ANNALS OF THE NEW YORK ACADEMY OF SCIENCES Volume XLVII. Art. 4. Pages 375-602

Editor

Roy Waldo Miner

Consulting Editor

David Nachmansohm

Associate Editor Michael Demarest

THE PHYSICO-CHEMICAL MECHANISM OF NERVE ACTIVITY

By

David Nachmansohn, Charles M. Berry, Oscar Bodansky, Frank

Brink, Jr., Detlev W. Bronk, M. Vertner Brown, C. W. Coaxes,

R. T. Cox, J. C. Eccles, Alfred Fessard, J. F. Fulton, R. AV.

Gerard, Alfred GilxMAN, D. E. Green, Joseph C. Hinsey,

Rudolf Hober, Martin G. Larrabee, and

Tracy J. Putnam

Q

11

.N5

new YORK

Published by the Academy

DFrrMBFR 15, 1946

Annals of The New York Academy of Sciences

Volume XLVII, Art. 4. Pages 375-602

December 15, 1946

THE PHYSICO-CHEMICAL MECHANISM OF NERVE ACTIVITY*

By

David Nachmansohn, Charles M. Berry, Oscar Bodansky, Frank

Brink, Jr., Detlev W. Bronk, M. Vertner Brown, C. W. Coates,

R. T. Cox, J. C. Eccles, Alfred Fessard, J. F. Fulton, R. W.

Gerard, Alfred Oilman, D. E. Green, Joseph C. Hinsey,

Rudolf Hober, Martin G. Larrabee, and

Tracy J. Putnam

CONTENTS

PAGE

Introduction to the Conference on Nerve Activity. By Tracy J. Putnam 377

The Membrane Theory. By Rudolf Hober 381

Chemical Mechanism of Nerve Activity. By David Nachmansohn 395

An Electrical Hypothesis of Synaptic and Neuro-Muscular Tr.-vnsmis-

siON. By J. C. Eccles 429

Chemical Excitation of Nkrve. By Frank Brink, Jr., Detlev W. Bronk,

AND Martin G. L.\rrabee 457

Electrical Characteristics of Electric Tissue. By R. T. Cox, C. W.

Coates, and M. Vertner Brown 487

Some Basic Aspects of the .Activity of Electric Plates. By .4.. Fessard. . 501

Physiological Function from the Standpoint of Enzyme Chemistry. By

D. E. Green '. 515

Cholinesterase. By Oscar Bodansky 521

The Effects of Drugs on Nerve Activity. By Alfred Gilman 519

The Recovery of Diameter and Impulse Conduction in Regenerating

Nerve Fibers. By Charles M. Berry .\nd Joseph C. Hinsey 553

Nerve Metabolism and Function. By R. W. Gerard 575

Conclusion. By J. F. Fulton 601

* This .series of papers is tlie result of a Conference on The Physico -Chemical Mechanism of Nerve Activity held by the Section of Biologv of The New York Academy of Sciences, February 8 and 9, 1946.

Publication made possible through a grant from the Conference EuWie&tiQjis Revolving Fund.

(375) Xv^3^^i47 N

fi

library) 5i

Copyright 1946

BY

The New York Academy of Sciences

INTRODUCTION TO THE CONFERENCE ON NERVE ACTIVITY

By Tracy J. Putnam Columbia University, New York, N. Y.

It is my pleasant duty to open a conference which is, in several respects, historic.

In the first place, this is, as far as I am aware, the first truly inter- national conference on a purely physiologic or, in the broad sense, medical subject, since the beginning of the war. I see, in the audience, five continents represented, and I can reproach the program commit- tee and myself only for failing to arrange for a delegate from Africa. It is my special privilege to welcome Dr. Eccles from Australia, Dr. Feng from China, Drs. Fessard, Couteaux, and Bugnard from France, and Dr. Bremer from Belgium, who have come so far for the special purpose of taking part in this symposium.

The fact that we are all met here from various corners of the earth to discuss problems of pure science with a humanitarian purpose sym- bolizes, it seems to me, the hope of this troubled world, the hope that civilization is beginning to recover from a desperate and destructive illness, which barely missed being fatal. A relapse might well be final. But here we are, ready, willing, and able to talk over some questions which are of great importance, but definitel}^ non-political. It is to be expected that differences of opinion will arise, and facts will be hotly debated. I feel certain, nevertheless, that the members of this confer- ence will be able to set an example for future international discussions, in agreeing on the criteria of truth and the means of arriving at an agreement on facts, with a broad tolerance towards possible means of interpretation. Let us hope that the United Nations Organization will take notice and be willing to learn.

This meeting opens a wholly new vista, in another sense, also. At the time when the German Army crossed the Polish border, in 1939, there seemed to be little hope of bridging the gap between the point of view that transmission of the nervous impulse was a purely electrical phenomenon, and on the other hand, the conception that the production of a specific chemical substance was the essential fact. The atmosphere of the war was not in the least conducive to placid scientific

(377) ' "^^

378 ANNALS NEW YORK ACADEMY OF SCIENCES

but here, six years have passed, and behold! the whole subject has be- come suddenly clearer. Our new insight into the problem has not come easily, and I must pay special tribute to our colleagues in Europe and in China, who have had the courage to carry on their investigations in the face of enormous discouragements and practical difficulties, and sometimes even in secrecy and at the peril of their lives.

It is amazing that so much has been accomplished under such ad- verse circumstances. The details of the physiology of the individual elements of the nervous system have, in the past, seemed most obscure, and we have had to infer the outlines of the metabolic processes which occurred there from data gathered from other tissues. But it now seems safe to say that our picture of the metabolism and the mecha- nism of action of neurones is more complete than our knowledge of any other tissue, and the methods of study which were originally con- fined to neurophysiology are being extended to other physiologic problems.

It is particularly gratifying to me that a clinical neurologist should be permitted to open this meeting. Clinical neurology used to be con- sidered a purely diagnostic specialty, a hopeless field of medicine, which consisted in little more than a meditation on disease. We are beginning now to be able to do a little more about the disorders of the nervous system, but we can make progress only as we possess insight. The physiologic methods of study which have been devised and applied by the distinguished scientists I see before me, and the facts they have elicited are, I am sure, the surest guide we possess to advances in therapeutic methods. This is a new chapter, not only in neurophysi- ology, but in pharmacology, clinical medicine, and, perhaps, even for the dark territories of psychiatry.

We are grateful to The New York Academy of Sciences, and especially to the executive secretary, Mrs. Miner, for having organized so eflfi- ciently and made possible this symposium. We were fortunate in hav- ing the support of the Rockefeller Foundation, and we express our gratitude to Dr. Lambert for his advice and active cooperation.

I should also like to thank very warmly Dr. Raymond Zwemer who, through his association with the State Department, helped us to over- come many difficulties.

A few words about the program. The purpose of the symposium is not to present recent data alone, but to give an opportunity for dis- cussing some of the fundamental aspects and problems. There is so much to say that we have filled the program perhaps unduly full, and still have been unable to find time for many investigators whom we

PUTNAM: INTRODUCTION 379

should all enjoy hearing. We hope, however, there will be enough time left for discussion, if we abide closely by our schedule. The chemical aspect of the subject has been as much stressed as the physical. It seems that we shall all have to get accustomed to terms like enzymes and coenzymes, as well as to positive and negative phases, and Weden- sky inhibition. At the end of the symposium, Dr. Gerard will try to integrate the different aspects which will be presented and discussed.

THE MEMBRANE THEORY

By Rudolf Hober University oj Pennsylvania, Philadelphia, Pennsylvania

The classical objects of the study of bioelectric phenomena are mus- cle and nerve. Resting, injury, and action potentials and currents are studied with both of them. On the basis of Wilhelm Ostwald's investi- gations upon the electric properties of artificial inorganic precipitation membranes (1890), the physiological membrane potentials have been looked upon as being special forms of Nernst concentration potentials; in other words, potentials arising when solutions of different electrolytes are separated by a membrane characterized by a more or less selective ion permeability. As it was from the beginning of the electrophysio- logical era, both nerve and muscle have been used for solving the basic problems, and information gained from one type is valuable for both. Therefore, although our object is primarily a discussion of physico- chemical mechanisms of nerve activity, muscle potentials will also be treated.

The Membrane Theory was established, in 1902, by Bernstein, when he ascribed the EMF of the locally injured muscle fiber to selective permeability to potassium ions present inside the fiber in a considerably greater concentration than outside. He conceived of the action poten- tial wave as a self-propagating depolarization by breakdown of this selective permeability. It was early assumed that this alteration is accompanied by chemical reactions.

THE ION PERMEABILITY OF THE RESTING MUSCLE

AND NERVE

In immediate connection with Bernstein's theory, in order to become acquainted with the general properties of the selective ion permeable membranes, it seemed to me that one of the main tasks was to study the effect of local application of the neutral inorganic salts upon the resting potential of muscle (Hober^).

A. The Inorganic Cations

At first, the alkali cations only were varied, and potassium was found to produce the strongest negative pole, similar to the effect of

(381)

382 ANNALS NEW YORK ACADEMY OF SCIENCES

cutting the muscle, but often as a reversible process and obviously in- dicative of the greatest permeating power. The other alkali cations appeared to be less effective, in this order: potassium, rubidium, sodium, lithium; potassium and rubidium producing negativity, as compared to sodium, while lithium produces positivity. This was interpreted as being due to swelling or shrinking of hydrophilic colloids, which were assumed to be the chief constituents of the plasma mem- brane. Later, the same series was met by Michaelis with non-colloidal, rigid, dried collodion membranes. His findings seemed to fit in best with the concept of an ionic sieve. For, taking into consideration the shells of water dipoles around the ions, the effective ionic volume ap- pears to be smallest with potassium, largest with lithium. But, as will be seen later, the differences are better correlated with adsorption, which, according to Gouy, Frumkin, and others, increases with decreas- ing hydration of the ions, potassium being most, lithium least, adsorbed. The membrane theory postulates that, as in the case of the suffi- ciently dried collodion membrane, which is permeable only to potassium ion, strength and direction of the injury potential are dependent upon the ratio of potassium inside to potassium outside. In other words, the surface of muscle or nerve behaves as a potassium-electrode, potassium inside being constant and about 20 to 40 times greater than potassium outside. Therefore, by raising potassium outside, the EMF of an injured (cut) muscle should be decreased to zero, if potassium outside is equal to potassium inside, and its direction should be re- versed, if potassium outside is greater than potassium inside. My own early experiments (1905) failed to show the reversal, because, in con- trast to the rigid, dried collodion membrane, the ion selectivity of the plasma membrane is lost, due to its colloidal behavior: in other words, due to the swelling and disintegrating, even to the cytolyzing effect of higher potassium, especially after some lapse of time. However, the postulate of a reversal complies with recent observations of Hodgkin and Huxley,' and of Curtis and Cole," in a particularly striking way. These authors, leaning upon Osterhout's* studies on the "impaled" giant plant cells ( Valonia) , pushed a microelectrode into the axoplasma of the giant nerve fiber of the squid, along its axis, so that its tip was placed just opposite to the outside electrode, and they thus measured the membrane potential directly across the wall. The potential was found, in the case of the squid nerve, to be, on an average, 50 mV. Then, upon raising potassium outside to about 18 times normal, the resting potential was decreased to zero, and upon raising it about 40 times normal, a reversal of 15 mV was observed. The corresponding

HOBER: THE MEMBRANE THEORY 383

procedure has been applied to single frog muscle fibers by Gerard, Carlson, and Graham.^

B. The Inorganic Anions

The colloidal behavior of the physiological membranes, further, is brought into evidence by the effect of the inorganic anions. In gen- eral, with regard to their role in physiology, the anions are less powerful than the cations. This is due to the prevalent negative charge of the colloidal aggregates, which repel the anions. Locally applied to nerve or muscle, the resting potential reveals the following anion series: thio- cyanide, iodide, bromide and chloride, sulfate, with thiocyanide ion producing a positive pole. In other words, a reversed injury potential occurs as a consequence of the anion adsorption on the membrane, which is greatest with thiocyanide, smallest with sulfate. This re- versal is an important point to be kept in mind for later discussion. However, after some lapse of time or after applying the salts in stronger solution, the anion series is reversed, thiocyanide and iodide forming a negative pole, thus resembling the effect of potassium ion. This, again, is significant, as due to a loosening, softening, and subsequent disintegrating action upon the hydrophilic colloidal membrane.^' ^

C. Organic Anions

These effects are related to those of a large group of organic anions: for instance, those of higher fatty acids starting with the 8-carbon atom chain, i.e., caprylic acid. These ions have a nonpolar-polar structure, the nonpolar, or organophilic and hydrophobic, part of the anion, mainly due to the adsorption affinities of the alkyl radicals, attaching to the organic material, e.g., to the particles of a Langmuir surface film of protein; the other polar or hydrophilic part, due to the carboxyl radicals with their cloud of water dipoles, anchoring in the water. It appears that a pull towards the water can be exhibited upon the organic material. The pull is stronger or weaker, corresponding to the relative hydroaffinity, so that colloidal particles composed of a variety of molecules, hke hemoglobin (which means, hem plus globin), visual purple, or chloroplastin, can be torn to pieces, a process termed de- naturation, solubilization, or detergency. On the other hand, coiled peptide chains can be uncoiled, and organic architectures of great com- plexity, such as a plasma membrane, can be loosened by the adsorptive pull, which has the effect of abolishing reversibly the selective ion permeability of the membrane and starting irreversible cytolysis.^ For example, sodium caprylate is applied locally to a muscle. The re-

384 ANNALS NEW YORK ACADEMY OF SCIENCES

suit of a very weak solution is simply a reversed resting potential, due to the anion attachment to the pores of the membrane; the effect of a stronger solution is a regular resting potential, due to reversible loosening; while the result of a still stronger solution is an irreversible disintegration, i.e., an injury potential.

D. Inorganic Plurivalent Cations

The occurrence of the well-known cation antagonism is another in- dication of the prevalence of anionic colloids in cell structure. An im- balance between the monovalent and the plurivalent cations shows up in numerous observations upon animal and plant cells, among others by alteration of their electrical properties, as, for example, ohmic resistance, or conductance, or excitability. Preponderance of mono- valent cations (sodium, potassium) is bound up in muscle and nerve with loss of normal selective cation permeability, due to increased hy- dration. However, this is compensated for by the consolidating effect of plurivalent cations, like calcium, strontium, barium, cobalt, manga- nese, nickel (Hober^).

THE CHANGE OF ION PERMEABILITY BY DC

Let us turn now to the old and complex phenomenon of electrotonus. It comprises a multitude of alterations of cell responses, effected by direct current and having their origin in changes of membrane polariza- tion. Especially well-known are the changes of excitability of nerve and muscle. Excitability is diminished at the anode and increased at the cathode, except that, beyond a certain current strength, the in- crease turns to a decrease, the so-called cathodic depression. Anelcc- trotonus and catelectrotonus are tied up with changes of resistance. By placing one electrode on an intact spot of the excised nerve, the other on a crushed end, the resistance is raised at the anode, dimin- ished at the cathode. Consequently, while sending alternating current through the preparation instead of direct current, a rectifier effect ap- pears. These and other observations can be explained, partly on the basis of ion distribution between the inside and the outside of the membranes, partly by taking again into account the colloidal prop- erties of the wall of the natural membranes. As to the first point, ac- cording to Bear and Schmitt,^ Cowan,^" Fenn and others," and Webb and Young,^ the axoplasm of the giant nerve fiber of the squid, for example, contains 4 to 5 times more inorganic cations (mainly potas- sium) than anions, and 18 times more potassium than the blood. Be-

HOBER: THE MEMBRANE THEORY 385

sides chloride, there are in the axoplasm small concentrations of phos- phate, sulfate, and lactate, but rather large amounts of organic anions of low mobility, possibly the anions of amino acids. These conditions are roughly reproduced in model experiments of Labes^^ and Ebbecke.^^ A membrane core-conductor is formed by a collodion tube, with pores wide enough to allow cations and anions to pass the wall. The tube is filled with a solution of potassium phosphate and is packed in gauze which has been wetted with a solution of sodium chloride. One electrode is placed inside, another outside. If direct current passes the membrane, a smaller resistance is encountered, when the current goes from within outwards, than when it goes in the opposite direction. The reason is that, with the outgoing current, the faster potassium inside and chloride outside are swept into the membrane and travel, there, with greater velocity than so- dium outside and phosphate inside, being driven by an ingoing current. If, instead of collodium, hydrophilic and negatively charged colloids are the membrane constituents, as they actually are under most physiological conditions, then additional swelling and increasing dis- persion occur at the cathode, as well as shrinking at the anode. Con- sequently, the polarizability of the membrane falls at the cathode and rises at the anode. Swelling causes in natural objects, such as muscle and nerve, greater excitability at the cathode, but as the current strength rises more and more, the higher excitability turns over to inexcitability, or, in other words, to cathodic depression. More spe- cifically, according to Blinks,^* the membrane polarization of a giant cell of the fresh water alga Nitella, comparable to nerve or muscle with its thread-like shape, drops down to zero, if exposed to the swelling effect of potassium chloride in sufficiently high concentration, and the cell does not respond any more to otherwise effective stimuli. How- ever, by applying, locally, an anode of rising strength to the depolarized Nitella cell, above a certain threshold value, the polarizability is re- stored, and a normal action potential can be elicited upon stimulation. Alternatively, with respect to cation antagonism, after excitability of a nerve has been suppressed by calcium, this stiffening effect is can- celled by the softening influence of a cathode, as shown by Woronzow," and more recently by Guttman and Cole.^^

We turn, now, to the discussion of natural changes of ion permeability during action. It has been accepted, for more than 40 years, that depolarization, which is brought about by injury, compares essentially to depolarization accompanying excitation, as indicated by the "nega- tive variation" of du Bois-Reymond. The negativity wave, therefore, is interpreted as a "breakdown of the membrane," by which the selec-

386 ANNALS NEW YORK ACADEMY OF SCIENCES

tive cation permeability is abolished. Correspondingly, it has been assumed that, during excitation, the potential fall is as great as is the resting potential, measured at best with the impaled nerve or muscle. But this is not true. Impedance measurements have shown the resist- ance to persist, to some extent, during excitation (Curtis and Cole) . In other words, the resting potential could be expected to be larger than the action potential. However, the contrary is true. Hodgkin and Huxley,- and Curtis and Cole," inserting a microelectrode into the axon, detected the potential change, during activity, to be even larger than that due to injury. For example, in the experiments of Curtis and Cole, the resting potential average is 51 mV, the action potential 108 mV.

Before discussing this interesting situation, attention will be turned briefly to a special problem. The word, breakdown, suggests leakage, and for this reason, activity could be expected to be accompanied by leakage, especially from the large surplus of well-diffusible potassium .normally retained in the axoplasm. However, such an escape from frog nerve, though often investigated, is doubtful, except following very prolonged stimulation (for example, 60 stimuli per second, for 1-3 hours, in the experiments of Arnett and Wilde, with Fenn).^^ However, this may be accounted for, by assuming that only a very small area of the surface of a myelinated nerve, the Ranvier nodes, is available for diffusion. This can be correlated with the experiments of Cole and Curtis, ^^ regarding impedance and membrane capacity of the squid nerve. Notwithstanding the fact that, during excitation, the resistance of the squid nerve falls off from 1000 ohm/square-cm. to only 25 ohm/square-cm., not more than 2% of the area is involved in the increase of permeability. This means that the remainder, about 98%, would be inactive. Another point is the fact that the state of excitation, in general, lasts only a very short time, measured in milliseconds. Very slowly reacting cells, therefore, may offer a greater chance to detect an ion escape. As a matter of fact, the con- ductivity of the water on the outside of the surface of a Nitella cell rises perceptibly, after several excitation waves have passed the slowly responding object, the excitation time being measured in tenths of a second (Cole and Curtis). Since depolarization is followed by re- polarization, the question arises, whether and how the ions which escape through the leaky membrane are recovered. It becomes in- creasingly clear that, in one way or the other, energy is utilized for this purpose. In other words, the physiological membranes are more than labile structures. Rather, they are, or can be, acting machineries.

HOBER: THE MEMBRANE THEORY

387

For example: According to Furusawa, Feng, and Shanes and Brown/^ during anoxia the polarization of crab nerve and its excitability fall off reversibly, but seem to be maintained in the presence of phos- phopyruvic acid, adenosintriphosphate, and thiamin; in other words, by establishing the normal glycolytic cycle. According to Hoagland and Davis,^° Nitella cells in the dark lose their intracellular chloride ions, through the protoplasmic wall, into the surrounding water and recapture them during exposure to light. Furthermore, according to J. E. Harris, ^^ potassium ion gets lost from human erythrocytes at low temperature, but re-enters at room temperature, after addition of glucose.

REVERSAL OF THE NORMAL ACTION POTENTIAL

I now come back to the lately-discovered fact, already mentioned, that the potential change during action does not equal the resting potential in magnitude, as it was assumed for many years. Rather, by overshooting the zero line, as shown in figure 1, the potential is

Figure 1. Potential of the internal electrode. The figure shows that the resting potential is — 44 mV. During activity, the potential overshoots to the positive side, +40 mV, so that the action potential wave amounts to 84 mV. ( Hodgrkin & Huxley^.)

momentarily reversed in sign, the outside of the membrane becoming negative to the inside. This reversal during passage of the impulse does not fit into the classical picture of the behavior of the active nerve membrane, and possibly indicates a special mechanism, which is super- imposed to the mechanism of the customary excitation depolarization. Figure 2 depicts three conditionsof the nerve membrane: (a) represents the normal polarization of a resting nerve membrane; (b) is indicative of the depolarized membrane, which, according to the ordinary view-

388 ANNALS NEW YORK ACADEMY OF SCIENCES

point, during activity is fairly equally permeable to cation and anion, to this extent resembling the situation in injury; while, in (c), an additional influence of organic anions has been taken into account, as has been discussed already by Hodgkin and Huxley.- These authors

d-,,.<j V.S+ _± 1 1 1 ± 1 1_ -its nV (a)

^"'1 -r-) :: :: :: :: :: :: ; 01^ (b)

a/opkt>n.

^--',^'-- ::l::i::t::i::t::i^ ^-fow (c)

Figure 2. Diagrams illustrating the reversal of the nerve potential.

have considered, among others, particularly the lactic acid anion, which, during activity, would penetrate the membrane from inside and pro- duce a negativity outside. However, this hypothesis is rejected by Hodgkin and Huxley themselves, as it would be hard to imagine the concentration and the mobility in the membrane of the lactate ion as being sufficient. Instead, I would prefer to pay attention, especially, to the organic nonpolar-polar, hydrophobic-hydrophilic anions, already mentioned, which possibly can be assumed to be present in the nerve membrane, or, rather, to be liberated as the excitation wave travels along the fiber. As stated earlier, such anions, locally applied to the outside of a muscle, call forth a reversed resting potential, whereas, if they originated during excitation inside, they would call forth a reversed action potential, due to the fact that the adsorption forces would draw these anions into the porous membrane, as shown in figure 2(c). Such a reversed resting potential has been found with the salts of higher fatty acids, alkyl sulfates, aryl sulfonates, and others. These experiments should be extended to nerves, especially to single nerve fibers like that of the squid, for the following reason:

Nonpolar-polar anions are abundantly preformed in the molecules of lipoids of the nervous system, chiefly in phospholipids and cerebrosides. Among their split products, the nonpolar-polar character is especially pronounced in the anions of fatty acids with long carbon atom chains, and, according to Langmuir and Adam, particularly in fatty acids

HOBER: THE MEMBRANE THEORY 389

with one to three double bonds, e.g., in oleic, linolic, linoleic, arachi- donic, nervonic, and oxynervonic acids. These long carbon atom chains of the lipoids, lecithin, kephalin, sphingomyelin, and cerebrosides seem to be existent, not only in the massive sheath of the myelinated nerve fibers, but, according to Young and Francis Schmitt,^^ also in the unmyelinated fibers of crabs and cephalopods (for instance, the squid nerve) , where the thickness of the sheath has been found to be as small as one per cent of the diameter of the axon, i.e., about 5/i, compared to 25 per cent of the diameter in vertebrates, as shown by Pumphrey and Young (plate 1). In the sheath of the unmyelinated fibers, the lipoids, though often not demonstrable by the customary staining with osmium tetroxide, can be detected with polarization optics (Bear and F. 0. Schmitt-^).

Now, looking upon the excitatory process from the standpoint of the old "Stromchen theory" of Hermann,^* it is at the boundary between the stimulated altered and the adjacent, unaltered region that small local circuit currents arise, flowing out of the unaltered region, which then secondarily gets altered as in a catelectrotonus, and flowing in at the originally stimulated region, which, thus, is inactivated as in an anelec- trotonus. Catelectrotonus, however, as mentioned before, means soften- ing the colloidal membrane and dispersing its structural aggregates by way of potassium and chloride ions and depolarization of the normal resting membrane. Anelectrotonus, on the other hand, means condensa- tion and re-polarization. The aforementioned increase of concentration of potassium, which happens to be produced in the membrane by the outflowing current, may then serve to liberate in the nerve membrane, directly, some of the nonpolar-polar anions, as, according to the well- known studies of G. L. Brown and W. Feldberg," acetylcholine is liber- ated by even a very small surplus of potassium (amounting to not more than 0.01 per cent) in the perfused ganglion cells, where it normally is fixed in a nondiffusible state. However, the mechanism of this re- lease is by no means clearer than that just suggested for the nonpolar- polar anions. Alternatively, the nonpolar-polar anions could possibly be liberated, indirectly, by an activation of lecithase A, an enzyme occurring in nerve tissue, which is known to set free the unsaturated, but not the saturated, fatty acids of the lipoid molecules. ^^

These are speculations, it is true. If, however, we refer them to the giant axon of the cephalopods, which was studied, in recent years, with most diversified and modern methods, it means that probably the alterations are bound up with the thin surface membrane which wraps up the voluminous column of axoplasm, and that this fine mem-

390 ANNALS NEW YORK ACADEMY OF SCIENCES

brane would be the site of a complex chemistry. Although the dis- cussion of the chemical side of nerve activity is beyond the scope of this paper, I should hke to conclude by turning to some interesting observations of von Muralt (1942), involving the appearance, during excitation, of a substance which may bring about the reversal of the membrane polarization.

When an excised frog sciatic is stimulated, electrically, at a certain frequency and simultaneously is dipped with a certain velocity into liquid air, several excitation waves must be caught and frozen along the nerve. When an extract of stimulated and unstimulated nerves, pulverized in the frozen state, is made up with eserinized frog Ringer or serum, it appears that, during excitation, the nerve has liberated minute amounts of several substances. One of these, by various tests, is identified as acetylcholine ; and a second substance is concentrated in the foam of the extract, which, from this sign of surface activity, possibly indicates the presence of a nonpolar-polar substance, whereas, in the foam from an acetylcholine-eserin-serum solution as a control, the acetylcholine fails to show an accumulation. Recently, by the same freezing method, von Muralt has intercepted a third substance, thiamin, which possibl)^ also is surface-inactive.-^ Certainly, these results are far from giving conclusive support to the concept that non- polar-polar substances, detectable by their surface activity, have been liberated during excitation. Even if they were, the liberation may be of minor significance, considering the fact that, according to Hopkins and Huxley, and to Curtis and Cole, the resting potentials of the giant nerve fibers vary little from one experiment to another, in contrast to a wide variability appearing in the size of their action potentials. In any case, this grouj) of observations emphasizes the urgent need to extend the study of chemical products, which are directly connected with nerve activity, beyond the demonstration of acetylcholine.

BIBLIOGRAPHY

1. Hbber, R.

1905. Pfliig. Arch. ges. Physiol. 106: .599.

2. Hodgkin, A. L., & A. F. Huxley

1939. Nature 144: 710. 1945. J. Physiol. 104: 176.

Webb, D. A.", & J. Z. Young

1940. .1. Physiol. 98: 299.

3. Curtis, H. J., & K. S. Cole

1942. .1. Cell. Comp. Physiol. 19: 135.

4. Osterhout, W. V., et al.

1927. J. Gen. Physiol. 11: 193.

HOBKR: THE MEMBRANE THEORY 391

5. Graham, J., G. R. Carlson, & R. W. Gerard

1942. Fed. Proe. 1: 31.

6. Hober, R., M. Andersch, J. Hober, & B. Nebel 1939. J. Cell. Comp. Physiol. 13: 195.

7. Hober, R. Unpublished Data.

8. Hober, R.

1917. Pfliig-. Arch. ^es. Physiol. 166: 531.

9. Bear, R. S., & F. O. Schmitt

1939. J. Cell. Comp. Physiol. 14: 205, 351.

10. Cowan, S. L.

1934. Proc. Roy. Soc. London B 115: 216.

11. Fenn, W. O., et al.

1934. Am. J. Physiol. 110: 74.

12. Labes, R., & H. Zain

1926. Arch. E.xp. Path. Pharm. 125.

1927. .Vrch. Exp. Path. Pharm. 126.

13. Ebbecke, U.

1933. Ergebn. Physiol. 35: 756.

1922. Pfliig. Arch', ges. Physiol. 195: 555.

14. Blinks, L. R.

1936. .J. Gen. Physiol. 20: 229.

15. Woronzow, D. S.

1924-1929. Pfliig. Arch. ges. Physiol. 203, 207, 210, 216, 223. Hober, R., & H. Strobe

1929. Pfliig. Arch. ges. Physiol. 222: 71.

16. Guttman, R., & K, S. Cole

1941. Proc. Soc. Exp. Biol. & Med. 48: 293. Guttman, R.

1940. .J. Gen. Physiol. 23:343.

17. Arnett, V., & S. Wilde

1941. J. Xeurophysiol. 4: 572.

18. Cole, K. S., & H. J. Curtis 1939. J. Gen. Physiol. 22: 649.

19. Furusawa, K.

1929. J. Physiol. 67: 325. Feng, T. P.

19.32. J. Physiol. 76:477. Shanes, A. M., & D. E. S. Brown

1942. J. Cell. Comp. Physiol. 19: 1.

20. Hoagland, D. R., & A. R. Davis

1923. J. Gen. Physiol. 5: 629.

21. Harris, J. E.

1941. J. Biol. Chem. 141:579.

22. Young, J. Z.

1936. Proc. Roy. Soc. London B 121: 319. Schmitt, F. O.

1936. Cold Spring Harbor Sympos. 4: 7. Pumphrey, R. J., & J. Z. ifoung 1938. J. Exp. Biol. 15:433.

23. Bear, A. S., & F. O. Schmitt

1937. J. CeU. Comp. Physiol. 9: 275. Young, J. Z.

1933. Cold Spring Harbor Sympos. 4:1.

24. Lillie, R. S.

1923. Protoplasmic Action and Nerve Action. Chicago Univ. Press. Chicago, Illinois.

392 ANNALS NEW YORK ACADEMY OF SCIENCES

25. Brown, G. L., & W. Feldberg

1936. J. Physiol. 83: 290.

26. Belfanti, S., A. Contardi, & A. Ercoli 1936. Erg. Enzymforsch. 6: 213.

Schmidt, G., B. Hershman, & S. J. Thannha3uer 1945. J. Biol. Chem. 161: 523.

27. von Muralt, A.

1942. Pflug. Arch. ges. Physiol. 245: 604.

28. Liechti, A., A. von Muralt, & M. Rsinsrt

1943. Hslvet. Physiol. Acta. 1: 79. von Muralt, A.

1945. Experientia 1(5).

IIOBER: THE MEMBRANE THEORY 393

PLATE 1

394 ANNALS NEW YORK ACADEMY OF SCIENCES

Plate 1 Large axon and small axons of stellar nerve of Sepia officinalis. (Young.^^)

Annals N. Y. Acad. Sn.

Vol.. XI.VII. AkT. 4. Pl.xTK 1

I. ax

sax

HOBKR: THE MEMBRAXK THEORY

CHEMICAL MECHANISM OF NERVE ACTIVITY

By David Nachmansohn*

Department oj Neurology, College of Physicians and Surgeons, Columbia Uni- versity, New York, N. Y.

INTRODUCTION

The electrical signs of nervous action were, for a century, the only manifestations studied by neurophysiologists. But the function of a living cell cannot be conceived in purely physical terms. This was clearly expressed by Gasser, when he compared the electric spikes to the ticks of the clock, both being only signs of activity.^ For a thor- ough understanding of the mechanism of nerve activity, a knowledge of the chemical reactions involved is essential. Biophysics and bio- chemistry are, consequently, of equal importance and inseparable in any attempt to solve the problem.

The special function of the nervous system is that of carrying mes- sages from one distant point of the body to another. This process may be subdivided into three successive phases: First, a stimulus reaching a neuron has to initiate an impulse. Second, the impulse once initiated has to be propagated along the axon. Finally, the impulse arriving at the nerve ending has to be transmitted either to a second neuron or to an effector cell. Early in this century, T. R. Elliot had the idea that the third phase, namely, the transmission of the nerve impulse from the nerve ending to the effector cell, may be carried out by a chemical compound released from the nerve ending and acting directly on the second unit. Elliot suggested that adrenaline may be the transmitter of the impulse from the sympathetic nerve end- ing to the effector cell.- He based this idea on the similarity between the action of adrenaline and the effect of stimulation of sympathetic nerves on the effector organ. Similar ideas were advanced subsequently by Dixon and Howell.

In 1921, Otto Loewi found that, following vagus stimulation of an isolated frog's heart, a compound appeared in the perfusion fluid which, when transmitted to a second heart, produced an effect similar to that of vagus stimulation. Accepting the basic idea of Elliot, Otto Loewi concluded that this compound, which was later identified with acetyl-

* Most of the work described in this lecture has been supported by grants from the Josiah Macy, Jr., Foundation and the Dazian Foundation for Medical Research.

(395)

396 ANNALS NEW YORK ACADEMY OF SCIENCES

choline (ACh), is actually released from the nerve ending and acts on the heart cell directly.^ The concept of "neurohumoral" transmission appeared enlightening in the case of the autonomic nerves and was widely accepted among physiologists.

In 1933, Dale tried to extend this idea of a "chemical mediator" of the nerve impulse to the neuromuscular junction and to the ganglionic synapse.* In this case, however, the theory encountered strong opposi- tion. In addition to many contradictions and difficulties, summarized by Eccles,^ there were two main objections. The first was the time factor. The transmission of nerve impulses across neuromuscular junctions and ganglionic synapses occurs in milliseconds. No evidence was available that the chemical process can occur at the high speed re- quired, and Dale admitted this difficulty. The second objection was still more fundamental. According to leading neurophysiologists, the excitable properties of axon and cell body are basically the same. The electric signs of nervous action do not support the assumption that the transmission of the nerve impulse along the axon differs, fundamentally, from that across the synapse.

The idea of a chemical' mediator, released at the nerve ending and acting directly on the second neuron, thus appeared to be unsatisfac- tory in many respects (Fulton'') .

NEW APPROACH

Recognition of two features of nervous action is essential to an under- standing of the problems and the difficulties involved: The high speed of the propagation of the impulse, and the smallness of the energy re- quired. In medullated mammahan nerve, the impulse travels at the rate of 100 meters per second, and the energy required per impulse per gram is less than 1/10 of a millionth of a small calorie. The recording of such an event offers many difficulties, even with the use of specialized physical methods. Only in the last twenty years have really adequate instruments become available for the analysis of physical aspects of nervous function.

It is obvious that the study of the chemical reactions connected with an event of this kind must offer even more serious difficulties. No ade- quate methods are available for directly determining chemical com- pounds appearing in such minute amounts and for such short periods of time. There is, however, another possible approach. Nearly all chemical reactions in the living cell are effectuated by enzymes. The study of enzymes in vitro has elucidated many chemical reactions.

NACHMANSOHN: CHEMICAL MECHANISM IN NERVES 397

known to occur in living cells, which could not be followed by direct chemical determination of the compounds metabolized. Especially for an event occurring with such a high speed as the propagation of the nerve impulse, analysis of the enzyme systems involved appeared to be the most promising approach.

Enzyme studies alone are, however, not sufficient for the elucidation of a biological mechanism, since there are so many simultaneous enzy- matic reactions in the complex system of the living cell. It is necessary to correlate enzyme activities with events in the intact cell recorded by physical methods. The most conspicuous example of such an ap- proach is the development of muscle physiology. Through the pioneer work of A. V. Hill and 0. Meyerhof, many physical and chemical changes have been correlated, and our concept of the mechanism of muscular contraction has, according to an expression of A. V. Hill, gone through a real "revolution."

The question of the role of ACh in the mechanism of nerve activity has been approached by the study of the enzyme systems involved in the formation and hydrolysis of the ester. On the basis of their be- havior in vitro, the activities of the enzymes could be correlated in dif- ferent ways with events in the living cell recorded by physical methods. The facts established show that the original theories of the role of ACh, and, more generally, the idea of "chemical mediation," have to be mod- ified. There is a strong body of evidence that the release and re- moval of ACh is an intracellular process, occurring at points along the neuronal surface and directly associated with the nerve action poten- tial. The agent, however, which transmits the impulse along the axon, as well as across the synapse, is the action potential.*'"^ Some of the most important features of these investigations may be briefly outlined.

I. CHOLINESTERASE

A. Time Factor

ACh is inactivated by the enzyme cholinesterase, which hydrolyzes the ester into choline and acetic acid. The first essential result of the studies of this enzyme has been the evidence of its high concentration in nerve tissue: Significant amounts of ACh may be split in milli- seconds; that is, a period of time of the order required for the passage of a nerve impulse. Consequently, the potential rate of ACh metab- olism is thus sufficiently high to permit the assumption that it parallels the rate of the electric changes and may, therefore, be directly con- nected with the nerve action potential.

398 ANNALS NEW YORK ACADEMY OF SCIENCES

The special case in which this problem of the time factor has been studied and received a satisfactory answer, is the frog's sartorius mus- cle.^° A small fraction of this muscle is free of nerve endings. By determining the concentration of cholinesterase in this part of the muscle, in the part containing nerve endings, and in the nerve fibers, it is possible to calculate the concentration of cholinesterase at the motor end-plates. Since the number of end-plates in a frog's sartorius is known, the amount of ACh which may be split during one milli- second at a single motor end-plate can be calculated. This turns out to be 1.6 X 10" molecules of the ester. About one-third of the enzyme at the motor end-plate appears to be localized inside the nerve ending. On the assumption that one molecule of ACh covers about 20-50 square A, the amount which may be hydrolyzed during one millisecond at one end-plate would cover a surface of 100-250 square microns.

A high concentration of cholinesterase, of an order of magnitude similar to that at motor end-plates, exists at all synapses, whether central or peripheral, mammalian or fish, vertebrate or invertebrate." In mammalian brain, for instance, 10^^ to 10^° molecules of ACh may be activated per gram of tissue during one millisecond. This corre- sponds to about 10-100 millions of square microns of neuronal surface.

These experiments removed one of the chief difficulties from the theory that ACh is involved in the transmission of nerve im- pulses. They established that the ester may be metabolized at the high speed required for a chemical reaction directly connected with such a rapid event.

The difference between synaptic region and fiber is, however, only quantitative. The concentration of cholinesterase is high everywhere in nerves, although it rises at the region of synapses.

B. Localization of Cholinesterase at the Neuronal Surface

The second essential feature is the localization of cholinesterase in the neuronal surface. Direct evidence for this localization has been offered with experiments on the giant axon of squid {Loligo pealii^-) . This axon has a diameter ranging from 0.5 to 1.0 mm. The axoplasm may be extruded and thus separated from the envelope. The envelope is formed of connective tissue, lipoid and plasma membrane. The axoplasm was found to be practically free of cholinesterase. The whole enzyme activity is in the envelope.

This exclusive localization of an enzyme in the neuronal surface has been found only in the case of cholinesterase. Respiratory enzymes are localized nearly completely in the axoplasm.^'' Bioelectric phe-

NACHMANSOHN: CHEMICAL MECHANISM IN NERVES 399

nomena occur at the surface. The high concentration of the enzyme at the surface suggested that ACh may be connected with conduction along the axon, as well as with transmission across the synapse. This view is consistent with the conclusion of neurophysiologists that the mechanism of these two events is fundamentally the same.

II. CORRELATION BETWEEN ENZYME ACTIVITY AND PHYSICAL EVENTS DURING NERVOUS FUNCTION

The high rate of ACh metabolism and the locahzation of the enzyme at the neuronal surface made possible the assumption that the ester is connected with the electrical manifestations of nerve activity. How- ever, suggestive as these facts may be, observations on enzymes, as pointed out before, do not permit an interpretation of the actual role of the substrate. For an understanding of the precise function of an enzyme, its activity has to be connected with events in the living cell which, in the case of nerve, can only be recorded by physical means. Such a relationship has been established in three different ways.

A. Parallelism Between the Voltage of the Action Potential and

Cholinesterase Activity

The first line of investigations in which a correlation between physi- cal and chemical processes was obtained, was in experiments on the electric fish. It was found that the activity of cholinesterase in the electric organ parallels exactly the voltage of the action potential.

The powerful electric discharge in these organs is identical in nature with the nerve action potential of ordinary nerves (A. V. HilP*). The only distinction is the arrangement of the nervous elements, the elec- tric plates in series. The potential difference developed by a single element is about 0.1 volt, which is the same order of magnitude as that found in ordinary nerves. In the species with the most powerful elec- tric organ known, Electrophonis electricus, the so-called electric eel, several thousand elements are arranged in series from the head to the caudal end of the organ. Thus, the voltage of a discharge amounts to 400-600 volts, on the average, and, in some specimens, more than 800 volts have been observed. In Torpedo, another species with a powerful electric organ, the elements are arranged in a dorso-ventral direction. Since it is a flat fish, the number of plates in series usually does not exceed 400 to 500, and, consequently, the discharge is only 30 to 60 volts, on the average.

In 1937, an extraordinarily high concentration of cholinesterase was found in the strong electric organ of Torpedo. In the following year,

400

ANNALS NEW YORK ACADEMY OF SCIENCES

a similar high concentration was found in the electric organ of Elec- trophorus electricus. The organs, in one hour, hydrolyze amounts of ACh equivalent to one to five times their own weight. In the larger specimens, the organs have a weight of several kilograms, so that the

m.

QCh.E. V^ 500- -J9^

400+ 15*2

300--II.4

200" 7.6

100

\

\ o \ \ \

V

3.8

0 Head end

1— 10

20 30-^ cm.

Caudal end

FiGUKB 1. Action potential and cholinesterase activity in the electric organ, specimen no. 1. Length of fish, 51 cm.

Abscissae : distance from the anterior end of the organ in cm. Ordinates: QCh.E. and V/cm. • average QCh.E. from a single piece of tissue. + average QCh.E. values from pieces of the same section. ■ V/cm.

amount of ACh which may be split in these organs may amount to several kilograms per hour or several milligrams in one-thousandth of a second. These are significant amounts. They make possible the assumption that ACh is directly connected with the action potential and may even generate it, for, in this case, the compound must appear and disappear in milliseconds. If speculation were to be excluded, the only means of removing this compound so rapidly would be by enzymatic action. The high concentration of a specific enzyme appeared partic-

NACHMANSOHN: CHEMICAL iM EC HAN ISM IN NERVES 401

ularly significant, in view of the cliemical composition of these organs: They contain 92 per cent of water and only 2 per cent of protein.

In the weak electric organ of the common Ray, the concentration is relatively low. If, in the three species mentioned, voltage and number of plates per centimeter are compared with the concentration of cholin- esterase, a close relationship becomes obvious. ^^' ^®

A more detailed analysis has been carried out on the electric organ of Electrophorus electriciis. This species is particularly favorable for such studies, since the number of plates per centimeter, and conse- quently, the voltage per centimeter, decrease from the head to the caudal end of the organ (plates 2 and 3). The cholinesterase activity

200

400

acH E

Figure 2. Correlation between voltage and cholinesterase activity. The voltage per cm. is plotted against the enzyme concentration. The dotted line is calculated from the data obtained with the method of least squares; the fully drawn line calculated on the assumption that the line goes through the 0 point.

decreases in the same proportion. If the electric changes are recorded and compared with the enzyme activity of the same section, a close parallelism is obtained between voltage and enzyme concentration (figure 1). This is found not only in regard to the variations which occur in the same specimen, but even for the variations between the individuals, which are quite considerable.^^

A great number of experiments have been carried out on fish of various sizes, covering a range of the action potential from 0.5 to 22.0 volts per centimeter. The quotient CH.E./V was found to be 20.7, with a standard deviation of only ±0.7 or 3.7 per cent. The standard deviation for a single measurement is d=5.1 or about 25 per cent. This is good uniformity for a quotient correlating physical and

402 ANNALS NEW YORK ACADEMY OF SCIENCES

chemical data. Of particular importance is the fact that the line correlating the two variables apparently goes through 0 (figure 2). This indicates a direct proportionality. The results are consistent with the concept that the physical and chemical processes recorded are directly associated and, consequently, interdependent.^^ Such a par- allelism has not been found with other compounds or enzyme activities known to be connected with nervous action.

The direct proportionality found between physical and ch,emical events is significant, in view of the changing morphological structure of the electric unit, the electric plate. If all plates were identical in structure, as e.g., in the case of the electric organ of Torpedo, the volt- age and the cholinesterase activity would be expected to be directly proportional to the number of plates. The situation is entirely dif- ferent in the electric organ of Electrophorus electricus, because the structure of the plates shows enormous variations. In spite of all variations of the visible structure, the voltage of each plate is the same, namely, close to 100 millivolts. It has, therefore, to be assumed that the "active membrane," with which the electric manifestations are connected and which is not yet well defined, does not change, but is similar in all plates. The direct proportionality found between voltage and enzyme activity suggests, then, that the physical and chem- ical events may be associated with the same membrane and that they may be functionally interdependent. Here again, the fact is important only in connection with the great number of other observations, espe- cially the extraordinarily high speed of the chemical process, without which the correlation observed would not have the same interest.

Two assumptions appear possible concerning the manner in which ACh may act: It may produce electromotive force directly by action on the surface, or it may decrease the resistance by increasing the permeability of the boundary. Resistance and electromotive force are closely related properties. So far, the evidence from experiments on nerves is in favor of a change in resistance and increased perme- ability. On the basis of alternating current impedance measurements carried out on the giant axon of squid, Cole and Curtis calculated that the resistance drops during the passage of the impulse from 1,000 ohms to about 25 ohms per square centimeter.^^ In experiments on the elec- tric tissue, a comparable drop in resistance was found by Cox, Coates, and Brown. 2° There is no conclusive evidence that electromotive force is actually produced during the passage of the impulse. One possible interpretation on the basis of the material available at present is, there- fore, the assumption that the parallelism found between voltage and

NACHMANSOHN: CHEMICAL MECHANISM IN NERVES 403

ACh metabolism may be due essentially to the effect of the ester on the resistance of the boundary or, which is equivalent, on its permeability.

Thus, we arrive at the following picture of the role which ACh may have in the mechanism of nerve activity : According to the membrane theory which is most widely accepted among physiologists, the nerve is surrounded by a polarized membrane. The polarized state of the membrane is due to a selective permeability to potassium ions which are many times more concentrated inside the axon than outside. Dur- ing the passage of the impulse, the permeability of the membrane to negative ions is increased, and a depolarization occurs. The rapid ap- pearance and removal of ACh may be an event essential for this change in permeability. The depolarized point becomes negative to the adja- cent region, and flow of current results. This flow of current stimu- lates the next following point. There again, ACh is released, and the whole process repeated. The impulse is thus propagated along the axon. Cholinesterase destroys the active ester very rapidly, and the state of polarization may hereby be restored.

At the nerve ending, other factors, like increased surface and de- creased resistance leading to a greater flow of current, may act in ad- dition. But the process is fundamentally identical, the transmitting agent being the flow of current. Whereas, in earlier theories, ACh was considered as a "neurohumoral" or "synaptic" transmitter, i.e., a substance released from the nerve ending and acting directly on a sec- ond neuron, in the new concept it is assumed that the transmitting agent is always the electric current, the action potential, but the re- lease of ACh is necessary for generating the current.

The picture is consistent with the idea of the propagation of the nerve impulse as developed by Keith Lucas and Adrian. It becomes unnecessary to assume that the transmission along the axon differs fun- damentally from that across the synapse. The assumption of a special mechanism at the synapse, different from that in the axon, as empha- sized before, was the chief difficulty which had to be overcome to reconcile the original theory with the conclusions of the electrophysi- ologists. This appeared necessary for any satisfactory answer to the problem. If it is true that physical methods alone are unable to ex- plain the mechanism in a living cell, it is equally true that conclusions based on chemical methods should not be in contradiction to those ob- tained with physical methods, in view of the much higher sensitivity of the latter.

The picture of the transmission of the nerve impulse across the syn- apse is, however, far from being complete, if only the flow of current

404 ANNALS NEW YORK ACADEMY OF SCIENCES

from the nerve ending to the second unit is considered. The observa- tions of Eccles and his associates have shown that the electric current set up by the pre-synaptic impulse initiates in the post-synaptic mem- brane a special junctional potential (end-plate potential or, more gen- erally, synaptic potential-^).

These findings have recently found a morphological correlate by the discovery of Couteaux that the sarcoplasm surrounding the presynap- tic nerve ending has a very peculiar structure.^^ It is similar to that described by several authors of the last century in the electroplasm which surrounds the nerve endings in the electric plates of electric fish and which shows a layer of "electric rods," the "palisades" of Remak, at that particular point.

The biochemical data support the assumption of a high rate of ACh metabolism in the post-synaptic membrane of the neuro-muscular junction. At the motor end-plate of guinea pig gastrocnemius, only one third of the cholinesterase was found to disappear within three to four weeks after section of the motor nerve.^^- ^^ The rest remained there for many months, a long time after the end-plate had been trans- formed into a sole plate. It appears, thus, probable that part of the high concentration of cholinesterase observed at the motor end-plate may be located at the post-synaptic membrane. The observations on the electric organ support the assumption of such a localization. The electric plates which form the electric organ are homologous to motor end-plates. The discharge in these organs can be considered as com- parable to the end-plate potential, that is, a response of the post-synap- tic membrane. The direct proportionality found between the voltage of the discharge and the cholinesterase activity is, therefore, another in- dication for the importance of ACh in the post-synaptic membrane.

Specificity of Cholinesterase

In all the experiments on the activity of the enzyme, it was assumed that cholinesterase is specific for ACh. In such a case, not only is the conclusion justified that the substrate metabolized is ACh, but also, the activity of a specific enzyme determined in vitro may well be used as an indication for the potential rate of metabolism of the substrate oc- curring in vivo.

It appeared imperative, therefore, to demonstrate the specificity of the enzyme for ACh in all those tissues which were used in the inves- tigations leading to the new concept. The ester linkage in ACh shows no peculiar properties. It has, therefore, to be expected that the ester can be hydrolyzed by other esterases and, on the other hand, that

NACHMANSOHN: CHEMICAL MECHANISM IN NERVES 405

cholinesterase can hydrolyze other esters. Specificity, in this case, would be expected on the basis of analogy to be only relative, not absolute: Cholinesterase might be expected to split ACh at a higher rate than other esters, whereas other esterases might be expected to be- have differently. By testing a number of substrates, a pattern has been obtained which makes it possible to distinguish specific cholin- esterase from other esterases. ^^

In the variety of nerve tissues which have been used as basis for establishing the new concept, the enzyme was found to be an esterase specific for ACh: viz., mammalian brain, lobster nerve, squid fiber con-

1001-

lOO 75 2S 3

I 16

NUC CAUD

OX

ACH PRO But me ben trib meBU

^00

a 00

I

KIDNEV

GUINEA PIG

100 2m 371 6 200 108 28

Figure 3. Pattern of cholinesterase (nucleus caudatus of ox) in presence of different substrates compared to that of an esterase (kidney) not specific for acetylcholine.

The columns represent the Q of the substrates, the Q of ACh being 100. Abbreviations: Pr = propionylcholine, Bu = butyrylrholine, Me = acetyl -j8 -methyl choline (mecholyi), Be = benzoyl- choline, Tr = tributyrin, Mb = methyl butyrate.

taining the giant axon, and the electric tissue. All show a similar pattern, typical for cholinesterase. Even then, rigid statements should be avoided. Occasional deviations in one or the other directions may be expected. Recent observations of Richards and Cutcomp^*' have revealed that the cholinesterase of bee brain splits acetyl-/3-methyl- choline at a higher rate than ACh, whereas, otherwise, the pattern was typical for cholinesterase. In contrast, the hydrolysis patterns of the esterase of other organs (liver, kidney, and pancreas) differ greatly from that of cholinesterase (figure 3). The esterase in these tissues shows several variations, but this could be expected, since the physio- logical substrate is unknown, and probably varies in the different or- gans. They should be referred to as unspecified, not as unspecific, esterases, because they may well be specific for substrates not as yet specified. ACh is metabolized at a high rate only in nerve tissue,

406

ANNALS NEW YORK ACADEMY OF SCIENCES

since only there is choline acetylase found. If the esterase in all nerve tissue shows a pattern so distinctly different from that of the esterases of other tissues, it is justifiable to consider this enzyme as a specific cholinesterase.

HOMOGENIZED ACH PR BU ME BE TR.

HDD

M.B.

100 97

22 0 0

M400

IIDQ

100 101 3 31 0 0

1 = 3.000

IIDQ

100 109 1 18 0 0

1:78,000

HDD

100 109 0 26 0 0 0

Figure 4. Rate of hydrolysis of different esters by the cholinesterase of the electric organ of Electrophorus electricus.

The first row gives the data obtained with a homogenized suspension of electric tissue. In such suspensions, 1 mg. of protein splits about 20-40 mgs. of ACh per hour.

The three following rows show the data obtained with increasing degrees of purity, 1 ing. of protein splitting 1,400, 3,000, and 78,000 of ACh per hour, respectively.

Of particular importance is the question of the enzyme present in the electric tissue. The interpretation given for the direct proportionality between voltage and enzyme activity is justified only if the enzyme is exclusively, or almost exclusively, specific cholinesterase. Only in that case can the proportionality be referred to as an interdependence between ACh metabolism and electric manifestations.

The enzyme extracted from the electric organ of Electrophorus elec- tricus has been purified by fractional ammonium sulfate precipitation. A high degree of purity may be obtained in this way. 1 milligram of protein is capable of splitting twenty to thirty thousand milligrams of ACh per hour. By further separation of the proteins by high speed centrifugation (ultracentrifuge) , in collaboration with Dr. K. G, Stern, a degree of purity has been obtained where 1 milligram of protein was able to split eighty thousand milligrams of ACh per hour. If the rates

NACHMANSOHN: CHEMICAL MECHANISM IN NERVES 407

of hydrolysis of different substrates by the highly purified enzyme are compared to those obtained with the homogenized suspension of electric tissue, the pattern obtained remains exactly the same throughout the whole process of purification (figure 4). Both have the pattern char- acteristic for cholinesterase. Thus, the correlation established between voltage and enzyme activity can be consequently referred to a corre- lation between ACh metabolized and voltage developed.

It may be noted that examination in the analytical ultracentrifuge in- dicates that the enzyme is a very large molecule. These are not yet final observations. If they could be confirmed, they would indicate that the turnover number of the enzyme is many milhons per minute and that one molecule of cholinesterase could split one molecule of ACh within a few microseconds.

B. The Energy Source of the Nerve Action Potential

The second line of investigations, in which enzyme activity could be correlated with events in the living cell recorded by physical methods, is based on the energy transformations involved and on thermodynamic considerations.

If the release and removal of ACh are associated with the primary alterations of the nerve membrane during the passage of the impulse, then the primary source of the chemical energy released during the re- covery process should be used for the resynthesis of ACh.

The most readily available source of energy in living cells is that released by energy-rich phosphate bonds. Phosphocreatine, the main "storehouse" of energy-rich phosphate bonds in muscle, is also present in nerves. The electric organ offers a suitable material for investigat- ing the chemical reactions which supply the energ}^ for the action po- tential. Both electric and chemical energy released are within the range of measurement, whereas, in ordinary nerves, such an analysis is difficult.

Measurements carried out on the electric organ of Electrophorus electricus have revealed that the chemical energy released by the break- down of phosphocreatine is adequate to account for the electric energy released by the action potential. The electric energy released exter- nally per gram and impulse, in large eels, was found to be 4 micro- calories. This is the maximum external energy which may be obtained, under the condition that the external resistance is approximately equal to the internal. The total electric energy is about 6 times as high as the external, or about 25 micro-calories. These data were obtained on eels of 170 to 180 cm. length. In medium-sized eels of 90 to 120

408 ANNALS NEW YORK ACADEMY OF SCIENCES

cm. length, the total electric energy released per gram and impulse was found to be 47 micro-calories, on the average. There are some as- sumptions, on which these figures are based, which will be discussed by Drs. Cox, Coates, and Brown. If we consider all probable assump- tions, these figures may possibly be revised downward by 15 per cent or upward up to 100 per cent.

Tested under the same conditions, the energy released by the break- down of phosphocreatine was found to be 32 micro-calories per gram and impulse, in the large eels (average of 15 experiments). In the medium-sized eels, the energy released by phosphocreatine was about 51 micro-calories (average of 15 experiments). The lactic acid forma- tion released about 17 /xcal., in the large, and 53 /xcal. in the medium- sized, eels per gram and impulse. The energy of the lactic acid is prob- ably used to rephosphorylate creatine, just as in muscle where the phosphopyruvic acid transfers its phosphate via adenosine triphosphate to creatine ("Parnas reaction"). The sum of the two reactions may, therefore, be used as indication for the energy released by phosphate bonds. It amounts to 49 /xcal. in the large, and 104 ;u,cal. in the medium-sized, eels. The figures are consistent with the conclusion that energy-rich phosphate bonds are adequate to account for the en- ergy of the action potential.

It appeared crucial to test whether or not energy-rich phosphate bonds are really the energy source of ACh formation. If this be the case, it would show that the energy of the primary recovery process is really used for the resynthesis of the compound which, by its release, supposedly initiates the nerve impulse. It would, therefore, at the same time, constitute a new support for the assumption that the primary "excitatory disturbance" which produces a propagated impulse may, indeed, be the release of the ester.

In confirmation of this assumption, a new enzyme, choline acetylase, could be extracted from brain in cell free solution, which, under strictly anaerobic conditions, in presence of adenosine triphosphate, forms ACh."-3°

The enzyme has been extracted from homogenized brain. From one gram of fresh rat or guinea pig brain, enzyme solutions were obtained which form 150-200 /xg. of ACh per hour. More recently, up to 250 ;ag./g./hr. were obtained.

Presence of eserine and fluoride is necessary to inhibit the action of cholinesterase and adenosine triphosphate, respectively. Inhibition of the latter enzyme is necessary, since, otherwise, the breakdown of adenosine triphosphate occurs too rapidly. Fluoride inhibits this

NACHMANSOHN: CHEMICAL MECHANISM IN NERVES 409

breakdown, but it does not interfere with the transfer of energy-rich phosphate bonds, as has been shown by Ochoa.^^

The enzyme has also been extracted from powder of acetone dried brain.2^'^° Extracts prepared from one gram of powder form 1.0-2.0 mgs. of ACh per hour. Since acetone inactivates chohnesterase, this enzyme is largely, or sometimes completely, inactivated in the extracts prepared from powder of acetone dried brain, so that addition of eserine may have either a small effect or practically none on the formation of ACh. Adenosine triphosphate is also removed in extracts from acetone dried brain. No addition of fluoride is, therefore, required. For in- stance: In one experiment, 820 ^g. of ACh were formed per gram and hour, with no eserine 780 ^g., and without fluoride 810 ftg. It has, thus, been demonstrated that the enzyme mechanism responsible for the formation of the ester is not identical with the hydrolyzing enzyme.

The enzyme requires the presence of potassium in high concentration, close to that found in brain. It contains active sulfhydryl groups which are readily inactivated by monoiodoacetic acid or copper in low concentration. The — SH groups are easily oxidized by air. On di- alysis, the enzyme rapidly loses its activity. Addition of potassium ion and 1 ( + ) glutamic acid or cysteine reactivates partly. 1 ( + ) alanine, also, has some effect; other amino acids have either a weak effect or none. Citric acid has an effect nearly as strong as glutamic acid, whereas dicarboxylic acids have practically no effect.^^- ^°

The longer the dialysis is carried on, the weaker is the reactivation by the compounds mentioned. The experiments suggest that choline acetylase requires, a coenzyme for its activity. The coenzyme has now been found. In contrast to the enzyme which occurs only in nerve tissue, the coenzyme has been extracted from brain, liver, heart, and skeletal muscle (Nachmansohn and Berman^^). The coenzyme has been purified to a certain degree by treatment with barium salt, which precipitates the coenzyme. The purification, however, is still in progress. The coenzyme not only reactivates the dialyzed enzyme, but increases considerably the undialyzed enzyme preparations. Marked activation has been obtained in this way, especially in extracts from lobster nerve, rabbit's optic nerve, ajid electric tissue. Of special interest is the evidence for the presence of choline acetylase in the optic nerve. The possibility of a role of ACh in sensory nerves has been a matter of controversy for many years, since the ester was not found in such nerves, whereas chohnesterase is present in concentra- tions in an order of magnitude similar to that in motor nerves. The presence of choline acetylase in the optic nerve is further support for

410 ANNALS NEW YORK ACADEMY OF SCIENCES

the assumption that ACh may have the same function there as in other nerves.

The oxidation products of amino acids, i.e., a-keto acids, have a strong inhibitory effect on the formation of ACh, when present in con- centrations of 10-^ to 10"* M. So far, pyruvic, phenylpyruvic, oxy- phenyl pyruvic acid, and a-keto glutaric acid have been tested.^^' •'*°

ACh formation has also been studied in extracts prepared from peripheral nerve fibers, in order to determine whether or not choline acetylase is present in the peripheral fibers, as well as in brain. •■'^' ^* This should be the case, if the new concept of the role of the ester in the axon is correct. It has been found that choline acetylase may be ex- tracted from peripheral nerve fibers, as well as from brain. The rate of formation of ACh in extracts prepared from the sciatic nerve of the rabbit was found to be 70 to 90 /xg. per gram and hour. The sciatic contains a large amount of inactive tissue (connective tissue, fat, and myelin). On the assumption that this tissue forms about two-thirds of the total weight, which is a conservative estimate, the amount of ACh which can be formed in the axon of the rabbit sciatic may, thus, be about 250 fxg. per gram per hour, and is probably higher.

It appeared of special interest to determine the activity of choline acetylase during degeneration, and to test how this metabolism is re- lated to the nerve function, i.e., to conductivity. Conduction is still maintained two days after section, whereas, after three days, it has disappeared. If the release of ACh is responsible for conductivity, formation of ACh should be possible at a rate not too far below normal, as long as the nerve is capable of conducting.

Forty-eight hours after the section of the sciatic, choline acetylase activity has decreased only about 20 to 25 per cent. After seventy- two hours, when conductivity has disappeared, the decrease is marked, but still about one-third of the enzyme is present. The results are consistent with the assumption that enzyme mechanism is required for conduction.

C. Nerve Action Potential and Inhibition of Cholinesterase

In a third line of investigation, cholinesterase activity and nerve action potential could be directly correlated in experiments on the peripheral axon. One of the essential facts in support of the theory of "neuro-humoral" or "synaptic" transmission was the observation that ACh, when applied to synaptic regions, may have a stimulating ac- tion. No action has yet been obtained with the ester, when applied to the axon. Lorente de N6^^ kept bullfrogs' sciatic nerve in a two

NACHMANSOHN: CHEMICAL MECHANISM IN NERVES 411

gram per cent solution of ACh for many hours, and did not find any effect on conductivity. He considers his failure to obtain an effect on the axon by ACh as proof against the new concept of the role of ACh in the mechanism of nerve activity. ACh is a quaternary ammonium salt. Such compounds are completely ionized and usually lipoid insoluble. Generally, they do not penetrate the lipoid membrane. Therefore, these compounds can be expected to have no effect on the axon, since axons are always surrounded by a lipoid membrane, even though it may be rather thin.

1

i

V

A

'iT

Figure 5. Effect of eserine on single fiber action potential (giant axon of squid).

Left: eyerine 0.002M, records (from top to bottom) at 0', 10', 25' (conduction abolished), 35'; sea water at 26'. Conduction distance less for last record, too short to demonstrate latency effect.

Right: eserine O.OIM, records at 0', 15', fiber then rinsed, and axoplasm analyzed chemically. Upper time scale applies to this experiment, lower to column at left, both 1000 c.p.s.

The problem has been approached in a different way: If ACh is the depolarizing agent and if the function of cholinesterase is to remove the active ester, so that polarization again becomes possible after the passage of the impulse, then inhibition of the enzyme should alter, and, in sufficiently high concentration, abolish, the nerve action potential. ^"^

Eserine is known to be a strong inhibitor of cholinesterase. This compound is a tertiary amine and may, therefore, if undissociated, penetrate the lipoid membrane. Experiments carried out on the giant

412 ANNALS NEW YORK ACADEMY OF SCIENCES

axon and on the fin nerve of squid have shown that eserine alters, and, in higher concentrations, abohshes, the nerve action potential. Within a few minutes in eserine, amplitude, length, and duration of the action potential recorded with the cathode ray oscillograph are markedly changed, and in 20 to 25 minutes, the conductivity has been abolished (figure 5) . When the nerves are put back into sea water, they c^uickly recover, and conductivity reappears. The reversibility of the effect is consistent with the fact that the inhibition of cholinesterase is easily reversible in vitro.

Strychnine, another inhibitor of cholinesterase, was also found to alter, and, in higher concentrations, to abolish, the nerve action poten- tial reversibly.

-^1

'/-

J

'^r

Figure 6. Effect of prostigmine on single fiber action potential (giant axon). Records before and after 45' in O.OIM.

Thus, a new relationship has been established between enzyme activ- ity and nerve action potential, in this case using the peripheral axon.

Prostigmine has, in vitro, the same effect as eserine, but it has no effect on the nerve action potential (figures 6 and 7). Prostigmine is like ACh, a quaternary ammonium salt, and it cannot penetrate the lipoid membrane. This has been demonstrated by the following ex- periment. The axoplasm of the nerves kept in eserine was extruded, and the presence of the compound was tested by the inhibitory effect on a purified cholinesterase solution. Even in thousand-fold dilution, the axoplasm from a portion of a single axon showed, by the inhibition

NACHMANSOHN: CHEMICAL MECHANISM IN NERVES 413

of esterase, easily detectable quantities of eserine. The axoplasm of nerves kept in prostigmine had no inhibitory effect on cholinesterase, even when undiluted.

Prostigmine, like ACh, has 3 methyl groups attached to the nitrogen. Drs. Bronk and Acheson have offered evidence that tetraethylammo- nium chloride acts on medullated nerve and, therefore, presumably en- ters it. This compound is also a quaternary ammonium salt and com-

MiNo

MS.

Figure 7. Effect of prostigmine on fin nerve. O.OIM. Traced from enlarged photographs.

Records before and after 83', 205', and 370' in

pletely dissociated. Although it is true that ionized compounds are not readily soluble in lipoids, the properties which decide lipoid solu- bility are far from well established. Frequently, when, in a com- pound, the ratio of C over N is increased, it becomes more lipoid soluble. In tetraethylammonium chloride, there is four times as much carbon as in tetramethylammonium chloride. The change from

414 ANNALS NEW YORK ACADEMY OF SCIENCES

methyl to ethyl affects profoundly the physico-chemical properties of a molecule. The difference between methyl and ethyl alcohol is well known, and need not be discussed here.

Nearly half a century ago, Michaelis showed that, if, in a certain dye, the ethyl groups were substituted by methyl groups, no staining in- side the living cell could be obtained. ^^ Since these groups are not part of the molecule which has the staining properties, the loss of staining power may be due to the impossibility of penetrating the cell, due to the substitution performed. It appears not surprising that a compound with 4 ethyl groups becomes lipoid soluble, in spite of nearly complete dissociation.

The inability to penetrate the lipoid membrane may explain why ACh and prostigmine, applied externally, act only on nerve endings which do not have a myelin sheath, but are inactive when applied to the axon. Only in electric tissue may the power of ACh to produce an action potential be demonstrated. Injection of ACh leads to changes in potential of the same direction as those observed during the discharge.^^ Electric tissue, however, is an accumulation of end- plates which, in contrast to the axons, are not protected by myeline and, therefore, do react. This may also be the explanation for the famous observation of Claude Bernard on the effect of curare, since, according to recent observations, the active principle of curare is a quaternary ammonium salt.^^' *°

The peculiar ability of the synapse to react to injected ACh can no longer be referred to a difference in the fundamental physico-chemical process underlying the propagation of the nerve impulse, but to the difference in histological structure.

Effect of Di-Isopropyl Fluorophosphate (DFP)*

Recently, a new inhibitor of cholinesterase, di-isopropyl fluoro- phosphate (DFP), became known, which can inhibit cholinesterase irreversibly. Tested on the fin nerve of squid, the compound has the same effect on the action potential as was observed with eserine, and at about the same concentration.^^ When the nerve is kept in a solu- tion of 2 mgs. of DFP per cc, the action potential is completely abol- ished in about 30 minutes. When the nerve is put back into sea water, the action potential comes back (figure 8) . These experiments sug- gested that, for relatively short periods and at low temperature, around 20° C, the inactivation of cholinesterase by DFP may be partly re-

* Most of the observations reported in this paragraph were carried out after the conference, but, since the effect of fluorophosphate on the action potential and its mechanism played an important role then, it appeared desirable to include these data in this paper.

NACHMANSOHN: CHEMICAL MECHANISM IN NERVES 415

versible. The cholinesterase activity in squid nerves, under the ex- perimental conditions used, could not be determined, since the season was too advanced, and no squids were available. Experiments were therefore carried out with the abdominal nerve cord of the lob- ster. This nerve preparation has a high cholinesterase activity and relatively satisfactory action potentials. The potentials recorded were exclusively those of the giant axons of the cord. The transmission across the synapses in the ganglia does not, therefore, enter into the picture.

Ftgurr 8. Effect of DFP on thp action potentinl of the fin nerve of sai"H. DFP 0.013M. First two records (from top to bottom): before, and after, 35' in DFP. The last two records : recovery after 60' and 215' in sea water.

When the nerve is immersed in a solution of 2 mgs. of DFP per cc, the action potential disappears within about 30 to 40 minutes, as in the case of squid nerve. If the nerve preparation is put back into sea water, the action potential reappears after some time. Nerves kept in DFP for additional periods after the abolition of the action potential show less complete recovery. Exposure of the nerve to DFP, for 90 minutes after the disappearance of the action potential, abolishes the response irreversibly.

Determinations of cholinesterase in these nerves reveal a striking parallelism between the recovery of the action potential and the re- appearance of cholinesterase (figure 9). The less complete the re-

416

^A^A^^L-S NEW YORK ACADEMY OF SCIENCES

h

-^^Iw* -> "'\L^ V*iv^ -r^Kr

'■' r

- v>

'IJ^

Figure 9. Reversibility of action potential and reappearance of cholinesterase in nerves ex- posed for varying periods of time to DFP, 0.013M.

The nerve whose action potentials are shown in Column 1 was transferred to sea water imme- diately after the action potential was abolished, and washed for one hour. The nerves of Column 2 to 4 were kept in DFP for 30', 60', and 90', after the action potential had disappeared, and then washed in sea water. The top line of each column shows the action potential in the untreated nerves. The second line shows the abolition of the response by DFP. The third line shows the degree of recovery after washing the nerve. The reappearance of cholinesterase activity is shown in the vertical bars of the fourth line. The CO2 output is 233, 129, 88.5, and 50 cmm. per 100 mgs. per hour.

covery of the action potential, the smaller is the amount of cholin- esterase activity. Even after complete and irreversible abolition of the action potential, a small amount of enzyme activity may still be detected. The experiments indicate that cholinesterase inhibition by DFP of cold-blooded animals is partly reversible, for a certain period of time.

This has been confirmed by observations on in vitro inhibition of cholinesterase solution. DFP was added, in two different concentra- tions, 0.1 and 0.5 [xg. per cc, to cholinesterase solution prepared from electric tissue. At the low concentration of DFP, the enzyme solution

NACHMANSOHN: CHEMICAL MECHANISM IN NERVES 417

liberated 520 cmm. CO2, instead of 790 cmm. CO2 per hour without DFP. The activity decreased slo"wly, over a period of hours. In the solution exposed to the stronger concentration of DFP, the activity was only 25 per cent and was nearly completely abolished after 30 minutes. If, after varying periods of exposure of the enzyme solution to DFP in greater concentration, the solution was diluted, part of the activity could be retained for a period of two to three hours (figure 10) . These experiments give additional evidence that the irreversible inhibition of cholinesterase by DFP is a slow process at low tem- perature.

MINUTES

Figure 10. Reversibility of cholinesterase inhibition bv DFP in vitro, tested by dilution effect, t = 9° C.

The cholinesterase solutidn used liberates 790 cmm. CO2 per hour. + — + Activity found in presence of 0.1 /ig. of DFP per cc. O — ^ O Activity found in presence of 0.5 fig. of DFP per cc. • — • Activity found after exposure to 0.5 /ig. of DFP per cc, for varj'ing periods of time,

and subsequent dilution to 0.1 ng. per cc. The part with the dotted lines indicates

the reversibility as a function of time.

Dr. Oilman presented observations on bullfrogs, in which it was found that, following injection of DFP, the action potential of the sciatic nerve may persist in the apparent absence of cholinesterase. The bullfrog sciatic nerve contains extremely small amounts of cholin- esterase. 100 mgs. of nerve (wet weight) liberate 40-50 cmm. CO2 per hour. Observations on lobster nerve indicate that the enzyme is present in about five times excess, since about 80 per cent may be re- moved while the action potential is unaffected. Even if, in the bull- frog sciatic nerve, the excess of enzyme is smaller when part of the

418 ANNALS NEW YORK ACADEMY OF SCIENCES

activity disappears, the measurement of the CO2 liberation falls into a range where precise evaluation becomes difficult. Moreover, in such a preparation, the retention of CO2 by the protein becomes an impor- tant factor. Finally, even in the thin lipoid membrane of the lobster nerve after prolonged washing, sufficient excess of DFP is retained to inhibit 20-40 per cent of the remaining esterase activity. At least this amount, if not more, may be retained in the relatively greater amount of myelin and fat in the bullfrog sciatic nerve. When this nerve is then ground, the retained DFP may come in contact with the cholinesterase and destroy a considerable fraction of the enzyme still present in the intact nerve.

DISCUSSION

It may be of interest to discuss the neuro-humoral theory in the light of recent developments, and to analyze the two basic experiments which form the main support for the hypothesis that the ester is actually liberated at the nerve ending and, having crossed the synapse or motor end-plates, acts directly on the second neuron or on the muscle fiber. The two observations are: (1) The stimulating action of ACh when applied to synaptic regions; (2) the appearance of the ester in the perfusion fluid, following nerve stimulation. It has just been explained why the effect of ACh applied externally is limited to the nerve ending. In any case, a stimulating effect is not necessarily a physiological effect, but may well be a pharmacological one. The same action may, indeed, be produced by other compounds. The observation of Otto Loewi that a compound appears in the perfusion fluid, following nerve stimulation, was important because it suggested that the compound may be con- nected with nerve activity. The importance of this observation need not be minimized because a quarter of a century later the original in- terpretation has to be changed. In fact, by the new development, the role of ACh became more general and more important than could orig- inally have been anticipated. The appearance of a compound in the per- fusion fluid, however, is not sufficient evidence for concluding that the compound acts outside the cell. Many compounds of intermediate cell metabolism may appear outside the cell. This is due to the fact that all enzymatic reactions follow a logarithmic curve. Therefore, if even the greatest part of a compound is rapidly metabolized by the intracellular enzymes, a small fraction may persist long enough to escape enzymatic action and leak out from the cell. This, apparently, may happen also to ACh, in spite of the high concentration of cholin-

NACHMANSOHN: CHEMICAL MECHANISM IN NERVES 419

esterase inside the cell, particularly when some kind of damage of the surface membrane is produced, as may be expected in the case of pro- longed perfusion or in other unphysiological conditions affecting either the membrane permeability or the cholinesterase activity.

In order to verify the assumption that the amount of ACh actually released from the nerve ending is sufficiently high to produce a stim- ulating effect on the second unit, Dale and his associates attempted two sets of experiments. They determined the minimum required to produce a stimulus and compared it to the amounts released. How- ever, in both cases tested, a puzzling discrepancy was found: In the case of the superior cervical ganglion, only 1/40,000 of the amount of ACh necessary to produce a single response appeared in the perfusion fluid per impulse. In the case of the muscle, only 1/100,000 of the amount of ACh necessary to produce a single twitch was collected. This difference is so considerable that the observations cannot be con- sidered as evidence for the idea that ACh is the direct transmitter of the impulse, especially in view of all the other obstacles.

The situation is further complicated by the fact that these infinitely small amounts of ACh can only be found in presence of eserine which should inhibit their destruction. The enzyme located at the neuro- nal surface forms a barrier for the crossing of the ester. Even without regarding the existing discrepancy, it is very difficult to believe that, under physiological conditions, that is, in the absence of eserine, the small amounts of ACh released can cross the barrier and still arrive in sufficient concentrations for producing a response. The small amounts found under these conditions are easily explained if we as- sume that ACh is released inside the cell, and that the amounts which appear in the perfusion fluid are those which have escaped hydrolysis and have been preserved, due to the presence of eserine.

Another question on which some comments may be useful, is that of the difference between the rates of ACh formation and hydrolysis. There are two instances in which these two rates may be compared on the basis of experimentally established data: The guinea pig brain and the rabbit sciatic nerve. In the first case, about 200 to 250 ;u,g. of ACh may be formed, whereas about 70 mgs. may be split per gram per hour. The rate of cholinesterase activity is, thus, about 300 to 350 times higher than that of choline acetylase. In rabbit sciatic nerve, the figures are about 100 /xg. per gram per hour and 15-20 mgs. per gram per hour, i.e., the rate of hydrolysis is about 150-200 times

420 ANNALS NEW YORK ACADEMY OF SCIENCES

as high as that of synthesis. It is doubtful whether these figures in- dicate the real difference of the possible rates of the two enzymes. Cholinesterase is an extremely stable enzyme. Its activity is deter- mined in a well ground and homogenized suspension of the tissue. It appears probable that the maximal possible activity is actually meas- ured in vitro. This is almost certainly not the case with choline acetylase. The enzyme is an extremely labile and a rather complex system which has to be extracted from the tissue. During the prepara- tion, part of the activity may have been lost. We do not know whether the conditions used at present are optimal or even close to optimal. Although the enzyme was discovered three years ago, the rates of formation obtained are still continuously increasing, since more and more factors are becoming known which activate the enzyme (Nachmansohn and Berman, unpublished experiments). In such a case, it is possible and, in fact, probable, that the activity in vivo may be considerably higher than that observed in the solution. A sharp distinction has, moreover, to be made between the potential and the actual rate. Rates of enzymes measured in vitro are potential rates. The actual rates in the living cell may be entirely different. Many enzymes are present in excess in the cell. An excess of 3 to 5 times above the actual requirement is nothing unusual. A 5-fold excess of cholinesterase above that necessary for function has been recently observed in the case of lobster nerve (Bullock et aL*^) . Other enzymes are in much greater excess, whereas, in some cases, the excess activity appears to be relatively small. Nothing is known, at present, as to whether or not choline acetylase is present in excess. Even if this is the case, it may be much smaller than that of cholinesterase.

For an understanding of the problem, the decisive difference which has to be considered is not the difference of rates, but the difference of function. There is a fundamental difference between the function of cholinesterase and that of choline acetylase. If the release of ACh is an essential event in the alterations of the membrane during the passage of the impulse, then the active ester has to be destroyed within a millisecond or less, so that the resting condition may be restored. Therefore, the enzyme which removes the active ester, cholinesterase, has to be very active, but only during this brief period, and may then be inactive until the passage of the next impulse. The formation of ACh, on the other hand, need not be such a rapid process. It is gen- erally assumed that the active ester is released from an inactive form. This is supported by the fact that the primary energy released during recovery is used for the synthesis of ACh, thus implying that the syn-

NACHMANSOHN: CHEMICAL MECHANISM IN NERVES 421

thesis is a slow recovery process. It is, therefore, not difficult to as- sume that the period required for the formation is longer than that for the hydrolysis of the same amount, according to the kind of nerve, its condition, temperature, and so on.

In the initial phase of nerve stimulation, the preformed ACh would act as a reserve and would make conductivity independent of the rate of ACh formation for a considerable length of time. Even the few /ig. of ACh found per gram of nerve would be sufficient to make possible the passage of several thousand impulses. The actual amount of pre- formed ACh in the living cell may be higher than that found experi- mentally, since it is possible that, during the destruction of the cell, a process necessary for the determination, a considerable part of the preformed ester is destroyed. A nerve should, therefore, be able to respond to stimulation for a considerable length of time, independent of the rate of ACh formation. Only in cases of prolonged stimulation should the rate of formation become the limiting factor. If all pre- formed ACh has been exhausted, and stimuli are applied to mammalian nerve every five milliseconds, then the amount synthesized in the intervals between stimuli should be sufficient for producing the necessary alterations in the membrane when released by a stimulus, and should be equivalent to the amount actually destroyed, during the passage of the impulse, by cholinesterase. Since, in mammalian nerve, the duration of the spike is only 0.5 millisecond and the cholinesterase may have acted only during part of this period, e.g., 0.1 or 0.2 milli- seconds, a difference of 25 to 50 times between the actual rate of cholin- esterase and that of choline acetylase activity would keep the nerve going indefinitely, if this were the only factor involved.

In summary, considering the difference between the rates of cholin- esterase and choline acetylase, we have to keep in mind: (1) that there is a fundamental difference of function; (2) that the cholinesterase activity determined is probably the maximum possible, whereas the choline acetylase activity found in vitro is almost certainly below the optimal rate in vivo; (3) that the excess of cholinesterase may be greater than that of choline acetylase. In view of this situation, the difference between the rates found does not offer any difficulty and, in fact, appears close to that which one would expect of these two enzymes so different in function and properties.

As to the criticism of Dr. Gerard, who resolutely rejects the con- cept presented, some of his main objections may be discussed briefly. (1) The high speed required for any chemical reaction associated with the transmission of the nerve impulse has been considereii-ler-^^P^S

422 ANNALS NEW YORK ACADEMY OF SCIENCES

time, by many leading physiologists as the chief difficulty for any chemical theory. It is gratifying to see that the evidence accumu- lated during the last ten years for the high rate of cholinesterase activ- ity appears to be so impressive that Dr. Gerard now sees in this high speed one of the main difficulties. He calculates, for example, that the ACh preformed, plus that synthesized, could not possibly supply the ester as fast as cholinesterase can split it.

Such an objection would only hold if the whole amount of cholin- esterase present were continuously and fully active. It appears likely, however, that, at any given moment, only part of the enzyme acts and only for extremely brief periods. The differences found between the rates of formation and removal of ACh appear, as pointed out before, to be well within the expected range.

(2) Still more puzzling to Dr. Gerard is the fact that, at the motor end-plate, there is 15,000 times more cholinesterase than in the sur- rounding muscle fiber, since there is no evidence for a great store or synthesis of ACh at this junction.

The difference between muscle fiber and end-plate is interesting, in view of the specialized localization. It is comparable to the distribution found in nerve between *surf ace and axoplasm, which is infinite. In absolute amounts, the ACh which can be metabolized per impulse per end-plate is 0.000002 fx,g. The formation of this amount does not re- quire a particularly powerful synthesizing system nor an intensive respiration. The energy required for the synthesis, even assuming a high frequency, would still amount to less than one per cent of the oxidative energy measured, a deviation which is far below the meas- urable range.

(3) Dr. Gerard assumes that the heat production by the ACh re- leased would amount to 10 per cent of the total heat, whereas the initial heat is only about 3 per cent.

The frog's sciatic nerve is suitable for such a calculation, since here more experimental data are available than in other cases. According to von Muralt, 0.0006 /xg. of ACh is released per gram per impulse.*^ This would amount to about 6 X 10"^ gram calories, which is 0.6 per cent of the total or 20 per cent of the initial heat.

(4) Finally, many other agents and enzymes are present in neurons, like adrenaline, thiamin, adenosine triphosphate, COo, and many others. Dr. Gerard asks how we can reasonably select ACh and assign to it alone an essential role in conduction of the nervous impulse. Un- doubtedly, there are other compounds and enzymes playing an essential role in nerve activity. The ACh cycle is evidence for that. But none

NACHMANSOHN: CHEMICAL MECHANISM IN NERVES 423

of these compounds shows the typical features of the ACh system, like the high speed, the exclusive localization in the surface, the parallelism with voltage, etc. These unique features of the ACh system make it possible to associate the ester more closely with the action potential !* than all other agents so far known.

SUMMARY AND CONCLUSION

In view of the complex nature of biological mechanisms, one or two facts, however well established and suggestive, would not be sufficient for any theory. However, if a great number of facts point in the same direction, then they support each other and potentiate the value of each of them. The essential facts established may be summarized: (1) The high concentration of cholinesterase in nerve tissue makes possible the removal of ACh at a speed comparable to that of the electric mani- festations. (2) Cholinesterase is localized everywhere at the neuronal surface where the bioelectrical phenomena occur. The exclusive local- ization in the surface contrasts strikingly with the localization of other enzymes. (3) Cholinesterase in nervous tissue (and in muscle) is distinctly different from all other tissue esterases occurring in the body. The enzyme, present in all types of nerves throughout the entire animal kingdom, shows similar properties. (4) A direct proportionality be- tween voltage and cholinesterase activity has been established in the electric organ of Electrophorus electricus. (5) The primary energy source of recovery after the passage of the impulse, namely, the energy- rich phosphate bonds of adenosine triphosphate, is used for ACh syn- thesis. (6) The formation of ACh by choline acetylase occurs at a high rate in the peripheral fibers, as well as in the brain. The enzyme has, so far, been found exclusively in nerve tissue. (7) Anticholin- esterases alter, and, in high concentrations, abolish, the nerve action potential. The abolition of the action potential is reversible, if the inhibition of cholinesterase is reversible; irreversible inhibition of cholinesterase abolishes the nerve action potential irreversibly.

These facts considered altogether make it highly probable that the release and removal of ACh is an intracellular event, directly associated with the nerve action potential.

The precise function of the ester is still a matter of interpretation. On the basis of the physical and chemical data available, one possible interpretation appears to be that the ester plays an essential role in the breakdown of the membrane resistance, occurring during the pass- age of the impulse. New facts may change the situation. A number

424 ANNALS NEW YORK ACADEMY OF SCIENCES

of questions still have to be answered before a satisfactory picture of the chemical mechanisms of nervous action can be obtained.

BIBLIOGRAPHY

1. Erlanger, J., & H. S. Gasser

1937. Electrical Signs of Nervous Activity. University of Pennsylvania Press. Philadelphia.

2. Elliot, T. R.

1905. J. Physiol. 32:401.

3. Loewi, O.

1932-1933. Harvey Lectures. 218.

4. Brown, G. L.

1937. Physiol. Rev. 17:485.

5. Eccles, J. C.

1937. Physiol. Rev. 17: 538.

6. Fulton, J. F.

1938, 1943. Physiology of the Nervous System. Oxford Univ. Press. New York.

7. Nachmansohn, D.

1945. In: Harris, R. S., & K.V, Thimann. Vitamins and Hormones. Aca- demic Press. New York. 3: 337.

8. Fulton, J. F., & D. Nachmansohn 1943. Science. 97:569.

9. Nachmansohn, D.

1946. In: Green, D. E. Currents in Biochemical Research. Interscience Pub- lishers. New York. 335.

10. Marnay, A., & D. Nachmansohn

1938. J. Physiol. 92:37.

11. Nachmansohn, D.

1940. Yale J. Biol. Med. 12: 565.

12. Boell, E. J., & D. Nachmansohn 1940. Science. 92: 513.

13. Nachmansohn, D., H. B. Steinbach, A. L. Machado, & S. Spiegelman 1943. J. Neurophysiol. 6: 203.

14. Hill, A. V.

1932. Chemical Wave Transmission in Nerve. Cambridge University Press. London.

15. Nachmansohn, D.

1940. Science. 91: 405.

16. Nachmansohn, D., & B. Meyerhof

1941. J.. Neurophysiol. 4 : 348.

17. Nachmansohn, D., R. T. Cox, C. W. Coates, & A. L. Machado 1913. J. Neurophysiol. 6: 203.

18. Nachmansohn, D., C. W. Coates, & M. A. Rothenberg 1946. J. Biol. Chem. 163: 39.

19. Cole, K. C, & H. T. Curtis

1939. J. Gen. Physiol. 22:649.

20. Cox, R. T., C. W. Coate?, & M. V. Brown

1945. J. Gen. Physiol. 28: 187.

21. Eccles, J. C.

1943. J. Physiol. 101:465.

22. Couteaux, R.

1944. C. R. Soc. biol. 138:976.

23. Couteaux, R., & D. Nachmansohn

1940. Proc. Soc. Exp. Biol. & Med. 43: 177.

NACHMANSOHN: CHEMICAL MECHANISM IN NERVES 425

24. Couteauz, R.

1942. Bull. biol. 76: 14.

26. Nachznansohn, D., & M. A. Rothenberg 1945. J. Biol. Chem. 158: 653.

26. Richards, A. G.. & L. K. Cutcomp 1945. J. Cell. Comp. Physiol. 26: 57.

27. Nachmansohn, D., & A. L. Machado

1943. J. Neurophysiol. 6: 397.

28. Nachmansohn, D., H. M. John, & H. Waelsch

1943. J. Biol. Chem. 150: 485.

29. Nachmansohn, D., & H. M. John

1944. Proc. Soc. Exp. Biol. & Med. 57: 361.

30. Nachmansohn, D., & H. M. John

1945. J. Biol. Chem. 158: 157.

31. Ochoa, J.

1941. J. Biol. Chem. 138: 751.

32. Nachmansohn, D., & M. Berman

1946. J. Biol. Chem. 165:551

33. Nachmansohn, D., & H. M. John

1945. Science. 102:250,

34. Nachmansohn, D., H. M. John, & M. Berman

1946. J. Biol. Chem. 133: 475. 36. Lorente de No, R.

1944. J. Cell. Comp. Physiol. 24: 85.

36. Bullock, T. H., D. Nachmansohn, & M. A. Rothenberg

1946. J. Neurophysiol. 9: 9.

37. Michaelis, L.

1900. Arch. mikr. Aiiat. 55: 565.

38. Feldborg, W., A. Fessard, & D. Nachmansohn 1940. J. Physiol. 97: 30.

39. King, H.

1935. J. Chem. Soc. London 2: 1381.

40. Wintersteiner, O., & J. D. Dutcher

1943. Science 97: 467.

41. Bullock, T. H., H. Grundfest, O. Nachmansohn, M. A. Rothenberg, & K. Sterling

1946. J. Neurophysiol. 9:253.

42. von Muralt, A.

1945. Experiential:!.

Annals N. Y. Acad. Sn.

Vol. XI.VII, Akt. 4, Pi.atk -2

5-6

H if- - ■-^■'■}fS^: '

40-41 •

55-56

NACHMANSOHN; CHKMICAL MKCHANISM OF NKRVK ACTIVITY

NACHMANSOHN: CHEMICAL MECHANISM IN NERVES 427

Plate 2

Changes of the morphological structure of the electric plates at different sec- tions of the electric organ of Electrophorus elcclricus.

The specimen used for this section was 114 cm. long. All sections are repro- duced with the same magnification (X 145). The numbers below each section indicate the distance in cm. from the anterior end of the organ. H — bead end, C ~ caudal end.

428 ANNALS NEW YORK ACADEMY OF SCIENCES

Plate 3

Changes of the morphological structure of the electric plates at different sec- tions of the electric organ of Electrophorus electncus.

The specimen used for this section was 57 cm. long (Xl45). The numbers below each section indicate the distance in cm. from the anterior end of the organ. H = head end, C = caudal end.

Annals N. Y. Acad. Sri.

Vol.. XI.VII. Aht. 4, Plate 3

H

3-5

NACHMANSOHN: CHEMICAL MKCHA.MSM OK NKRVF. ACIIVITY

.V

AN ELECTRICAL HYPOTHESIS OF SYNAPTIC AND NEURO-MUSCULAR TRANSMISSION

By J. C. EccLES

University of Otago, Dunedin, New Zealand

1. PRESENT THEORETICAL POSITION

This paper will be restricted to the synapses of ganglia and the spinal cord and to the neuro-muscular junctions of skeletal muscle (hence- forth, collectively referred to as synapses), because, physiologically, they form a fairly homogeneous group. A preliminary report has al- ready been published.^* There is now good evidence that the trans- mission of impulses, at all these synapses, is mediated by catelectrotonic potentials set up at the synaptic membrane of the post-synaptic cell — the end-plate potentials of skeletal muscle-^' ^^' ^*' ^^' ^°' " and the synaptic potentials of ganglion cells^"' ^^ and motoneurones.^' ^^' ^^ We may, therefore, subdivide the problem of synaptic transmission into two Problems: (a) the mechanism whereby impulses in pre-synaptic nerve fibers set up catelectrotonic synaptic potentials in the post- synaptic cell; and (b) the initiation of impulses in the post-synaptic cell by these synaptic potentials. As is well known, the existing hy- potheses relating to Problem (a) are chemical (acetylcholine), or elec- trical, or some combination thereof.*' -''' *^^ Problem (b) has, hitherto, been regarded as just a part of the general problem of impulse initiation by catelectrotonus. However, there is evidence of a unique mechanism in the case of the only synapse worked on in detail.^"

There is some resemblance between these two stages of synaptic transmission and the two "boundary faces" postulated by Buchthal and Lindhard,^' ^ to explain the two stages of neuro-muscular block pro- duced by curare and acetylcholine.

In their existing form, both hypotheses relating to Problem (a) are unsatisfactory :

(i) Originally, the acetylcholine hypothesis simply stated that a pre- synaptic impulse liberated at the synapse a sudden jet of acetylcholine, which excited the post-synaptic cell by acting on specific receptors ;^^ thus set up the synaptic potential, according to present views; and was ickly removed by the locally concentrated cholinesterase.®- ^' "• ^^ The usual failure to detect acetylcholine in venous blood collected from

(429)

430 ANNALS NEW YORK ACADEMY OF SCIENCES

eserinized, stimulated ganglia or muscle suggested the additional hy- pothesis that, normally, acetylcholine is removed by being rebuilt rapidly to a precursor and that cholinesterase merely acts as a barrier, to prevent diffusion of acetylcholine away from the synapse.**^ Further additions to this hypothesis were needed, in order to explain the effects of eserine and curarine on end-plate potentials.* It is unsatisfactory that the acetylcholine hypothesis has had to be reconciled with new experimental evidence, by thus making subsidiary ad hoc hypotheses, which have not been independently testable. The most recent develop- ment of the acetylcholine hypothesis*'^ is essentially a special type of the electrical hypothesis, for it postulates that electrical transmission across the synapse excites the postsynaptic liberation of acetylcholine, which, in turn, sets up the synaptic potential.

(ii) Most expressions of the electrical hypothesis of synaptic trans- mission have merely stated that the electrical currents of the pre- synaptic impulses set up impulses in the post-synaptic cell, much as one segment of a nerve excites the next.^°' ^^' ^°' ^^' '^^ The attempt at pre- cise formulation by Lapicque (the isochronism hypothesis) has had to be modified so much by the recognition of the significance of addi- tional factors (relative durations of the pre- and post-synaptic re- sponses,"'^ the rheobase of the post-synaptic celP^) , that it now states little more than the above vague formulation. Thus, the electrical hypothesis is unsatisfactory (indeed, virtually useless), because it is so vaguely expressed that it fails to give predictions that would be a fertile source of experimental tests.

2. RECENT EXPERIMENTAL SUPPORT FOR AN ELECTRICAL HYPOTHESIS

The need for a more developed electrical hypothesis has now become urgent, because the following recent investigations have indicated that acetylcholine plays but a subsidiary role at ganglionic synapses, and a negligible role at spinal cord synapses. With muscle, too, there are indications that electrical transmission may play an important part.

(i) A detailed study^- -'^ of the electrical responses of eserinized ganglia (normal and curarized) revealed an excitatory action, pro- longed for several seconds after repetitive stimulation and attributable to acetylcholine. However, this prolonged action was so weak that the summation of about 20 volleys in (juick succession was needed to make it sufficiently strong to excite normal, fully eserinized ganglion

• Cf. Eccles, J. C, B. Katz, & S. W. Kuffler.": 227-8.

ECCLES: ELECTRICAL THEORIES OF TRANSMISSION 431

cells to discharge. The usual transmission mechanism was due to an excitatory action, unaffected by eserine and brief enough to be at- tributable to the action currents of the pre-ganglionic impulses. ^^' '^

(ii) A similar study of synaptic potentials of motoneurones excited through mono-synaptic reflex pathways of the spinal cord (frog, cat) has failed to detect even such a subsidiary role for acetylcholine trans- mission.^^ Furthermore, it has been found that synaptic transmission of the frog's spinal cord is unaffected by prolonged soaking (several hours) in high concentrations of acetylcholine (up to 1 in 5,000) . Still higher concentrations have an anesthetic action which, initially, is reversible. The isolated oxygenated cord (anesthetized or unanes- thetized) is soaked for 30 min. in a strong anti-cholinesterase (1 in 50,000 eserine), and then the acetylcholine is added to the solution." These experiments would appear to falsify the hypothesis that acetyl- choline plays a major role in synaptic transmission in the spinal cord. However, too much emphasis should not be placed on these latter ex- periments, until they are repeated with prostigraine as an anti-cholin- esterase (cf. iii, below).

(iii) Just as with sympathetic ganglia,^'' the responses of curarized, eserinized (or prostigminized) muscles to repetitive stimulation are sharply distinguishable into a prolonged end-plate potential which is attributable to acetylcholine, and an initial, very brief, end-plate potential, but little lengthened by anti-cholinesterases.^^' ^'^ It seems probable that, as with ganglia, the small, apparent lengthening of the initial phase by anti-cholinesterases may be attributable to some ad- mixture of the prolonged acetylcholine phase, and that the initial phase may be excited by the action currents of pre-synaptic impulses (cf. PART 8, ii). Acetylcholine blocks neuro-muscular transmission,^' ^ pre- sumably by catelectrotonic blockage, but, despite a relatively high acetylcholine background (1 in 200,000), pre-synaptic volleys still set up large end-plate potentials, even larger than in curarized muscle.^® When performing these experiments by soaking frog's sartorii in acetyl- choline solutions, prostigmine is used as an anti-cholinesterase. Eser- ine is ineffective, probably because acetylcholine competes with it for the cholinesterase."*

3. EXPERIMENTAL BASIS FOR ELECTRICAL HYPOTHESIS

In recent years, important advances have been made in the investiga- tion of nerve and muscle fibers, and an electrical hypothesis of trans- mission must be based on the following evidence:

* Cf. Eccles, J. C, B. Katz, 8c S. W. Kufller.^^: 225.

432

ANNALS NEW YORK ACADEMY OF SCIENCES

A. The Electrical Properties of the Surface Membranes and the

Changes Produced by Catelectrotonus, Anelectrotonus,

Local Responses, and Propagated Impulses

Quantitative measurements liave been made of resistance, electro- motive force, capacity, and rectification. The great diminution of the two former during the excited phase of the impulse has been described for nerve and muscle, vertebrate and invertebrate.^' ^^' "' ^^' ^^' ^^' **'' *'' There is no good evidence that the large inductance of cephalopod nerve^"' ^* is present in normal vertebrate nerve or muscle. The phe- nomenon of para-resonance is simply explained in terms of the two excitation constants of nerve.* In contrast with cephalopod nerve, there is, in frog muscle, no appreciable lag between a sudden change in potential and the associated change in the resistance of the mem- brane.*^ Such a lag in cephalopod nerve has been attributed to the large membrane inductance.^- On present evidence, vertebrate nerve and muscle may, therefore, be regarded as having a negligible induc- tance, and may be provisionally schematized, as in figure 1.

Figure L Diagram showing probable electrical characteristics of ner/e and muscle membranes, rx and n being, respectively, the external and internal longitudinal resistances ; C, E, and R, the capacity, battery, and resistance of the membrane.

B. Local Responses

All grades of active local responses, short of propagated, all-or- nothing impulses, have been shown to exist in nerve and muscle,^' ^''' ^^' 44, 4G, 47, 52, 70, 79 q^^^ jjj^y j^g explalucd as due to the limited area ex- cited"® and/or to the low intensity of the excitation. ^^ In refractory, anesthetized, or deteriorated nerve or muscle, these local responses may be very large.*^' *'■ ®^' ^°' ''^ It appears probable that active local re- sponses differ from passive electrotonic changes (including rectifica- tion), jiist as with the propagated impulse, in that they are caused by a temporary diminution, extinction, or even reversal, of the membrane battery.^' i^' "• '^

* Katz, B.«: 28.

ECCLES: ELECTRICAL THEORIES OF TRANSMISSION

433

C. Reactions of Ephapses (Artificial Synapses)

The double axon preparations^' ■**' *^ are particularly relevant to electrical action across synapses. Since there has been excellent corre- lation between the effects predicted by the "local current" theory of nerve conduction and the effects observed, it may be concluded that these effects are caused by electrical current flow across the ephapse. 48, 49, 64 According to the geometry of the ephaptic contact, three main types of effect are exerted on the resting fiber by an impulse in the ac- tive fiber :^

(i) At regions where fibers are contiguous for some distance on either side — for maximum effect, at least half a wave length. Here the currents generated by the impulse have, in turn, anodal, cathodal, and anodal action on the resting fiber.*^* ''^ Figure 2 shows that the

Figure 2. Diagram of two contiguous fibers, showing the current flow generated by impulse in active lower fiber and its penetration of the resting fiber (cf. Katz & Schmitt, figure 2''*). As the impulse (shown above) propagates along the active fiber, any point on the resting fiber is sub- jected, in turn, to effects A1C1C2A2. Active part of impulse shown by hatched area in this and subsequent figures.

cathodal phase is really double, being due to currents generated by the membrane battery of the active fiber, at first ahead (Ci), and then in the wake, of the impulse (Ca). Thus, the sequence of action is A1C1C2A2, as the impulse sweeps past a point on the resting fiber. As Katz and Schmitt pointed out, the current penetrating and acting on the resting fiber is virtually a mirror image of the penetrating current of the active fiber, and hence, has an intensity-time course correspond- ing to the second derivative of the monophasic action potential, with respect to the time coordinate (d^P/dt^). Since the curvature of the wave front of the impulse is at least twice as sharp as that of its wake, Ai and Ci will be at least twice as large as C2 and A2, will be correspondingly longer in duration.*

• Cf. Katz, B., ft O. H. Sdunltt.*^ Ftgijees 5Aji and 6, Curve 1.

434

ANNALS NEW YORK ACADEMY OF SCIENCES

(ii) At regions where the resting fiber is not affected by the approach of the impulse, but only by its immediate juxtaposition and its propaga- tion. As shown in figure 3a, this occurs when the impulse propagates from an electrically insulated region of the active fiber to a region where it is in contiguity with the resting fiber. Effects Ai and C2 are prevented by the insulation, the interaction being due to effects Ci and

Figure 3a. Diagram showing current flow at junctional zone of two previously separated fibers, (i) Impulse at junction give.s Ci effect on resting fiber; (ii) after further propagation, wake of impulse gives A2 effect.

(i)

(ii)

#i^^

se

E*==F

air.

w^-

Figure 3b. Penetrating current generated by impulse arising in one fiber, (i) Ci effect, as im- pulse is initiated ; (ii) A2 effect, in wake of impulse, propagating in both directions from site of origin.

(i)

(ii)

%' nCi
	1 1		1 1
\	^////

<f-

C^- N^g
	1 1		A 1 1
	y//^	_ _/

Figure 3c. Two fibers connected by double salt bridge, (i) Impulse opposite proximal arm; (ii) impulse opposite distal arm, showing C1A2 action on resting fiber at proximal arm and A1C2 action at distal arm.

A2* A similar effect would also be produced at a region in a passive fiber adjacent to the origin of the impulse in the active fiber (figure 3b), and at the proximal arm of a double salt bridge (figure 3c). ^^

(iii) At regions where the resting fiber is influenced by the approach and juxtaposition of an impulse, but not by its propagation. Figure 4

* Cf . Arvanltakl, A.' Figure 4 III ; Marrazzi, A. S.. 8t R. Iiorente de XTd.^" Figure 3.

ECCLES: ELECTRICAL THEORIES OF TRANSMISSION 435

(also FIGURE 3c, for distal arm of double salt bridge) shows that this occurs under the converse conditions to those above, and that effects

C2.^-~ ~ N

e^^?^^

Figure 4a. Converse diagram to figure 3a, showing current flow at zone of separation of two previously contiguous fibers, fi) Impulse approaching bifurcation gives Ai effect; (ii) impulse at bifurcation gives C2 effect.

(i)

(ii)

t'^l ""^' ^>i^ ^^^,

r^W^

3E

&K

Figure 4b. (i) Two impulses approaching a collision give Ai effect on resting fiber; and (ii) at collision of impulses, 02 effect on resting fiber.

Ci and A2 are prevented, the action being due to effects Ai and C2.* At synaptic regions, a similar electrical action would be exerted by the pre-synaptic impulse on the post-synaptic membrane (cf. figure 5), since the surrounding conducting medium provides a pathway for cur- rents generated by the approaching impulse. There is, of course, no departing impulse {boutons de passage excepted). In this connection, Arvanitaki's results^ are of especial interest, for it was only in ephaptic situations giving A1C2 effects that she observed appreciable local re- sponses of the resting fiber. She concluded that, in all other condi- tions, the terminal A, effect suppressed any active response of the rest- ing fiber.

The approximate time-course of the penetrating current at a synapse may be derived from the monophasic spike potential at the pre-synaptic terminal, by considering the flow of current, much as Katz and Schmitt*^ did for two parallel fibers. The longitudinal current flowing in the external circuit is proportional to the first derivative of the monophasic potential, dP/dt. Immediately proximal to the end of the pre-synaptic fiber, all the longitudinal current is provided by the current penetrating the terminal end. Hence, this penetrating current is also proportional to dP/dt,t and not to d'P/dt^ as occurs along the length of the fiber. A

* Cf. Arvanitakl, A.^ Figure 4 IT.

t Cf. Marrazzl, A. S., & XI. Iiorente de JSf6^: 89.

436

ANNALS NEW YORK ACADEMY OF SCIENCES

similar time-course may be assumed for the current penetrating the closely adjacent post-synaptic membrane (see figure 5). The first derivative of the monophasic potential gives, of course, the expected diphasic effect, A1C2.

Thus, it may be concluded that, so far as they go, ephaptic investi- gations lend support to the hypothesis that an excitatory action would be exerted by impulses terminating at synapses. However, with ephapses, this excitatory action is normally too weak to initiate im- pulses in the resting fiber. For example, Katz and Schmitt*^ find that the maximum C effect is never as much as 20% of threshold, and Arvanitaki^ has to sensitize the resting fiber by decalcification, in order to increase the local response sufficiently for impulse initiation.*

/

T^

\^ ^

/

(a)

(b)

Figure 5. Diagrams of current flow at a schematic synapse with pre-synaptic impulse ap- proaching synapse in (a), and at synapse in (b). Note reversal of current flow, the focal Ai effect being followed by the focal C2 effect at the synaptic region of the post-synaptic membrane.

The Hering effect, ephaptic transmission adjacent to killed or injured regions,^^' *^' " may also be explained as due to A1C2 stimulation of fibers rendered sensitive by the catelectrotonus'* prevailing close to the injured region. Tests of excitability changes, 4 ram. from the killed

* Cf . Arvanitaki, A.> Figure S.

ECCLES: ELECTRICAL THEORIES OF TRANSMISSION 437

end, showed a terminal A2 effect,^* but, presumably, this would dis- appear, closer to the killed end. The ephaptic experiments, in general, show that special conditions must prevail at synaptic contacts, if elec- trical excitation is to be adequate for synaptic transmission (cf. part 8) .

D. Special Properties of the Synaptic Region

So far, such investigation has been restricted to the isolated neuro- muscular junction. When electrical recording is effectively localized to the end-plate region of the muscle, it has been shown that the end- plate potential set up by a nerve impulse rises smoothly to the full height of the spike potential,''" w^ithout showing the sudden inflection characteristic of impulse initiation.*^- ^-' "^ The impulse appears to be initiated, a little later, by an adjacent region of the membrane, when it reaches a critical intensity of catelectrotonus. Progressive curarization progressively diminishes the end-plate potential; the im- pulse initiation occurs adjacently, after the longer delay ensuing before the lower end-plate potential builds up the critical catelectrotonus ; and eventually, transmission fails. It may, therefore, be assumed that the end-plate region of the muscle is speciahzed to give "local responses" of high and graduated intensities, without the sudden incursion of the all-or-nothing "breakdown" of resistance and battery that occurs with impulse initiation. ^° This evidence of unique electrical properties of the end-plate is relatable to its well-known, unique, pharmacological properties. ^^' ^*' ^^ In the isolated preparation, Kuffler^^ has failed to detect the large resting potential (positive or negative) between the surface of the end-plate and that of the muscle fiber that has been described by Buchthal and Lindhard.^

4. INITIAL ASSUMPTIONS OF HYPOTHESIS

The following three initial assumptions of the electrical hypothesis are based on the evidence of the preceding four sections, together with the conventional histological picture (they form, as it were, a model of a synapse whose functional operation will be discussed in part 5) :

A. That the geometrical situation at the synapse may be schematic- ally represented by the pre-synaptic fiber ending as a cylindrical mem- brane, with a closed end in close apposition to the large plane surface membrane of the post-synaptic cell, as is shown in section in figure 5. Histologists are now fairly generally agreed that a transverse mem- brane exists at the synapse,^" and there is also electrical evidence* of a

* Eccles, J. C.2»: 352.

438 ANNALS NEW YORK ACADEMY OF SCIENCES

highly resistant transverse membrane. This evidence for ganglionic synapses also obtains for neuro-muscular junctions.

B. That, in general, the surface membranes of figure 5 have the electrical properties demonstrated for peripheral nerve and muscle membranes: resistance, electromotive force, capacity, and rectification, as shown in figure 1. There are no direct observations on nerve cells, but they resemble nerve fibers in their electrical excitability and in the propagation of impulses from a nerve cell to its axon,^''' ^^' **° and vice versa. ^°' ^^' *^''' "^ It may also be assumed that both the exterior and the interior of the cells are good conducting media, and that the resting potential of the post-synaptic membrane is identical with that for the remainder of the post-synaptic cell.^^

C. That the synaptic region of the post-synaptic cell has unique electrical properties, in that cathodal polarization (lowering of resting charge) sets up a graduated "local response," with a temporarily ir- reversible and large diminution of electromotive force and resistance, but not the all-or-nothing membrane "breakdown" characteristic of the propagated impulse (cf. part 3, B and D). Direct evidence is only available for the end-plate region,^" but the assumption is extended to the synaptic regions of nerve cells.

5. DEVELOPMENT OF HYPOTHESIS ON THE BASIS

OF THIS MODEL

It appears that, assuming A and B, we have to expect that the cur- rent generated by an impulse propagating up to the terminal of the pre- synaptic fiber will, in part, penetrate the post-synaptic cell and give a diphasic action (cf. part 3, C, iii). Firstly, there will be an anodal focus, Ai, at the synaptic region, with a cathodal surround (figures 5a and 6a). Then, when the active region of the impulse reaches the terminal, current flow will reverse, giving a cathodal focus, C2, with an anodal surround (figures 5b and 6b). The penetrating current will be limited by polarization of the membrane and, in the initial phase, by the increasing resistance of the localized anode (rectification effect) . On account of its much larger area, the membrane resistance (and pene- trating current density) at the cathodal surround will be so much lower that its simultaneous diminution by the rectification effect will be relatively insignificant in tending to increase the flow of penetrating current. However, in the second phase, the situation is reversed, be- cause, on account of the high current density, the lowering of the initi- ally high resistance at the localized cathode will have a preponderant effect in increasing the flow of current. Hence, due to rectification, the

ECCLES: ELECTRICAL THEORIES OF TRANSMISSION 439

over-all resistance offered to the penetrating current will be much lower in the second phase than in the first, the current being, as it were, canalized through the localized low resistance at the cathodal focus. It should be noted that, in this way, rectification will diminish, at the synapse, the depressing action of the relatively high intensity Ai, and then increase the stimulating action of the relatively low intensity Cz (cf. FIGURE 7b). The effectiveness of this discriminative action of rectification is illustrated for bipolar stimulation, by Cole.* It should be even more effective for the unipolar type of stimulation that occurs at the synapse. It is evident that, if the membrane had a high induct- ance in series with R (figure 1), the brief penetrating currents would be much less intense, and the rectification correspondingly less effective. The polarization of the membrane, in the first and second stages, is shown diagrammatically in figures 6a and 6b. Note the wider spread of anelectrotonus, Ai, than C2, and the reversal of potential gradients along the inner side of the membrane, corresponding to the reversal of the "core currents" (cf. figures 5a and 5b). Note, also, that, at the dotted lines separating the anelectrotonic and catelectrotonic areas, the curves of the inner and outer membrane potentials are inflected, as would be expected for zero density of penetrating current. The catelec- trotonic focus shown in figure 6b will not immediately develop the pre-synaptic current flow reverses. The anodal polarization in figure 6a takes some time to be removed by the local current flow, as well as by the reversed penetrating currents, and further time is needed to charge the membrane condensers to the fully-developed cathodal focus in figure 6b (cf. figure 7b). If, at this latter stage, the external elec- tric field, applied by the impulse in the pre-synaptic terminal, were sud- denly removed, the membrane would immediately revert to the poten- tial distribution of figure 6c (assuming that the internal and external media have equal longitudinal resistances; i.e., that ri and r2 of figure 1 are equal). If no local response is set up, i.e., if the membrane ex- hibits only its electrotonic properties, local current flow would quickly cause the anodal surround to discharge into, and repolarize, the cathodal focus, and the membrane would quickly revert to the normal, uniformly charged, condition. Thus, under such circumstances, with the usual disposition of electrodes for recording responses at the synaptic region (one close to the synapse and one distally on the post- synaptic cell) , there would be recorded merely a brief diphasic poten- tial, attributable to currents generated by the pre-synaptic impulse and but little modified by the passive properties of the post-synaptic cell.

* Cole. K. S." FiCDKE 3.

440

ANNALS NEW YORK ACADEMY OF SCIENCES

(a)

(b)

(c)

(d)

Figure 6. Graphs of spatial distribution of potentials on the outer and inner sides of post- synaptic membrane, the synaptic region being in center, i.e., potentials are ordinates and dis- tances abscissae. The reference potential is given by a distal region of the membrane, the outer side being shown above the inner. The normal resting condition is shown by the broken lines separated by the resting potential, (a) Initial Ai focus at synapse with low intensity cathodal sur- round, (b) Reversed phase with C2 synaptic focus and anodal surround (cf. figure 5b). (c) Membrane potentials, when external field generated by pre-synaptic impulse is removed, the spatial distribution of the potentials across the membrane being identical with those of (b). (d) Potentials after generation of local response at synapse, with the catelectrotonus (the synap- tic potential) spreading thence over the post-synaptic membrane. Hatched area shows specialized synaptic zone of membrane. Impulse initiation occurs outside this zone, for example at the arrow.

There would be no synaptic potential, with its characteristic long duration.

The additional assumption (part 4, C) is necessary, in order to ex- plain the origin of the synaptic potential. It postulates that, when above a critical intensity, the cathodal focus evokes at the synaptic

ECCLES: ELECTRICAL THEORIES OF TRANSMISSION 441

region of the post-synaptic cell an intense ''local response," which runs through a cycle of increasing and decreasing intensity, much as does the spike of a propagating impulse.''^' ^^' ^^^ ^°' '^ This local response, thus, would outlast the second phase of the penetrating current-flow, and provide a relatively enduring focus, of very low polarization (pre- sumably due to diminution or extinction of the membrane battery) and resistance, through which adjacent regions of the post-synaptic mem- brane proceed to discharge. Such a local response is actually ob- served at an ephapse giving A1C2 interaction.^ Figuee 6d shows the internal and external potentials of a fully-developed local response at the end-plate region, where Kuffler^° finds the potential as high as the spike potential. It is shown as zero transverse membrane potential, since it is not known if reversal of potential occurs with the muscle spike. Also, in figure 6d, the anodal surround (of figure 6c) has given place to a catelectrotonic surround of diminished polarization, which spreads spatially, according to "core conductor" theory. With muscle, it appears that the all-or-nothing spike arises when the mem- brane adjacent to the end-plate is critically depolarized,^" e.g., at the arrow (figure 6d). The synapses of ganglion cells and of moto- neurones of the spinal cord have also been observed to generate such catelectrotonic potentials (synaptic potentials), spreading spatially, ac- cording to core conductor theory.-' ^^' ^^ It has further been shown that, as with the end-plate potential,^*'- " these potentials have a time- course, which may be interpreted as due to a brief, active polarization and a passive exponential decay governed by the electric time con- stant (the product CR, in figure 1) of the membrane.^-' -^ In the present hypothesis, the active depolarizing action is provided by the local response of the synaptic region of the post-synaptic cell, not directly by the currents generated by the pre-synaptic impulse, as has hitherto been assumed in hypotheses of electrical transmission.^"' ^^' ®^' 77, 78 There have, however, been suggestions of a possible involvement of a local response.*

The recent hypothesis of Nachmansohn*'^ is relevant to assumption C, for it would postulate that the local response is due to the action of acetylcholine liberated by the post-synaptic membrane at a critical in- tensity of catelectrotonus {i.e., at X in figure 7c). Such an assump- tion is readily assimilable to the present electrical hypothesis, but its general application to nerve impulse transmission would seem to be falsified by Lorente de No's finding^^ that this transmission is unaf- fected by high concentrations of acetylcholine.

* Eccles, J. C.21: 369; Lorente de N6, B.««: 449; Arvanltaki, A.^: 103.

442 ANNALS NEW YORK ACADEMY OF SCIENCES

The further problem of synaptic transmission concerns the initia- tion of propagated impulses by the catelectrotonic synaptic potential (see PART 1, Problem (b)). The observations of Kuffler'" on the iso- lated neuro-muscular junction indicate that the initiation of the im- pulse occurs in the muscle membrane adjacent to the end-plate region (cf. FIGURE 6d). No such intimate observations have been made for synapses in ganglia and the spinal cord, but it may be assumed that the catelectrotonic potentials from diverse synaptic regions sum by electrotonic spread. So far, only the over-all, summed potential has been observed after electrotonic spread along the axon (the so- called synaptic potential.^' ''*' ^2. 23, 25) j^ setting up the discharge of impulses, this synaptic potential appears to act just as a catelectro- tonus, the discharge occurring at a critical degree of depolarization. The synaptic potential provides a satisfactory explanation of all the phenomena hitherto attributed to the central excitatory state.- 22,25 The "detonator response"^"' ^^ need no longer be considered as a sepa- rate entity, for that hypothesis was based on experiments now explic- able, in part, by the flow of penetrating current, as in figure 5, and, in part, by the postulated local response of the post-synaptic mem- brane.^^' ^^' ^^

Thus, the sequence of events in synaptic transmission is envisaged as:

(1) Impulse in pre-synaptic nerve fiber generates a current which gives a diphasic effect at the synaptic region of the post-synaptic cell, with a total duration of probably not more than 1 msec, in mammalian muscle and the spinal cord; initial anodal focus, with cathodal sur- round; more intense cathodal focus, with anodal surround.

(2) This cathodal focus sets up a brief and intense local response at the synaptic region.

(3) From this local response, a catelectrotonus spreads decrementally over the post-synaptic cell membrane.

(4) A propagated impulse is set up in the post-synaptic cell, if this catelectrotonus is above a critical value. If it is below, then, as the local response subsides, the catelectrotonic surround decays passively.

6. APPLICATIONS OF THE HYPOTHESIS

The hypothesis offers an explanation of the following observations on synaptic transmission:

A. Irreversibility of Synaptic Transmission

This may be explained in the following three ways: (i) An impulse, artificially set up in the post-synaptic cell and fired antidromically at

ECCLES: ELECTRICAL THEORIES OF TRANSMISSION 443

the synapse, would, in general, propagate past the synapse and so exert on the pre-synaptic nerve terminals the full sequence of AiCiC2A2, the terminal A2 cutting short any excitation of the pre-synaptic terminal by CiCo;^ (iil the asymmetry of the pre- and post-synaptic elements, both as regards relative size and convergence relationship;^ (iii) the pre-synaptic terminal may not have the special excitatory properties postulated for the synaptic region of the post-synaptic cell (assumption C). Of these, (i) is susceptible to test with the neuro-muscular junc- tion. Antidromic transmission across synapses has been observed only under the special conditions provided by the prolonged end-plate nega- tivity set up by nerw impulses in eserinized muscle.^' There, the nerve terminal would be sensitized by the currents generated by the localized end-plate negativity.* Moreover, some of the muscle impulses may be blocked at the end-plate by the catelectrotonus.^^- ^^ Thus, such anti- dromic transmission occurs under conditions resembling those causing ephaptic transmission, close to a killed or injured region of nerve (the Hering effect, part 3, C, iii). The present hypothesis would predict that antidromic synaptic transmission would be greatly facilitated by colliding two nmscle impulses at the end-plate region (cf. figure 4b). The pre-synaptic fiber would then be subjected to the greatly increased excitatory action of double strength A1C2 stimulation, and antidromic transmission sliould occur under much less favorable predisposing con- ditions.

B. Synaptic Delay

If the initiation of the post-synaptic impulse is always caused by mediation of a synaptic potential of the post-synaptic cell, then true synaptic delay measures the interval between the time of ar- rival at the synapse of the fore-front of the pre-synaptic impulse and the initiation of the synaptic potential. In figure 7a, the time- course of the action potential at the pre-synaptic terminal is shown, and below it (the dotted line in figure 7b), the first derivative, which gives the approximate time-course of the current penetrating the post- synaptic membrane (part 3, C, iii). Allowance for rectification action is made in the broken line of figure 7b. On account of the electric time constant of this membrane, its potential change (the continuous line in figure 7b) is shown lagging behind the current which produces it.f Now, according to the hypothesis, the post-synaptic membrane initiates a local response when the catelectrotonus reaches a critical value, e.g., at the point X, in figure 7c. As shown in figure 6d, this

* Cf Eccles, J. C, & J. !■. Malcolni.3" Figure 15a, t Cf. Katz, B., & O. H. Schmitt."' Figube 6.

444

ANNALS NEW YORK ACADEMY OF SCIENCES

(b)

(c)

(d)

Figure 7. (a) Monophasic pre-synaptic action potential, (b) Its first derivative (dotted line) giving the time-course of the post-synapfic penetrating current (part 3, C), anodal currents being plotted downwards. The broken line shows the modification produced in this current if resistance is doubled at the anodal focus and halved at the cathodal focus (rectification). The continuous line gives the approximate time-course of the post-synaptic membrane potential so produced, al- lowance being made for the electric time constant of the membrane (cf. Katz & Schmitf", figure 6). (c) Post-s>naptic membrane potential shown as in (b). At X, the local response of the syn- aptic membrane is initiated by the catelectrotonic phase, and it is shown running a time-course rather slower than a spike, (d) Neglecting the passive electrotonic changes of (b) and (c), this local response is plotted together with the time-course of the resulting spreading catelectrotonus, shown as the broken line (the synaptic potential) recorded from the post-synaptic membrane adja- cent to the synapse (e.f/., at arrow, in figure 6d). The rising phase of sin initiated impulse is also shown (dotted line), AZ being its synaptic delay.

ECCLES: ELECTRICAL THEORIES OF TRANSMISSION 445

local response provides the active region for setting up a spreading catelectrotonus — the synaptic potential. Thus, the synaptic delay (in FIGURE 7, c and d), is represented by the interval AX. The value of 0.6 msec, which has been observed for this synaptic delay at mamma- lian neuro-muscular junctions'*^' ^* and motoneurone synapses"^ (frog, 1.3 msec.) accords well with tlie duration of the pre-synaptic action potential (figure 7a), when it is remembered that there is probably some slowing of time-course, as the impulse propagates into the fine pre-synaptic terminals.'*® As shown in figure 7d, a further delay, XZ, usually 0.2 to 0.3 msec, is involved in the building up of the synaptic potential to the threshold, for initiating an impulse at Z."- ^^' ^^

On the basis of figure 7, the hypothesis offers a satisfactory explana- tion of all the experimental findings on synaptic delay. For example: (i) By facilitation, synaptic delay cannot be shortened below a limit- ing value of about 0.5 msec, for central synapses.^'-* In figure 7b and 7c, AY would be the minimal interval at which excitation could occur, under optimal conditions of facilitation, (ii) Synaptic delay can, however, be further shortened by the direct excitatory action of a pre- ceding subliminal induction shock. '"^ By its depolarizing action, the shock would diminish the time lag between reversal of current and re- versal of potential, and so shorten synaptic delay to less than AY (figure 7b). (iii) The longer synaptic delay with sympathetic gang- lia (about five times longer) correlates with the longer duration of the pre-synaptic spike.^" which sets the time scale throughout figure 7. (iv) The upper limiting value of synaptic delay, for example, about 1.0 to 1.5 msec, for mammalian central synapses'^' "• ^^' ®°' " and neuro- muscular junctions,-*'' ^^ has been correlated with the time of the rising phase of the synaptic potential,^'*' ^'^ and thus, according to figure 7d, to the duration of the local response of the post-synaptic membrane (cf. C, below), (v) Synaptic delay (neuro-muscular in frog)"^ has, as would be expected from figure 7, approximately the same tempera- ture coefficient (2.1) as has the duration of the spike potential.

C. Time-Course of the Active Phase of the Synaptic Potential

The time-course of a local response is but little slower than the spike potential.'*^' ^-' °^' ^°' ''^ According to the hypothesis, therefore, the brief phase of active polarization (determined by analysis of the synap- tic potential) should have a time-course somewhat slower than the spike of the post-synaptic cell. This accords well with the findings on ganglion cells,^^ motoneurones,^^ and muscles.^® Furthermore, the tem- perature coefficient of this "active phase" is approximately the same as for a spike.^^

446 ANNALS NEW YORK ACADEMY OF SCIENCES

D. The Brief Period of Low Resistance

This is in addition to that attributable to catelectrotonus, during the initial, "active" phase of the end-plate potential.'*^ The postulated local response of the end-plate region of the muscle would produce just such an additional fall of resistance, running the same time-course as the active phase.*

E. Slow Catelectrotonic Potentials in the Pre-Synaptic Fibers^* ^^

Such potentials have only been observed with synapses in the central nervous system (the dorsal root potentials). The present hypothesis has been extended to explain the production of these potentials, by making the additional assumption that the terminal region of the pre- synaptic fiber resembles the post-synaptic area, in being specialized to give local responses. The catelectrotonic focus provided by the local response of the post-synaptic membrane sets up the current, which ex- cites the pre-synaptic terminal to give a local response, which, in turn, acts as a focus, setting up the spreading catelectrotonus of the dorsal root potential. ^°

7. DIFFICULTIES OF THE HYPOTHESIS

The electrical hypothesis encounters difficulties in explaining the fol- lowing experimental observations, but possible lines of reconciliation are suggested.

A. Synaptic Block Produced by Curarine in Skeletal Muscle

and Ganglia

Curarine acts as a specific depressant of the excitatory responses evoked in motor end-plates and ganglion cells by acetylcholine and related substances.^' ''' ^^' "^' ^^' ^* Thus, the acetylcholine hypothe- sis provides an obvious explanation of the synaptic blockage produced by curarine. Now curarine causes such blockage by depressing the synaptic potential. ^^' ^^' ^^' ^^ Hence, according to the electrical hy- pothesis, the simplest explanation of the blockage would be, that there is depression of the local response set up by the cathodal focus (cf. FIGURES 6d and 7c) ; i.e., that curarine depresses the electrical excita- bility of the post-synaptic membrane, as well as its acetylcholine ex- citability. In a recent attempt to test this, by electrical stimulation of the motor end-plate in the isolated nerve-muscle fiber preparation, before and after curarization, the initiation of propagated muscle im-

♦ Cf. Katz, B." FiGtniB 11.

ECCLES: ELECTRICAL THEORIES OF TRANSMISSION 447

pulses was used as the criterion of end-plate excitability. Yet, if the end-plate region reacts by local responses, rather than by propagating impulses (part 4, C), it seems probable that this investigation tested the excitability, not of the end-plate {i.e., of the post-synaptic mem- brane), but of the membrane adjacent thereto, confirming the previ- ously observed absence, there, of curare action. '^^' '^- '* Thus, it pro- vides no evidence for, or against, a specific depression of electrical ex- citability of the post-synaptic membrane. Alternatively, if curarine blocks solely by its known depressant action on the local potentials set up by acetylcholine,"' ^^' ^* then, in very deep curarization, the re- sidual catelectrotonic effects produced by the cathodal focus, i.e., by electrical transmission, should be observable, uncomplicated by chem- ical transmission. This inference is particularly pertinent in the case of sympathetic ganglia, where the synaptic potential is virtually abol- ished by deep curarization, and yet other evidence suggests that acetyl- choline transmission plays but a minor role (part 2, ii) .^^

B. Action of Anti-Cholinesterases on Synaptic Transmission with Skeletal Muscle and Sympathetic Ganglia

Anti-cholinesterases (eserine, prostigmine) delay the summit of the curarized end-plate potential and slow its decline,^^- ^■*' ^^ effects which, undoubtedly, are attributable to a prolongation of the active depolar- izing agent. With rapid, repetitive stimulation, a still greater effect is observed, the end-plate potential persisting for several seconds, both in curarized and normal muscle."- ^^ With sympathetic ganglia (normal or curarized), anti-cholinesterases also cause a prolonged synaptic po- tential to appear, after rapid, repetitive stimulation, but this prolonged potential is sharply distinguishable as a special addition to the other- wise unaltered synaptic potential.-^ Presumably, both with ganglia and muscle, the prolonged potential is due to acetylcholine liberated by pre-synaptic impulses.* With ganglia, it has been argued that, since the initial, brief transmitter action, setting up the synaptic potential, is un- affected by anti-cholinesterases, it is not due to acetylcholine.^^ Simi- larly, with muscle, the eserinized (or prostigminized) end-plate poten- tial appears to be the partly fused compound of a brief, initial phase, but little, if at all, lengthened by the anti-cholinesterase and the pro- longed phase (certainly due to acetylcholine) .f Thus, the electrical hypothesis would attribute the effect of anti-cholinesterases to an in- tensification and great prolongation of the normally small transmitter action of acetylcholine. Incidentally, it may be noted that such an

• However, cf. ITachmansolin, D."*

t Eccles, J. C, B. Xatz, 8c S. W. Kuffler.^" Figure 5.

448 ANNALS NEW YORK ACADEMY OF SCIENCES

acetylcholine effect would account well for the latter part of the rela- tively long transmitter action observed at normal neuro-muscular junc- tions (5 msec, frog;^^ 6 msec, cat^^.) A similar investigation of the synaptic potentials of motoneurones fails to reveal any such effect of anti-cholinesterases. Hence, it would appear that acetylcholine plays a negligible role at such synapses. These experiments indicate that the synapses of nerve-muscle, sympathetic ganglia, and the spinal cord form a series of decreasing significance for acetylcholine transmission. There has been no reference to other effects of anti-cholinesterases on synaptic transmission, e.g., after-discharge,*'' "' -^' ^^ local contrac- ture,^' ^•^' ^^ lengthening of refractory period, ^^' ^^ or catelectrotonic block,^' ^^' ^^' 2^ because these are all secondarily produced by the pro- longed and intensified synaptic potential.

C. Repetitive Synaptic Transmission There is but little diminution of the synaptic potential set up in curarized mammalian muscle^'' or anesthetized motoneurones^'^ by a second pre-synaptic volley, at the shortest intervals after the first. With curarized ganglia, the second potential is usually a little in- creased.-' The great increase with the frog's end-plate potential^*'' ^*' ^■^' '^ raises a further, as yet insoluble, problem. Since local responses are followed by refractory periods,*-' ^" assumption C of the hypothesis (part 4) would predict a considerable diminution of a second synaptic potential, at short intervals. However, it must be remembered that the above observations relate to synaptic potentials diminished suffi- ciently for synaptic blockage, and that small local responses set up much less refractoriness than large responses."' '" A large diminution of synaptic potential is observed, when synaptic transmission is not blocked,-**' -"' ^'^ and has, hitherto, been attributed to the refractoriness of the post-synaptic cell. Nevertheless, it may be, in part, due to re- fractoriness, following the postulated large local responses of the spe- cialized post-synaptic membrane (part 4, C). This latter explanation is supported by the observation that the diminution is lessened by sub- paralytic curarization.*

8. TESTING THE HYPOTHESIS

It has been shown that the hypothesis gives a satisfactory explana- tion of all experiments investigating the temporal factors in synaptic transmission. It is otherwise with the intensity factors. Before the hypothesis can be regarded as well established, it has to be shown that

* Bccles, J. C, & S. W. Xaffler'»: 505.

ECCLES: ELECTRICAL THEORIES OF TRANSMISSION 449

the currents generated by the pre-synaptic impulses (figure 5) excite the post-synaptic membrane sufficiently to produce the observed synap- tic potentials and initiation of impulses. Such an intensity of action is not normally attained, even by the most efficient ephapse/' *^ but there are possibly three factors increasing the efficiency of the synapse:

(i) The contact of the pre- and post-synaptic membranes is so inti- mate^° that virtually all the current penetrating the former must pene- trate the latter (in contrast to the estimated value of one third for the ephapse*^) .

(ii) In the synapse, a large expansion in the area of contact is pro- duced by the swelling and branching of the pre-synaptic terminals; also, with nerve cells, by the multiplicity of closely adjacent synapses.^"

(iii) If the post-synaptic membrane has special excitatory proper- ties (part 4, C), the efficiency of electrical excitation may be thereby increased.

By stimulating just beyond the region where a nerve volley is blocked, Hodgkin" (cf. also Lorente de N6'''°) showed that, with critical condi- tions for blockage, the threshold may be lowered to only 10% of nor- mal, i.e., the nerve volley still provides as much as 90% of the threshold electrical stimulus, beyond the blockage. Hence, there is a high prob- ability that, in the absence of block, it provides an electrical stimulus adequate to excite, i.e., that the transmission of nerve impulses is elec- trical. It should be possible to apply a similar test to the curarized end-plate of the isolated nerve-muscle fiber preparation. It has, of course, been shown that there is a lowering of threshold, during the end-plate potential,*^- ^^ as would be expected for a catelectrotonus, however produced. The present test would explore, instead, the brief interval of pre-synaptic current flow, particularly that preceding the origin of the synaptic potential {i.e., AX in figure 7C). Such a test has already given suggestive results with motoneurones,^* but, in order to be convincing, there should be an accurate location of the stimulat- ing electrode on the synaptic region of the post-synaptic membrane, and, at present, this seems possible only with the isolated nerve- muscle fiber (cf. Kufiler^*). An attempt on the whole sartorius was unsuccessful.*

Crucial testing of the hypothesis will also be provided by further pharmacological experiments on synaptic transmission: particularly the action of anti-cholinesterases, and the effects of various background concentrations of acetylcholine, potassium, and calcium. Predictions leading to tests have already been mentioned in the preceding sections.

♦ Katz, B.«: K9t;.

450 ANNALS NEW YORK ACADKMY OF SCIENCES

111 addition, iiiHtlifiiiatical treatment of the flow of penetrating current at a schematized synapse should be possible, and would give more pre- cise predictions for experimental testing. Since the hypothesis is based on the investigations on nerve and muscle fibers outlined in PART 3 (particularly the ephaptic experimentsj , further developments of this work are of immediate relevance as tests of the hypothesis, providing data on which will depend its development, or modification, or rejection in whole or in part.

Finally, it may be stated that a recommendation for the hypothesis is its systematization of synapses and neuro-muscular junctions in a series (neuro-muscular junctions, ganglionic synapses, and central synapses) , exhibiting a progressive replacement of acetylcholine trans- mission by electrical transmission. To the beginning of such a series could be added those special modifications of ganglionic and neuro- muscular synapses, seen, respectively, with the chromaffin organs (su- prarenal medulla) and electric organs, where synaptic transmission seems to be wholly due to acetylcholine.

9. SUMMARY

• Transmission of impulses across synapses of the spinal cord, sympa- thetic ganglia, and skeletal muscle, involves a dual problem: (a) the setting up of synaptic, catelectrotonic potentials in the post-synaptic cell; and (b) the initiation of impulses in the post-synaptic cell by such potentials. Evidence is given that, in their present form, both the chemical (acetylcholine) and electrical hypotheses relating to Prob- lem (a) are unsatisfactory. Furthermore, recent experiments are cited which indicate that acetylcholine plays a negligible part as a synaptic transmitter with motoneurones ; a subsidiary role with sym- pathetic ganglia, and possibly also with skeletal muscle. Hence, it is desirable to attempt a more precise formulation of the electrical hy- pothesis.

. The present attempt is based, mainly, on four lines of recent inves- tigation :

(1) The electrical properties of surface membranes: resistance, elec- tromotive force, capacity, and rectification. Inductance is neglected, because it is doubtful if the high values of cephalopod axons obtain for vertebrate nerve and muscle.

(2) The existence of active load responses. These may be very large in refractory, or anesthetized, or deteriorated nerve.

(3) The electrical actions occnrring (uto.s.s artificial synapses [ephap- ses) . There are shown to be three main types of ejih apses, the synapse

ECCLES: ELECTRICAL THEORIES OF TRANSMISSION 451

being a special example of the type with an initial anodal and terminal cathodal action; the only type in which a significant excitatory action is exerted.

1.4) Electrical recording from the isolated neuro-muscular junction shows that the motor end-plate is specialized to give local responses without the all-or-nothing breakdown of propagated impulses. Im- pulse initiation appears to be produced by a secondary catelectrotonus in the surrounding membrane.

The present hypothesis makes three main assumptions:

A. A schematized formulation is made of the essential geometrical relationship of the membranes of the pre- and post-synaptic elements, as revealed by histological and electrical investigation.

B. The electrical properties of the pre- and post-synaptic surface membranes resemble those observed for peripheral nerve and muscle (see (.1), above j.

C. The membrane of the immediate post-synaptic region is special- ized, so that large and graduated local responses are set up by catelec- trotonic polarization (see (2) and (4), above).

On these basic assumptions, it is shown that a pre-synaptic im- pulse sets up electric currents exerting an initial anodal and later cathodal action on the post-synaptic membrane. The latter action, in- tensified by rectification, sets up a local response (part 4, C), which, in turn, acts as a relatively prolonged cathodal focus, from which spreads, electrotonically, the synaptic potential of the effector cell. Finally, the initiation of impulses by this synaptic potential appears to be ex- plicable, simply, as the action of a catelectrotonus.

This hypothesis is shown to offer satisfactory explanations of many fundamental observations on synaptic transmission: irreversibility; synaptic delay; time-course of junctional potential; brief impedance loss at end-plates; dorsal root potentials of the spinal cord; some of which were hitherto inexplicable in detail.

On the other hand, the hypothesis encounters difficulties in explain- ing the actions of curare and of anti-cholinesterases on synaptic trans- mission in ganglia and skeletal muscle. The action of curare may be explained, if it is assumed that it depresses the electrical excitability of the post-synaptic membrane, as well as its pharmacological excitabil- ity. It is argued that this assumption has not yet been tested. The action of anti-cholinesterases is attributed to the intensification and prolongation of the action of acetylcholine, to which the hypothesis as- cribes a subsidiary role, as a transmitter at synapses of ganglia and skeletal muscle. A further difficulty appears to arise in the explana-

452 ANNALS NEW YORK ACADEMY OF SCIENCES

tion of rapid, repetitive, synaptic transmission. The postulated local responses should be followed by refractory periods, but a possible ex- planation is suggested.

The testing of the hypothesis is shown, especially, to concern the further investigation of the special electrical properties assumed for the post-synaptic membrane ; also, the attempt to discover how far the postulated electrical actions can account quantitatively for the post- synaptic stimulation. It has been shown (above) that the observed temporal course of the post-synaptic stimulation is satisfactorily ex- plained. In addition, further pharmacological investigation is neces- sary to test the explanation attributing a subsidiary role to acetyl- choline transmission. It is evident that further work on the electrical properties of membranes on local responses, and on ephaptic transmis- sion, will provide additional tests of the hypothesis.

BIBLIOGRAPHY

1. Arvanitaki, A.

1942. J. Neurophysiol. 5: 89.

2. Barron, D. H., & B. H. C. Matthews

1938. J. Physiol. 92:276.

3. Bishop, G. H.

1941. Ann. Rev. Physiol. 3: 1.

3a. Bremer, F., V. Bonnet, & J. Moldaver

1942. J. Arch. Int. Physiol. 52: 1.

4. Bronk, D. W.

1939. J. Neurophysiol. 2: 380.

5. Brown, G. L.

1937. J. Physiol. 89:438.

6. Brown, G. L.

1937. Physiol. Rev. 17: 485.

7. Brown, G. L., H. H. Dale, & W. Feldberg

1936. J. Physiol. 87:394.

8. Buchthal, F., & J. Lindhard

1939. 'i'lie Physiology of Striated Muscle Fiber. Munksgaard. Copenhagen.

9. Buchthal, F., & J. Lindhard 1942. Act. Physiol. Skand. 4: 136.

10. Cajal, S. R.

1934. Trav. Lab. Rech. Biol. Madrid 29: 1.

11. Clark, A. J.

1936. Proc. Roy. Soc. London B 121 : 580.

12. Cole, K. S.

1941. J. Gen. Physiol. 25: 29.

13. Cole, K. S., & R. F. Baker 1941. J. Gen. Physiol. 24: 535.

14. Cole, K. S., & R. F. Baker 1941. J. Gen. Physiol. 24: 771.

15. Cole, K. S., & H. J. Curtis

1938. J. Gen. Physiol. 22: 37.

16. Cole, K. S., & H. J. Curtis

1939. J. Gen. Physiol. 22:649.

ECCLES: ELECTRICAL THEORIES OF TRANSMISSION 453

17. Cole, K. S., & H. J. Curtis

1941. J. Gen. Physiol. 24: 551.

18. Cowan, S. L.

1936. J. Physiol. 88: 30.

19. Dale, H. H.

1937. Harvey Lectures 32 : 229.

20. Eccles, J. C.

1936. Ergebn. Physiol. 38: 339.

21. Eccles, J. C.

1939. Ann. Rev. Physiol. 1: 363.

22. Eccles, J. C.

1943. J. Physiol. 101:465.

23. Eccles, J. C.

1944. J. Physiol. 103:27.

24. Eccles, J. C.

1945. Nature 156: 680.

25. Eccles, J. C.

1946. J. Neurophysiol. 9: 87

26. Eccles, J. C, B. Katz, & S. W. Kuffler

1941. J. Neurophysiol. 4: 362.

27. Eccles, J. C, B. Katz, & S. W. Kuffler

1942. J. Neurophysiol. 5: 221, 225, 227-228.

28. Eccles, J. C, «fe S. W. Kuffler 1941. J. Neurophysiol. 4: 402.

29. Eccles, J. C, & S. W. Kuffler 1941. J. Neurophysiol. 4: 486.

30. Eccles, J. C, & J. L. Malcolm Sent in to J. Neurophysiol.

31. Eccles, J. C, & J. L. Malcolm Unpublished Observations.

32. Erlanger, J.

1939. J. Neurophysiol. 2: 370.

33. Feng, T. P.

1937. Chin. J. Physiol. 12: 331.

34. Feng, T. P.

1940. Chin. J. Physiol. 15:367.

35. Feng, T. P.

1941. Biol. Symp. 3: 121.

36. Fillenz, M., & M. Hanafin Unpublished Observations.

37. Gerard, R. W.

1942. Ann. Rev. Physiol. 4: 329.

38. Granit, R., & C. R. Skoglund 1945. J. Physiol. 103:435.

39. Grundfest, H.

1932. J. Physiol. 76: 95.

40. Grundfest, H.

1939. Am. J. Physiol. 127: 252.

41. Hodgkin, A. L.

1937. J. Physiol. 90: 183.

42. Hodgkin, A. L.

1938. Proc. Roy. Soc. London B 126: 87.

43. Jasper, H. H., & A. M. Monnier 1938. J. Cell. Comp. Physiol. 11: 259.

44. Katz, B.

1937. Proc. Roy. Soc. London B 124: 244.

454 ANNALS NEW YORK ACADEMY OF SCIENCES

45. Katz, B.

1939. J. Physiol. 95: 286.

46. Katz, B.

1939. Electric Excitation of Nerve. Oxford Univ. Press.

47. Katz, B.

1942. J. Neurophysiol. 5: 169.

48. Katz, B., & O. H. Schmitt

1939. J. Phy.siol. 97: 471.

49. Katz, B., & O. H. Schmitt

1942. J. Physiol. 100: 369.

50. Kuffler, S. W.

1942. J. Neurophy.siol. 5: 18.

51. Kuffler, S. W.

1942. J. Neurophysiol. 5: 199.

52. Kuffler, S. W.

1942. J. Neurophysiol. 5: 309.

53. Kuffler, S. W.

1943. J. Neurophysiol. 6: 99.

54. Kuffler, S. W.

1945. J. Neurophysiol. 8: 77.

66. Langley, J. N.

1907. J. Physiol. 36: 347.

56. Lloyd, D. P. C.

1943. J. Neurophysiol. 6: 143.

57. Lloyd, D. P. C.

1943. J. Neurophysiol. 6: 293.

58. Lorente de No, R.

1935. J. Cell. Comp. Physiol. 7: 47.

59. Lorente de No, R.

1938. J. Neurophysiol. 1: 187.

60. Lorente de No, R.

1939. J. Neurophysiol. 2: 402.

61. Lorente de No, R.

1944. J. Cell. Comp. Physiol. 24: 85.

62. Macintosh, F. C.

1941. J. Physiol. 99:436.

63. Marmont, G.

1941. J. Physiol. 133: 376P.

64. Marrazzi, A. S., & R. Lorente de No

1944. J. Neurophysiol. 7: 83

66. Monnier, A. M.

1934. L'e.xcitation 61ectrique des tissus. Hermann. Paris

66. Monnier, A. M.

1936. Cold Spr. Harb. Symp. 4:111.

67. Nachmansohn, D.

1940. Yale J. Biol. Med. 12: 565.

68. Nachmansohn, D.

1945. Vitamins and Hormones 3: 337.

69. Osterhout, W. J. V., & S. E. Hill

1930. J. G3n. Physiol. 13: 547.

70. Pumphrey, R. J., O. H. Schmitt, & J. Z. Young 1940. J. Physiol. 98: 47.

71. Renshaw, B.

1940. J. Neurophysiol. 3: 373.

72. Renshaw, B.

1942. J. Neurophysiol. 5: 23.'^

ECCLES: ELECTRICAL THEORIES OF TRANSMISSION 455

73. Renshaw, B., & P. O. Therman

1941. Am. J. Physiol. 133: 98.

74. Rosenblueth, A.

1911. Am. J. Physiol. 132: 119.

75. Rushton, W. A. H. 19.33. J. Physiol. 77: 337.

76. Rushton, W. A. H.

1937. Proc. Roy. Soo. London B 124: 201.

77. Schaefer, H., & P. Haass

1939. PHug. Arch. ges. Physiol. 242 : 364.

78. Schaefer, H., P. Scholmerich, & P. Haass

1938. PHug. Arch. ges. Physiol. 241: 310.

79. Schmitt, F. O., & O. H. Schmitt

1940. J. Physiol. 98: 26.

CHEMICAL EXCITATION OF NERVE*

By Frank Brink, jR.,t Detlev W. Bronk, and Martin G. Larrabee

Eldridge Reeves Johnson Research Foundation, University of Pennsylvania,

Philadelphia, Pennsylvania

One of the noteworthy characteristics of neurones is their sensitivity to changes in the chemical environment. Even within the relatively protected interior of the body, the properties of nerves are subject to modification by variations in the composition of the body fluids. In- deed, the alterations of irritability and the trains of nerve impulses, which are the result of changes in the chemical environment, are among the most important factors involved in the regulation of the activity of the organism. This is one of the significant reasons for studying the chemical activation of nerve. A second reason derives from the current interest in the role of chemical agents in the mechanism of syn- aptic transmission. Furthermore, the investigation of the effects of various chemical agents is one of the most fruitful sources of informa- tion regarding the role of the several chemical components of the nerve structure and of the chemical processes involved in nervous action.

This last consideration suggests that the most significant chemical agents for use in the study of the processes of activation are those which have an important part in the normal structure of nerve. Cal- cium is such an element. Potassium is another; it modifies the action of calcium, to which it is closely related in the regulation of nerve ac- tion, and it has a marked influence on the electric potential difference across the interfaces at which the nerve comes in contact with its en- vironment. Finally, the effects of acetylcholine on the initiation and conduction of the nerve impulse make an important and timely subject for investigation in such a study as this. It is with the effects of these agents that we shall be primarily concerned. There are others of significance for a general study of this problem, but from these three we can derive many of the basic phenomena involved in chemical excitation.

The changes in the functional characteristics of a nerve caused by an alteration of its chemical environment are due to the consequent

* The experimental work reported here has been generously supported by grants from the Supreme Council, Scottish Rite Masons, and from the American Philosophical Society, t Fellow of the Lalor Foundation.

(457)

458 ANNALS NEW YORK ACADEMY OF SCIENCES

changes in the chemical constitution of the cell or axon. Thus, the in- creased irritability that is induced by surrounding a nerve with a cal- cium-deficient fluid follows a decrease of calcium within the nerve. In order that the effects of such changes may be investigated, it is con- venient to have available a solution which will maintain nerve in a stable functional state for long periods of time, and to which the effects of other environmental solutions may be referred. It is customary to choose for reference a solution having a salt content and pH approxi- mating that of the animal's body fluids. By direct test, it has been found that a frog nerve can be kept in a solution at pH 7.2 (phosphate buffer) containing sodium chloride (116 mM), potassium chloride (2.0 mM), and calcium chloride (1.8 mM) for many hours, with no sig- nificant change in excitability or in rate of aerobic oxidation. Squid nerve, which we have also employed, maintains a similar stable func- tional state in sea water (Woods Hole) at pH 8.0 or in a solution con- taining sodium chloride (405 mM), potassium chloride (11 mM), and calcium chloride (70 mM). Modifications of these solutions have been used as the experimental means of chemical activation.

The calcium ion concentration of the environmental fluid is espe- cially important in determining the excitability of nerve. This familiar phenomenon (cf., e.g., MisskeM can be studied quantitatively and under quickly reversible conditions in squid giant axons or in bundles of frog axons from which the perineurium has been removed. Under those cir- cumstances, diffusion equilibrium between the axons and the surround- ing fluid is attained relatively quickly. In figure 1, the threshold strength of direct current necessary to initiate an impulse, which is the rheobase, is plotted as a function of the concentration of calcium chloride in the fluid bathing a giant axon of the squid. A similar re- lation is obtained for the a fibers in a frog sciatic nerve (figure 2).

The increased excitability produced by the action of solutions having a low concentration of calcium chloride is presumably due to the dif- fusion of Ca^+ from the cell structure. Indeed, Tipton' has shown by chemical analysis that as much as 40 per cent of the total calcium of frog nerve is in diffusion equilibrium with the surrounding fluid, some of this diffusible calcium being in the cell phase. His evidence for intracellular precipitation of added calcium is a further indication that changes in the calcium chloride content of the bathing fluids lead to changes in the cellular content. The spatial distribution of these changes in cellular calcium are unknown.

When frog nerve is equilibrated with solutions containing from 1.0 mM to 0.3 mM calcium chloride, or when squid nerve is equilibrated

BRINK AND OTHERS: CHEMICAL EXCITATION OF NERVE 459 100

o

X CO

FiGTJRB 1. Threshold of giant axon of Squid in isotonic solutions containing various concentra- tions of CaCh. Threshold (rheobase) as per cent of threshold of nerve in Woods Hole sea water. Concentration in millLmoles per liter. The arrow indicates concentration at which spon- taneous activity began.

FiGDEE 2. Threshold of a fibers of sciatic nerve of frog with various concentrations (millimoles per liter) of CaCh in bathing fluid. Threshold as per cent of rheobase of nerve m reference aolution described in text.

460 ANNALS NEW YORK ACADEMY OF SCIENCES

with solutions containing from about 70 mM to 10 mM calcium chlo- ride, there is a certain degree of irritability corresponding to each con- centration. This is measured as the minimal strength of current nec- essary to initiate a conducted impulse. It may be thought of as an index of the stability of the excitable portion of the nerve structure. We shall subsequently refer to the fact that the rate of oxygen con- sumption of nerve is also modified by changes in its calcium content. Here, it is pertinent to remark that the variations of oxygen consump- tion occur within this same range of calcium concentrations in which there are measurable changes of irritability. In the case of frog nerve, moderate increases in calcium above 2.0 mM do not cause a further appreciable decrease of irritability nor a further decrease of oxygen consumption. At very much higher concentrations, above 15 mM, the irritability again decreases,^ and there is a further fall in the oxygen consumption.* If the concentration of calcium be lowered beyond 0.3 mM, or 10 mM in the case of squid nerve, it is no longer possible to measure the irritability in terms of the strength of current necessary to initiate an impulse. At these levels, the nerve structure has been so much modified, its stability lowered so much, that it goes through periodically-recurring cycles of change, with consequent, self-initiated trains of propagated impulses.^

II

The response of nerve to the exciting action of an electric current can be studied in a nerve trunk or in a bundle of fibers. The stimulus is under the control of the experimenter, and all of the fibers are ex- cited simultaneously. Accordingly, the action potential recorded from the aggregate of fibers of a given type is a fairly accurate representa- tion of the sequence of events in each fiber, provided temporal disper- sion, due to differences in conduction velocity, is avoided.

The situation is quite different in the case of chemical excitation. The altered chemical environment modifies the properties of the fibers, so that a sequence of cychc events develops in each fiber, with a fre- quency that is determined by the characteristics of the fiber. Because these intrinsic characteristics differ, the frequency of the impulses dis- charged from a chemically treated region varies from fiber to fiber. Furthermore, the properties of the fiber may change from moment to moment, so that the sequence of impulses is not truly periodic. Finally, the times of initiation of impulses in one fiber are independent of the timing of these events in the other fibers, in contrast to the externally determined synchronization imposed by electric stimuli. Because of

BRINK AND OTHERS: CHEMICAL EXCITATION OF NERVE 461

these considerations, the action potentials recorded from a bundle of numerous fibers reveal little of what is occurring in the individual units (figure 3). Under these conditions, the investigation of the proc- esses of excitation and response encounters the same difficulties experi- enced in the study of groups of sensory endings or motor nerve cells (Adrian f Adrian and Zotterman ;^ Adrian and Bronk*) .

The solution here is the same as there : that is, to isolate and measure the activity in a single fiber. Only when this is done can one observe the more or less rhythmic train of impulses discharged from the chem- ically modified region (figure 3). The difficulties inherent in this

i

Figure 3. Above: Impulses recorded from branch of sciatic nerve of frog stimulated by topical ap- plication of isotonic sodium citrate. Below: Impulses recorded from single « fiber dissected from this nerve. Time in 1/5 seconds.

experimental procedure partly explain the relative paucity of our knowledge regarding the nature of chemical excitation.

It is worthy of emphasis that an axon possesses the capacity (as does a cell body or sensory ending) for transforming the continuous environ- mental action of a physical or chemical agent into a series of recurrine events which are made manifest as nerve impulses.

It has been said that the frequency of impulses developed varies from fiber to fiber, and depends upon the intrinsic characteristics of each

462

ANNALS NEW YORK ACADEMY OF SCIENCES

fiber. Obviously, these properties will be modified by the chemical excitant, qualitatively by the nature of the chemical change, and to a degree that is determined by the amount of substance added to, or re- moved from, the nerve structure. Therefore, it is not surprising to find that the number of impulses discharged per second from the treated region depends upon the concentration of the calcium ion, as shown in FIGURE 4.

Figure 4. The average frequency of impulse;^ recorded from a single fiber depends UDon the concentration of calcium ions. Upper record, concentration of Ca* is 0.3 mM ; middle, 0.1 mM ; lower, no calcium. In this experiment, the 0.1 mM calcium was applied first, then tiie 0.3 mM, and finally the calcium-free solution. Time in 1/5 seconds.

The frequency of impulses initiated by a given reduction of calcium ion concentration, or by other chemically stimulating media, also de- pends upon the previous duration of the chemical action. The electric threshold begins to fall almost at once after the application of the solution (figure 5). Further time is required for changes in the intra- cellular processes which must precede the development of conducted impulses. Indeed, the first impulse may not develop for some minutes, and then at a time when the threshold has fallen to zero. This same gradual loss of stability, continuing further, is manifest in the pro- gressive increase in the average frequency of impulses.

When calcium is removed from a nerve, by diff'usion into a solution containing less than the normal amount of calcium chloride, the dis-

BRINK AND OTHERS: CHEMICAL EXCITATION OF NERVE 463

charge of impulses referred to above begins slowly. If the calcium ion concentration is reduced, by adding a calcium-binding agent such as sodium citrate, the discharge begins more quickly and continues longer. Because of this, we have used for many of our experiments a stimulating fluid which contained sodium citrate in place of some of the sodium chloride. When the calcium ion concentration is thus re- duced below about 0.4 mM, impulses are initiated. This is the concen-

FiGUHE 5. The threshold of an axon decreases with time after topical application (upper arrow) of a solution which lowers the calcium ion concentration. Repetitive activity begins at 14 min. after rheobase has decreased below five per cent of its initial value. Threshold measured with cathode on calcium-deficient region of nerve.

tration level for activation, whether the calcium be removed by the action of citrate or by the simple process of diffusion. This, and other evidence, suggest that a principal factor in citrate excitation is the low- ered calcium ion concentration.

We have already stated that the frequency of impulses initiated by the removal of calcium is, at any time, dependent upon the duration of the previous action of the stimulating fluid. The time-course of development of the impulse discharge is also largely influenced by the previous chemical treatment of the nerve. Usually, the impulses be- gin to occur, at random intervals, when the rheobase has fallen to about 5 per cent of its initial value. Thereafter, the impulses are discharged in groups, which gradually merge into a more or less regular train when the frequency reaches about 150 per second. Such a gradual increase in the frequency of impulses is shown in one of the curves of figure 6.

The development of activity is not always so gradual. Sometimes, the initial frequency may be high, and then decline to a lower level that is sustained for some hours, with the development of hundreds of

464

ANNALS NEW YORK ACADEMY OF SCIENCES

200

> u

Z

u

D Of

u q: ii.

O 10 20

MINUTES

Figure 6. Time-course of discharge (impulses per second in a single fiber), after topical appli- cation of isotonic solution containing sodium citrate (35 mM) and sodium chloride.

Open circles : gradual increase of frequency during first period of chemical excitation.

Squares: response to second application of same solution, after intervening 2 hours in reference fluid. ^ . ,

Filled circles: response to a third application, after another hour in Ringer s fluid.

200

0

20

40 60

MINUTES

80

100

FiatiBE 7. The discharge of impulses continuing at high frequency for many minutes during topical application of isotonic sodium citrate. Approximately 540,000 impulses were produced by the single fiber, during the activity plotted in this figure.

BRINK AND OTHERS: CHEMICAL EXCITATION OF NERVE 465

thousands of impulses (figure 7). This initial high frequency dis- charge usually occurs after a previous period of chemical excitation that had been arrested by the restoration of calcium (figure 6). It also occurs during the actual restoration of calcium.^

A nerve which has been once modified by the withdrawal of cal- cium continues to give such a response of high initial frequency to a successive activation, even though it has been in unmodified Ringer's fluid for many hours (figure 6). Whether the impulses start at a high frequency that declines, or whether the frequency gradually increases, the final, sustained average frequency is about the same for a given stimulating fluid. This frequency is, to an important degree, deter- mined by the calcium content of the nerve, and our experiments also suggest that it is, in part, dependent upon the rate of removal of cal- cium. This latter factor may be especially important in the determina- tion of the transient changes of frequency.

Ill

The resting metabolism of nerve has long been thought of as neces- sary for maintaining the organization of its unstable structure against the tendency of the structure to become disorganized. In accordance with this view, the less stable structure resulting from the withdrawal of calcium should have a higher metabolic requirement for its main- tenance. This increased metabolism has been observed.^" We, too, have made such measurements of the oxygen consumption of nerve from which varying amounts of calcium have been withdrawn, while, at the same time, measuring the excitability and recording any impulses that were initiated." Alterations of calcium content in the nerve, suffi- cient to cause a lowered threshold to electric stimuli, but insufficient to cause the rhythmic discharge of impulses, induce an increased oxygen consumption (figures 8 and 9). As the calcium content is further reduced, the oxygen consumption increases still more. Finally, a level of calcium content may be reached which is sufficiently low to cause the rhythmic discharge of impulses, and associated with this calcium content there is a still higher oxygen consumption (figure 8) .

This progressive increase- of oxygen consumption of nerve with de- creasing concentrations of calcium, starting at calcium levels too great to permit the development of spontaneous activity, raises the question as to the meaning of the term, "resting oxygen consumption." The oxygen consumption of axons which are resting, in the sense of not

conducting impulses, may be quite different in different chemical environments.

46G

ANNALS NEW YORK ACADEMY OF SCIENCES

NO ACTIVITY

Ob

10

15

20

CA WILLIMOLCS PER LITER

Figure 8. The respiration (as per cent of value in reference solution) of a frog sciatic nerve increasing when the concentration of calcium chloride in bathing fluid is below 1.0 mM. There is an appreciable increase at a concentration which is not low enough to initiate impulses.

300

50 Threshold

Figure 9. Respiration of frog sciatic nerve in relation to threshold of a fibers in same nerve. Changes produced by equilibration in isosmotic solutions containing 1.0, 0.5, and 0.25 mM CaCh. Both respiration and threshold (rheobase) are expressed as per cent of their values (marked bj solid square) in reference solution (containing 2.0 mM CaCh).

Open circles : A nei-ve showing a small per cent change in respiration has a correspondingly small per cent change in rheobase.

Solid cu'cles : Another nerve, showing a large per cent change in respiration, has a large per cent change in rheobase.

BRINK AND OTHERS: CHEMICAL EXCITATION OF NERVE 467

An increased rate of oxidation is essential for the initiation of im- pulses. That is shown by the effects of an oxidation-inhibiting agent, such as sodium azide. A portion of a nerve trunk was placed in sodium citrate, and impulses which were thus developed were recorded in one of the fibers coming from the chemically activated region. Sodium azide was then applied to the citrate-treated portion of the nerve, in a concentration that was sufficient to suppress the chemical excitation. For this, a concentration of azide which restored the oxidation to a normal rate was adequate (figure 10). After the rhythmic discharge

300

200

a.

100 -

AZIDE

2 3

MILLIMOLES PER LITER

Figure 10. The respiration of a calcium-deficient nei-ve (per cent of value in reference solu- tion) suppressed by sodium azide, as is the respiration of a normal nerve.

of impulses had thus been inhibited, it was still possible, for several hours to send a high frequency train of impulses, initiated by electric stimuli, through the calcium-deficient and azide-treated length of nerve. The initiation of impulses by a calcium-deficient region of nerve re- quires a higher rate of oxidation than is necessary for the maintenance of the capacity of such a calcium-deficient region to conduct impulses. Certain specific chemical changes in the constitution of nerve and in- creased metabolic rate both appear to be necessary for the initiation of rhythmic activity. One without the other is an inadequate condition for self-excitation.

468 ANNALS NEW YORK ACADEMY OF SCIENCES

Although there are many instances of a close parallelism between the stability of nerve, as measured by its electric threshold or by the spontaneous development of impulses, and its rate of respiration, there are exceptions. For instance, the rhythmic activity may be abolished by the application of potassium chloride, which, at the same time, in- creases the rate of respiration. Another instance of such a lack of parallelism is revealed when a nerve is returned to its normal fluid environment, after treatment with sodium citrate. The rhythmic ac- tivity is promptly suppressed, and the threshold becomes normal, long before there is a corresponding recovery of the original, normal rate of respiration.

IV

There has been a persistent notion that the initiation of trains of impulses from a chemically activated portion of an axon or from a sense organ under a constant stimulus is due to a gradient of electric potential at the site where the impulses originate. Indeed, Adrian^^ found that there was a gradient of 10 mV between a normal portion of nerve and an injured region from which impulses were discharged. Accordingly, he attributed the excitation to this demarcation potential. Furthermore, Erlanger and Blair^* and Fessard^^ caused the rhythmic discharge of impulses by the passage of constant currents. Finally, Katz^*^ and Arvanitaki^^ found that the duration of such an electrically induced repetitive discharge could be much prolonged by reducing the calcium content of the nerve.

Because of these considerations, we have carefully searched for some causal relation between a potential gradient developed at the site of calcium removal and the chemical initiation of impulses. To do this, one of a pair of non-polarizable electrodes was placed in contact with the chemically altered region of the nerve; the second was in contact with an adjacent, untreated portion. We have found that the activity develops in the calcium-deficient part of the nerve, without the appear- ance of an appreciable longitudinal potential gradient.^^ However, this part of the nerve is very sensitive to weak currents. It is, therefore, possible that potential differences of less than a millivolt might be in- volved in the mechanism of chemical excitation.

Accordingly, experiments were devised to study quantitatively the relation between the frequency of conducted impulses and changes in membrane polarization produced in the hyper-excitable portion of the axon by certain additional chemical agents. Increasing the proportion of potassium chloride in the solution of sodium citrate used to excite

BRINK AND OTHERS: CHEMICAL EXCITATION OF NERVE 469

the nerve makes the calcium-deficient region negative to the adjacent parts of the cell. Under these conditions, the associated current flow is inward across the plasma membrane in the hyper-excitable region of the axon. The conducted impulses still occur, but at a reduced frequency, compared with the activity initiated by a solution contain- ing less potassium chloride. The frequency of response is lower, the

3

=! -I

o kl «>	200
a.
CO UJ	100

s

20 30

MINUTES

FiGUBE 11. The depression by KCl of the activity produced by topical application of isotonic sodium citrate is related to the associated decrease in membrane polarization.

Upper curve is potential difference between treated region of nerve and normal part of nerve. At the first arrow, the solution was changed from isotonic sodium citrate to one containing the same amount of citrate, but with 15 mM K in place of some of the sodium. At the second arrow, the ner\'e was returned to a solution containing isotonic sodium citrate without potassium. At the third arrow, the solution containing potassium was again applied to the nerve.

The lower curve shows change in frequency of response recorded from a single fiber in this nerve. A similar suppression of activity during the first cycle of depolarization was observed, but not recorded.

higher the concentration of potassium chloride, and, therefore, de- creases, as the degree of depolarization increases. When the calcium- deficient part of the cell is made sufficiently negative, the activity is suppressed, but it begins again as this depolarization is removed by washing out the potassium chloride (figure 11). Conversely, if this region of the nerve is made positive to adjacent parts, as by a solution of sodium thiocyanate, the frequency of the impulses is increased.

Thus, small differences of potential between a normal and hyper- irritable region of an axon modify the frequency of the impulses dis-

470

ANNALS NEW YORK ACADEMY OF SCIENCES

charged from that region, despite the fact that the impulses originate without the mediation of such potential gradients. The influence of these gradients on the excitability of calcium-deficient nerve, as meas- ured by changes in the frequency of impulse discharge, is in agreement with the usual effects of current flow from an external source: depres- sion at the anode and excitation at the cathode.

Although the chemical excitation of nerve does not depend upon the development of a steady current flow, the discharge of chemically initiated impulses can be modified by an externally imposed potential

§250

o u

UJ (O

cm h!200

UJ

-J

a. ZI50

100

i

POLARIZE

RECORD

10

20

30

40

50

SECONDS

Figure 12. The frequency of response in a single fiber stimulated by topical application of isotonic sodium citrate is reduced wlien the treated region is anodally polarized (first arrow). When circuit is opened (second arrow), there is a transient increase in response. The current passed into nerve in the treated region as shown in the diagram.

gradient. This was first reported by Fessard,^^ who observed such an effect in crab nerves which had been excited to activity by the applica- tion of alcohol or sodium thiocyanate. To study this problem further, we have passed a polarizing current through a calcium-depleted region of frog nerve, during the period of constant frequency of discharge. When the direction of current flow is such that it enters the nerve fiber in the chemically activated region, the average frequency of im- pulses is reduced for a brief time. As shown in figure 12, only a slight depression continues after a few seconds. When the polarizing current is terminated, there is a temporary increase in the frequency of im- pulses from the chemically activated region, followed by a return to the frequency that preceded the beginning of the current flow.

BRINK AND OTHERS: CHEMICAL EXCITATION OF NERVE 471

If the current is in the opposite direction, so that it flows out of the fiber in the calcium-deficient region, the sequence of frequency changes is reversed. Then, as the current starts, there is a transient increase of frequency, followed by a temporary depression when the current is interrupted.

Still obscure are the cellular mechanisms which account for an in- crease in the frequency of impulses from calcium-deficient nerve, when positively charged ions move outward across the fiber interface, or for

-130 L

Figure 13. The maximum change in frequency (impulses per second) in a chemically excited frog a fiber is proportional to the magnitude of polarizing current (in microamperes). Nerve ex- cited as described in figure 12.

Positive current : anode in the treated region ; negative current : cathode in treated region.

a decrease of frequency, when the ionic movement is reversed. The effects are, however, consistent with the long-established fact that a cathodally polarized region of nerve is more irritable, while anodally polarized nerve is less irritable. Accordingly, the effects are also in agreement with the view that an agent which reduces the stability of the nerve structure increases the frequency of chemically induced activity.

The magnitude of the transient increase or decrease in the frequency of impulses, caused by the passage of an electric current through a chemically activated nerve, depends upon the strength of current. Fig- ure 13 shows that there is, indeed, a linear relationship between the current strength and the maximal increase or decrease of impulse fre- quency caused by the current flow, within certain limits. This figure

472

ANNALS NEW YORK ACADEMY OF SCIENCES

also reveals the significant fact that there is, apparently, no minimal current strength that must be exceeded before the rhythmic discharge from the chemically sensitized nerve is modified. Any change in the direct current, no matter how small, flowing across the membrane of these chemically modified nerves, alters the rhythmic activity of the nerve and is reflected in the altered frequency of the propagated im- pulses. This is in contrast to the limiting threshold of current strength necessary for the excitation of a conducted impulse in a nerve with normal calcium content.

We have already said that potassium chloride causes a decrease in the frequency of the impulses developed in a calcium-deficient portion of

o 100 -

z

8

Ul (0

(T bJ Q.

m

bJ O)

-I

2

40 60

CONCENTRATION

Figure 14. Frequency of response due to action of isotonic solutions containing various propor- tions of sodium chloride and sodium citrate. Concentration in terms of per cent of isotonic solu- tion of sodium citrate.

a nerve. This is the effect of potassium chloride when its action has reached a steady state, but it is preceded by a transient increase in the number of impulses discharged per second. This stimulating action of potassium chloride also occurs in nerves with normal calcium content, but there it is of even shorter duration. Both the maximum frequency and the duration of the impulse discharge, caused by an increase of potassium chloride, are greater, the lower the calcium content of the fluid bathing the nerve. -° Thus, the removal of calcium from a nerve makes it more sensitive to the transient stimulating action of a moder- ate increase in the concentration of potassium chloride.

This increased sensitivity of calcium-deficient nerve to other chem- ical agents is further revealed in the experiments shown in figure 14. There, the nerve was made active by bathing a portion of it in isotonic

BRINK AND OTHERS: CHEMICAL EXCITATION OF NERVE 473

sodium chloride. When the frequency of response had become con- stant, the sodium chloride solution was replaced by one in which some of the sodium chloride had been replaced by sodium citrate. The fre- quency of impulse discharge then increased, as is shown. AVith each further increase in the proportion of sodium citrate, there was a further increase in the average number of impulses per second. When the nerve was subsequently returned to solutions containing successively smaller concentrations of citrate, there was a parallel decrease in the impulse frequency. Finally, in isotonic sodium chloride, the initial low degree of activity was resumed. This decrease in average fre- quency, associated with the return to isotonic sodium chloride, is obvi- ously not due to a restoration of calcium to the nerve. It seems prob- able, therefore, that a nerve made active by removal of calcium is sen- sitive to changes in the concentration of the citrate ion. This contrasts with the previously mentioned lack of effect of citrate upon a nerve in the presence of Ringer's proportion of calcium ions.

In a similar manner, the stimulating action of sodium thiocyanate is enhanced by first removing some of the calcium from the nerve. Also, tetraethyl ammonium chloride will stimulate frog nerve,^^ and we have found that its effectiveness in initiating impulses is greater, if the axon is sensitized by preliminary removal of some of the calcium.

Another quaternary ammonium salt of interest in this discussion is acetylcholine. Lorente de No has shown^^ that it does not alter the membrane potential of frog nerve, even in massive concentrations. On the other hand, Nachmansohn argues for the possibility of such an ac- tion, on the grounds that cholinesterase-inhibiting agents, which should permit the accumulation of acetylcholine, do cause depolarization of squid nerve. ^^ In our experience, acetylcholine does not induce a dis- charge of impulses when it is applied to the axons of a peripheral frog nerve trunk, and we have not been able to increase the frequency of chemically excited impulses by adding acetylcholine to the calcium- deficient fluid. Also, we have investigated the effects of this substance on mammalian nerve, by perfusing the stellate ganglion of a cat. In no case have we found any evidence that impulses are thus initiated in the pre-synaptic fibers within the ganglion, even though as much as 500 micro-grams of acetylcholine were added to each cc. of perfusion fluid. This was determined by observing that no impulses were discharged over the fibers of the preganghonic trunk (figure 15). Finally, there remains the contrasting and significant observation, that much lower concentrations of acetylcholine do cause the discharge of rhythmically recurring impulses in the post-synaptic neurons. The cell bodies or the

474

ANNALS NEW YORK ACADEMY OF SCIENCES

POSTGANGLIONIC ^ iJi^kUiitLi^tllM k *^^^^

NERVE mmm^>ikmmm»m* i^^nfff^Wm^ .p '0°

ACETYLCHOLINE

^G. /CC.

PREGANGLIONIC NERVE

MM^MIMMMMWMMMMMHMt) MMMMMMIMIWII

^ACETYLCHOLINE

100 \iQ>JCC.

PREGANGLIONIC

ACETYLCHOLINE

NtnVt "^ oOOpG./cc.

PREGANGLIONIC NERVE

*/mf0if^i0mmimmm

SODIUM CITRATE 380 uG. /cc.

Figure 15. Responses recorded from the preganglionic and postganglionic nerves of a cat's stellate sympathetic ganglion, during perfusion with acetylcholine and sodium citrate. Control records in the absence of a chemical excitant in the left hand column. Time: 0.1 sec.

immediately contiguous portions of their axons differ from axons in general in some way that makes them sensitive to the action of this agent.

This is a striking example of a specificity of nerve structure involved in the process of chemical excitation. It has been suggested that the basis of this differential action is the presence or absence of a myelin sheath that would prevent the rapid penetration of the acetylcholine." That is not likely to be the explanation of the contrasting effects in the experiments just reported, for the terminal portions of the pre-synaptic fibers within the ganglion are considered to be non-myelinated/"* as are the post-synaptic neurons.

Certain chemical agents, such as acetylcholine, are highly specific, with regard to the type of nerve structure they excite. Others, of which citrate and calcium-deficient solutions are examples, are quite general in their action.^^ Thus, a reduction of calcium ions in the perfusing fluid, or the addition of sodium citrate, causes the discharge of recurring impulses in both the pre-synaptic and the post-synaptic neurons, as in axons generally (figure 15).

The character of the response of ganglion cells to acetylcholine is, in many respects, analogous to the response of peripheral axons to chem- ical excitation. For each cell, there is a threshold concentration that must be exceeded before impulses are developed. This threshold dif- fers from cell to cell, but it is usually less than 25 micrograms of acetyl-

BRINK AND OTHERS: CHEMICAL EXCITATION OF NERVE 475

choline per cc, during continuous perfusion with solutions containing no inhibitor of cholinesterase. When the threshold for a cell is exceeded, it discharges impulses with a regular rhythm, and for extended periods that we have observed to be as long as an hour. The frequency of this

B

Figure 16. Impulses discharged from a single sympathetic ganglion cell in response to acetyl- choline in concentrations of A:25; B :50 ; C:100 micrograms per cc. Cat's stellate ganglion per- fused with a modified Ringer's fluid containing acet.\lcholine, but no inhibitor of cholinesterase. Impulses recorded from a fine strand of the postganglionic nerve. Time in seconds.

discharge increases with increased concentrations of acetylcholine (fig- ure 16) . Finally, at concentrations of about 200 micrograms per cc, the excitatory action ceases, the discharge of impulses is arrested, and the ganglion cells cannot be stimulated by volleys of preganglionic impulses.

The response of nerve to chemical excitants depends upon the totality of environmental agents. This has been emphasized before. It is a fact that is illustrated by the effects of the combined action of acetyl-

476

ANNALS NEW YORK ACADEMY OF SCIENCES

choline and other chemical agents on ganglion cells. For example, the frequency of impulses initiated by a certain concentration of acetyl- choline may be reduced by increasing the concentration of calcium in the perfusion fluid, or by reducing the concentration of potassium (fig- ure 17). Conversely, the frequency of discharge may be increased by

2 X NORMAL CALCIUM

5 X NORMAL POTASSIUM

i«iw

'■°'""'-iiiii|i|iiiiiii liiii II mill mil mil

NORMAL

!ilillllillllllllllllilli||l|lllll|llllllllli44^

4

NO CALCIUM

NO POTASSIUM

Figure 17. Impulses discharged from a single cell in a cat's stellate ganglion, during perfusion with six different solutions, all containing the same amounts of acetylcholine (40 /xg. 1 cc), but different concentrations of calcium and potassium. Time in seconds.

lowering the concentration of calcium or by augmenting the concentra- tion of potassium. These effects of calcium and potassium on the rhythmic action initiated by another chemical agent might be antici- pated from our knowledge of their effects on the electrical excitability of axons. It is, perhaps, worthy of comment that, regardless of whether the acetylcholine, calcium, and potassium act upon the same or different parts of the irritable mechanism, their combined effects become mani- fest in a modification of the rhythmic process which initiates the propagated impulses.

This repetitive process, which is a latent characteristic of nerve, is revealed in the discharge of impulses initiated by sensory stimulation or by chemical action. It is also to be observed in the periodic activity of nerve cells which are excited by the arrival of impulses in adjacent pre-synaptic endings. When the preganglionic fibers entering a sym- pathetic ganglion are stimulated by repetitive electrical shocks of high frequency, the cells discharge repetitively, but at a much lower fre-

BRINK AND OTHERS: CHEMICAL EXCITATION OF NERVE 477

quency. Furthermore, the cells di.scharge their impulses in no fixed temporal relation to the incoming, excitatory impulses. In short, the cells which are activated through tiie pre-synaptic endings initiate im- pulses at a Irequency that depends upcjn the characteristics of each cell, as well as upon the frequency and tlie number of pre-synaptic im-

FiGURE 18. Impulses discharged by a few sympathetic ganglion cells in response to stimu- lation of the preganglionic nerve at a frequency of 50 per second. The ganglion was perfused with a modified Ringer's solution containing various amounts of calcium chloride:

Uppermost record, 4.4 mM ; middle record, 2.2 mM ; bottom record, 1.1 mM. The middle rec- ord represents the normal level of calcium.

,Time in 0.1 seconds.

pulses.^' If these cellular characteristics are modified by any means (by nerve impulses or by chemical agents), the rhythmic processes are altered, and this modifies the frequency of their action.

Such a modification of the rhythmic response of a nerve cell to neural activation can be accomplished, as would be expected, by varying the concentrations of calcium ions in the synaptic regions (figure 18). If activity is excited in a ganglion cell by trains of pre-synaptic impulses, the frequency is decreased by raising the concentration ojf calcium in the perfusion fluid. Alternatively, the frequency of the impulses dis--

478 ANNALS NEW YORK ACADEMY OF SCIENCES

charged from the gangUon cells can be increased by decreasing the cal- cium concentration. It must be said that a further reduction of cal- cium in the perfusing solution may cause a complete block of ganglion cell excitation by impuls3s in the pre-synaptic fibers.

VI

Soms evidence regarding the nature of the cellular events which cause the more or less rhythmic discharge of impulses from a chemically excited region of nerve can be derived from a consideration of the temporal distribution of the impulses. Such a study suggests that there is a rhythmic excitatory process in nerve, of a fairly constant frequency, which may or may not produce an impulse each cycle. Definite evi- dence from several sources is now available for the existence of such a process. The role it plays in the regulation of the frequency of con- ducted impulses will be discussed in the following pages.

The earliest work on the discharge of impulses in single neurons re- vealed a temporal distribution of impulses that was more or less regu- lar, but not quite periodic. Thus, one of us in 1928, when commenting on the failure of the discharge from a fatigued muscle tension receptor, remarked that "one or more impulses drop out of an otherwise fairly regular series, the impulses becoming more and more scattered. "^^ The longer intervals were observed to be approximately equal multiples of the shortest time interval between successive impulses. Adrian observed a similar phenomenon in the discharge of injured mammalian nerve fibers, ^^ and such irregular intermittence appears in Pumphrey's"^ rec- ords of impulses from taste receptors. More recently, this occasional omission of impulses from an otherwise regular series was observed dur- ing the repetitive discharge caused by super-threshold direct current excitation (Erlanger and Blair," Fessard^^).

This same irregularity in the temporal distribution of impulses is a prominent characteristic in our experiments upon chemically excited axons. This will have been evident in some of the preceding records, but, for the more precise analysis of this phenomenon, additional ex- periments will be presented. The fibers were excited by removing Ca""* from a short length of nerve, by means of sodium citrate, as previously described. The measurements were made on records taken when the nerve was producing impulses at a constant average frequency. Under these circumstances, the temporal distribution may be regular or ir- regular.

The magnitudes of the time intervals between successive impulses in ^ certain series obtained in the above manner are plotted in figure 19,

BRINK AND OTHERS: CHEMICAL EXCITATION OF NERVE 479

as a function of the number of the impulse in the sequence. Obviously, the intervals are grouped about certain values, which are 6, 12, and 18 milliseconds. These same data are presented in the form of a dis- tribution plot in FIGURE 20, where the number of intervals in a certain range is plotted against the length of the interval. Since the ordinate is a measure of the probability of occurrence of the interval indicated

52-

"I- 30 29 ;6 27 26 29 24 23 22 21 n 20

a

- 19

S '■'

f 12

• • ••• •.

• • • •

■■■■''■''■' ■ ' ' ' ■

30 40 50

NUMBER OF IMPULSE IN TRAIN

I I I I I I I 1 I I I I I I I

Figure 19. The intPirals bptv.een impulses ipiorded from a chemirallv exritp<I .=infr'e filipr (fiog) .Tre aplHoxiiiKitelv intpfrial in I't.p'ps <if ;i !r;ist intni\nl. In lliis filipr, the least interval was about 6 milliseconds. Stimulation by localized rcmoxal of calcium fiom the axon.

on the axis of abscissae, it is obvious that all intervals are not equally probable. The most probable values are 6, 12, and 18 milliseconds. In all the frog fibers thus far examined, the most probable values of the least interval are in the range of 3-6 milliseconds.

The aliquot relations between these most probable values for Ihc fiber just cited suggest, again, that the longer intervals are due to the omission of one or more impulses from an otherwise continuous series. Such omissions could be due to failure of the impulses to be initiated in the chemically excited region, or to blocked conduction between that region and the recording electrodes. If the nerve is stimulated at high frequency by repeated electric shocks, the impulses travel over the nerve fiber and through the treated region. Consequently, there is no reason why each impulse initiated by chemical excitation should not, likewise, be conducted to the recording electrodes. We conclude,

480

ANNALS NEW YORK ACADEMY OF SCIENCES

therefore, tliat tlie omission of impulses from the series is not due to conduction block. The longer intervals in the records must be due to failure of one or more impulses to be initiated.

100 - 90 -

80 -

70 -

<n

-J

>60

K UJ

K

?50

ii. o

UJ '

30

20 -

10 -

0 -

J — I I I I l_i.

> ' ( I l_l I I — I — I — I — I — I — I J. <

8 10 12 14 16 DURATION Of INTERVAL (MS.)

18

20 22 24

FiGURG 20. Diagram representing the same data as in figure 19, but extended to over 500 suc- cessive intervals between impulses. Ordinates are number of intervals having a value in each 0.4 millisecond range. Inten-als longer than 24 ms. not shown.

It is difficult to account for these observations, except on the assump- tion that there is, in this nerve fiber, some rhythmic process, with an average period of 6 milliseconds, that maintains its rhythmic quality in- dependently of the initiation of impulses, once the impulses are started.

Arvanitaki^' and Hodgkin'^ have presented evidence that an impulse, initiated by electrical stimulation in unmyelinated nerve, develops from a local electrical response which occurs at the site of stimulation. Arvanitaki also showed^^ that this local response, which is elicited by electrical stimulation, may be cyclic in nerAcs deprived of calcium. Using the giant nerve fiber of the squid, wo have studied the develop- ment of this local electrical response in nerves excited solely by the removal of calchnn. We had two objectives: (1) to see if the local response appears before the initiation of the first impulse in a train, and (2) to ascertain whether a rhythmic local response could be pro- duced chemically, and independently of conducted impulses.

BRINK AND OTHERS: CHEMICAL EXCITATION OF NERVE 481

When part of a squid giant fiber is deprived of calcium ions, trains of impulses are initiated in this region and conducted to the recording electrodes. The nerve behaves, in' this respect, exactly like the myelinated nerve of a frog, but here conditions are more favorable for recording the difference of potential between the chemically excited region of the fiber and a remote portion. When this is done, it is found that the conducted impulses are preceded by a series of local periodic potential changes of variable amplitude and relatively constant fre- quency^^ (figure 21). The spacing between adjacent peaks of the

Figure 21. Local electrical response, recorded from a chemically excited region of giant axon of squid, is oscillatory and precedes the conducted impulses. The last ten oscillations on the right of the record initiated propagated impulses, which are much larger in amplitude than shown. Stimulation by topical application of isotonic sodium chloride.

local response, just before the conducted impulses appear, is the same as the spacing between the conducted impulses. It is obvious, as Arvanitaki concluded,^^ that the frequency of conducted impulses along the giant axon is determined by the frequency of the local excitatory process.

If relatively little calcium is removed from the nerve, local periodic electric changes may be observed which do not initiate propagated im- pulses. The local process is an essential part of the excitatory mecha- nism, but the cyclic changes initiate impulses only when a given cycle is of sufficient magnitude. Furthermore, the frequency of the local process is essentially independent of whether or not a conducted impulse is initiated by each cycle.

Figure 22 gives the frequencies of conducted impulses observed in nerves treated with solutions containing different concentrations of Ca"^, or with sodium citrate. The frequencies vary from 250 to 400 cycles per sec. This range is comparable to that which is characteristic of the undamped natural frequency of the nerve membrane, as calculated by Cole and Baker^^ from impedance measurements on squid nerve. This

482

ANNALS NEW YORK ACADEMY OF SCIENCES

parallelism between the range of frequencies in the local rhythmic re- sponse, the fundamental frequency in the trains of chemically initiated impulses, and the undamped natural frequency of the resting nerve membrane, support Cole's suggestion^* that it is the structural charac- teristics of the membrane which govern the periodic activity of nerve.

Two frequencies of nerve action have been described in the foregoing discussions of the response of nerve to chemical excitation. One is the average number of impulses conducted along the nerve per second; the other is the fundamental and relatively constant frequency of the excitatory process, which has, in the case of squid nerve, been identi- fied with the local electric response.

Nerve	A	B	C
5 nun* Calcium	No Calcium	No Calcium. 80 mm. Sodium Citrate
1	260
2	250
3	300
4	250	300
2G0	290	400
b		260	310
6			340
7		260	290
8		300	330
9			240
10			290

Solutions are modified sea water. Cal- cium and magnesium omitted in B and C. Magnesium omitted in A. KCl concen- tration and pH are same as in sea water.

Figure 22. Values for the fundamental frequency obsei-ved in 10 giant axons of the squid (Loligo pealii), estimated from frequency of conducted impulses. Stimulation by topical application of in- dicated isotonic solutions.

The relation between these two frequencies is illustrated by the fol- lowing experiment, which makes use of the fact that a polarizing cur- rent may modify the average frequency of impulses discharged from a calcium-deficient region of nerve. In figure 13, the outward flow of current across the chemically altered nerve membrane caused a tran- sient increase of the average impulse frequency. The distribution plot for the intervals between impulses from the non-polarized nerve is shown in the lower half of figure 23. The intervals between some im- pulses were 3.2 milliseconds; other impulses recurred at intervals which were about two times this value.

In accordance with the concepts which have been developed in this section, we may say that there was a rhythmic excitatory process, in the chemically modified portion of the nerve, of a fairly constant frequency.

BRINK AND OTHERS: CHEMICAL EXCITATION OF NERVE 483

50 r

40

30

20

10

0 -

1

J I I I I I'll l_J

8

10

u. 50 O

2 40

CD

30

20

10

0

J ' I I I I I I I I I

8

10

DURATION OF INTERVAL (MS.)

Figure 23. The average number of impulses per second in a chemically excited single fiber (frog) can be markedly increased, without much change in the fundamental penorl.

Lower graph : Frequency of occurrence in a train of impulses of intervals having values indi- cite.l en I lu- abscis.-iie. Stimulation Uv topical app.ication ol a fo.ution containjig sodium citrate. The average number of conducted impulses per second was about 220.

Upper graph : Similar anals sis ol a consecutive series of inipu..'-es recorded at height of increa^^ed re^non^'e evoked hv cathodal po'arization of rhcmically treated region (see figures 12 and 13). The average number per second of conducted impiilses was about 305.

Some of these cycles of local change failed to initiate conducted im- pulses, and, accordingly, some of the intervals between successive im- pulses were integral multiples of the least time interval of 3.2 milli- seconds. The average impulse frequency was 220 per second.

484 ANNALS NEW YORK ACADEMY OF SCIENCES

A similar analysis of frequency relations, during the time of in- creased response caused by the cathodal polarization, was then made. The distribution of intervals between impulses is plotted in the upper part of FIGURE 23. The total number of impulses there considered is the same as in the lower plot. During this time, the most probable period for the excitatory process was only slightly changed, but the probability that impulses would recur at the longer intervals was then practically zero. The action of the polarizing current caused more im- pulses to be discharged at the basic interval, and this increased the aver- age frequency to 305 per second. In general, changes in this average frequency of impulses, caused by superimposed chemical or physical agents, occur with only slight modification of the most probable least interval. It should be pointed out, however, that, under certain con- ditions of intense stimulation, this interval can be decreased.

Since the temporal distribution of impulses initiated by chemically excited nerve appears to be regulated by a local rhythmic process, which is determined by the intrinsic characteristics of the nerve, the possibil- ity arises that the same mechanism may govern the discharge of im- pulses from naturally excited sense organs and motor nerve cells. Whether the mechanisms described above do have such a general sig- nificance, must wait upon further investigations.

BIBLIOGRAPHY

1. Misske, B.

1930. Biochem. Zeit. 219: 320.

2. Lipton, S. R.

1934. Am. J. Physiol. 109: 457.

3. Blumenfeldt, E.

1925. Biochem. Zeit. 156: 236.

4. Gerard, R. W.

1930. Proc. See. Exp. Biol. & Med. 27: 1052.

5. Brink, F., T. Sjostrand, & D. W. Bronk 1939. Am. J. Physiol. 126: P442.

6. Adrian, E. D.

1932. The Mechanism of Nervous Action. Univ. of Pennsylvania Press.

7. Adrian, E. D., & Y. Zotterman

1926. J. Physiol. 61: 151.

8. Adrian, E. D., & D. W. Bronk

1928. J. Physiol. 66:81.

9. Brink, F., & D. W. Bronk

1937. Proc. Soc. Exp. Biol. & Med. 37: 94.

10. Chang, T. H., M. Shaffer, & R. W. Gerard

1935. Am. J. Physiol. Ill: 681.

11. Davies, P. W., & F. Brink

1941. Am. J. Physiol. 133: P257.

12. Bronk, D. W., F. Brink, & P. W. Davies 1941. Am. J. Physiol. 133: P224.

BRINK AND OTHERS: CHEMICAL EXCITATION OF NERVE 485

13. Adrian, E. D.

1930. Proc. Roy. Soc. London B 106: 596.

14. Erlanger, J., & E. A. Blair

1935. Am. J. Physiol. 114: 328.

16. Fessard, A.

1936. Propri^t^s Rythmiques de la Matifere Vivante. I. Nerfs Myelinfeses. Hermann & Cie, Paris.

16. Katz, B.

1936. J. Physiol. 88: 239.

17. Arvanitaki, A.

1939. Arch. Int. de Physiol. 49: 209.

18. Sjostrand, T., F. Brink, & D. W. Bronk 1938. Proc. Soc. Exp. Biol. & Med. 38: 918.

19. Fessard, A.

1936. Propri^tes Rythmiques de la Matiere Vivante. II. Nerfs Non Mye- lineses. Hermann & Cie, Paris.

20. Brink, F., T. Sjostrand, & D. W. Bronk 1938. Am. J. Physiol. 123: P22.

21. Cowan, S. L., & W. G. Walter

1938. J. Physiol. 91: 101.

22. Lorente de No, R.

1944. J. Cell. & Comp. Physiol. 24: 85.

23. Bullock, T. H., D. Nachmansohn, & M. A. Rothenberg 1946. J. Neurophysiol. 9: 9.

24. Ranson, S. W., & P. R. Billingsley 1918. J. Comp. Neurol. 29: 313.

25. Bronk, D. W.

1939. Symposium on the Synapse. C. C. Thomas, Baltimore.

26. Feldberg, W., & A. Vartiainen 1935. J. Physiol. 83: 103.

27. Larrabee, M. G., & D. W. Bronk Unpublished.

28. Bronk, D. W.

1929. J. Physiol. 67:270.

29. Pumphrey, R. J.

1935. J. Cell. & Comp. Physiol. 6: 457.

30. Erlanger, J.

1937. Electrical Signs of Nervous Activity. University of Pennsylvania Press, Philadelphia.

31. Hodgkin, A. L.

1938. Proc. Roy. Soc. London B 126: 87.

32. Brink, F., & D. W. Bronk 1941. Am. J. Physiol. 133: P222.

33. Cole, K. S., & R. F. Baker 1940-1941. J. Gen. Physiol. 24: 771.

34. Cole, K. S.

1941-1942. J. Gen. Physiol. 25: 29.

ELECTRICAL CHARACTERISTICS OF ELECTRIC TISSUE

By R. T. Cox, C. AV. Coaxes, and M. Vertner Brown

The Department oj Physics, The Johns Hopkins University, Baltimore, Maryland;

The New York Aquarium, New York Zoological Society; and the

Department of Physics, College of the City of New York.

The group of electric fishes comprises a number of very different varieties, both fresh water and marine. All of them possess special organs capable of producing transient electric discharges, which, in some species, are quite weak, but in others, are powerful enough to give a severe shock. These organs vary widely among the different species in their shape and size and in their position and orientation in the body of the fish. They are alike in having a common unit of structure, the electroplax.

The arrangement of the electroplaxes has its highest geometrical regu- larity in the electric rays, Torpedo and Narcine}' ^ In the electric or- gans of these genera, they are piled in columns, an average one of which contains about 400 electroplaxes in Torpedo marmorata and per- haps 300 in Narcine brasiliensis. Each column extends from the ven- tral to the dorsal surface of the body. A number of them, side by side, form each of the two electric organs, which lie in the disk-like body of the fish to the right and left of the body cavity, just outside the line of gill slits. In each organ, there are four or five hundred columns in Torpedo marmorata and Narcine brasiliensis, and about a thousand in T. occidentalis. During the discharge, the current traverses each organ in the direction from its ventral to its dorsal face. Thus, the columns of electroplaxes discharge in parallel, while, within each column, the electroplaxes act in series (figure 1).

In Torpedo and Narcine alike, the electric tissue comprises about one sixth of the whole volume of the fish. In the electric eel, Electrophorus electricus, it makes, by contrast, about one half. Organs of such a size must conform, in part, to the shape of the fish, and hence there cannot be so regular an arrangement of the electroplaxes as in the rays. It is customary to distinguish in Electrophorus three pairs of organs: the main organs, which extend along the posterior four-fifths of the length of the fish ; the much smaller organs of Hunter, which lie under the main organs along their entire length; and the organs of Sachs, which lie

(487)

488

ANNALS NEW YORK ACADEMY OF SCIENCER

Figure 1. Embryo Narcine brasUiensis, dorsal and ventral views. Negative faces of electric organs at e, e. (From Zoologica.)

a

over the main organs in the posterior half of the fish. The organs of Hunter are separated from the main organs only by a thin layer of muscle, and the tissue is identical in adjacent parts of the two pairs of organs. It seems more reasonable to regard the organs of Hunter as parts of the main organs than as a distinct pair (figure 2) .

cox AND OTHERS: ELECTRIC TISSUE 489

The main organs are of nearly uniform cross-section for some dis- tance from their anterior end, but they taper toward the tail, conform- ing to the ventral surface of the body and, in the posterior portion, to the under surface of the organs of Sachs. Although this tapering pre- vents the arrangement of electroplaxes in uniform columns, the series- parallel array already noted in the rays is, nevertheless, clearly dis- cernible. In Electrophorus, the axis of polarity is along the length of the fish. Thus, the organs, in comparison with those of the rays, are very much elongated along the line of series connection of the electro- plaxes.

^m^^P^^0J

'M"^

Drawing: by Ralph Graeter

Figure 2. Electrophorus, with skin removed to show the electric organs. A, main organs; B, organs of Sachs, overlapping the main organs; C, organs of Hunter. (From Bull. N. Y. Zool. Soc.)

This contrast between the most highly specialized electric fishes of the fresh-water and marine groups has a significance which seems to have been noticed first by du Bois-Reymond.^ The combination of a fixed number of electromotive elements to supply power to an external circuit of given resistance is a well-known problem in the theory of electric networks. The solution of the problem shows that maximum power will be delivered to the external circuit by the combination of the elements in a series-parallel array, such that the resistance of the combination is equal to that of the external circuit. Thus, if the external resistance is high, more electromotive elements will be joined in series; if it is low, more will be joined in parallel. Fresh water has a much higher specific resistance than sea water. Consequently, if the condition for maximum external power is equally approximated in the different genera, the organs of the fresh-water fishes will be elongated, and those of the marine fishes will be flattened, along the axis of polar- ity. Most of the varieties confirm such a generalization. The electric skates, which are marine fishes with weak electric organs elongated in the direction of the axis of polarity, make a rather puzzling exception.

The arrangement of the electroplaxes, by means of which the main organs of Electrophorus are accommodated to the tapering body of the fish, is simple and rather interesting.^' * It is best described in terms of a transverse slice just thick enough to contain a single layer of elec- troplaxes. Near the anterior end of the organs of a fish about 1 meter

490 ANNALS NEW YORK ACADEMY OF SCIENCES

long, the thickness of this single electroplax layer is about .01 cm. In fish of this length, the cross-section of the organs near the anterior end has an average area of about 30 cm^. Hence, the volume of the single electropax layer is 0.3 cm-''. The organs taper caudally, but, as the cross section decreases, there is a compensating increase in the thickness of the layer, and the volume is nearly uniform over most of the length of the organs. The structure is much as if uniform layers were assem- bled in a long column, and then the column were drawn out thin toward one end, the layers being changed in shape, but not in volume.

These long organs found in Electrophorus offer remarkable advan- tages in the study of the action of electric tissue in the living fish. The series array of electroplax layers is accessible for electrical connection all along its length, rather than only at the ends as in the rays. The variation in structure makes it possible to compare in the same speci- men the electrical characteristics of electroplax layers of very different dimensions. Also, Electrophorus, which comes to the surface to breathe, can be kept for some time out of water without injury, and the elec- trical characteristics of its tissue remain constant during an interval in which it can produce a thousand or more electric impulses.

In our observations, the fish is removed from the water and laid in a dry wooden trough. Electrodes made of aluminum strip 1 cm. wide may be placed in any of a number of slots in this trough. These make contact with the skin adjacent to the electric organs and, when they are connected to a cathode-ray oscillograph, it is possible to record photographically the discharge of the part of the organs included be- tween the electrodes (figure 3). The measurements made with the oscillograph are found not to vary appreciably with the area of con- tact between skin and electrode, provided this area is not less than a few square centimeters. Of course, no appreciable dependence on the choice of a metal for the electrodes is to be expected, since the voltages measured are very much greater than any contact potential differences.

When the electrodes are at the extremities of the main organs of a mature specimen, and the external circuit is open, so that there is no electric current outside the body of the fish, the average peak voltage is about 370 volts. ^ The highest voltage we have measured is 550. There is also a discharge of much lower voltage, which is evidently pro- duced by the organs of Sachs, since it is observed only when some part of these organs lies between the electrodes. In immature specimens, the voltages are smaller. The voltage of the main organs increases with their length, at an average rate of 8 volts per cm., until the organs attain a length of about 50 cm. The organs may ultimately attain

cox AND OTHERS: ELECTRIC TISSUE 491

three times this length, but, in any group of longer specimens, the varia- tions in voltage appear to be random.

By measuring the peak voltage between electrodes 5 or 10 cm. apart at different places along the organs, it is possible to compare the volt- age per cm. in different parts. At the anterior end, where the number of electroplax layers is greatest, the voltage per cm. is also greatest. It decreases caudally, as the electroplax layers thicken. The voltage per electroplax layer is roughly uniform along the organs. In four

T

t »

Figure 3. Oscillographic traces of the discharge of Electrophorus:

(a) An impulse from Sachs' organs followed by five impulses from the main organs; sweep period, 50 msec.

(b) Impulses from the main organs, superimposed by successive sweeps; sweep period, 4 msec. (From Zoologica.)

specimens, values from 0.11 to 0.16 volt were found at the anterior ends of the organs. Somewhat lower values are found in the posterior parts, but, over most of the length, the voltage per electroplax layer is 0.1 volt or more.^' ^ Values around 0.1 volt per electroplax layer are found also in Narcine brasiliensis. In Narcine, however, and also in Torpedo, the voltage of the organs varies widely with the condition of the fish. When a conductor is connected between the electrodes, so that the electric tissue produces an external current, the peak voltage is lower than with the external circuit open. If care is taken not to tire the fish, the voltages obtained with a given resistance are reproducible. The resistance R of the external conductor being known, the external cur- rent / is found from the measured voltage V, by the relation, / = V/R. When conductors of successively lower resistance are employed, the voltage continues to decrease, as the current increases. The results of

492

ANNALS NEW YORK ACADEMY OF SCIENCES

such a series of measurements are most conveniently shown by plotting the values of the voltage against the values of the current, one plotted point representing the voltage and current obtained with a given re- sistance (figure 4). It is found that the points lie near a straight line

lOO .

amperes

Figure 4. Peak voltage vs. peak current in external circuits of different resistance joining elec- trodes on main organs of Electrophorus. (From J. Gen. Physiol.)

with both the electric eel and the electric rays.* The meaning of this result is that the electric tissue, at least at the peak of the discharge and within the uncertainty of the measurements, can be described elec- trically in terms of electromotive force and ohmic resistance.

If the external resistance were made negligibly small, the voltage also would be negligible. The corresponding current, estimated by extrapolating the straight hne of the graph to zero voltage, is the maxi- mum current of the organs. This maximum current varies from one specimen to another and, in Electrophorus, it varies between different parts of the main organs. In an average specimen, around 1 m. in length, it is about 1 amp. at the anterior end of the main organs. Values of about 4 amp. have been found in adult specimens of Narcine brasiliensis, and a value of 120 amp. was roughly estimated in a single, very large specimen of Torpedo occidentalis.^' *• ^

These great variations are due more to differences in the cross-sec- tions of the organs than to differences in the electrical characteristics of the tissue. The maximum current per unit area of the electroplax layer was found to have values in Electrophorus from .02 to .06 amp. per cm.^, the average being about .04. In Narcine brasiliensis, the value is about 0.1 and, in the specimen of Torpedo occidentalis just mentioned, it was about 0.2 amp. per cm.^ (In this calculation and

cox AND OTHERS: ELECTRIC TISSUE 493

others to follow, it is implied that the two paired organs discharge simul- taneously. In Narcine and Torpedo, where the organs are far apart, a simple experiment shows this is true. The evidence in respect to Electrophoru.s points to the same conclusion, but it is not certain.)

If opening the external circuit prevented any current in the organs, the voltage measured with the circuit open would be equal to the elec- tromotive force of the part of the organs included between the elec- trodes. Then, the maximum voltage per electroplax layer would be equal to the electromotive force of the layer, and its quotient by the maximum current per cm.^ would be the resistance of 1 cm.- of the layer at the peak of the discharge. However, even with the external circuit open, there must be closed circuits within the body of the fish, during the discharge. Consequently, the voltage per electroplax layer must be somewhat less than the electromotive force of the layer. Its quotient by the maximum current per cm.- is still of some significance as a lower limit for the resistance of 1 cm.^ of electroplax layer. It seems likely, also, that this lower limit is not very much less than the actual value. In Electrophonis, minimum values thus found for the resistance of 1 cm.- of electroplax layer have varied in different speci- mens between 2 and 5 ohms. In two specimens of Narcine brasiliensis, the values were about 1 ohm.

It is interesting that, in Electrophorus, the resistance of unit area of electroplax layer does not increase caudally, although the thickness of the layer increases about ten-fold from the anterior to the posterior end of the main organs. This suggests that the resistance resides prin- cipally at boundaries in the electric tissue.

From the observations considered thus far, it is seen that the single electroplax layer in Electrophorus has characteristic electric quantities which are roughly uniform, in spite of wide variations in the arrange- ment of the electroplaxes and in the size of the fish observed. Also, in those cases in which comparison has been possible, it is found that these quantities have the same order of magnitude in Narcine and Torpedo as in Electrophorus.

We have studied, in a number of different specimens of Electrophorus, the variation during an impulse of the electrical characteristics of the tissue.* The oscillograph was connected, as already described, to electrodes placed 10 cm. apart against the main organs of each speci- men studied. Impulses with the external circuit open and closed, through resistances from 400 to 50 ohms, were recorded photographic- ally. With each specimen and each value of the external resistance, measurements were made on a number of oscillographic traces at each

494

ANNALS NEW YORK ACADEMY OF SCIENCES

of several short time intervals after the beginning of the impulse. Val- u s of the voltage measured at any one interval, with a given fish and a given external resistance, were then averaged. As with the measure- ments at the peak of the impulse, the results are conveniently shown by l)l()tting the voltage against the external current. A typical set of measurements is thus shown in figure 5. The points along any one

-as

O ampefes 0.5

1.0

Figure 5. Each line shows voltage vs. current at one instant, in external circuits of different resistance joining electrodes on main organs of Electro jhorus. (From J. Gen. Physiol.)

line show values of voltage and current obtained with different external resistances, at the same interval after the beginning of the impulse. The graph farthest to the top and right shows the measurements at the peak and, thus, corresponds to the single graph shown in figure 4. The other two graphs show the measurements at two later instants during the impulse.

Although a straight line cannot be drawn precisely through the plotted points of the measurements at a given interval, we have been unable to detect, in the series of observations, as a whole, any system- atic deviation from a linear relation. It appears, therefore, that the tissue can be described electrically in terms of electromotive force and ohmic resistance, not only at the peak of the impulse, but at other times as well.

A simple diagram for such a description is shown in figure 6. In reference to this figure, let E denote the electromotive force of the part of the organs included between the electrodes at p and q, and let r de-

cox AND OTHERS: ELECTRIC TISSUE

495

note the internal resistance. The current in whatever circuits are closed within the body of the fish is treated, somewhat arbitrarily, as traversing a single path of resistance, R\

R'

t

i.

fA/^W

LVWvVWJ

R

Figure 6. Simple diagram for describing the impulse from the main organs of Electrophorus. (From J. Gen. Physiol.)

Let I, I, and /' denote the currents in r, R, and R', respectively. Since the current i branches to form the currents / and /', it follows that:

i = i + r.

Let V denote the voltage measured by the oscillograph, connected at p and q. This voltage may be reckoned in any of the three branches of the network, and, thus, we obtain three expressions for V, as follows: V = IR, V = I'R', V = E - ir.

Eliminating i and /', among these four equations, we obtain two ex- pressions for the external current / in terms of the voltage V:

/ =

I = -- V

R ' r ' \r ' R'

The first of these equations is used to find the current / from the measured voltage V, by means of the known resistance R. The values of / and V are plotted as in figure 5. The other equation is then used to interpret the graph so obtained. If the resistances r and R' are ohmic, they are constants in this equation, which is then a linear rela»- tion between / and V, such as is actually found to exist. If, in this equation, we let V = E, we find the corresponding value of / to be —E/R\ If we suppose that E and R' have the same values at dif- ferent instants during the discharge, this equation states that the graphs for the different instants, when extrapolated to negative values of /, will intersect at a point. The co-ordinates of this point, moreover, will determine the values of E and R' .

Actually, the lines shown in the figure, which are typical of those obtained from measurements after the peak of the impulse has been attained, do nearly meet in a point. The fact that the graphs of the

496 ANNALS NEW YORK ACADEMY OF SCIENCES

measurements at later instants are steeper than the graph of the peak values, indicates a rise in the internal resistance r, after the peak is passed.

These results led us, earlier, to suppose that the electromotive force might be constant throughout the impulse and between impulses as well, the discharge being caused by a transient drop in the resistance from a very high resting value.' At that time, we had not succeeded in plotting graphs of voltage and current at instants during the brief interval of rising voltage. Both the steepness of the rising phase of the oscillographic trace and its consequent faintness in the photo- graphs made measurement difficult in this interval. Measurements which we have made more i-ecently iiave obliged us to reconsider our earlier opinion. The graphs of voltage and current obtained from measurements during the interval of rising voltage do not meet at a point. Moreover, even during the interval of falling voltage, we find that deviations, which were formerly within our estimated errors of measurement and which we, therefore, supposed were accidental, ap- pear consistently in the later observations.

The variation in resistance during the interval of falling voltage seems, in any case, well established. It seems probable, also, that the electromotive force is at least approximately constant during this phase. Our immediate object is an estimate of the total electric energy pro- duced in an impulse. Fortunately for this purpose, the time after the attainment of peak voltage is most of the duration of the impulse. Although the changes in the electrical characteristics, during the brief phase of rapidly rising voltage, remain uncertain, the assumptions made about them in the calculation of the energy can be varied widely, with- out changing the result by more than about 10 per cent.

From the equations already given, it follows that the current i traversing the electric tissue is related to the current /, measured in the external circuit by the equation :

7 = (1 + R/R/)r.

In this equation, E is known, and R' is determined by the intersection of the voltage-current graphs. Thus, the current in the electric tissue is found.

The charge q which passes through the tissue in one impulse is given by :

q = jidt,

where t denotes the time, and the integration is performed over the dur- ation of the impulse. The integration can easily be done graphically.

cox AND OTHERS: ELECTRIC TISSUE 497

The charge passing through 1 cm.^ of electroplax layer is found by dividing q by the cross-sectional area of the electric organs. (The measurements were made at the anterior end, where the cross-section is nearly uniform.) Since the fish on which the measurements were made were not killed, the cross-section had to be determined indirectly from external measurements. Two methods were employed. In the first method, the girth of the fish was measured. The cross-section of the organs was then estimated by comparison with measurements on dissected fish, on the assumption that the cross-section of the organs bears a constant ratio to the square of the girth. In the second method, a simple mechanical device was employed to trace the outline of the cross-section of the live fish. The area enclosed by this outline was measured, and the cross-section of the organs was taken as 59 per cent of the total area, this percentage having been obtained from meas- urements on a number of dissected specimens. When both methods were used, the agreement between the results was fairly good, the values determined by the two methods showing a mean deviation of around eight per cent.

(However, the use of only the first method, in another experiment, led to a rather serious error." The number used then as the ratio of the cross-sectional area of the organs to the square of the girth of the fish was obtained from measurements on only two specimens. Also, the girth of these sections was not measured in the same way as on the live fish, and this led to a further discrepancy, which was increased when the girth was squared. The correction of the resulting error to accord with our new measurements on a larger number of specimens requires that the values of electric energy per gm. and impulse given in the paper referred to should be reduced about 40 per cent. Instead of inferring, as we did in a subsequent paper,^ that the total electric energy is about equal to that of the breakdown of phosphocreatine and the production of lactic acid, we should now infer that the electric energy is about six tenths of the sum of the energies of these two chem- ical processes. The correction brings this result into fair agreement with that reported by Nachmansohn, elsewhere in this volume.)

The electromotive force of the part of the organs included between the electrodes was determined by the point of intersection of the volt- age-current graphs. This quantity was divided by the distance be- tween the elctrodes, to give the electromotive force per cm. along the column of electroplax layers. The product of the electromotive force per cm., regarded as constant during the impulse, by the charge travers- ing one cm.^ of electroplax layer, is the total electric energy per cm.^ produced in one impulse.

498 ANNALS NEW YORK ACADEMY OF SCIENCES

The charge passing through the organs, and hence, also, the energy, depend on the resistance of the external circuit, as well as on the elec- trical characteristics of the tissue. For comparison with the chemical measurements, it is, of course, essential that the external resistance should be the same in the electrical, as in the chemical, experiments. Otherwise, the choice of an external resistance is, within limits, un- important. The fish on which the electrical measurements have been made form three groups according to size, with average lengths of 67, 103, and 180 cm. We have used an external resistance of 200 ohms with the two groups of smaller length, and of 100 ohms with the other group. These resistances were roughly the same as the internal re- sistances, averaged over the time of the impulse, of the part of the or- gans between the electrodes. Consequently, the condition of the ex- periment approximated the requirement for maximum energy in the external circuit.

The results are summarized in table 1, which shows, for each speci- men and for the average of each group, the electromotive force per cm., the charge passing in one impulse through 1 cm.^ of electroplax layer, and the total electric energy produced in one impulse in 1 cm.^ of elec- tric tissue. With the electrical units employed, the product of the elec- tromotive force per cm. by the charge per cm.^ would give the energy per cm.^ in microjoules. For more convenient comparison with the chemical energies reported by Nachmansohn, the energies have been given, instead, in microcalories. Since the tissue has nearly unit spe- cific gravity, the energy per cm.^ may be taken as the energy per gm., without serious error.

It has already been mentioned that the assumption made in the cal- culations, that the right and left organs discharge simultaneously, though probable, is still unproved. If the fact should be that the or- gans discharge separately, then the given values of the charge per cm.^ and the energy per cm.^ would have to be doubled. The same correc- tion would have to be applied to the values of the chemical energy, and, therefore, the comparison of the electrical and chemical energies made by Nachmansohn is valid in either case.

The individual values of electromotive force per cm. show a mean deviation of 12 per cent from the average of the group. The mean deviation of the charge per cm.^ is 15 per cent, and that of the energy per cm.^ is 23 per cent. The risk of a serious uncertainty in the aver- ages, beyond that indicated by these deviations, depends on the possi- bility that the equations used in the computation are seriously in error. The evidence by which they were justified has already been given.

cox AND OTHERS: ELECTRIC TISSUE

Table 1 Electrical Measurements on Eleclrophorus

499

	Length of fish (cm.)	Cross- section of electric organs (cm.2)	Electro- motive force per cm. (volts)	Charge per cm. 2 and impulse (micro- coulombs)	Energy per cm.' and impulse (micro- calories)
External resistance, 200 ohms	60 62 65 68 68 70 72 74	10 12 13 10 16 14 13 13	20 15 15 19 23 20 21 20	11 20 18 22 21 20 21 19	55 72 64 99 118 95 108 89
Ave.	67	13	19	19	88
E.xternal resistance, 200 ohms	90 93 96 100 102 104 107 112 121	28 33 23 32 38 31 31 36 27	16 13 13 18 14 12 18 12 15	18 12 14 14 12 10 13 11 16	71 35 43 59 40 29 54 32 57
Ave.	103	31	15	13	47
External resistance, 100 ohms	164 169 186 187 187 188	67 70 70 67 72 66	15 12 11 10 13 10	5.0 9.5 8.0 8.7 11.3 10.8	18 27 21 20 35 26
Ave.	183	69	12	8.9	24

It seems unlikely that the actual values of the electrical quantities should be much lower than those given in the table. The voltage meas- ured with the external circuit open sets a lower limit to the possible value of the electromotive force. The maximum measured voltage is, on the average for all the specimens, 77 per cent of the value found for the electromotive force. It seems reasonable that the current in cir- cuits closed in the body of the fish should cause a voltage drop of 23 per cent. Similarly, the charge passing through the external resist- ance, which is obtained directly from simple oscillographic measure- ments, sets a lower limit to the possible magnitude of the charge tra-

500 ANNALS NEW YORK ACADEMY OF SCIENCES

versing the tissue. This lower limit is, on the average, 82 per cent of the computed total charge. It seems unlikely that the external charge is, actually, much nearer than this to the total charge. From these considerations, we should judge it improbable that our assumptions involve a systematic error, whereby the actual values of the energy should be consistently more than about 15 per cent lower than those found by the method we have used.

The question as to how much the actual values may exceed those we find, presents a greater uncertainty. Although, as was just explained, it is difficult to propose an equivalent network for the electric organs which will dissipate much less energy than we allow for in internal currents, there is no difficulty in proposing a network to dissipate more energy. This follows from Thevenin's theorem of electric networks, according to which any electromotive force inferred from external measurements may be regarded, alternatively, as the open-circuit volt- age of a concealed network containing a higher electromotive force. On the other hand, our calculations ascribe to the single electroplax layer an electromotive force about as high as any that are found at boundaries in bioelectric phenomena. This suggests that the actual values of the electrical quantities are not very much greater than those we calculate. For the energy, which is the most uncertain of these quantities, twice the calculated value appears to be a safe upper limit.

REFERENCES

1. Gotch, F.

1900. The Physiology of Electrical Organs. Textbook of Physiology, Vol. 2. Young J. Pentland, Edinburgh and London; MacMillan, New York. (This article contains references to a number of original sources.)

2. Cox, R. T., & C. M. Breder, Jr.

1943. Observations on the electric discharge of Narcine brasiliensis (Olfers). Zoologica 28: 45.

3. Cox, R. T., W. A. Rosenblith, J. A. Cutler, R. S. Mathews, & C. W. Coates

1940. A comparison of some electrical and anatomical characteristics of the electric eel, Electrophorus electricus (Linnaeus). Zoologica 25: 553.

4. Cox, R. T., C. W. Coates, & M. V. Brown

1945 Relations between the structure, electrical characteristics, and chemical processes of electric tissue. J. Gen. Physiol. 28: 187.

5. Coates, C. W., & R. T. Cox

1945. A comparison of length and voltage in the electric eel, Electrophorus electricus (Linnaeus). Zoologica. 30: 89.

6. Coates, C. W., & R. T. Cox

1942. 01)servations on the electric discharge of Torpedo occidentalis. Zoologica. 27:25.

7. Nachmansohn, D., R. T. Cox, C. W. Coates, & A. L. Machado

1943. Action potential and enzyme activity in the electric organ oi Electrophorus electricus. II. Phosphocreatine as energy source of the action potential. J. Neurophysiol. 6: 383.

SOME BASIC ASPECTS OF THE ACTIVITY OF ELECTRIC PLATES

By Alfred Fessard Institut Marey, Paris, France

The most conspicuous manifestations of the activity of electric or- gans are: (a) the electrical discharge itself; (b) variations of the elec- trical conductivity of the tissue; (c) thermal effects; (d) chemical (metabolic) transformations.

Every individual feature of the total discharge is more or less closely associated with the species of fish examined, with its shape and dimen- sions, and with the structural arrangement of the physiological units (or electric plates) composing the organ. However, on the cellular dimension scale, the behavior of these plates shows a striking unity, and is generally considered as an ordinary bioelectric phenomenon. The classical conceptions of nerve physiology are applicable here, and these, combined with our rapidly progressing knowledge of neuromus- cular transmission mechanisms, have been used for explaining the pro- duction of electricity in these organs. Here, briefly exposed for each essential manifestation of activity, are the most significant facts in favor of the unitary conception, together with some details concerning our experimental contribution to this problem.

A. The most recent determinations of the electromotive force per plate show, on Electrophorus electricus, a fairly uniform value: viz., 0.14 volts (Coates and Cox^^). This is in agreement with the highest value found by Curtis and Cole^*' for the action potential from the squid giant axon. As for the time course of the discharge, non-typical shapes are often observed (Cox, Coates, and Brown^^), which may wrongly be considered as representing the elementary process. Our own research in this field has definitely convinced us that these non- typical wave shapes are due to statistical dispersion effects, the causes of which are low velocity at the periphery and differences in length of winding nerve endings.

In our experiments (most of them still unpublished), we used organs of Torpedo marmorata and of Raia undulata. Our purpose was to record the discharge of a single plate and then to interpret the complex discharges in terms of their components. Columns of electric tissue

(501)

I

502 ANNALS NEW YORK ACADEMY OF SCIENCES

were carefully isolated, and transverse slices about 1 millimeter thick cut out with fine scissors. Such fragments contain 10 to 20 plates, many of them damaged by the dissection. The intact ones retain a good excitability for hours. By transverse electrical stimulation at threshold values, it is sometimes possible to record responses which, be- yond all doubt, are elementary (plate 4A, lower record). They obey the all-or-nothing law and are the conspicuous components of the com- plex waves obtained with slightly higher intensities. In Torpedo, they last 1.75-2.25 msec, (rising phase, 0.5-0.7 msec.) ; in skates (Auger and Fessard^), they are much longer, about 12 msec, (rising phase, 2-2.5 msec). Now the discharge of a whole column is noticeably of longer duration (plate 4A, middle record). In spite of the shortness of the nervous tracts contained in these small pieces of tissue, latencies vary, from less than 1 msec, to more than 4 msec, and dispersion of com- ponents lengthens the wave duration to 2 or 3 times its elementary value. The natural discharge starting off the whole organ is hardly longer. The long distance command is transmitted by high velocity fibers of large diameter, and these do not introduce such an important shift in components as do the thin nerve branches at the periphery. The elementary electrical process is, therefore, comparable to a nerve action potential, never being diphasic, as conduction is absent. Posi- tive after-potentials have never been observed. Mention must be made of some variations in the declining phase, which is sometimes longer than usual, especially in skates. In these cases, the discharge is more like an end-plate potential than an action potential.

B. The analogy between the electric organs and other excitable systems retains its value when the electrical conductivity is studied during the discharge. Using the impedance bridge method, applied by Cole and Curtis^^ to the squid giant axon, we have observed^ a transient drop of impedance during the activity of Torpedo and Raia organs. Recently, Cox, Coates, and Brown^^ have obtained indirect evidence of a diminution in ohmic resistance during the discharge of Electroph- orus electricus.

Plate 4B shows one of our records obtained with a double beam cathode ray oscillograph. The upper record is the discharge of an isolated column {Raia), placed in one branch of an impedance bridge; the lower one shows the reappearance of the 15,000 cycles oscillation feeding the bridge. The two phenomena start simultaneously, instead of showing a shift as in the nerve, a fact easily interpreted as a lack of conduction The impedance change is slight, less than 10 per cent, but

FESSARD: ACTIVITY OF ELECTRIC PLATES 503

here the presence of inactive tissue around the plates renders illusory any attempt to introduce quantitative measurements.

C. In observations unfortunately never repeated, Bernstein and Tschermak^" have shown an initial cooling of the organ, when the dis- charge was externally derived through a resistance. They interpreted this fact as being inconsistent with a chemical mechanism of the energy supply. Meyerhof-" criticized this interpretation, without, however, denying the fact, which should be reinvestigated with the more per- fected methods now available.

D. Recent investigations into the biochemistry of electric organs proved their metabolism to be quite similar to that of nerve tissue or striated muscle. Cholinesterase has been shown to be very abundant in electric tissue (concerning this significant presence of cholinesterase in electric organs, as at all neuronal surfaces, see Nachmansohn^"). This fact implies that acetylcholine plays an essential role here, as elsewhere.

It was under the impetus of Nachmansohn that research was under- taken by Feldberg, Fessard, and Nachmansohn to detect the prob- able presence of acetylcholine in electric organs, and to study the part it plays in the production of the discharge. In these experiments, posi- tive results were obtained in Torpedo organs. During stimulation, the ester appears in the perfusate, and arterial injections of micro-doses of acetylcholine produce long-lasting electrical changes. ^^' ^®

The next step was when Nachmansohn and his collaborators" dem- onstrated that cholinesterase is localized at the active surfaces in elec- tric organs, its concentration being strictly correlated with the maxi- mum voltage and, consequently, like the electromotive force, with the number of plates per unit length. Now the acetylcholine release is able to start the chain of reactions, beginning with the phosphorylated compounds acting in nerve or muscle metabolism, which we now know to be present, together with the associated enzymatic system, in the electric organs (Baldwin and Needham,* Kisch,^^ Nachmansohn et al.^^). The research in this field has now reached a quantitative as- pect, and the energy liberated by the breakdown of phosphocreatine and the formation of lactic acid during the discharge can be compared with the electrical energy released (Nachmansohn et oL'^) .

In concluding this short survey, we can say that there is a striking convergence of data, allowing the electric plate to be put side-by-side with the nerve and muscle units, from the point of view of their electro-

504 ANNALS NEW YORK ACADEMY OF SCIENCES

chemical properties. However, some uncertainties persist as to the particular mode of production of their electrical discharges. We are still ignorant of the way in which acetylcholine may act at an inter- face to generate electricity. This ignorance is general, but we do not even know, in electric organs, where this active interface lies; and, to assign a definite physiological significance to the plate, we are faced with at least three different views. Although we are far from being able to give a satisfactory answer to these three debated questions, we shall briefly discuss the last two in the light of the experimental evi- dence we have obtained, up to the present, in our research on the Torpedo.

I. PHYSIOLOGICAL SIGNIFICANCE OF THE ELECTRIC

PLATE

A. According to a current view, the electric plate is considered as an element of a true effector, and this implies the notion of its physiological individuality. As in the case of the muscle, this special effector would be normally set into activity through a "relay" mechanism, and should show, by direct stimulation, an excitability of its own. This com- parison with striated muscle is all the more justified, as both effectors have a common embryonic origin (the Malapterurus organ excepted). On the other hand, functional analogies seem to exist between the electrical discharge and the muscular contraction (Marey^^) .

B. However, the regression of all vestiges of striation in the adult stage of the more powerful electric organs; the absence of myosin (re- placed by mucin^) among the proteins of electrical tissues; and, above all, the simultaneous disappearance of direct and indirect excitability, under different conditions (nerve degeneration, fatigue, cooling) have thrown serious doubts upon the value of the analogy. Some authors have gone so far as to consider the possibility of a purely nervous ori- gin of the discharge. Gotch^^ wrote that "the excitatory electromotive change may be nothing more than the fact that when an excitatory process travels down a nerve, the nerve trunk becomes negative to its terminal cross-section." The maximum value of a nerve action po- tential is the same as that of an electric plate, and it is suggestive to note that the elementary plate discharge and the single fiber action potential of the nerve commanding the organ have exactly the same duration (plate 4A, upper part) . Furthermore, the chemical data are far from being opposed to this conception, which tends to reduce the role of the plate to that of a simple support for a richly expanding nervous branching.

FESSARD: ACTIVITY OF ELECTRIC PLATES 505

C. Nevertheless, as Rosenberg points out in his Review on the sub- ject,^'^ it seems unlikely that the plate has only this passive role of support. According to our views, the main difficulty of a purely nervous theory lies in the fact that the orientation of the discharge does not agree with the symmetries offered by the nerve distribution.

An electric organ is sometimes described as an accumulation of modi- fied motor end-plates. This view is more in agreement than any other with the ontogenic facts and with the analogies suggested by histology. There are many nuclei in the plate near the innervated face, as in the sarcoplasmic sole of the striated muscle. Couteaux recently described, at the myoneural junction, a rod-like structure which is strikingly analogous to that long believed to be specific of electric organs.^*

The existence, now well established, of a localized response at the nerve-muscle junction (end-plate potential, e.p.p.), preceding the muscle fiber propagated impulse, renders the analogy still more evi- dent. This e.p.p., like the discharge, is accompanied by an impedance change that follows the same time-course as that observed in electric organs (Katz^*) : i.e., a non-delayed rising phase and a maximum ef- fect near the inflexion point of the potential variation.

Our experiments on small isolated fragments contribute to show that Hypothesis A cannot be retained, as it is really impossible to isolate the plate as a functional effector unit. They are also more in agreement with Hypothesis C than with B.

a. AVe thought it useful, at first, to revert to the degeneration test, for the observations mentioned by Garten^^ were not sufficient in num- ber, and the methods for electrical detection have improved since that time. 30 animals were operated on and examined at different inter- vals after nerve sections on one side. Some survived more than 2 months, and this was sufficient to detect histological signs of alteration in the terminals (Fessard and Pezard-°). Such signs began to appear on our Torpedoes only 5 or 6 weeks after the operation (average tem- perature 14° C). Before that, the organs were found excitable, al- though needing more and more current. Excitability in any form (electrical, chemical, mechanical) totally disappears after about 7 weeks. Osmic acid staining then shows fragmentation of the last branches. Deprived of its terminal innervation, the electric organ is decidedly incapable of activity.

b. No sound conclusion can be deduced from the old results on poisoning by curare. Most of the previous experimenters (namely^

506 ANNALS NEW YORK ACADEMY OF SCIENCES

Moreau, Babuchin, Gotch) denied its action, and this was, at first, also our opinion, as 24 hour immersions in 1 per cent curare solutions had no marked effect. In other experiments, we had noted that hght cuts made with a razor blade along the longitudinal surface of a column did not interfere with its capacity for delivering good responses. We thought that this treatment might facilitate the penetration of drugs, and we immediately got a positive result (Auger and Fessard^). 1 per cent solutions acting during 1-2 hours abolish all excitability. As op- posed to the striated muscle, the electric organ becomes inexcitable after curarization.

c. If the electric plate is a real functional unit, its activity must be greatly impaired or completely suppressed by severe mutilation. Iso- lated columns were divided lengthwise into three narrow strips, each plate being thus fragmented into 3 parts, and severely damaged. This is a complementary situation, compared to that of the nerve degenera- tion experiments, the plate itself being practically destroyed, but the finer nerve tracts at the endings being only partially damaged. In spite of this drastic treatment, the preparation remains excitable and gives discharges of smaller amplitudes, but of normal shape.

d. One may object to the strict vahdity of arguments (a) and (6), as they concern situations in which the plate is modified in some way. For instance, after nerve degeneration, the organ shows some reduc- tion in thickness, and the curare poisoning may have altered the prop- erties of the plate. Results obtained with isolated intact columns may supply indirect, but more satisfactory, evidence.

Using very strong electrical stimuli, we had expected to get a true effector response, as the nerve impulses would arrive during the re- fractory period of this hypothetical effector unit. Different electrode positions were tried, the results of which we observed from the point of view of threshold, latency, amplitude, and components of complex waves. The results show that any of these parameters (and the varia- tions thereof) depends upon, and can only be explained by, the char- acteristics of the nerve supply pattern. They appear to be determined by the symmetry of the nerve distribution, not by that of the plates. These are some of our observations:

1. No difference in latency or in the form of the discharge can be observed in supra-maximal longitudinal excitation, whether the cur- rent is or is not in the direction of the discharge. No systematic dif- ferences in threshold values were found.

FESSARD: ACTIVITY OF ELECTRIC PLATES 507

2. In the lengthwise stimulation of Torpedo columns, the latency can never be reduced to less than a certain minimum (3 to 5 msec), however strong the current (even if we approach the lethal value). This can be explained by the special distribution of nerves in Torpedo columns, if one admits that excitation is always localized at the bend- ing points of the nerve branches. The situation is comparable to that in which Rushton^* made his observations on excitation of bent nerves. Before entering the plates, the nerve branches run along the edges of the prismatic column, then each axon bends at right angles and, by mul- tiple division at the bending point, sends small transverse twigs to sev- eral plates. This point {"bouquet de Wagner") is at the same time a Ranvier-node, and we assume, as most likely, that it is the most distal one from which excitation can start in longitudinal stimulation.

3. This assumption is confirmed by the fact that the latency is not irreducible. Strong transverse stimulations lower it to less than 1 msec. (Auger and Fessard^), clearly showing that the long latencies are due to nerve conduction in the plate plane, and not to some elabo- ration process in the plate itself. When the intensity is lowered, the transverse latency increases regularly, but never exceeds, in normal conditions, the shortest latency observed in longitudinal stimulation. This is perfectly comprehensible, if we localize the excitation at bend- ing points nearer and nearer the "bouquets de Wagner," provided that we adopt the current opinion, according to which the threshold values diminish, the further we are from the nerve extremities.

e. The non-existence of a relay action similar to that of the neuro- muscular command is further indicated by the absence of repetitive re- sponse when acetylcholine is injected intra-arterially into an isolated organ,^®' ^^ although we have shown that this drug exerts, in this case, a marked depolarizing effect.

/. Other drugs were introduced into the interior of the plates by the same technique as described in (6) (curare poisoning): eserine(10"*), which lengthens up to more than 4 times the declining phase of the ele- mentary discharge; atropin (lO"'^), which suppresses all excitability in 1-2 hours; curare, which has the same effect as atropin, but in doses ten times larger. During the course of both intoxications, the threshold intensity progressively rises. The duration of the elementary dis- charge does not change or even become shorter. These data confirm the cholinergic nature of the nerves supplying the organ. They also add supplementary evidence in favor of the similitude between the electric discharge and the end-plate potential (cf. Kuffler^*'' ").

508 ANNALS NEW YORK ACADEMY OF SCIENCES

II. ORIGIN AND LOCALIZATION OF THE ELECTROMOTIVE

FORCE

These are largely a matter of speculation, as they now concern the level of molecular organization. Thermodynamic data on one hand, experiments with microelectrodes on the other, would be most useful. Awaiting these, we must content ourselves with discussing the points on v/hich experimental results are available: for instance, the most im- portant problem of plate polarization in the resting state.

a. The classical hypothesis is that of Bernstein.'' This postulates a permanent superficial polarization of the plate boundaries. The arrival of a nerve impulse results in a local transient annulment of this polari- zation, in accordance with the general assumption. Now, only one side of each plate is innervated and capable of being depolarized. As, at rest, the potential difference between the two extremities of an elec- tric organ is approximately zero, in spite of the coupling in series of the plates, one must suppose an exact compensation of the electromo- tive ft)rce developed on one side of each plate by that developed on the other side (figure 2, Schema I). We thus arrive at that strange con- ception of two distant polarized layers, endowed with different prop- el ties and yet electrically charged in exactly the same way. Their properties are different, because one is supposed to discharge itself through a sudden internal leak due to collapse, while the other starts discharging without collapse into the external medium. Yet not a sign of a double evolution of potential can be observed in the course of the elementary discharge. On the other hand, these opposite layers are not situated in similar regions, from the point of view of tissue structure and chemical environment. It is very unlikely that they should develop the same electromotive force.

b. Another hypothesis has been recently proposed by Cox, Coates, and Vertncr Brown,' '^ who assume a constantly present electromotive force, non-compens.ited by another opposed electromotive force, but hidden by tlie high resistivity of an interface. This is not conceivable, in our opinion, without caj^acitive properties by which the interface appears as passively charged (fijure 2, Schema ITl. According to the present concept, "the discharge would be started by a very large and rapid drop in the resistance." We have seen that this drop in resist- ance really exists, but we cannot conceive of the resting voltage, sup- posedly present, being lowered, say from 500 volts to 5 millivolts, by the simple interposition of biological membranes, the resistances of which are, at the highest estimate, 1000 ohms/cm.^ in the nerve interfaces

FESSARD: ACTIVITY OF ELECTRIC PLATES 509

(Cole and Hodgkin^^). Furthermore, electrometric determinations on isolated portions freed from the internal derivations normally present in the intact animal, should reveal much higher resting potentials. This was never observed.

c. It appears to us that the following question should be resolved first: Do polarized layers really exist in the plate, previous to its state of activity? To prove this, it is necessary to communicate in some way with the interior of the plates.

1. We have tried piercing slowly a column from the electric organ of Torpedo with a fine metallic electrode and have observed small re- petitive discharges, due to a mechanical excitation, which we have shown to be caused by irritation of the nerve twigs encountered.^ When the electrode is extremely fine, a number of plates may be perforated without being excited and without giving rise to those systematic vari- ations in potential we might expect when passing through one plate to the following.

2. We also took small groups of columns, one of which we slit lat- erally with the edge of a heated blade. The measurements were started immediately with the ordinary method of opposition, one electrode be- ing placed on the killed region, the other as far as possible from it, on the intact tissue of the same column. The resting potential had to be dissociated from the long-lasting residuals of activity, following the excitation produced by the lesion (cf. Gotch^-). Three methods were tried: first, allowing the residuals to vanish; second, diminishing their disturbing effect by a transverse arrangement of the electrodes; third, using a degenerated preparation.

The results in these 3 cases are exemplified in figure 1. In figure 1 (2), the A electrode, being a little more dorsal than B, is positive at the start, according to the direction of the discharges. However, it rapidly reverts to its steady potential value, which is negative, rela- tively, to B. In (3), no initial discharge is present as expected.

In all cases, no value higher than 5 mv. has been obtained for this rest potential. This is 20 to 30 times less than the elementary dis- charge. We cannot believe that such a discrepancy can be completely due to a shunt effect.

The preceding results, incomplete as they are, throw a serious doubt on the value of the first and second hypotheses. The alleged perma- nent polarization may not, therefore, exist, at least not at sufficient strength to play the more important part in the discharge. This sug- gests a third hypothesis that we formulated once,* and according to

510

ANNALS NEW YORK ACADEMY OF SCIENCES

B

m

i

i

ded.

o

I

I

.4.

•minutes

FiGURB 1. Measurements of potential differences (Va-Vb) immediately following a localized injury (hatched zone).

Torpedo organ, 3 different cases (see text).

which the activity in the electric organ simply consists in the transient appearance of a membrane polarization, rather than in the depolariza- tion of a previously polarized surface. This is the meaning of the schema represented in figure 2, Schema III. However, it is difficult

	M
+	—		—	+
4-	—	1	—	+
+	—	—	+
+			—	+

Figure 2. Hypothetical schemas proposed to explain the production of an electromotive force in electric organs.

N indicates the innervated side of the plate. The dotted lines represent what is supposed to collapse during activity'.

The dotted lines in II are intended to show a passively charged membrane in contrast with other interfaces, which are supposed to be actively polarized by an internal electromotive force.

The figure makes it clear that IV may be described as a synthesis of I, II, and III.

to believe in the formation of a polarized layer at the moment when the responsible membrane collapses, as is indicated by the drop of its re- sistance. This schema, like the other two, appears really far too simple.

FESSARD: ACTIVITY OF ELECTRIC PLATES 511

Finally, the schema that we shall tentatively propose as the most representative of our present knowledge and the most promising as help for future research, is the one shown in figure 2, Schema IV. It is supposed to represent a complex molecular structure of the same type as that used by modern biophysicists in their hypotheses on the consti- tution of the molecular membranes (Danielli and Davson^^. Such double layer leaflets are built up with lipoid and protein molecules, and are widespread at the cell surfaces. Similar arrangements have been assumed to be present, for instance, in the rods of the retina, which has well-known electrical properties.

Figure 2 makes it clear how this last hypothetical schema may be considered as a synthesis of the other three. In effect, it borrows an idea from each of the preceding theories. It is like the first, inasmuch as it admits the presence of two opposing polarized layers, of which only one can be neutralized. It borrows from the second the idea that a high resistivity layer exists (probably made of oriented lipoid chains in the intermediate region), which collapses during the discharge; and it is in agreement with the third, in admitting that the electromotive force is strictly localized at the innervated face of the plate, where it becomes apparent during the short period when one of the layers is depolarized.

BIBLIOGRAPHY

1. Auger, D., & A. Fessaxd

1938. C. R. Soc. Biol. 128: 1067.

2. Auger, D., & A. Fessard

1939. Ibid. 131:765.

3. Auger, D., & A. Fessard 1939. Ann. de Physiol. 15: 261.

4. Auger, D., & A. Fessard

1939. Livro de Homenagem Prof. A. e M. Ozorio de Almeida. 25. Rio de Ja- neiro, Brazil.

6. Auger, D., & A. Fessard 1941. C. R. Soc. Biol. 135: 76.

6. Auger, D., & A. Fessard 1929. Ibid. 102:582.

7. Bailey, K.

1939. Biochem. J. 33 : 255.

8. Baldwin, E., & D. Needham

1937. Proc. Roy. Soc. London B 122: 197.

9. Bernstein, J.

1912. Elektrobiologie. 118.

10. Bernstein, J., & A. Tschermak 1906. Pflueger's Arch. 112: 439.

11. Coates, C, & R. Cox

1945. Zoologica30:89.

512 ANNALS NEW YORK ACADEMY OF SCIENCES

12. Cole, K., & H. Curtis

1939. J. Cien. Physiol. 22: 649.

13. Cole, K., & A. Hod?kin 1939. .J. Gen. Physiol. 22: 671.

14. Couteaux, R.

194.5. C. R. Soc. Biol. 139:641.

15. Cox, R., C. Coates, & V. Brown 1945. J. Gen. Physiol. 28: 187.

16. Curtis, H., & K. Cole

1942. .J. Gell. & Comp. Physiol. 19: 135.

17. Danielli, J., & H. Davson

1943. The Permeability of Natural Membranes. Cambridge.

18. Feldberg, W., A. Fessard, & D. Nachmansohn

1939. J. Physiol. 97:2.

19. Feldberg, W., & A. Fe3sard 1942. J. Physiol. 101:200.

20. Fessard, A., & A. Pezard

1940. C. R. Soc. Biol. 134: 525.

21. Garten, S.

1910. Handb. d. vergl. Physiol. 105.

22. Gotch, F.

1887. Phil. Trans. B. 487.

23. Gotch, F.

1888. Phil. Trans. B. 329.

24. Katz, B.

1942. J. Neurophysiol. 5: 169

25. Kisch, B.

1930. Biochem. Z. 225: 183.

26. Kuffler, S.

1942. J. Neurophysiol. 5: 18.

27. Kuffler, S.

1943. Ibid. 6:99.

28. Marey, E.

1887. Travau.x du Laboratoire de 1' Institut Marey 3: 1.

29. Meyaihof, O.

1926. Thermodynamik des Lebensprozesses. Handb. d. Physik 11: 254.

30. Nachmansohn, D.

1945. Vitamins and Hormones 3: 337.

31. Nachmansohn, D., R. Cox, C. Coates, & A. Machado

1942. J. Neurophysiol. 5: 499.

32. Nachmansohn, D., R. Cox, C. Coates, & A. Machado

1943. Ibid. 6:383.

33. Rosenberg, H.

1918. Handb. der norm. u. pathol. Physiol. 8: 876.

34. Rushton, W.

1927. J. Physiol. 63:3.57.

FESSARD: ACTIVITY OF ELECTRIC PLATES 513

PLATE 4

514 ANNALS NEW YORK ACADEMY OF SCIENCES

Plate 4

A. Middle record: response of an isolated column (Torpedo organ) to an elec- trical stimulus directed along its main axis.

Ivower : response of a small fragment of tissue ; transverse stimulation ; thresh- old intensity.

Upper: single fiber action potential (electric organ nerve). Time scales in milliseconds.

B. Double beam oscillograph record.

Above: discharge of a fragment of electric tissue (Raia).

Below: impedance change test with an alternating ciuTent of 15,000 cycles.

The whole activity wave lasts 12 milliseconds.

Annals N. Y. Acad. Scr.

Vol.. XI.VII, Aht. 4, PuTK i

FKSSARl): ACTIVITY OK Kl.KCTRIC PI.ATK;-

PHYSIOLOGICAL FUNCTION FROM THE STANDPOINT OF ENZYME CHEMISTRY

By D. E. Green

Departments of Medicine and Biochemistry, College of Physicians and Surgeons, Coluinbia University, New York, N. Y.

It is a curious fact tliat, although there is general recognition and agreement that the cell is a chemical system, none the less the full im- plications of this truism have yet to be appreciated in some fields of physiological investigation. Perhaps the explanation is to be found in the preoccupation of biochemistry, until very recently, with problems of the structure of cellular constituents and with their estimation. Classical biochemistry represented to the physiologist the extension of histology to the chemical field. The study of what we may call chem- ical morphology was hardly calculated to attract physiologists or to arouse their interest in the chemical basis of physiological function. However, the interest of biochemistry has been shifting gradually from the purely structural problems to the dynamic chemical events of the cell. Our present knowledge of the chemical mechanisms of the cell has grown sufficiently for it to be ignored no longer by those who are concerned with the study of physiological function.

We may conceive of the cell as a chemical factory in which literally thousands of chemical reactions take place, cheek by jowl, without mutual interference. Some of these reactions are concerned in the syn- thesis of structural components of the cell and others in providing chemical energy for carrying on the activities of the cell. Practically without exception, these reactions do not proceed spontaneously. They require the presence of protein catalysts, which we call enzymes. Each enzyme is distinct, chemically, from all the others, and is uniquely specialized for its particular catalysis. If this picture of the cell is correct, then it follows that all dynamic activities including physiolog- ical function must be reducible to terms of enzyme chemistry. In other words, physiological function and enzyme chemistry are two sides of the same coin. I hope, in the short time at my disposal, to marshal the available evidence which justifies this interpretation.

In the syndromes of avitaminosis, we observe profound morphological and physiological abnormalities. The recognizable signs of each of the vitamin deficiencies are too well known to be discussed here. The point

(515)

516 ANNALS NEW YORK ACADEMY OF SCIENCES

I want to make, however, is that these abnormalities which cover the en- tire gamut of physiological dysfunction can be explained completely in terms of certain enzyme systems. Vitamin Bi, B2, Bg, and the P-P factor have all been shown to be the prosthetic groups of certain enzyme systems. When these vitamins are not available in the diet, the active enzymes cannot be formed in the cell. In consequence of the failure of these enzyme systems to function properly, an abnormal physiological situation develops, which, if uncorrected, will lead to death. Which organ first registers the effect of a particular deficiency is determined by the amount of the reserves of enzymes containing the vitamin and by the relative importance of this set of enzymes in the economy of the organ. Thus, in Bi deficiency in the pigeon, the brain is the first organ to register disturbed function, presumably because there are no reserves of this vitamin in the brain and because the active enzyme formed by the vitamin plays a key role in the metab- olism of brain. One may well raise the point that, if, as in the avita- minosis, the causal link between the physiological disturbance and the effect on enzyme systems is unquestioned, then surely there is a good case for assuming the same link between the normal physiology and enzyme systems.

Woolley^ has introduced the use of anti-vitamin reagents which, by virtue of their structural resemblance to the vitamins, are able to pre- vent the vitamins either from being incorporated into enzymes or from functioning as prosthetic groups. These anti-vitamins produce, in a relatively short period, the same syndromes which arise from de- priving the diet of an animal of a particular vitamin. Not only are these anti-vitamins valuable for speeding up the process of avita- minosis, but the profound pharmacological effects which they induce permit correlations between the action of reagents on certain enzyme systems and the pharmacological consequences. In other words, the anti-vitamins have focussed attention on the fact that the effects of certain, if not all, pharmacological agents can be explained completely in terms of effects on enzyme systems.^' ^ In recent years, a rich litera- ture has grown up to deal with this correlation. There are now at least 14 instances in which the pharmacological effects of certain reagents can be explained in terms of a specific effect on an enzyme system. Thus, iodoacetic acid, which induces muscle rigor, has been found to inhibit, in minute concentration, the triosephosphoric dehydrogenase which catalyzes an essential step in lactic acid formation. This paralysis of the triosephosphoric dehydrogenase accounts for all the pharmacolog- ical effects produced by iodoacetic acid. Fluoroacetic acid, the highly

GREEN : PHYtilOlJJGICAL FUNCTION, ENZ YME C HEM 1ST R Y 5^7

toxic agent discovered in Chemical Warfare Research, has been shown to inhibit the enzyme systems involved in the metaboHsm of acetic acid. The classical pharmacological reagents, strychnine, eserine, and prostigmine, have been shown to exert their effects exclusively by virtue of their paralysis of cholinesterase. The effects of cyanide on cyto- chrome oxidase, fluoride on enolase, and chlorine on the triosephos- phoric dehydrogenase, are other examples of this correlation. Perhaps even more unexpected, has been the identification of various toxins with enzymes. Thus, spreading factor, the agent which facilitates the rapid diffusion of injected substances through dermal tissue, has been shown to be identical with hyaluronidase, a mucolytic splitting enzyme. The hemolytic principle of Clostridium welchii toxin and that of snake venom have been shown to be lecithinases, and the hemolytic effects are completely explicable in terms of their ability to weaken the lipoid membrane of the red blood cell by hydrolysis of the lecithin contained therein. During the war, some English workers, led by McFarlane,* identified one of the toxins produced by the gas gangrene organism as collagenase, a proteolytic enzyme which dissolves the connective tissue sheath of muscle. The action of this enzyme explains the pulping of muscle observable in advanced cases of gas gangrene poisoning. At the present moment, it would be premature to assume that all specific pharmacological agents which work at high dilutions are active by virtue of their effects on enzyme systems. On the other hand, it is pertinent to point out that no other principle of mechanism has been established for any pharmacological agent which has been studied. Apart from the dictates of caution, there is no good reason not to anticipate that, eventually, all effects of specific pharmacological agents will be reducible to terms of enzyme chemistry.

The study of endocrines has always been one of the most active fields of physiological investigation, and it is of interest to inquire to what extent hormones can be related to enzyme phenomena. Until quite recently, hormones were held up as notable exceptions to the rule that substances which act at high dilutions must be enzymes or parts of enzymes, or must specifically affect some enzyme system. Some recent research, however, fails to confirm the hormones as exceptions to the enzyme-trace substance thesis. No doubt, everyone is aware of the epoch-making discovery of Cori and his group,^ that one of the hor- mones of the anterior pituitary inhibits the action of hexokinase, and that, in turn, this inhibition is released by insulin. We have here a clear blueprint for the way in which hormone antagonism can be effected. A key enzyme system which controls some metabolic process

518 ANNALS NEW YORK ACADEMY OF SCIENCES

can be regulated by a set of hormones, one of which inhibits, while the other releases the inhibition. All students of endocrinology have long been aware that hormones regulate metabolic processes, and it is not surprising to find in one instance, at any rate, that the regulation oper- ates at the level of the enzyme systems. Houssay and his colleagues in the Argentine have presented cogent evidence that renin, a kidney hormone, is a type of proteolytic enzyme which hydrolyzes one of the plasma proteins to form a pressor substance. In this instance, the hormone regulates metabolic processes by actually assuming an enzy- matic role.

In still another direction, there has been confirmation that enzymic phenomena underlie essential physiological processes. The brilliant work of Beadle*' and his school have made it abundantly clear that the regulation of growth and development by the hereditary units of the cell, viz., the genes, is exercised through control of enzyme systems. They have shown that each gene determines the synthesis, probably, of a single enzyme. Whereas some of the hormones regulate metabolic reactions by slowing up or speeding up an enzyme reaction, genes regulate by determining the synthesis of an enzyme. Remarkably little is known of the mechanisms by which enzymes are synthesized, but it would appear that, whatever the mechanism, the genie material will be implicated.

The mere recognition that enzymic phenomena underlie physiological function is, of course, only the first step in the biochemical analysis. Obviously, the exercise of physiological function requires a source of energy, and the energy must arise in enzyme-catalyzed reactions. But how is the energy converted into the manifold forms required by the cell? How is chemical energy converted into mechanical energy of contraction or electrical energy of nervous conduction (to mention two examples) ? There are no transforming elements in the cell, such as the storage battery. Until recently, this problem of energy con- versions was shrouded in deepest fog, but some light has managed to penetrate. It now appears, from the work of various laboratories, that the contraction of the myosin molecule may be coupled with the enzymatic hydrolysis of adenosine triphosphate. The contraction of muscle is now visualized as the integration of the single contractions of myosin molecules arranged in linear series. The picture is, of course, very crude, and probably will be modified by further research. However, if the basic facts are correct, then we have a blueprint for visualizing energy transfers at the enzyme level. Adenosine triphos- phate represents a readily tapped supply of the chemical energy gen-

GREEN: PHYSIOLOGICAL FUNCTION, ENZYME CHEMISTRY 5^9

erated by the process of glycolysis. Because of the close proximity of adenosine triphosphatase and myosin, some of the chemical energy of hydrolysis is absorbed by the myosin molecule, which then under- goes simultaneous contraction. In other words, myosin is acting as a kind of transformer element for the conversion of chemical energy to mechanical energy.

We have to consider the possibility that, just as myosin is specialized for muscular contraction, chlorophyll or visual purple for photochem- ical reactions, and hemoglobin for oxygen transfer, so there may be one or more proteins in nerve specialized for the reactions which underlie the propagation of a nerve impulse. The knowledge that acetylcholine and adrenaline are the chemical agents involved in nerve conduction, is merely the introduction to the problem of mechanism. Undoubtedly, these substances react with special proteins. It is the transformations which these special proteins then undergo that is the basis of the phe- nomenon of nerve transmission.

REFERENCES

1. Woolley, D. W.

1945. Science. 100: 579.

2. Green, D. E.

1946. Currents in Biochemical Research: 149. Interscience Publishers, Inc., New York.

3. Green, D. E.

1941. Advances in Enzvmology. 1: 177. Interscience Publishers, Inc., New York.

4. McFarlane, R. G., & J. D. MacLennan 1945. Lancet 2: 328.

5. Price, W. H., Carl F. Cori, & Sidney P. Colowick

1945. J. Biol. Chem. 160: 633.

6. Beadle, G. W.

1946. Currents in Biochemical Research: 1. Interscience Publishers, Inc., New York.

CHOLINESTERASE

By Oscar Bodansky*

Medical Division, Chemical Warfare Service, Edgewood Arsenal, Maryland^

On the basis of his studies on the pharmacology of acetylcholine, in 1914, Dale stated: "In the blood at body temperatures it seems not im- probable that an esterase contributes to the removal of the active ester from circulation."^ In 1926, Loewi and Navratil observed that acetylcholine, as well as "vagus substance," was rendered inactive by in- cubation with heart extract.^ However, such inactivation did not occur after the heart extract had been heated or subjected to ultra- violet irradiation. These observations by Loewi and Navratil in- augurated the study of the enzyme, cholinesterase.

A heat-labile substance which is capable of hydrolyzing acetylcholine is found very widely distributed in the organs and fluids of the body. It cannot be assumed that, apart from this common property of hy- drolyzing acetylcholine, the other properties of this enzyme are the same in all these tissues. Our present discussion should most fit- tingly concern itself with the properties of this enzyme as found in nervous tissue, and should determine the extent to which these proper- ties play a role in nerve activity. However, most of the data avail- able for discussion describe chiefly the properties of cholinesterase found in serum, red cells, and, to a lesser extent, in the whole brain. The extent to which these data apply to the properties of cholinesterase, at synapses and in other nerve tissue, should be carefully evaluated.

RELATION BETWEEN SUBSTRATE CONCENTRATION AND

REACTION VELOCITY

We shall first turn our attention to the relation between the rate of action of cholinesterase and the concentration of the substrate, acetyl- choline. Examination of the data shows that two types of relation- ships hold. The first type appears to follow the Michaelis-Menten formulation:^

-^ = ^ (1)

where v is the reaction velocity at substrate concentration S, F^ax is the maximum reaction velocity occurring at infinite substrate con-

* Lt. Colonel, M. C, A. U. S.

t Present address: Dept. of Pharmacology, Cornell University Medical College, New York, N. Y.

(521)

522 ANNALS NEW YORK ACADEMY OF SCIENCES

centration, and Kg is the dissociation constant of the intermediate enzyme-substrate complex.

The second type of relation obtaining among some cholinesterases appears to be one in which inhibition of reaction velocity occurs at higher substrate concentrations. This relation has been found to hold for some other enzymes, such as lipases, catalase, oxygenase.* The reactions between substrate and enzyme may be formulated as follows: E + S ?^ ES (active)

ES + (71 - l)S:f± ESn (inactive)

ES ^E + P.

The relation between substrate and reaction velocity has been devel- oped by Haldane^ for the reaction where n = 2, as follows:

' max *-'

V =

S" S + A% + ^

A2

(2)

where K2 is the dissociation constant of the inactive enzyme-substrate

r p on r oin— i

compound - — rlF^^ — When the velocity is plotted against the log-

[Ebn\

arithm of the substrate concentration, a bell-shaped curve is obtained. The values for K,, Ko may be obtained by transformation of the above equations, according to the method of Lineweaver and Burk.* Since there are very few and incomplete data in the literature for cholinesterases which follow this type of substrate-reaction velocity relationship, we shall not attempt such an evaluation.

In TABLE 1, are listed the results of investigations on the relationship between concentration of the substrate, acetylcholine, and the velocity of cholinesterase action. It may be seen that the chohnesterases present in human and dog serum, and in the cat superior cervical gang- lion, show increasing rates of reaction, with increasing substrate con- centrations, which become asymptotic to a maximal rate at infinite substrate concentration. These values of the dissociation constants (molar concentration at which one-half maximal reaction velocity oc- curs) range, in general, from 1 to 1.7 X 10"^ In contrast, the cholin- esterases from the red cells of man, sheep, horse, and ox, and from the brain of the mouse and dog, do not show increasing reaction velocities with increasing substrate concentration. According to Mendel and Rudney," and to Alles and Hawes,'' they show optimal activity at about 1 X 10-^M; at concentrations higher than this, the reaction velocities decrease.

On the other hand, values for the optimal concentration obtained by Zeller and Bissegger,^^ and by Nachmansohn and Rothenberg," are

BOD AN SKY: CHOLINESTERASE

523

Table 1

Relation between Reaction Velocity of Cholinesterase Action and Acetyl- choline Concentration

In accord with Michaelis-M enten forvuda

Enzyme source

Investigator

Dissociation constant

Dog serum Dog serum Dog serum Human serum Human serum Cat superior

cervical ganglion

Goldstein^

Eadie^

Wright & Sabine*

Click'

Wright & Sabine*

Click i«

1.25 X 10-3 1.7 X 10-3 0.26 X 10-3

1.1 X 10-3

1.2 X 10-3

1 X 10-3

Inhibition at higher substrate concentrations (bell-shaped curve)

Enzyme source	Investigator	Optimal concentration (M)
Red cell: human Red cell: human Red cell: human sheep horse ox Red cell: human Brain: mouse dog Brain: human Brain: mouse Brain areas: ox	AUes & Hawes" Zeller & Bissegger'^ Mendel & Rudney'3 Nachmansohn & Rothenberg'^ Mendel & Rudney" Zeller & Bissegger^^ Nachmansohn & Rothenberg" Nachmansohn & Rothenberg'^	0.25 X 10-^ 4.4 X 10-3 <1.5 X 10-^ ca. 5 X 10-3 <1.5 X 10^ 4.1 X 10-3 ca. 6 X 10-3 ca. 8 X 10-3

much higher. As Mendel and Rudney^'^ have pointed out, and as will be seen later, the concentration of electrolyte influences the optimal concentration of substrate.

The relation existing between reaction velocity and substrate con- centration of acetylcholine may be of physiologic significance. Glick^° found that, under optimal substrate conditions, pH 7.4 and 38°, 0.10 y acetylcholine chloride was split per second, per milligram of cat supe- rior cervical ganglion. Brown and Feldberg^** found that the concen- tration of acetylcholine in the superior cervical ganglion of the cat was

524 ANNALS NEW YORK ACADEMY OF SCIENCES

22.5y/gm. of tissue. If conditions of maximal substrate concentration are assumed to exist, 0.225 seconds would be required to split the acetylcholine. If the assumption is made, that lower substrate concen- trations exist, then a longer period would be required. Brown and Feldberg also found that O.ly acetylcholine was liberated from a gang- lion weighing 12.9 mg. and perfused with eserinized Lockes' solution, during the first 5 minutes of preganglionic stimulation at 17 per second. According to Glick, if the substrate concentration were maximal, 78 milliseconds would be necessary to hydrolyze this amount, or 1.5 milli- seconds for splitting the acetylcholine liberated by one nerve impulse. These times will be longer, if lower substrate and enzyme concentra- tions are assumed. The time for hydrolysis, at the minimum rate, was 8 seconds, and localization of enzyme and substrate within the ganglion cell would have to be assumed, to explain enzymatic destruc- tion of acetylcholine liberated by nerve impulses within the span of the refractory period of 2 milliseconds.

EFFECT OF pH AND ELECTROLYTES ON CHOLINESTERASE ACTIVITY

The effect of pH on enzyme activity is, of course, well known. Bern- heim & Bernheim,'^ employing a pharmacological procedure, found a pH optimum of 8.4 for the serum and brain cholinesterases of some lower animals. Glick'-'' ^^ found practically this same value, 8.4 to 8.5 for the cholinesterases of human serum, pig's gastric mucosa, horse serum, and cat brain. Calculations from the shape of the pH activity curve, for these various cholinesterases, show that the activity at 7.4 is about 65 to 70 per cent that of the activity at optimal pH. A some- what lower optimal pH, 7.5-8.0, has been reported for red cell cholin- csterase."

The effect of various ions on the activity of cholinesterase has been studied by several groups of investigators. The activating effect of Ca"'"' and Mg"''" is well established and, except perhaps for the magni- tudes of the degree of activation, appears independent of the source of the enzyme. Thus, Nachmansohn^-' found that 4 X 10"^ M Ca"""" in- creased the activity of dialyzcd Torpedo electric organ cholinesterase 5-fold, and 4 X 10"^ Mg"'^ increased the activity about 8- to 9-fold. Massart and Du Fait-" found' that horse serum, which lost 40 per cent of its cholinesterase activity, on dialysis, regained its normal activity in the presence of 2 X 10-' M Mg++ or Ca++. Mn++ has been found to activate considerably both dialyzed Torpedo electric organ cholin-

BOD AN SKY: CHOLINESTERASE 525

esterase and dialyzed horse serum cholinesterase. Ba"^"^ activates Tor- pedo cholinesterase, but not dialyzed horse serum cholinesterase.

There has been considerable disagreement, regarding the activating effects of Na+ and K"". But here, as in other respects, these disagree- ments are resolved, if it is recognized that the studies have been carried out on cholinesterases from various sources. Thus, 1 X 10~^ MK"" inhibits Mendel and Rudney's purified horse serum cholinesterase 40 per cent at a substrate concentration of .0015 M acetylcholine, 15 per cent at .03 M.^^ Glick-^ found, at .02 M substrate concentration, that Na+ and K+ did not have any regular effect on horse serum cholin- esterase activity, but both increased rabbit serum cholinesterase activ- ity 25 per cent at 0.3 M Na+ or K+, 35 per cent at 0.5 M, and 40 per cent at 1.0 M. Nachmansohn^^ reported that both Na"" and K+ activ- ated Torpedo cholinesterase.

There appears to be a relationship between the effect of Na* and K"" and the effect of substrate concentration on reaction velocity. Thus, in human serum where the substrate-reaction velocity relationship fol- lows the Michaelis-Menten formulation, 0.16 M NaCl inhibited the cholinesterase about 10 to 20 per cent, at substrate concentrations ranging from 1 X lO"** to 4 X 10~*. On the other hand, among red cell cholinesterases, in which reaction velocity decreases at higher substrate concentrations (Haldane formulation), 0.16 M NaCl, inhibited at low substrate concentrations, activated at substrate concentrations greater than about 1 X 10"^'^ M. Alles' and Hawes' results" show, and Mendel and Rudney^'^ have emphasized this point particularly, that, for those cholinesterases which follow the Haldane formulation for the relation- ship between substrate' concentration and reaction velocity, the presence of Na"^ or K"" not only increases the value of the optimal reaction rate, but also shifts the optimum to higher substrate concentrations.

So far as anions are concerned, cyanide has been found to have no effect on horse or human serum cholinesterase. Oxalate, fluoride, and citrate inhibit dialyzed horse serum cholinesterase to the extent of 30 per cent at 0.002 M concentrations of these ions, 60 per cent at 0.02 M.^"

INHIBITION OF CHOLINESTERASE ACTIVITY

■ The inhibiting effects of various compounds on enzyme activity have been generally formulated in terms of an inactive, but reversible, enzyme inhibitor complex.^' ^ If the inhibition is non-competitive, then

v' K,

V K, + I

(3)

526 ANNALS NEW YORK ACADEMY OF SCIENCES

Where v' is the velocity in the presence of the inhibitor, v in its ab- sence, I is the concentration of the inhibitor, and Kj is the dissociation constant of the enzyme-inhibitor complex.

If the inhibition is competitive, then the following expression holds:

^ + A'/ 1 +

(-i)

where the terms have the meanings as described in equations 1 and 3. Kj may be calculated from the value of the dissociation constant, K's, in the presence of a constant concentration of inhibitor, as follows:

K: = ^r^— (4')

The Michaelis-Menten derivation is based on the assumption that the concentration of enzyme centers is constant and, as compared with the concentration of any substance with which it could combine, so small that it may be neglected. Recently, Straus and Goldstein," elaborating upon the ideas of Easson and Stedman,^^ have submitted a more general formulation for the effect of an inhibitor, which takes into account those possibilities in which the concentration of enzyme centers may not be negligible. In the presence of a large excess of substrate,

/ = ^. + iE . (5)

1 — I

total inhibitor free inhibitor bound inhibitor where I is the concentration of total inhibitor, combined and free ; i is the fraction of total enzyme combined with inhibitor; E is the concen- tration of total enzyme; and Kj is the dissociation constant of the enzyme-inhibitor complex. When the above equation is divided by Kj, the following expression is obtained:

/' - ^. + iE' (6)

where /' = I/Kj and E' = E/Kj.

Simplifications of these equations are possible, under conditions where E' is very small, or very large, so that the other term on the right hand side of the equation may be neglected.

The implication of equation 6 is that the degree of inhibition de- pends upon the value of £", namely, the ratio of the concentration of enzyme centers to the dissociation constant of the complex. This may be illustrated by taking values from a theoretical plot by Straus and

BOD AN SKY: CHOLINESTERASE 527

Goldstein. Thus, the log values of /', at i = 0.5 (50 per cent inhibi- tion) are as follows, for various values of E':

E' log r

0.1 0.00

10 0.79

100 1.71

1000 2.70

It may, thus, be seen that dilution of the enzyme influences the £" value, and hence, the extent of inhibition.

The values of Kj and E may be calculated from the experimental determination of the inhibition, i, at various concentrations, /, of in- hibitor. Equation 5 may be transposed, as follows:

l = KjX j^. + E (7)

A plot of - against :j -. should, therefore, yield a straight line, the

slope of which would be Kj and the intercept on the F-axis would be E. Goldstein'' has also developed an expression for competitive equilib- rium between enzyme, substrate, and inhibitor:

/'

total free combined

Where /' = I/Kr, S' = S/Ks, E'l = E/Ki,

Es = E/Ks, and a = 1 — z or fractional activity of the enzyme.

Another type of derivation is possible, if E{, the amount of free enzyme, is considered negligible in comparison with the amount of enzyme combined with inhibitor and substrate. Then:

/' =iS'-aEs')l- -]^{l-a)Er (9)

total free combined.

Various simplifications of equations 8 or 9 are possible, depending upon whether we assume Ej' or Eg- to be small enough to be neglected, or so large that other terms not involving them become negligible.

The investigations on in vitro inhibition may now be summarized. In TABLE 2, are shown those results which have been formulated, in terms of the dissociation constants of an inhibitor-enzyme complex, in accordance with the equations already discussed. Several points of interest may be noted. First, the dissociation constants of the enzyme complexes of physostigmine and prostigmine are of a low order of mag- nitude, about 10"^ to 10"^, as compared with the dissociation constants of the cholinesterase-morphine derivative complexes, 10"^ to 10~*. The

528

ANNALS NEW YORK ACADEMY OF SCIENCES

values of Eadie^ seem extraordinarily low, but these dissociation con- stants have been calculated on the assumption that one molecule of enzyme combines with two of inhibitor. Goldstein*' has discussed this point and has claimed that Eadie's method of determining reaction velocity involved a constant error. The titration of released acetic acid, for twenty minutes, immediately following the addition of enzyme

Table 2

Dissociation Constants of Cholinesterase-Inhibitoh Complexes

Dissociation constants calculated by method of Michaelis-Menten, except for Straus & Goldstein values. Values for Ki at 37-38° .

Inhibitor	Source of cholinesterase	Investigator	Dissociation constant, Ki
Physostigmine Physostigmine Physostigmine Prostigmine Morphine Morphine Dilaudid Codeine Desomorphine	Horse serum, 22% Dog serum, 4.54% Dog serum Dog serum Human serum Dog serum Human serum Human serum Human serum	Straus & Goldstein22 Goldstein^ Eadie' Eadie' Wright & Sabine* Eadie' Wright & Sabine* Wright & Sabine* Wright & Sabine*	3.7 X 10-* 3.11 X 10-* 3 X 10-'* (n = 2) 2 xlO-iMn = 2) 8.1 X 10-* 14.6 X 10-* 1.2 X 10-3 4.2 X 10-* 1.8 X 10-*

to a substrate-inhibitor mixture, corresponded to a stage of the reaction where equilibrium had not yet been established, and where the reaction velocities were higher than at equilibrium. This discrepancy was greatest for concentrations of inhibitor producing moderate inhibitions.

A second point of interest is the difference in the dissociation con- stants of the complexes of the same inhibitor with cholinesterases of different sources. Although the difference is not marked, it may be seen that the constants for the morphine complex are 8.1 X 10~* for human serum cholinesterase, and 14.6 X 10"* for dog serum cholin- esterase. The occurrence, among different cholinesterases, of differing sensitivities to inhibition by the same compound, will be discussed further.

In TABLE 3, are shown the concentrations of cholinesterase inhibitors giving 50 per cent of the uninhibited cholinesterase activity. It will be readily recognized that these values are equivalent to the values of the dissociation constants, calculated by the Michaelis-Menten ex- pression for non-competitive equilibrium. This table illustrates the order of inhibition of different compounds. It may be seen that human serum, rabbit serum, and human muscle cholinesterases are less sensi- tive to inhibition by physostigmine than is the horse serum cholin-

BODANSKY: CHOLIN ESTERASE

529

Table 3

Concentrations op Cholinesterase Inhibitors Giving 50 Per Cent of

Uninhibited Velocity

Inhibitor	Source of Cholinesterase	Investigator	Concentration of Inhibitor
Physostigmine	Horse serum, 4.5%	Collier & AUen^^	2.5	M X 10-«
Physostigmine	Horse serum, 22.2%	Ellis, Plachte, & Straus^^	5	X 10-«
Physostigmine	Horse serum, 11.1%	Ellis, Plachte, & Straus^^	2.5	X 10-«
Physostigmine	Human serum, 2.5%	Mazur & Bodanskv^^	2	X 10-^
Physostigmine	Rabbit serum, 12.5%	Mazur & Bodansky-*	1.2	X 10-^
Physostigmine	Human muscle	Mazur & Bodansky^"	8	X 10-^
Methylene Bhie	Horse serum	Collier & Allen^^	1.2	X 10-«
Methylene Blue	Horse serum	Massart & DuFait"	6	X 10-«
Acriflayine	Horse serum	Collier & Allen^^	6.6	X 10-^
Phenothiazone	Horse serum	Collier & Allen^^	6.7	X 10-^
Thionin	Horse serum	Massart & DuFait"	11	X 10-'
Thiamin	Horse, rat serum	Glick^s	1.8	X 10-3
Isopropyl antipyrin	Guinea pig serum	Zeller^s	2	X 10-3
Isopropyl antipyrin	Horse serum	Zeller"	0.92	! X 10-3

esterase. The basic dyes, methylene blue, acriflavine, phenothiazone, and thionin, are somewhat less potent inhibitors than physostigmine. Next, in order of potency of inhibition, are thiamin and the antipyrines. Table 4 shows that several antipyrines at a concentration of 3 X 10-^ M inhibit human serum cholinesterase from 30 to 80 per cent. Sulfon- amides at this concentration inhibit only slightly. Acid dyes {e.g., Congo red) and p-phenylenediamine have been reported to have no inhibiting action at 2 X 10"* M.-'

Table 4

Inhibition of Human Serum Cholinesterase Activity by Antipyrines and

Sulfonamides (Zeller)

Inhibitor	Concentration M.	Inhibition per cent
Antipyrine Aminoantipyrine Dimethylaminoantipyrine Isopropylantipyrine Sulfanilamide Irgamid Sulfathiazole Benzyl sulfanilamide	3 X 10-3 3 X 10-3 3 X 10-3 3 X 10-3 2.5 X 10-3 2.5 X 10-3 2.5 X 10-3 2.5 X 10-3	51 30 38 80 -1 13 18 9

530

ANNALS NEW YORK ACADEMY OF SCIENCES

It is, of course, always of interest to attempt a correlation between chemical structure of inhibitors and enzyme action. In general, com- pounds which resemble the substrate, chemically, inhibit enzyme action, presumably by combining with the same chemical groupings on the enzyme molecule. However, it must be stressed that other chemical groups on the inhibitor molecule may influence this combination. The evaluation of this influence requires considerable experimentation. In the case of cholinesterase, relatively few data are available.

Acetylcholine is a quaternary ammonium compound. A survey of the inhibitors which we have discussed shows that, although no strict correlation can be drawn, between structure and degree of inhibition, the closer the inhibitor comes to possessing a completely alkylated nitrogen grouping, the greater is its inhibition. Thus, prostigmine, one of the most effective inhibitors, has a quaternary ammonium grouping. Physostigmine has two tertiary amine groupings. Methylene blue, which is also a potent inhibitor, may be considered to have a cjuater- nary ammonium grouping. Thionin, which is the unalkylated con- gener of methylene blue, is a less powerful inhibitor; about 20-100 times as great a concentration is needed to produce 50 per cent inhibition.

That an alkylated amino grouping is important in inhibition, seems to be generally true. Thus, the pyrazolone derivatives (antipyrines), morphine, caffeine, percaine, are all moderate inhibitors. Unalkylated amino groupings tend to make a compound a poor inhibitor: for exam- ple, diphenylamine does not inhibit at 2 X 10"* M. Acid groupings (COOH, SO3, H, OH), apparently tend to negate the inhibiting power of alkylated amino groups. Examples of this occur in the sulfon- amides and the acid dyes.

CH3

CH3 O

\ ^ N-C-0-

/

CH,

/

-N\

CH3NH-C-O

SO4

CH3 CH3

CH3

CH,

prostigmine S

CH3 CH3 physostigmine

N'

w

N methylene blue

H2N

BOD AN SKY: CHOLINESTERASE

531

CH= 0=C

=C.CH3 N.CH3 N— CeHs

O

H

/

C2H5

C-N-CH2-CH2-N -0C4HC,

vv

N

antipyrine CH3-N C=0

percame

CHs

CH

H2N

-<^ \-SO2-NH2

caffeine

sulfanilamide

It must be noted, however, that the relationship between the degree of inhibition and the chemical structure of the inhibitors also depends upon the source of the cholinesterase. Table 5 shows that percaine,

Table 5

Effect of Drugs on Inhibition of Various Cholinesterases (Zeller and Bissegger")

	Concentration M	Per cent inhibition
Inhibitor	Human serum	Human red cells	Human brain
Percaine Irgamid Isopropyl antipyrin Morphine Caffeine	6 X 10-3 6 X 10-3 6 X 10-3 6 X 10-3 6 X 10-3	94 46 81 66 4	25 4 26 76 42	IS 3 66 40

irgamid, and isopropyl antipyrine inhibit human serum cholinesterase much more markedly than human red cell or human brain cholin- esterase. In contrast, morphine inhibits these three cholinesterases to about the same extent, and caffeine inhibits human serum cholinesterase only slightly, as compared with its effect on human red cell and human brain cholinesterase. Nachmansohn and Schneeman^" have observed that caffeine inhibits nervous tissue cholinesterases much more mark- edly than those of horse serum and guinea pig pancreas. In contrast, the inhibitions of these latter cholinesterases by quinine, quinidine, cocaine, and lobeline are more marked than those of nervous tissue.

In addition to the inhibitors which we have discussed above, a num- ber of other substances have been reported to possess an inhibiting ac-

532 ANNALS NEW YORK ACADEMY OF SCIENCES

tion: vitamin K, ether, chloral, and some hormone preparations, such as estrone, progestin, testosterone, etc.^^- ^^ Observations on these in- hibitions are not yet detailed, and their exact significance is not yet defined. It should be recognized that, in general, any enzyme is sub- ject to in vitro inhibition by a great number of compounds, particularly at relatively high concentration. Such inhibitions probably do re- flect an interaction of the inhibitor with the enzyme molecule, but the in vivo significance of such inhibitions is very often cjuestionable.

There is evidence to indicate that the inhibitors which we have al- ready described form reversible enzyme-inhibitor complexes. Matt- hes," for example, first showed that dialysis of a mixture of cholin- esterase and physostigmine resulted in the restoration of the enzyme activity. Similarly, dilution of cholinesterase physostigmine mixtures results in a relative increase of cholinesterase activity, presumably as the result of the dissociation of the inactive enzyme-inhibitor com- plgx.22, 26 Zeller has shown that the inhibition of cholinesterase by pyrazolons and sulfonamides is similarly reversible, by dialysis of the corresponding enzyme-inhibitor complexes. ^^

IN VITRO AND IN VIVO INHIBITION BY DIISOPROPYL- FLUOROPHOSPHATE (DFP)

We should now like to present a description of the properties of a compound, typical of an entire group, which, in contrast to the inhibi- tors we have described above, forms a combination with cholinesterase which it has, so far, not been found possible to reverse. This com- pound is diisopropyl-fluorophosphate. It is one of a group of alkyl fluorophosphates first described by Lange and Krueger.^"* During the war, it was regarded as a potential chemical warfare agent, and its properties were first investigated by British workers. Adrian, Mc- Combie, B. A. Kilby, and M. Kilby^^- ^"^ noted the similarity between the cholinergic effects of the fluorophosphates and those of physostig- mine. Mackworth" found that incubation of the alkyl fluorophos- phates with horse serum cholinesterase resulted in the inactivation of the enzyme, and that dialysis of the fluorophosphate-cholinesterase mixture did not result in any restoration of cholinesterase activity. Our interest in the mechanism of the anticholinesterase action was first aroused when we noted that, upon exposure of men to very low concen- trations of this agent, the serum cholinesterase was very markedly re- duced to 2 to 5 per cent of the pre-exposure value, in spite of the fact that there were only slight or doubtful systemic symptoms. The

BOD AN SKY: CHOLINESTERASE

533

in vitro and in vivo inactivation of cholinesterase by DFP has been studied by Mazur and Bodansky,^® and the results of these studies will now be briefly described.

In Vitro Inhibition of Cholinesterase Activity by Diisopropyl-Fluoro- phosphate and by Physostigmine. The inhibition of the activities of serum, red cell, muscle, and brain cholinesterases of the rabbit, monkey, and man were determined, at various concentrations of DFP and physostigmine. In order to obtain a general measure of the extent of inhibition of the different cholinesterases, the relative velocities were plotted against the negative logarithm of the molar concentration of DFP or physostigmine. The negative log molar concentration at which 50 per cent inhibition occurred, was termed the pCi value. These values for various enzyme preparations are shown in table 6.

Table 6 Sensitivity of Various Cholinesterases to Inhibition by Diisopropyl-Fluoro-

PHOSPHATE and PhYSOSTIGMINE

The values are expressed in terms of the negative log of the concentration of in- hibitor required to produce a oO per cent inhibition of cholinesterase activity (pCi).

Serum

Red cells

Muscle

Brain

Diisopropyl-fl uorophosphate

Rabbit	4.1	5.2		5.5
Monkey {M. rhesus)	7.8	5.5		5.5
Human	7.7	5.4	5.4	6.0
Horse	8.3
Horse* (purified)	8.1

Physostigmine

* Horse serum cholinesterase (purified), prepared by Drs. Northrop and Kunitz, according to directions of Mendel and Rudney.

It may be seen that, of the various serum cholinesterases studied, that of the rabbit was least sensitive to inhibition by DFP. Thus, a negative log molar concentration of 4.1 of DFP was necessary for 50 per cent inhibition of rabbit serum cholinesterase, whereas concentra- tions of about one ten thousandth as much (negative log molar values of 7.7 to 8.3) gave 50 per cent inhibition of monkey, human, and horse serum cholinesterase activity. The various red cell cholinesterases

534 ANNALS NEW YORK ACADEMY OF SCIENCES

showed approximately the same degree of sensitivity toward inhibition by DFP (pCi values of 5.2 to 5.5), whereas human brain cholinesterase (pCi = 6.0) was somewhat more sensitive than monkey or rabbit brain chohnesterase (pCi = 5.5) to inhibition by DFP. Purified cholin- esterase of the electric eel gave a pCi value of 4.1.

It is of interest to compare the sensitivities of the various tissue cholinesterases from one species to inhibition by DFP. Rabbit brain and red cell cholinesterases showed a greater sensitivity than did serum cholinesterase; brain showed the greatest sensitivity. Thus, a 50 per cent inhibition of serum cholinesterase activity occurred at a negative log of the molar concentration of DFP of 4.1, whereas the same degree of red cell and brain cholinesterase inhibition occurred at values of 5.2 and 5.5, respectively. Monkey serum cholinesterase was much more sensitive to inhibition by DFP than red cell or brain cholinesterase. Human serum cholinesterase was much more sensitive to inhibition by DFP than human red cell or brain cholinesterase. This picture is similar to that found in the monkey, and is in marked contrast to that observed in the rabbit.

Table 6 also shows the sensitivities of rabbit serum, human serum, and muscle cholinesterases to inhibition by physostigmine. It may be seen that rabbit serum cholinesterase was more sensitive to inhibition by physostigmine (pCi = 5.9) than by DFP (pCi = 4.1), whereas the reverse was true with human serum cholinesterase.

The possibility existed that the differences in sensitivity of different cholinesterases to inhibition by DFP were due to materials, other than the enzymes themselves, present in the preparations. Table 6 shows that a purified horse serum cholinesterase preparation had, within experimental error, the same pCi value as horse serum itself. Heat- inactivated extracts of one tissue, added to a tissue possessing cholin- esterase activity, did not alter the sensitivity of the latter to inhibition by DFP. Thus, human brain extract was heated to destroy its cholin- esterase activity, and added to human serum. The pCi value for the mixture was 7.7, the same as that found for human serum cholinesterase itself.

In Vivo Inhibition of Cholinesterase Activity by DFP. The extent to which various cholinesterases are inhibited in vivo, after administra- tion of DFP, may be considered to depend, not only on the sensitivity of the particular tissue cholinesterase to inhibition by DFP, but also on the localization and, hence, of the concentration of DFP in the tissue. In rabbits exposed to DFP vapor, severe muscular tremors and death

BODANSKY: CHOLINESTERASE

535

occurred at the higher exposures; at lower exposures, no symptoms ex- cept miosis occurred. In most instances, the decrease in red blood cell cholinesterase activity was greater than that in serum eholinesterase activity. It will be recalled that in vitro rabbit red cell cholinesterase is more sensitive than serum cholinesterase to inhibition by DFP. When rhesus monkeys were exposed to DFP, the plasma cholinesterase activity was decreased to only 1 to 5 per cent of normal, at almost all exposures, whereas the red cell cholinesterase activity showed only slight decreases at the lower exposures. This marked difference in the extent of decrease paralleled the considerable in vitro difference be- tween the sensitivities of monkey red cell and serum cholinesterases to inhibition by DFP.

The effect of intravenously injected DFP on cholinesterase activity, in the rabbit, is shown in table 7. It can be seen that, in most in-

Table 7

The Effect of Intravenously Injected Diisopropyl-Fluorophosphate on Rabbit Plasma, Red Cells, and Brain Cholinesterase Activity In Vivo

		Blood sample time	Relative cholinesterase
Animal	Dose mg.		activity*		T?pmnrkK
number
	per kg.	min.	Plasma per cent	Red cells per cent	Brain per cent
327	3.0	4	0	0	0	Died immediately
326	0.3	23	15	7	12	Muscle tremors
328	0.3	26	5	0	5	Muscle tremors
491	0.1	26	54	41	59	No symptoms
492	0.1	25	37	29	57	No symptoms
330	0.05	27	60	19	74	No symptoms
331	0.05	43	51	29	73	No symptoms

* These values are per cent of the pre-exposure values.

stances, the red cell cholinesterase activity is more markedly reduced than that of the serum cholinesterase. This difference corresponds to the in vitro difference in sensitivities. The brain cholinesterase activ- ities were reduced to about the same extent as the serum cholinesterase activities at the lower doses, and slightly less at the higher doses. This finding did not correspond to the in vitro sensitivity of brain cholin- esterase to inhibition by DFP. The brain cholinesterase activities, corresponding to the appearance of symptoms, were less than about 60 per cent of normal.

The effect of intravenous injections of DFP in the monkey are shown in TABLE 8. A dose of 0.3 mg. per kg. was fatal in 10 minutes. The serum, red blood cell, and brain cholinesterases were reduced to very

536

ANNALS NEW YORK ACADEMY OF SCIENCES

low or zero levels of activity. Essentially the same results were ob- tained at doses of 0.2 and 0.25 mg. per kg., except that the survival period was longer. At a dose of 0.1 mg. per kg., the animal survived, although the serum and red cell cholinesterase activities were very low. At 0.02 mg. per kg., the serum cholinesterase activity was reduced to a very low level, 2 per cent of normal, whereas the brain cholinesterase activity was decreased only slightly, to 78 per cent of normal. This

Table 8

The Effect of Intravenously Injected Diisoproptl-Fluorophosphate on Mon- key Plasma, Red Cells, and Brain Cholinesterase Activity In Vivo

Animal number	Dose mg. per. kg.	Blood sample time min.	Relative cholinesterase activity Per cent of pre-injedion value	Remarks
	Plasma	Red cells	Brain
4 6 3 1 2 5	0.3 0.25 0.2 0.1 0.1 0.02	10 27 2 hrs. 24 3H hrs. 60	0 0 0 0 0 2	2 1 0 2 10 14	0 0 0 0 78	Died in 10 minutes Died in 33 minutes Severe symptoms, sacrificed No symptoms Muscle tremors, diarrhea No symptoms

marked difference between the decreases of serum and brain cholin- esterase was similar to that obtained at low exposures of monkeys to DFP vapor, and parallels strikingly the in vitro difference in sensitivity between the monkey serum and brain cholinesterases to inhibition by DFP. As will be seen presently, these findings are quite similar to those obtained in man. Although the red blood cell cholinesterase was of about the same order of in vitro sensitivity as the brain cholin- esterase, it was reduced in vivo, at the very low dose, to a greater degree.

In the monkey, symptoms were absent, even when the serum cholin- esterase levels were zero. In most instances, the occurrence of severe symptoms or death was associated with zero levels of brain cholin- esterase activity. After exposure to, or injection of, DFP, the serum cholinesterase returned very slowly to normal. In the few instances in which this return was followed, about 50 per cent of the normal serum cholinesterase activity was regained in about 7 days.

Men were also exposed to DFP vapor. One group of 7 men was ex- posed to a concentration of 19 micrograms per liter, for 8% minutes;

BODANSKY: CHOLINESTERASE

537

a second group of 6, to a concentration of 27.1 micrograms per liter, for 9 minutes; and 2 men, to 28.8 micrograms per liter, for 10 minutes, 40 seconds, and 27 micrograms per liter for 6 minutes, 20 seconds, re- spectively. The symptoms were extremely mild. All of the men showed miosis and most of them complained of a slight feeling of tight- ness in the chest, lasting for several hours. The following symp- toms were observed occasionally: increased nasal secretion, nausea, salivation.

Table 9

Effect of Inh.alation of Diisopropyl-Fluorophosphate Vapor on Serum Cholixesterase Activity In Vivo in Max

■ -B Subject	Concen- tration of DFP vapor	Duration of exposure	Relative cholinesterase activity at various intervals after exposure*
5-30 min.	Days
	1	2	3	4	6	8	10	15	17
	y per 1.	min.	4 3 2 3 1	13 7 14 12 7			29 28 3 31 30		58 53 55 50 52		71 77 72 69 68
J.H. R.L. W.B. M.G. J.P.M.	19 19 19 19 19	8.7 8.7 8.7 8.7 8.7

* These values are per cent of the pre-exposure values

Table 9 shows the decreases in serum cholinesterase activity, imme- diately after exposure, and the rate of return of the activity to normal. It may be seen that, immediately after gassing, the serum cholinesterase activity decreased to about 1 to 5 per cent of the pre-exposure value. The rate of return to normal was very slow. On the average, the activ- ity returned to about 30 per cent, in four days; to about 50 per cent, in 8 days; and to about 70 per cent, in 15 days. The red cell cholin- esterase activities of several of these men were determined immediately after exposure and were found to be only slightly decreased below pre- exposure values. These results show, therefore, a correlation with the in vitro sensitivities of human serum and red cell activities.

Rate of Restoration of Rabbit Plasma, Red Cell, and Brain Cholin- esterase Activities after Poisoning with DFP. The slow rate of re- generation of serum cholinesterase activities, demonstrated above in man and monkey, raised the question as to the rate of regeneration of brain cholinesterase activity. The average brain cholinesterase activ- ity Vas first determined in a series of normal rabbits. Each of a group

538 ANNALS NEW YORK ACADEMY OF SCIENCES

of about 50 rabbits was injected with 0.3 mg. DFP per kg., and blood samples were taken, before injection, for determination of normal plasma and red cell cholinesterase. At this dose, the rabbits developed tremors within about 15 minutes after injection and continued to have these tremors throughout the day. About 10 per cent of the animals died. The survivijig animals were free of symptoms the day following injection. At suitable intervals after injection, blood was taken for de- termination of plasma and red cell cholinesterase activities. At these or other times after injection, 1 to 6 rabbits were selected for sacrifice. Brains were removed, within 5 minutes after death, and the brain cholinesterase activities determined. The plasma cholinesterase ac- tivity returned to normal values in about 5 days. The red cell cholin- esterase activity returned to normal somewhat more slowly. It was about 50 per cent of normal in about 5 days and attained the normal, pre-injection level in 10 days. The rate of recovery of brain cholin- esterase activity was exceedingly slow. Ten days after injection, the brain cholinesterase activity was about 50 to 60 per cent of normal. Twenty to thirty days after injection, it was about 60 to 70 per cent of normal. Fifty days after injection, the brain cholinesterase activity had returned to 90 per cent of normal.

Attempts at Reversal of DFP Inhibition of Cholinesterase. It has been shown that the inhibitions of phosphatase by amino acids,^^ of pepsin by proteolytic digestion products,^'' and of cholinesterase by physostigmine^^' ^^ may be reversed by subjecting the enzyme-inhibitor mixture to dialysis or dilution. In the present study, the serum and brain extracts of rabbits injected with 0.3 mg./kg. DFP were dialyzed against several changes of saline, for about 24 hours. Rabbit plasma was also treated in vitro with DFP and then dialyzed, for 24 hours. In neither type of experiment was there any increase in activity of the inactivated cholinesterase. The in vitro results are in agreement with those of Mackworth.^^ Dilution of mixtures of cholinesterase and fluorophosphate failed to show any relative increase in enzyme activ- ity. This was in contrast to the results obtained on dilution of physo- stigmine-cholinesterase mixtures.

In view of the difference among the cholinesterases of different tis- sues to inhibition by DFP, it would be unjustified to draw any conclu- sions from our data concerning the sensitivity to inhibition of cholin- esterases, at autonomic effector organs, ganglia, or myoneural junctions. According to the concept of chemical transmission of nervous impulses, the extent of cholinesterase inhibition, at these sites, should be corre-

BODANSKY: CHOLINESTERASE 539

lated with the appearance of various chohnergic symptoms. In gen- eral, in the monkey or rabbit, such cholinergic symptoms as muscular tremors, salivation, and diarrhea were associated with low red cell and brain cholinesterase activity, and death was associated with zero brain cholinesterase activity. However, this association is to be regarded as fortuitous. Conversely, it should be emphasized that depression of serum cholinesterase activity does not necessarily indicate the appear- ance of cholinergic symptoms. In monkey and man, for example, the sei'um cholinesterase activity may be reduced to extremely low levels, without the manifestations of such symptoms.

The persistence of low serum, red cell, and brain cholinesterase ac- tivity in the rabbit, for periods of 5, 10, and 60 days, respectively, and of low serum cholinesterase activities in the monkey and man, for periods of at least one to two weeks, offers evidence in support of the irreversibility of inactivation in vivo. Hall and Ettinger'*° found that, after injections of physostigmine in the dog, the serum cholinesterase activity dropped to 10 to 25 per cent of normal in about a half hour and returned to normal in two hours. This prompt restoration of nor- mal activity may well be expected in the case of a readily reversible inhibitor-enzyme complex. On the other hand, the long periods of time necessary for the restoration of normal cholinesterase activity, after exposure to DFP vapor or injections with DFP solutions, in the instances mentioned, are of the same order of magnitude as those nec- essary for the regeneration of protein/^ and would seem to indicate a synthesis of enzyme protein.

DFP has already been proved to be of considerable value as a tool in investigative work. Its anticholinesterase action has made it a candi- date for clinical trials in Myasthenia gravis and glaucoma.*^"^'' The marked degree to which it may inhibit cholinesterase activity in vivo, and the character of this inhibition, have, as we have seen, permitted studies of the rate of regeneration of various tissue cholinesterases. They open the way to further studies on the way in which diet, drugs, or the existence of various pathological lesions influence the regenera- tion of cholinesterase. We have also seen that DFP has permitted more incisive studies into the role of cholinesterase in nerve transmis- sion."' '^^

SPECIFICITY

Specificity of Cholinesterase Action. The question of the specificity of the cholinesterase activities of various tissue extracts has claimed considerable attention. Although certain general distinctions between

540 ANNALS NEW YORK ACADEMY OF SCIENCES

the pattern of certain of the cholinesterases may be made, it will be shown that there are a number of exceptions to any strict classification.

Easson and Stedman^' proposed the following criteria for the specific- ity of the cholinesterase action of different sera: (a) relative action of the serum on choline and non-choline esters; (b) inhibition of the ester- splitting action by prostigmine; (c) hydrolysis of mixed substrates. Thus, human serum was considered to contain a specific cholinesterase, because its action on non-choline esters was about l/80th of that to- wards butyrylcholine, and because both actions were inhibited to the same extent by the same concentration (10"'' M) of prostigmine. On the other hand, guinea pig serum was considered to contain a specific cholinesterase and a non-choline, ester-splitting esterase; because the rate of actions on butyrylcholine and methylbutyrate were of about the same magnitude; because 10~^ M prostigmine inhibited, markedly, only the action on butyrylcholine; and because the actions on a mixture of butyrylcholine and methylbutyrate were equal to the sum of the action on each. We have found that judgment as to the relative action of a serum or tissue extract on acetylcholine and non-choline esters may de- pend considerably on the particular esters employed. For example, if the actions of human and rabbit brain extracts on acetylcholine are compared with that on monoacetin, it is found that there is relatively little hydrolysis of the latter. It might be concluded that these extracts contain, chiefly, specific cholinesterase. On the other hand, if the ac- tion is compared with that on triacetin, it is found that there is consid- erable hydrolysis of this ester, and it might just as readily be concluded that the content of non-specific esterase is very high.

Mendel and Rudney^^ stated that there were two different cholin- esterases in the body : one of which acted exclusively on choline esters ; and the other, a non-specific enzyme, which split both choline and non- choline esters. These they termed "true" and "pseudo"-cholinesterases, respectively. In addition to some of the criteria for specificity em- ployed by Easson and Stedman,*^ Mendel and Rudney^^ pointed out that the non-specific or "pseudo"-cholinesterases exhibited maximal ac- tivity at high concentrations of the substrate, acetylcholine, whereas the "true" cholinesterase showed optimal activity at low substrate con- centrations. According to these authors, inhibition of both choline and non-choline ester hydrolysis constitutes a criterion for distinguish- ing the cholinesterase as "pseudo." We have found that the anti- cholinesterase compound, DFP, markedly inhibits the mouse brain hydrolyses of both acetylcholine and triacetin. According to the cri- teria of Mendel and Rudney, this finding should classify mouse brain

BOD AN SKY: CHOLIN ESTERASE 541

cholinesterase as a "pseudo"-cholinesterase. Yet, these authors had classified it as a "true" cholinesterase, since its hydrolysis of acetyl- choline was inhibited by physostigmine, whereas the hydrolysis of non- choline esters was unaffected by this compound.

According to Mendel's and Rudney's criteria, the red cells of sev- eral species (human, horse, sheep, ox) contained two enzymes: a spe- cific or "true" cholinesterase and a non-choline ester-splitting enzyme. Mendel and Rudney classified red cell cholinesterase as "true" cholin- esterase, because a purified preparation hydrolyzed acetylcholine, but failed to hydrolyze non-choline esters, as exemplified by methylbuty- rate or tributyrin. We have found that such a purified preparation splits triacetin, a non-choline ester, which Mendel and Rudney did not test. ■

Nachmansohn and Rothenberg" have inclined towards the view that specificity is relative, and that tissue extracts containing the specific cholinesterase split acetylcholine at a higher rate than other esters. According to these investigators, the esterase in all nerve tissue is either exclusively or predominantly cholinesterase. Results which we have obtained confirm those of Nachmansohn and Rothenberg, except for the ester (triacetin) , which was not tested by these investigators. This ester was hydrolyzed more rapidly than acetylcholine.

The results on several aspects of the action and inhibition of various cholinesterases have been summarized in table 10. Although the data are not complete, it may be seen that, except in two respects, the cholin- esterases may be divided into two general groups. In the first group, the enzymes follow the Michaelis-Menten formulation for the relation- ship between reaction velocity and substrate concentrations. There is no inhibition by caffeine ; a marked inhibition by percaine ; no activa- tion by Na"" and K+; and failure to hydrolyze acetyl-B-methyl cho- line. In the second group, inhibition occurs at higher substrate con- centrations; there are also: inhibition by caffeine; slight inhibition by percaine; activation by Na"" and K""; and the ability to hydrolyze acetyl- B-methyl chohne. However, as already pointed out, DFP inhibits the non-choline ester hydrolysis of the enzymes, in both of these groups. Although DFP seems to inhibit the cholinesterases of the first group more markedly, there is considerable variation in sensitivity to inhibi- tion. Moreover, there is no sharp distinction between the enzymes of these two groups, with respect to the ratio of velocities at which they hydrolyze triacetin and acetylcholine. This latter finding may be ex- plained by assuming that there are varying amounts of non-acetyL: choline hydro lyzing esterases in these preparations. Y\\ri\\QYJff^^% f q

542 ANNALS NEW YORK ACADEMY OF SCIENCES

indicated to explore the actions of various cholinesterases, with respect to the criteria indicated in Table 10.

IN VIVO CHANGES OF CHOLINESTERASE

Alterations of Cholinesterase Activity in Disease. Considerable clin- ical and pharmacological investigation into the in vivo changes of cholinesterase activity has been conducted, with a view to determining the physiological significance of cholinesterase. The pathological con- dition which has attracted most attention, in this respect, is Myasthenia gravis. In this condition, which is characterized by muscle weakness and inclination to fatigue, it has been postulated that there is a defi- ciency of acetylcholine at the neuromuscular junction. Such a postu- lated deficiency may, of course, be brought about by a failure to syn- thesize acetylcholine, or by an excessive amount, or excessive activity, of cholinesterase at the neuromuscular junctions. That the latter mechanism is operative has been assumed, because .of the finding that prostigmine, an inhibitor of cholinesterase activity in vitro and of serum cholinesterase activity in vivo, results in clinical improvement.*® There is, however, no conclusive evidence of increased cholinesterase concentration or activity at the myoneural junctions. A number of investigators**^"^^ have failed to find increased cholinesterase activity in the serum in Myasthenia gravis, and, although the cholinesterase activity of muscle may perhaps not be regarded as too specific, there has been a similar inability to find increase of cholinesterase activity in muscle.^'' Other explanations of the physiological fault in Myas- thenia gravis and of the action of various drugs have been submitted by Gammon, Harvey, and Masland.^'

There are, however, several conditions in which definite changes in serum cholinesterase activity have been reported. There is fairly general agreement that debilitating diseases, such as tuberculosis, can- cer, and liver disease, are characterized by low serum cholinesterase activities. °®"'^* For example, Faber'^* found ranges of 65 to 150 units in normal men and 57 to 184 in normal women. In acute hepatitis, values ranging from 41 to 51 units were obtained; in liver cirrhosis, activities ranging from 77 to 92; in cancer, from 33 to 99; and in uremia, from 32 to 56 units. High serum cholinesterase activities have been re- ported in hyperthyroidism.*^'*' ®^

The author does not know, however, of any evidence to indicate that the low serum cholinesterase values found in debilitating diseases are of any special significance, so far as transmission of nervous impulses is concerned. Indeed, these low cholinesterase activities appear to be

BOD AN SKY: CHOLINESTERASE

543

Ed

m

<

Eh

g

o

w o

M

o 2 J 1>

o

;?; o

<! o

CO

H

CM

03
OJt*-
	O t^ -f	o	o CO
<M OO	lO	(M CO
	o do	d	oi d
> ^ ■
_^
--''>> a^ >>
>>j3 d-?
Acet -met 3holi ydrc	+ OO	+	+ +
« -^
c
W.2
-1-^
i~>	+ 0	+	+
^•1
o
o3
	tc a;
	•s.s
	>i-;5 a3
	-— c O t<
O		+		+ +
"-5	^i-
!-2 '-S	ffi§
.S
Q	. .^ ^	^ cot^ t^	^	o
If	■*'odi> N^	id	d
	lO
4J C
c 2	+
9X!	+	+	+1
S'^	+
(^ c
*^H
-, r^
(U .J
C CI
'S!-^		+	+
	o	+	+
o^
• .-H
i >>c
s3-*^ O
cfi O ^
-Q "S "S
3 > ^
rJ2
		fl
	>> -	03
	si 6^	a 3	03 (C S 3 3x2 -a o £
	s		s
		-«
	^	«	s~
		^	05	oq

544 ANNALS NEW YORK ACADEMY OF SCIENCES

merely a reflection of the state of the serum proteins. Faber^'' has found that there is a direct proportionality between serum cholin- esterase activities and the concentration of serum albumin, but not between these activities and the concentration of total serum protein or that of the serum globulin. This, of course, does not imply that cholinesterase is an albumin. Faber"^ has noted that, in proteinurias, the serum cholinesterase activities remain high, relative to the concen- tration of the serum albumin. There is an indication, in these observa- tions, that the formation of serum cholinesterase parallels that of serum albumin.

We have already discussed the effects of DFP on the in vivo activity of serum, red cell, and brain cholinesterases. There are, in the litera- ture, similar studies on other drugs. Perhaps one of the most detailed is that of Schutz,^^' *'^' '° concerning the effect of barbiturates. This investigator noted that the prolonged administration of these drugs, in man and animals, resulted in marked decreases of serum cholinesterase activity, although these drugs in concentrations of about 0.01 M do not inhibit the in vitro serum cholinesterase activity. He also ob- served that such prolonged administration in animals resulted in in vivo decreases of brain, spinal cord, and muscle cholinesterase activity. His explanation was that the barbiturates decreased the activity of the cholinergic system and, hence, the demand for cholinesterase. Such an explanation is, of course, teleological and demands direct proof of decreased synthesis of the cholinesterases involved.

CONCLUSION

As a conclusion to this review, it may be of value to emphasize certain general points. The literature, as well as our own data, indicates that we cannot speak of "one" cholinesterase, identical in its properties, no matter where it may be found. Within one species, the enzyme differs, in certain respects, from tissue to tissue, and the enzyme of a given tis- sue may differ from species to species. Perhaps, then, it would be more proper to speak of a "family" of cholinesterases, the members of which resemble each other in some attributes and differ in others. Classifi- cation into certain groups may now be possible, but even within such groups, differences in properties may occur. Considerably more ex- perimental work with various criteria of enzyme action is necessary, in order to achieve a more satisfactory classification. The writer believes that many of the controversies on the properties of cholinesterase which have occurred in the literature will be resolved, if the foregoing consid- erations are kept in mind.

BODANSKY: CHOLINESTERASE 545

It would also appear that these considerations will be of aid in elucidating various physiological and physiopathological problems. It has already been pointed out that the cholinesterase activity of the serum is not necessarily an indicator of the cholinesterase activity of the brain. Attempts to define the role of cholinesterase activity in the transmission of nerve impulses must concern themselves with a study of the properties of cholinesterase, at various loci of the nervous system, and with a correlation of the in vivo alterations of the activity of these cholinesterases, in response to administration of drugs or to other fac- tors which have an effect on nerve activity. Finally, it would seem that the compound, DFP, because of its capacity for decreasing mark- edly, and apparently with a considerable degree of irreversibility, the content of cholinesterase in various tissues, may serve as a most useful tool for elucidating the role of cholinesterase and the factors which influence its synthesis and degradation.

BIBLIOGRAPHY

1. Dale, H. H.

1914. J. Pharm. Exp. Therap. 6: 147.

2. Loewi, O., & £. Navratil

1926. Pfliig. Arch. ges. Physiol. 214: 678, 689.

3. Michaelis, L., & M. L. Menten

1913. Biochem. L. 49:333.

4. Linaweaver, H., & D. Burk 1934. J. Am. Chem. Soc. 56 : 658.

6. Haldane, J. B. S.

1930. Enzymes. Longmans, Green & Co. London, New York and Toronto. 85.

6. Goldstein, A.

1944. J. Gen. Physiol. 27: 529.

7. Eadie, G. S.

1942. J. Biol. Chem. 146:85. 1941. J. Biol. Chem. 138: 597.

8. Wright, C. I., & J. C. Sabine

1943. J. Pharm. Exp. Therap. 78: 375.

9. Glick, D.

1937. Biochem. J. 31: 521.

10. Glick, D.

1938. J. Gen. Physiol. 21: 431.

11. Alles, G. A., «fe R. C. Hawes

1940. J. Biol. Chem. 133: 375.

12. Zeller, E. A., & A. Bissegger 1943. Helvet. Chim. Acta 23: 1619.

13. Mendel, B., & H. Rudney 1943. Biochem. J. 37: 59.

14. Nachmansohn, D., & M. A. Rothenberg

1945. J. Biol. Chem. 158: 653.

15. Mendel, B., & H. Rudney 1945. Science 100: 499.

546 ANNALS NEW YORK ACADEMY OF SCIENCES

16. Brown, G. L., & W. Feldberg

1936. J. Physiol. 88:265.

17. Bernheim, F., & M. L. C. Bernheim

1936. J. Pharm. Exp. Therap. 57: 427.

18. Glick, D.

1937. J. Gen. Physiol. 21: 289.

19. Nachmansohn, D.

1940. Nature 145:513.

20. Massart, L., & R. P. DuFait

1939. Bull. Soc. Chim. 21: 1039.

21. Glick, D.

1941. Nature 148: 662.

22. Straus, O. H., & A. Goldstein 1943. J. Gen. Physiol. 26: 559.

23. Easson, L. H., & £. Stedman

1936. Proc. Roy. Soc. London B 127: 142.

24. Collier, H. B., & D. E. Allen

1942. Canad. J. Res. 20B: 189.

25. Ellis, S., F. L. Plachte, & O. H. Straus

1943. J. Pharm. Exp. Therap. 79:295.

26. Mazur, A., & O. Bodansky 1946. J. Biol. Chem. 163:261.

27. Massart, L., & R. P. DuFait

1940. Enzymol. 9:364.

28. Glick, D.

1939. J. Pharm. Exp. Therap. 65: 389.

29. Zeller, E. A.

1942. Helvet. Chim. Acta 25: 216, 1099.

30. Nachmansohn, D., & H. Schneeman

1945. J. Biol. Chem. 159: 239.

31. Torda, C, & H. G. Wolff

1944. Proc. Soc. Exp. Biol. & Med. 57: 236.

32. Heim, F., & W. Rhode

1943. Arch. Exp. Path. Pharm. 202: 216.

33. Matthes, J.

1930. J. Physiol. 70: 338.

34. Lange, W., & G. v. Krueger

1932. Ber. 65: 1598.

35. McCombie, H., E. D. Adrian, B. A. Kilby, & M. Kilby

1941. Personal communication.

36. Adrian, E. D., B. A. Kilby, & M. Kilby

1942. Personal communication.

37. Mackworth, J. F.

1942. Personal communication.

38. Bodansky, O.

1933. J. Biol. Chem. 101: 641.

1936. J. Biol. Chem. 115: 101.

39. Northrop, J.

1921. J. Gen. Physiol. 2: 487.

40. Hall, G. E., & G. H. Ettinger

1937. J. Pharm. Exp. Therap. 59: 29.

41. Kerr, W. J., & G. H. Whipple 1919. J. Physiol. 47:356.

42. Leopold, I. H., & J. H. Comroe

1946. Fed. Proc. 6(11) : 190.

BODANSKY: CHOLIN ESTERASE 547

43. Comroe, J. H., J. Todd, G. Gammon, G. B. Koelle, & A. Gilman

1946. Fed. Proc. 5(11): 172.

44. Harvey, A. M., B. F. Jones, S. Talbot, & D. Grob 1946. Fed. Proc. 5(11) : 182.

45. Rothenberg, M. A., & D. Nachmansohn 1946. Fed. Proc. 5(11) : 199.

46. Crescitelli, F., G. B. Koelle, & A. Gilman 1946. Fed. Proc. 5(11): 172.

47. Easson, L. H., & E. Stedman 1937. Biochem. J. 31: 1723.

48. Walker, M. B.

1935. Proc. Roy. Soc. Med. 28: 759.

49. McGeorge, M.

1937. Lancet 1:69.

50. Milhorat, A. T.

1938. J. Clin. Invest. 17:649.

51. Freudenberg, R., & F. K. Redlich

1938. Arch. Exp. Path. Pharm. 188: 645.

52. Poncher, H. B., & H. W. Wade

1939. Arch. Neurol. Psychiat. 41: 1127.

53. Stedman, E., & W. R. Russell

1937. Biochem. J. 31: 1987.

54. Odom, G., C. K. Russell, & D. McEachern 1943. Brain 66:1.

55. Stoner, H. B., & A. Wilson 1943. J. Physiol. 102: 1.

56. Jones, W. C, & W. C. Stadie

1939. Quart. J. Exp. Physiol. 29: 63.

57. Gammon, G. D., A. M. Harvey, & R. L. Masland 1941. Biol. Symposia 3: 291.

58. Glick, D.

1941. Biol. Symposia 5: 213.

59. Antopol, W., A. Schifrin, & L. Tuchman

1938. Proc. Soc. Exp. Biol. & Med. 38: 363.

60. McArdle, B.

1940. Quart. J. Med. 9: 107.

61. Vahlquist, B.

1935. Skand. Arch. Physiol. 72: 133.

62. Scoz, G., & C. Cattaneo 1937. Enzymol. 4: 157.

63. Vincent, D., & J. de Prat

1942. C. R. Soc. de Biol. 136:821.

64. Faber, M.

1943. Acta Med. Skand. 114: 59.

65. Antopol, W., L. Tuchman, & A. Schifrin

937. Proc. Soc. Exp. Biol. & Med. 36: 46.

66. Faber, M.

1943. Acta Med. Skand. 114: 72.

67. Faber, M.

1943. Acta Med. Skand. 115: 475.

68. Schutz, F.

1943. Quart. J. Exp. Med. 33 : 35.

69. Schutz, F.

1941. Nature 148: 725.

70. Schutz, F.

1942. jj^ J. Physiol._102: 259, 269.

THE EFFECTS OF DRUGS ON NERVE ACTIVITY

By Alfred Oilman*

Pharmacology Section, Medical Division, Chemical Warfare Service, Edgewood

Arsenal, Maryland

I cannot help but feel somewhat apologetic for having consented to contribute to this symposium. Indeed, as the date for this conference approached, it became more and more apparent that I had become in- volved in an extremely paradoxical situation. On first thought, it would seem reasonable to request a pharmacologist to discuss the sub- ject of the effects of drugs on nerve activity. However, a moment's reflection, which unfortunately, from my point of view, was too long delayed, results in the irrefutable conclusion that it is the neuro- physiologist who should be addressing the pharmacologists on this subject.

Although the pharmacologist has, as his ultimate objective, the elucidation of the fundamental mechanism of action of drugs on cells, he is continually frustrated by the limitations of his own technics. The very nature of the subject of pharmacology, which borders on' so many medical disciplines, almost precludes the possibility of the investigator in this field engaging in the basic research which is essential for the reaching of his objective, except, possibly, in a chosen, limited field. It is from the neurophysiologist, therefore, that the answer to many of the basic problems of the actions of drugs on the nervous system can be expected.

If one wishes to indulge in oversimplification, the entire subject of the effects of drugs on nerve activity can be summarized in a few moments or even in a single sentence. There is no phase of nerve activity which cannot be profoundly affected by drugs. Effector cells can be com- pletely released from nervous control or, conversely, the effects of nerve impulses can be faithfully mimicked; conduction in nerve fibers can be completely blocked; synaptic transmission can be interrupted or en- hanced ; cord transection can be simulated ; selected centers in the brain can be stimulated or depressed. It only remains to name the drugs associated with these actions.

Any further amplification would result in a textbook discussion, in which, in a more or less orderly fashion, the actions of drugs could

• Major, Sn-C, A. U. S.

(549)

550 ANNALS NEW YORK ACADEMY OF SCIENCES

be classified in such general terms as local and general anesthetics, analgesics, central stimulants, blocking agents, etc. For the purpose of understanding the therapeutic application of drugs modifying nervous activity, such information may be adequate. However, from the standpoint of the contributions that drugs can make toward the solution of basic problems in neurophysiology, our understanding of their mechanism of action is as yet inadequate.

Drugs have proved to be invaluable tools in many fields of biological and medical research, a statement which the neurophysiologist would be the last to deny. Their value is readily appreciated, when one con- siders one of the few basic and general statements which can be made concerning the fundamental mechanism of action of drugs: namely, that drugs cannot impart new functions to cells or tissues, but can only modify, i.e., stimulate, depress, or block, functions which are the fundamental properties of that cell or tissue. Thus, when a drug pro- duces general or local anesthesia, has a convulsant action, blocks synaptic transmission, stimulates chemoreceptors, or affects nervous activity in any of a variety of ways, no matter how extreme the re- sponse, the assurance is justified, until an exception to the general rule can be proved, that the drug in question has modified a normal cellular function.

Examples of how chemical agents, foreign to the body, have con- tributed to physiological concepts are numerous. Indeed, through these the subject matter of neurophysiology has been enriched. Surely, it cannot be mere coincidence that so many basic contributions to the concept of the chemical transmission of the nerve impulse had their origin in pharmacological laboratories? Rather, the knowledge that chemical agents could mimic, in end-organs, the effect of nerve stimulation served as the basic stimulus for the search for evidence of chemical mediation. Is it not possible that, in some drug, still inade- quately explored by the neurophysiologist, lies the answer to a basic neurophysiological problem? As a possible example, let us consider the local anesthetics. The local anesthetic action of cocaine was first demonstrated in 1884. This type of action has been shown to be exhibited by a variety of chemical structures, but the tertiary amino esters of benzoic acid and para-amino benzoic acid, as a group, are capable of blocking transmission. These compounds show no respect for any classification of nerves, but block cholinergic and adrenergic, sensory and motor fibers in an indiscriminate manner, which points to a basic action on a fundamental mechanism of transmission, shared by all nerves. Is it nnt reasonable to suppose that, by inquiring more

OILMAN: EFFECTS OF DRV OS ON NERVE ACTIVITY 551

deeply into the action of drugs such as the local anesthetics on a cellular level, information may be gained on the mechanism of the propagation of the nerve impulse?

The full realization of the contributions that drugs can make to the elucidation of fundamental physiological mechanisms, can only result from the cooperative research efforts of the pharmacologist and those investigators who are focusing their interests more intensively on a specific field. This is readily appreciated, in so far as the full exploita- tion of the therapeutic applications of a drug is concerned. The pharmacologist, who, in the course of an investigation on central de- pressants, finds a compound with new and significant anti-convulsive properties, will invariably refer the compound to a clinical neurologist, if he deems it worthy of consideration as an anti-epileptic. During the course of chemical warfare research, compounds were studied which, to the pharmacologist, suggested themselves as potential thera- peutic agents in the treatment of such unrelated conditions as Hodg- kin's disease, mercury poisoning, Myasthenia gravis, and glaucoma. In every instance, the prediction of therapeutic worth which was based upon laboratory analyses proved correct, but it was only through the efforts of clinicians, highly specialized in their particular field, that the full appreciation of the therapeutic value of these particular agents was realized.

Drugs are constantly following the path from laboratory to clinic, and many of the outstanding medical accomplishments of the past decade are the result of this cooperative effort. However, drugs can only make their full contribution to the science of medicine when they also follow another, more fundamental, and possibly more important path : namely, from the laboratory of the pharmacologist to the labora- tories of investigators working on those physiological problems which attempt to define biological processes in their most fundamental terms. That drugs are not being fully exploited, in this respect, is a regrettable fact. Even in the field of neurophysiology, where drugs have proveH such valuable research tools, a full realization of their potential contri- butions has not been reached. The tendency, rather, has been to accept drugs with known actions and to employ them for these actions, rather than to investigate unknown mechanisms of drug action as probes into physiological processes. This occurs despite the fact that acetylcholine and physostigmine, two drugs which are better understood than any other compounds affecting nerve action, have paved the way toward revolutionary concepts in an understanding of synaptic transmission.

New agents affecting the nervous system are constantly being de-

552 ANNALS NEW YORK ACADEMY OF SCIENCES

scribed. They are conveniently labeled as convulsants, depressants, etc., and sporadic attempts are made to localize their action, following which, investigation is considered to be complete. I wish to take this opportunity to present three new compounds which have come to the attention of those who, during the past few years, have been workin:? in the field of chemical warfare. These are highly toxic agents which must, by definition, exert profound effects on basic cellular mechanisms. As a result of the cooperative research program, associated with the war effort, these compounds have received more fundamental study, in the course of a few years, by groups with more divergent interest, than is usually the case with better-known and more widely-employed drugs. These three compounds have been selected for discussion from a large group, because of the profound actions which they exert on the nervous system.

The first of these compounds is the sodium salt of fluoroacetic acid. This agent has had an interesting history. It represents, in its prac- tical applications, one of the many fruitful by-products of chemical warfare research. Sodium fluoroacetate was screened by the Fish and Wildlife Service and has proved to be by far the most outstanding of all rodenticides. Now known as Compound 1080, it promises to be an important agent in the control of the spread of disease and the eco- nomic wastage caused by rodents. Yet the prediction is not unjustified that the compound may prove to be of even more significant value as a laboratory research tool, assuming equal importance, in this respect, with the iodoacetates.

Although the fluoroacetate ion resembles iodoacetate in structure, it shares none of its chemical or physiological properties. However, both halogenated acetates exert profound effects on cellular metabolism. Highly provocative are the observations of Barron and co-workers,^ that the oxidation of acetate by yeast, rat kidney suspensions, and heart slices is inhibited by fluoroacetate. When pyruvate is used as the oxidizable substrate, there is an accumulation of acetate, after the addition of fluoroacetate, and the synthesis of carbohydrate from pyruvate is completely inhibited. Barron has advanced the reasonable hypothesis that fluoroacetate, because of its close structural relation- ship to acetate, blocks, by competitive inhibition, enzyme systems con- cerned with the utilization of acetate.

When one considers the basic importance of acetate metabolism, it is of extreme interest to inquire into the pharmacological actions of a compound which, possibly, interferes with the utilization of this essen- tial metabolite. These actions have been investigated by Chenoweth

OILMAN: EFFECTS OF DRUGS ON NERVE ACTIVITY 553

and Gilman.^ Species vary greatly, both in their response and toler- ance to fluoroacetate. In general, the fluoroacetate ion possesses two main pharmacological actions. In some animals, it affects primarily the heart. Energy metabolism and conduction are so disturbed that pulsus alternans, A-V block, frequent ectopic ventricular beats, and eventually, ventricular fibrillation result. Most herbivorous animals, as well as those species of primates that have been studied, are sus- ceptible to the cardiac actions of the fluoroacetates. The actions of the fluoroacetates on the nervous system are even more striking. Follow- ing a latent period of approximately one hour, animals become progi'es- sively more excited, and eventually exhibit severe epileptiform con- vulsions which continue, uninterrupted, until death. Dogs and cats respond in this manner. Most of those species which exhibit the car- diac actions show no evidence of central stimulation. However, primates, although succumbing to the cardiac action, may show mild epileptiform convulsions. In regard to susceptibility, the lethal con- vulsive dose in the dog is approximately 0.1 mg./Kg. The lethal dose in primates is approximately 100 times as great.

When one considers the descriptive pharmacology of fluoroacetate in the light of its possible fundamental mechanism of action, certain ques- tions come immediately to mind. Are differences in species response due to different metabolic patterns in their nervous tissue? Observa- tions of Chenoweth and co-workers support the view that the primate myocardium is uniquely dependent upon the utilization of acetate, for adequate function. Similar investigations may reveal that the same is true of the nervous system in the case of the dog and cat. It has been shown by Tepperman and Mazur^ that, in the presence of fluoro- acetate, acetylation is greatly enhanced, presumably because of the high concentration of available acetate. Could this finding possibly be concerned with the convulsive action of the fluoroacetate ion? These are but a few of the problems, pertinent to the nervous system, that have been raised by the preliminary investigations of this drug. By the proper utilization of fluoroacetate as a research tool, it may be possible to relate specific disturbances in the metabolism of nervous tissue to functional abnormalities. In this respect, it is of interest to note that the electroencephalogram obtained during a fluoroacetate- induced convulsion is almost identical to that of a petit mal epileptic seizure.

The second compound to be discussed is diisopropyl-fluorophosphate. This compound represents a new type of anticholinesterase. Not only does it depart, in its chemical configuration, from previously studied

554 ANNALS NEW YORK ACADEMY. OF SCIENCES

anticholinesterase agents, but it also differs in its fundamental mecha- nism of action, in that the inhibition of cholinesterase is irreversible. What is more, diisopropyl-fluorophosphate is highly lipoid-soluble and rapidly gains access to nervous tissue.

In diisopropyl-jfiuorophosphate, the neurophysiologist has at his command a research tool in which the fundamental mechanism of ac- tion is known. Thus, if the major premise of my introductory re- marks is to hold true, the application of this compound to problems of neurophysiology should help to prove or disprove fundamental con- cepts of nerve function.

The advantages of an irreversible anticholinesterase, as a research tool, are at once evident. Following the action of diisopropyl-fluoro- phosphate, the cholinesterase activity of a tissue can be restored only by resynthesis of enzyme. Moreover, the agent can be administered, and the response of a tissue studied. Following this, the tissue can be removed and ground ; the homogenate appropriately diluted ; and the absolute cholinesterase activity determined ; an approach which cannot be employed with a compound such as physostigmine, due to the re- versible nature of its inhibition. Thus, for the first time, a highly quan- titative approach to the problems of the role of cholinesterase and acetylcholine in the transmission of the nerve impulse is available.

Diisopropyl-fluorophosphate has already followed the path from the laboratory to the clinic. Comroe and associates,^ at the University of Pennsylvania, and Harvey and co-workers,^^ at Johns Hopkins, have employed this agent in the treatment of Myasthenia gravis. The therapeutic efficacy of this type of compound, as well as its limitations, has already been demonstrated. Of even greater interest, will be the more fundamental data, from these studies, which may shed light on the defect in transmission associated with this myopathy.

Basic laboratory studies, employing diisopropyl-fluorophosphate as a research tool, have also begun. I should like to report, in some detail, the experiments of Crescitelli and co-workers,® designed to elucidate the possible role of acetylcholine in the conduction of the nerve impulse along the nerve fiber. The background literature to this problem has recently been summarized by Loewi^ and by Feldberg,^ and need not be repeated here. Mention should be made, however, of the studies of Cowan,^ Lorente de No,^" Hertz," and Cantoni and Loewi,^^ in which either physostigmine or acetylcholine failed to exert a significant effect on transmission in the nerve fiber. However, the availability of an irreversible inhibitor of cholinesterase, which afforded an oppor- tunity to correlate nerve function with quantitative data on cholin-

OILMAN: EFFECTS OF DRUGS ON NERVE ACTIVITY 555

esterase concentration, prompted a repetition of this type of study. Moreover, advantage was taken of the opportunity to compare the effects of a reversible (physostigmine) and an irreversible anticholin- esterase. The nerve action potential was employed as an index of effect on transmission. It was argued that, whereas both types of anti- cholinesterase agent should affect the nerve action potential in the same manner, assuming acetylcholine to play a major role in trans- mission, the action of physostigmine should prove reversible, that of diisopropyl-fluorophosphate, irreversible. The possibility that phy- sostigmine might not gain access to those structures accessible to di- isopropyl-fluorophosphate was avoided, as far as possible, by employ- ing the alkaloid base, as well as the salicylate salt.

Two types of experiments were performed. In the first, isolated nerves of bull frogs and of cats were mounted in a moist chamber, placed in a constant temperature bath of appropriate temperature. A portion of the nerve was looped into a small chamber containing Ringer's solution. Following the recording of control action potentials, the effects of the various drugs were ascertained. When the isolated nerves of the cat or the bull frog were exposed to 0.01 molar physostig- mine salicylate, no detectable change in the action potential was ob- served. However, when the solution containing the salicylate salt of physostigmine was replaced by the alkaloidal base, the action potential disappeared within a period of 10 minutes. Washing the nerve with Ringer's solution restored the action potential completely. Thus, the water-soluble salicylate salt was devoid of action, whereas the lipoid- soluble alkaloidal base blocked transmission.

Similar experiments were then performed, by exposing the nerve to 0.02 molar diisopropyl-fluorophosphate. Again, the action poten- tial disappeared within a few minutes. It only remained to demon- strate the irreversibility of this block, in order to attribute the effect to the inactivation of cholinesterase. However, washing the nerve restored the action potential, despite the fact that the action of di- isopropyl-fluorophosphate was supposedly irreversible. In view of this surprising result, experiments were performed to determine the extent of the wash necessary to restore the nerve and the speed at which the action potential returned. During the course of these studies, it was observed that it was only necessary to remove the nerve from contact with the solution containing diisopropyl-fluorophosphate, to restore the action potential completely. Thus, the conduction defect could not have been related to an inhibition of cholinesterase.

It was not possible, in the experiments on the isolated nerve, to meas- ure accurately the extent of inhibition of cholinesterase, for the reason

556 ANNALS NEW YORK ACADEMY OF SCIENCES

that only a very small segment of nerve was exposed. It was, thus, impossible adequately to wash the nerve, so as to preclude the possi- bility of mechanical transfer of sufficient diisopropyl-fluorophosphate to inactivate cholinesterase during the preparation of the nerve for enzymatic studies. For this reason, a series of experiments was per- formed in which a large dose of diisopropyl-fluorophosphate was in jected into the ventral lymph sac of frogs and allowed to reach the nerve, by way of the circulation. After a suitable interval, the nerves were dissected and their transmission characteristics studied and com- pared with control frogs. Following this, the cholinesterase content of the control and experimental nerves was studied. Despite the fact that the nerves of the experimental frogs were completely devoid of cholinesterase, the transmission of the nerve impulse, as determined by the characteristics of the action potential, in response to single and repetitive stimuli, was unaffected. This finding casts serious doubt on the role of acetylcholine as a de-polarizer, in the processes of con- duction along the axon.

Loewi, in his recent review, quotes Dale as having once remarked that it was unreasonable to suppose that nature would provide for the liberation in the ganglion of acetylcholine, the most powerful stimulant of ganglionic cells, for the sole purpose of fooling physiologists. What, then, is the function of cholinesterase in nerve fibers, which Nachman- sohn and his co-workers have shown so conclusively to be concen- trated at the surface, rather than in the axoplasm? The answer is not yet forthcoming. However, in a drug like fluorophosphate, it is possi- ble, by localized injection, to reduce the concentration of cholinesterase in a chosen tissue to negligible amounts. Thus, we have a research tool which may provide the answer to these basic problems.

The third agent will be discussed only very briefly. It shares with diisopropyl-fluorophosphate the ability irreversibly to inactivate cholinesterase. It differs from diisopropyl-fluorophosphate in pos- sessing a more outstanding action on the nervous system. Certain spe- cies, in particular cats and dogs, exhibit severe convulsions, which have their onset within a few minutes after the intravenous injection of the drug and which persist until death. The fact that an anticholinesterase agent possesses such extreme convulsant action could possibly be attributed to coincidence. However, there is one finding which points to an intimate relationship between convulsions and the chemical mediation of central synaptic transmission. If the animals receive a therapeutic dose of atropine, before the administration of this anti- cholinesterase, no convulsions are observed, and complete protection

OILMAN: EFFECTS OF DRUOS ON NERVE ACTIVITY 557

is afforded from what would, otherwise, be a lethal dose. Moreover, if the agent is administered and the convulsions are allowed to progress to their peak intensity, the intravenous injection of atropine stops all convulsive activity within 30 seconds, and the animal appears nor- mal, as soon as it recovers from its exhaustion. It should be empha- sized that the doses of atropine that exert this anticon\^lsant action are of a small magnitude and, in themselves, exert no demonstrable central effects. No other central stimulant can be inhibited in this manner. The effect of atropine, in blocking the convulsant action of this anticholinesterase, is as dramatic and as complete as is the effect of atropine in blocking the reception of post-ganglionic cholinergic im- pulses by autonomic effector cells. It seems certain that, in this com- pound, there is a research tool which can make a significant contribu- tion to the fundamental problems of central synaptic transmission.

During the past few years, the group of investigators at Edgewood Arsenal has been engaged in a cooperative research effort, in which toxic war gases or, in other words, highly active drugs, were the focal point of their investigations. Their efforts were coordinated with extensive programs of numerous academic groups. From the point of view of the pharmacologist, this elaborate approach to the mechanism of drug action has proved to be an illuminating experience. One cannot fail to be impressed by the fact that, as the story of each agent unfolded, its potential value toward the solution of fundamental problems in phy- siology and biochemistry was more and more appreciated.

I have departed from a routine discussion of the effects of drugs on nervous activity, to present to you three new agents which have re- sulted from this program. It may be predicted that, as research tools, they will prove invaluable. If so, then the contention made earlier will have been fulfilled: that, by tracing the actions of drugs to their cellular mechanisms, basic physiological processes will be revealed.

REFERENCES

1. Barron, E. S, G., G. R. Bartlett, & G. Kalnitsky 1946. Fed. Proc. 5(11): 120.

2. Chenoweth, M. B., & A. Gilman 1946. J. Pharm. Exp. Therap. 87; 90.

3. Tepperman, J., & A. Mazur Per.sonal Comiiunication.

4. Comroe, J. H., Jr., J. Todd, G. Gammon, G. B. Koelle, & A. Gilman 1946. Fed. Proc. 5(11) : 172.

5. Harvey, A. McG., B, F. Jones, S. Talbot, & D. Grob

1946. Fed. Proc. 5(11): 182.

6. Crescitelli, F. N., G. B. Koelle, & A. Gilman

1946. J. Neurophysiol. 9:24.

558 ANNALS NEW YORK ACADEMY OF SCIENCES

7. Loewi, O.

1945. J. Mount Sinai Hosp. 12: 851.

8. Feldberg, W.

1945. Physiol. Rev. 25: 596.

9. Cowan, S. L.

1938. J. Physiol. 93:215.

10. Lorente de No, R.

1944. J. Cell. Comp. Physiol. 24: 86.

11. Hertz, H.

1945. J. Physiol. 104: 1.

12. Cantoni, G. L., & O. Loewi

1944. J. Pharm. Exp. Therap. 81: 67.

THE RECOVERY OF DIAMETER AND IMPULSE

CONDUCTION IN REGENERATING

NERVE FIBERS

By Charles M. Berry and Joseph C. Hinsey

Department oj Ayiatomy, Cornell University Medical College, New York, N. Y.

The primary purpose of these experiments was to study those prop- erties of regenerating fibers which could be observed oscillographically and related to histological controls. Therefore, the contours of the action potentials, the conduction velocities of the impulses, and fiber diameters were followed in a series of cat nerves. The recovery of these properties might be considered to be a process of reconstitution or maturation, as opposed to the longitudinal outgrowth of the fibers, and since these properties continued to change over long periods of regenera- tion, measured in years, the experiments were spread over a wide range of time, from a few days to more than three years.

Having observed the effects of time (which is undoubtedly the most important factor in the reconstitution of fibers distal to a suture) , we extended the experiments to include the effects of crushing the nerves compared to section and suture; the effects of delaying the suture after transection; and the effects of cross-suturing nerves containing fibers of different fiber diameters. The importance of these factors has been reemphasized in recent publications. The growth of fiber diameters in the distal stump over a one year period has been carefully plotted by Gutmann and Sanders.^ Furthermore, they showed differences in recovery between crushed and sutured nerves. The influences of phy- sical stresses in the '''union" tissue described by Weiss- show the im- portance of the type of junction between central and distal stumps. The effects of delaying the suture after section of a peripheral nerve were studied by Holmes and Young,^ and the effects of cross-suturing visceral and somatic nerves were reported by Simpson and Young.* Young and his co-workers have paid special attention to the connective tissue sheath diameters in the distal stump.

METHODS AND RESULTS

Action Potentials from Regenerating Nerves Cathode-ray oscillographs were taken from regenerating tibial, peroneal, and saphenous nerves of 64 cats. The nerves were transected

(559)

560 ANNALS NEW YORK ACADEMY OF SCIENCES

with a sharp blade and immediately sutured with silk. Each nerve was allowed to regenerate for a determined interval, from 17 to 1363 days, and was then excised and placed on electrodes in a moist chamber at 38° C.

The regenerating fibers were able to conduct impulses after a very brief regeneration period, and at 17 days, action potentials were re- corded from the distal stump of one tibial nerve within a distance of 2 cm. from the suture. This potential was small, less than 10 micro- volts in amplitude, and was conducted very slowly at a maximum of 0.9 meters per second. The potentials recorded from fibers which had regenerated for longer periods were of greater amplitude and were con- ducted at greater velocities. The increase in conduction velocity was rapid in the first few days and, by 36 days, velocities of 17 m.p.s. were recorded, as shown in plate 5 A. This record was taken from a mono- polar electrode placed 3.5 cm. distal to the suture, and shows a maxi- mum conduction velocity of 17 m.p.s. and a spike amplitude of 25 microvolts. The maximum conduction velocity continued to increase with the time allowed for regeneration, but at an ever-decreasing rate of recovery. Thus, at 50 days, 25 m.p.s. were attained; at 100 days, 40 m.p.s.; at 200 days, 60 m.p.s.; at 365 days, 70 m.p.s. Beyond 544 days, no further recovery of conduction velocity was found, and at the long period of 1363 days, only 80 m.p.s. were attained. The record in PLATE 5 C was taken from the distal stump of a tibial nerve 1363 days after suture, and can be compared with the record from the opposite, normal, tibial nerve of the same animal in plate 5 B, in order to deter- mine the degree of recovery. In plate 5, B and C, the conduction dis- tance was 8 cm., but in the record from the regenerated nerve, the dis- tance or time between the shock artifact and the beginning of the spike is greater than in the normal record, and shows that the 80 m.p.s. repre- sent less than 80% recovery toward the normal conduction velocity. Similar results were obtained from the peroneal and saphenous nei'ves.

Two other observations can be made from plate 5, B and C. Firstly, the amplitude of the spike is less in the regenerated nerve, and secondly, the spike in plate 5 C is not as complex. The lack of recovery of all the components of the spike was even more obvious in records from the saphenous nerves, where the normal potential is more complex and consists of a double or triple peaked alpha wave and distinct beta, gamma, and delta waves. Even after long periods of regeneration, the saphenous nerve did not recover these wave components and showed only an initial peak which leveled off into a long tail.

BERRY— HINSEY: RECOVERY, REGENERATING NERVES 551

An accurate method of determining the conduction velocities of the most rapidly conducting fibers is illustrated in figure 1. Several records were taken along a regenerating nerve at various conduction distances, and either the stimulating or the recording electrodes were placed at a fixed point along the nerve. Then the other electrodes were moved stepwise, to provide a greater conduction distance for each suc- cessive record. Thus, in figure 1, the stimulating electrodes were placed 4.6 cm. distal to the suture, and the pair of recording electrodes was placed at a variety of points both distal and proximal to the

lOr-

i

0 c o

in

•0

c

o

0

;3 c

o

o

8-

-		/
-	• ^^
1	1 1 1 1 1 1 1 1	1 1 1 1 1 1

2-

0.5 10

Conduction time - msec.

1.5

"'^ Central — 8 end

-7

-6

*-5

) Suture

-3

■^^ Distal <-jL end

Figure 1. Graph of conduction distance against conduction time of the action potentials, from a saphenous nerve, 58 days after suture. The diagram on the right sliows that the stimulating electrodes were held stationary on the distal stump, while the recording electrodes were moved. The conduction velocity jumped from 23 m./s. in the distal stump to 50 m./s. in the central stump. (Berry, Grundfest, & Hinsey.')

suture. At each distance, a record was taken, and the conduction time was measured between the shock and the initial rise of the spike. When this time was plotted against conduction distance, as in figure 1, the slope of the line indicated the maximum conduction velocity.

The change in the slope of the line, in figure 1, occurred at the suture line which shows that the distal outgrowths conduct more slowly than their parent fibers in the central stump. The actual velocities from this saphenous nerve, 58 days after suture, were 23 m.p.s. distal to the suture, and these same fibers central to the suture conducted at 50 m.p.s. The continuity of the plotted line and its straight contour cen- tral to the suture demonstrates that the change in velocity was recorded from identical fibers on both sides of the suture.

Assuming that conduction velocity is related to the fiber diameter, the electrical method proves that the small fibers of the distal stump are not the result of a selective ability of only the small fibers of the

562 ANNALS NEW YORK ACADEMY OF SCIENCES

central stump to grow out into the distal stump, before the larger fibers can grow. Conversely, it demonstrates that large fibers of the central stump send out small extensions into the distal stump, which then ma- ture and take on the action potential characteristic of smaller fibers, regardless of their origin.

A consistent finding, not illustrated in figure 1, was that the conduc- tion velocity of the distal fibers was less at greater distances from the suture.

Fiber Diameter Measurements

The regenerating tibial, peroneal, and saphenous nerves which had been excised and used for action potential experiments were fixed in osmic acid and studied microscopically. The most obvious change in the character of the regenerating nerves, as they were allowed to grow for longer and longer periods, was the gradual increase in fiber diam- eters. The outside diameters, including both axis cylinders and myelin sheaths, were measured with a movable, ocular micrometer. To insure random sampling, the fibers were measured in horizontal bands, with approximately 500 fibers measured in each nerve. The results were plotted along fiber distribution curves, as illustrated in figure 2. These nine histograms were picked from a series of regenerating tibial nerves, to show the diameter characteristics of the fibers in the distal segments at different time intervals following suture. Since the diameters were smaller, at greater distances from the suture, the portions of tibial nerves studied in each of the histograms in figure 2 were taken from similar levels, just beyond the upper branches to the gastrocnemius muscles, 3 to 5 cm. distal to the sutures.

The shift in distribution from left to right in the histograms of FIGURE 2 shows rapid diameter growth between 33, 59, and 127 days after transection and suture. At longer regeneration times of 207, 318, and 420 days, the fibers continued to mature, but more slowly. At 544, 901, and 1363 post-operative days, there was negligible increase in fiber diameter, but a complete recovery of the normal fiber size was never attained. Even at 1363 post-operative days, the large group of fibers between 9 and 20 micra had not appeared. Similar lack of com- plete recovery of fiber size was found in the peroneal and saphenous nerves.

Particular attention was paid to the measurement of the largest fibers in each nerve, since the maximum conduction velocity had al- ready been determined accurately, and it was a reasonable assumption that the largest fibers were responsible for the maximum conduction velocity. The growth of the largest fibers is illustrated in figure 3

BERRY— HINSEY: RECOVERY, REGENERATING NERVES 563

40-
»- 30- Z UJ			33 DAYS
O 20- UJ Q- 10-		-|
2		-i:)-

UJ

CD 20

FIBER DIAMETER IN MICRA

Figure 2. Reconstitution of fiber diameter with regeneration time. Each fiber diameter distri- bution graph is from a different nerve, excised on the indicated number of days after transection and suture.

(6-1

14-

< \2-\

a:

o

q:

UJ

<

6-

tr 4.

bJ

m u. 2

-r

T

300 600 900

REGENERATION TIME IN DAYS

1200

FiCTRB 3. Outside diameters of the largest fibers in the distal stumps of sutured tibial nerves. The fibers do not attain the normal maximum diameters, above 20 micra.

564

ANNALS NEW YORK ACADEMY OF SCIENCES

as a graph of fiber diameter against the amount of time allowed for re- generation after suture of the tibial nerves. The leveling off of the diameter growth curve at a level between 14 and 16 micra again illus- trates the incomplete "maturation" of fibers, even after long periods of regeneration.

The Relationship between the Conduction Velocity and the

Fiber Diameter

The results from both the action potential and the fiber diameter studies showed a gradual recovery, which tapered off with no complete

100

"--Tibial • ^Peroneal A = Saphenous

6 8 10 12

Tibep diametep - /j.

Figure 4. Graph showing the relationship between the maximum conduction velocities and the outside diameters of regenerating fibers of the nerves: tibial (crosses), peroneal (circles), and saphenous (triangles). (Berry, Grundfest, & HinseyJ)

return of either the velocity of impulse conduction or the size of the fibers. The actual relationship between the two functions is shown in FIGURE 4, where the maximum conduction velocity is plotted against

BERRY— HINSEY: RECOVERY, REGENERATING NERVES 565

the maximum fiber diameter for each nerve. The linear relationship between the two functions is thus illustrated in a new way, and con- firms the contention of Gasser and Grundfest,^ that the relationship is a linear one. Less deviation from the straight line in figure 4 was found at the lower end of the graph, if the inside diameter (without myelin sheath) was measured instead of outside diameter.

The Difference in Recovery between Sutured and Crushed Nerves

In a small series of 10 cats, the tibial, peroneal, and saphenous nerves were crushed with thin, flat-surfaced forceps, and the nerves were al- lowed to regenerate for determined intervals. The purpose of these experiments was to determine whether the recovery of the action poten- tial and fiber diameter would occur in the same way as had been ob- served in the sutured nerves.

25

UJ

o

UJ

Q.

CO QC UJ CO

o

cr

UJ CD

20

15-

10

5-

"1 — T — I — i — I — I — I — 1~ I — I — I — I I I I 1 I r 23456789 123456789

FIBER DIAMETER IN MICRA

10

FiGUBE 5. Comparison of fiber distribution according to diameter. Histogram on the left, after suture. Histogram on the right, after crush. Both tibial nerves were allowed to regenerate for 94 days.

The action potential records from the crushed nerves showed greater recovery of conduction velocity and magnitude of the spikes than was found in records from comparable regions of sutured nerves, taken after the same amount of time had been allowed for regeneration. The fibers also grew in diameter more rapidly in the crushed nerves. In FIGURE 5, the histogram on the left is from the distal stump of the tibial nerve, 94 days after suturing, that on the right, from a tibial nerve, 94 days after crushing. Shift of the graph to the right, in the c

566 ANNALS NEW YORK ACADEMY OF SCIENCES

nerve compared to the sutured, demonstrates the more rapid recovery of fiber size in the crushed nerves. These experiments were not carried beyond 200 days, and the ultimate amount of recovery of the normal fiber distributions was not determined for crushed nerves.

The Effects of Delayed Suture on Fiber Diameter Growth

The method of studying the "maturation" of the fibers by action potential records and diameter measurements was used to determine the influence of delaying the suture after transection. An operative procedure similar to that of Holmes and Young^ was devised, to allow the tibial nerve to degenerate after section and remain free of fibers during a determined delay period. The adjacent peroneal nerve was then transected, and the freshly cut central stump was sutured to the old distal remnant of the tibial. This cross-suture was used to limit the study to effects of delay in the distal segment.

In the same animal, a reliable control was provided in each experi- ment by suturing the peroneal nerve of the opposite leg to the tibial nerve. This was done in exactly the same manner as on the delayed side, but, in this case, there was no delay between section and suture.

The procedure may be summarized as follows: The right tibial nerve was exposed, and a long segment removed from the sciatic notch to the popliteal space. To insure the absence of regeneration during the de- lay period, the cut nerve was exposed at 6-month intervals. Then, after a delay of 14 to 476 days, a second operation was performed, in which the right peroneal was sectioned and the central end sutured in the old distal tibial. At this same time, the left peroneal was sectioned and sutured to the distal tibial as a control. 105 to 440 days were then allowed for regeneration before the terminal experiment.

The shapes of the action potentials and the maximum conduction velocities in the distal segments, after delay periods of 14, 21, 28, 56, 84, and 180 days, were similar to those recorded from the nerves of the opposite leg which had been sutured without delay. In each of these experiments, 105 days were allowed for regeneration after suture. The diameter distribution of the fibers was the same on both sides, as illus- trated in FIGURE 6, A and B. The histogram in figure 6 B was from the right tibial, 3 cm. distal to the suture, which was delayed 84 days, and after which the nerve regenerated for 105 days before the fibers were measured. The control from the same animal is shown in figure 6 A. Both nerves were analyzed 105 days after suture. The similarity of the two histograms was also found with 14, 21, 28, 56, and 180 days, and indicates that such delay periods had no influence on the diameter growth of fibers, under these experimental conditions.

BERRY— HINSEY: RECOVERY, REGENERATING NERVES 567

A suture delayed for a longer period of 253 days resulted in less re- constitution of the fiber diameters in the peripheral stump than that found in the control. The histogram in figure 6 D was from the dis- tal tibial, 3 cm. distal to the suture, which was delayed 253 days.

DIAMETER IN MICRA

Figure 6. The effects of delayed suture on fiber diameter growth in the distal stumps of re- generating nerves. Control sutures with no delay, on the left (A, C, E) ; delayed sutures, on the right (B, D, F).

A is from th^ left tibial nerve, 105 days after immediate suture; B is from the right nerve of the same cat, 105 days after a suture which was delayed 84 days; C and D, 440 days after suture with 253 days delay in D ; E and F, 337 days after suture with 476 days delay in F.

Figure 6 C was from the same level of the opposite tibial, with no delay. Both nerves were allowed to regenerate for 440 days after suture, and it must be pointed out that the longer regeneration period may be an important factor. The fibers in figure 6 D were smaller throughout, with a maximum diameter under 11 micra. On the control side, figure 6 C, the fibers were generally larger, with a few reaching 14 micra.

The effects of delay were even more marked after 476 days, as shown in FIGURE 6, E and F. The histogram in F, after 476 days delay, shows most of the fibers to be less than 4 or 5 micra, while the histogram of

568 ANNALS NEW YORK ACADEMY OF SCIENCES

the control in E shows much larger fibers throughout (figure 6) . Both nerves were allowed to regenerate 337 days after suture.

The maximum delay which caused no change in fiber growth could not be determined accurately from these experiments, because the regeneration time was not kept constant. However, in those experi- ments with 253 and 440 days delay, approximately a year was allowed for regeneration. These experiments were more conclusive, and indi- cated that such delay periods restrict the diameter growth of the re- generating fibers. Not only were the largest caliber fibers limited in growth, but the whole fiber distribution curve was altered.

The Effects of Cross- Suturing Nerves of Different Fiber Caliber

On the basis of cross-suture experiments, Simpson and Young'* de- scribed a restrictive influence on fiber diameter growth by very small Schwann tubes. Using a similar approach, Hammond and Hinsey® cross-sutured the hypoglossal nerve and the cervical sympathetic trunk. The choice of these nerves was fortunate, since they contain much different tube diameter distributions, and they are situated close ^ together for easy manipulation for cross-suturing.

Fiber diameter distributions of the normal hypoglossal nerves, meas- ured from osmic preparations, showed large fibers, between 2.3 and 17 micra in outside diameter, with a unimodal peak between 6.5 and 8.5 micra. Calculation from 11 hypoglossal nerves showed the median fiber diameter to be 7.7 micra, with only 3% of the fibers smaller than 4.5 micra. Similar observation on the cervical sympathetic trunk at its rostral end showed relatively small fibers, between 1.2 and 8.4 micra in diameter, with a unimodal peak at 2.5 to 3.5 micra. The median diameter calculated from 7 experiments was 3.3 micra, and only 7% of the fibers were larger than 4.5 micra.

In order to compare the effects of suture of a nerve with large fibers into a nerve with smaller fibers and the effects of a control experiment in which the nerve with large fibers was sutured into its own distal stump, it was necessary to run simultaneous experiments, with the nerves excised and measured at the same intervals after suture. There- fore, in one set of experiments, the hypoglossal nerves were sectioned and immediately joined to their own distal stumps. The fiber distribu- tion in the distal stumps of these hypoglossal-to-hypoglossal sutures was determined as illustrated in the top row of histograms in figure 7. The progress of diameter growth of the fibers is indicated by a shift of the curves from left to right, as the time of regeneration increased to 216, 250, 300, and 365 days. This increase is similar to that found

BERRY— HINSEY: RECOVERY, REGENERATING NERVES 569

in the experiments on the tibial, peroneal, and saphenous nerves, and about 80% recovery of normal fiber diameter was found at 365 days.

.5 3.5 6.5 9.5 .5

FIBER DIAMETER

I I I I I I I I I I I I

,5 3.5 6.5 9.5

IN MICRA

Figure 7. The top row of fiber distribution graphs is from dista! stumps, after simple suture of the hypoglossal nei-\-e; bottom row, from the distal stump, when the central end of the hypo- glossal was sutured to the cervical .sympathetic trunk.

The time allowed for regeneration in both the top and bottom rows is indicated in days. (Modified from Hammond & Hinsey.")

With the diameter studies of the hypoglossal-to-hypoglossal series as a basis for comparison, the suture of the hypoglossal nerve to the cervical sympathetic trunk produced remarkably different results. The lower row of histograms in figure 7 shows the fiber distributions in the distal stumps (cervical sympathetic), at the same intervals of regeneration as those in the upper row of hypoglossal-to-hypoglossal experiments. At 216 days, the histogram of the distal segment shown in the lower row was not much different than that of the distal segment in the upper row. However, at 250, 300, and 365 days, the fibers did not continue to grow as in the simple hypoglossal-to-hypoglossal suture, but, instead, the diameters decreased. By 365 days, the distal segment of the hypoglossal-to-cervical sympathetic cross-suture took on the diameter characteristics of the normal cervical sympathetic trunk, in- stead of the hypoglossal nerve.

The converse experiments, those of cross-suturing the cervical sym- pathetic trunk into the hypoglossal nerve, were carried out to find out if the larger Schwann tubes of the distal segment would allow the re-

570 ANNALS NEW YORK ACADEMY OF SCIENCES

generating fibers to expand beyond the diameters of their parent fibers in the cervical sympathetic trunk. The results showed no expansion, and the distal fibers tended to recover the characteristics of the cervical sympathetic trunk.

DISCUSSION

The growth of fiber diameter is part of a process of reconstitution or maturation of the fibers and can be considered separately from the longitudinal growth of the fibers toward the periphery. Although this latter process of outgrowth has been the subject of numerous studies, the diameter growth was not investigated intensively until the re- search of Gutmann and Sanders^ on rabbit nerves. They described a gradual increase in fiber diameter for the first year of regeneration after suture, without recovery of either the maximum diameter or of the bimodal fiber distribution. Our experiments confirm these findings and, in addition, show that the increase in diameter con- tinues beyond one year to at least 544 days. Also, at extremely long times, up to 1363 days after suture, complete recovery of diameter is not attained, nor is the bimodal distribution of fibers. The fact that crushing allows much more rapid reconstitution of fiber diameter con- firms their findings (Gutmann and Sanders^).

The recovery of impulse conduction velocity and action potential characteristics is also an important part of this reconstruction or ma- turation process in regenerating nerve fibers. In fact, the proper coordination of nerve functions might be impossible if their conduction velocities are not regained, in spite of proper peripheral connections. These combined electrical and microscopic experiments showed that the conduction velocities and action potentials recover slowly, at the same rates, as the fiber diameters increase. The conduction velocities were actually compared to the fiber diameters of the distal stump, and the same linear relationship between these two functions was found as ex- pected for normal nerves, as reported by Gasser and Grundfest.^ Therefore, the electrical characteristics of the regenerating outgrowths in the distal stumps were found to be related solely to the reconstitu- tion of fiber diameter and did not otherwise depend on the type or size of the parent fiber in the central stump. Further data on the ac- tion potentials from regenerating nerves have been reported by Berry, Grundfest, and Hinsey.'^

A long delay between sectioning the nerve and subsequent suture was shown to impede the usual reconstitution of fiber diameter and con-

BERRY— HINSEY: RECOVERY, REGENERATING NERVES 571

duction velocity. Holmes and Young^ have described this phenomenon and have shown that the connective tissue tubes in the distal stump undergo shrinkage during the delay period. The experiments reported here show that delay causes even greater influence than they descril)ed. This difference in results is probably due to the fact that our experi- ments allowed the nerves to regenerate for much longer periods, during which the restrictive influences could be more strongly exerted. How- ever, the results show that brief delay periods have little effect, but that delays of 253 and 476 days produced considerable interference with fiber reconstitution. Unfortunately, the exact delay times between no effect and slight effect could not be determined from these experi- ments. The introduction of control cross-sutures, without delay in nerves of the opposite leg of the same animals, seems to rule out the factors operating at the suture line, or differences in peripheral re-in- nervation (unless atrophy of the muscle is considered), which can in- fluence the fiber reconstitution. Presumably, therefore, only differ- ences in the condition of the connective tissue and Schwann tubes of the distal stump are responsible for the results in these experiments.

The influence of the connective tissue or Schwann tubes in the distal stump on the fiber growth has been recently emphasized by Sanders and Young,^ who found that the motor branches of a sutured, mixed nerve contained fibers of larger caliber than the sensory branches. Also, Simpson and Young'* cross-sutured somatic nerves into the splanchnic and anterior mesenteric nerves and suggested that the restriction in fiber diameter which resulted might be due, in part, to the small size of the peripheral tubes. The results reported here by Hammond and Hinsey'' showed this same restriction in cross-sutures of the hypoglossal and cervical sympathetic. However, in these experiments, the nerves were allowed to regenerate for longer periods than reported by Simpson and Young,^ and an additional phenomenon was disclosed. At 216 days after cross-suture of the hypoglossal to the cervical sympathetic, the recovery of fiber diameter was slightly less than that obtained in control, hypoglossal-to-hypoglossal, sutures. At 250, 300, and 365 days, the fibers not only showed greater restriction of growth, but actu- ally the caliber of the fibers found distally became smaller than they were at 216 days. There is no conclusive explanation of this apparent reversal of diameter growth, but two facts might be mentioned. First, the final histogram (figure 7), at 365 days, resembled that of the orig- inal distal stump before operation, and, perhaps, the small tubes com- pressed or killed off the larger fibers. Secondly, it must be recognized that the hypoglossal fibers could not reach proper end organs by grow-

572 ANNALS NEW YORK ACADEMY OF SCIENCES

ing down the cervical sympathetic trunk. Simpson and Young^ showed the importance of these peripheral connections by cutting a regener- ating nerve peripheral to the original suture, which prevented the re- establishment of peripheral connections. Weiss and Taylor^ also found evidence that fibers were smaller when re-innervation of the end organs was prevented.

SUMMARY

1. Excised, distal stumps of tibial, peroneal, and saphenous nerves of cats were studied oscillographically and microscopically, at intervals, up to 1363 days after transection and suture.

2. The processes of maturation or reconstitution of fiber diameter and impulse conduction velocity continued over a long period of at least 544 days. The regenerating fibers never completely recovered.

3. Crushed nerves recovered fiber diameter and conduction velocity more rapidly than sutured nerves.

4. Delay between transection and suture of more than 6 months in- terfered with the reconstitution of the regenerating fibers.

5. Cross-suturing a nerve containing large fibers into a distal stump containing small connective tissue or Schwann tubes resulted in re- striction of fiber diameter growth.

REFERENCES

1. Gutmann, E., & F. K. Sanders

1943. J. Physiol. 101:489.

2. Weiss P.

1944. ' J. Nemo.surg. 1: 400.

3. Holmes, W., & J. Z. Young

1942. J. Anat. 77: 63.

4. Simpson, S. A., & J. Z. Young

1945. J. Anat. 79:48.

5. Gasser, H. S., & H. Grundfest 1939. Am. J. Physiol. 127: 393.

6. Hammond, W. S., & J. C, Hinsey 194.'). J. Comp. Neurol. 83: 79.

7. Berry, C. M., H. Grundfest, & J. C. Hinsey 1944. J. Neurophysiol. 7: 103.

8. Sanders, E. K,, & J. Z. Young 1944. J. Physiol. 103: 119.

9. Weiss, P., & A. C. Taylor

1944. J. exp. Zool. 95:233.

BERRY—HLXSEY: RECOVERY, REGENERATING NERVES 573

PLATE 5

574 ANNALS NEW YORK ACADEMY OF SCIENCES

Plate 5

Action potentials recorded from excised distal stumps of regenerating tibial nerves.

A was recorded after 36 days of regeneration with a monopolar electrode placed 3.5 cm. distal to the suture, and the conduction distance was 3.5 cm. The spike is approximately 25 microvolts, and the time signals under A are 1.7 millisec. per cycle.

B was recorded from a normal tibial nerve with a conduction distance of 8 cm., and B is a control record for C.

C is from the opposite regenerating tibial, 1363 days after suture with the same 8 cm. conduction distance, and the same amplification as B. The time line for B and C is 1 millisec. per cycle.

Annals N. Y. Ac\n. Sci.

Vol. XLVII, Aht. i, Platk

\^ '^"M^s^s^w^

BERRY AND HINSEY; RECOVERY OF CONDLTTION IN FIBERS

NERVE METABOLISM AND FUNCTION = A CRITIQUE OF THE ROLE OF ACETYLCHOLINE

By R. W. Gerard

Department of Physiology, The University of Chicago, Chicago, Illinois.

INTRODUCTION

Clearly, the acetylcholine system is the theme around which these papers have been arranged. The various hypotheses as to its functional significance, and especially the one regarding it as an essential com- ponent in conduction in the nerve fiber, have proven most fertile in re- search suggestions — witness the many studies here reported and the animated discussion of them. Yet, I must close with the judg- ment, on the basis of what has been said here, that this hypothesis has now exhausted its usefulness.

May I first offer, as evidence of my own long sympathy to the' view I shall shortly be dissecting, a quotation or two from my early writings?

"It remains to correlate this material [on heat and metabolism] with some actual mechanism of conduction. The current view that activity of one portion of a nerve fiber is the stimulus to the adjacent portion and so along the entire fiber has much to support it, especially in the form developed by Lillie. Recent evidence indicates that conduction itself may be analyzed into two phases occurring repeatedly in succes- sion. The first is an explosive type of chemical change in a portion of the membrane surrounding the nerve fiber, and it leads, probably by local potentials, to ion movements within the fiber, which constitute the second phase. Local concentration of ions against an adjacent por- tion of membrane initiates here the explosive change, and so on. Prob- ably the ion movements are associated with only a small fraction of the energy changes, and with the behavior of the membrane during and after conduction" (P- 499^).

"... In this way, it is obvious, a wave of electric and chemical change must spread along the nerve fiber in both directions from the point first stimulated. This is the nerve impulse, a propagated excitation. . . . Certain steps in this development are hypothetical, and it must be recognized that the picture has been simplified to a merest skeleton.

* This paper is essentially as presented on February 9, 1946. Later developments of any kind have not been introduced into the discussion.

(575)

576 ANNALS NEW YORK ACADEMY OF SCIENCES

Tlie action potential, for example, may not represent a passive de- polarization but a potential actively produced by the chemical reac- tions. But whatever the details, it is highly probable that the nerve impulse consists basically of a local membrane change of a chemical and physical nature, which leads to a flow of ions, or current, which in turn starts the local membrane change at adjacent points" (p. 64^).

"... Either the same kind of ion migration and chemical response which represents successive activation of one region of the nerve fibre by another must also take place at the synapse, or it is conceivable that the end of the axone acts as a miniature gland and, when stim- ulated, produces some chemical which is able to excite an adjacent or neighboring dendrite" (p. 74^) .

"In the nervous system itself, a similar mechanism has been consid- ered by several workers. The end of an axone is at least an unspecial- ized end organ, often a complicated one (as, for example, in the olfac- tory glomeruli) and might activate the dendrite or cell body on which it impinges via chemical as well as electrical changes. The transmis- sion from cell to cell by means of action potentials has long been the orthodox view, and emphasis on the chemical possibilities has had a novel flavor. In fact, however, the conduction along a nerve fibre involves excitation of a resting region by an active one, and both elec- trical and chemical components are present in the mechanism of propa- gation. At the ending, which is specialized, at least anatomically, either or both components might well be exaggerated to facilitate trans- mission over a critical region. Long-enduring action or depolarization potentials or special chemical accumulation might equally well be utilized in various situations and (except for familiarity with the one idea) one seems as likely as the other" (p. 546^) .

GENERAL BACKGROUND

The Role of Metabolism

Nerve fibers, like whole neurones or any other cells, depend on a maintained metabolism to survive and to function. This was strongly indicated when it was found,'* near the start of this century, that nerve conduction failed in the absence of oxygen; and was proved when nerve respiration and heat production, at rest and on activity, were successfully measured by several workers in the mid-twenties.'"*- ^ The next question is: For what result is metabolism essential? Or, What agencies link the chemical reactions with the physiological conse- quences? In general, the answer is clear enough: Metabolism liberates

GERARD: NERVE METABOLISM AND FUNCTION 577

energy to do necessary work, such as to maintain polarization across a leaky membrane, or it removes unwanted substances, or produces re- (lui)'ed ones. Further, the change in concentration of a substance may or may not be a needed step in the event of functioning, an indis- pensable gear in the cell machine.

In a particular case, it is often a teasing problem to determine just what role a metabolic event plays in a tissue's function. The formation of lactate in muscle contraction is a perfect illustration. When this relation was first established, early this century, lactic acid was at once assigned the key role of initiating shortening. It was an essential gear and, perhaps by changing surface tension due to acidity, engaged the shortening mechanism. Its removal or neutralization permitted re- laxation. Later, attention to energy balance emphasized that glycolysis could supply the energy required in anaerobic contraction and that this reaction was largely rewound with oxidative metabolism in oxygen. It was an easy assumption, then, even under aerobic conditions when no lactate change was found, that there was a rapid formation and destruction of this substance. Indeed, lactic acid was considered the essential link between metabolism, of which it was a necessaiy inter- mediate, and contraction, of which it was a necessary cause, and it was supposedly involved in both energetics and mechanics.

lodoacetic acid, alactic contractions, phosphocreatin and adenosine triphosphate changes, and, finally, the use of lipid fuels (not to mention myosin) , changed all that.*^ Muscle did not require lactic or any other acid to shorten it ; lactate is not part of the machinery. The CrP and ATP breakdown supplied the early energy needed for contraction, heat, and work; lactate formation is not an immediate energy source. Mod- erate exercise with good oxygenation involved no lactate change and little carbohydrate loss; lactate is in no way necessary to contraction. It is just one of the many initial or intermediate fuels available to the engine under normal working conditions, and its accumulation anaero- bically is, in a sense, a sign of failure to complete the initiated oxidations.

In a particular case, further, it is well to note that historical acci- dents greatly influence the trend of our scientific thought and research. Acetylcholine first came to attention as a pharmacologic agent; ATP, as an intracellular substance involved in important metabolic sequences. The great experimental sweeps were, accordingly, oriented differently in the two cases. Yet ATP also has profound pharmacological ac- tions,'- **• *• and ACh may well prove to be an important compon^ cell metabolic systems in general. This point will require later.

578 ANNALS NEW YORK ACADEMY OF SCIENCES

Neural Metabolism

Finally, I shall recall the gross metabolic picture of neural tissue. ^""^^ For nerve, in contrast to muscle, the heat and respira- tion balance of rest and action were first established. Lactic acid, next studied, seemed to be excluded from any role except as an anaerobic end-product, but it was later shown to serve as an effective substitute fuel when sugar oxidation was interfered with by iodoacetic acid; and anaerobic glycolysis was similarly established as a source of useful energy. Yet oxidations, by oxygen or an oxidizing reserve, re- main of especial importance in nerve, for the long and large delayed heat production cannot be dissociated from the initial heat by anoxia or by any other maneuver tried. What fuel or fuels are oxidized, is largely unknown. At rest, nerve may destroy more carbohydrate than could be fully oxidized, while the R. Q. hovers at 0.8; and CHO utiliza- tion can taper off to zero while O2 consumption is maintained unaltered. Even brain, with a resting R. Q. of 1.0 and a CHO fuel, can shift to another substrate which fully supports respiration. During activity, the much increased oxidative metabolism of nerve is not supported by CHO. The R. Q. of the extra respiration does rise to nearly 1.0 in tetanized nerves, but CHO loss is not increased. Some rise in acid- soluble phosphorus and in ammonia-liberation occurs, suggesting the degradation of phospholipins or phosphoproteins; but the problem is still wide open. The lipo-protein changes in rods on illumination (mentioned by Wald) and the Swedish work^^ on nucleoprotein de- crease in fatigued nerve (to which Schmitt called attention), are ob- servations challenging a resolution of this enduring uncertainty.

THE NERVE MACHINE

The resting metabolism of nerve is essential to keeping the tissue functionable. The normal resting potential, for example, falls when respiration is prevented^'' and even more rapidly when glycolysis is also blocked. ^^ When an action is evoked, the cell machinery whirs, physical and chemical changes occur, an impulse is propagated, and, finally, a cycle is completed and the machine fully reset. The events associated with activity are known in moderate detail, and it will be helpful to outline this sequence. Since so much attention has been given by investigators to the early and the electrical phenomena of response, may I emphasize that all the phenomena are closely coupled together. A single impulse, gone by in a millisecond, is yet irrevocably followed by a rise in heat liberation and in oxygen consumption which

GERARD: NERVE METABOLISM AND FUNCTION 579

endure for minutes. Conduction fails when respiration^' ^^ or gly- colysis^^ is disturbed, although, whether this is a result of interference directly with active metabolism, or is secondary to interference with resting metabolism, is not clear. •" A mechanism, partly in terms of phosphate intermediates, for insuring the one-to-one relation between early and late events, was suggested some time ago-° and is still useful. But, before pursuing this aspect, what of conduction itself?

Depolarization

Electric currents, applied to nerve or muscle, excite at the cathode, where ion movements are such as to depolarize the polarized mem- brane. The most direct evidence for the preexisting membrane poten- tial and for its diminution by trans-membrane currents is that from impaled single ncrve"^' '^ and muscle fibers. ^^ Membrane potentials up to £0 mV. have been obtained from resting units; and excitation is easily achieved with a cathode outside and anode inside the fiber, but even 100- fold greater currents in the reverse direction are ineffec- tive. Further, recalling the uniquely high sensitivity of these tissues to electric currents and the generation of electric changes when non- electric stimuli are applied, it seems probable that membrane depolar- ization by ion movements is the initial step in all forms of natural excitation of nerve and muscle. AVhether excitation results most di- rectly from a potential, impedance, or other, change, and to what crit- ical level, is a separate and secondary problem.

Active Membrane Participation

There is much evidence that the nerve membrane does not passively follow the imposed depolarization, at least when applied currents are more than a few per cent of threshold, but responds with active changes. These changes are almost certainly chemical as well as physical. The decreased impedance is suggestive, but perhaps not convincing, on this point. The existence of prepotentials (with depolarizing shocks, but not with equal ones in the reverse direction) in invertebrate^*- ^^ and vertebrate nerve^^ has been several times referred to in this publication. The fact that these often oscillate, and that the oscillations can incre- ment without additional external change,-* has been emphasized here by the report of Bronk and Brink, and by Cole's discussion. The oscil- lation period, 4 to 5 msec, observed in Ca-depleted nerves (Bronk), fits satisfactorily with the physical constants of the membrane, men- tioned by Curtis, which should lead to resonance at about 250 cycles per sec. But such physical factors control only the period of oscilla-

580 ANNALS NEW YORK ACADEMY OF SCIENCES

tion, and cannot supply the energy to maintain, even less to increase, it. Energy, presumably liberated from metabolic events, must supply the drive, while physical conditions only modulate its flow. The analogy used in connection with the similar problem of electric oscillations in the isolated frog brain-' may be used again. Air pressure, from the motor, drives most windshield wipers, and their beat does rise and fall with this; but the beat is much more under the control of a valve, which determines when each stroke is tripped off.

Another set of facts bespeaks, even more strongly, the intervention of a chemical step this early in the excitation process. In tortoise auricle,^^ crab nerve, 2*^ and even frog nerve--' (see also Gerard,^" and following discussion), an opposed electric shock, delivered between a supra-threshold shock and the start of the resulting propagated re- sponse, can nullify the response. While it may be possible for the purely physical changes, produced by a pulse in an appropriate network, to surge on to a peak after the pulse has passed (as Curtis suggested in connection with the reversed action potential), such a physical in- terpretation is under the burden of offering positive evidence in the case of the cooled auricle, where a reverse shock given 20 msec, after an effective one is still able to abort the response.

The Discontinuous Response

When the local membrane changes have progressed sufficiently, a full- fledged action appears and propagates. This response, as several speakers have emphasized, is not a continuation of the earlier processes, but a new and explosive group of events. Here, even more surely than in the preceding phase, chemical as well as physical changes are in- volved. The resting membrane potential shifts abruptly, not merely toward or to neutrality, but to an inverted magnitude which can much exceed the original level. ^^' "> ^^ Perhaps, as Curtis suggests, this is only a physical overshoot, rather than a newly-developed, oppositely- oriented, and chemically-active membrane battery; but the burden of proof seems to be clearly on the adherents to such a physical view. Cole's comment, that the reversed action potential can vary in mag- nitude independently of the resting potential, certainly favors more the positive conclusion. Hober's suggestion, that a fatty acid is re- leased by activated lecithinase and, reaching the inside of the mem- brane, reverses its potential, just as caproic acid does when placed on the outside, is an example of the chemical, active-membrane-change viewpoint. (This particular example is not fully satisfying, however; for, if the non-polar chains enter the membrane lipids and the polar

GERARD: NERVE METABOLISM AND FUNCTION 581

earboxyl groups form a negatively charged layer in the aqueous phase, this could shift the outer membrane surface charge from positive to negative, but could hardly shift the inner surface charge from nega- tive to positive.)

The well-known high temperature coefficients of excitation also speak, though admittedly in an uncertain voice, for chemical components in the process. If elongated molecules in a loose palisade in the mem- brane are merely bent about, during stimulation, then they must make quite a sudden fall when a threshold is reached, and must also start a vigorous series of changes. For, whether the main chemical reactions of metabolism accelerate during, or only after, the explosive membrane response, they are locked to it in an essentially invariable sequence. And, finally, the important and complex impedance, and potential, variations which accompany or follow the spike surely indicate proc- esses beyond simple ion movements or dielectric strains. The action potential spike represents more than a passive depolarization of a pre- viously charged membrane. It is an active physico-chemical process, still unknown in its details.

Local Currents

Whatever the events in an activated membrane region, there remains no doubt as to the mechanism of projjagation along a nerve or muscle fiber. Voltage differences between active and not-yet-active areas must lead to current flow between them and to catelectronic depolariza- tion of the latter. That such currents are a sufficient mechanism for propagation is certain from the experiments in which the nerve im- pulse is made to jump a block one or two millimeters Iqng.^^- ^'^' ^^ Even normally, propagation is probably by similar saltations from node to node, in medullated fibers.'*' '^^' ^®

Immediate Recovery

During the absolute refractory period, often under a millisecond, the membrane must at least recover toward its normal potential, impedance, and other properties, so that it is again activable. While the anodal action of the eddy currents sweeping on ahead may contribute to this restoration, this is obviously insufficient. Energy has been dissipated and must be made good from sources beyond the currents which help dissipate it. There can be no reasonable doubt that the complex of initial and immediately-subsequent recovery, with the reversing thresh- olds and potentials already well known,'' is dependent on one or more of the energj^-yielding metabolic reactions; perhaps on ATP breakdown.

582 ANNALS NEW YORK ACADEMY OF SCIENCES

Full Restoration

For completeness, although it is far removed from the direct problem of propagation, I mention, finally, the delayed recovery processes. We often forget that the increased respiration of activity persists a half- hour or more after a brief tetanus of nerve,^^ that the delayed heat production is similarly prolonged,^" and that considerable after-poten- tials may endure for comparable periods.''" And, as late recovery lags progressively further behind in a continuously-driven nerve, its re- sponse capacity falls to a lower equilibrium level. Irritability, velocity, chemical, thermal, and electrical response per impulse, etc., fall during a maintained tetanization.'*^

Now, with this outline of nerve action before us, I should like to con- sider the questions around which so much of this symposium has re- volved: (1) How does excitation engage metabolism; (2) what is the roh of the acetylcholine system; and (3) in what respects does junc- tional transmission differ from that in a fiber?

THE LINKAGE OF ACTION TO METABOLISM

Given the externally-applied stimulus energy, given even the propa- gated membrane response, the insistent question remains as to how one event induces the next and, especially, how chemical changes are made to follow the physical ones. This was asked by Grundfest, discussed by Green, and exemplified by Ochoa's contribution. It was considered for nerve, in some detail, a decade ago,^'' and is today being clarified in the case of muscle.*^ In muscle, the extra step of mechanical response offers both an additional problem, of how the membrane response leads to the myosin response, and an additional line of attack on the general case. Since metabolic details are far more numerous for muscle than for nerve, I shall choose illustrations freely also from the former material, in confidence that the principles they illustrate are equally valid for both tissues.

Ion Action

In the homogeneous liquid phase of a heterogeneous system like tis- sue, electric currents, applied as external stimuli or generated in the course of the active response, mean ion movements and only ion move- ments. Where these ion streams encounter interfaces — membranes, micelles, molecular palisades, etc. — ions can accumulate or decrease. A local change in ion concentration at molecular or structural surfaces of a cell must be the initial consequence of an electric stimulus and ap-

GERARD: NERVE METABOLISM AND FUNCTION 583

pears to be the only possible first link to a metabolic chain of events. (Electron shifts within single molecules or lattices, if such occur, would themselves follow the shift of charged ions, unless large electromagnetic fields were applied— fields that seem beyond a range of possible bio- logical significance.) However, changed ion concentration is easily sufficient to initiate other local chemical changes.

Altered metabolism means changes in the rates of chemical reactions. Not only are quantitative increases or decreases of total metabolism the sum of similar variations in the rates of the component reactions; but also qualitative changes are the resultant of increase in rates of certain reactions and decreases in others. The rate of a given reaction is determined by the concentration of active reactants and products and by the catalytic conditions (temperature, water, ions, etc.), especially by the enzyme activity. Reactant concentrations can change only as a result of an antecedent change in another chemical system which produces them — which gets us no further in our problem — or of a spatial redistribution. If reactants are themselves ions, and so moved by the stimulating currents, this could be a direct result of stimulation. In most cases, however, as emphasized by Hober, such a redistribution would also demand prior changes in the system to increase the physical availability, changes in membrane barriers, surface adsorption, and the like. These might also be a direct effect of the stimulus, as the rota- tion of a polar molecule, but are more likely to be secondary to more extensive chemical changes.

In contrast to the relatively unpromising situation for substrate al- teration, a modification of enzyme activity, and so of metabolism, by ion changes is both theoretically probable and experimentally estab- lished. Besides a direct ion effect on the activity of given enzyme molecules, there exist the other possibilities of activating pro-enzymes (Ca on prothrombin), removing inhibitors (phosphate or citrate bind- ing calcium), and adding accelerators (Cu on thiol oxidation*^). Such ion effects are richly present in biological systems, as well as in the non-living systems mentioned by Dr. Alexander, and it may be useful to itemize some that are important in muscle and nerve tissiies.^^' ^^

Magnesium ion either is essential to, or materially hastens, a number of key reactions in carbohydrate degradation, while local increase in its concentration would suffice to initiate or accelerate them. The phos- phorylation of glucose to hexose-6-phosphate by hexokinase, a reaction of especial importance in neural tissue which "prefers" glucose to glyco- gen as a fuel, requires Mg""*; as does, also, the shift of phosphate from the 1 to 6 position by glucophosphomutase.^'* The further phosphory-

584 ANNALS NEW YORK ACADEMY OF SCIENCES

lation of this substance, an intermediate in the glycogen as well as glucose reaction chain, to fructose 1,6-diphosphate, also requires Mg^*. The later change in phosphoglyceric acid to the energy-rich phos- phoenol-pyruvic acid involves a magnesium combination with enolase, and fluoride inhibition of this reaction depends on displacement of Mg^* by a fluoride complex. ^^ Magnesium or manganese, as well as calcium (K inhibits), is an essential component of the system which forms ATP and pyruvic acid from the phosphopyruvic acid and a lower adenosin phosphate.*" This ion is again reported necessary for the splitting of ATP to ADP by myosin,^' although most workers*^' *''• ^° find Mg"""^ inhibitory here. Mg*"" also inhibits the shift from 3-phos- phoglyceric acid to 2-phosphoglyceric acid by phosphotriose mutase.^^

Calcium ion, as mentioned, is required for the formation of ATP from phosphopyruvic acid. It is also involved in splitting a phosphate from ATP by myosin or other ATP-ases'*^' ^'^' -'^ and, perhaps, in the accompanying shortening of the myosin fibers. The splitting of acetyl- phosphate to acetate and phosphate is accelerated by Ca"^"^.^* And, again, to mention an inhibitory action, Ca""* interferes with the forma- tion of acetylcholine from choline. ^^

The formation of ATP is thus influenced by all three of the major cellular cations (Mg, Ca, K) ; its destruction, by at least two (Mg, Ca). ATP, in turn, is critical in both fat and carbohvdrate oxidation and may be one of the regulators of metabolism. Thus, a lowered ATP concentration might favor utilization of carbohydrate over fat'^*' and glycolysis over respiration.'^' The abrupt shift of muscle metabolism, on vigorous contraction, in just these directions, may. then, be due to the fall in ATP concentration and this, in turn, to the movement of ions to or from the critical enzyme surfaces.

Another example of ion importance, especially of K+, is offered by recent myosin studies. In an appropriate system, myosin B contracts in 0.1 M KCl and relaxes when the K"" is doubled in concentration^*; and an antagonism between K"^ and Ca*+, or Mg^^ and Ca""*, on myosin action as ATP-ase and on myosin extension, has been repeatedly noted. Potassium also increases the content of creatin phosphate in muscle (while Ca"^"" decreases it),^"' •^"' ^" as well as of stable phosphate esters in nerve,**^ perhaps by its ability to enhance CrP formation when pyruvic acid is oxidized.''- This ion also aids both the synthesis and the liberation of ACh and, conversely, ACh (or ATP) can release

Tr+ 63, .55, 64

It is not difficult to trace connections between these catalytic actions of the tissue cations and the physiologic effects which ion changes pro-

GERARD: NERVE METABOLISM AND FUNCTION 585

duce; but, at present, relating specific actions to specific effects would be mainly guesswork. It will suffice to recall that K^ increase leads to such effects as: a rise, passing into a severe fall, for the irritability, membrane potential, and electrotonic spread in nerve fibers, and, per- haps, for their oxygen consumption, as Brink mentioned; and a fall, from the start, for the spike (only slight), after-potential, conduction velocity, and recovery rate. Similar changes have been observed less fully in muscle and in the central nervous system: moderately in- creased K+, for example,^^' 2'' '^^ increases the fast electrical activity of cat or frog brain and prolongs the after-discharge on stimulation of deep cerebellar nuclei (see also ^^) . There are, thus, ample roads from current flow, through altered ion concentration and chemical reaction rates, to physiological responses. The problem is not to find connec- tions, but rather to identify the few important actualities among the many conceivable possibilities. This brings us back to acetylcholine.

THE ROLE OF THE ACETYLCHOLINE SYSTEM

That ACh is formed and destroyed as an integral part of impulse propagation in nerve fibers, has been suggested by several workers"^' '^^ and strongly supported by Nachmansohn.*'" He has summarized his arguments here : ACh is present in nerve and is released on stimulation (though its leaving the cell is accidental) ; cholinesterase (ChE) in nerve is highly active and specific, and choline acetylase (ChA) is also rich in nerve; a close parallelism exists, in the electric organ, be- tween potential and ChE activity; energy relations, in electric organ and nerve, are satisfactory for ACh synthesis via CrP, etc.; various drug actions, though always in danger of misinterpretation, especially where penetration through a membrane is involved, do support the im- portance of ACh in conduction. Just what the role of ACh is, seems less defined. Earlier, Nachmansohn supposed that the stimulus liber- ated ACh directly and that this caused the membrane depolarization. Now, recognizing that the stimulus itself must lead to depolarization, he suggests that ACh is responsible for the loss of resistance in the membrane — certainly, a step for which a chemical mechanism would be welcome. Beutner and Barnes have also emphasized a function for ACh, both in producing the action potential and in lowering mem- brane resistance.

Quantitative Relations

The calculations (Nachmansohn), that the ChE at a motor end- plate is powerful enough to split a complete layer of ACh in a milli-

586 ANNALS NEW YORK ACADEMY OF SCIENCES

second (perhaps even in a few microseconds, which Cole has mentioned as a more appropriate time for the impedance changes in nerve), and that the enzyme activity in the C.N.S. would split a layer covering 10^ sq. mu. per gm., are impressive. There is no question of strong esterase activity. However, it is worth noting, for the C.N.S., that even the surface of the nuclei in a gram of brain, let alone the whole neuronal surface, is some 4 X 10^° sq. mu.'^'' The difficulties appear when one examines the rest of the system, to see how well it can keep 'up with the esterase. Let us assume, with Nachmansohn, that some 2000 cal/M are required to esterify choline to ACh, and calculate, from his data on enzyme content, Feldberg's summary'^ of ACh content and liberation, and the figures of myself and others on heat and metab- olism, the over-all balance for nerve and brain.

Nachmansohn estimates that, in mammalian brain, ChE can split up to 10"^ molecules of ACh per millisecond per gram fresh tissue. This amounts to 6 millimoles per hour per gram. In terms of Qcit: values, reduced to these same units (mM, hr, gm.), less ACh could be split: between 0.3 mM for cortex, and 3.0 for caudate nucleus or sympathetic ganglia. For mammalian nerve, the rate calculates to 0.06; for frog nerve (20°), to 0.05; and for white matter, to 0.02. In contrast, the maximum rate of ACh synthesis (in tissue brei in N., with ATP and all necessary accessories) is 0.001 mM for mammalian brain and 0.0005 for nerve, still in these same units. In both mammalian brain and nerve, therefore, ChE activity is over 1000 times, perhaps over 5000 times, as great as ChA activity, and similar relations will probably be found for the frog. Neural enzymes can split ACh by three or four magnitudes faster than they can build it.

This calculation is made, of course, for maximum rates and over long time intervals, and requires further consideration. If, for example, the synthesis normally continues evenly in time, but the hydrolysis occurs only in brief bursts associated with activity, the discrepancy in rates might be unimportant. But this will not hold. First, whether ACh be associated with the potential or impedance changes of a nerve action, the rise is far more rapid than the fall, and the need for an explosive release of the agent is even more imperative than for an explosive de- struction. If, therefore, ACh is synthesized and destroyed in the course of each nerve action, ChA should actually be several-fold more active than ChE, instead of a thousand-fold less active. Let us make the more favorable assumption, however, that ACh need not actually be synthesized for each impulse, but only be released from a store. Then, though used in bursts, its formation could be continuous. Even so, there remain fatal discrepancies.

GERARD: NERVE METABOLISM AND FUNCTION 587

The total ACh present in whole brain is, keeping to millimoles per gram fresh tissue, about 2 X 10"^ for mammals and twice as much for the frog. White matter contains ten-fold less, but the value for mam- malian mixed nerve is close to 3 X 10'^. Dorsal roots contain, at most, one-twentieth of this amount; frog nerve,"' even less, 10~®. All the ACh in mammalian brain could, therefore, be destroyed by ChE in about 50 milliseconds, and would require a minute to be synthesized by ChA. For mammalian nerve, the stored ACh could last less than two seconds and would require over three minutes to replace ; for frog nerve, the ACh could last about 65 milliseconds, and for mammalian white matter (not to mention dorsal root), a third of a second. Yet nerve, including roots and central tracts, can maintain activity for hours, conducting hundreds of impulses per second ; and activity of the central grey can also long outlast the possible time limits. Clearly, then, neither ACh storage nor synthesis, nor both combined, could possibly (unless an entirely different order of ChA exists in vivo than has been found in extracts) supply this substrate as fast as ChE can split it.

Of course, an enzyme is not always kept saturated with substrate. However, this at once undermines the many arguments that have been made, from high ChE concentration, for the possibility of rapid rise and fall of ACh concentration; it also throws into question the significance of high local ChE concentrations. The striking finding, for example, that ChE is 15,000 times or more as concentrated in the end-plate re- gion as in the adjoining nerve or muscle, adds confusion rather than insight. The end-plate potential falls much less rapidly than that of nerve or muscle, and there is no evidence of a great store or synthesis of ACh there. How, then, can the tremendous ChE activity be recon- ciled with any current theories relating ACh to neural functioning? When a 2000 horse-power engine is found in a half-ton truck, one must suspect it is there for some other reason than to supply ordinary motive power.

Let us agree, however, that ChE is not kept fully saturated, and con- tinue with these calculations. The cat cervical sympathetic ganglion releases ACh to perfusing fluid, on preganglionic stimulation. Again in millimoles per gram, the ACh content of the ganglion, before or after several hours' tetanus, is about 10"^, although five times this amount has been released during the activity period. The rate of re- lease falls with continued activity, but holds up better when some blood is present. A maximum of 10"* is liberated in five minutes' tetanus at 17 per second or, per impulse, about 2 X 10~^ (cf. "• ^^). The ACh es-

588 ANNALS NEW YORK ACADEMY OF SCIENCES

caping from the ganglion, per impulse, is thus only about one-fiftieth of the amount that the ChE present could split in one millisecond. If ChA activity is taken to be one-thousandth that of ChE, the ACh synthesized in 60 ms. (the interval between impulses) could equal that released or exceed it two- or three-fold. This, incidentally, leaves no place for the often-assumed existence of a much greater ACh turnover within active units than is reflected in the amount escaping from them.

Isolated frog nerve, according to von Muralt,*^' actually increases its ACh content on tetanization, from lO"*' at rest to 1.5 X 10"*^ while ac- tive, and the ACh increase per impulse calculates to 6 X 10"^ from von Muralt's figures, to 10~" from Lissak's.^^ (In the latter experiments, only the ACh diffusing from the cut ends of a stimulated nerve was measured.) For cat gastrocnemius, assuming a weight of 20 grams, the ACh released by a single maximal twitch evoked by the nerve is 6 X 10"^^ mM/gm.^* Nerve can, of course, conduct several hundred impulses per second for long periods, but we might conservatively cal- culate with 50 per second, or 20 milliseconds total time available per impulse. Frog nerve ChE could split in this period 3 X 10"'^ mM/gm. of ACh: five times the amount von Muralt finds liberated and 30,000 times Lissak's figure. If, again, ChA is only one-thousandth as active, it could easily supply ACh at the rate demanded by Lissak but would fall short of von Muralt's figure by 100-fold. On the basis of such an analysis, a nerve should be able to conduct an impulse only once in two seconds. Von Muralt's value, incidentally, is far more in ac- cord with that for the ganglion, both in absolute amount and in rela- tion to ChE activity, and it is also more probably correct on method- ological grounds. But it cannot be right if the assumed ChA activity is remotely correct.

Perhaps, then, all these discrepancies result from falsely low ChA values. This enzyme system might easily have been seriously injured during tissue extraction and, thus, be far more active in vivo. Let us make this assumption, and allow a ChA activity sufficient to equal ChE activity or, giving ACh the most favorable conditions, an activity sufficient only to cover the ACh actually released on stimulation. Note, however, that even this excludes any greater ACh formation and sub- sequent destruction, within the cell or outside it, beyond the measured formation. If this greater turnover is allowed, by assuming ChA activity to equal ChE, the following relationships reveal still more in- tolerable discrepancies.

The formation of one millimole of ACh requires, we have agreed, some 2 calories. The sympathetic ganglion, releasing ACh on stimula-

GERARD: NERVE METABOLISM AND FUNCTION 589

tion at the rate of 10~^ mM/gra./hr., would liberate 2 X 10~^ cal./gm./hr. For frog nerve, at 50 impulses per second (and at this frequency the energy per impulse is fully 80% of that at zero frequency) , von Muralt's value gives 10"' mM ACh or 0.02 cal. (If ChE were working at full capacity, the heat liberated just from ACh splitting would be 6 cal. for the ganglion, 0.1 cal. for frog nerve!) But the measured total heat production of frog nerve is, in these units, 0.1 cal. at rest and 0.18 at maximal activity; for mammalian cortex (using the highest values of Q02 reported"") , the resting energy release is 25 cal. and that of maxi- mum activity perhaps 50 cal. These brain values are probably much too high for the ganglion (probably three-fold"^''), but this gives every ad- vantage to ACh. The actual ACh released in nerve during activity would thus, duj-ing its normal hydrolysis by ChE, account for over 10% of the total heat of nerve activity. Yet, only 3% of this heat is initial heat, immediately related to the events of conduction. Moreover, other exothermic reactions are surely involved, even with ACh itself — in its formation, liberation, neutralization, etc. — before that of its destruction. (And again, if ChE were fully active, the ACh hydrolysis heat alone would account for more than the full extra heat production of active nerve!)

An examination, further, of actual chemical reactions involved in the synthesis of ACh raises added difficulties. The initial energy'- source for ACh synthesis is considered to be CrP. During maximal frog nerve activity, less than 13 mgm. % of CrP is split in an hour; enough to account, at best, for 0.007 cal.^", far below the needs for ACh. But, of course, CrP is resynthesized by energy from other metabolic reac- tions, so this does not mean too much. The total fuel turnover, how- ever, does set an inescapable limit. For bullfrog nerve, 6 mgm. % of carbohydrate disappears per gram per hour at rest or activity ;''® for the small frog nerve, this might be 10 mgm. %, or 6 X 10~^ mM/gm./hr. On complete oxidation, this could yield a maximum of 0.02 mM of CrP, if all the energy available to form high-energ\' phosphate bands (3 per atom of oxygen) were so directed. Thus, the total nerve metabolism could just comfortably synthesize ACh at the rate it is reported actually to form during activity (.01 mM ACh from .02 CrP), and could not begin to supply energy to synthesize it at the rate ChE can destroy it. (Actually, the picture is worse than here presented, because the maxi- mum heat of activity is 0.18 cal./gm./hr. for frog nerve, whereas the assumed carbohydrate oxidation would yield 0.4.)

It may also deserve thought that, while the esterase is located in the membrane of the giant nerve fiber, the oxidizing enzyme systems are

590 ANNALS NEW YORK ACADEMY OF SCIENCES

distributed through its volume. It seems impossible that any con- siderable fraction of the oxidative energy released in the core of such a fiber could be utilized to drive reactions in its surface, up to 400 micra distant; and it seems unlikely even in the usual medullated fibers, up to 20 micra in diameter. Along the same lines, I know of no evi- dence for the intensive respiration at neuro-myal junctions which would be demanded to keep up with the terrific ChE activity. There is definite evidence against an intense respiration at syn- aptic regions in the central nervous system, despite their greater relative surfaces for fiber propagation and junctional transmission.'^'' (A high ChE and DPN concentration in the synaptic layers of the retina, however, has recently been reported.")

A final calculation, dealing with materials rather than energetics, is only suggestive. Most workers have tacitly or explicitly assumed that the acetate formed by hydrolysis of ACh was not re-utilized for ACh synthesis. Indeed, ACh has been found by Lipton"^ to form only from pyruvate in oxygen or an acetate source (acetoacetic or citric acids) in nitrogen, under present in vitro conditions. This would exclude full re-utilization of acetate, even in nitrogen, unless the reaction, 2 acetate -^ acetoacetate, is fully reversible. Lipmann has just indicated that synthesis from acetate may be possible when additional com- ponents, of coenzyme character, are added, and this would make easier a cyclic use of acetate. Without cyclic use, a molecule of glucose would have to be lost for every two of ACh formed and hydrolyzed — or twenty- fold the actual rate in nerve, 0.01 mM ACh; 0.0006 glucose. The accompanying heat production would have to be similarly outrageous, in comparison with the factual rate.

Such quantitative considerations are admittedly rough, with little attention to detailed conditions (temperature, species, rate of stimula- tion, etc.), but the order of magnitude cannot be far off. They demon- strate conclusively, I believe, that ChE cannot possibly exert its full activity on ACh in neural tissues and suggest that other meanings for its presence and action be sought. Further, even the less drastic rates and amounts reported for other phases of an ACh system lead to severe quantitative strains on the total metabolism of nerve or brain. But still other difficulties have been brought out in this publication.

Drug Action

Drugs, especially esterase inhibitors, have been widely used in study- ing the ACh system, and their actions have been much discussed dur- ing this symposium. The point of greatest debate has been the ques-

GERARD: NERVE METABOLISM AND FUNCTION 591

tion of permeability; for, of course, the absence of an expected effect in vivo could easily be due to a failure of the added substance to pene- trate to the vulnerable region. However, the evidence marshalled seems to be conclusive that esterase can be inactivated, or ACh con- tent increased, without serious disturbance of function of nerve or muscle.

All agree that eserine, a tertiary amine, can enter nerve and muscle, and Nachmansohn makes the point that the action potential of squid nerve can be abohshed by soaking in this drug. (The fall of the ac- tion potential does not show the great prolongation one might expect if ACh removal were interfered with.) Yet Cantoni and Loewi have re- ported"^ that a frog can be eserinized in vivo so that nerve ChE activity is abolished, while nerve conduction remains undisturbed. (Con- versely, intravenous ChE blocks the pupillary reflex in rats.*°) A comparable result with the even more powerful, and irreversible, in- hibitor, diisopropyl-fluorophosphate, has just been presented by Oilman and by Bodansky. Both in vivo and in vitro, though with some anomalies in behavior, this agent has been shown to inactivate entirely ChE while leaving nerve conduction and action potentials intact. Al- though detailed criticisms have been made, especially by Talbot, the major fact remains, as in the eserine experiments, that conduction with- out esterase is possible.* Again, veratrm can inhibit ChE,^^ yet it does not^- influence muscle or the neuromyal junction, including its sensitiv- ity to added ACh, except for a late and independent negativity.

The inability of ACh, added in large concentration to the surround- ing medium, to depolarize nerve or otherwise to disturb conduction, has been reemphasized by the new experiments of Lorente de No and of Bronk. Nachmansohn has urged that ACh, a quarternary ion, cannot penetrate the lipoid membrane of nerve fibers, except at their naked terminals, thus accounting for these negative results. Yet ACh does leave nerve trunks on stimulation and should, similarly, be able to enter under combined anoxia and stimulation. Further, both Bronk and Atcheson have presented clear evidence that tetraethylammonium

* The results of Gilman and the Edgewood workers have since been challenged by Nachman- sohn and his colleagues. Both groups reported work at the April meeting of the Federation, and their full papers have since appeared (J. Neurophysiol. June). Work done in the interval in my laboratory fully supports the conclusions of the Edgewood group.

Frog sciatics were immersed in peanut oil, with or without DFP, resting on stimulating and lead-off electrodes. Action potentials fail in a few minutes or remain normal for hours, depending on the drug concentration. Conduction, when lost, is not restored in fresh oil. A nerve exposed for an hour to a non-depressing concentration of DFP, washed, ground, and assayed for cholin- esterase by its rate of destruction of added acetylcholine (tested on the frog's rectus), shows no cholinesterase activity. A companion nerve continues to conduct well, while remaining in the same DFP solution. Further, when a washed, poisoned nerve is ground together with an untreated one, the homogenate assays at the average cholinesterase activity of the two nerves taken separately. The DFP inactivation of cholinesterase occurs, therefore, prior to the grinding. Clearly, conduc- tion is possible in nerve lacking cholinesterase.

592 ANNALS NEW YORK ACADEMY OF SCIENCES

chloride, another quarternary, acts powerfully on medullated nerve and so, presumably, penetrates easily.

A seemingly crucial test has been carried out in the last few weeks in our laboratory by Miss Graham. ACh Br (1 : 1000 in isotonic KCl with vital red) was injected into single muscle fibers of the eserinized frog sartorius with a micropipette, and the membrane potential meas- ured. With one electrode inside the fiber and another outside, mem- brane potentials of 40 to 80 mV are regularly obtained. Injection of a large drop of isotonic KCl with or without ACh, large enough to fill the fiber cross-section and spread one to three diameters along its length, will immediately lower the potential by one- to two-thirds; but a smaller drop, not filling the entire cross-section, has little effect; only 7% fall, in one fully satisfactory experiment with ACh. It seems, then, that ACh inside the membrane does not depolarize it, as postulated by Nachmansohn and by Beutner and Barnes.

A number of other points have been raised here, which must also be kept in mind. Bodansky has emphasized the existence of a family of esterases, even in different neural structures of the same species: e.g., the enzymes in cervical sympathetic ganglion and in brain show different substrate and ion concentration optima, different equations relating concentration to activity, different substrate selectivities, etc. The last point is especially important, since a criterion urged for discrim- inating between "true" and "pseudo" esterase is the relative inactivity of the true enzyme with tributyrin. Yet the "true" esterase of brain splits triacetin up to six times faster than it does ACh.

The distribution of ChE or ACh, or both, in various organs and tis- sues has also been mentioned by several discussants. Rosenblueth asked about conduction in adrenergic nerve fibers, which lack the ACh system ; Hoagland made a similar point about Nitella, which conducts independently of ACh; another discussant mentioned a recent report that ChE is absent in the electric organ of Malaptorurus ; and the caro- tid body, although specifically sensitive to ACh, is reported to lack ChE.^^ ChE is also absent from the iris sphincter of the amphibian eye, while present in its cornea and in the turtle's sphincter.^* Con- versely, parts of the ACh system are richly present in spleen, placenta, cornea, potatoes, and some bacteria, where any relation to neural func- tion is nearly, or quite, impossible. And finally, in this connection, many other agents act on, and other enzymes are present in, neurones. Adrenalin keeps up the action potential in isolated cat nerve ;^^ ATP stimulates smooth, as well as striped, muscle ;^^ carbonic anhydrase is interestingly distributed in the brain ;^^ CO2 has marked and differential

GERARD: NERVE METABOLISM AND FUNCTION 593

actions on the nervous system; and so on. Thiamin is reported^^"^^ to affect ACh action and synthesis and to be hberated from pre- cursor in relatively large amounts from stimulated frog nerve even to be the transmitter. Cocarboxylase, like ChE, is concentrated in the nerve membrane.®- (For discussion of further recent evidence, see Gerard and Libet.") I do not see how we can reasonably select the ACh system from all this welter and just assign to it an essential role in conduction of the nerve impulse.

THE PROBLEM OF JUNCTIONAL TRANSMISSION

In the time available, the problem of junctional transmission, pre- sented mainly by Eccles, can only be touched upon, and even so the case of autonomic effectors, the classical one of neurohumoral action, which has not been before us, will be omitted. As for the neuro-myal junction, the unquestioned facts, that ChE is more concentrated there than elsewhere in the muscle fiber by a factor of 10* (Nachmansohn), and that this region is more sensitive to added ACh by a similar factor (Kuffler), are impressive; along with the potentiating and prolonging action of eserine, long known for junction as well as ganglion. I am, personally, less convinced of a transmitter role of ACh at the junction than I was a few years back, but do not consider that the evidence is crucial in either direction. (The observation®^ that the lizard muscle fiber can respond to nerve stimulation at a time when ACh applied to the end-plate is ineffective, although such ACh does cause contrac- tion when first administered, has not been explained. Also, the end- plate potential, often supposed to be set up by ACh liberation, is pres- ent in invertebrate, as in vertebrate, muscle; but the end-plate region in the former is not sensitive to ACh or to curare.®*) Discussion here has been mostly on central synapses, and the reader should consider these.

As elaborated by Eccles, the currents that flow between an active fiber region and an inactive one, whether in the same or another unit, do account for the usual activation phenomena. The results of Ar- vanitaki-* and of many other recent experimenters®^' ®®' ®^ show that threshold changes and transmission from unit to unit in simple sys- tems are accurately and quantitatively explicable in terms of the meas- ured currents and the known geometry. Whether Eccles' detailed anal- ysis of the situation at a synapse will hold up as well with time as he was able to defend it here, we do not know, but there is every reason to push such thinking further. (Some difficulties are: the very variable structures which are found in synapses, where the two units may meet as

594 ANNALS NEW YORK ACADEMY OF SCIENCES

parallel fibers, spirals of one on another, multiple contacts, etc., as well as the orthodox small end-foot stuck on a large surface, like a match stick on a cheese; the problem of multiple synapses on a cell body and the relative area of end-feet and their surround; the presence of ir- reciprocal conduction with protoplasmic continuity, as in an asymmet- rically compressed sartorius muscle. Neither these, nor the spatial theory of inhibition, nor the need for regarding the E. E. G. as an oscil- lating somatic potential, problems which have received attention in this publication and elsewhere,^* can be here expanded.)

The evidence for a transmitter role of ACh in the central nervous system, on the contrary, is inferential and conflicting. Those who have read Feldberg's recent review of this question'^ must have been impressed by the poor case that can be made. Added eserine, or ACh, or both, may increase the activity of a brain region, or depress it, or cause negligible change. The failure of ACh to alter frog cord re- flexes, mentioned by Eccles, is a case in point. Or ACh may excite, while eserine depresses. Atropine, on the whole, does nothing. The two compounds mentioned by Oilman, both powerful anti-esterases and both able to produce convulsions, one of which is completely antidoted by atropine, while the other is uninfluenced by it, afford an instance of the conflicting facts in this area. Strychnine is supposed to exert its action by blocking ChE, yet Tobias^^ has found the ACh content of frog and rat brains and cords decreased, if anything, by strychnine. Nembutal, conversely, increases the ACh content, although, as Bo- dansky mentioned, it also lowers ChE activity. If ACh is an agent for evoking neurone activity, it should increase the oxygen consump- tion of brain. Lipton has recently made Q02 measurements on rat brain slices, at my request, and found no influence of eserine (10"^) alone with eserine and ACh (10"^), at most a better maintenance of the usual initial rate. Incidentally, the only other observations I have found on the influence of ACh on respiration are one"° showing an increase in salivary gland oxygen consumption, and a forgotten one from my own laboratory^^^ showing a marked decrease in the oxygen consumption of nerve (uneserinized).

The best basis for invoking chemicals in synaptic transmission is that synaptic potentials, like those of the end-plate, may last much longer than could any reasonable physical discharge period for mem- branes with capacitances and resistances in the range known. Then one invokes some active depolarization process, as for nerve; and then this must be explained, by a chemical reaction of some sort. To be sure, chemical activity is involved, as in nerve, and quantitatively more in-

GERARD: NERVE METABOLISM AND FUNCTION 595

tense. But is the chemical ACh? Quien sabef It should be recalled that undrugged nerve also has an enduring after-potential, which can increase in intensity for several minutes and persist for ten or more.

40

CONCLUSION

Dr. Nachmansohn skillfully and generously organized the extraor- dinarily successful conference of which this is the result, to bring forth much current evidence and a full range of judgments bearing on the significance of ACh, as well as of electrical changes, for the functioning of nerve and other tissues. With these facts and arguments before us, we must conclude that ACh is not critically involved in nerve conduc- tion, and we must be reserved in assigning it a role in junctional trans- mission, particularly within the nervous system. This is progress and should lead to greater progress. Our thinking and our consequent ex- perimentation now can be directed along new lines.

This is not to say that the hypotheses which must be relinquished have been worthless, nor that the ACh system is unimportant. Hy- potheses are not true or false (who can assert absolute truth?) ; they are useful or useless. They do or do not suggest investigations which reveal new facts, facts which discriminate between alternate views or which fill in gaps of felt ignorance or which suggest new interpretations and experiments. By such standards, the various ACh hypotheses have been good ; they have been abundantly fruitful. But this fruit is ripe, and it is time for the seed of a new idea to be germinated. Fresh fruit will then ripen with time and the present crop not be husbanded until it rots or dries up.

What a new and fertile approach may be, I do not know. ACh and the enzymes that operate in the system can hardly be present adven- titiously. Nature no more evolved the ACh system to mislead bio- chemists than it evolved the giant nerve fiber to aid physiologists. ACh has some significance to cells. Perhaps this system is a fragment of a universally important metabolic mechanism, dealing, if one must hazard a particular guess, with the manipulation of lipid molecules. Such facts or statements as the following maj'- serve as clues. ACh prevents the splitting of CrP by muscle juice ;^°- choline lack increases the turnover of phosphohpids;^°^ ACh can replace Ca in enabhng myosin to split ^'pp.52 ^Qi^ jg |.j-jg Qj^jy system able to capture energy via both respira- tory and glycolytic reactions,^^ and so is related to both respiration and carbohydrate utilization rates. ^°*' "^ It would still be possible for evolution to have selected this fragment of a more general system for special emphasis and functioning in particular situations; to serve, for

596 ^A^A^.4L^ NEW YORK ACADEMY OF SCIENCES

example, as a transmitter at parasympathetic endings. After all, ACh is an ion with rather striking physico-chemical properties. In just such fashion have the ubiquitous respiratory hemins of cells been se- lected for the special function of transporting oxygen. The parallel evolution of hemoglobin in utterly separate phyla, as the vertebrates and annelids, was a great mystery before the discovery that such re- lated molecules as cytochrome are almost universally present in cells. I am suggesting, then, that ACh may extend further and have more importance in cell functioning than has yet been seriously considered and that any particular role it plays in transmission is a secondary and derivative one. In arguing, as I have, for renouncing the belief that ACh has any direct function in nerve conduction and in transmission at many junctions, I am inviting those who work with the ACh sys- tem to emerge from the chrysalis which they have outgrown and to seek fresher and greater fields of intellectual nourishment.

BIBLIOGRAPHY

1. Gerard, R. W.

1927. Science. 66: 495.

2. Gerard, R. W.

1931. Quart. Rev. Biol. 6: 59.

3. Gerard, R. W.

1932. Physiol. Rev. 12: 469.

4. Fillie, H.

1908. Zeitsch. f. d. AUgem. Physiol. 8: 492.

5. Feng, T. P.

1936. Ergeb. Physiol. 38: 73.

6. Wilhelmi, A. E.

1946. In: Fulton, Howell's Textbook of Physiology. See p. 56. Saunders. Philadelphia.

7. Drury, A. N.

1932. J. Physiol. 74: 147.

8. Green, H. N., & H. B. Storer 1944. Brit. J. Exp. Path. 25: 150.

9. Buchtal, F., A. Deutsch, & G. G. Knappsrs 1944. Nature. 153: 774.

10. Gerard, R. W., & N. Tupikova

1938. J. Cell. Comp. Physiol. 12: 325.

11. Quastel, J. H.

1939. Physiol. Rev. 19: 135. "^

12. Garard, R. W.

1937. Ann. Rev. Biochem. 6: 419.

13. Gerard, R. W.

1938. Arch. Neurol. & Psych. (Chicago). 40: 985.

14. Gerard, R. W., & B. Libet

.1946. Progress in Neuropsychology and Psychiatry. New York.

15. Hyden, H.

1943. Acta Physiol. Scand. 6: Suppl. 17.

GERARD: NERVE METABOLISM AND FUNCTION 597

16. Gerard, R. W.

1930. J. Physiol. 92:498.

17. Shanes, A. M., & D. E. S. Brown

1942. .1. Oil. C;omp. Physiol. 19: I.

18. Ronzoni, £.

19.31. .1. Biol. Chem. 92: iii.

19. Chang, T. H., & R. W. Gerard

1933. J. Physiol. 104: 291.

20. Gerard, R. W.

1934. Science. 79: Suppl. 20.

21. Curtis, H. J., & K. S. Cole

1940. .J. CxiW. Comp. Physiol. 15: 147.

22. Hodgekin, A. L., & A. F. Huxley

194.5. .1. Phy,siol. 104: 17(1

23. Graham, J., G. R. Carlson, & R. W. Gerard

1942. Fed. Proc. 1:29.

24. Arvanitaki, A. J.

1942. .J. Neurophys. 5: 89. 26. Hodgekin, A. L.

1938. Proc. Hoy. Soc. Loridou H 126: 87.

26. Katz, B.

1939. Electrical Excitation of Xerve. Oxford Univ. Press. London.

27. Libet, B., & R. W. Gerard 1939. J. Neurophys. 2: 1.53.

28. CoUe, J.

1934. Compt. rend. Soc. hiol. 117: 99.

29. Gerard, R. W.

1934. J. Physiol. 83:24.

30. Gerard, R. W.

193fi. Cold Spring Harbor Symposia. 8: 194.

31. Tasaki, I.

1939. .J. Physiol. 127:211.

32. Hodgekin, A. L.

1937. .1. Physiol. 90: 183, 211.

33. Blair, E. A., & J. Erlanger 1939. .J. Physiol. 126:97.

34. Lillie, R. S.

1925. .1. Gen. Physiol. 7: 473.

35. Marshall, W. H., & R. W. Gerard 1933. .J. Physiol. 104:. 58(J.

36. Hursh, J. B.

1939. .J. Physiol. 127: 131.

37. Gasser, H. S.

1937. In: Gasser & Erlanger, Electrical Signs of Nervous Activity. Univ. of Pennsylvania Press. Philadelphia.

38. Gerard, R. W.

1927. J. Physiol. 82:381.

39. Hill, A. V.

1933. Proc. Roy. Soc. London B 113: 34.5.

40. Gerard, R. W.

1930. .J. Physiol. 93:337.

41. Gerard, R. W., & W. H. Marshall 1933. J. Physiol. 104:. 53.5.

42. Kuffler, S. W.

1946. Ann. N. Y. Acad. Sci. 47 (.5). (In Press.)

598 ANNALS NEW YORK ACADEMY OF SCIENCES

43. Kharasch, M. R., R. R. Legault, A. B. Wilder, & R. W. Gerard

1936. J. Biol. Cheni. 113: 537, 557.

44. Colowick, S. P., & E. W. Sutherland

1942. J. Biol. C'hem. 144: 423.

45. Warburg, O., & W. Christian 1942. Biochem. Z. 310:384.

46. Boyer, P. D., H. A. Lardy, & P. H. Phillips 1942. .J. Biol. Chem. 146: 675.

47. Bailey, K.

1942. Biochem. J. 36: 121.

48. Engelhardt, W. A., & M. N. Lyubimova

1942. Biochimia. 7: 205.

49. DuBois, K. P., H. G. Albaum, & V. R. Potter

1943. .1. B'nA. C'hem. 147: 699.

50. Greville, G. D., & H. Schwann 1943. Nature. 152: 81.

51. Utter, M. F., & C. H. Workman

1942. J. Biol. Chem. 146:289.

52. DuBois, K. P., & V. R. Potter

1943. J. Biol. Chem. 148: 451.

53. DuBois, K. P., & V. R. Potter

1943. J. Biol. Chem. 150: 185.

54. Lipmann, F., & L. J. Tuttle

1944. J. Biol. Chem. 153: 571.

55. Feldberg, W., & T. Mann

1945. J. Physiol. 104: 8.

56. Lardy, H. A., & C. A. Elvehjem 1945. Ann. Rev. Biochem. 14: 1.

57. Evans, E. A,

1944. Ami. Rev. Biochem. 13: 187.

58. Szent-Gyorgyi, A.

1945. Acta Physiol. Scand. 9: Suppl. 25.

59. Hegenauer, A. H., W. O. Fenn, & D. M. Cobb

1934. J. Cell. Comp. Physiol. 4: .505.

60. Tipton, S. R.

1936. J. Cell. Comp. Physiol. 7: 433.

61. Gerard, R. W., & N. Tupikova

1939. J. Cell. Comp. Phy.siol. 13: 1.

62. Boyer, P. D., H. A, Lardy, & P. H. Phillips

1943. J. Biol. Chem. 149: .529.

63. Schallek, W.

1945. J. Cell. Comp. Physiol. 26: 15.

64. Montigel, C.

1943. Helv. Physiol. Pharm. Acta. 1: c. 47.

65. Dubner, H. N., & R. W. Gerard

1939. J. Neurophysiol. 2: 142.

66. Fenn, W. O.

1940. Physiol. Rev. 20: 377.

67. von Muralt, A.

1942. PHlig. Arch. 245: 604.

68. Lissak, K.

1939. J. Physiol. 127:263.

69. Nachmansohn, D.

1945. Vitamins & Hormones. 3: 337.

70. Pearce, J., & R. W. Gerard

1942. .1. Physiol. 136:49.

GERARD: NERVE METABOLISM AND FUNCTION 599

71. Feldberg, W.

1945. Physiol. Rev. 26: .596.

72. Brown, G. L., & W. Feldberg 1936. .J. Physiol. 88:26.5.

73. Mackintosh, F. C. 1938. J. Physiol. 94: 155.

74. Brown, G. L., H. H. Dale, & W. Feldberg

1936. J. Physiol. 87:394.

75. Gerard, R. W.

1937. Proc. Assoc. Res. Nerv. & Ment. Dis. 18: 316.

76. Holmes, E. G., R. W. Gerard, & E. I. Solomon 1930. .J. Physiol. 93:342.

76a. Tobias, J. M., D. B. Clark, & R. W. Gerard

1942. Fed. Proc. 1:85.

77. Anfinsen, C. B.

1944. J. Biol. Chem. 152: 267, 285.

78. Lipton, M. A. Unpublished data.

79. Cantoni, G. L., & O. Loewi

1944. J. Pharm. Exp. Therap. 81: 67.

80. Mendel, B., & R. D. Hawkins

1943. J. Neurophysiol. 6: 431.

81. Nachmansohn, D., & H. Schnermann

1945. J. Biol. Chem. 159: 239.

82. Kuffler, S. W.

1945. .J. Neurophysiol. 8: 113.

83. Hollinshead, W. B., & C. H. Sawyer 1945. Anat. Rec. 91: 282.

84. Armstrong, P. B.

1943. J. Cell. Comp. Physiol. 22: 1.

85. Bulbring, E., & D. Whitteridge

1941. J. Physiol. 99:201.

86. Buchtal, F., & G. Kahlson

1944. Acta Physiol. Scand. 8: 317, 325.

87. Ashby, W.

1944. J. Biol. Chem. 156: 323. 331.

88. von Muralt, A.

1945. E.xperientia. 1(5): 1.

89. Torda, C, H. Wolff

1944. Proc. Soc. Exp. Biol. & Med. 53: 88.

90. Liechti, A., A. von Muralt, & M. Reinert

1943. Helv. Physiol. Acta. 1: 79.

91. Wyss, F.

1944. Helv. Physiol. Acta. 2: 121.

92. Nachmansohn, D., & H. B. Steinbach

1942. ,J. Neurophysiol. 5: 109.

93. Buchtal, F., & J. Lindhard 1937. J. Physiol. 90:82.

94. Katz, B., & S. W. Kuffler Personal Communication.

95. Marrazzi, A. S., & R. Lorente de No 1944. J. Neurophysiol. 7: 83.

96. Rosenblueth, A.

1944. J. Physiol. 141 : 196.

97. Skoglund, C. R.

1945. J. Neurophysiol. 8: 365. 377.

600 ANNALS NEW YORK ACADEMY OF SCIENCES

98. Gerard, R. W.

1941. Ohio J. Sci. 41: 160.

99. Tobias, J. M., M. A. Lipton, & A. Lepinat

1946. Proc. Soc. Exp. Biol. & Med. 61: 51.

100. Brock, N., H. Druckrey, & W. Loch

1943. Biochem. Z. 313 : 300.

101. Shaffer, M., T. H. Chang, & R. W. Gerard 193.^). J. Physiol. 111:697.

102. Abdon, N. O., & N. A. Nielsen 1937. Skand. Arch. Physiol. 77: 13.

103. Patterson, J. M., N. B. Keevil, & E, W. McHenry

1944. J. Biol. Chem. 153: 489.

104. Welch, J. H., & J. E. Hyde

1944. J. Neurophysiol. 7:41.

106. Sanz, M. C.

1943. Pflus. Arch. 246: 596.

CONCLUSION

Remarks Made at a Dinner in the Hotel Astor, New York, after the

Conference on The Mechanism of Nerve Activity, Sponsored by

The New York Academy of Sciences, February 9, 1946.

By J. F. Fulton

Sterling Frojessor of Physiology, Yale University School of Medicine,

New Haven, Connecticut

No set speeches have been planned for this evening, since those who arranged the Conference wished to keep our proceedings entirely in- formal. The Committee has asked me to express our most sincere thanks to The New York Academy of Sciences and, particularly, to Mrs. Miner and her gracious assistants, who have done so much, both before and during the Conference, to make it outstandingly successful.

In this connection, I must also mention the man — and I do this without instructions from the Committee — who originally conceived the idea of having the Conference and who, with Tracy Putnam's ener- getic backing, has been so largely responsible for working out the de- tails. David Nachmansohn came to the United States, in the summer of 1939, under the sponsorship of the Dazian Foundation, and since this is something of a family party, I will, perhaps, be forgiven for telling you a family secret. The Dazian Foundation had wished to sponsor a physiologist from Europe. David Nachmansohn was chosen, and I can only say that American Physiology has been vastly stimulated by his presence in this country. He and his wife have made a solid place for themselves here; and, in the language of George Eliot, David, "through his mild persistence, has urged Man's thoughts to vaster is- sues."

We feel particularly fortunate in being able to welcome so many dis- tinguished colleagues from abroad, this having been made possible by the vision of the Rockefeller Foundation and of the Commission for Relief in Belgium. Our colleagues from France bring us heartening news of the revival of their laboratories and of their faith in the uni- versal fellowship of scientific men. We are also happy to see Pro- fessor Augusto Pi-Sufier of Barcelona and Caracas, and his son, Dr. Jaime Pi-Sufier. Also, Arturo Rosenblueth from Mexico. In Doctors Hober and Michealis, we have distinguished representatives of the

(601)

602 ANNALS NEW YORK ACADEMY OF SCIENCES

highest traditions of German science, men who could never surrender their faith in academic freedom.

Frederic Bremer exemphfies all that we most admire in his country- men: loyalty, self-reliance, humor, industry, and, with it all, a burning zeal for research that sustained him in his vigorous way of life during the lean years through which he has just passed. AVhen conditions in his university laboratory made work impossible in 1943, he retired to the cellar of his house for nearly two years, and there continued his experimentation and his writing. Virtue cannot be enhanced by calling it to public notice and I do so now, not to add luster to Frederic Bremer's position in the world of science, but rather that others may take inspiration from his faith and his example. He states modestly that he has merely followed in the pathways of those he seeks to emulate. Many of us here share with him a common devotion to two of his masters: one was Harvey Gushing; the other, Sir Gharles Sher- rington. Sir Gharles, whom Bremer has recently seen, is now in his ninth decade. Bremer found him hard at work, bringing out a biog- raphy of Jean Fernel, the sixteenth century physician and humanist; while he is somewhat crippled by arthritis, his mind remains ever vigorous.

Another pupil of Sherrington is John Eccles, whose industry, like that of Bremer's, is phenomenal. In a very short space of time, he has had eight children, and, not content with bringing up a large family at home, he also created a laboratory family of loyal associates: Hebbel Hoff and David Lloyd who were his pupils at Oxford, and Stephen Kuffler (whom you have all enjoyed hearing at the Gonference), Ber- nard Katz, and many younger men whose names we are beginning to see in the literature. Ghandler Brooks of Baltimore permits me to tell you that he, too, is going presently to New Zealand, to experience for a year the stimulating atmosphere of Eccles' laboratory. Gharacter- istic of the Eccles family, Mrs. Eccles has extended a cordial invitation to Dr. and Mrs. Brooks to live with them while they are in Dunedin, should they have difficulty in finding accommodations. I could tell you more about Jack Eccles, but since he is a good friend of mine I shall spare him, the more so since he knows much too much about me to make it safe to indulge in blackmail.

My pleasant duty in closing is to propose a toast. Since we did not wish to obligate anyone to speak, it seemed inappropriate to single out our guests, for they have been one with the Gonference. But it has seemed highly appropriate to ask you to drink a standing toast to the man who has probably influenced our thinking more profoundly than anyone now living — Sir Gharles Sherrington.