Mitchell, J.F., 1971. The effect of bicuculline, metrazol, picrotoxin and strychnine on the release of (3H) G ABA from rat brain slices. J. Neurochem. 18, 2441–2446. [71] Johnston, G.A.R., 1996. GABAC receptors: relatively simple transmitter-gated ion channels. Trends Pharm. Sci. 17, 319. [90] Johnston, G.A.R., Iversen, L.L., 1971. Glycine uptake in rat cerebral nervous system slices and homogenates: Evidence for different uptake systems in spinal cord and cerebral cortex. J. Neurochem. 18, 1951–1961. [85] Johnston, G.A.R., Krogsgaard-Larsen, P., Stephanson, A.L., 1976. Betel nut constituents as inhibitors of gamma-aminobutyric acid uptake. Nature. 258, 627–628. [87] Jonas, P., Major, G., Sakmann, B., 1993. Quantal components of unitary EPSCs at the mossy fibre synapse on CA3 pyramidal cells of rat hippocampus. J. Physiol. 472, 615–663. [195, 197] Jones, G., Meier, T., Lichtsteiner, M., Witzemann, V., Sakmann, B., Brenner, H.R., 1997. Induction by agrin of ectopic and functional postsynaptic-like membrane in innervated muscle. Proc. Nat.Acad. Sci. USA. 94(6), 2654–2659. [273] Joyce, J.N., 1993. The dopamine hypothesis of schizophrenia: limbic interactions with serotonin and norepinephrine. Psychopharmacol. 112 (Suppl 1), S16–S34. [101] Joyce, J.N., Lexow, N., Bird, E., Winokur, A., 1988. Organizaton of dopamine D1 and D2 receptors in human striatum: receptor autoradiographic studies in Huntington’s disease and schizophrenia. Synapse. 2, 546–557. [101] Jule, Y., 1980. Nerve mediated descending inhibition in the proximal colon of the rabbit J. Physiol. 309, 245–252. [123] Junge, D., 1967. Multi-ionic action potentials in molluscan giant neurones. Nature. 215, 546–548. [142]
308
REFERENCES
Kalsner, S., 1980. Limitations of presynaptic adrenoceptor theory: the characteristics of the effects of noradrenaline and phenoxybenzamine on stimulated-induced efflux of [3H] noradrenaline in vas deferens. J. Pharmacol. Exp. Therapeut. 212, 232–239. [73] Kandel, E.R., Spencer, W.A., Brindley, F.J., 1961. Electrophysiology of hippocampal neurones. I. Sequential inva-sion and synaptic organization. J. Neurophysiol. 24, 225– 242. [211] Kaprielian, Z., Patterson, P.H., 1993. Surface and cytoskeletal markers of rostrocaudal position in the mammalian nervous system. J. Neurosci. 13(6), 2495–2508. [283] Kaprielian, Z., Cho, K.O., Hadjiargyrou, M., Patterson, P.H., 1995. CD9, a major platelet cell surface glycoprotein, is a ROCA antigen and is expressed in the nervous system. J. Neurosci. 15(1 Pt 2), 562–573. [283] Katada, T., Ui, M., 1981. In vitro effects of islet-activating protein on cultured rat pancreatic islets. Enhancement of insulin secretion, adenosine 3’:5’-monophosphate accumulation and 45Ca flux. J. Biochem. 89, 979–990. [77] Katz, B., 1936a. Multiple response to constant current in frog’s medullated nerve. J. Physiol. 88, 239–255. [34] Katz, B., 1936b. Neuromuscular transmission in crabs. J. Physiol. 87, 199–221. [34] Katz, B., 1939. The ‘anti-curare’ action of a subthreshold catelectrotonus. J. Physiol. 95, 286–304. [34] Katz, B., 1978. Archibald Vivian Hill. Biograph. Mem. Fell. Roy. Soc. Lond. 24, 71–149. [34] Katz, B., 1981. Concluding Remarks. In: Chemical Neurotrans-mission: 75 years. Stjarne, L., Hedquist, P., Lagercrantz, H., Wennmalm, A. (eds.). Academic Press, London. [108] Katz, B., 1986. Bayliss-Starling Memorial Lecture. Reminis-cences of a Physiologist, 50 years after. J. Physiol. 370, 1–12. [34] Katz, B., 1996. The Fenn Lecture—Neural transmitter release: from quantal secretion to exocytosis and beyond. J. Neurocytol. 25, 677–686. [137] Katz, B., Kuffler, S.W., 1942. The effect of calcium on the neuro-muscular junction. J. Neurophysiol. 7, 17–26. [137] Katz, B., Miledi, R., 1965. The measurement of synaptic delay, and the time course of acetylcholine release at the neuromuscular junction. Proc. Roy. Soc. Lond. Ser. B. 161, 483– 495. [179, 180] Katz, B., Miledi, R., 1967a. The timing of calcium action during neuromuscular transmission . J. Physiol. 189, 535– 544. [145, 147] Katz, B., Miledi, R., 1967b. A study of synaptic transmission in the absence of nerve impulses. J. Physiol. 192, 407–436. [39, 40, 144, 145] Katz, B., Miledi, R., 1967c. Modification of transmitter release by electrical interference with motor nerve endings. Proc. R. Soc. Lond.Series B: Biol. Sci. 167, 1–7. [145] Katz, B., Miledi, R., 1967d. Tetrodotoxin and neuromuscular transmission Proc. R.Soc. Lond. B. 167, 8–22. [145] Katz, B., Miledi, R., 1967e. The release of acetylcholine from nerve endings by graded electric pulses. Proc. R. Soc. Lond.-Series B: Biol. Sci. 167, 23–38. [144, 145] Katz, B., Miledi, R., 1969. Tetrodotoxin-resistant electric activity in presynaptic terminals. J. Physiol. 203, 459–487. [146– 148] Katz, B., Miledi, R., 1970a. Further study of the role of calcium in synaptic transmission. J. Physiol. 207, 789–801.[147, 148] Katz, B., Miledi, R., 1970b. Membrane noise produced by acetylcholine. Nature. 226, 962–963. [61, 62] Katz, B., Miledi, R., 1971a. Further observations on acetylcholine noise. Nature New Biol. 232, 124–126. [61] Katz, B., Miledi, R., 1973. The characteristics of ‘end-plate noise’ produced by different depolarizing drugs. J. Physiol. 230, 707–717. [62] Kebabian, J.W., Greengard, P., 1971. Dopamine-sensitive adenyl cyclase: possible role in synaptic transmission. Science. 174, 1346–1349. [95] Kebabian, J.W., Petzold, G.L., Greengard, P., 1972. Dopamine-sensitive adenylate cyclase in caudate nucleus of rat brain, and its similarity to the ‘dopamine receptor’. Proc. Nat. Acad. Sci. USA. 69, 2145–2149. [95, 96] Kebabian, J.W., Calne, D.B., 1979. Multiple receptors for dopamine. Nature. 277, 93–96. [95, 98] Keef, K.D., Shuttleworth, C.W., Xue, C., Bayguinov, O., Publicover, N.G., Sanders, K.M., 1994. Relationship between nitric oxide and vasoactive intestinal polypeptide in enteric inhibitory neurotransmission. Neuropharmacol. 33, 1303–1314. [129] Keef, K.D., Du, C., Ward, S.M., McGregor, B., Sanders, K.M., 1993. Enteric inhibitory neural regulation of human colonic circular muscle: role of nitric oxide. Gastroenterol. 105, 1009–1016. [128] Keenan, M.J., Niedergerke, R., 1967. Intracellular sodium concentration and resting sodium fluxes of the frog heart ventricle. J. Physiol. 188, 235–260. [142] Keja, J.A., Stoof, J.C., Kits, K.S., 1992. Dopamine D2 receptor stimulation differentially affects voltage-activated calcium channels in rat pituitary melanotropic cells. J. Physiol. 450, 409–435. [103] Kelling, G., 1903. Untersuchungen uber die Spannungszustanade der Bauchwand, der Magen-under der Darmwand. Z. Biol. 44, 161–267. [111] Kemp, M.C., Kuonen, D.R., Sutton, A., Roberts, P.J., 1988. Rat brain NADPH-dependent diaphorase. A possible relationship to cytochrome P450 reductase. Biochem. Pharmacol. 37, 3063–3070. [132] Kendall, M., Stuart, A., Ord, J.K., 1987. Advanced Theory of Statistics, 5th Ed. Vol.1 Charles Griffith & Coy Ltd. London. [174, 199–202] Kerr, L.M., Yoshikami, D., 1984. A venom peptide with a novel presynaptic blocking action. Nature. 308, 282–284. [153, 155] Kirpekar, S.M., Puig, M., 1971. Effect of flow-stop on noradrenaline release from normal spleens and spleens treated with cocaine, phentolamine or phenoxybenzamine. Brit. J. Pharmacol. 43, 359–369. [69, 77] Kitiyakara, A., Angevine, D.M., 1963. A study of the pattern of post-embryonic growth of M.Gracilis in mice. Devel. Biol. 8, 322–340. [238] Kleiman, R.S., Reichardt, L.F., 1996. Testing the agrin hypothesis. Cell. 85, 461–464. [273] Ko, C.P., 1981. Electrophysiological and freeze-fracture studies of changes following denervation at frog neuromuscular junctions. J. Physiol. 321, 627–639. [275] Ko, C.P., 1984. Regeneration of the active zone at the frog neuromuscular junction. J. Cell Biol. 98(5), 1685–1695. [275]
REFERENCES
309
Ko, C.P., 1985. Formation of the active zone at developing neuromuscular junctions in larval and adult bullfrogs. J. Neurocytol. 14(3), 487–512. [275] Ko, C.P., Chen, L., 1996. Synaptic remodeling revealed by repeated in vivo observations and electron microscopy of identified frog neuromuscular junctions. J. Neurosci. 16(5), 1780–1790. [277] Koelle, G.B., 1961. A proposed dual neurohumoral role of acetylcholine: its function at the pre- and post-synaptic sites. Nature. 190, 208–211. [67, 69] Koenig, J., 1963. Innervation motrice experimentale d’une portion du muscle strie normalement depourvue de plaques motrices chez le rat. C.R. hebdomad. Seances Acad. Sci. 256, 2918. [235] Koh, D.S., Hille, B., 1997. Modulation by neurotransmitters of catecholamine secretion from sympathetic ganglion neurons detected by amperometry. Proc. Natn. Acad. Sci. U.S.A. 94, 1506–1511. [158] Koh, S.D., Campbell, J.D., Carl, A., Sanders, K.M., 1995. Nitric oxide activates multiple potassium channels in canine colonic smooth muscle. J. Physiol. 489, 735–743. [130] Kohn, A., 1903a. Das chromaffine Gewebe. Allgemeine Wien medizinisch Zeitung. XLVII, 503–515. [50] Kohn, A., 1903b. Die Paraganglien. Archiv fur microskopische Anatomie und Entwicklungsgeschichte, Bonn LXII–263– 365. [50] Kolliker, A., 1856a. Note sur l’action du curare sur le systeme nerveux. Comptes rendus Acad. Sci 43, 791. [18] Kolliker, A., 1856b, Virchow’s Arch. 10, 3. [18] Kolliker, J.A., 1896. Handbuch de Gewebelehre des Menschen; 1852; 1896. 6th Edition, Vol. 2, Nervensystem des Menschen und derThiere. W.Engelmann, Leipzig; 1896; 1852. [16, 18] Komatsu, Y., Toyama, K., Maeda, J., Sakaguchi, H., 1981. Long-term potentiation investigated in a slice preparation of striate cortex of young kittens. Neurosci Letts. 26, 269– 274. [214] Korn, H., Mallet, A., Triller, A., Faber, D.S., 1982. Transmission at a central inhibitory synapse. II. Quantal description of release, with a physical correlate for binomial n. J. Neurophysiol. 48(3), 679–707. [187, 189, 190, 195, 197] Korneliussen, H., Jansen, J.K.S., 1976. Morphological aspects of the elimination of polyneuronal innervation of skeletal muscle fibres in newborn rats. J. Neurocytol. 5, 591–604. [242] Kosterlitz, H.W., 1967. Intrinsic intestinal reflexes. Amer. J. Digest. Dis., New Series. 12, 245–254. [121, 123] Kosterlitz, H.W., Watt, A.J., 1964. Adrenergic receptors in the guinea-pig ileum. J. Physiol. 177, 11P–12P. [125] Kostyuk, P.G., Krishtal, O.A., 1977. Effects of calcium and calcium-chelating agents on the inward and outward current in the membrane of mollusc neurones. J. Physiol. 270, 569–580. [151] Kostyuk, P.G., Krishtal, O.A., Pidoplichko, V.I., 1977a. Asymmetrical displacement currents in nerve cell membrane and effect of internal fluoride. Nature. 267, 70–72. [151] Kostyuk, P.G., Krishtal, O.A., Shakhovalov, Yu. A., 1977b. Separation of calcium and sodium currents in the somatic membrane of mollusc neurones. J. Physiol. 270, 545– 568. [150, 151] Kretschmann, H.J., Kammrad, G., Kranthausen, I., Sauer, B., Wingert, F., 1986. Growth of the hippocampal formation in man. Bibloth. Anatom. 28, 27–52. [217] Krishtal, O.A., Pidoplichko, V.I., 1975. Intracellular perfusion of Helix neurones. Neirofiziologiya 7, 327–329. (Neurophysiology, Engl. Transl. 7, 258–259). [151] Krnjevic, K., 1970. Glutamate and gamma-aminobutyric acid in brain. Nature. 228, 119–124. [85] Krnjevic, K., Schwartz, S., 1967. The action of gammaaminobutyric acid on cortical neurones. Exp. Brain Res. 3, 320–336. [85] Krogsgaard-Larsen, P., Honoré, T., Hansen, J.J., Curtis, D.R., Lodge, D., 1980. New class of glutamate agonists structurally related to ibotenic acid. Nature 284, 64–66. [88] Kuffler, S.W., 1942. Electrical potential changes at an isolated nerve-muscle junction. J. Neurophysiol. 5, 18–26. [36] Kuffler, S.W., Eyzaguirre, C., 1955. Synaptic inhibition in an isolated nerve cell. J. Gen. Physiol. 39, 155–184. [115, 118] Kuffler, S.W., Edwards, C., 1958. Mechanisms of gamma aminobutyric acid (GABA) action and its relation to synaptic inhibition. J. Neurophysiol. 21, 589–610. [83, 85] Kuhne, W., 1862. über die peripherischen Endorgane der motorischen Nerven. Leipzig: Engelmann. [15, 18] Kuhne, W., 1888. On the origin of causation of vital movement. Proc. Roy. Soc. Lond. Ser. B. 44. 427. [15, 18] Kuhne, W., 1869. In: Manual of Human and Comparative Histology. Stricker, S. (ed.), The New Sydenham Society, London, pp. 1870–1873, 147–169. [11, 15] Kullberg, R.W., Lentz, T.L., Cohen, M.W., 1977. Development of the myotomal neuromuscular junction in Xenopus laevis: an electrophysiological and fine-structural study. Develop. Biol. 60, 101–129. [242] Kullmann, D.M., Martin, R.L., Redman, S.J., 1989. Reduction by general anaesthetics of group Ia excitatory postsynaptic potentials and currents in the cat spinal cord. J. Physiol. 412, 277–296. [189, 190] Kuntz, A., 1922. On the occurrence of reflex arcs in the my enteric and submucous plexuses. Anat. Rec. 24, 193–210. [125] Kuonen, D.R., Kemp, M.C., Roberts, P.J., 1988. Demonstration and biochemical characterisation of rat brain NADPH-dependent diaphorase. J. Neurochem. 50, 1017–1025. [132] Kuriyama, H., 1964. Effect of calcium and magnesium on neuromuscular transmission in the hypogastric nerve-vas deferens preparation of the guinea pig. J. Physiol. 175, 211–230. [145] Laborit, H., Huguenard, P., Alluaume, R., 1952. Un nouveau stabilisateur vegetatif (LE 4560 RP). Presse. Med. 60, 206– 208. [91] Lai, R., Watanabe, Y, Yoshida, H., 1983. Effect of isletactivating protein (IAP) on contractile responses of rat vas deferens: evidence for inhibitory participation of GTP binding regulating protein in the 2-adrenoceptormediated response. Eur. J. Pharmacol. 90, 453–456. [77] Landmesser, L., 1971. Contractile and electrical responses of vagus-innervated frog sartorius muscles. J. Physiol. 213, 707–725. [238]
310
REFERENCES
Landmesser, L., 1972. Pharmacological properties, cholinesterase activity and anatomy of nerve-muscle junctions in vagus-innervated frog sartorius. J. Physiol. 220, 243– 256. [236] Langer, S.Z., Adler, E., Enero, M.A., Stefano, F.J.E., 1971. The role of the alpha-receptors in regulating noradrenaline overflow by nerve stimulation. Proc. Int. Union Physiol. Sci. 9, 335. [72] Langer, S.Z., Adler-Graschinsky, E., Giorgi, O., 1977. Physiological significance of alpha-adrenoceptor-mediated negative feedback mechanism regulating noradrenaline release during nerve stimulation. Nature. 265, 648–650. [71] Langer, S.Z., Vogt, M., 1971. Noradrenaline release from isolated muscles of the nictitating membrane of the cat. J. Physiol 214, 159–171. [72] Langer, S.Z., 1974. Presynaptic regulation of catecholamine release. Biochem. Pharm. 23, 1793–1800. [70, 77] Langer, S.Z., 1970. The metabolism of (3H) noradrenaline released by electrical stimulation from the isolated nicitating membrane of the cat and from the vas deferens of the rat. J. Physiol. 208, 515–546. [72] Langley, J.N., 1878. On the physiology of the salivary secretion. Part II. On the mutual antagonism of atropin an dpilocarpin having especial reference to their relations in the sub-maxillary gland of the cat. J. Physiol. I, 339–369. [51] Langley, J.N., 1892. On the origin from the spinal cord of the cervical and upper thoracic sympathetic fibres, with some observations in white and grey rami communicantes. Phil. Trans. Roy. Soc. CLXXXIII, 85–124. [278] Langley, J.N., 1895. Note on regeneration of preganglionic fibres of the sympathetic. J. Physiol. XVII, 280–284. [233, 234] Langley, J.N., 1897. On the regeneration of preganglionic and of postganglionic visceral nerve fibres. J. Physiol. XXII, 215–230. [281] Langley, J.N., 1898a. On the union of cranial autonomic (visceral) fibres with the nerve cells of the superior cervical ganglion. J. Physiol. XXIII, 240–270. [107] Langley, J.N., 1898b. On inhibitory fibres in the vagus for the end of the oesophagus and the stomach. J. Physiol. XXIII, 407–414. [108, 111, 120] Langley, J.N., 1900. Notes on the regeneration of the preganglionic fibres in the sympathetic system. J. Physiol. XXV 417–426. [278] Langley, J.N., 1901. Observations on the physiological action of extracts of the supra-renal bodies. J. Physiol. 27, 237–256. [25, 26, 47, 48, 51, 53, 65, 107] Langley, J.N., 1905. On the reaction of cells and of nerve endings to certain poisons, chiefly as regards the reaction of striated muscle to nicotine and to curari. J. Physiol. XXXIII, 374–413. [50–53, 107] Langley, J.N., 1906. Croonian lecture of the Royal Society: On nerve endings and on special excitable substances in cells. Proc. Roy. Soc. Lond. B. LXXVIII, 170–194. [27, 46, 53, 107] Langley, J.N., 1907. On the contraction of muscle, chiefly in relation to the presence of ‘receptive’ substances. Part 1. J. Physiol XXXVI, 347–384. [51, 53] Langley, J.N., 1908. On the contraction of muscle, chiefly in relation to the presence of ‘receptive’ substance. Part III. The reaction of frogs muscle to nicotine after denervation. J. Physiol. XXXVII, 285–300. [54] Langley, J.N., 1921. The Autonomic Nervous System, Part 1. W.Heffer and Sons Ltd., Cambridge. [105, 107] Langley, J.N., Anderson, H.K., 1895. On the innervation of the pelvic and adjoining viscera. Part 1. The lower portion of the intestine. J. Physiol. XVIII, 67–105. [113] Langley, J.N., Sherrington, C.S., 1884. Secondary degeneration of nerve tracts following removal of the cortex of the cerebrum in the dog. J. Physiol. 5, 49–65. [25] Lapicque, L., 1926. L’excitabilite en fonction du temps. Les Presses Universitaires de France. [59] Lapicque, L., 1934. Neuro-muscular isochronism and chronological theory of curarization. J. Physiol. 81, 113–145. [59, 60] Lapicque, L., Lapicque Mme., 1906. Comparaison de l’excitabilite du muscle a celle de son nerf moteur. Comptes Rendus hebdomadaires des Seances et Memoires de la Societe de Biologie, Paris 60, 898–901. [59] Larkman, A.U., Jack, J.J., Stratford, K.J., 1997. Assessment of the reliability of amplitude histograms from excitatory synapses in rat hippocampal CA1 in vitro. J. Physiol. 505(Pt 2), 443–456. [389] Laskowski, M.B., Sanes, J.R., 1987. Topographic mapping of motor pools onto skeletal muscles. J. Neurosci. 7(1), 252– 260. [264] Laskowski, M.B., Sanes, J.R., 1988. Topographically selective reinnervation of adult mammalian skeletal muscles. J. Neurosci . 8(8), 3094–3099. [264] Lea, T.J., Usherwood, P.N.R., 1973. The site of action of ibotenic acid and the identification of two populations of glutamate receptors on insect muscle fibres. Comp. Gen. Pharmac. 4, 333–350. [86] Leach, J.P., Brodie, M.J., 1998. Tiagabine. Lancet 351, 203– 207. [87] Lee, C.Y., 1972. Chemistry and pharmacology of polypeptide toxins in snake venoms. Annu. Rev. Pharmacol. 12, 265– 286. [63] Lee, C.Y., Chang, C.C., 1966. Modes of actions of purified toxins from elapid venoms on neuromuscular transmission. Memorias do Instituto Butantan. 33, 555–572. [63] Lee, H.K., Shuttleworth, C.W., Sanders, K.M., 1995. Tachykinins activate nonselective cation currents in canine colonic myocytes. Amer. J. Physiol. 269, C1394–1401. [132] Lee, K.S., Akaike N., Brown, A.M., 1977. Trypsin inhibits the action of tetrodotoxin on neurones. Nature. 265, 751– 753. [151] Lefebvre, R.A., Hasrat, J., Gobert, A., 1992. Influence of NG-nitro-L-arginine methyl ester on vagally induced gastric relaxation in the anaesthetized rat. Brit. J. Pharmacol. 105, 315–320. [128] Lehre, K.P., Levy, K.M., Ottersen, O.P., Storm-Mathisen, J., Danbolt, N.C., 1995. Differential expression of two glial glutamate transporters in the rat brain: quantitative and immunocytochemical observations. J. Neurosci. 15, 1835– 1853. [128] Lei, S.Z., Pan, Z.H., Aggarwal, S.K., Chen, H.S.V., Hartman, J., Sucher, N.J., Lipton, S.A., 1992. Effect of nitric oxide production on the redox modulatory site of the NMDA receptor-channel complex. Neuron 8, 1087–1099. [131]
REFERENCES
311
Letinsky, M.S., Morrison-Graham, K., 1980. Structure of developing frog neuromuscular junctions. J. Neurocytol. 9, 321–342. [242] Leveque, C., Hoshino, T., David, P., Shoji-Kasai, Y., Leys, K., Omori, A., Lang, B., el Far, O., Sato, K., Martin-Moutot, N., NewsomDavis, J., Takahashi, M., Seagar, M.J., 1992. The synaptic vesicle protein synaptotagmin associates with calcium channels and is a putative Lambert-Eaton myasthenic syndrome antigen. Proc. Natn. Acad. Sci. U.S.A. 89, 3625–3629. [163] Levey, M.S., Jacob, M.H., 1996. Changes in the regulatory effects of cell-cell interactions on neuronal AChR subunit transcript levels after synapse formation. J. Neurosci. 16(21), 6878–6885. [279] Levey, M.S., Brumwell, C.L., Dryer, S.E., Jacob, M.H., 1995. Innervation and target tissue interactions differentially regulate acetylcholine receptor subunit mRNA levels in developing neurons in situ. Neuron. 14(1), 153–162. [279] Levitan, E.S., Kramer, R.H., 1990. Neuropeptide modulation of single calcium and potassium channels detected with a new patch clamp configuration. Nature. 348, 545–547. [131] Levy., L.M., Lehre, K.P., Rolstadt, B., Danbolt, N.C., 1993. A monoclonal antibody raised against a [Na+, K+]-coupled L-glutamate transporter purified from rat brain confirms glial cells localization. FEBS Lett. 317, 79–84. [87] Lewandowsky, M., 1899. Ueber die wirkung des Nebennieren-extractes auf die glatten Muskeln, in Besonderen des Auges. Arch. Anat. Physiol. 360–366. [48] Leysen, J.E., Janssen, P.M., Schotte, A., Luyten, W.H., Megens, A.A., 1993. Interaction of antipsychotic drugs with neurotransmitter receptor sites in vitro and in vivo in relation to pharmacological and clinical effects: role of 5HT2 receptors. Psychopharmacol. 112 (Suppl 1), S40–S54. [102] Li, C., Davletov, B., Sudhof, T.C., 1995a. Distinct Ca2+ and Sr2+ binding properties of synaptotagmins. Definition of candidate Ca2+ sensors for the fast and slow components of neurotransmitter release. J. Biol. Chem. 270, 24898–24902. [162] Li, C., Ullrich, B., Zhang, J.Z., Anderson, R.G., Brose, N., Sudhof, T.C., 1995b. Ca(2+)-dependent and -independent activities of neural and non-neural synaptotagmins. Nature. 375, 594–599. [162] Li, C.G., Rand, M.J., 1989a. Evidence for a role of nitric oxide in the neurotransmitter system mediating relaxation of the rat anococcygeus muscle. Clin. Exper. Pharmacol. Physiol. 16, 933–938. [131] Li, C.G., Rand, M.J., 1989b. Prejunctional inhibition of nonadrenergic non-cholinergic transmission in the rat anococcygeus muscle. Eur. J. Pharmacol. 168, 107–110. [131] Li, C.G., Rand, M.J., 1990. Nitric oxide and vasoactive intestinal polypeptide mediate non-adrenergic, non-cholinergic inhibitory transmission to smooth muscle of the rat gastric fundus. Eur. J. Pharmacol. 191, 303–309. [128, 129] Libet, B., Tosaka, T., 1970. Dopamine as a synaptic transmitter and modulator in sympathetic ganglia: a different mode of synaptic action. Proc. Nat. Acad. Sci. USA. 67, 667–673. [95] Lichtman, J.W., 1980. On the predominantly single innervation of submandibular ganglion cells in the rat. J. Physiol. 302, 121–130. [278] Lichtman, J.W., Purves, D., 1980. The elimination of redundant preganglionic innervation to hamster sympathetic ganglion cells in early post-natal life. J. Physiol. 301, 213–228. [278] Lichtman, J.W., Purves, D., Yip, J.W., 1979. On the purpose of selective innervation of guinea-pig superior cervical ganglion cells. J. Physiol 292, 69–84. [278] Liddell, E.G.T., 1960. The Discovery of Reflexes. Clarendon Press, Oxford. [22] Liddell, E.G.T., Sherrington, C.S., 1925. Further observations on myotatic reflexes. Proc. Roy. Soc. Lond. B. 97, 267–283. [27] Liddell, E.G.T., Sherrington, C.S., 1924. Reflexes in response to stretch (myotatic reflexes). Proc. Roy. Soc. Lond. B.96, 212–242. [27] Liestol, K., Maehlen, J., Nja, A., 1987. Two types of synaptic selectivity and their interrelation during sprouting in the guinea-pig superior cervical ganglion . J. Physiol. 384, 233– 245. [278] Lieth, E., Fallon, J.R., 1993. Muscle agrin: neural regulation and localization at nerve-induced acetylcholine receptor clusters. J. Neurosci. 13(6), 2509–2514. [273] Lin, Y.Q., Brain, K.L., Nichol, K.A., Morgan, J.J., Bennett, M.R., 1996. Vesicle-associated proteins and calcium in nerve terminals of chick ciliary ganglia during development of facilitation. J. Physiol. 497, 639–656. [162] Lindgren, B.W., 1993. Statistical Theory. 4th Ed. Chapman and Hall. New York. [175] Ling, G., Gerard, R.W., 1949. The normal membrane potential of frog sartorius fibres. J. Cell. Comp. Physiol. 34, 383– 396. [36] Littleton, J.T., Bellen, H.J., 1995. Synaptotagmin controls and modulates synaptic-vesicle fusion in a Ca(2+)-dependent manner. Trends NeuroSci. 18, 177–183. [160, 161] Littleton, J.T., Stern, M., Perin, M., Bellen, H.J., 1994. Calcium dependence of neurotransmitter release and rate of spontaneous vesicle fusions are altered in Drosophila synaptotagmin mutants. Proc. Natn. Acad. Sci. U.S.A. 91, 10888– 10892. [160] Littleton, J.T., Stern, M., Schulze, K., Perin, M., Bellen, H.J., 1993. Mutational analysis of Drosophila synaptotagmin demonstrates its essential role in Ca(2+)-activated neurotransmitter release. Cell. 74, 1125–1134. [160] Liu, L.X., Kapatos, G., Chiodo, L.A., 1992. DA autoreceptor modulation of different calcium currents. Soc. Neurosci. Abstr. 18, 1516. [103] Lledo, P.M., Legendre, P., Israel, J.M., Vincent, J.D., 1990. Dopamine inhibits two characterized voltage-dependent calcium currents in identified rat lactotroph cells. Endocrin. 127, 990–1001. [103] Llewellyn-Smith, I.J., Furness, J.B., Gibbins, I.L., Costa, M., 1988. Quantitative ultrastructural analysis of enkephalin-, substance P−, and VIPimmunoreactive nerve fibers in the circular muscle of the guinea pig small intestine. J. Comp. Neurol. 272, 139–148. [131] Llewellyn-Smith, I.J., Song, Z.M., Costa, M., Bredt, D.S., Snyder, S.H., 1992. Ultrastructural localization of nitric oxide synthase immunoreactivity in guinea-pig enteric neurons. Brain Res. 577, 337–342. [132] Llinas, R., Blinks, J.R., Nicholson, C., 1972. Calcium transient in presynaptic terminal of squid giant synapse: detection with aequorin. Science 176, 1127–1129. [147, 148]
312
REFERENCES
Llinas, R., Steinberg, I.Z., Walton, K., 1981. Relationship between presynaptic calcium current and postsynaptic potential in squid giant synapse. Biophys. J. 33, 323–351. [158, 163] Llinas, R., Sugimori, M., Silver, R.B., 1992a. Presynaptic calcium concentration microdomains and transmitter release. J. Physiol Paris 86, 135–138. [163] Llinas, R., Sugimori, M., Hillman, D.E., Cherksey, B., 1992b. Distribution and functional significance of the P-type, voltage-dependent Ca2 + channels in the mammalian central nervous system. Trends NeuroSci. 15, 351–355. [155, 161] Llinas, R., Sugimori, M., Silver, R.B., 1992c. Microdomains of high calcium concentration in a presynaptic terminal. Science. 256, 677–679. [163] Llinas, R., Sugimori, M., Lin, J.W., Cherksey, B., 1989. Blocking and isolation of a calcium channel from neurons in mammals and cephalopods utilizing a toxin fraction (FTX) from funnel-web spider poison. Proc. Natn. Acad. Sci. U.S.A. 86, 1689–1693. [153, 155, 157, 163] Llinas, R.R., Sugimori, M., Silver, R.B., 1994. Localization of calcium concentration microdomains at the active zone in the squid giant synapse. Adv. Second Messenger & Phosphoprotein Res. 29, 133–137. [160] Lloyd, D.P.C., 1946. Facilitation and inhibition of spinal motoneurones. J. Neurophysiol. 9, 421–438. [26, 36] Locke, F.S., 1894. Notiz uber den Einfluss physiologischer Kochsalzlosung auf die elektrische Erregbarkeit von Muskel und Nerv. Zbl Physiol. 8, 166–167. [40, 135, 136] Lodge, D., 1997. Subtypes of glutamate receptors. Historical perspectives on their pharmacological differentiation. In: The Ionotropic Glutamate Receptors. Monaghan, D.T., Wenthold, R.J. (eds.), Humana Press Inc., Totowa, NJ, USA, pp. 1–38. [90] Lodge, D., Headley, P.M., Curtis, D.R, 1978. Selective antagonism by D-alpha-aminodipate of amino acid and synaptic excitation of cat spinal neurones. Brain Res. 152, 603–608. [88] Loeb, J.A., Fischbach, G.D., 1997. Neurotrophic factors increase neuregulin expression in embryonic ventral spinal cord neurons. J. Neurosci. 17(4), 1416–1424. [279] Loewi, O., 1935. The Ferrier lecture. Problems connected with the principle of humoral transmission of nervous impulses. Proc. Roy. Soc. Lond., B. 119, 299–316. [27] Loewi, O., 1921. Uber humorale Ubertragbarkeit der Herznervenwirkung. I.Mitteilung. Pflugers . Archiv. Fur. Die. Gesmnte. Physiologie. 189:239–242. [107] Loewi, O., Navratil, E., (1926a) Uber humorale Ubertragbarkeit der Herznervenwirkung. X. Uber das Schicksal des Vagusstoffs. Pflugers Arch. Ges. Physiol. Mensch. Tiere. 214, 678–688. [56, 57] Loewi, O., Navratil, E., 1926b. Uber humorale Ubertragbarkeit der Herznervenwirkung. XI. Uber den Mechanismus der Vaguswirkung von Physostigmin und Ergotamin. Pflugers Arch. ges. Physiol. Mensch. Tiere. 214, 689–696. [56, 57, 58] Logan, W.J., Snyder, S.H., 1971. Unique high affinity systems for glycine, glutamic and aspartic acids in central nervous tissue. Nature. 234, 297–299. [85] Lomo, T., 1966. Frequency potentiation of excitatory synaptic activity in the dentate area of the hippocampal formation. Acta. Physiol. Scand. 68, Suppl 277, 128. [211] Loo, D.D., Kaunitz, J.D., 1989. Ca2+ and cAMP activate K+ channels in the basolateral membrane of crypt cells isolated from rabbit distal colon. J. Mem. Biol 110, 19–28. [131] Lorente de No, R., 1949. On the effect of certain quaternary ammonium ions upon frog nerve. J. Cell. Comp. Physiol. 33, Suppl, 1–231. [141] Lovinger, D.M., Wong, K.L., Murakami, K., Routtenberg, A., 1987. Protein kinase C inhibitors eliminate hippocampal long-term potentiation. Brain Res. 436, 177–183. [225] Lucas, D.R., Newhouse, J.P., 1957. Toxic effects of Na-L-glutamate on the inner layers of the retina. A.M.A. Arch. Ophthalmol. 58, 193–204. [85] Lucas, K., 1917. The conduction of the nervous impulse. Monographs on Physiology. London: Longmans. [30] Lucas, K., Mines, G.R., 1907. Temperature and excitability. J. Physiol. XXXVI, 334–346. [59] Luebke, J.I., Dunlap, K., Turner, T.J., 1993. Multiple calcium channel types control glutamatergic synaptic transmission in the hippocampus . Neuron. 11, 895–902. [155] Luff, A.R., Proske, U., 1979. Properties of motor units of the frog iliofibularis muscle. Amer. J. Physiol 236, C35–42. [252] Lullman-Rauch, R., 1971. The regeneration of neuromuscular junctions during spontaneous reinnervation of the rat diaphragm. Z. Zellforsch. Mikrosk. Anat. 121, 593–608. [235] Lux, H.D., Nagy, K., 1981. Single channel Ca2+ currents in Helix pomatia neurons. Pflugers Arch.-Eur. J. Physiol. 391, 252–254. [151, 152] Lynch, G., Baudry, M., 1984. The biochemistry of memory: anew and specific hypothesis. Science. 224, 1057–1063. [227] Lynch, G., Halpain, S., Baudry, M., 1982. Effects of high-frequency synaptic stimulation on glumate receptor binding studied with a modified in vitro hippocampal slice preparation. Brain Res. 244, 101–111. [220, 225] Lynch, G., Larson, J., Kelso, S., Barrionuevo, G., Schottler, F., 1983. Intracellular injections of EGTA block induction of hippocampal longterm potentiation. Nature. 305, 719– 721. [221] Lynch, M.A., Bliss, T.V., 1986a. Noradrenaline modulates the release of [14C] glutamate from dentate but not from CA1/CA3 slices of rat hippocampus. Neuropharmacol. 25, 493–498. [223] Lynch, M.A., Bliss, T.V., 1986b. Long-term potentiation of synaptic transmission in the hippocampus of the rat; effect of calmodulin and oleoyl-acetyl-glycerol on release of [3H]glutamate. Neurosci. Letts. 65, 171–176. [228]
REFERENCES
313
Lynch, M.A., Errington, M.L., Bliss, T.V., 1985. Long-term potentiation of synaptic transmission in the dentate gyrus: increased release of [14C]glutamate without increase in receptor binding. Neurosci. Letts. 62, 123–129. [223] Lynch, K., Ko, C.P., 1983. Presynaptic active zones at neuromuscular junctions of larval frogs. Develop. Biol. 97(1), 10–18. [275] Lyster, D.J., Bywater, R.A., Taylor, G.S., Watson, M.J., 1992. Effects of a nitric oxide synthase inhibitor on noncholinergic junction potentials in the circular muscle of the guinea pig ileum. J. Auton. Nerv. Sys. 41, 187–196. [127] Ma, E., Morgan, R., Godfrey, E.W., 1994. Distribution of agrin mRNAs in the chick embryo nervous system. J. Neurosci. 14(5 Pt 2), 2943–2952. [279] Ma, E., Morgan, R., Godfrey, E.W., 1995. Agrin mRNA variants are differentially regulated in developing chick embryo spinal cord and sensory ganglia J. Neurobiol. 26(4), 585–597. [279] Maas, A.J., Den Hertog, A., 1979. The effect of apamin on the smooth muscle cells of the guinea pig taenia coli. Eur. J. Pharmacol. 58, 151–156. [127, 130] Maas, A.J., Den Hertog, A., Ras, R., Van den Akker, J., 1980. The action of apamin on guinea-pig taenia caeci. Eur. J. Pharmacol. 67, 265–274. [127, 130, 131] MacDonald, J.F., Porietis, A.V, Wojtowicz, J.M., 1982. L-Aspartate acid induces a region of negative slope conductance in the currentvoltage relationship of cultured spinal cord neurones. Brain Res. 237, 248–253. [221] Mackenzie, I., Burnstock, G., 1980. Evidence against vasoactive intestinal polypeptide being the non-adrenergic, noncholinergic inhibitory transmitter released from nerves supplying the smooth muscle of the guinea-pig taenia coli. Eur. J. Pharmacol. 67, 255–264. [129] Maehlen, J., Nja, A., 1984. Rearrangement of synapses on guinea-pig sympathetic ganglion cells after partial interruption of the preganglionic nerve. J. Physiol. 348, 43–56. [278] Maggi, C.A., Manzini, S., Meli, A., 1984. Differential effects of vasodilators on the mobilization of calcium pools during contractions of rabbit ear artery induced by noradrenaline and high potassium. J. Pharm. & Pharmacol. 36, 27–30. [132] Magill-Solc, C., McMahan, U.J., 1988. Synthesis and transport of agrin-like molecules in motor neurons. J. Exper. Biol. 153, 1–10. [271] Magleby, K.L., Miller, D.C., 1981. Is the quantum of transmitter release composed of subunits? A critical analysis in the mouse and frog. J. Physiol. 311, 267–287. [190, 193] Maines, M.D., 1988. Heme oxygenase: function, multiplicity, regulatory mechanisms, and clinical applications. FASEB J. 2, 2557–2568. [130] Makhlouf, G.M., Grider, J.R., 1993. Nonadrenergic noncholinergic inhibitory transmitters of the gut. News Physiol. Sci. 8, 195–199. [127–129] Malenka, R.C., Kauer, J.A., Perkel, D.J., Mauk, M.D., Kelly, P.T., Nicoll, R.A., Waxman, M.N., 1989. An essential role for postsynaptic calmodulin and protein kinase activity in long-term potentiation. Nature. 340, 554–557. [229] Malenka R.C., Kauer, J.A., Zucker, R.S., Nicoll, R.A., 1988. Postsynaptic calcium is sufficient for potentiation of hippocampal synaptic transmission. Science. 242, 81–84. [220, 221] Malenka, R.C., Madison, D.V., Nicoll, R.A., 1986. Potentiation of synaptic transmission in the hippocampus by phorbol esters. Nature. 321, 175–177. [228] Malik, R., Bennett, M.R., 1987. Loss of polyneuronal innervation and establishment of a topographical map in the glutaeus muscle of Bufo marinus during generation of secondary muscle cells. Dev. Brain Res. 431, 173–189. [242, 246, 252] Malinow, R., Madison, D.V., Tsien, R.W., 1988. Persistent protein kinase activity underlying long-term potentiation. Nature. 335, 820–824. [225] Malinow, R., Miller, J.P., 1986. Postsynaptic hyperpolarization during conditioning reversibly blocks induction of long-term potentiation. Nature. 320, 529–530. [221] Malinow, R., Schulman, H., Tsien, R.W., 1989. Inhibition of postsynaptic PKC or CaMKII blocks induction but not expression of LTP. Science. 245, 862–866. [228, 229] Malinow, R., Tsien, R.W., 1990. Identification and localizing protein kinases necessary for LTP. Ad. in Exp. Med. A & Biol. 268, 301–305. [225] Mall, F., 1896. A study of the intestinal contraction. Johns Hopkins Hosp. Rep. 1, 37–75. [120] Mann, P.J.G., Tennenbaum, M., Quastel, J.H., 1939. Acetylcholine metabolism in the central nervous system. The effects of potassium and other cations on acetylcholine liberation. Biochem. J. 33, 822–835. [135, 137] Manske, R.H.F., 1933. Alkaloids of fumaraceous plants III. A new alkaloid, bicuculline, and its constitution. Can. J. Res. 8, 142–146. [85] Marshall, I., Nasmyth, P.A., Shepperson, N.B., 1978. Presynaptic alpha-adrenoceptors and [3H]-noradrenaline overflow from the mouse vas deferens. Brit. J. Pharmacol. 62, 382P–383P. [75] Martin, P.T., Ettinger, AJ., Sanes, J.R., 1995. A synaptic localization domain in the synaptic cleft protein laminin beta 2 (s-laminin). Science. 269(5222), 362–363. [275] Martin, P.T., Sanes, J.R., 1995. Role for a synapse-specific carbohydrate in agrin-induced clustering of acetylcholine receptors. Neuron. 14 (4), 743–754. [274] Martin-Moutot, N., Charvin, N., Leveque, C., Sato, K., Nishiki, T., Kozaki, S., Takahashi, M., Seagar, M., 1996. Interaction of SNARE complexes with P/Q-type calcium channels in rat cerebellar synaptosomes. J. Biol. Chem. 271, 6567–6570. [163] Martinou, J.C., Falls, D.L., Fischbach, G.D, Merlie, J.P., 1991. Acetylcholine receptor-inducing activity stimulates expression of the epsilon-subunit gene of the muscle acetylcholine receptor. Proc. Nat. Acad. Sci. USA. 88(17), 7669–7673. [274] Martinson, J., 1965. Vagal relaxation of the stomach. Acta Physiol. Scand. 64, 453–462. [114, 120, 238] Martinson, J., Muren, A., 1963. Excitatory and inhibitory effects of vagus stimulation on gastric motility in the cat. Acta Physiol. Scand. 57, 309–316. [113–115, 120]
314
REFERENCES
Matsumoto, T., Nakane, M., Pollock, J.S., Kuk, I.E., Forstermann, U., 1993. A correlation between soluble brain nitric oxide synthase and NADPH-diaphorase activity is only seen after exposure of the tissue to fixative. Neurosci. Letts. 155, 61–64. [132] Matteson, D.R., Kriebel, M.E., Llados, F., 1981, A statistical model indicates that miniature end-plate potentials and unitary evoked end-plate potentials are composed of subunits. J. Theor. Biol. 90(3), 337–363. [184, 186, 189, 194] Matteucci, C., 1838. Sur le courant électrique ou propre de la grénouille. Bibl. Univ. Geneve. 15, 157–168. [12] Matteucci, C., 1842a. Deuxième mémoire sur le courant électrique propre de la grénouille et sur celur des animaux a sang chaud. Ann. Chim. et Phys. 6(Ser.3), 301–339. [12] Matteucci, C., 1842b. Sur une phénomène physiologique produite par les muscles en contraction. Compt. Rend. Acad. Sc., Paris. 4, 797. [12] Matteucci, C., 1844. Traité des phénomènes électro-physiologiques des animaux suivi d’etudes anatomiques sur le systneme nerveux et sur l’organe électrique de la torpille par Paul Savi. Paris: Fortin, Masson et C.i.e. [12] Matteucci, C., 1848. Lecons sur les Phénomènes Physiques des Corps Vivants, translated by Clet. Paris: Masson, 1847; (English translation by Jonathan Pereira. Lectures on the physical phenomena of living beings. Philadelphia: Lea and Blanchard). [12] Matteucci, C., Humboldt, F.H.A., 1843. Sur le courant électrique des muscles des animaux vivants ou réclimment tués. Compt. rend. Acad. Sc. Paris 16, 197 [12] Matthes, K., 1930. The action of blood on acetylcholine. J. Physiol. LXX: 338–348. [57] Matthew, W.D., Tsavaler, L., Reichardt, L.F., 1981. Identification of a synaptic vesicle-specific membrane protein with a wide distribution in neuronal and neurosecretory tissue. J. Cell Biol. 91, 257–269. [158] May, W.P., 1904. The innervation of the sphincters and musculature of the stomach. J. Physiol. XXXI, 260–271. [111, 112] Mayer, M.L., Westbrook, G.L., Guthrie, P.B., 1984. Voltage-dependent block by Mg2+ of NMDA responses in spinal cord neurones. Nature. 309, 261–263. [221] McArdle, J.J., 1975. Complex endplate potentials at the regenerating neuromuscular junction of the rat. Exp. Neurol. 49, 629–638. [233, 239, 242, 254, 255] McAvoy, M., Smith, M.A., Fujii, J.T., 1996. Agrin mRNA expression in the developing chick Edinger-Westphal nucleus. Vis. Neurosci. 13 (2), 293–301. [279] McConalogue, K., Furness, J.B., Vremec, M.A., Hols, J.J., Tornoe, K., Marley, P.D., 1995. Histochemical, pharmacological, biochemical and chromatographic evidence that pituitary adenylyl cyclase activating peptide is involved in inhibitory neurotransmission in the taenia of the guinea-pig caecum. J. Auton. Nerv. Sys. 50, 311–322. [130] McCrea, E.D’A., McSwiney, B.A., Stopford, J.S.B., 1925. The effect on the stomach of stimulation of the peripheral end of the vagus nerve. Quart. J. Exp. Physiol. 15, 201– 233. [113] McLennan, H., Liu, J., 1982. The action of six antagonists of the excitatory amino acids on neurones of the rat spinal cord. Exp. Brain. Res. 45, 151–156. [218] McLennan, H., 1957. A comparison of some physiological properties of an inhibitory factor from brain (Factor I) and of gammaaminobutyric acid and related components. J. Physiol. 139, 79–86. [85] McLennan, H., 1970. Inhibition of long duration in the cerebral cortex. A quantitative difference between excitatory amino acids. Exp. Brain Res. 10, 417–426. [85] McLennan, H., Lodge, D., 1979. The antagonism of amino acid-induced excitation of spinal neurones in the cat. Brain Res. 169, 83–90. [88] McLennan, I.S., 1983. Differentiation of muscle fiber types in the chicken hindlimb. Develop Biol. 97, 222–228. [260] McMahan, U.J., 1990. The agrin hypothesis. Cold Spring Harbor Symp Quant. Biol 55, 407–418. [242, 273] McMahan, U.J., Horton, S.E., Werle, M.J., Honig, L.S., Kroger, S., Escher, G., Ruegg, M.A., 1992. Agrin isoforms and their role in synaptogenesis. Current Opinion in Cell Biology. 4(5), 869–874. [273] McNaughton, B.L., Barnes, C.A., Rao, G., Baldwin, J., Rasmussen, M., 1986. Long-term enhancement of hippocampal synaptic transmission and the acquisition of spatial information. J. Neurosci. 6, 563–571. [230] McNaughton, B.L., Douglas, R.M., Goddard, G.V., 1978. Synaptic enhancement in fascia dentata: cooperativity among coactive afferents. Brain Res. 157, 277–293. [217, 218, 221] McSwiney, B.A., 1931. Innervation of the stomach. Physiol. Rev. 11, 478–514. [111] McSwiney, B.A., Brown, G.L., 1926. Reversal of the action of adrenaline. J. Physiol. LXII, 52–64. [110] McSwiney, B.A., Robson, J.M., 1929. The response of smooth muscle to stimulation of the vagus nerve. J. Physiol. LXVIII, 124–131. [112, 113] Meador-Woodruff, J.H., Grandy, D.K., Van Tol, H.H., Damask, S.P., Little, K.Y., Civelk, O., Watson, S.J., 1994. Dopamine receptor gene expression in the human medial temporal lobe. Neuropsychopharmacol. 10, 239–248. [101] Meaney, J.A, Balcar, V.J., Rothstein, J.D., Jeffrey, P.J., 1998. Glutamate transport in cultures from developing avian cerebellum: Presence of GLT-1 immunoreactivity in Purkinje neurons. J.Neurosci. Res. 54, 595–603. [87] Meltzer, H.Y., Nash, J.F., 1991. Effects of antipsychotic drugs on serotonin receptors. Pharmacol. Rev. 43, 587–604. [102] Mennerick, S., Dhond, R.P., Benz, A., Xu, W., Rothstein, J.D., Danbolt, N.C., Isenberg, K.E., Zorumski, C.F., 1998. Neuronal expression of the glutmate transporter GLT-1 in hippocampal microcultures. J. Neurosci. 18, 4490–4499. [87] Meriney, S.D., Wolowske, B., Ezzati, E., Grinnell, A.D., 1996. Low calcium-induced disruption of active zone structure and function at the frog neuromuscular junction. Synapse. 24(1), 1–11. [275, 276] Meyer, D., Birchmeier, C., 1994. Distinct isoforms of neuregulin are expressed in mesenchymal and neuronal cells during mouse development. Proc. Nat. Acad. Sci. USA. 91(3), 1064–1068. [279]
REFERENCES
315
Mikami, A., Imoto, K., Tanabe, T., Niidome, T., Mori, Y., Takeshima, H., Narumiya, S., Numa, S., 1989. Primary structure and functional expression of the cardiac dihydropyridine-sensitive calcium channel. Nature. 340, 230–233. [155] Mikoshiba, K., Fukuda, M., Moreira, J.E., Lewis P.M., Sugimori, M., Niinobe, M., Llinas, R., 1995. Role of the C2A domain of synaptotagmin in transmitter release as determined by specific antibody injection into the squid giant synapse preterminal. Proc. Natn. Acad. Sci. U.S.A. 92, 10703–10707. [160] Miledi, R., 1960. Properties of regenerating neuromuscular synapses in the frog. J. Physiol. 154, 190. [236] Miledi, R., 1963. Formation of extra nerve-muscle junctions in innervated muscle. Nature. 199, 1191–1192. [238] Miledi, R., 1973. Transmitter release induced by injection of calcium ions into nerve terminals. Proc. R. Soc. Lond.— Series B: Biol. Sci. 183, 421–425. [147, 148, 151] Milner, B., 1972. Disorders of learning and memory after temporal lobe lesions in man. Clin. Neursurg. 19, 421–446. [212] Mintz, I.M., Adams, M.E., Bean, B.P., 1992a. P-type calcium channels in rat central and peripheral neurones. Neuron. 9, 85–95. [155] Mintz, I.M., Venema, V.J., Swiderek, K.M., Lee, T.D., Bean, B.P., Adams, M.E., 1992b. P-type calcium channels blocked by the spider toxin omega-Aga-IVA. Nature. 355, 827–829. [155] Mitchell, J.F., Srinivasan, V., 1969. The release of [3H] gamma-aminobutyric acid from the surface of the brain by electrical stimuli which produce synaptic inhibition. J. Physiol. 204, 23P–24P. [85] Miyata, Y, Yoshioka, K., 1980. Selective elimination of motor nerve terminals in the rat soleus muscle during development. J. Physiol 309, 631–646. [242] Mizuta, M., Inagaki, N., Nemoto, Y, Matsukura, S., Takahashi, M., Seino, S., 1994. Synaptotagmin III is a novel isoform of rat synaptotagmin expressed in endocrine and neuronal cells. J. Biol Chem. 269, 11675–11678. [162] Mochida, S., Saisu, H, Kobayashi, H., Abe, T., 1995. Impairment of syntaxin by botulinum neurotoxin C1 or antibodies inhibits acetylcholine release but not Ca2+ channel activity. Neurosci. 65, 905–915. [160] Monaghan, D.T., Holets, V.R., Toy, D.W., Cotman, C.W., 1983. Anatomical distributions of four pharmacologically distinct 3H-Lglutamate binding sites. Nature. 306, 176–179. [230] Mondadori, C., Weiskrantz, L., Buerki, H., Petschke, F., Fagg, G.E., 1989. NMDA receptor antagonists can enhance or impair learning performance in animals. Exp. Brain. Res. 75, 449–456. [230] Monnier, A.M., Bacq, Z.M., 1935. Recherches sur la physiologie et la pharmacologie du systeme nerveux autonome. XVI Dualite du mecanisme de la transmission neuromusculaire de l’excitation chez le muscle lisse. Arch. Int. Physiol 40, 485–510. [32] Morat, J.P., 1893. Sur quelques particularites de l’innervation motrice de l’estomac et de l’intestin. Arch. Physiol norm, et path. (Paris) V, 142–153. [113] Mori, Y., Friedrich, T., Kim, M.S., Mikami, A., Nakai, J., Ruth, P., Bosse, E., Hofmann, F., Flockerzi, V., Furuichi, T., Mikoshiba, K., Imoto, K., Tanabe, T., Numa, S., 1991. Primary structure and functional expression from complementary DNA of a brain calcium channel. Nature. 350, 398–402. [155] Morris, R.G., Anderson, E., Lynch, G.S., Baudry, M., 1986. Selective impairment of learning and blockade of long-term potentiation by an N-methyl-D-aspartate receptor antagonist, AP5. Nature. 319, 774–776. [230, 238] Morrison-Graham, K., 1983. An anatomical and electrophysiological study of synapse elimination at the developing frog neuromuscular junction. Develop. Biol. 90, 298–311. [242, 254] Moscoso, L.M., Chu, G.C., Gautam, M., Noakes, P.G., Merlie, J.P., Sanes J.R., 1995a. Synapse-associated expression of an acetylcholine receptor-inducing protein, ARIA/heregulin, and its putative receptors, ErbB2 and ErbB3, in developing mammalian muscle. Dev. Biol. 172(1), 158–169. [274] Moscoso, L.M., Merlie, J.P., Sanes, J.R., 1995b. N-CAM, 43K-rapsyn, and S-laminin mRNAs are concentrated at synaptic sites in muscle fibers. Mol. & Cell Neurosci. 6(1), 80–89. [275] Mule, F., Gerarci, A., Serio, R., Posterino, A., 1992. On the peptidergic hypothesis for non-adrenergic non-cholinergic innervation in the rat duodenum. J. Autonom. Pharmacol 12, 81–83. [129] Muller, J.M., Schlitter, E., Bein, H.J., 1952. Reserpen, der sedative Wirkstoff aus Rauwolfia serpentina Benth. Experientia 8, 388–391. [91] Müller, J., 1840. Handbuch der Physiologie des Menschen fur Vorlesungen, 4th Ed. 2 vols. I:583. J.Holscher: Koblenz. [14] Müller, J.M., 1851. Manuel de Physiologie. Paris: Ballière. [14] Mulsch, A., Mordvintcev, P., Vanin, A.F., Busse, R., 1991. The potent vasodilating and guanylyl cyclase activating dinitrosyl-iron(II) complex is stored in a protein-bound form in vascular tissue and is released by thiols. FEBS Letts. 294, 252–256. [132] Murray, A.M., Hyde, T.M., Knable, M.B., Herman, M.M., Bigelow, L.B., Carter, J.M., Weinberger, D.R., Kleinman, J.E., 1995. Distribution of putative D4 dopamine receptors in postmortem striatum from patients with schizophrenia. J. Neurosci. 15, 2186–2191. [101] Murray, J., Du, C., Ledlow, A., Bates, J.N., Conklin, J.L., 1991. Nitric oxide: mediator of nonadrenergic noncholinergic responses of opossum esophageal muscle. Amer. J. Physiol 261, G401–406. [132] Murthy, K.S., Makhlouf, G.M., 1994. Vasoactive intestinal peptide/pituitary adenylate cyclase-activating peptide-de-pendent activation of membrane-bound NO synthase in smooth muscle mediated by pertussis toxin-sensitive Gil-2. J. Biol Chem. 269, 15977–15980. [129] Murthy, K.S., Zhang, K.M., Jin, J.G., Grider, J.R., Makhlouf, G.M., 1993. VIP-mediated G protein-coupled Ca2+ influx activates a constitutive NOS in dispersed gastric muscle cells. Amer. J. Physiol 265, G660–G671. [129] Musgrave, I., Marley, P., Majewski, H., 1987. Pertussis toxin does not attenuate alpha 2-adrenoceptor mediated inhibition of noradrenaline release in mouse atria. Naunyn-Schmiedebergs Arch. Pharmacol. 336, 280–286. [77] Naito, S., Ueda, T., 1985. Characterization of glutamate uptake into synaptic vesicles. J. Neurochem. 44, 689–693. [87]
316
REFERENCES
Nanko, S., Hattori, M., Ikeda, K., Sasaki, T., Kazamatsuri, H., Kuwata, S., 1993. Dopamine D4 receptor polymorphism and schizophrenia. Lancet. 341, 689–690. [99] Nansen, F., 1887. The Structure and combination of the Histological Elements of the Central Nervous System. Bergen: Bergen Mus. Aarsberetning. [18] Narahashi, T., Deguchi, T., Urakawa, N., Ohkubo, Y., 1960. Stabilization and rectification of muscle fiber membrane by tetrodotoxin. Am. J. Physiol. 198, 934–938. [145] Nastuk, M.A., Fallon, J.R., 1993. Agrin and the molecular choreography of synapse formation. Trends. Neurosci. 16(2), 72–76. [273] Neher, E., Sakmann, B., 1976. Single-channel currents recorded from membrane of denervated frog muscle fibres. Nature. 260, 799–802. [61, 62, 151] Neher, E., Sakmann, B., Steinbach, J.H., 1978. The extracellular patch clamp: a method for resolving currents through individual open channels in biological membranes. Pflugers Arch.—Eur. J. Physiol. 375, 219–228. [151] Nernst, W., 1888. Zur kinetik der in lösung befindlichen körper: Theorie der diffusion. Z. Chem. Phys. 613–637. [15] Nernst, W., 1908. Zur Theòrie des elektrischem reizes. Pflügers Arch. d. ges. Physiol. 122, 275–314. [15] Neve, K.A., Henningsen, R.A., Bunzow, J.R., Civelli, O., 1989. Functional characterization of rat dopamine D-2 receptor cDNA expressed in a mammalian cell line. Mol. Pharmacol. 36, 446–451. [98] Nguyen, T.V., Kosofsky, B.E., Birnbaum, R., Cohen, B.M., Hyman, S.E., 1992. Differential expression of c-fos and zif268 in rat striatum after haloperidol, clozapine, and amphetamine. Proc. Nat. Acad. Sci. USA. 89, 4270–4274. [104] Nichol, K.A., Chan, N., Davey, D.F., Bennett, M.R., 1995. Location of nitric oxide synthase in the developing avian ciliary ganglion. J. Auton. Nerv. Sys. 51, 91–102. [132] Niel, J.P., Bywater, R.A., Taylor, G.S., 1983a. Apamin-resistant post-stimulus hyperpolarization in the circular muscle of the guinea-pig ileum. J. Auton. Nerv. Sys. 9, 565–569. [127] Niel, J.P., Bywater, R.A., Taylor, G.S., 1983b. Effect of substance P on non-cholinergic fast and slow post-stimulus depolarization in the guinea-pig ileum. J. Auton. Nerv. Syst. 9, 573–584. [132] Niemegeers, C.J.E., Janssen, P.A.J., 1979. A systematic study of the pharmacological activities of dopamine antagonists. Life Sci. 24, 2201–2216. [102] Nilius, B., Hess, P., Lansman, J.B., Tsien, R.W., 1985. A novel type of cardiac calcium channel in ventricular cells. Nature. 316, 443–446. [153, 154] Nishi, R., Willard, A.L., 1985. Neurons dissociated from rat myenteric plexus retain differentiated properties when grown in cell culture. Neurosci. 16, 187–199. [131] Nitkin, R.M., Smith, M.A., Magill, C., Fallon, J.R., Yao, Y.M., Wallace, B.G., McMahan, U.J., 1987. Identification of agrin, a synaptic organizing protein from Torpedo electric organ. J. Cell Biol. 105, 2471–2478. [271] Niznik, N.B., 1987. Dopamine receptors: molecular structure and function. Molec. Cell Endocrinol. 54, 1–22. [95] Niznik, H.B., Jarvie, K.R., Brown, E.M., 1989. Dopamine D1 receptors of the calf parathyroid gland: identification of a ligand binding subunit with lower apparent molecular weight but similar primary structure to neuronal D1-receptors. Biochem. 28, 6925–6930. [95] Nja, A., Purves, D., 1977a. Specific innervation of guinea-pig superior cervical ganglion cells by preganglionic fibres arising from different levels of the spinal cord. J. Physiol. 264(2), 565–583. [278, 281] Nja, A., Purves, D., 1977b. Reinnervation of guinea-pig superior cervical ganglion cells by preganglionic fibres arising from different levels of the spinal cord. J. Physiol. 272(2), 633–651. [278, 281] Nja, A., Purves, D., 1978. Specificity of initial synaptic contacts made on guinea-pig superior cervical ganglion cells during regeneration of the cervical sympathetic trunk. J. Physiol. 281, 45–62. [278] Noakes, P.G., Bennett, M.R., 1987. Growth of axons into developing muscles of the chick forelimb is preceded by cells that stain with Schwann cell antibodies. J. Comp. Neurol. 259(3), 330–347. [283] Noakes, P.G., Bennett, M.R., Davey, D.F., 1983. Growth of segmental nerves to the developing rat diaphragm: absence of pioneer axons. J. Comp. Neurol. 218(4), 365–377. [283] Noakes, P.G., Gautam, M., Mudd, J., Sanes, J.R., Merlie, J.P., 1995. Aberrant differentiation of neuromuscular junctions in mice lackingslaminin/laminin beta 2. Nature. 374(6519), 258–262. [275] Noakes, P.G., Bennett, M.R., Stratford, J., 1998. Migration of Schwann cells and axons into developing chick forelimb muscles following removal of either the neural tube or the neural crest. J. Comp. Neurol. 277(2), 214–233. [283] Noakes, P.G., Everett, A.W., Bennett, M.R., 1986. The growth of muscle nerves in relation to the formation of primary myotube in the developing chick forelimb. J. Comp. Neurol. 248, 245–256. [240] Nobili, C.L., 1824. über einen neuen Galvanometer. J. Chem. u. Phys. 45, 249–260. [12] Nobili, L., 1828. Comparison entre les deux galvanometres les plus sensibles, la grenouille et le multiplicateur a deux arguilles, suivie de quelques resultats noveaux. Ann. Chim. et Phys. 38, 225–245. [12] Nonet, M.L., Grundahl, K., Meyer, B.J., Rand, J.B., 1993. Synaptic function is impaired but not eliminated in C. elegans mutants lacking synaptotagmin. Cell. 73, 1291–1305. [160] Nonomura, Y., Hotta, Y., Ohashi, H., 1966. Tetrodotoxin and manganese ions: effects on electrical activity and tension in taenia coli of guinea pig. Science. 152, 97–99. [145] Norberg, K.A., 1964. Adrenergic innervation of the intestinal wall studied by fluorescence microscopy. Int. J. Neuropharmacol. 3, 379–382. [125] Nowak, L., Bregestovski, P., Ascher, P., Herbet, A., Prochiantz, A., 1984. Magnesium gates glutamate-activated channels in mouse central neurones. Nature. 307, 462–465. [219, 221]
REFERENCES
317
Nowycky, M.C., Fox, A.P., Tsien, R.W., 1985a. Three types of neuronal calcium with different calcium agonist sensitivity. Nature. 316, 440–443. [153, 154] Nowycky, M.C., Fox, A.P., Tsien, R.W., 1985b. Long-opening mode of gating of neuronal calcium channels and its promotion by the dihydropyridine calcium agonist Bay K 8644. Proc. Natn. Acad. Sci. U.S.A. 82, 2178–2182. [153] Nyback, H., Sedvall, G., 1970. Further studies on the accumulation and disappearance of catecholamines formed from tyrosine-14C in mouse brain. Effect of some phenothiazine analogues. Eur. J. Pharmacol. 10, 193–205. [92] Nystrom, R.R., Ko, C.P., 1988. Disruption of active zones in frog neuromuscular junctions following treatment with proteolytic enzymes. J. Neurocytol. 17(1), 63–71. [276] Obata, K., Takeda, K., 1969. Release of gamma-aminobutyric acid into the fourth ventricle induced by stimulation of the cat’s cerebellum. J. Neurochem. 16, 1043–1047. [85] Obata, K., Ito, M, Ochi, R., Sato, N., 1967. Pharmacological properties of the postsynaptic inhibition of Purkinje cell axons and the action of gamma-aminobutyric acid on Deiter’s neurones. Exp. Brain Res. 4, 43–57. [85] O’Brien, R.A.D., Ostberg, A.J.C., Vrbova, G., 1978. Observations on the elimination of polyneuronal innervation in developing mammalian skeletal muscle. J. Physiol. 282, 571–582. [242, 245–247] O’Connor, L.T., Lauterborn, J.C., Gall, C.M., Smith, M.A., 1994. Localization and alternative splicing of agrin mRNA in adult rat brain: transcripts encoding isoforms that aggregate acetylcholine receptors are not restricted to cholinergic regions. J. Neurosci. 14, 1141–1152. [273, 279] O’Connor, V.M., Shamotienko, O., Grishin, E. and Betz, H., 1993. On the structure of the ‘synaptosecretosome’. Evidence for a neurexin/ synaptotagmin/syntaxin/Ca2+ channel complex. FEBS Lett. 326(1–3), 255–260. [163, 276] Oersted, H.C., 1820. Galvanic magnetism. Phil. Mag. 56, 394. Ogston AG., 1955. Removal of acetylcholine from a limited volume by diffusion. J. Physiol. 128, 222–223. [32, 36] Ohga, A., Nakazato, Y., Saito, K., 1970. Considerations of the efferent nervous mechanism of the vago-vagal reflex relaxation of the stomach in the dog. Jpn. J. Pharmacol. 20, 116–130. [113] O’Keefe, J., 1976. Place units in the hippocampus of the freely moving rat. Exp. Neurol. 51, 78–109. [232] O’Keefe, J., Nagel, L., 1978. The hippocampus as a Cognitive Map. Oxford: Clarendon Press. [230, 232] Olesko, K.M., Holmes, F.L., 1993. Experiment, quantification and discovery: Helmholtz’s early physiological researches, 1843–1850. In: Cahan Hermann von Helmholtz and the Foundations of Nineteenth-Century Science. Univ. California Press, Berkeley. [15] Oliver, G., Schafer, E.A., 1895. The physiological effects of extracts of the suprarenal capsules. J. Physiol. 18, 230– 276. [65, 66] Olivera, B.M., Miljanich, G.P., Ramachandran, J., Adams, M.E., 1994. Calcium channel diversity and neurotransmitter release: the omegaconotoxins and omega-agatoxins. Annu. Rev. Biochem. 63, 823–867. [153] Olney, J.W., Ho, O.L., Rhee, V., 1971. Cytotoxic effects of acidic and sulphur containing amino acids on the infant mouse central nervous system. Exp. Brain Res. 14, 61–76. [85] Olney, J.W., Sharpe, L.G., Feigin, R.D., 1972. Glutamate-induced brain damage in infant primates. J. Neuropath, exp. Neurol. 31, 464–488. [85] Olverman, H.J., Jones, A.W., Watkins., J.C., 1984. L-Glutamate has higher affinity than other aminoacids for [3H]-D-AP5 binding sites in rat brain membranes. Nature. 307, 460–462. [88] O’Malley, K.L., Harmon, S., Tang, L., Todd, R.D., 1992. The rat dopamine D4 receptor: sequence, gene structure, and demonstration of expression in the cardiovascular system. New Biologist. 4, 137–146. [98] Ontell, M.A., Dunn, R.F., 1978. Neonatal muscle growth: a quantitative study. Am. J. Anat. 152, 539–556. [242] Openchowski, T. von, 1889. Zur pathologischen Anatomie der geschwurigen Prozesse im Magendarmtractus. Arch. Path. Anat. Physiol. CXVII, 347–356. [111, 120] Oppenheim, R.W., Chu-Wang, I.W., 1983. Aspects of naturally occurring motoneurone death in the chick spinal cord during embryonic development. In: Somatic and Autonomic Nerve-Muscle Interactions. Burnstock, G., O’Brien, R., Vrbova, G., (ed.). Elsevier Publishers pp. 57–107. [242, 268] Otis, T.S., Kavanaugh, M.P., Jahr, C.E., 1997. Postsynaptic glutamate transport at the climbing fibres-Purkinje cell synapses. Science. 211, 1515–1518. [87] Ozaki, H., Blondfield, D.P., Hori, M., Publicover, N.G., Kato, I., Sanders, K.M., 1992. Spontaneous release of nitric oxide inhibits electrical, Ca2+ and mechanical transients in canine gastric smooth muscle. J. Physiol. 445, 231–247. [131] Palade, G.E., 1954. Electron microscope observations of interneuronal and neuromuscular synapses. Anat. Rec. 118, 335– 336. [40] Paton, W.D.M., 1954. The response of the guinea-pig ileum to electrical stimulation by coaxial electrodes. J. Physiol. 127, 40P–41P. [114] Paton, W.D.M., Vane, J.R., 1963. An analysis of the responses of the isolated stomach to electrical stimulation and to drugs. J. Physiol. 165, 10–46. [111, 113, 114, 120] Patthy, L., Nikolics, K., 1993. Functions of agrin-related proteins. Trends. NeuroSci. 16(2), 76–81. [273] Patthy, L., Nikolics, K., 1994. Agrin-like proteins of the neuromuscular junction. Neurochem. Int. 24, 301–316. [273] Peck, A.L., 1943. ‘Symphyton pneuma’, Appendix B, pp. 576– 593. Translation of Aristotles ‘De Generatione Animalium’, Cambridge, London. [3] Perin, M.S., Brose, N., Jahn, R., Sudhof, T.C., 1991. Domain structure of synaptotagmin (p65). J. Biol. Chem. 266, 623– 629. [160] Perin, M.S., Fried, V.A., Mignery, G.A., Jahn, R., Sudhof, T.C., 1990. Phospholipid binding by a synaptic vesicle protein homologous to the regulatory region of protein kinase C. Nature. 345, 260–263. [162] Pettigrew, A., Lindeman, R., Bennett, M.R., 1979. The development of the segmental innervation of the chick forelimb. J. Embryol. Exp. Morphol. 49, 115–137. [242]
318
REFERENCES
Phillips, W.D., Bennett, M.R., 1984. Differentiation of fiber types in wing muscles during embryonic development: effect of neural tube removal. Develop. Biol. 106, 457– 468. [260, 262] Phillips, W.D., Bennett, M.R., 1987. Elimination of distributed acetylcholine receptor clusters from developing fast-twitch fibres in an avian muscle. J. Neurocytol. 16, 1–10. [242] Phillips, W.D., Lai, K., Bennett, M.R., 1985. Spatial distribution and size of acetylcholine receptor clusters determined by motor nerves in developing chick muscles. J. Neurocytol. 14, 309–325. [242] Phillis, J.W., Krnjevic, K., 1961. Sensitivity of cortical neurones to acetylcholine. Experentia. 17, 469. [81] Piccolino, M., 1998. Animal electricity and the birth of electrophysiology: the legacy of Luigi Galvani. Brain Res. Bull. 46, 381–407. [14] Pin, J.-P., Duvoisin, R., 1995. The metabotropic receptors: structure and function. Neuropharmacol. 34, 1–26. [88] Piochet, P., 1996. The discovery of chlorpromazine and the place of psychopharmacology in the history of psychiatry. In: The Psychopharmacologists, Healy, D., (ed.). pp 1–28. Chapman & Hall, New York. [91] Plato, 1892. The Dialogues of Plato. B.Jowett, 3rd ed. 5 vols. Oxford: Clarendon Press. [1] Pockett, S., 1981. Elimination of polyneuronal innervation in proximal and distal leg muscles of chick embryos. Develop. Brain Res. 1, 299–302. [242] Polak, R.L., 1971. Stimulating action of atropine on the release of acetylcholine by rat cerebral cortex in vitro. Brit. J. Pharmacol. 41, 600–606. [69, 71] Poo, M.M., 1981. In situ electrophoresis of membrane components. Ann. Rev. Biophys. Bioeng. 10, 245–276. [247] Poo, M.M., 1985. Mobility and localization of proteins in excitable membranes. Ann. Rev. Neurosci. 8, 369–406. [247] Posner, M.I., Raichle, M.E., 1994. ‘Images of Mind’. W.H. Freeman & Company, New York. [2] Press, W.H., Flannery, B.P., Teukolsky, S.A., Vetterling, W.T., 1986. Numerical Recipes. Cambridge University Press. Cambridge. [194, 209] Prestige, M.C., 1967. The control of cell numbers in the lumbar ventral horns during the development of Xenopus laevis tadpoles. J. Embryol. Exp. Morph. 18, 359–387. [242] Prisko, L., 1963. Short-term memory in focal cerebral damage (PhD Thesis), pg 126, McGill University, Montreal. [211] Prokop, A., Landgraf, M., Rushton, E., Broadie, K., Bate, M., 1996. Presynaptic development at the Drosophila neuromuscular junction: assembly and localization of presynaptic active zones. Neuron. 17(4), 617–626. [276] Propst, J.W., Herrera, A.A., Ko, C.P., 1986. A comparison of active zone structure in frog neuromuscular junctions from two fast muscles with different synaptic efficacy. J. Neurocytol. 15(4), 525–534. [275] Propst, J.W., Ko, C.P., 1987. Correlations between active zone ultrastructure and synaptic function studied with freeze-fracture of physiologically identified neuromuscular junctions. J. Neurosci. 7(11), 3654–3664. [275] Protti, D.A., Sanchez, V.A., Cherksey, B.D., Sugimori, M., Llinas, R., Uchitel, O.D., 1993. Mammalian neuromuscular transmission blocked by funnel web toxin. Ann. N.Y. Acad. Sci. 681, 405–407. [158] Publicover, N.G., Hammond, E.M., Sanders, K.M., 1993. Amplification of nitric oxide signalling by interstitial cells isolated from canine colon. Proc. Nat. Acad. Sci. USA. 90, 2087–2091. [128, 129, 131] Purkinje, J., 1837a. Anatomisch-phisiologische Verhandlungen, pp. 177–180. Bericht über die Versammlung deutscher Naturforscher und Arzte in Prag. Prague Pt. 3 sec 5. [15, 16] Purkinje, J.E., 1837b. Neueste Beobachtungen über die Struktur des Gehirns. Opera Omnia 2, 88. [15] Purkinje, J.E., 1837c. Neueste Untersuchungen aus der Nervenund Hirnanatomie. Opera Omnia 3, 45–49. [15] Purpura, D.P., Girado, M., Grundfest, H., 1957. Selective blockade of excitatory synapses in the cat brain by gamma-aminobutyric acid. Science. 125, 1200–1202. [79, 80, 83] Purpura, D.P., Girado, M., Smith, T.G., Callan, D.A., Grundfest, H., 1959. Structure-activity determinants of pharmacological effects of amino acids and related components on central synapses. J. Neurochem. 3, 238–268. [79] Purves, D., 1976. Competitive and non-competitive reinnervation of mammalian sympathetic neurones by native and foreign fibres. J. Physiol. 261(2), 453–475. [278, 279] Purves, D., 1977. The formation and maintenance of synaptic connections. In: Function and formation of neural systems. G.S.Stent, (ed.). Berlin. Dahlem Konferenzen, pp. 21–49. [242] Purves, D., 1986. The trophic theory of neural connections. TINS 9, 486–489. [242] Purves, D., Thompson, W., 1979. The effects of post-ganglionic axotomy on selective synaptic connexions in the superior cervical ganglion of the guinea-pig. J. Physiol. 297, 95– 110. [278] Purves, D., Wigston, D.J., 1983. Neural units in the superior cervical ganglion of the guinea-pig. J. Physiol. 334, 169– 178. [278] Purves, D., Hume, R.I., 1981. The relation of postsynaptic geometry to the number of presynaptic axons that innervate autonomic ganglion cells. J. Neurosci. 1(5), 441–452. [277] Purves, D., Lichtman, J.W., 1980. Elimination of synapses in the developing nervous system. Science. 210, 153–157. [242] Purves, D., Lichtman, J.W., 1987. Synaptic sites on reinnervated nerve cells visualized at two different times in living mice. J. Neurosci. 7 (5), 1492–1497. [277] Purves, D., Thompson, W., Yip, J.W., 1981. Re-innervation of ganglia transplanted to the neck from different levels of the guinea-pig sympathetic chain. J. Physiol. 313, 49– 63. [278] Qu, Z., Huganir, R.L., 1994. Comparison of innervation and agrin-induced tyrosine phosphorylation of the nicotinic acetylcholine receptor. J. Neurosci. 14(11 Pt 2), 6834– 6841. [273] Racine, R.J., Milgram, N.W., Hafner, S., 1983. Long-term potentiation phenomena in the rat limbic forebrain. Brain. Res. 260, 217–231. [214, 215]
REFERENCES
319
Rae, M.G., Muir, T.C., 1996. Neuronal mediators of inhibitory junction potentials and relaxation in the guinea-pig internal anal sphincter. J. Physiol. 493, 517–527. [130] Ramagopal, M.V., Leighton, H.J., 1989. Effects of NG-monomethyl-L-arginine on field stimulation-induced decreases in cytosolic Ca2+ levels and relaxation in the rat anococcygeus muscle. Eur. J. Pharmacol. 174, 297–299. [131] Rand, M.J., Mitchelson, F., 1986. The guts of the matter: contribution of studies on smooth muscle to discoveries in pharmacology. In: Discoveries in Pharmacology, Vol. 3 Pharmacological Methods, Receptors and Chemotherapy, Ch. 2, Part 1. Parnham, M.J., Bruinvels, J. (eds.). Elsevier, Amsterdam. [120] Ranson, S.W., Foley, J.O., Alpert, C.D., 1933. Observations on the structure of the vagus nerve. Amer. J. Anat. 53, 289– 315. [113] Rasmussen, C.E., Willshaw, D.J., 1993. Presynaptic and postsynaptic competition in models for the development of neuromuscular connections. Biol. Cybern. 68(5), 409–419. [240, 243, 252, 253, 258] Redfern, P.A., 1970. Neuromuscular transmission in new-born rats. J. Physiol. 209, 701–709. [238, 239, 242] Regehr, W.G., Mintz, I.M., 1994. Participation of multiple calcium channel types in transmission at single climbing fiber to Purkinje cell synapses. Neuron. 12, 605–613. [155] Reid, C.A., Clements, J.A., Bekkers, J.M., 1997. Nonuniform distribution of calcium channel subtypes on presynaptic terminals of excitatory synapses in hippocampal cultures. J. Neurosci. 17, 2738–2745. [155] Reist, N.E., Werle, M.J., McMahan, U.J., 1992. Agrin released by motor neurons induces the aggregation of acetylcholine receptors at neuromuscular junctions. Neuron. 8(5), 865– 868. [271] Reuter, H., 1967. The dependence of slow inward current in Purkinje fibres on the extracellular calcium-concentration. J. Physiol. 192, 479–492. [142, 143] Reuter, H., 1973. Divalent cations as charge carriers in excitable membranes. In: Progress in Biophysics and Molecular Biology, vol. 26, pp. 1–43. Butler, J.A.V., Noble, D., (eds.). Pergamon, Oxford. [142] Reuter, H., Scholz, H., 1968. On the influence of the extracellular calcium concentration on membrane potential and contraction of isolated heart preparations during graded depolarization. Pflugers Arch. ges. Physiol. 300, 87–107. [142, 143] Reuter, H., Stevens, C.F., Tsien, R.W., Yellen, G., 1982. Properties of single calcium channels in cardiac cell culture. Nature. 297, 501–504. [152, 153] Reymann, K.G., Frey, U., Jork, R., Matthies, H., 1988. Polymyxin B, an inhibitor of protein kinase C, prevents the maintenance of synaptic long-term potentiation in hippocampal CA1 neurones. Brain Res. 440, 305–314. [225] Reymann, K.G., Matthies, H.K., Schulzeck, K., Matthies, H., 1989. N-methyl-D-aspartate receptor activation is required for the induction of both early and late phases of long-term potentiation in rat hippocampal slices. Neurosci Letts. 96, 96–101. [218] Reynolds, G.P., 1997. 5-HT receptors, genes and antipsychotic mechanisms. J Psychopharm. 11, 26–27. [103] Reynolds, G.P., Mason, S.L., 1994. Are striatal dopamine D4 receptors increased in schizophrenia? J. Neurochem. 63, 1576–1577. [101] Ribchester, R.R., 1988a. Activity-dependent and -independent synaptic interactions during reinnervation of partially denervated rat muscle. J. Physiol. 401, 53–75. [258] Ribchester, R.R., 1988b. Competitive elimination of neuromuscular synapses. Nature. 331(6151), 21–22. [258] Ribchester, R.R., Taxt, T., 1983. Motor unit size and synaptic competition in rat lumbrical muscles reinnervated by active and inactive motor axons. J. Physiol. 344, 89–111. [254, 258] Rich, M., Lichtman, J.W., 1989. Motor nerve terminal loss from degenerating muscle fibers. Neuron. 3(6), 677–688. [276] Rich, M.M., Colman, H., Lichtman, J.W., 1994. In vivo imaging shows loss of synaptic sites from neuromuscular junctions in a model of myasthenia gravis. Neurol. 44(11), 2138– 2145. [277] Richardson, K.C., 1962. The fine structure of autonomic nerve endings in smooth muscle of the rat vas deferens. J. Anat. 96, 427–442. [36] Ridge, R.M.A.P., Betz, W.J., 1984. The effect of selective, chronic stimulation on motor unit size in developing rat muscle. J. Neurosci. 4, 2614–2620. [268, 269] Ridge, R.M.A.P., Thomson, A.M., 1980. Polyneuronal innervation: mechanical properties of overlapping motor units in a small foot muscle of Xenopus laevis. J. Physiol. 306, 29–39. [246, 252] Riley, D.A., 1977. Spontaneous elimination of nerve terminals from the endplates of developing skeletal myofibres. Brain Res . 134, 279–285. [242] Riley, D.A., 1981. Ultrastructural evidence of axon retraction during the spontaneous elimination of polyneuronal innervation of the rat soleus muscle. J. Neurocytol. 10, 425–440. [242] Robbins, J., 1959. The excitation and inhibition of crustacean muscle by amino acids. J. Physiol. 148, 39–50. [79] Roberts, E., Frankel, S., 1951. Glutamic acid decarboxylase in brain. J. Biol Chem. 188, 789–795. [79] Roberts, E., Frankel, S. Harman, P.J., 1950. Amino acids of nervous tissue. Proc. Soc. Exp. Biol. (N.Y.) 74, 383–387. [79] Roberts, P.J., 1974. Glutamate receptors in the rat central nervous system. Nature. 252, 399–401. [88] Robertson, J.D., 1956. The ultrastructure of a reptilian myoneural junction. J. Biophysic. & Biochem. Cytol. 2, 381–394. [36, 40] Robertson, B.E., Schubert, R., Hescheler, J., Nelson, M.T., 1993. cGMP-dependent protein kinase activates Ca-activated K channels in cerebral artery smooth muscle cells. Amer. J. Physiol. 265, C299–C303. [131] Robinson, I.M., Yamada, M., Carrion-Vazquez, M., Lennon, V.A., Fernandez, J.M., 1996. Specialized release zones in chromaffin cells examined with pulsed-laser imaging. Cell Cal. 20, 181–201. [161] Robinson, J., 1976. Estimation of parameters for a model of transmitter release at synapses. Biometric. 32, 61–68. [170– 173, 183, 186] Robinson, M.B., Dowd, L.A., 1997. Heterogeneity and functional subtypes of sodium-dependent glutamate transporters in the mammalian central nervous system. Adv. Pharmacol. 37, 69–115. [87]
320
REFERENCES
Robitaille, R., Adler, E.M., Charlton, M.P., 1993. Calcium channels and calcium-gated potassium channels at the frog neuromuscular junction . J. Physiol. Paris. 87(1), 15–24. [275] Roman, F., Staubli, U., Lynch, G., 1987. Evidence for synaptic potentiation in a cortical network during learning. Brain Res. 418, 221–226. [230] Roots, B.I., 1983. Neurofilament accumulation induced in synapses by leupeptin. Sci. 221, 971–972. [242] Rosenthal, J.L., Taraskevich, P.A., 1977. Reduction of multiaxonal innervation at the neuromuscular junction of the rat during development. J. Physiol. 270, 299–310. [242] Roseth, S., Fykse, E.M., Fonnum, F., 1995. Uptake of L-glutamate into rat brain synaptic vesicles: effect of inhibitors that bind specifically to the glutamate transporter. J. Neurochem. 65, 96–103. [87] Rossum, J.M. van, 1966. The significance of dopamine-receptor blockade for the mechanism of action of neuroleptic drugs. Arch. Int. Pharmacodyn. Ther. 160, 492–494. [92, 93] Rothman, J.E., 1994. Mechanism of intracellular protein transport. Nature. 372, 55–63. [180] Rothstein, J.D., Martin, L., Levey, A.I., Dykes-Hoberg, M., Jin, L., Wu, D., Nash, N., Kuncl, R.W., 1994. Localization of glia and neuronal transporters. Neuron. 13, 713–725. [87] Rotshenker, S., McMahan, U.J., 1976. Altered patterns of innervation in frog muscle after denervation. J. Neurocytol. 5, 719–730. [242] Roufogalis, B.D., Thornton, M., Wade, D.N., 1976. Specificity of the dopamine sensitive adenylate cyclase for antipsychotic antagonists. Life Sci. 19, 927–934. [95] Routtenberg, A., Colley, P., Linden, D., Lovinger, D., Murakami, K., Sheu, F.S., 1986. Phorbol ester promotes growth of synaptic plasticity. Brain Res. 378, 374–378. [228] Rubin, E., 1985. Development of the rat superior cervical ganglion: ingrowth of preganglionic axons. J. Neurosci. 5(3), 685–696. [277] Ruegg, M.A., Tsim, K.W., Horton, S.E., Kroger, S., Escher, G., Gensch, E.M., McMahan, U.J., 1992. The agrin gene codes for a family of basal lamina proteins that differ in function and distribution. Neuron. 8(4), 691–699. [273] Rupp, F., Payan, D.G., Magill-Solc, C., Cowan, D.M., Scheller, R.H., 1991. Structure and expression of a rat agrin. Neuron. 6(5), 811–823. [273] Rushton, W.A.H., 1930. Excitable substances in the nervemuscle complex. J. Physiol. LXX, 317–337. [59] Rushton, W.A.H., 1932. Identification of the gamma excitability in muscle. J. Physiol. LXXV, 161–189. [59] Saffrey, M.J., Bailey, D.J., Burnstock, G., 1991. Growth of enteric neurones from isolated myenteric ganglia in dissociated cell culture. Cell & Tiss. Res. 265, 527–534. [131] Saffrey, M.J., Hassal, C.J., Hoyle, C.H., Belai, A., Moss, J., Schmidt, H.H., Forstermann, U., Murad, E, Burnstock, G., 1992. Colocalization of nitric oxide synthase andNADPH-diaphorase in cultured myenteric neurones. NeuroReport. 3, 333–336. [131] Saito, A., Zacks, S.I., 1969. Fine structure of neuromuscular junctions after nerve sections and implantation of nerve in denervated muscle. Exp. Mol Pathol. 10, 256–273. [235] Salganicoff, L., De Robertis, E., 1965. Subcellular distribution of the enzymes of the glutamic acid, glutamin and gaminobutyric acid cycles in rat brain. J. Neurochem. 12, 287–309. [85] Sanders, K.M., Ward, S.M., 1992. Nitric oxide as a mediator of nonadrenergic noncholinergic neurotransmission. Amer. J. Physiol 262, G379–G392. [127] Sanders, K.M., Ward, S.M., Thornbury, K.D., Dalziel, H.H., Westfall, D.P., Carl, A., 1992. Nitric oxide as a non-adrenergic, noncholinergic neurotransmitter in the gastrointestinal tract. Jpn. J. Pharmacol. 58, 220P–225P. [129] Sanes, J.R., Chiu, A.Y., 1983. The basal lamina of the neuromuscular junction. Cold Spring Harbor Symp. Quant. Biol. 48, 667–678. [242] Sanes, J.R., Hall, Z.W., 1979. Antibodies that bind specifically to synaptic sites on muscle fiber basal lamina. J. Cell Biol. 83, 357–370. [242] Sanes, J.R., Johnson, Y.R., Kotzbauer, P.T., Mudd, J., Hanley, T., Martinou, J.C., Merlie, J.P., 1991. Selective expression of an acetylcholine receptor-lacZ transgene in synaptic nuclei of adult muscle fibers. Develop. 113(4), 1181–1191. [277] Sanes, J.R., Marshall, L.M., McMahan, U.J., 1978. Reinnervation of muscle fiber basal lamina after removal of myofibers. Differentiation of regenerating axons at original synaptic sites. J. Cell Biol 78, 176–198. [270, 271, 275, 276] Sano, K., Enomoto, K., Maeno, T., 1987. Effects of synthetic omega-conotoxin, a new type Ca2+ antagonist, on frog and mouse neuromuscular transmission. Eur. J. Pharmacol 141, 235–241. [158] Savaria, D., Lanoue, C., Cadieux, A., Rousseau, E., 1992. Large conducting potassium channel reconstituted from airway smooth muscle. Amer. J. Physiol. 262, L327–L336. [131] Schattenberg, P.J., 1973. Untersuchungen uber das Langenwachstum der Skelettmuskulatur von Fischen. Z. Zellforsch. Mikrosk. Anat. 143, 587–596. [238] Schlaepfer, W.W., Hasler, M.B., 1979. Characterization of the calcium-induced disruption of neurofilaments in rat peripheral nerve. Brain Res. 168, 299–309. [242] Schmutz, J., Eichenberger, E., 1982. Clozapine. In: Chronicles of Drug Discovery. Vol. 1, D Lednicer (ed), Wiley, New York, 39–59. [99, 102] Scholz, H., 1969. Differences in contractural character of ventricular and atrial preparations from homeotherm hearts. Pflugers Arch.—Eur. J. Physiol 312, 63–81. [142] Scholz, H., Reuter, H., 1969. Electrophysiological studies of the action of tetrodotoxin on the heart. [German] Elektrophysiologische Versuche zur Wirkung von Tetrodotoxin am Herzen. Naunyn-Schmied. Arch. Pharmakol exp. Pathol 263, 248. [142] Schousboe, A., 1981. Transport and metabolism of glutamate and GABA in neurons and glial cells. Int. Rev. Neurobiol 22, 1–45. [87]
REFERENCES
321
Schultze, M., 1870. General character of nervous tissue. Vol. 1, pp. 147–186, in Manual of Human and Comparative Histology. S. Stricker, S., (ed.). The New Sydenham Society, London). [17] Schwartzkroin, P.A., Wester, K., 1975. Long-lasting facilitation of a synaptic potential following tetanization in the in vitro hippocampal slice. Brain Res. 89, 107–119. [211] Schworer, H., Clemens, A., Katsoulis, S., Kohler, H., Creutzfeldt, W., Schmidt, W.E., 1993. Pituitary adenylate cyclase-activating peptide is a potent modulator of human colonic motility. Scand. J. Gastroenterol 28, 625–632. [129] Schworer, H., Katsoulis, S., Creutzfeldt, W, Schmidt, WE., 1992. Pituitary adenylate cyclase activating peptide, a novel VIP-like gut-brain peptide, relaxes the guinea-pig taenia caeci via apamin-sensitive potassium channels. Naunyn-Schmiedebergs Arch. Pharmacol 346, 511–514. [129, 131] Scoville, W.B., Milner, B., 1957. Loss of recent memory after bilateral hippocampal lesions. J. Neurol. Neurosurg. Psychiat. 20, 11–21. [211, 212] Seeman, P., 1980. Dopamine receptors. Pharmacol. Rev. 32, 229–313. [95] Seeman, P., 1981. Dopamine receptors in post-mortem schizophrenic brains. Lancet. 1(8229), 1103. [98] Seeman, P., 1992. Dopamine receptor sequences. Therapeutic levels of neuroleptics occupy D2 receptors, clozapine occupies D4. Neuropsychopharmacol 7, 261–284. [97] Seeman, P., Bzowej, N.H., Guan, H.C., Bergeron, C., Reynolds, G., Bird, E.D., Riederer, P., Jellinger, K., Tourtellotte, W.W, 1987. Human brain D1 and D2 dopamine receptors in schizophrenia, Alzheimer’s, Parkinson’s, and Huntington’s diseases. Neuropsychopharmacol 1, 5–15. [99] Seeman, P., Chau-Wong, M., Tedesco, J., Wong, K., 1975. Brain receptors for antipsychotic drugs and dopamine: direct binding assays. Proc. Nat. Acad. Sci. USA. 72, 4376–4380. [95–97] Seeman, P., Chau-Wong, M., Tedesco, J., Wong, K., 1976. Dopamine receptors in human and calf brains, using [3H]apomorphine and an antipsychotic drug. Proc. Nat. Acad. Sci. USA. 73, 4354–4358. [99] Seeman, P., Guan, H.C., Van Tol, H.H., 1993. Dopamine D4 receptors elevated in schizophrenia. Nature. 365, 441–445. [101] Seeman, P., Niznik, H.B., 1990. Dopamine receptors and transporters in Parkinson’s disease and schizophrenia. FASEB J. 4, 2737–2744. [95, 102] Seeman, P., Niznik, H.B., Guan, H.C., Booth, G., Ulpian, C., 1989. Link between D1 and D2 dopamine receptors is reduced in schizophrenia and Huntingdon diseased brain. Proc. Nat. Acad. Sci. USA. 86, 10156–10160. [102] Seeman, P., Van Tol, H.H.M., 1994. Dopamine receptor pharmacology. Trends Pharmacol Sci. 15, 264–270. [97] Seeman, P., Van Tol, H.H.M., 1995a. Dopamine D4-like receptor elevation in schizophrenia: cloned D2 and D4 receptors cannot be discriminated by raclopride competition against [3H] nemonapride. J. Neurochem. 64, 1413–1415. [101] Seeman, P., Van Tol, H.H.M., 1995b. Deriving the therapeutic concentrations for clozapine and haloperidol: the apparent dissociation constant of a neuroleptic at the dopamine D2 or D4 receptor varies with the affinity of the competing radioligand. Eur. J. Pharmacol 291, 59–66. [101] Selbie, L.A., Hayes, G., Shine, J., 1989. The major dopamine D2 receptor: molecular analysis of the human D2A subtype. DNA. 8, 683–689. [98] Semba, T., Fujii, K., Kimura, N., 1964. The vagal inhibitory responses of the stomach to stimulation of the dog’s medulla oblongata. Jpn. J. Physiol 14, 319–327. [112–114] Sen, S., Bose, K.C., 1931. Rauwolfia serpentina, a new Indian drug for insanity and high blood pressure. Indian Med, World, 2, 194–198. [91] Serio, R., Mule, F., Postorino, A., 1995. Nonadrenergic, noncholinergic inhibitory junction potentials in rat proximal colon: role of nitric oxide. Can. J. Physiol. Pharmacol. 73, 79–84. [128] Shaikh, S., Collier, D., Kerwin, R.W., Pilowsky, L.S., Gill, M., Xu, W.M., Thornton, A., 1993. Dopamine D4 receptor subtypes and response to clozapine. Lancet. 341, 116. [99] Sheng Z.H., Rettig, J., Takahashi, M., Catterall, W.A., 1994. Identification of a syntaxin-binding site on N-type calcium channels. Neuron. 13, 1303–1313. [163] Shepherd, G.M., 1991. Foundation of the Neuron Doctrine. Oxford University Press, Oxford. [11, 16, 17, 21] Sherrington, C.S., 1897. The central nervous system. Vol. 3. In: Foster, M., (ed.). A Textbook of Physiology, 7th ed. London: Macmillan. [25] Sherrington, CS., 1906. Integrative action of the nervous system. New Haven, Yale University Press. [25] Sherrington, C.S., 1925. Remarks on some aspects of reflex inhibition. Proc. Roy. Soc. Lond. B.97, 519–545. [27] Sherrington, C.S., to Fulton, J,. Letter 25th Dec. 1937. In: Swazey, J. Reflexes and motor integration: Sherrington s concept of integrative action. Cambridge, Mass: Harvard University Press, 1969. [22, 23] Sherrington, C.S., to Schäfer, E.A., Letter 27th Nov. 1897a. In the Sharpey-Schäfer papers in the Contemporary Medical Archives Centres, the Wellcome Institute for the History of Medicine, reference PP/ESS/B21/8. [22, 23] Sherrington, C.S., to Schäfer, E.A., Letter 1st Dec. 1897b. In the Sharpey-Schäfer papers in the Contemporary Medical Archives Centres, the Wellcome Institute for the History of Medicine Reference PP/ESS/B21/9. [22, 23] Shimomura, O., Johnson, F.H., Saiga, Y., 1962. Extraction, purification and properties of aequorin, a bioluminescent protein from the luminous hydromedusan, Aequorea. J. Cell. Comp. Physiol. 59, 223–239. [148] Shuba, M.F., Vladimirova, I.A., 1980. Effect of apamin on the electrical responses of smooth muscle to adenosine 5’-triphosphate and to non-adrenergic, non-cholinergic nerve stimulation. Neurosci. 5, 853–859. [130] Shukla, P.L., Aitken, J.T., 1963. Formation of motor endplates in denervated voluntary muscles of the rat. J. Anat. 97, 152P. [235] Shuttleworth, C.W., Sanders, K.M., 1996. Involvement of nitric oxide in neuromuscular transmission in canine proximal colon. Proc. Soc. Exper. Biol. & Med. 211, 16–23. [129]
322
REFERENCES
Shuttleworth, C.W., Xue, C., Ward, S.M., de Vente, J., Sanders, K.M., 1993. Immunohistochemical localization of 3’, 5’-cyclic guanosine monophosphate in the canine proximal colon: responses to nitric oxide and electrical stimulation of enteric inhibitory neurons. Neurosci. 56, 513–522. [129] Sibley, D.R., Monsma, F.J. Jr, 1992. Molecular biology of dopamine receptor. Trends. Pharmacol. Sci. 13, 61–69. [103] Sigworth, F.J., Neher, E., 1980. Single Na+ channel currents observed in cultured rat muscle cells. Nature. 287, 447–449. [151] Simon, S.M., Llinas, R.R., 1985. Compartmentalization of the submembrane calcium activity during calcium influx and its significance in transmitter release. Biophys. J. 48, 485–498. [163] Simon, S.M., Sugimori, M., Llinas, R., 1984. Modelling of submembranous calcium-concentration changes and their relation to rate of presynaptic transmitter release in the squid giant synapse . Biophys. J. 45, 264a. [163] Skarsfeldt, T., 1988. Differential effects after repeated treatment with haloperidol, clozapine, thioridazine and tefludazine on SNC and VTA dopamine neurones in rats. Life Sci. 42, 1037–1044. [99] Skrede, K.K., Malthe-Sorenssen, D., 1981a. Increased resting and evoked release of transmitter following repetitive electrical tetanization in hippocampus: a biochemical correlate to long-lasting synaptic potentiation. Brain Res. 208, 436– 441. [223] Skrede, K.K., Malthe-Sorenssen, D., 1981b. Differential release of D-[3H]aspartate and [14C]gamma-aminobutyric acid following activation of commissural fibres in a longitudinal slice preparation of guinea pig hippocampus. Neurosci. Letts. 21, 71–76. [223] Sladeczek, F., Pin, J.-R, Recasens, M., Bockaert., J., Weiss, S., 1985. Glutamate stimulates inositol phosphate formation in striatal neurones. Nature. 317, 717–719. [88] Smith, A.B., Cunnane, T.C., 1996. Omega-conotoxin GVIA-resistant neurotransmitter release in postganglionic sympathetic nerve terminals. Neurosci. 70, 817–824. [158] Smith, B.R., Wojtowicz, J.M., Atwood, H.L., 1991. Maximum likelihood estimation of non-uniform transmitter release probabilities at the crayfish neuromuscular junction. J. Theor. Biol. 150(4), 457–472. [189, 190] Smits, G.J., Lefebvre, R.A., 1994. Tachykinin receptors involved in the contractile effect of the natural tachykinins in the rat gastric fundus. J. Auton. Pharmacol. 14, 383–392. [132] Snutch, T.P., Reiner, P.B., 1992. Ca2+ channels: diversity of form and function. Curr. Opin. Neurobiol. 2, 247–253. [153] Snutch, T.P., Leonard, J.P., Gilbert, M.M., Lester, H.A., Davidson, N., 1990. Rat brain expresses a heterogeneous family of calcium channels. Proc. Natn. Acad. Sci. U.S.A. 87, 3391–3395. [155] Snyder, G.L., Girault, J.A., Chen, J.Y., Czernik, A.J., Kebabian, J.W., Nathanson, J.A., Greengard, P., 1992. Phosphorylation of DARPP-32 and protein phosphatase inhibitor-1 in rat choroid plexus: regulation by factors other than dopamine. J. Neurosci. 12, 3071–3083. [103] So, K.L., Pun, S., Wan, D.C., Tsim, K.W., 1996. Cerebellar granule cells express a specific isoform of agrin that lacks the acetylcholine receptor aggregating activity. FEBS Letts. 379(1), 63–68. [279] Soediono, P., Burnstock, G., 1994., Contribution of ATP and nitric oxide to NANC inhibitory transmission in rat pyloric sphincter. Brit. J. Pharmacol. 113, 681–686. [130] Soha, J.M., Callaway, E.M., Van Essen, D.C., 1989. Lack of fiber type selectivity during reinnervation of neonatal rabbit soleus muscle. Develop. Biol. 131(2), 401–414. [262] Sohal, G.S., 1981. Ultrastructural observations on the establishment of neuromuscular unctions following prevention of embryonic death of motoneurones. Exp. Neurol. 71, 625–634. [247] Sohal, G.S., Creazzo, T.L., Oblak, T.G., 1979. Effects of chronic paralysis with alpha-bungarotoxin on development of innervation. Exp. Neurol. 66, 619–628. [247] Soileau, L.C., Silberstein, L., Blau, H.M., Thompson, W.J., 1987. Reinnervation of muscle fiber types in the newborn rat soleus. J. Neurosci. 7(12), 4176–4194. [262] Sokoloff, P., Giros, B., Martres, M.P., Bouthenet, M.L., Schwartz, J.C., 1990. Molecular cloning and characterization of a novel dopamine receptor (D3) as a target for neuroleptics. Nature. 347, 146–151. [98, 101] Song, Z.M., Brookes, S.J., Llewellyn-Smith, I.J., Costa, M., 1995. Ultrastructural studies of the myenteric plexus and smooth muscle in organotypic cultures of the guinea-pig small intestine. Cell & Tiss. Res. 280, 627–637. [131] Spano, P.P., Govoni, S., Travucci, M., 1978. Studies on the pharmacological properties of dopamine receptors in various areas of the central nervous system. Adv. Biochem. Psychopharmacol. 19, 155–165. [95] Srihari, T., Vrbova, G., 1978. The role of muscle activity in the differentiation of neuromuscular junctions in slow and fast chick muscles. J. Neurocytol. 7, 529–540. [247] Stack, J., Surprenant, A., 1991. Dopamine actions on calcium currents, potassium currents and hormone release in rat melanotrophs. J. Physiol. 439, 37–58. [103] Standen, N.B., 1975. Voltage-clamp studies of the calcium inward current in an identified snail neurone: comparison with the sodium inward current. J. Physiol. 249, 253–268. [151] Starke, K., 1971. Influence of a-receptor stimulants on noradrenaline release. Naturwissenschaften 58, 420. [69, 71, 72, 77] Stark, M., Bauer, A.J., Szurszewski, J.H., 1991. Effect of nitric oxide on circular muscle of the canine small intestine. J. Physiol. 444, 743–761. [127] Stark, M.E., Szurszeweski, J.H., 1992. Role of nitric oxide in gastrointestinal and hepatic function and disease. Gastroenterol. 103, 1928–1949. [127] Stark, M.E., Bauer, AJ., Sarr, M.G., Szurszewski, J.H., 1993. Nitric oxide mediates inhibitory nerve input in human and canine jejunum. Gastroenterol. 104, 398–409. [132]
REFERENCES
323
Starke, K., Gothert, M., Kilbinger, H., 1989. Modulation of neurotransmitter release by presynaptic autoreceptors. Physiol Revs. 69, 864–989. [72, 77] Starke, K., Montel, H., Schumann, H.J., 1971. Influence of cocaine and phenoxbenzamine on noradrenaline uptake and release. NaunynSchmiedebergs Arch. Pharmakol. 271, 181– 192. [72] Stern, J.R., Eggleston, L.V., Hems, R., Krebs, H.A., 1949. Accumulation of glutamic acid in isolated brain tissue. Biochem. J. 44, 410–418. [85] Sternini, C., Su, D., Gamp, P.D., Bunnett, N.W., 1995. Cellular sites of expression of the neurokinin-1 receptor in the rat gastrointestinal tract. J. Comp. Neurol. 358, 531–540. [132] Stevens, C.F., 1968. Synaptic physiology. Proc. I.E.E.E. 56, 916–930. [179] Stevens, C.F., Wang, Y., 1993. Reversal of long-term potentiation by inhibitors of haem oxygenase. Nature. 364, 147–149. [130] Stille, G., Lauener, H., Eichenberger, E., 1971. The pharmacology of 8-chloro-11-(4-methyl-1-piperazinyl)-5H-dibenzo(b, e)(1, 4) diazepine (clozapine). Farmaco-Edizione Pratica. 26, 603–625. [100] Stille, G., Lauener, H., Eichenberger, E., Hunziker, J., Schmutz, K., 1965. Arzneim.-Forsch. (Drug. Res.) 15, 841. [100] Stolz, R, 1904. Uber Adrenalin und Alkylaminoacetobrenzcatechin. Ber. Dtsch. Chem. Ges. 37, 4149–4154. [65] Stoof, J.C., Kebabian, J.W., 1981. Opposing roles for D1 and D2 dopamine receptors in efflux of cyclic AMP from rat neostriatum. Nature. 294, 366–368. [95] Storey, D.F., McCulloch, M.W., Standford-Starr, C.A., Rand, M.J., 1981. Conditions required for the inhibitory feedback loop in noradrenergic transmission. Nature. 293, 62–65. [73] Story, D.D., Briley, M.S., Langer, S.Z., 1979. The effects of chemical sympathectomy with 6-hydroxydopamine on alpha-adrenoceptor and muscarinic cholinoceptor binding in rat heart ventricle. Eur. J. Pharmacol. 57, 423–426. [71, 72] Strange, P.G., 1994. Dopamine D4 receptors: curiouser and curiouser. Trends Pharmacol. Sci. 15, 317–319. [102] Stratford, K.J., Jack, J.J., Larkman, A.U., 1997. Calibration of an auto-correlation-based method for determining amplitude histogram reliability and quantal size. J. Physiol. 50, 425–442. [193, 194] Stricker, C., Redman, S.J., 1994. Statistical Models of Synaptic Transmission Evaluated Using the Expectation-Maximization Algorithm. Biophy. J. 67, 656–670. [210] Stricker, C., Field, A.C., Redman, S.J., 1996. Statistical Analysis of amplitude fluctuations in EPSCs evoked in rat CA1 pyramidal neurones in vitro. J. Physiol. 490, 419–441. [197] Stricker, C., Redman, S.J., Daley, D., 1994. Statistical analysis of Synaptic Transmission: Model Discrimination and Confidence Limits. Biophys. J. 67, 532–547. [209] Su, C., Bevan, I.A., 1970. The release of H3 norepinephrine in arterial strips studied by the technique of supervision and transmural stimulation. J. Pharmac. Exp. Ther. 172, 62–71. [72] Subramony, P., Dryer, S.E., 1997. Neuregulins stimulate the functional expression of Ca2+-activated K+ channels in developing chicken parasympathetic neurons. Proc. Nat. Acad. Sci. USA. 94(11), 5934–5938. [279] Sudhof, T.C., 1995. The synaptic vesicle cycle: a cascade of protein-protein interactions. Nature. 375, 645–653. [163, 180] Sudhof, T.C., Rizo, J., 1996. Synaptotagmins: C2-domain proteins that regulate membrane traffic. Neuron. 17, 379–388. [161] Sugiyama, J., Bowen, D.C., Hall, Z.W., 1994. Dystroglycan binds nerve and muscle agrin. Neuron. 13(1), 103–115. [274] Sunahara, R.K., Guan, H.C., O’Dowd, B.F., Seeman, P., Laurier, L.G., Ng, G., George, S.R., Torchia, J., Van Tol, H.H.M, Niznik, H.B., 1991. Cloning of the gene for a human dopamine D5 receptor with higher affinity for dopamine DL Nature. 350, 614–619. [98] Suzue, T., Kaprielian, Z., Patterson, P.H., 1990. A monoclonal antibody that defines rostrocaudal gradients in the mammalian nervous system. Neuron. 5(4), 421–431. [282, 283] Svensson, T.H., Tung, C.S., Grenhoff, J., 1989. The 5-HT2 antagonist ritanserin blocks the effect of pre-frontal cortex inactivation on rat A10 dopamine neurones in vivo. Acta. Physiol. Scand. 136, 497–498. [103] Swammerdam, Jan. 1737. Biblia Naturae. 2 vols. Ed. by Boerhaave. Severinus, Amsterdam, 1737 and 1738. (English translation by Th. Floyd. London, 1758). (Facsimile edition, G.Lindeboom, 1982). [6, 10] Szewczak, S.M., Behar, J., Billett, G., Hillemeier, C., Rhim, B.Y., Biancani, P., 1990. VIP-induced alterations in cAMP and inositol phosphates in the lower esophageal sphincter. Amer. J. Physiol. 259, G239–G244. [129] Takagaki, G., 1996. The dawn of excitatory amino acid research in Japan. The pioneering work by Professor Takashi Hayashi. Neurochem. Internat. 29, 225–229. [79] Takahashi, M., Sarantis, M., Attwell, D., 1996. Postsynaptic glutamate uptake in rat cerebeilar Purkinje cells. J. Physiol. 497, 523–530. [87] Takahashi, T., Momiyama, A., 1993. Different types of calcium channels mediate central synaptic transmission. Nature. 366, 156–158. [155] Takamine, J., 1902. The isolation of the active principle of the suprarenal gland. J. Physiol. 27, Pxxix–xxx. [65] Takamine, J., 1901. The isolation of the active principle of the suprarenal gland. J. Physiol. XXVII, xxix–xxx. [47] Tansey, E.M., 1997. Not committing barbarisms: Sherrington and the Synapse, 1897. Brain Res. Bull. 44, 211–212. [23] Tansey, E.M., 1991. Chemical neurotransmission in the autonomic nervous system: Sir Henry Dale and acetylcholine. Clin. Auton. Res. 1, 63–72. [57] Taraskevich, P.S., Douglas, W.W., 1990. Dopamine (D2) or gamma-aminobutyric acid (GABAB) receptor activation hyperpolarizes rat melanotrophs and pertussis toxin blocks these responses and the accompanying fall in [Ca2+]i. Neurosci. Lett. 112, 205–209. [103] Tarlov, S.R., Langer, S.Z., 1971. The fate of 3 H-norepinephrine released from isolated atria and vas deferens: effect of field stimulation. J. Pharmacol Exper. Therapeut. 179. 186–197. [72]
324
REFERENCES
Taxt, T., 1983. Local and systemic effects of tetrodotoxin on the formation and elimination of synapses in reinnervated adult rat muscle. J. Physiol. 340, 175–194. [242, 252, 254, 255, 258, 259] Taxt, T., Ding, R., Jansen, J.K.S., 1983. A note on the elimination of polyneuronal innervation of skeletal muscles in neonatal rats. Acta. Physiol. Scand. 117, 557–560. [242, 245–247, 252] Tello, J.F., 1907. Degeneration et regeneration des plaques motrices. Trab. Lab. Invest. Biol. Univ. Madrid, vol. V.117– 149. [233, 234] Tello, J.F., 1917. Genesis de las terminaciones nerviosas motrices y sensitivas. Trab. Lab. Invest. Biol. Univ. Madrid. 15, 101– 199. [238–240, 242] Thoenen, H., Hughes, R.A., Sendtner, M., 1993. Trophic support of motoneurons: physiological, pathophysiological, and therapeutic implications. Exp. Neurol. 124(1), 47–55. [277] Thoman, E., Wetzel, A., Levine, S., 1968. Learning in the neonatal rat. Animal Behav. 16, 54–57. [217] Thomas, P., Surprenant, A., Almers, W., 1990. Cytosolic Ca2+, exocytosis, and endocytosis in single melanotrophs of the rat pituitary. Neuron 5, 723–733. [103] Thomas, W.S., Jacob, M.H., O’Dowd, D.K., Smith, M.A., 1995. Agrin gene expression in ciliary ganglion neurons following preganglionic denervation and postganglionic axotomy. Dev. Biol. 168(2), 662–669. [279] Thompson, W., 1978. Reinnervation of partially denervated rat soleus muscle. Acta. Physiol. Scand. 103, 81–91. [242, 254, 257, 258] Thompson, W., 1983a. Synapse elimination in neonatal rat muscle is sensitive to pattern of muscle use. Nature. 302, 614–616. [247, 249, 264] Thompson, W.J., 1983b. Lack of segmental selectivity in elimination of synapses from soleus muscle of newborn rats. J. Physiol . 335, 343–352. [247, 249, 264] Thompson, W.J., 1985. Activity and synapse elimination at the neuromuscular junction. Cell. Mol. Neurobiol. 5, 1–2, 167– 182. [246–248] Thompson, W., Kuffler, D.P., Jansen, J.K.S., 1979. The effect of prolonged, reversible block of nerve impulses on the elimination of polyneuronal innervation of new-born rat skeletal muscle fibers. Neurosci. 4, 271–281. [247] Thompson, W.J., Soileau, L.C., Balice-Gordon, R.J., Sutton, L.A., 1987. Selective innervation of types of fibres in developing rat muscle. J. Exper. Biol 132, 249–263. [262] Thompson, W.J., Sutton, L.A., Riley, D.A., 1984. Fibre type composition of single motor units during synapse elimination in neonatal rat soleus muscle. Nature. 309(5970), 709– 711. [262, 263] Thornbury, K.D., Donaghy, K.M., Peake, J., 1995. Characteristics of the NANC post-stimulus (‘rebound’) contraction of the urinary bladder neck muscle in sheep. Brit. J. Pharmacol 116, 2451–2456. [132, 133] Thornbury, K.D., Ward, S.M., Dalziel, H.H., Carl, A., Westfall, D.P., Sanders, K.M., 1991. Nitric oxide and nitrosocysteine mimic nonadrenergic, noncholinergic hyperpolarization in canine proximal colon. Amer. J. Physiol 261, G553– G557. [132] Tian, M., Jacobson, C., Gee, S.H., Campbell, K.P., Carbonetto, S., Jucker, M., 1996. Dystroglycan in the cerebellum is a laminin alpha 2chain binding protein at the glial-vascular interface and is expressed in Purkinje cells. Europ. J. Neurosci. 8(12), 2739–2747. [279] Timmermans, P.B., van Zwieten, P.A., 1981. Mini-review. The postsynaptic alpha 2-adrenoceptor. J. Autonom. Pharmacol 1, 171–183. [72] Toda, N., Baba, H., Okamura, T., 1990. Role of nitric oxide in non-adrenergic, non-cholinergic nerve-mediated relaxation in dog duodenal longitudinal muscle strips. Jpn. J. Pharmacol 53, 281–284. [128] Tomita, T., 1972. Conductance change during the inhibitory potential in the guinea-pig taenia coli. J. Physiol 225, 693– 703. [118, 127] Torphy, T.J., Fine, C.F., Burman, M., Barnette, M.S., Ormsbee, H.S., 1986. Lower esophageal sphincter relaxation is associated with increased cyclic nucleotide content. Amer. J. Physiol 251, G786–G793. [128] Tottrup, A., Svane, D., Forman, A., 1991. Nitric oxide mediating NANC inhibition in opossum lower esophageal sphincter. Amer. J. Physiol 260, G385–G389. [128, 132] Trendelenburg, P., 1917. Physiologische und pharmakologische Versuche uber die Dunndarm peristaltik. Arch. Expl. Pathol Pharmacol. 81, 55–129. [123] Trifaro, J.M., Fournier, S., Novas, M.L., 1989. The p65 protein is a calmodulin-binding protein present in several types of secretory vesicles. Neurosci. 29, 1–8. [160] Trischmann, U., Isenberg, G., 1989. Ca2+-activated K+ channels contribute to the resting potential of vascular myocytes. Pflugers Archiv. Eur. J. Physiol. 414, AS 184. [130] Trischmann, U., Klockner, U., Isenberg, H.G., Utz, J., Ullrich, V., 1991. Carbon monoxide inhibits depolarization-induced Ca rise and increases cyclic GMP in visceral smooth muscle cells. Biochem. Pharmacol 41, 237–241. [130] Tsim, K.W., Ruegg, M.A., Escher, G., Kroger, S., McMahan, U.J., 1992. cDNA that encodes active agrin. Neuron. 8(4), 677–689. [273, 279] Turner, T.J., Dunlap, K., 1995. Pharmacological characterization of presynaptic calcium channels using subsecond biochemical measurements of synaptosomal neurosecretion. Neuropharmacol 34, 1469–1478. [155] Twarog, B.M., Page, I.H., 1953. Serotonin content of some mammalian tissues and urine and method for its determination. Am. J. Physiol 175, 157–161. [92] Udenfriend, S., 1950. Identification of gamma-aminobutyric acid in brain by isotope derivative method. J. Biol Chem. 187, 65–69. [79] Ugedo, L., Grenhoff, J., Svensson, T.H., 1989. Ritanserin, a 5-HT2 receptor antagonist, activates midbrain dopamine neurons by blocking serotonergic inhibition. Psychopharmacol 98. 45–50. [103] Ullrich, B., Li, C., Zhang, J.Z., McMahon, H., Anderson, R.G., Geppert, M., Sudhof, T.C., 1994. Functional properties of multiple synaptotagmins in brain. Neuron. 13, 1281– 1291. [162]
REFERENCES
325
Usdin, T.B., Fischbach, G.D., 1986. Purification and characterization of a polypeptide from chick brain that promotes the accumulation of acetylcholine receptors in chick myotubes. J. Cell Biol 103(2), 493–507. [274] Utz, J., Ullrich, V., 1991. Carbon monoxide relaxes ileal smooth muscle through activation of guanylate cyclase. Biochem. Pharmacol 41, 1195–1201. [130] Vahabzadeh, A., Fillenz, M., 1992. Studies on the origin of rat hippocampal dihydroxyphenolacetic acid using micro-dialysis. Neurosci. Lett. 136, 51–55. [102] Valentijn, J.A., Louiset, E., Vaudry, H., Cazin, L., 1991. Dopamine regulates the electrical activity of frog melanotrophs through a G protein-mediated mechanism. Neurosci. 44, 85–95. [103] Valenzuela, D.M., Stitt, T.N., DiStefano, P.S., Rojas, E., Mattsson, K., Compton, D.L., Nunez, L., Park, J.S., Stark, J.L., Gies, D.R., Thomas, S., LeBeau, M.M., Fernald, A.A., Copeland, N.G., Jenkins, N.A., Burden, S.J., Glass, D.J., Yancopoulos, G.D., 1995. Receptor tyrosine kinase specific for the skeletal muscle lineage: expression in embryonic muscle, at the neuromuscular junction, and after injury. Neuron. 15(3), 573–584. [274] Van Braam-Houckgeest, G., 1874. Zweite Miltheilung uber Magen-und Darmperistaltik. Pflugers Arch. VIII, 163–170. [125] Van der Kloot, W., 1988. The kinetics of quantal releases during end-plate currents at the frog neuromuscular junction. J. Physiol. 402, 605–626. [185] Van Der Kloot, W.G., Robbins, J., Cooke, I., 1958. Blocking by picrotoxin of peripheral inhibition in crayfish. Science. 127, 521–522. [79] Van Essen, D., Jansen, J.K.S., 1974. Reinnervation of the rat diaphragm during perfusion with a-bungarotoxin. Acta. Physiol. Scand. 91, 571–573. [242, 262] Van Essen, D.C., Gordon, H., Soha, J.M., Fraser, S.E., 1990. Synaptic dynamics at the neuromuscular junction: mechanisms and models. J. Neurobiol. 21(1), 223–249. [246] Van Harreveld, A., 1959. Compounds in brain extracts causing spreading depression of cerebral cortical activity and contraction of crustacean muscle. J. Neurochem. 3, 300–315. [79] van Leeuwenhoek, A., 1685. ‘An extract of a letter from Mr. Anthony van Leeuwenhoek, Fellow of the Royal Society, concerning the parts of the brain of several animals’. Philos. Trans. Roy. Soc. Lond. 15, 883–895. [8, 11] van Leeuwenhoek, A., 1717. Epistolae physiologicae super compluribus naturae arcanis. (Letter No. 32 in Latin addressed to Domino Abrahamo van Bleiswyk. March 2, 1717). Beman, Delft, 1719; 1717. [8] Van Tol, H.H., Bunzow, J.R., Guan, H.C., Sunahara, R.K., Seeman, P., Niznik, H.B., Civelli, O., 1991. Cloning of the gene for a human dopamine D4 receptor with high affinity for the antipsychotic clozapine. Nature. 350, 610– 614. [98, 99, 101] Van Tol, H.H., Wu, C.M., Guan, H.C., Ohara, K., Bunzow, J.R., Civelli. O., Kennedy, J., Seeman, P., Niznik, H.B., Jovanovic, V, 1992. Multiple dopamine D4 receptor variants in the human population. Nature. 358, 149–162. [99] Vautrin, J., 1986. Subunits in quantal transmission at the mouse neuromuscular junction: tests of peak intervals in amplitude distributions. J. Theoret. Biol. 120(3), 363–370. [190, 389] Veach, H.O., 1925. Studies on the innervation of smooth muscle. I. Vagus effects on the lower end of the esophagus, cardia and stomach of the cat and the stomach and lung of the turtle in relation to Wedensky inhibition. Amer. J. Physiol. 71, 229–264. [113] Vogt, M., 1954. The concentration of sympathin in different parts of the central nervous system under normal conditions and after the administration of drugs. J. Physiol. 123, 451– 481. [92] Volta, A., 1918a. Le opere di Alessandro Volta (edizione nazionale). Vol. I. Milano: Hoepli. [12] Volta, A., 1918b. Le opere di Allesandro Volta (edizione nazionale). Vol. II, Milano, Hospli. [12] von Euler, U.S., 1946. A specific sympathomimetic ergone in adrenergic nerve fibers (Sympathin) and its relations to adrenaline and nor adrenaline. Acta. Physiol. Scand. 12, 73–97. [38, 66, 67] von Euler, U.S., Hillarp, N.A., 1956. Evidence for the presence of noradrenaline in submicroscopic structures of adrenergic axons. Nature. 177, 44–45. [40] von Humboldt, A., 1797. ‘Versuche über die gereiztle Muskel undNervenfasser. Rottmann, Posen, & Decker, Berlin. [9, 12] Voyvodic, J.T., 1989. Peripheral target regulation of dendritic geometry in the rat superior cervical ganglion. J. Neurosci. 9(6), 1997–2010. [277] Vulpian, E.F., 1866. Lecons sur la Physiologie generale et comparee du Systeme nerveux. (Bailliere, Paris), pp. 920. [45] Wagner, R., 1847. Fortgesetzte Untersuchungen über die Verbreitung des Nerven im Elektrischen Organ des Zitterochens. Nachr. von d. G.A. Univers, u.d. Kgl. Gesellsch. d. Wissensch. zu Göttingen. Nov. [15] Waldeyer, H.W., 1891. Uber einige neuere Forschungen im Gebiete der Anatomie des Centralnerven Systems. Deutsche med. Wochenschrift 17, 1213–1218. [18] Wallace, B.G., 1989. Agrin-induced specializations contain cytoplasmic, membrane, and extracellular matrix-associated components of the postsynaptic apparatus. J. Neurosci. 9(4), 1294–1302. [271] Wallace, B.G., 1994. Staurosporine inhibits agrin-induced acetylcholine receptor phosphorylation and aggregation. J. Cell Biol. 125(3), 661–668. [273] Wallace, B.G., 1995. Regulation of the interaction of nicotinic acetylcholine receptors with the cytoskeleton by agrin-activated protein tyrosine kinase. J. Cell Biol. 128(6), 1121– 1129. [274] Wallace, B.G., Qu, Z., Huganir, R.L., 1991. Agrin induces phosphorylation of the nicotinic acetylcholine receptor. Neuron. 6(6), 869–878. [273] Walmsley, B., Edwards, F.R., Tracey, D.J., 1988. Nonuniform release probabilities underlie quantal synaptic transmission at a mammalia excitatory central synapse. J. Neurophysiol. 60(3), 889–908. [177, 189, 190]
326
REFERENCES
Wang, T., Xie, K., Lu, B., 1995. Neurotrophins promote maturation of developing neuromuscular synapses. J. Neurosci. 15(7 Pt 1), 4796–4805. [277] Ward, S.M., Dalziel, H.H., Bradley, M.E., Buxton, I.L., Keef, K., Westfall, D.P., Sanders, K.M., 1992a. Involvement of cyclic GMP in nonadrenergic, non-cholinergic inhibitory neurotransmission in dog proximal colon. Brit. J. Pharmacol. 107, 1075–1082. [128] Ward, S.M., McKeen, E.S., Sanders, K.M., 1992b. Role of nitric oxide in non-adrenergic, non-cholinergic inhibitory junction potentials in canine ileocolonic sphincter. Brit. J. Pharmacol. 105, 776–782. [132] Ward, S.M., Xue, C., Shuttleworth, C.W., Bredt, D.S., Snyder, S.H., Sanders, K.M., 1992c. NADPH diaphorase and nitric oxide synthase colocalization in enteric neurons of canine proximal colon. Amer. J. Physiol. 263, G277–G284. [127, 128] Waterman, S.A., 1997. Role of N-, P- and Q-type voltage-gated calcium channels in transmitter release from sympathetic neurones in the mouse isolated vas deferens. Br. J. Pharmacol. 120, 393–398. [158] Waterman, S.A., Costa, M., 1994. The role of enteric inhibitory motoneurons in peristalsis in the isolated guinea-pig small intestine . J. Physiol. 477, 459–468. [127] Watkins, J.C., Evans, R.H., 1981. Excitatory amino acid transmitters. Ann. Rev. Pharmacol. & Tox. 21, 165–204. [218] Watson, M.J., Bywater, R.A.R., Taylor, G.S., Lang, R.J., 1996. Effects of nitric oxide (no) and no donors on the membrane conductance of circular smooth muscle cells of the guineapig proximal colon. Br. J.Pharmacol. 118, 1605–1614. [131] Weber, E., 1846. Muskelbewegung in Handworterbuch der Physiologie mit Ruchsicht auf physiologische Pathologie. Ed. R.Wagner. 4 vols. 3: 28–29, Braunschweig: Vieweg. [14] Weiner, D.M., Levey, A.I., Sunahara, R.K., Niznik, H.B., O’Dowd, B.F., Seeman, P., Brann, M.R., 1991. D1 and D2 dopamine receptor mRNA in rat brain. Proc. Nat. Acad. Sci. USA. 88, 1859–1863. [101] Weinshank, R.L., Adham, N., Macchi, M., Olsen, M.A., Branchek, T.A., Hartig, P.R., 1991. Molecular cloning and characterization of a high affinity dopamine receptor (D1 beta) and its pseudogene. J. Biol. Chem. 266, 22427–22435. [98] Welch, A.D., Henderson, V.E., 1934. A comparative study of hydrastine, bicuculline and adlumine. J. Pharmacol. Exp. Ther. 51, 482–491. [85] Werle, M.J., Herrera, A.A., 1988. Synaptic competition and the elimination of polyneuronal innervation following reinnervation of adult frog sartorius muscles. J. Neurobiol. 19(5), 465–481. [254] Werman, R., 1965. The specificity of molecular processes involved in neural transmission. J. Theoret. Biol. 9, 471– 477. [82] Werman, R., Davidoff, R.A., Aprison, M.H., 1967. Inhibition of motoneurones by iontophoresis of glycine. Nature. 214, 681–683. [81–84] Wernig, A., 1975. Estimates of statistical release parameters from crayfish and frog neuromuscular junctions. J. Physiol 244(1), 207–221. [173] Wessler, I., 1992. Acetylcholine at motor nerves: storage, release and presynaptic modulation by autoreceptors and adrenoceptors. Internat. Rev. Neurobiol. 4, 283–384. [67] Wessler, I., Dooley, D.J., Werhand, J., Schlemmer, F., 1990. Differential effects of calcium channel antagonists (omega-conotoxin GVIA, nifedipine, verapamil) on the electricallyevoked release of [3H]acetylcholine from the my enteric plexus, phrenic nerve and neocortex of rats. Naunyn-Schmied. Arch. Pharmacol. 341, 288–294. [155] Wheeler, D.B., Randall, A., Tsien, R.W., 1994a. Roles of N-type and Q-type Ca2+ channels in supporting hippocampal synaptic transmission. Science. 264, 107–111. [153, 155] Wheeler, D.B., Sather, W.A., Randall, A., Tsien, R.W., 1994b. Distinctive properties of a neuronal calcium channel and its contribution to excitatory synaptic transmission in the central nervous system. Adv. Second Messenger & Phosphoprotein Res. 29, 155–171. [153, 155] Whitaker, L., 1914. On the Poisson Law of Small Numbers. Biometrika, 10, 36–71. [170] White, N.P., 1996. Plato’s metaphysical epistemology. In: The Cambridge Companion to Plato . Kraut, R., (ed.). pp. 277– 310; 1996. Cambridge University Press. [1] Wigston, D.J., Sanes, J.R., 1985. Selective reinnervation of intercostal muscles transplanted from different segmental levels to a common site. J. Neurosci. 5(5), 1208–1221. [264, 278] Wigstrom, H., Gustafsson, B., 1985a. On long-lasting potentiation in the hippocampus: a proposed mechanism for its dependence on coincident pre- and postsynaptic activity. Acta. Physiol. Scand. 123, 519–522. [221] Wigstrom, H., Gustafsson, B., 1985b. Facilitation of hippocampal long-lasting potentiation by GABA antagonists. Acta. Physiol. Scand. 125, 159–172. [221] Wigstrom, H., Gustafsson, B., Huang, Y.Y., 1986a. Mode of action of excitatory amino acid receptor antagonists on hippocampal longlasting potentiation. Neurosci. 17, 1105– 1115. [218, 222] Wigstrom, H., Gustafsson, B., Huang, Y.Y., Abraham, W.C., 1986b. Hippocampal long-term potentiation is induced by pairing single afferent volleys with intracellularly injected depolarizing current pulses. Acta. Physiol. Scand. 126, 317– 319. [221, 222] Williams, J.H., Errington, M.L., Lynch, M.A., Bliss, T.V., 1989. Arachidonic acid induces a long-term activity-dependent enhancement of synaptic transmission in the hippocampus. Nature. 341, 739–742. [225] Williams, J.T., Henderson, G., North, R.A., 1985. Characterisation of alpha 2-adrenoceptors which increase potassium conductance in rat locus coeruleus neurones. Neurosci. 14, 95–101. [75, 77] Williams, M.E., Feldman, D.H., McCue, A.F., Brenner, R., Velicelebi, G., Ellis, S.B., Harpold, M.M., 1992. Structure and functional expression of alpha 1, alpha 2, and beta subunits of a novel human neuronal calcium channel subtype. Neuron. 8, 71–84. [155] Williams, P.E., Goldspink, G., 1971 Longitudinal growth of striated muscle fibres. J. Cell Sci. 9, 751–767. [238] Williams, P.E., Goldspink, G., 1973. The effect of immobilization on the longitudinal growth of striated muscle fibres. J.Anat. 116, 45–55. [238]
REFERENCES
327
Wilson, D.A., 1984. A comparison of the postnatal development of post-activation potentiation in the neocortex and dentate gyrus of the rat. Brain Res. 318, 61–68. [216, 217] Wilson, D.A., Racine, R.J., 1983. The postnatal development of post-activation potentiation in the rat neocortex. Brain Res. 283, 271–276. [217] Wolf, G., Wurdig, S., Schunzel, G., 1992. Nitric oxide synthase in rat brain is predominantly located at neuronal endoplasmic reticulum; an electron microscopic demonstration of NADPHdiaphorase activity. Neurosci. Letts. 147, 63–66. [132] Wong, D.F., Wagner, H.N. Jr, Tune, L.E., Dannals, R.F., Pearlson, G.D., Links, J.M., Tamminga, C.A., Broussolle, E.P., Ravert, H.T., Wilson, A.A., Thomas Young, J.K., Malat, J., Williams, J.A., O’Tuama, L.A., Snyder, S.H., Kuhar, M.J., Gjedde, A., 1986. Positron emission tomography reveals elevated D2 dopamine receptors in drug-naive schizophrenics. Sci. 234, 1558–1563. [99] Woodward, J.J., Rezazadeh, S.M., Leslie, S.W., 1988. Differential sensitivity of synaptosomal calcium entry and endogenous dopamine release to omega-conotoxin. Brain Res. 475, 141–145. [155] Wright, C.E., Angus, J.A., 1996. Effects of N-, P- and Q-type neuronal calcium channel antagonists on mammalian peripheral neurotransmission. Br. J. Pharmacol. 119, 49–56. [158] Xue, C., Pollock, J., Schmidt, H.H., Ward, S.M., Sanders, K.M., 1994. Expression of nitric oxide synthase immunoreactivity by interstitial cells of the canine proximal colon. J. Auton. Nerv. Sys. 49, 1–14. [127] Young, H.M., McConalogue, K., Furness, J.B., Vente, J., 1993. Nitric oxide targets in the guinea-pig intestine identified by induction of cyclic GMP immunoreactivity. Neurosci. 55, 583–596. [127] Zagorodnyuk, V, Maggi, C.A., 1994. Electrophysiological evidence for different release mechanism of ATP and NO as inhibitory NANC transmitters in guinea-pig colon. Brit. J. Pharmacol. 112, 1077–1082. [130] Zelena, J., Jirmanova, I., 1973. Ultrastructure of chicken slow muscle after nerve cross union. Exp. Neurol. 38, 272–285. [262] Zhang, J.F., Randall, A.D., Ellinor, P.T., Horne,W.A., Sather, W.A., Tanabe, T., Schwarz, T.L., Tsien, R.W., 1993. Distinctive pharmacology and kinetics of cloned neuronal Ca2+ channels and their possible counterparts in mammalian CNS neurons. Neuropharmacol. 32, 1075–1088. [157] Zhu, X., Lai, C., Thomas, S., Burden, S.J., 1995. Neuregulin receptors, erbB3 and erbB4, are localized at neuromuscular synapses. EMBO J. 14(23), 5842–5848. [274] Zhuo, M., Small, S.A., Kandel, E.R., Hawkins, R.D., 1993. Nitric oxide and carbon monoxide produce activity-dependent long-term synaptic enhancement in hippocampus. Sci. 260, 1946–1950. [130] Ziskind-Conhaim, L., Dennis, M.J., 1981. Development of rat neuromuscular junctions in organ culture. Develop. Biol. 85, 243–251. [242] Zoran, M.J., Doyle, R.T., Haydon, P.G., 1990. Target-dependent induction of secretory capabilities in an identified motoneuron during synaptogenesis. Develop. Biol. 138(1), 202–213. [276] Zoran, M.J., Doyle, R.T., Haydon, P.G., 1991. Target contact regulates the calcium responsiveness of the secretory machinery during synaptogenesis. Neuron. 6(1), 145–151. [276] Zoran, M.J., Funte, L.R., Kater, S.B., Haydon, P.G., 1993. Neuron-muscle contact changes presynaptic resting calcium set-point. Develop. Biol. 158, 163–171. [276] Zucker, R.S., 1973. Changes in the statistics of transmitter release during facilitation. J. Physiol. 229(3), 787–810. [170, 173] Zucker, R.S., Fogelson, A.L., 1986. Relationship between transmitter release and presynaptic calcium influx when calcium enters through discrete channels. Proc. Natn. Acad. Sci. U.S.A. 83, 3032–3036. [163]
Illustration Acknowledgements
Fig. 1.1(A) Reprinted with permission of Bayerische Staatsbibliothek Fig. 1.1(B) Reprinted with permission of the Biblioteca Nazionale Centrale, Florence. Fig. 1.1(C) Reprinted with permission of the Incunabula Collection of the National Library of Medicine, Bethesda. Fig. 1.1(D) Courtesy of the Royal Collection © 2000, Her Majesty Queen Elizabeth II Fig. 1.2(B) Reproduced with kind permission of the Director of the Vatican Library Fig. 1.2(C) Copyright (1996) Elsevier Pergamon Fig. 1.2(D) Reproduced with permission of the University of Bologna Library. Fig. 1.3(A) Reproduced with kind permission of Harvard University Press. Fig. 1.3(B), (C) Reproduced by kind permission of Raven Press. Fig. 1.3(D) Reproduced with permission of the American Physiological Society. Fig. 1.4(A) Reproduced with kind permission of the Boerhaave Museum. Fig. 1.4(E), (F) Reproduced with permission of Raven Press. Fig. 1.5(A) Reproduced with permission of the Boerhaave Museum. Fig. 2.2(B) This figure was first published in the BMJ (Dale, 1934) and is reproduced by permission of the BMJ. Fig. 2.3(A), (C) © Springer—Verlag Fig. 2.4(B) Reprinted with permission of the American Physiological Society. Fig. 2.5(A) Reprinted with permission of Oxford University Press. Fig. 2.5(B) Reprinted with permission for Nature, Brooks and Eccles (1947), Copyright (1947) Macmillan Magazines Limited. Fig. 2.6(B)b © Springer—Verlag Fig. 2.7(B) © Springer—Verlag Fig. 2.7(C)b © Springer—Verlag Fig. 2.7(D) Reprinted with permission of the American Physiological Society. Fig. 2.8(D) By Copyright permission of the Rockefeller Press. Fig. 2.8(E) Reprinted with permission of Cambridge University Press. Fig. 2.8(F) Copyright (1956) With permission from Elsevier Science. Fig. 2.10(A), (B), (C) Reprinted with permission of Liverpool University Press. Fig. 2.10(D) © Springer—Verlag Fig. 3.1(A) Reprinted with permission form Oxford University Press. Fig. 3.1(B) By permission of the President and Council of the Royal Society Fig. 3.1(C) By permission of the President and Council of the Royal Society Fig. 3.1(D) Reprinted with permission form the National Library of Medicine Fig. 3.3(A) © Springer—Verlag Fig. 3.5(A), (B), (C) Reprinted with permission of the American Society for Pharmacology and Experimental Therapeutics. Fig. 3.5(D), (E), (F), (G), (H), (I) © Springer—Verlag Fig. 3.8(A) Reprinted with permission from Nature, Katz and Miledi (1970), Copyright (1970) Macmillan Magazines Limited. Fig. 3.8(B) Reprinted with permission from Nature, Katz and Miledi (1971), Copyright (1971) Macmillan Magazines Limited. Fig. 3.8(C), (D) Reprinted with permission from Nature, Neher and Sakmann (1976), Copyright (1976) Macmillan Magazines Limited. Fig. 4.2(D) © Springer—Verlag Fig. 4.3(B) Reprinted with permission from Nature, Koelle (1961), Copyright (1961) Macmillan Magazines Limited. Fig. 4.3(E) © Springer—Verlag Fig. 4.4(B) © Springer—Verlag
ILLUSTRATION ACKNOWLEDGEMENTS
329
Fig. 4.4(F) Reprinted with permission from Nature, Langer et al., (1977), Copyright (1977) Macmillan Magazines Limited. Fig. 4.5(A), (B), (F) © Springer—Verlag Fig. 4.6 (E) © Springer—Verlag Fig. 5.1(A) Reprinted with permission from Purpura et al., 1957. Copyright 1957 American Association for the Advancement of Science. Fig. 5.1(C), (D) Reprinted with permission from Nature, Curtis et al., (1970), Copyright (1970) Macmillan Magazines Limited. Fig. 5.2 Copyright Lippincolt Williams & Wilkins. Fig. 5.3 Copyright Lippincolt Williams & Wilkins. Fig. 5.4(A), (B) Reprinted with permission from Nature, Werman et al., (1967), Copyright (1967) Macmillan Magazines Limited. Fig. 5.4(C) Reprinted with permission from Nature, Curtis et al., (1967), Copyright (1967) Macmillan Magazines Limited. Fig. 5.4(D) Copyright Springer-Verlag. Fig. 5.5(A) Copyright (1973) with permission from Elsevier Science. Fig. 5.5(B) Copyright (1972) with permission from Elsevier Science. Fig. 5.5(C) Reprinted with permission from Nature, Biscoe et al., (1977), Copyright (1977) Macmillan Magazines Limited. Fig. 5.5(D) Copyright (1978) with permission from Elsevier Science. Fig. 5.5(E) Copyright (1979) with permission from Elsevier Science. Fig. 5.6(B) Copyright (1982) with permission from Elsevier Science. Fig. 5.6(C) Reprinted with permission from Nature, Oliverman et al., (1984), Copyright (1984) Macmillan Magazines Limited. Fig. 5.6(D) With permission from Elsevier Science. Fig. 6.1(A) © 1963 Munskgaard International Publishers Ltd., Copenhagen, Denmark. Fig. 6.1(B) Reprinted from Anden et al., 1964, with permission from Elsevier Science. Fig. 6.4(B) Reprinted from Seeman, 1992, with permission from Elsevier Science. Fig. 6.4 (C), (D), (E) Reprinted from Seeman and van Tol, 1994, with permission from Elsevier Science. Fig. 6.5 (A), (B) Reprinted with permission from Nature, Kebabian and Caine (1979), Copyright (1979) Macmillan Magazines Limited. Fig. 6.6 With permission, from Annual Reviews of Neuroscience, Volume 16, © 1993, by Annual Reviews Inc. Fig. 6.7 © John Wiley and Sons Inc. Fig. 6.8 Reprinted with Farde, 1996, with permission from Elsevier Science. Fig. 6.9 © Springer—Verlag Fig. 7.2(A) This figure was first published in the BMJ (Dale, 1934) and is reproduced by permission of the BMJ Fig. 7.6(B) Reprinted from the Int. J. Neuropharmacology (Burnstock, Campbell, Bennett and Holman, 1964). Copyright (1964) with kind permission form Elsevier Science Ltd, The Boulevard, Langford Lane, Kidlington OX5 1GB UK. Fig. 7.7(A), (B) Reprinted with permission from Nature, Burnstock et al., (1963), Copyright (1963) Macmillan Magazines Limited. Fig. 7.7(C) © Springer—Verlag Fig. 7.8(C) Reprinted with permission from Nature, Anderson and Eccles., (1962), Copyright (1962) Macmillan Magazines Limited. Fig. 7.8(D) By Copyright permission of the Rockefeller University Press. Fig. 7.9(B), (C) © Plenum Press Fig. 7.10(D) © Society for Neuroscience Fig. 7.12(B) By Copyright permission of the Rockefeller University Press. Fig. 7.14(A) Copyright (1993) National Academy of Sciences, U.S.A. Fig. 8.1(D) Reproduced with permission from Mann et al., 1939, Biochemical Journal, 33, 822–835. © the Biochemical Society. Fig. 8.3(B), (C) By Copyright permission of the Rockefeller University Press. Fig. 8.4(B) © Springer—Verlag Fig. 8.5(A) By permission of the President and Council of the Royal Society Fig. 8.6(B) Reprinted with permission from Llinas et al., 1972. Copyright 1972 American Association for the Advancement of Science. Fig. 8.8(B) Reprinted with permission from Nature, Akarke et al., (1978), Copyright (1978) Macmillan Magazines Limited. Fig. 8.9(A) © Springer—Verlag Fig. 8.9(C) Reprinted with permission from Nature, Brown et al., (1982), Copyright (1982) Macmillan Magazines Limited.
330
ILLUSTRATION ACKNOWLEDGEMENTS
Fig. 8.9(D) Reprinted with permission from Nature, Reuter et al, (1982), Copyright (1982) Macmillan Magazines Limited. Fig. 8.10(A) Reprinted with permission from Nature, Nowycky et al., (1985a), Copyright (1985) Macmillan Magazines Limited. Fig. 8.10(B) Reprinted with permission from Nature, Nilius et al., (1985), Copyright (1985) Macmillan Magazines Limited. Fig. 8.11(E) Copyright (1994) Elsevier Pergamon Fig. 8.13(B) © Cell Press (1996) Fig. 8.13(C) © Churchill Livingstone Inc (1996). Fig. 8.13(D) Reprinted with permission from Llinas et al., 1992b. Copyright 1992 American Association for the Advancement of Science. Fig. 9.4 (B), (C) By Copyright permission of the Rockfeller University Press. Fig. 9.5(A) © 1968 IEEE. Fig. 9.6(A), (B) By permission of the President and Council of the Royal Society. Fig. 9.9(A) By permission of the Biophysical Society. Fig. 10.1(A), (B) These figures were first published in the BMJ (Scoville and Milner, 1957) and are reproduced by permission of the BMJ. Fig. 10.3(A), (B) Reprinted from Racine et al., (1983), with permission from Elsevier Science. Fig. 10.4(A) Copyright © (1979) by the American Psychological Association. Reprinted with permission. Fig. 10.5 Reprinted from McNaughton et al., (1978), with permission from Elsevier Science. Fig. 10.6(B) Reprinted from Harris et al., (1984), with permission from Elsevier Science. Fig. 10.6(C) Reprinted from Harris and Cotman (1986), with permission for Elsevier Science. Fig. 10.6(D) Reprinted with permission from Nature, Nowak et al., (1984), Copyright (1984) Macmillan Magazines Limited. Fig. 10.7(A) Reprinted with permission from Nature, Lynch et al., (1983), Copyright (1983) Macmillan Magazines Limited. Fig. 10.7(B) Reprinted with permission from Malenka et al., 1988. Copyright 1988 American Association for the Advancement of Science. Fig. 10.8(A) Reprinted with permission from Nature, Malinow and Miller, 1986, Copyright (1986) Macmillan Magazines Limited. Fig. 10.9(B), (C) Reprinted with permission from Nature, Dolphin et al., 1982, Copyright (1982) Macmillan Magazines Limited. Fig. 10.10(A), (B) Reprinted with permission from Duffy et al., 1981. Copyright 1981 American Association for the Advancement of Science. Fig. 10.11(A) Reprinted with permission from Lynch and Baudry, 1984. Copyright 1984 American Association for the Advancement of Science. Fig. 10.11(B) Reprinted with permission from Nature, Davies et al., 1989, Copyright (1989) Macmillan Magazines Limited. Fig. 10.12 Reprinted from Bar et al., 1980, with permission from Elsevier Science. Fig. 10.13(A) Reprinted with permission from Malinow et al., 1989. Copyright 1989 American Association for the Advancement of Science. Fig. 10.13(B) Reprinted with permission from Nature, Malenka et al., 1989. Copyright (1989) Macmillan Magazines Limited. Fig. 10.14(A) Reprinted with permission from Nature, Morris et al., 1986. Copyright (1986) Macmillan Magazines Limited. Fig. 10.14(B) Reprinted from Roman et al., 1987, with permission from Elsevier Science. Fig. 10.14(C)Reprinted with permission from Nature, Castro et al., 1989. Copyright (1989) Macmillan Magazines Limited. Fig. 10.15 Reprinted with permission from Brown and McAfee, 1982. Copyright 1982 American Association for the Advancement of Science. Fig. 11.2(A), (B) Reprinted with permission from Cambridge University Press Fig. 11.2(C) Reprinted with permission from Cold Spring Harbor Laboratory Press. Fig. 11.3 (C), (D) Reprinted with permission from Cold Spring Harbor Laboratory Press. Fig. 11.6 By permission of the President and Council of the Royal Society. Fig. 11.7(A) By permission of the President and Council of the Royal Society. Fig. 11.11(A) By permission of the President and Council of the Royal Society. Fig. 11.20(A) By permission of the President and Council of the Royal Society. Fig. 11.22(A), (B), (C), (D) By Copyright permission of the Rockfeller University Press. Fig. 11.23(A), (B), (C) © Cell Press, 1996 Fig. 11.25(A), (B) © Cell Press, 1995 Fig. 11.25(C), (D) © (1995) Society for Neuroscience.
ILLUSTRATION ACKNOWLEDGEMENTS
Epilogue Fig. Reprinted with permission of McGill University Press.
331
Acknowledgements
Chapters in this book have been published as reviews in the following journals. Chap 1:
Chap 2:
Chap 3:
Chap 4:
Chap 5:
Chap 6:
Chap 7:
Chap 8:
Chap 9:
Chap 10:
Brain Research Bulletin 50(2): 95±118 (1999) The early history of the synapse: From Plato to Sherrington. M.R.Bennett Copyright (1999) Reprinted with permission from Elsevier Science Advances in Second Messenger and Phosphoprotein Research, Molecular and Cellular Mechanisms of Transmitter Release 29:1±29 (1994). Eds L.Stjärne, P.Greengard, S.Grillner, T Hökfelt, D.Ottoson. Raven Press. The concept of neurotransmitter release. M.R.Bennett Copyright (1994) Reprinted with permission from Lippincott, Williams & Wilkins Neuropharmacology 39(4): 523±546 (2000) The concept of transmitter receptors: 100 years on. M.R.Bennett Copyright (2000) Reprinted with permission Elsevier Science Clinical Autonomic Research 9(3): 145±159 (1999) One hundred years of adrenaline: the discovery of autoreceptors. M.R.Bennett Copyright (1999) Reprinted with permission from Lippincott, Williams & Wilkins Neurochemistry International 35(4): 269±280 (1999) Forty years of amino acid transmission in the brain. M.R.Bennett and V.J.Balcar Copyright (1999) Reprinted with permission from Elsevier Science Journal of Psychopharmacology 12(3): 289±304 (1998) Manoaminergic synapses and schizophrenia: 45 years of neuroleptics. M.R.Bennett Copyright (1998) British Association for Psychopharmacology and reprinted by permission of Sage Publications Ltd, London, UK Progress in Neurobiology 52(3): 159±195 (1997) Non-adrenergic non-cholinergic (NANC) transmission to smooth muscle: 35 years on. M.R.Bennett Copyright (1997) Reprinted with permission from Elsevier Science Progress in Neurobiology 59(3): 2443–277 (1999) The concept of a calcium sensor in transmitter release. M.R.Bennett Copyright (1999) Reprinted with permission from Elsevier Science Progress in Neurobiology 60(6): 545±606 (2000) Statistics of transmitter release at nerve terminals. M.R.Bennett and J.L.Kearns Copyright (2000) Reprinted with permission from Elsevier Science Progress in Neurobiology 60(2): 109±137 (2000) The concept of long term potentiation of transmission at synapses. M.R.Bennett
333
Chap 11:
Copyright (2000) Reprinted with permission from Elsevier Science Progress in Neurobiology 57(2): 225–287 (1999) Synapse formation molecules in muscle and autonomic ganglia: the dual constraint hypothesis. M.R.Bennett Copyright (1999) Reprinted with permission from Elsevier Science
Index
acetylcholine appears on splanchnic nerve stimulation, 58 appears on stimulation of nerves to ganglia, 58 discovered in the spleen, 57 esterase in blood, 57 has complementary effects to adrenaline, 54 identified in ergot, 108 in the spleen, 108 original source, 27 receptor redistribution, 276 receptor toxins, 63 release dependent on calcium, 137 release from ganglia dependent on calcium, 135 released from stimulated muscle, 59 as the parasympathetic transmitter, 108 action currents, 12 action potential conducts into muscle, 18 not due to sodium in smooth muscle, 145 not due to sodium in crab muscle, 141 velocity measured, 14 active zone formation, 275 adrenaline action not dependent on sympathetic nerves, 50 action on stomach depends on its tone, 111 acts directly on muscle, 50 acts on muscle like sympathetic nerves, 47 and acetylcholine as the only transmitters, 34 and sympathin compared, 65 released at nerve terminals, 27, 63, 107 released by nerves in toad’s heart, 67 released by sympathetic nerves suggested, 27 synthesized, 65 adrenergic blocking drugs, increase overflow of noradrenaline, 67 agrin hypothesis and synapse formation, 273 in ganglia, 279 isoforms found in the nervous system, 273 as a synapse formation molecule, 271 alkaloid effects on muscle, 54 alpha—adrenoceptors, 71 activate G proteins, 77 exist on central neurons, 73 can increase potassium conductance, 75 and beta adrenergic receptors, 67 inhibit noradrenaline release, 71 alpha—adrenoceptors antagonists can enhance noradrenaline overflow, 71
can potentiate overflow on denervation, 72 amino acid excitants and alpha—decarboxylation, 81 AMPA and kainate receptors, 88 animal current discovered, 12 animal spirit particles conduct from brain to muscle, 7 particles transmitted into muscles, 4 is a corpuscular juice, 8 becomes electricity, 12 antagonism between curare and nicotine, 53 antipsychotic drugs depolarization block of dopamine neurons, 104 ARIA, a neuregulin in acetylcholine receptor regulation, 274 astatic galvanometer invented, 12 ATP as an IJP transmitter, 129 atropine not block vagus effects on intestine, 108 resistance of vagal excitation to stomach, 108 resistance of intestine to nerve stimulation, 111 resistant excitatory responses in muscle, 118 autocorrelation (AC) assessing components in histograms, 190 scores in tests of peaks in quantal data, 194 autonomic nervous system defined, 47, 107 autoreceptors concept, 67, 119 activated endogenously, 73 decrease calcium influx, 73 distinguished from uptake pumps, 69 in the central nervous system, 73 on synaptosomes, 73 bicuculline blocks inhibition at crayfish muscle, 85 blocks strychnine resistant inhibition, 85 is a GABA receptor antagonist, 85 binomial and Poisson statistics for quantal release, 298 distributions and deconvolution, 187 statistic error analysis, 170 brain and spinal cord, 1 butaclamol blocks binding of dopamine, 95 is a neuroleptic, 95 calcium
334
INDEX
determines probability of quantal release, 137 determines amplitude of endplate potential, 137 enters nerve terminals, 147, 148 measured with indicators in terminals, 148 hypothesis for transmitter release, 147 influx imaged at the active zone, 163 injection into terminals releases transmitter, 148 ions act on the inside of nerve terminals, 40 must be present for transmitter release, 40, 147 calcium action potential identified, 145 in crustacean muscle after buffer injection, 141 in crustacean muscle in TEA, 141 calcium action on the inside of terminals, 145 calcium channels alpha 1 subunit, 155 insertion at growing synaptic terminals, 276 types differ at different synapses, 155, 158 types L, N and T delineated, 153 distinguished by their pharmacology, 155 L-type blocked by dihydropyridines, 153 N-type blocked by specific toxins, 153 P-type, 153 Q-type, 153 R-type, 153 calcium component of cardiac muscle action potential, 142 of intestinal muscle action potential, 142 calcium conductance of single channels, 151 of L-, N- and T- type channels, 153 calcium cooperativity mechanism involved in transmitter release, 145 required for transmitter release, 147 calcium current component in molluscan impulses, 142 in molluscan neurons, 151 noise in molluscan neurons, 151 in voltage clamped cardiac muscle, 142 calcium required for acetylcholine release from ganglia, 135 for liberation of acetylcholine in the brain, 135 for transmission at the endplate, 137 calcium sensor for transmitter release, 139 possesses multiple calcium binding sites, 158 catecholamine granules in vesicles, 40 nerves identified by fluorescence, 95 central excitatory state, 27, 29, 30, 34, 38 central inhibitory state of the heart determined, 30 central inhibitory state, 29, 30, 36, 38 chemical transmission and transmitters, 38, 81, 114 at synapses, 25 and receptors, 107 concept, 50 in the heart, 27 chemiotactic concept and specificity molecules, 283 in synapse formation, 233
chemoreceptor concept, 44, 47 chlorpromazine analog promethazine is tranquilizer, 91 analogs synthesized, 91 has antipsychotic power, 91 used as an antipsychotic, 91 clinical effects, 91 use in schizophrenia, 91 cholinesterase at endplates of muscle, 62 in ganglia, 59 chronaxie affected by curare and acetylcholine, 59 circular smooth muscle relaxation at a bolus, 120 clozapine acts on a wide range of receptors, 102 does not block striatal dopamine receptors, 99 does not have Parkinsonian-like side effects, 99 identified as a neuroleptic agent, 99 inhibits mesolimbic dopamine neurons, 99 CO as a transmitter at NANC synapses, 130 collaterals formed by motoneurones in muscle, 254 compound binomial statistic, 171 for quantal release at crayfish junctions, 170 compound non-uniform binomial statistics for quantal release, 169 compound Poisson statistics, 169 and quantal release, 167 fails for quantal release at crab junctions, 170 concept of a junction between nerve and muscle, 45 of action current and action potential, 14 of a resting potential, 14 of chemical transmission, 50 of the synapse, 50 conduction along nerves from brain, 3 and transmission, 3, 8 and transmission particles, 4 cortical field potentials modified by amino acids, 79 cross-reinnervation of myofibres, 238 curare acts on junction between nerve and muscle, 45 acts on nerve terminals, 45 effects on nerve and muscle, 45 current flows between cut and intact end of muscle, 12 D2 dopamine receptor effects mediated by G proteins, 98 bind chlorpromazine and haloperidol, 95 in striatum elevated in schizophrenics, 99 in the cingulate and temporal cortex, 101 degeneration studies establish neuron doctrine, 20 studies on Purkinje cells, 20 dendrites discovered, 18 denervation leaves endplate intact, 23 dopamine as a transmitter in ganglia, 95 binding blocked by butaclamol, 95 dopamine D1 receptor distinguished from D2, 95, 98
335
336
INDEX
activation leads to diverse effects, 103 activate immediate early gene expression, 104 classified, 95 site of action of neuroleptics, 102 modulation of calcium currents, 103 knockout mice phenotype, 98 activation, 103 increase in the brain of schizophrenics, 101 dopamine D2-like receptors high in striatum, 99 dopamine D4 receptor cloned, 98, 99 increase in brain of schizophrenics, 101 dopamine diffuses large distances on release, 102 elevates adenylate cyclase, 95 in the brain, 92 receptor blockade is action of neuroleptics, 92 receptors cloned, 98 turnover accelerated by butyrophenones, 92 in ventral tegmentum and substantia nigra, 95 effects blocked by chlorpromazine, 95 drugs form compounds with substances in cells, 51 dual-constraint theory for synapse formation, 243 ectopic synapses formed after injury, 238 electric fluid flow along nerves, 10 electrical signs of acetylcholine receptor channels, 62 of transmission at endplates, 62 electrical theory of transmission, 23 of central inhibitory transmission, 36 of inhibition, 30 electrical transmission in ganglia proposed, 32 in smooth muscle proposed, 32 electricity generated by nerve and muscle, 10 electronmicroscope images of terminals, 40 elementary event due to acetylcholine, 62 end bulbs identified as nerve terminals, 18 on dendrites, 18 possess membranes, 23 transmission different to conduction, 22 endplate biological noise, 165 defined, 15 endplate potential amplitude calcium dependent, 40 composed of units, 165 recorded in frog muscle, 34 size depends on calcium, 135, 137 4th power dependence on calcium, 40 enteric inhibitory neurons not adrenergic, 114 enteric nervous system defined, 107 ergot fungus contains acetylcholine, 54 eserine and leech muscle assay for acetylcholine, 57 potentiates transmission through ganglia, 58 excitability of nerve and muscle, 59 excitatory amino acid
antagonists, 88 transmitter, L-aspartate, 88 excitatory and inhibitory synapses, 27 excitatory state at endplates determined, 34 failure to identify acetylcholine in the body, 54 final cause, 1, 3, 4 flexion reflex and inhibition, 27 flow of current between dissimilar metals, 12 form of a thing, 1 frictional machine, 9, 10 functional relation between nerve and muscle, 50 GABA action on cortex antagonised by picrotoxin, 79 and glycine uptake systems, 85 effects on cortex, possible inhibitory, 79 has depressant action at crayfish junctions, 83 has depressant effect on motoneurones, 79 and inhibition in central nervous system, 85 inhibitory transmitter in crustacean muscle, 79 GABA receptor antagonist enhances overflow of GABA, 71 types, 90 GABA synthesizing enzymes associated with synapses, 85 high in cortex, 79 gamma distribution for quantal amplitudes, 171 gastrointestinal tract, its autonomic innervation, 125 glutamate can be neurotoxic, 85 effects on cortex, 79 glutamate receptor metabotropic, 88 antagonists, 87 diversity, 87 cloned, 90 glutamate transport by glial cells, 87 glutamic acid excitant effect on motoneurones, 79 is active principle of rabbit pallium, 79 glycine depressant effects on motoneurones, 81 is inhibitory transmitter at motoneurones, 82 growth cones of motor axons, 240 growth factor molecules and synapse formation, 277 haloperidol binding correlated with clinical potencies, 95 identified as major neuroleptic, 92 synthesized, 91, 92 heart possesses an endogenous esterase, 57 hippocampal damage and memory loss, 211 IJP, inhibitory junction potential in smooth muscle, 120 and peristalsis, 123 blocked by tetrodotoxin, 118 fast and slow components, 127 fast component blocked by apamin, 127 fast component in taenia coli due to ATP, 130 generation by nitric oxide, 127–129
INDEX
in the taenia coli, transmitter identity, 128 is potassium dependent, 117 on distension of the gut, 123 impulse conduction, 13, 38 induction of associative LTP, 218 infinite velocity of the action current, 14 inhibitin created on vagal stimulation, 54 inhibitory descending pathway in the intestine, 125 mechanisms in peristalsis, 123 postsynaptic potentials, 38, 41, 82, 84, 189 transmitters in the intestine, 127 integral, implicated in anchoring terminals, 276 interruption of heart beat by extracts, 27 interstitial cells of Cajal and the IJP, 129 iontophoretic techniques in pharmacology, 79 junction potentials action of alpha receptor antagonists, 119 kinetic model of quantal secretion, 180 release of a quantum, 179 law of the intestine, 110 L-Dopa effects antagonized by chlorpromazine, 92 L-glutamic acid transmitter at crayfish muscle, 79 LTP, long term potentiation and changes in dendritic spines, 221 and increased release of transmitter, 223 and memory, 214 and spine enlargement, 223 and the role of calcium in induction, 221 and calcium calmodulin, 228 development, 217 discovered in the hippocampus, 211 in hippocampus may be related to memory, 214 in senescence, 214 in the limbic system, 214 in the medial geniculate nucleus, 214 in visual cortex, 214 increases postsynaptic receptors, 225 induction, 221 lasts for weeks in rabbit hippocampus, 211 maintenance and metabolic changes, 223 maintenance involves protein kinases, 225 NMDA receptors and magnesium block, 221 non-associative, 230 role in memory and learning, 230 role of protein kinase C in maintenance, 228 throughout the limbic forebrain, 214 MASC, myotube-associated specific component, 274 required for agrin binding, 274 maximum likelihood estimation methods in the analysis of quantal release, 185 in statistical models of quantal release, 186 mechanism of conduction and transmission, 8 of generation of action current, 15
of conduction, 4 of transmission, 4 mechanistic philosophy, 4 membrane noise due to acetylcholine, 62 motoneurone death and synapse formation, 268 motor and sensory nerves, 3 muscle contracted by injecting acetylcholine, 59 generates currents when twitching, 12 MuSK, a muscle receptor tyrosine kinase in synapse formation, 274 a synapse formation molecule, 273 NANC, nonadrenergic noncholinergic, 120, 127 excitatory junctions in the intestine, 132 excitatory transmitters in the intestine, 132 neurons identified, 131 and potassium permeability changes, 127 transmitter storage in nerve terminals, 132 negative variation in muscle current during a tetanus, 12 of action current on stimulation, 12 nerve and muscle possess different excitabilities, 59 composed of many cylinders, 8 composition given microscopic description, 8 conduction mechanism, 14 cylinders contain glutinous material, 8 described, 4 fibres originate from neurons, 15 membranes are polarised at rest, 15 to muscle synapse named, 23 neuregulins in ganglia, 279 neurofibrils conduct action potentials, 20 extend from end bulbs into neurons, 18 neuroleptics act on D2 and 5-HT2 receptors, 102 block dopamine receptors, 92 block monoamine receptors, 92 defined, 91 possess a wide range of activities, 102 neuron doctrine, 20, 22 first identified, 15 neurotransmitters as particles or granules, 40 nicotine acts on a substance in muscle, 51 and curare on muscle, 53 nitric oxide activates potassium channels, 130 NMDA and glutamic acid, 81 NMDA receptors, 88 distribution in the brain, 88 and magnesium ions, 88 at excitatory synapses, 88 gated by magnesium, 221 non-NMDA receptors at excitatory synapses, 88 non-uniform binomial statistics maximun likelihood estimation criteria, 190 probability of release, 174 noradrenaline
337
338
INDEX
and acetylcholine not sole transmitters, 120 is released from sympathetic nerves, 38, 67 present in the brain, 92 stored in granules, 40 stored in vesicles, 40 uptake and receptors at nerve endings, 69 parasympathetic nerve origins, 107 and supra-renal extract distinguished, 47 peristalsis described, 123 giving muscle contractions, 123 physostigmine potentiates the effects of acetylcholine, 58 pituitary adenylyl cyclase-activating peptide, 130 Poisson versus binomial statistics for quantal release, 169 polarisation of the neuron doctrine, 22 polyneuronal innervation elimination, 242 and increases in impulse traffic, 247 and lack of impulse traffic, 246 during development, 244 during reinnervation, 250 in developing ganglia, 277, 278 polyneuronal innervation of myofibres, 242 of synaptic sites, 240 excess, 254 potassium theory of the membrane potential, 15 presynaptic alpha 1 adrenoceptors, 72 and postsynaptic adrenoceptors, 72 terminals possess receptors, 67 probability of quantal release and density for synaptic delays, 180 under a depolarization and calcium, 139 differs at different sites in a terminal, 177 from a single release site, 182 from n release sites, 183 and secretion, 38, 177, 183 at single central nerve terminals, 190 psyche, 1, 3, 4 psychic pneuma replaced by particles, 4 quantal release at group 1 afferents is non-uniform, 177 at forming motor nerve terminals, 171 and quantitative description of kinetics, 182 hypothesis, 38, 298 over a time interval, 182 reaction equations for kinetics, 180, 182 models discriminated between, 194 rational psyche, 1 reason for a thing being as it is, 1 for existence of a thing, 1 rebound excitation and NANC transmitters, 132 firing in muscle on vagal stimulation, 118 firing of action potential in muscle, 118
receptive side chains become receptors, 45 receptive substance (receptor), 43 first defined, 51 and cells, 53 does not bind drugs, 47 located near nerve endings, 51 reinnervation of ganglia, 278 of mature muscles, 258 of myofibres, 236 of truncated muscles, 260 reserpine depletes monoamines in the brain, 92 isolated from Rauwolfia serpentina, 91 reduces serotonin in the brain, 92 used in psychiatric medicine, 91 antipsychosis and depletion of monoamines, 92 risperidone acts on D2 and 5-HT2 receptors, 103 ROCA1, a Schwann cell molecule, 283 sarcomere addition at the end of myofibres, 238 Schwann cell tubes, axons grow down, 233 secretosome concept, 163 segmental reinnervation of ganglia, 278 serotonin blocks hallucinogenic effects of LSD, 92 identified in the brain, 92 receptors acted on by neuroleptics, 102 side chain theory of cellular action, 45 silver stain introduced, 18 single calcium channels identified with patch clamp, 151 S-laminin, in motor terminal differentiation, 275 spinal cord—leg preparation, 10 spontaneous miniature endplate potentials, 38, 137, 165, 277 discovered, 38 independent of calcium, 135 statistical analysis of quantal release, 38 of quantal delays, 179 striatal distribution of receptor types, 101 strychnine as a specific antagonist of glycine action, 82 effects on electrical activity of cortex, 79 studies on reflex contractions and synapses, 25 substance P in intestine and brain, 118 supra-renal extract acts like sympathetic nerves, 47, 65 named epinephrin, 65 acts on muscle not nerve terminals, 47 sympathetic nerves release adrenaline, 67 origins, 107 do not synapse on intestinal smooth muscle, 125 to the intestine do not give IJPs, 125 sympathin as demythelated adrenaline, 67 sympathomemitic amine effects on effectors, 65 ‘synapse’ name introduced, 23 synapse formation and motor unit development, 250 in autonomic ganglia, 277 in partially denervated neonatal muscles, 254
INDEX
involves a postsynaptic source, 242 involves presynaptic source, 242 molecules identified, 271 molecules with different affinities, 262 on developing myofibres, 238 on ganglion cell dendrites, 277 receptor molecules on myofibres, 240 dual constraint hypothesis, 240 synaptic basal lamina and Schwann cell growth, 277 in synapse formation, 271 synaptic sites on ganglion cells, 277 form at random on developing myofibres, 242 contain information for synapse formation, 233 reinnervated, 233 synaptic terminal withdrawal, 252 synaptic transmission in ganglia, 32 synaptotagmin epitopel as the low affinity calcium sensor, 162 epitope 3 as a calcium sensor, 162 associated with vesicles, 158 binding sites for phospholipids, 162 C2a domain antibody effects, 160 C2a domain binds calcium, 162 domains, 160 mutants and failure in transmission, 160 mutants possess low calcium cooperativity, 160 syntaxin and calcium channels linked, 163 topographical information in synapse formation, 264 synaptic map formation, 260 transmission dependent on calcium ions, 135 described, 4 different to conduction, 22 into muscle, 4 is irreversible, 22 not due to flow of particles into muscle, 10 of animal spirits from nerve to muscle, 7 transmitter release calcium dependent, 40 vagal inhibitory nerves do not release adrenaline, 114 nerves to stomach and inhibitory reflexes, 113 reflex to stomach and deglutition, 113 vagus both excitatory and inhibitory to intestine, 110 contains inhibitor nerves to stomach, 108, 113 releases acetylcholine in stomach, 108 to the heart, characteristics, 30 Vagusstoff, 30, 44, 56–58 is acetylcholine, 57 released by vagus nerve in heart, 57 released from nerves in ganglia, 58 vasoactive intestinal peptide and the IJP, 129 vesicle hypothesis, 36–40, 165 vital pneuma, 1, 3, 7 voltage-dependent ion channel concept, 139
339