presumably proceeds until the impulse reaches the nerve terminals. Here acetylcholine (ACh) is liberated (Dale, Feldberg & Vogt, 1936) and, by a

Transcription

1 73 J. Physiol. (I952) i i8, THE EFFECT OF SODIUM IONS ON NEUROMUSCULAR TRANSMISSION BY P. FATT AND B. KATZ From the Biophysics and Physiology Departments, University College, London (Received 4 February 1952) During the passage of an impulse in a nerve or muscle fibre, sodium enters the fibre and potassium is subsequently released in equivalent amount (Hodgkin, 1951). When an impulse travels along a motor axon, this cation exchange presumably proceeds until the impulse reaches the nerve terminals. Here acetylcholine (ACh) is liberated (Dale, Feldberg & Vogt, 1936) and, by a specific action on the motor end-plate, locally short-circuits the muscle membrane (Fatt & Katz, 1951). In this way, a propagated wave is initiated in the muscle fibre, involving a renewed sodium/potassium exchange of the same type as found in the nerve axon (Nastuk & Hodgkin, 1950). The mechanism by which ACh, rather than-or in addition to-potassium, is released from motor nerve endings remains unknown. A simple working hypothesis would be to invoke a mechanism in the terminal axon membrane similar to that in the nerve or muscle fibre: that is, a type of electric membrane activity whose rising phase is due to the entry of sodium ions into the terminal, but whose return is brought about by an equivalent efflux of ACh+ ions from the interior of the nerve endings. According to this concept, the essential difference between the terminal membrane of the motor nerve and the ordinary axon membrane would be that a depolarization is followed, after a transient rise of sodium flux, by a specific rise of ACh+, rather than K+, permeability. A crucial test of this hypothesis would require an accurate measurement of the amounts of sodium and acetylchollne which flow across motor nerve terminals during an impulse. While this seems at present impracticable, there are other experiments which are suggested by this hypothesis. It would be expected, for instance, that the amount of ACh released during the impulse, and therefore the size of the end-plate potential (e.p.p.), should be related to the concentration of sodium ions in the extracellular fluid. The object of this paper is to provide quantitative information on the effect of sodium ions during neuromuscular transmission. It can be shown that the

2 74 P. FATT AND B. KATZ size of the e.p.p. at a single nerve-muscle junction depends upon the sodium concentration in the surrounding fluid, but these changes might be ascribed to an effect of sodium on end-plate receptors rather than on ACh release from nerve endings. We have attempted to distinguish between these factors by combining two methods: (i) measuring the e.p.p. (in response to a nerve impulse) in various ionic environments, and (ii) measuring the response of the end-plate to applied ACh under the same conditions. This is an indirect approach, and the conclusions require verification by direct measurements of ACh output. The effects of sodium (and of calcium) which are described below wi be discussed in relation to the hypothesis that ACh may be released in exchange for sodium ions across the nerve terminals. METHODS The experiments were made on isolated muscles and, usually, on single nerve-muscle junctions of the frog (Rana temporaria). In most experiments an intracellular recording electrode was used, of the type described by Ling & Gerard (1949) and Nastuk & Hodgkin (1950). The adaptation of this technique to a study of neuromuscular transmission has been described in detail by Fatt & Katz (1951). The main observations consisted of measurements of the e.p.p. and of the depolarization of the end-plate produced by applied ACh. In the latter type of experiment two alternative methods were employed: (a) the surface p.d. between the nerve-free end of a muscle and a region containing end-plates was measured in the way described by Fatt (1950); (b) the resting potentials of individual end-plates were determined methodis with an intracellular electrode before and after applying ACh to the bath. The first technically easier and allows one to make a complete series of observations on one muscle, changing the bath concentrations several times. It suffers, however, from the disadvantage that the membrane potential at the end-plates isunknown, and that measurements made with different salt solutions depend upon the variable short-circuiting factor of the interstitial fluid. The second method is more direct, but it does not usually allow one to make many measurements on any one end-plate. A comparison of the responses of different end-plates suffers from the large variability of their individual reactions to ACh. Another difficulty was that during repeated insertions of the micro-electrode at one end-plate the depolarization produced by ACh was often found to decline, i.e. the resting potential gradually increased although the bath concentration of ACh remained constant. The experiments of Fatt (1950) indicate that this effect is more marked in sodiumdeficient than normal Ringer, and the present method would therefore tend to give relatively too small depolarizations in low sodium solutions. No allowance could be made to, correct for this, but qualitatively it would appear to strengthen the argument presented below. In au the experiments of this group the depolarization produced by ACh was measured, abouti min or more elapsing between the time of application and the first measurement. Some caution must be used in comparing the prolonged membrane response to applied ACh with the almost instantaneous effect of ACh released by the nerve impulse. An attempt was made to apply ACh veryrapidly to individual end-plates by means of a special micropipette, but various technical difficulties made these attempts unsuccessful. The concentration of ACh used was 1-5x 10-7 (acetylcholine chloride in Ringer), except with curarized muscles when 10 times larger ACh doses were applied. In this relatively low range of concentrations, the depolarization is approximately proportional to the dose of ACh (see, for example, Table 3). Errors due to external fluid resistance. A smau error is introduced when recording with an intracellular electrode because of a potential drop in the fluid bath outside the fibre (Nastuk & Hodgkin, 1950; Fatt & Katz, 1951). When sodium is replaced by sucrose the conductivity of the

3 SODIUM AND NERVE-MUSCLE TRANSMISSION bath is diminished and the external potential drop increases. A reduction of sodium concentration to one-fifth increased the external potential by varying amounts, usually from about 3% to about 8% of the internally recorded potential change. In comparing the records of e.p.p.'s in different sodium solutions, an allowance was therefore made to correct for this error. We used no correction for normal Ringer, and a maximum correction factor of 1 05 for records obtained in one-fifth sodium. RESULTS A. Neuromuscular block resulting from sodium deficiency When a nerve-sartorius preparation is immersed in a sodium-deficient solution (modified Ringer's fluid in which four-fifths of NaCl has been replaced by osmotically equivalent sucrose), neuromuscular block develops, and after about 30 min a simple e.p.p. is obtained, similar to that in a curarized muscle. 75 Fig. 1. End-plate potentials recorded with intracellular electrode. Neuromuscular block due to reduction of Na concentration to one-fifth. Three records from same muscle fibre: a, 0-2 mm from end-plate; b, at end-plate; c, 0-4 mm away. Time, msec. S, stimulus artifact. There is, however, an important dlifference between this type of neuromuscular block and curarization. With curarine the end-plate becomes insensitive and fails to be depolarizedl by ACh. A lowering of sodium, however, to 20% of the normal concentration has only a relatively small effect on the end-plate response to ACh (Fatt, 1950; D below). This indicates that the transmission block in 'low sodium' is due to dliminished ACh output from the nerve endings rather than to reduced ACh action on the end-plate. To obtain more quantitative information intracellular recordling had to be usedl (Fig. 1). The e.p.p. in sodium-deficient muscle has been briefly described in a previous paper where an internal recording electrode was used (Fatt & Katz, 1951). The amplitude varied at different end-plates, the largest being about 30 mv. The time course is somewhat slower than in curarized muscle, the rising phase at the 'focal' position occupying about 2 msec and the time from the start to half-dlecline about 6 msec, at 200 C. These time values

4 76 7P. FATT AND B. KATZ increase more than tenfold when the preparation is treated with a cholinesterase inhibitor (prostigmine bromide 10-6), the effect being much more striking than in curarized muscle (Fatt & Katz, 1951). To obtain complete nerve-muscle block, muscles were soaked for min in solutions containing one-quarter or one-fifth of the normal sodium concentration. This level is not much higher than that at which conduction in intra-muscular nerve branches fails, and with more prolonged soaking there was evidence of nerve block at many places, the e.p.p. response at individual end-plates becoming intermittent and eventually disappearing altogether, without showing a gradual decline in size. B. Effect of sodium concentration on the end-plate potential in normal (non-curarized) muscle In Fig. 2 an experiment is illustrated in which the sodium level was lowered to one-half. At this concentration, neuromuscular transmission is still maintained, but the rate of rise of the e.p.p. is noticeably reduced, and the spike origin is therefore delayed. Fig. 2. Effect of Na concentration on e.p.p. (non-curarized muscle, prostigmine bromide 10-6). a and b, same end-plate, using '100 Na' and '50 Na' respectively; c, another end-plate, using '25 Na'. Time, msec. To find the relation between e.p.p. and sodium concentration the following procedure was used. Several end-plates were located during preliminary curarization as described by Fatt & Katz (1951). Curarine was then removed from the bath and, usually, prostigmine bromide 10-6 was added, in order to make the conditions similar to those described in D. After a further period of soaking (30-45 min) e.p.p.'s were recorded from the previously selected points, using reduced and normal sodium levels and allowing min

5 I SODIUM AND NERVE-MUSCLE TRANSMISSION 77 soaking for each change of solution (cf. Fig. 5). Whenever possible, a third, return series was made. The lower sodium level (giving a smaller e.p.p.) was taken first, so that any experimental damage to the fibre would tend to obscure, rather than exaggerate, the effects which we are describing. To compare the magnitudes of the e.p.p.'s at various sodium concentrations, the e.p.p. was measured at a fixed interval after its start. This gives one a measure of the rate of rise rather than of the peak of the e.p.p., the latter being obscured by the muscle spike. In some experiments, the sodium concentration was 150 (U ~0C 2. ~~~~8Extra sucrose ( 150 Relative sodium concentration (%) Fig. 3. Relation between size of e.p,p. and Na concentration (non-curarized muscle). Relative scales, normal Na and e.p.p. =_ 100%. HIollow circles: mean e.p.p. size with various Na concentrations. Full circle and broken line: effect of hypertonic solution, using exrtra sucrose (102 mms) instead of sodium. Dotted. line: probable effect of excess siodium, allowing for hypertonicity. S.EF. of mean shown by horizontal bars (when exceeding radius of circle). Numbers indicate number of experiments. raised up to 50 /% above that of Ringer's solution, and in separate control exrperiments the effect of the increased osmotic pressure was checked by adding equivalent amounts of sucrose to the Ringer bath. The results of these experiments are illustrated in Fig. 2 and summarized in Fig. 3, which shows the mean ratios of e.p.p.'s at various sodium concentrations and the standard errors of these values. The values given in Fig. 3 for the lowest sodium concentration ('one-fifth') were obtained from a different set of exrperiments described in a previous paper (Fatt &s Katz, 1951). They do not represent 'paired' exrperiments, but the mean of a large number of 'unpaired' observations. C. Effect of sodiuxm concentration on end-plate potential in curarized mutscle A similar series of experiments was made on curarized muscles. This had technical advantages, for in the absence of spikes the full size and time cour-se-

6 78 P. FATT AND B. KATZ of the e.p.p. were seen, and in the absence of twitches many more records could be obtained from any one end-plate. Each experiment was completed by returning from the altered to the initial sodium level (e.g. Fig. 4), and in some experiments the temporal development of the sodium effect was traced on an individual end-plate, as shown in Fig. 5. It is noteworthy that the full Fig. 4. Effect of Na concentration on e.p.p. in curarized muscle. Two experiments. Relative Na concentrations (normal _ 100) shown in figure. '50 Na' '25 Na' Minutes Fig. 5. Successive measurements of e.p.p. on same end-plate (curarized). Ordinates: amplitude of e.p.p. Abscissae: time of recording. development took some min, in spite of the fact that we were recording from superficial muscle fibres. A summary of the results on curarized muscles is shown in Fig. 6. The relation between e.p.p. (peak size) and sodium concentration is almost linear over a wide range and, for small concentration changes, its slope is greater than that found in the non-curarized muscle (for an explanation, see p. 82 below). The time course of the e.p.p. was not noticeably affected by a variation of sodium concentration (Fig. 4).

7 SODIUM AND NERVE-MUSCLE TRANSMISSION D. The effect of applied acetylcholine at various sodium concentrations The experiments show that there is a consistent relation between e.p.p. size and sodium concentration. If we accept the view that the e.p.p. results from a reaction between ACh and end-plate, it remains to be decided whether a reduction of the e.p.p. at a low sodium level is due to a diminished respons of the end-plate to a given amount of ACh or to a diminished output of AC] from the nerve endings. The most direct approach would be to assay thl quantities of ACh released into the perfusion fluid, as was done by Dale et al, (1936). We have used an indirect method by testing the response of th ioo CL Relatie sodim conentratont(% Fig. 6. Relation between amplitude of e.p.p. and Na concentration in curarized muscle. Co-ordinats, etc., as in Fig. 3. end-plate to applied ACh at various sodium concentrations. If the depolarization so obtained is independent of the variations of sodium level employed in the preceding exrperiments, then this would strongly support the view that a change in ACh output is the deciding factor. In a few exrperiments end-plate depolarizations were measured with exrternal electrodes on m. exrt. long, dig. IV and sartorius (see Methods). It was foulnd that a reduction of the sodium concentration to one-half or one-quarter did not appreciably diminish the ACh effect and often increased it. The increase is, no doubt, to be exrplained by the increased resistance, and reduced shunting effect, of the extracellular fluid. To allow for such shunting, we assumed that in normal Ringer's solution the exrtracellular resistance is equal to the internal resistance of the fibres (cf. Katz, 1948). With this allowance, we calculate (Table 1) that the depolarization of the end-plate is reduced to O886 in 'halfsodium' and to O869 in 'quarter-sodium'.

8 80 P. FATT AND B. KATZ More extensive measurements were made with the intracellular electrode. End-plates were located in sartorius muscles, as described above (preliminary curarization, removal of curarine and addition of prostigmine). The resting potential was determined at the end-plate before and after application of a fixed dose of ACh (5 x 10-7, all values referring to concentrations of acetylcholine chloride), and again after removing ACh from the bath solution, the depolarization being the difference (AE) of the resting potentials. In twentythree experiments complete measurements could be made on one end-plate, using two different sodium concentrations (100 and 25%, the sequence being varied in individual experiments). In these 'paired' experiments the depolarization of the end-plate was diminished, at the low sodium level, to 0-68 (s.e. of mean ). In several other experiments the end-plates did not survive all these manipulations, but measurements at any one sodium level (100 or 25 %) were completed. Taking the average depolarization in all experiments at a fixed sodium concentration, we found the following values: with '100% Na', the mean resting potential E was 89 mv, and the mean depolarization AE, TABLE 1. Effect of sodium concentration on depolarizing effect of acetylcholine External measurement on m. ext. long. dig. IV (5 expts.) and m. sartorius (1 expt.). The Na concentration in Ringer, and the depolarization observed with this concentration, have been taken as 100. ACh was applied in concentrations varying between 1 and 5 x Prostigmine bromide Estimated reduction, Rel. depolarization, Rel. Na due to external fluid corrected for concentration Rel. depolarization shunt shunting *85 53 produced by 5 x 10-7 ACh was 23 mv (forty-one experiments, s.e. of mean mv); with '25% Na ', the resting potential was 91 mv, and the depolarization, uncorrected for external p.d., 16-1 mv (thirty-one experiments, S.E mv). The reduction of AE in '25% sodium' was to 0 7 (+0 08), or to if we consider the fractional depolarization AE/E. If we use a correction for the external potential drop around the depolarized end-plates, we obtain, for AE/E, 0-72 instead of Thus, while a lowering of the sodium level to one-quarter has some effect on the end-plate reaction to applied ACh, the observed depression-to about 0 7-is considerably less than that of the e.p.p., which was to It would appear, therefore, that the main action of sodium lack is one on the ACh-release mechanism, rather than on the end-plate receptors. Certain reservations must be made when comparing the two sets of experiments, for the conditions during the application of ACh are not the same as those during the release from the nerve terminals (cf. Methods). One of the points which arise in this connexion may be discussed in more

9 SODIUM AND NERVE-MUSCLE TRANSMISSION 81 detail: in a previous paper (Fatt & Katz, 1951) it was suggested that the depolarization of the end-plate is due to a short-circuiting of the muscle membrane by ACh. It can be shown that, as a consequence, a large depolarization (arising from a very low resistance at the end-plate) is established more quickly than a small depolarization (which is not associated with such a low leakage resistance at the end-plate). The theoretical effect can be seen in fig. 32 of the previous paper, a part of which is reproduced in Fig. 7. If a reduction of the sodium level were to change the end-plate response, to a given dose of ACh, from curve A to B, then the observed effect would depend upon the moment at which it is measured: it would, for instance, be considerably greater during the initial half-millisecond (which would correspond to measurements of the e.p.p.) than during the final steady state (corresponding to the observations with applied ACh). Fortunately, this objection loses its force when we are dealing with relatively small depolarizations such as those described above (AR =23 and 16 mv, or 26 and 18% respectively of the resting potential). If we calculate the initial time course of the depolarization (from eqn. 6, Fatt & Katz, 1951), we find that at t =0-5 msec the depolarization would be 4-35 mv (100% Na) and 2-74 mv 90c mv I msec Fig. 7. Theoretical effect on membrane potential of suddenly applying a shunt of 10,000 or 100,000 Ql across end-plate membrane (cf. Fatt & Katz, 1951, fig. 32). Ordinates: depolarization, mv. Abscissae: time, msec. TABLE 2. Calculated short-circuit resistance at the end-plate Resistance, in ohms Reduction factor of end-plate conductance '100 Na' '25 Na' ('25 Na'/'100 Na') e.p.p. 35, , ACh applied 574, , (5 x 10-7) (25% Na), showing a reduction factor of 0-63 compared with the observed final value of Hence, the present argument is not seriously affected by these considerations. One might argue that the calculated 'leakage resistances' give a better quantitative indication of the intensity of ACh action than the observed potential changes. An approximate calculation was made for both types of experiments and the results are listed in Table 2. E. Decurarizing action of sodium ions It is clear from the preceding experiments that the effect of sodium ions is not restricted to one specific stage of neuromuscular transmission. The main action appears to be concerned with the output of ACh from the nerve endings, but there is also an effect on the sensitivity of the end-plate to ACh. Another PH. CXVIII. 6

10 82 P. FATT AND B. KATZ unexpected action of sodium ions was found when its effect on curarized endplates was studied. When testing the local electric response of a normal frog muscle (m. ext. long. dig. IV) to applied ACh, it is found that the end-plate depolarization is blbt little affected by moderate reductions of sodium concentration. In Table 3, for instance, two experiments are summarized, in which a lowering of sodium to one-third actually increased the observed depolarization by a factor of 1F03 and 1-13 respectively (diminished shunting). When the experiment was made after curarization, a very different result was obtained, there being a marked depression of the end-plate response (to 0-52 and 0-74 respectively). This result shows that the action of curarine is augmented by sodium-lack and, conversely, that there is an antagonism between sodium ions and curarine. It is of interest that no such interaction has been observed between calcium ions and curarine (Castillo & Stark, 1952). Whatever the theoretical significance of this observation, it helps to explain why a change of sodium concentration has a more drastic effect on the size of the e.p.p. in curarized muscle (Fig. 6) than in normal muscle (Fig. 3). TABLE 3. Interaction between sodium and curare External measurements of end-plate depolarization on m. ext. long. dig. IV. Drugs used: prostigmine bromide 10-6; D-tubocurarine chloride and acetylcholine chloride in concentrations shown below. Temp Na concentration in Ringer 100. External depolarization (mv) Ratio of ACh depolarizations conen. '100 Na' '33 Na' '100 Na' ('33 Na'/'100 Na') Expt. 1: A. Curarine 6 x x O B. Curarine removed x Expt. 2: A. Without curarine x B. Curarine 6 x x x F. Effect of calcium and potassium ions on end-plate potential size The effect of calcium on nerve-muscle transmission has been studied by several workers (see Castillo & Stark, 1952), and it appears that its action on the e.p.p. is very similar to that of sodium. In a series of experiments, the same procedure was employed as in B, varying the calcium concentration between 0-45 and 3-6 mm (the usual concentration in frog's Ringer is 1-8 mm, the concentration used in most of the previous experiments was 3-6 mm; cf. Fatt & Katz, 1951). Changes in calcium concentration had two distinct effects: (i) The 'threshold point' at which the e.p.p. gives rise to a muscle spike increases with the calcium level. Thus it is shown in Table 4 that with 3-6 mm-ca

11 SODIUM AND NERVE-MUSCLE TRANSMISSION 83 the 'e.p.p. threshold' (cf. Fatt & Katz, 1951, fig. 20) was 25% higher, and with 0 9 mm-ca 12% lower, than in normal Ringer's solution. (ii) The size of the e.p.p. varies with the calcium concentration, the results of thirt -one experiments being summarized in Fig. 8. The general form of this relation is similar to that obtained for sodium (Fig. 3), if we compare the effects of equal percentage changes of ion concentrations. (In terms of equal molar concentration changes calcium ions are, of course, about 50 times more effective than sodium.) 150 TV 0 = ~~~~~ Relative calcium concentration (%) Fig. 8. Effect of Ca concentration on e.p.p. (non-curarized muscle). Co-ordinates, etc., as in Fig. 3, except that abscissae show relative calcium concentrations (normal = 1.8 mm =_ 100). TABLE 4. Effect of calcium concentration on height of end-plate 'step' Summary of series of paired measurements, made on same end-plates. Ratio of end-plate step ('altered Ca'/'normal Ca') Ca concn. (mm) ('normal' = 1.8 mm) Resting potential, mean value (mv) End-plate step, mean, (mv) f.A Mean ratio S.E. of mean 1-25 t ± No. of expts. 9 At low calcium levels (0.45 mm) the relation between e.p.p. size and calcium concentration becomes steep, and in this region an interesting seasonal variation was observed. The results shown in Fig. 8 were obtained in well-fed 'winter frogs'. In another series of experiments, made during the summer months on emaciated animals, a lowering of calcium concentration to 0 45 mm was found to reduce the e.p.p. well below threshold, to an amplitude of only 6-2 6

12 84 P. FATT AND B. KATZ a few millivolts. Under these conditions successive e.p.p. responses varied in a striking 'quantal' fashion, illustrated previously (Fatt & Katz, 1952, fig. 9). This observation is of interest because it indicates that, at reduced calcium levels, intermittent block occurs at individual nerve endings, or at even smaller membrane patches which are concerned with the release of ACh. Such a 'quantal' blocking effect was never observed with low sodium levels: there was sometimes intermittent total failure of the e.p.p. response (p. 76), but this was an all-or-none effect indicating a conduction block in the motor axon. Perhaps the most significant difference between sodium and calcium actions is that observed during a study of the spontaneous activity at the nervemuscle junction (Fatt & Katz, 1952). It was shown that in the resting fibre a spontaneous discharge of minute e.p.p.'s occurs arising from activity at individual nerve terminals. The amplitude ofthese miniature e.p.p.'s diminished when the sodium concentration was lowered, but remained constant when the calcium concentration was reduced. It seems probable that the effect of calcium on the size of the e.p.p. is of a 'quantal' nature, involving a variation in the number of responding terminal units (rather than a variation in the number of ACh molecules released by each of them). In nine experiments the effect on the size of the e.p.p. of doubling the potassium concentration (from 2 to 4 mm) was studied. There was only a small and statistically insignificant change in the mean value of the e.p.p. which was altered by a factor of 103 (S.E ). Higher concentrations of potassium are known to reduce resting and end-plate potentials and to lead to nervemuscle block (cf. Coppee, 1943). A transient large increase of the e.p.p., as a result of a raised potassium level, has been described in curarized frog muscle by Walker & Laporte (1947), but with the present technique such a transient change could not have been detected. DISCUSSION The results indicate that the output of ACh from active motor nerve endings depends upon the concentration of sodium ions in the extracellular fluid. The same conclusion, however, applies to calcium ions, and the question arises what the exact role of these ions may be in the normal process of nerve-muscle transmission. The hypothesis was put forward that ACh may be released in exchange for sodium which enters the nerve terminal during its electric activity. This idea is certainly compatible with all existing evidence. But the present experiments do not provide unequivocal evidence for it, for the relation between the amount of sodium which enters and the amount of ACh which leaves the nerve ending may be of a more indirect kind. For example, the change of the membrane potential brought about by the entry of sodium ions may lead to a chemical reaction in the course of which ACh is liberated, while the membrane potential itself may be restored, quite independently, by the release of

13 SODIUM AND NERVE-MUSCLE TRANSMISSION 85 potassium ions from the nerve terminal. It is impossible to decide this question without accurate information on the size and surface area ofthe active terminals and the quantity of the ions exchanging across them. If the active surface area is somewhere between 10 and 10,000 U2, and the transfer of sodium during one impulse amounts to about 3 x mole/cm2 (cf. Keynes & Lewis, 1950), then an equivalent release of ACh would provide a quantity of between 3 x and 3 x mole per end-plate per impulse. According to Acheson's (1948) estimate, about mole of ACh per end-plate per impulse has been recovered in perfusion experiments, but this figure would give an underestimate of the amount released, because of various losses inherent in the method. A comparison of the effects of sodium and calcium ions has revealed a general similarity, but there remains an important difference in detail, namely that calcium affects the e.p.p., and presumably the output of ACh from nerve endings, in discrete steps, while the effect of sodium appears to be continuously graded. Even with sodium there must, of course, be discontinuities at the molecular level; calcium, however, changes the ACh output in quanta containing large numbers of molecules (cf. Fatt & Katz, 1952). It is of special interest that in a model of nerve activity suggested by Hodgkin and Huxley (see Hodgkin, Huxley & Katz, 1949) an important and 'synergic' role was assigned to calcium and sodium ions. In this model the membrane contains a calcium compound which dissociates during excitation and liberates 'carrier' molecules each capable of transporting a large number of sodium ions into the axon. This provides, therefore, a system in which one calcium ion is responsible for the transfer of a large aggregate of sodium ions. If ACh+ ions are released in exchange for sodium ions, then a system like that proposed by Hodgkin and Huxley would provide a basis for the fact that a change of calcium concentration affects ACh release in an aggregate manner, while a change of sodium concentration does so only in a continuous, molecular, gradation. There is, at present, one principal difficulty in accepting this kind of explanation: if calcium ions are responsible individually for a quantal release of ACh which produces a miniature e.p.p. (Fatt & Katz, 1952), then a lowering of calcium concentration would be expected to lead not only to a reduced size of the e.p.p., but also to a reduced rate of the spontaneous miniature discharge; this, however, was not always observed (Fatt & Katz, 1952). It is difficult, therefore, to dismiss the view that calcium-lack may act simply by blocking nerve impulses at pre-terminal points (e.g. at bifurcations of the axon) without affecting spontaneous activity at the terminals themselves. It is impossible to decide between this and other interpretations until direct evidence can be obtained concerning the properties of motor nerve endings.

14 86 P. FATT AND B. KATZ SUMMARY 1. The effect of sodium ions on neuromuscular transmission has been studied using the intracellular recording technique. Special attention was paid to the effects of sodium on the electric responses of single end-plates (i) to a motor nerve impulse, and (ii) to an applied dose of acetylcholine. 2. When the external sodium concentration is reduced to one-fifth of that in normal Ringer the end-plate potential falls to subthreshold magnitude and neuromuscular block ensues. 3. The relation between e.p.p. and sodium concentration was studied in normal and curarized muscle. In non-curarized muscle, lowering of sodium concentration to one-half diminishes the e.p.p. to about two-thirds; lowering of sodium to one-quarter diminishes the e.p.p. to less than one-third. 4. The electric reaction of the end-plate to applied acetylcholine (ACh) is also affected by changes in sodium concentration but to a lesser extent, e.g. lowering of sodium concentration to one-quarter reduces the end-plate depolarization, caused by a given dose of ACh, to about The neuromuscular block during sodium deficiency appears to be due mainly to a reduction in the ACh output from active motor nerve endings, similar to the block produced by lack of calcium, but in contrast to the effect of curarine. 6. A hypothesis is discussed according to which the release of ACh from the nerve endings occurs as a consequence of the entry of sodium ions into the terminal, and is to be regarded as a cation exchange process, analogous to the release of potassium ions from active nerve or muscle fibres. We are indebted to Prof. A. V. Hill for the facilities provided in his laboratory and to Mr J. L. Parkinson for his unfailing help. This work was carried out with the aid of a grant for scientific assistance made by the Medical Research Council. REFERENCES ACHESON, G. H. (1948). Physiology of neuromuscular junctions. Chemical aspects. Fed. Proc. 7, CASTILLO, J. DEL & STARK, L. (1952). The effect ofcalcium ions on the motor end-plate potentials. J. Physiol. 116, COPPftE, G. (1943). La transmission neuro-musculaire: Curarisation, d6curarisation et renforcement a la jonction myo-neurale. Arch. int. Physiol. 53, DALE, H. H., FELDBERG, W. & VoGT, M. (1936). Release of acetylcholine at voluntary motor nerve endings. J. Phy8iol. 86, FArT, P. (1950). The electromotive action of acetylcholine at the motor end-plate. J. Physiol. 111, FATT, P. & KATZ, B. (1951). An analysis of the end-plate potential recorded with an intracellular electrode. J. Phy8iol. 115, FATT, P. & KATZ, B. (1952). Spontaneous subthreshold activity at motor nerve endings. J. Physiol. 117, HODGKI, A. L. (1951). The ionic basis of electrical activity in nerve and muscle. Bi,ol. Rev. 26,

15 SODIUM AND NERVE-MUSCLE TRANSMISSION 87 HODGxIN, A. L., HUXLEY, A. F. & KATZ, B. (1949). Ionic currents underlying activity in the giant axon of the squid. Arch. Sci. phy8iol. 8, KATZ, B. (1948). The electrical properties of the muscle fibre membrane. Proc. Roy. Soc. B, 185, KEYNES, R. D. & LEWis, P. R. (1950). Determination of the ionic exchange during nervous activity by activation analysis. Nature, Lond., 165, 809. LUNG, G. & GmAD, R. W. (1949). The normal membrane potential of frog sartorius fibres. J. cell. comp. Phy8iol. 34, NASTUK, W. L. & HODGKN, A. L. (1950). The electrical activity of single muscle fibres. J. cell. comp. Phy8iol. 85, WAIKm, S. M. & LAPORTE, Y. (1947). Effects of potassium on the end-plate potential and neuromuscular transmission in the curarized semitendinosus of the frog. J. Neurophy8iol. 10,