PARAMETERS OF TRANSMITTER RELEASE IN SQUID NEURONAL SYNAPSES MILTON PETER CHARLTON THE UNIVERSITY OF TEXAS AT AUSTIN Marine Science Institute PORT ARANSAS MARINE LABORATORY Port Aransas, Texas 78373 PARAMETERS OF TRANSMITTER RELEASE IN SQUID NEURONAL SYNAPSES MILTON PETER CHARLTON PARAMETERS OF TRANSMITTER RELEASE IN SQUID NEURONAL SYNAPSES APPROVED BY SUPERVISORY COMMITTEE: WIS IS PARAMETERS OF TRANSMITTER RELEASE IN SQUID NEURONAL SYNAPSES by MILTON PETER CHARLTON, B.Sc., M.Sc DISSERTATION Presented to the Faculty of the Graduate School of The University of Texas at Austin in Partial Fulfillment of the Requirements for the Degree of DOCTOR OF PHILOSOPHY THE UNIVERSITY OF TEXAS AT AUSTIN December, 197& ACKNOWLEDGEMENTS I would like to thank Dr. George Bittner for his constant enthusiasm and assistance during this work. I was supported at various times during this research by NSF and NIH grants to Dr. Bittner and by fellowships from the National Research Council of Canada and the Grass Foundation. I would also like to thank the faculty and staff of the University of Texas, Marine Science Institute and the staff of the Marine Biological Laboratory, Woods Hole, Massachusetts, for providing laboratory space and animals. Finally, I thank my wife, Connie, who typed drafts of this thesis and without whose emotional support this work would have been impossible. TABLE OF CONTENTS Page Chapter 1: Facilitation of Transmitter Release at Squid Synapses 1 Chapter 2: Facilitation and Changes in Pre-synaptic Potentials 31 Summary . , : 65 Figures . 69 Bibliography 113 CHAPTER I FACILITATION OF TRANSMITTER RELEASE AT SQUID SYNAPSES Introduction Successive action potentials in a pre-synaptic axon terminal often evoke responses of increasing amplitude in the post-synaptic cell. This phenomenon, called facilitation, is of interest to physiologists for several reasons. First, facilitation is a common feature of chemical synapses and hence may represent a basic aspect of the mechanism of transmitter release (del Castillo and Katz, 1954; Dudel and Kuffler, 1961; Kuno, 1964; Charlton and Bittner, 1974; also see Table III). Knowledge of the mechanism of facilitation may enable greater understanding of the transmitter release process. Second, facilitation could play an important integrative role by affecting the relationship between pre-synaptic firing frequency and post-synaptic depolarization. Facilitation at the squid giant synapse has been noted by other workers but has not been extensively quantified (Takeuchi and Takeuchi, 1962, Bloedel et al, 1966; Miledi and Slater, 1966; Katz and Miledi, 1967; Charlton and Bittner, 1974, Kusano and Landau, 1975). In fact, most of the available data ( cf Charlton and Bittner, 1974) indicate that the time course of facilitation in squid synapses is quite different from that reported for other synapses, such as frog neuromuscular junctions (Mallart and Martin, 1967). The disparity in the data might be explained by the fact that the experiments on frog synapses used solutions containing high magnesium and low calcium concentrations to depress transmitter release and to enhance facilitation, while the squid experiments employed normal calcium concentrations and repetitive stimulation to depress transmitter release to subthreshold levels. Thus it is unclear whether Takeuchi and Takeuchi (1962) or Miledi and Slater (1966) studied the same kind of facilitation as was studied in the frog. In the present experiments, solutions of low calcium concentrations were used to reduce transmitter release to subthreshold levels and to enhance the detection of facilitation. I have shown that the magnituder time course of decay and summation of facilitation are similar in squid and other animals. This result lends generality to findings in the squid concerning the mechanism of facilitation discussed in this and the following chapter. Although its mechanism is obscure, facilitation is known to involve an increased secretion of transmitter quanta by the presynaptic cell and is not due to changes in properties of the postsynaptic cell (del Castillo and Katz, 1954; Liley, 1956; Dudel and Kuffler, 1961; Kuno, 1964; Martin and Pilar, 1964; Kuno and Weakley, 1972 a, b). Various hypotheses for the mechanism of facilitation involve changes in major factors governing transmitter release such as pre-synaptic depolarization and the availability of calcium ions. In the residual calcium hypothesis, some of the calcium which enters a nerve terminal during an action potential lingers in the terminal and increases the probability of transmitter release during subsequent action potentials (Katz and Miledi, 1965 b, 1968; Rahamimoff, 1968, 1973; Miledi and Thies, 1971; Rahamimoff and Yaari, 1973; Younkin, 1974). Other hypotheses for the mechanism of facilitation propose that slow changes in membrane potential (Lang and Atwood, 1973; Dudel, 1971) or increases in action potential amplitude, duration or invasion could account for facilitation (Dudel and Kuffler, 1961; Dudel, 1971; Takeuchi and Takeuchi, 1962; Bittner, 1968; for reviews see Lang and Atwood, 1973; Rahamimoff, 1973; Balnave and Gage, 1974; Zucker, 1974 a, b). In order to test these hypotheses, it is necessary to directly measure and modify these parameters in the pre-synaptic terminal. I chose the squid giant synapse to study facilitation because its extremely large size enables one, by placing microelectrodes directly in the pre-synaptic terminal, to measure and manipulate some of the above variables (Takeuchi and Takeuchi, 1962; Miledi, 1973; Llinas and Nicholson, 1975). This preparation was used to test the residual-calcium hypothesis by altering extracellular calcium concentration and temperature, and to test the membrane-potential hypotheses by recording and manipulating presynaptic membrane potentials. Materials and Methods The squid Lolliguncula brevis was obtained near the University of Texas Marine Science Institute, Port Aransas, Texas, where all experiments on this species were performed. These“squid were found in large numbers all year and were caught by bottom trawl. Undamaged animals survived indefinitely when kept in running, clarified seawater, in a circular fiberglass tank (diameter, 2.5 m; depth, 0.8 m) and fed live or frozen fish or shrimp. The work on Loligo pealei was done at the Marine Biological Laboratory, Woods Hole, Massachusetts. Stellate ganglia and nerves were isolated as described by Miledi (1967) and Arnold et al (1974). Since decapitation often elicited sustained mantle contractions, only the tentacles and mouth-parts were cut off during dissection. This procedure avoided repetitive firing of the pre-synaptic axons which synapse on the third-order, post-synaptic giant cells originating in the stellate ganglion (Young, 1939). Excised ganglia were bathed in flowing artificial seawater containing 3 mM sodium bicarbonate, 54 mM magnesium, 467.5 mM sodium, 10 mM calcium and 10 mM potassium with chloride as the anion (pH 7.6). Salines were oxygenated by bubbling 99.5% , 0.5% before and during experiments. This gas mixture allowed adequate oxygenation and avoided the rise in pH that occurred when 100% was used. Saline flowed through a regulated Peltier-effect cooler before it entered the experimental chamber w’here temperature was measured by a thermistor. Once the synapses were visually identified, the preparations were bathed in salines having calcium concentrations of 2-6 mM, with appropriate modification of sodium concentration to keep osmotic pressure constant. Preparations kept in calcium concentrations less than 4 mM seemed prone to damage by microelectrodes and often lost excitability within an hour. In order to obtain small post-synaptic potentials (< 5 mV) and to ensure prolonged synaptic function, I kept the calcium concentration in most L. pealei experiments above 4 mM and added a small amount (< 5 mM) of manganese to depress transmitter release. Manganese is much more potent than magnesium as a competitive inhibitor of calcium action in transmitter release, yet has little effect on the amplitude of spontaneous miniature potentials (Katz and Miledi, 1969 a; Meiri and Rahamimoff, 1972; Balnave and Gage, 1973). Concentrations of manganese approaching the normal concentration of magnesium may cause an increase in the rate of spontaneous miniature potentials in toad muscle (Balnave and Gage, 1973), but in our experiments the manganese concentration was only a small fraction of the normal magnesium concentration. Since the squid synapses were quite variable in their response to decreased calcium concentration and equilibration times were shorter for increases in ion concentration than decreases, the addition of manganese proved to be a very convenient way to adjust transmitter output. Data were recorded after stable responses were obtained in new salines (20-90 minutes). Synapses of giant, accessory giant and non-giant presynaptic fibers were examined to determine whether different synapses in the same animal or homologous synapses in different species had similar facilitation properties. In this paper I have classified all but the second-order pre-synaptic giant cell as presynaptic non-giant cells. Non-giant axons which form the proximal synapses (Young, 1939) were used when the pre-synaptic giant axon was nonfunctional. In some cases I was able to identify the "accessory giant axon" as the active axon, but in others I simply knew that the active axon was not the pre-synaptic giant axon and was therefore one of at least three other axons which form synapses on the post-synaptic giant cell. Pre-synaptic axons were stimulated via extracellular wire electrodes applied to the pre-ganglionic nerve or by intracellular microelectrodes. Extracellular stimulation of the pre-synaptic giant axon was confirmed by simultaneous intracellular recording of pre-synaptic potentials and post-synaptic potentials (psp’s). Synaptic input from non-giant pre-synaptic axons at proximal synapses was avoided by careful positioning of stimulating electrodes and by selectively pinching or crushing the pre-ganglionic nerve in regions not containing the pre-synaptic giant axon. The postsynaptic potentials and synaptic delays were carefully monitored while the extracellular stimulus strength and polarity were varied to ensure that only one synapse was activated. Unless otherwise indicated, all psp's were recorded from the third-order giant cell which was located in the most caudal stellar nerve. Pre- and post-synaptic giant cells were penetrated under visual control with microelectrodes (5-10 MQ) containing 3M KCI and 0.5 M K-citrate. These electrodes were always inserted within the zone of synaptic overlap of the pre- and post-synaptic giant axons. Pre-synaptic action potentials and psp’s were displayed using conventional electrophysiological techniques. Twenty or forty responses were stored and averaged in a computer of average transients (CAT 1000, Mnemotron Corp.), and averaged responses were plotted on a chart recorder. The number of responses averaged was chosen to reduce the effect of variation in junction potential size while allowing the experiment to be completed in a reasonable length of time. An interval of 10 sec was allowed between recording sweeps when twin-pulse stimuli or short trains were delivered. Psp's rising from the falling phase of the previous potential were measured from the peak of the former to the projected tail of the latter and corrected for non-linearity according to the procedure of Martin (1955). Experiments were performed only on unfatigued preparations in which normal synaptic transmission could be demonstrated. Synapses usually regained their original transmission properties when placed in normal calcium saline at the end of an 8-hour experiment conducted in low calcium saline containing manganese. Results Amount and Decay of Facilitation Synaptic potentials (psp’s) in post-synaptic cells are due to the action of transmitter chemical released by the pre-synaptic cell. Variation in the amplitude of the psp almost always indicates variation in the amount of transmitter released (Katz, 1969). Thus, changes in pre-synaptic cell activity such as facilitation can be detected by measurements made in the post-synaptic cell. I used the second pulse of a stimulus pair to test for the amount of facilitation remaining from that contributed by the first, or conditioning pulse. The time course of the decay of this facilitation was determined by varying the interval between the two pulses. (Each stimulus pulse caused a pre-synaptic action potential which resulted in a psp.) Facilitation was defined as the ratio between the amplitudes of the second and the first (Vs) psp’s minus 1, i.e.: Vi - v 0 (equation 1) (Mallart and Martin, 1967; Magleby, 1973 a). A typical output plot of a pair of psp’s from the averaging computer is shown in Figure 1 (insert) in which the interval between stimuli was about 10 msec. The second psp was clearly larger than the first, and, as the interval between the two stimuli was increased, the value of f decreased. The relation between the interpulse interval (t) and f is shown in the lower part of Figure 1. The points between 5 and 15 msec generally fall on a straight line when plotted on semi-log coordinates. Due to the refractory period of the pre-synaptic axons, stimulation could not be produced reliably at intervals of less than 5 msec. Therefore, the facilitation at zero time or the maximum facilitation (F ) produced by the first impulse could only be estimated by extrapolating to the ordinate, a line through the points between 5 and 15 msec. This analysis assumes that the relation between f and t is the same for short intervals as it is for longer intervals, although this may not be the case in other systems when artificial depolarizations are given at very short intervals (Katz and Miledi, 1968). From the graph of f versus t, I derived a value for and a value for the time constant of decay (T ) which was defined as the time required for to decay to Since the relation between f and t appears to be a straight line on a semi-log plot, it can be described by the exponential equation: f = (equation 2) Similar relations have been found in frog (Mallart and Martin, 1967; Magleby, 1973 a), toad (Balnave and Gage, 1974) and crayfish neuromuscular junctions (Linder, 1974; Zucker, 1974 a; Bittner and Sewell, 1976). The slope of the decay curve of f almost always decreased at stimulus intervals greater than 10 msec after which f decayed at a slower rate (? 2 ) than (Figs. 1-4, 7). The changes in appeared at 15 to 20 msec at 20°C in giant and non-giant pre-synaptic cells of L. brevis (Figs. 1,7) while in L. pealei giant fibers the change in slope occurred between 10-15 msec at 20°C (Figs. 2-4). The time to the occurrence of the break in the decay curve in many preparations was quite constant at any given temperature but increased at lower temperatures (Figs. 2,3). In most preparations the decay of f was non-monotonic, i.e., the change in the slope of the decay curve was accompanied by an increase in facilitation (Figs. 1,4, 6,7). A similar result has been found in the frog and crayfish neuromuscular junctions by Mallart and Martin (1967) and by Bittner and Sewell (1976). Facilitation during the second component of decay was quite small and difficult to quantify. Values of facilitation at the peak of the hump ranged from 0.15 to 0.2 while the second time constant of decay (T ) had values between 22 and 63 msec. In pre-synaptic giant cells in both species of squid, the average values of and determined in a number of preparations were significantly larger and smaller, respectively, than the average values found in non-giant cells (Table I). In four of five ganglia in which data were collected from both giant and non-giant synapses, the giant synapse had the smaller but in only two of the five ganglia was larger in the giant synapse. In the other three cases, Table I: Means and standard deviations (SD) of facilitation parameters in giant and non-giant synapses of L. brevis and L. pealei. Means determined from preparations in different concen++ ++ trations of extracellular calcium ions [Ca ] 0 and [Mn ] o as noted. Temperature ±l°C. n = number of preparations tested. Significant differences determined by Student’s t-test (L. brevis) and by Multiple Comparisons for non-balanced two-way classification (Ferguson, 1967) for L_. pealei. was identical in one and in two was slightly smaller in the giant than in the non-giant synapse. In two preparations using L. pealei, facilitation was measured by simultaneously recording psp’s from more than one postsynaptic giant cell in the same ganglion. Since the second-order, pre-synaptic giant cell branches and forms synapses with all the third-order, post-synaptic giant cells, this experiment tested for variations in facilitation between synapses of the same pre-synaptic cell. Figure 4A, B shows that facilitation in the pre-synaptic synapse onto the largest (seventh) giant post-synaptic cell was identical to facilitation measured in the pre-giant synapse onto the third and fourth post-synaptic giant cells. In the other animal, the giant synapse onto the seventh post-synaptic giant cell had less than half the initial facilitation and had a time constant of decay twice as long = 0.3, = 14) as that found in the giant synapse onto the fourth post-synaptic giant cell = 0.8, = 7). Effect of Temperature on Facilitation If facilitation were due to a persistent increase in intracellular calcium concentration ("residual calcium"), the decay of facilitation might be due to a gradual decrease in calcium concentration. The rate of decay of facilitation (T^), and possibly of residual calcium, would not be expected to be very temperature sensitive if the decay were due to a simple diffusion process. In order to test this hypothesis I measured facilitation at various temperatures (Table I). While changes in temperature had variable and insignificant effects on (Table I, compare Figs. 2 and 3), reduction of temperature always caused a reduction in rhe rate of decay of facilitation in L. pealei giant synapses (Table I). These results are similar to those found by Balnave and Gage (1970, 1974) in the toad neuromuscular junction although the extrapolated of the decay rate (between 20 and 12°C) was about 2 in my experiments and about 4 in theirs. Due to the increase in T , the average facilitation at twin-pulse intervals of 10 msec was 33% greater at 15°C than at 20 °C in L* pealei giant synapses despite the fact that the average values of at 15°C and 20°C were not significantly different (Table I). The change in slope (hump) of the decay curve occurred earlier (t = 10 to 15 msec at 20°C) at warm temperatures than at colder temperatures (t = 20 to 25 msec at 15°C) in L. pealei giant synapses (Table I). In some preparations facilitation was reduced at short intervals at cold temperatures (Fig. 2). Reduction in temperature in the range of 20°C to 15°C usually caused a slight increase in the amplitude of the psp, presumably due to the increased height and duration of the pre-synaptic action potentials. However, psp amplitude decreased at temperatures below 15°C despite increases in action potential height and duration. Figure 5 is a tracing of two pairs of averaged action potentials in a pre-synaptic terminal at 15°C and 12°C. The action potential recorded at lower temperatures showed a slower conduction velocity as indicated by the increased time between the stimulus artifact and the arrival of the action potential at the end of the terminal. The psp’s were 2.2 mV at 15°C and 1.5 mV at 12°C. The synaptic delay was longer at lower temperatures and the rate of rise, rate of fall and rate of recovery from the hyperpolarizing-after-potential were all slower at the lower temperature. Similar changes are evident in records of Lester (1970). It should also be noted that the second action potential of each pair shown in Figure 5 rose to a level more positive than the first. The difference between the peak voltages reached by the second and first action potentials at 15°C was 1.2 mV while at 12°C the difference was 3.5 mV measured from resting potential. Had the total amplitude of the second action potentials been measured, these differences would have been even greater. Facilitation at this interval (10 msec) paralleled the changes in action potential size and was 0.54 at 15°C and 0.72 at 12°C. A more complete account of pre-synaptic action potentials and their relationship to facilitation will be given in the following chapter. Effect of Changes in lonic Concentrations on Facilitation Facilitation requires the presence of extracellular calcium ion (Katz and Miledi, 1968; Younkin, 1974) and may be due to "residual calcium" remaining after nerve action potentials (Katz and Miledi, 1965 b, 1968; Rahamimoff, 1968, 1973; Miledi and Thies, 1971; Rahamimoff and Yaari, 1973; Younkin, 1974). With this in mind, I measured facilitation in various calcium and manganese concentrations to determine if these manipulations could affect the production and decay of facilitation and, presumably, residual calcium. An increase in [Ca ] o or decrease in [Mn ‘ ] o was always accompanied by large increases in psp amplitude (Table II) but changes in facilitation parameters Tj) under these conditions were not consistent between preparations. For example, in three experiments an increase in [Ca ]o or decrease in [Mn ]o led to an increase in and a decrease in Tj (Fig. 4); in one preparation there were small increases in both and in one preparation there was a decrease in and little apparent change in (Fig. 6) ; yet another preparation showed a decrease in and an increase in Relatively few experiments involving changes in ion concentration were attempted since the equilibration time for these changes was often as long as 1.5 hours. Effects of Repetitive Stimulation on Facilitation Short trains of stimuli were delivered to giant or non-giant pre-synaptic axons and responses monitored in the largest postsynaptic giant cell. Facilitation was measured by comparing the size of successive psp* s (V ) to the size of the first psp of the train (Vq): _ V n - V 0 f n v v 0 (equation 3) Table II: Facilitation Parameters in Different lonic Concentrations (Values in parentheses from fourth post-synaptic giant axon.) If the interval between stimuli within a train is At, then f (nAt) represents the amount of facilitation present after n intervals of length At. Facilitation was found to accumulate during repetitive stimulation in a manner indicating that increases in facilitation became smaller and smaller, i.e.: [f nAt " f (n-l)At ] n “ As an aid in describing the accumulation of facilitation, I compared the data to predictions of a linear summation model devised by Mallart and Martin (1967). The linear summation model assumes that each stimulus results in the production of an identical amount of facilitation which decays in the same manner as that following a single stimulus and which sums linearly with facilitation remaining from the preceding events (Mallart and Martin, 1967, Magleby, 1973 a). If only the first components of facilitation (Fj, Tp found in the twin-pulse study are considered, this model predicts that, with repetitive stimulation, facilitation should reach a maximum value (K) described by: K = F}/ (exp(At/Tp-1) (equation 4) and the growth of facilitation during the train should be described by the error function: f = K d“ ex P(-nAt/T )) (equation 5) In this analysis, the delayed small increase in facilitation and its subsequent slow decay are ignored because they should contribute little to summation during short trains of stimuli. In Figures 2 t 6,7, 8 the observed growth of facilitation during short trains of stimuli (dots) is plotted and compared to the predicted growth (solid line) calculated from the error function (equation 5). In most cases the experimental data points were quite close to the predicted curve. The rate of growth of facilitation during a train and the maximum facilitation attained were both greater at higher frequencies of stimulation as predicted by the linear summation model (Figs. 2,6, 7,8). Figure 2 shows an experiment on a L_. pealei giant fiber in which the predicted rise in facilitation was greater than the observed data at At = 6 msec. This deviation can be explained by the fact that the amount of facilitation detected in this preparation at short intervals (6 msec) was somewhat depressed (Fig. 2A) and should therefore not sum to as high a level as predicted using and calculated from longer pulse intervals. It should be noted that the summation prediction at At = 10 msec in the same cell was quite accurate. Summation in an L. brevis non-giant fiber (Fig. 6) was predicted equally well in ionic conditions in which the psp amplitude was varied about twofold by the addition of manganese. In aL. pealei giant fiber (Fig. 8), two synapses summed facilitation in about the same manner, although the predicted rise was greater than the data when [Ca ++ ] o was lowered (Fig. 8B). Although some exceptions are noted above, in most cases the magnitude and growth of facilitation during repetitive stimulation was adequately predicted by the linear summation model using the parameters and derived from a twin-pulse experiment on the same preparation. Discussion Facilitation Following a Single Impulse The data indicate that, following a single action potential in the squid, the amount of facilitation produced and its subsequent decay are qualitatively and quantitatively similar to that found at many synapses in other organisms (Table III). The values for zero time facilitation (Fp are within a remarkably small range (0.4-5), considering the diversity of the preparations. Values of show a greater range, with for the squid giant synapse being more rapid than most. The decay of facilitation followed a dual exponential time course in squid, as it does in most of the preparations mentioned in Table 111. The two phases of decay of facilitation are commonly thought to begin immediately after an action potential (Magleby, 1973 a; Zucker, 1974 a; Linder, 1973, 1974) and can be represented by: f = expC-t/Tp +F 2 exp(-t/T 2 ) (equation 6) This description would be an oversimplification of much of our data (Figs. 1,3, 4,6, 7) and that of Mallart and Martin (1967) or Bittner and Sewell (1976) who reported that the second component of decay is often preceded by a slight additional facilitation so that the composite decay curve displays a hump. This can be represented approximately by: f = F exp(-t/T ) + F exp (-(t-T /T )) 1 12 max 2 (equation 7) assuming that the second component of facilitation rises abruptly to a maximum (F ) at t = T (negative values of (t-T ) are ignored) 2 max max and decays thereafter with a time constant, Equation 7 is a discontinuous function and thus predicts two values of f for one value of t at t = T , which is a condition not detected in our max experiments. Although data from some other preparations do not always show this hump, it might be obscured in experiments where responses are recorded from a population of synapses or if the parameters and T are calculated, using linear summation assumptions, from the incremental increases in psp amplitude which accompany repetitive stimulation (Magleby, 1973 a). The observation that there are two time constants of decay with an increase in facilitation associated with the second component suggests that the decay of facilitation after a single action potential involves more than the simple decay by different time courses of two substances or the decay by different time courses of the same substance from two compartments. Either of these possibilities would produce a smooth decrease in facilitation. The time course of facilitation found in my experiments is quite different from that found by Takeuchi and Takeuchi (1962) and Miledi and Slater (1966). A possible explanation for this disparity is that I used low [Ca ++ ] o salines which enhanced the possibility of detecting facilitation (Mallart and Martin, 1967; Kusano and Landau, 1975) while Takeuchi and Takeuchi (1962) and Miledi and Slater (1966) all employed normal [Ca ++ ] o salines and used repetitive stimulation to depress transmitter release to sub-threshold levels —a procedure which would have depleted transmitter stores drastically. It is thus possible that the facilitation measured by Takeuchi and Takeuchi (1962) and Miledi and Slater (1966) was complicated by the recovery from depression which occurred in the interval between twin-pulse stimuli. It is also important to note that facilitation in the ++ giant synapse has been detected indirectly at normal [Ca ] o (Kusano and Landau, 1972). This observation rules out the possibility that facilitation in the squid synapse is solely an artifact produced by low [Ca ++ ] o solutions. As an adjunct to another study on depression, Kusano and Landau (1975) also studied facilitation in ++ the squid synapse at low [Ca ] o . Since these authors stimulated at fairly long intervals (20 msec), they were almost certainly observing the second component of facilitation and its decay. This facilitation (Kusano and Landau, 1975) was only 0.14 after four stimuli and decayed exponentially with a half decay time of 20 msec at 16-18°C (time constant of decay about 29 msec), a value which agrees well with our measurements of time constants of 22-63 msec for the second component of decay in Loligo giant synapses at 20°C. The giant synapses had significantly more facilitation and a faster time constant of decay than did non-giant synapses (Table I). Although these two synapses are structurally dissimilar at the gross level (Young, 1939, 1973), there is no a priori reason why their facilitation properties should be different. When more than one synapse of a pre-giant cell was tested, facilitation parameters were remarkably similar (Figs. 4,8) even though the diameters of the terminals varied by as much as threefold. These results are in contrast to the findings in some crayfish motorneurons where synapses of the same pre-synaptic cell on different post-synaptic cells can exhibit quite different properties (Bittner, 1968; Linder, 1974; Bittner and Sewell, 1976). Growth of Facilitation During Repetitive Stimulation Repetitive stimulation of pre-synaptic axons to the stellate ganglion usually (cf Fig. 2) produced the increases in facilitation predicted by the linear summation model (Figs. 6-8) devised by Mallart and Martin (1967) to describe facilitation in the frog neuromuscular junction. The results are generally consistent with the hypothesis that (a) each action potential of a train produces an equal amount of facilitation which (b) decays in a manner similar to that following a single action potential and (c) adds linearly to that produced by previous action potentials. This simple model appears to apply quite well to synapses such as those in frog (Mallart and Martin, 1967; Magleby, 1973 a) and squid which release large quantities of transmitter at normal calcium concentrations and which usually induce action potentials in post-synaptic cells. In contrast, synapses in crab and crayfish muscles (a) normally release far less transmitter per impulse than squid or frog svnapses, (b) usually induce only graded post-synaptic responses and (c) exhibit summation of facilitation very different from that predicted by the linear summation model (Linder, 1973, 1974; Zucker, 1974; Bittner and Sewell, 1976). (My experiments examined only the summation of facilitation due to the first components of facilitation (F , . With longer durations of stimulation, the analysis would have been complicated by the second component of facilitation (F , T ) and by 2 2 the possible introduction of additional potentiating factors such as those reported by Mallart and Martin (1967) and Magleby (1973 a b) at frog synapses.) The Effect of Temperature on Facilitation While the production of facilitation at zero time (Fp is not affected by changes in temperature/ the decay of facilitation pro-' ceeds at a slower rate as temperature is lowered (Figs. 2,3). The extrapolated of the first component of decay (T ) is about two, a value which is approximately one-half that found by Balnave and Gage (1974) in the toad neuromuscular junction. Since most diffusion processes have a less than two (Giese, 1962), it is unlikely that the decay of facilitation is brought about solely by diffusion of an activating agent (such as calcium ion) away from its site of action. Although is not significantly different at 20°C and 15°C, the observed facilitation at 5-30 msec after the conditioning pulse is usually larger at the lower temperature because the time constant of decay is greater (Figs. 2,3). Temperature decreases in other synapses (Hubbard et al., 1971; Ward et al., 1972) have been associated with decreases in the probability of transmitter release. However in squid, transmitter release is fairly constant for decreases in temperature of 5°C, probably because of increases in the duration and amplitude of presynaptic action potentials (Fig. 5). Thus, during rapid repetitive activation of synapses at low temperatures, when mobilization and transmitter-release probability might be expected to be decreased (Hubbard et al., 1971), these two mechanisms (slower decay of facilitation and increase in action potential amplitude and duration) might serve to maintain transmitter release at a high level at some synapses in ectothermic animals. However, the adaptive significance of facilitation in the squid giant fiber system is not clear. Since the giant synapse has a safety factor for transmission of action potentials of about 5 (Katz and Miledi, 1967; personal observations) at normal [Ca ] o , transmitter release would have to be drastically reduced to cause failure of transmission. Furthermore, since the cycling rate of mantle contractions (and presumably giant fiber action potentials) is about 1 Hz (Arnold et al., 1974), the facilitation described here would not be expected to sum and affect transmitter release. Mechanisms Producing Facilitation There are many ways in which facilitation at synapses could be produced. It is possible, for instance, that facilitation in the squid synapses could be due to an increase in sensitivity of the post-synaptic receptor. This possibility has been ruled out in synapses in which one can record spontaneous miniature potentials thought to be caused by the liberation of individual quanta of trans mitter. In both frog (del Castillo and Katz, 1954) and crayfish neuromuscular junctions (Dudel and Kuffler, 1961; Bittner, 1968; Bittner and Harrison, 1970), the average amplitude of spontaneous miniature potentials does not change following a normal psp. It is thus unlikely that, in these preparations, facilitation is due to changes in receptor sensitivity. Since spontaneous miniature potentials are difficult to record in squid synapses (Takeuchi and Takeuchi, 1962; Miledi, 1967) it is not possible to completely rule out a post-synaptic mechanism for facilitation in this preparation. However, the many similarities between facilitation in squid synapses and facilitation in frog, toad and crayfish synapses make it highly unlikely that facilitation in the squid is due to a postsynaptic mechanism. First, the amount and time course of decay of facilitation is similar in squid, frog, toad and crayfish synapses (Table III). Second, the production of facilitation (F ) but not its decay (Tp is temperature compensated in both squid and toad synapses (Balnave and Gage, 1970, 1974), while the growth of facilitation is similar in squid and froo synapses (Mallart and Martin, 1967; Magleby, 1973 a). Furthermore, a post-synaptic mechanism for facilitation would have to differ at giant and non-giant synapses on the same post-synaptic cell since the time course for facilitation at these synapses is different (Table I). Since facilitation appears to be dependent on the presence of extracellular calcium ions (Katz and Miledi, 1968; Younkin, 1974), it has been proposed that some of the calcium which enters a nerve terminal during an action potential lingers in the terminal and thus increases the probability of transmitter release to a subsequent action potential (Katz and Miledi, 1968; Rahamimoff, 1968, 1973; Miledi and Thies, 1971; Rahamimoff and Yaari, 1973). The amount of facilitation detected by the second of a pair of action potentials would then reflect the amount of Ca ++ admitted by the first action potential and its decrease in concentration with time. Some authors have also indicated that the residual calcium theory predicts that increase in [Ca ++ ] o should be accompanied by decreases in facilitation This result has been found in frog synapses at certain levels of transmitter release and at certain stimulus intervals (Rahamimoff, 1968, 1973; Mallart and Martin, 1968). Conversely, there was an increase in facilitation in four of six squid experiments in which [Ca ++ ]© or [Mn ++ ] o were varied to give two to twelvefold increases in psp amplitude. Furthermore, changes in [Ca ++ ] 0 seem to have little or no affect on facilitation in crayfish or crab neuromuscular junctions (Zucker, 1974 a; Linder, 1973). These data indicate that if residual calcium is responsible for facilitation, its mechanism of action is quite different in different synapses. Facilitation could also be due to a secondary release of calcium from intracellular stores such as mitochondria and subsurface cisternae which are known to have calcium binding sites (Landis et al., 1973; Llinas and Nicholson, 1975; Baker et al., 1971). A similar mechanism may cause release of calcium from sarcoplasmic reticulum in muscle cells (Ford and Podolsky, 1970; Endo et al., 1970) and from photoreceptor outer segments (Hendricks et al., 1974). Such a mechanism might be dependent on extracellular calcium and could produce the second component of facilitation (the hump) by a delayed release of Ca ++ . This secondary release of Ca might be small at t = 0 and then accelerate to a maximum after which Ca ++ could be removed by an uptake mechanism to yield the slow decay of facilitation. Recently/ Stinnakre and Taue (1973) found that successive action potentials in neuron cell bodies of Aplysia admit increasing quantities of Ca (as determined by light emission from the Ca - dependent photoprotein/ aequorin) and have suggested that this mechanism might be responsible for facilitation of transmitter output (see also Zucker, 1974 a). The facilitated light output was accompanied by progressive increases in the peak voltage of the action potentials, presumably due to increases in calcium conductance. Although progressive increases in peak voltage of action potentials are not seen in the squid pre-synaptic terminal, this observation does not rule out changes in Ca ++ conductance as a possible facilitation mechanism since Ca conductance in squid • '■f* "4" terminals is small compared to conductances of Na and K and would be expected to have little effect on membrane voltages (Katz and Miledi, 1969 b). Finally, facilitation at the squid synapse could be produced by an increase in the amplitude of the second of a pair of presynaptic action potentials (Takeuchi and Takeuchi, 1962). In my experiments, both facilitation and the difference in amplitude between the first and second action potentials were greatest at short stimulus intervals and both increased as temperature was lowered (Fig. 5). Increases in action potential amplitude might be important in some cases since the relation between depolarization and transmitter release can be quite steep at the peak of normal action potentials (Katz and Miledi, 1970). The next chapter examines the possibility that changes in pre-synaptic action potentials cause facilitation and it is shown that these changes are not sufficient to explain all the facilitation. Synapse n T F X ±(SD) T 1(SD) Time to °C msec msec L. brevis Giant 9 20° 1.51(1.2) 5.11(1.9) 15-20 Non-giant 13 20° 0.9±(0.9) 10.61(4.4) 15-20 [Ca ++ ]o=2.5 - 5 mN, [Mn ++ ] o = 0 - 5 r nM giant > non-giant significant at .05 level giant < non-giant significant at .05 level ♦ Table I: continued Synapse n T T^±(SD) Time to T 2 °C msec msec L. pealei Giant 19 20° 1.2±(0.7) 6.5±(3.6) 10-15 8 15° 0.9±(0.5) 9.5±(2.8) 20-25 1 12° 1.6 11.9 Non-giant 3 20° 0.6±(0.1) 10.3±(2.9) 2 15° 0.7±(0.8) 12.3±(0.6) [Ca ++ ] Q = 3 - 6 mM, [Mn ++ ] Q = 0 - 6 mM giant > non-giant significant at .05 level giant < non-giant significant at .05 level 20°C < 15°C significant at .05 level Synapse T [Ca ++ ] o [Mn ++ ]o PSP F 1 T 1 - °C mM mM mV msec L. pealei Giant 19.1° 4 2 6.8 0.7 6.8 4 4 1.6 0.4 21.7 Giant 19.6° 6 2 6.9 2.7 (2.7) 3.7 (3.6) 4 2 1.6 0.9 (1.1) 5.8 (5.5) Giant 19.4° 5 3 8.0 0.7 6.5 5 5 4.0 0.7 6.1 L. brevis Non-giant 19.5° 4 0 1.6 0.5 10.0 4 5 0.9 0.7 10.0 Non-giant 20.5° 4 0 1.8 3.8 4.3 2 0 .15 1.1 4.7 Giant 17.5° 3.5 0 ——— 0.9 9.7 2.5 0 — 1.4 8.3 Preparation Temp. °C F 1 m 1 1 Authors FROG - NMJ 22 1.27 35 Mallart and Martin, 1967 FROG - NMJ 20 0.8 50 Magleby, 1973a FROG - NMJ 19-23 0.7-1.1 12 Rahamimoff, 1968 TOAD - NMJ 21 2.0 34 Balnave and Gage, 1970 TOAD - NMJ 19 2.7 45 Balnave and Gage, 1974 GUINEA PIG - SCG 37 0.4 200 McLachlan, 1975 CRAB - NMJ-FA 8 2.9 8.5 Linder, 1973 CRAB - NMJ-SA 8 2.9 12.4 Linder, 1973 CRAYFISH - NMJ-SC 20 3.3 19 Zucker, 1974a CRAYFISH - NMJ-SC 20 4.6 32 Bittner and Sewell, 1976 CRAYFISH - NMJ-SD 20 1.8 180 Bittner and Sewell, 1976 CRAYFISH - NMJ-SD 17 0.5-4.0 >160 Linder, 1974 CRAYFISH - NMJ-SC 17 2.7 20-30 Linder, 1974 MONKEY - CMS 37 0.85 10 Muir and Porter, 1973 SQUID - Non-giant 20 0.74 10.5 -average of both species SQUID - Giant 20 1.35 5.8 -average of both species Table III: Comparison of Initial Facilitation Parameters in Various Preparations. NMJ = Neuromuscular junction; SCG = Superior cervical ganglion; FA, SA = Fast, slow axon; SD, SC = Superficial distal and central muscle fibers; CMS - Corticomotoneuronal synapse CHAPTER II FACILITATION AND CHANGES IN PRE-SYNAPTIC POTENTIALS Introduction - Successive action potentials release increasing amounts of transmitter from pre-synaptic terminals. This phenomenon, called facilitation, has been described in the first chapter. In most synapses, facilitation has not been associated with changes in action potentials recorded extracellularly from the pre-synaptic terminal (see references in Zucker, 1974). However, an increase in the total amplitude of the second of a pair of action potentials has been observed in intracellular recordings from squid synapses and has been proposed to account for facilitation at these synapses (Takeuchi and Takeuchi, 1962; cf Miledi and Slater, 1966). This increase in total amplitude of the second spike at squid synapses could consist of an increase (hyperpolarization) in the membrane potential at the beginning (foot) and/or an increase in the depolarization at the peak (peak voltage) of the second action potential relative to the first. Hyperpolarization, for example, results in increased transmitter release by action potentials (Takeuchi and Takeuchi, 1962; Miledi and Slater, 1966; Bloedel et al., 1966) and hence facilitation might be due to successive action potentials arising from the hyperpolarization remaining from preceding action potentials (Dudel, 1971). Furthermore, since increases in peak depolarization are known to increase transmitter release (Bloedel et al., 1966), an increase in peak voltage in the second of a pair of action potentials could also cause facilitation. I have performed a set of experiments to dete.vmine if any of these changes which occur in normal action potentials could explain the observed facilitation. By the application of hyperpolarizing current to pre-synaptic terminals, I have shown that the afterhyperpolarization which normally follows an action potential cannot explain the observed amounts of facilitation which normally occur when these terminals are placed in low [Ca ++ ] o . Furthermore, the increase in peak voltage which normally occurs in the second action potential of a pair is shown to contribute little to twin-pulse facilitation. While it is not entirely certain whether changes in some combinations of these parameters might account for a significant portion of twin-pulse facilitation, it is clear that the growth of post-synaptic potentials (psp’s) and facilitation during trains of stimuli cannot be explained by progressive changes in any combination of peak voltage, total amplitude or hyperpolarization. In another set of experiments, I have attempted to determine the mechanism of action of artificial conditioning hyperpolarizations since this procedure has been shown to eliminate facilitation in squid synapses (Miledi and Slater, 1966) and to reduce the facilitation normally seen in rat neuromuscular junctions (Hubbard and Willis, 1962, 1968). I have found that the reduction of facilitation seen in squid pre-synaptic terminals that are artificially hyperpolarized can in large part be explained by a reduction in the amplitude of the "testing” pre-synaptic action potentials relative to the "conditioning" action potentials. Finally, it was of interest to determine whether the full voltage swing of an action potential is necessary to trigger facilitation. I report that, in contrast to evoked transmitter release, the facilitation produced by an action potential is barely affected by large reductions in total amplitude or peak voltage of the action potential. Materials and Methods All of these experiments were performed at the Marine Biological Laboratories, Woods Hole, Massachusetts, U.S.A., using the squid IL. pealei. Dissection techniques and other procedures were as described in the first chapter. Microelectrodes were placed into both pre-synaptic terminals and post-synaptic cells in their region of synaptic overlap. I used fresh, unfatigued synapses in low [Ca ++ ] o salines and added a small amount of Mn ++ to depress transmitter release. Data were collected and averaged by a computer of average transients (CAT, Mnemotron Corp.) and plotted on a chart recorder (Figs. 11, 15) or photographed directly from the oscilloscope screen (Figs. 10, 12-14). (Twenty or forty averaging sweeps were taken with 10 sec between sweeps). Facilitation (f) was defined as: v n - v o f = v o equation (1) , where V was the amplitude of the psp evoked by the first or conditioning pulse and was the amplitude of the nth psp. I used a modified Howland current pump (Fig. 9) to apply constant intracellular depolarizing or hyperpolarizing currents to nerve terminals or to initiate action potentials by short pulses of depolarizing current. This usage required that this device be able to pass square pulses of constant current through high-resistance microelectrodes in the presence of high circuit capacitance. In order to produce such currents, was trimmed to minimize current changes with load resistance changes while adjusted the relationship between control voltage ( v in ) and output current (I QUt )• The variable capacitor provided positive feedback and was adjusted for optimum frequency response, although the setting of determined to a great extent the amount of capacitance required in The power supply for the operational amplifier (Teledyne-Philbrick model 1022) was offset to allow large voltage swings for depolarizing current. Results Action Potentials During Pairs and Trains of Stimuli I have confirmed (Fig. 10) previous reports (Takeuchi and Takeuchi, 1962) that the second of a pair of action potentials in a squid pre-synaptic terminal,often has a greater total amplitude than the first. Takeuchi and Takeuchi (1962) proposed that this phenomenon could explain the observed facilitation of transmitter release; however, Miledi and Slater (1966) reported that facilitation in the squid synapse could be produced with no change or with a decrease in the total amplitude of the second of a pair of pre-synaptic action potentials. I have also confirmed these data but have noted that the greatest facilitation was usually produced at short intra-pair intervals (Fig. 15) during which total amplitude and peak voltage at the second action potential were maximally increased compared to the first spike. I have now extended these observations to include, for the first time, short trains of equal-interval stimuli. It was found that increases in psp amplitude and facilitation were never accompanied by progressive increases in any combination of total amplitude, peak voltage, level of after-hyperpolarization, and duration of action potentials after the second in the train. The peak voltages of the second and successive action potentials were somewhat higher than the peak voltages of the first action potential but did not continue to increase as the train progressed. In fact, increasing facilitation during trains was accompanied by decreases in the total amplitude of successive action potentials after the second pulse in the train (Figs. 11, 15). Hence, naturally occurring changes in any of these variables during trains of action potentials cannot be responsible for the observed facilitation. However, facilitation might be affected by voltage changes other than those listed above. Although there were no increases in duration or peak voltage of action potentials after the second pulse of a train, the voltage levels at the beginning of the rising phase (the foot) of successive action potentials occurring after the second in the train were all hyperpolarized with respect to the foot of the first action potential (Fig. 11). Membrane hyperpolarization decreased from a maximum value after the first pulse and some hyperpolarization existed during each interspike interval. Since there was some average value of hyperpolarization during trains of action potentials and since hyperpolarization acts with a slow time course to increase transmitter release (Miledi and Slater, 1966; Dudel, 1971), the possibility existed that the growth of psp’s during repetitive stimulation could have been due to a gradually developing effect of after-hyperpolarization following each spike. Furthermore, changes in the total amplitude or in the voltage level at the foot or peak of each action potential might have had independent effects on facilitation and each might have been affected in a rather complex fashion by the maintained after-hyperpolarizations. Consequently, I designed experiments to manipulate these variables in order to determine their relative effects on the ability of these synapses to release transmitter and to facilitate. Effect of Naturally Occurring After-hyperpolarizations on Facilitation When a pair of action potentials was elicited at a brief intra-pair interval, the foot of the second action potential began at a voltage slightly hyperpolarized with respect to the resting membrane potential. Since transmitter release has been shown to increase following a conditioning hyperpolarization in squid synapses (Takeuchi and Takeuchi, 1962; Miledi and Slater, 1966), the observed facilitation during twin-pulse or brief trains of stimuli having short intra-pulse intervals might have resulted from the hyperpolarization following the first pulse or from the maintained hyperpolarization during a train (Dudel, 1971). This hypothesis was tested directly by comparing the amplitudes of the psp evoked by the second action potential in a normal terminal (n*) with the first psp evoked by the action potential during a conditioning hyperpolarization (h) in the same terminal (Figs. 12, 13). Injection of hyperpolarizing current was begun more than 10 sec before action potentials were elicited; the current was maintained as long as action potentials were generated during a particular experiment using one stimulus interval (such an experiment could last as long as 5 minutes). No consistent differences were noted in the effects of hyperpolarization on trains of psp's recorded at various times during this maintained hyperpolarization. This observation indicated that the effects of hyperpolarization were constant after the first 10 sec had elapsed. The current-passing and voltagerecording electrodes were at opposite ends of the region of synaptic overlap with the recording electrode nearer the tip of the terminal (see Methods). Therefore, the average hyperpolarization given to the entire synaptic region was, due to spatial decrement of the imposed current, almost certainly more than the measured hyperpolarization indicated for each figure. The data indicate that artificially imposed hyperpolarizations can increase psp amplitude but that the naturally occurring hyperpolarizations following normal action potentials are not responsible for the facilitation observed in twin-pulse experiments. For example, the second normal spike (N*) in Figure 12 started from a potential 4mV hyperpolarized with respect to the first normal pulse and produced a psp only slightly smaller than the first psp produced by an action potential evoked after the terminal was artificially hyperpolarized by 13 mV. A similar relationship was seen in other preparations (compare (N’) and (H) in Fig. 13). In all cases, increasing amounts of conditioning hyperpolarizations gave increasing amounts of transmitter release to the first evoked impulse; this observation indicated that the effects of hyperpolarization were not maximal at the voltage levels which occurred after normal action potentials (Fig. 12). In order to make the psp following an artificial hyperpolarization equal to the psp produced by the second in a pair of normal action potentials, I had to use much more hyperpolarization than that which occurred following the first normal action potential. These results show that the hyperpolarization following a normal action potential was not responsible for most of the observed facilitation at the second pulse. Similarly, the data indicate that hyperpolarizations cannot account for the growth of facilitation which occurs during a train of pulses. If the maintained average hyperpolarization during a short train of action potentials acts with a slow time course to increase transmitter release, then the maximum amount of this effect should be mimicked by a maintained artificial hyperpolarization applied to the terminal. In the experiment of Figures 11c and lid, 24 mV of hyperpolarization was applied to a pre-synaptic terminal beginning 10 sec prior to stimulation. It was obvious that the first psp produced by an artificially hyperpolarized terminal (Fig. 11D) was smaller than the psp evoked by the last spike in a normal train (Fig. 11B), even though the artificial hyperpolarization (Fig. 11C) had a greater amplitude and duration than the summation of the naturally occurring after-hyperpolarizations following each spike in the train (Fig. 11A). Therefore, the data from these two experiments show that neither post-spike nor maintained hyperpolarization is responsible for much of the observed facilitation at the second or subsequent action potentials in normal terminals. Effect of Naturally Occurring Changes in Peak Voltage on Facilitation The peak voltage of an action potential has a powerful effect on transmitter release (Katz and Miledi, 1967) . Increases in this parameter in the second action potential of a pair or in successive action potentials of a train could be responsible for facilitation. Although there were no progressive increases in the peak voltages of action potentials in a train, I often observed a small increase in peak voltage at the second action potential of a pair. Hence, I attempted to determine whether this increase accounted for all the observed facilitation in twin-pulse experiments. For example, in one preparation shown in Table IV, a pair of normal action potentials (interval = 7 msec) of peak height 73 and 76 mV produced psp's of 1.2 and 2.2 mV respectively (facilitation = 0.81); that is, an increase of 3 mV in peak voltage was associated with an increase in psp amplitude of ImV in this normal terminal. However, when the peak voltage of an action potential was artificially increased to 83 mV by superposition of a depolarizing pulse, the resulting psp was only 1 mV larger than that produced by the 73 mV action potential. Thus, the increase in peak voltage in the normal pair of action potentials should not have been entirely responsible for all of the increased amplitude of the second psp. Furthermore, increases in the peak voltage of action potentials produced by depolarizing currents would have been greater at the current-passing electrode than at the voltage-sensing electrode due to spatial decrement of current along the length of the terminal. Consequently, the average amplitudes of the artificially increased action potentials in the whole terminal were probably greater than those presented in Table IV, whereas normal action potentials probably underwent little change in amplitude over the same length of the terminal. Therefore, small changes in peak voltage occurring in normal terminals would probably have had even less effect on psp amplitude than the effect presented for artificial pulses in Table IV. The results of these latter inaccuracies strengthen the hypothesis that changes in peak voltage at normal terminals cannot entirely account for facilitation. In fact, the data are consistent with the hypothesis that naturally occurring changes in peak voltage have relatively little effect on twin-pulse facilitation in normal terminals. In this same terminal, when depolarizing current pulses were applied during both action potentials, the psp at the second pulse was facilitated by 34% compared to the first pulse, even though the peak voltage at the second pulse was increased to only 82 mV compared to 83 mV at the first pulse (Table IV). These data again suggest that changes in peak voltage of several millivolts are not entirely responsible for twin-pulse facilitation. Effect of Naturally Occurring Simultaneous Changes of Both Hyperpolarization and Peak Voltage on Facilitation For technical reasons I did not obtain single records showing that a combination of the naturally occurring after-hyperpolarization and the naturally occurring increase in peak voltage of the second action potential in a pair could not account for all the twin-pulse facilitation. However, calculations made from data in Figures 12 and 16 and Table IV (nAP , nAP 2 ), all from the same preparation, showed that a combination of naturally occurring increases in hyperpolarization and peak voltage in the second action potential of a pair could only have produced about 50% of the observed facilitation. For example, in the pair of action potentials nAP. and nAP on Table IV, 1 2 the peak voltage of the first action potential was 73 mV, while the peak voltage of the second action potential was 76 mV. These action potentials produced psp’s of 1.2 and 2.2 mV, respectively (f = 0.81). An artificially increased action potential of 83 mV peak voltage produced a psp of 2.3 mV. The psp expected to be produced by the 76 mV action potential was found by interpolation on a graph of psp amplitude versus action potential peak voltage. The 76 mV action potential would have been expected to produce a psp of about 1.5 mV. Therefore, facilitation produced by the increase in peak voltage from 73 to 76 mV would have been f = (1.5/1.2) - 1 = 0.25 The foot of the second action potential in Table IV) was hyperpolarized BmV with respect to the resting potential. The effect of this naturally occurring hyperpolarization was determined from Figure 16. Since the average amplitude of psp’s had declined between the obtaining of the data of Figure 16 (Fig. 13) and the data of Table IV, we could not use the psp amplitudes indicated on Figure 16. Instead, we calculated the facilitation (fractional increase) in psp amplitude which would have been produced by an 8 mV hyperpolarization, that is, psp produced by an action potential hyperpolarized 8 mV psp produced by a normal action potential Facilitation calculated in this way was 0.16. Therefore, the total facilitation which could have been produced by the combination of the 8 mV hyperpolarization and the increase of 3 mV in peak voltage at was 0.25 + 0.16 = 0.41. The actual facilitation measured was 0.81. Therefore, if it is assumed that the facilitatory effects of the normally occurring after-hyperpolarization and the increase in peak amplitude at the second action potential simply sum, they should have produced at most about 50% of the observed facilitation. Similar data were obtained in another preparation. Effect of Conditioning Hyperpolarizations on Facilitation When artificial hyperpolarization was applied to pre-synaptic terminals, all psp's became larger. However, the amount of facilitation was decreased compared to that produced when the same terminals did not receive a conditioning hyperpolarization (Figs. 11 and 15). The decrease in facilitation was greater when greater conditioning hyperpolarizations were used (Figs. 12 and 13), but facilitation was never entirely abolished, even when terminals were artificially hyperpolarized as much as 60 mV (Fig. 14, cf Miledi and Slater, 1966). Obviously, facilitation could still occur even though the total amplitude of both action potentials in a pair had been increased. The reduction of facilitation by conditioning hyperpolarizations was also greater at short intra-pair stimulus intervals than at longer intervals (Fig. 15) . Psp’s continued to grow when trains of stimuli were given to artificially hyperpolarized terminals; the rate of increase of facilitation (but not its magnitude) appeared similar to that which occurred in normal terminals (Figs. 11, 15). It should also be noted that, as in normal terminals, facilitation increased during repetitive stimulation despite progressive decreases in peak voltage and postspike hyperpolarization. Since the direction of these changes in peak voltage and post-spike hyperpolarization would be expected to reduce transmitter release, this result again implies that some factor other than these voltage changes must have been responsible for the observed facilitation. Mechanism of Action of Conditioning Hyperpolarization on Facilitation If it could be shown that conditioning hyperpolarization per se had a direct effect to reduce facilitation, then this fact would have to be incorporated into models of the mechanism of facilitation. I therefore attempted to ascertain whether the effect of conditioning hyperpolarization on other variables (psp amplitude, peak voltage or the voltage at the foot of the second action potential) could account for the observed reduction in facilitation. It was immediately obvious that all of the psp's produced by artificially hyperpolarized terminals were larger than those produced by the same terminal when it was not hyperpolarized (Figs. 11-16). Since facilitation is reduced at higher levels of transmitter output in frog neuromuscular synapses (Mallart and Martin, 1968; Rahaminoff, 1968), the possibility exists that a similar increase in psp amplitude might account for all of the decreased facilitation in artificially hyperpolarized terminals. However, three lines of evidence indicate that this hypothesis is not correct. First, the experiments were done in low [Ca ++ ] o and the amount of transmitter released by hyperpolarized terminals was only a small fraction of that released at normal [Ca ++ ] o . Hence, there should have been little depletion of the transmitter available for release by subsequent action potentials. Second, as preparations became equilibrated to low [Ca ++ ] o , facilitation at a given interval was often larger at higher levels of transmitter release in non-hyperpolarized terminals than at the lower levels of release produced by the same terminals w’hen artificially hyperpolarized following equilibration. Third, when transmitter release was increased by superimposing depolarizing current pulses of various strengths and durations on normal action potentials, facilitation was not reduced as much as when conditioning hyperpolarizing currents were used to increase transmitter release. For example, in one preparation 24 mV of hyperpolarization caused an increase in psp amplitude from 1.2 to 2.6 mV and a reduction in facilitation at twin-pulse intervals of 7 msec from 0.81 to 0.09 (Table IV). However, an increase in psp amplitude from 1.2 to 2.6 mV by the addition of 27 mV of depolarization to the normal action potential reduced the facilitation from 0.81 to 0.31 (Table IV). Hence, the increase in psp amplitude which accompanied conditioning hyperpolarization should not have been responsible for all of the decrease in facilitation seen in hyperpolarized terminals. However, in this particular experiment the increased psp amplitude might have accounted for over half of the reduced facilitation. Compared to the peak voltage reached by normal action potentials, the peak voltage of the first action potential of a pair or train (Figs. 11, 12, 14) was usually increased by conditioning hyper polarization. This effect was presumably due to a decrease in sodium inactivation (Hodgkin and Huxley, 1952). However, the peak voltages of the second and subsequent action potentials were about the same as the peak voltages reached by terminals without conditioning hyperpolarization and hence the former were reduced with respect to the peak voltage reached by the first action potential in an artificially hyperpolarized terminal. This reduction in peak voltage of the second or successive action potentials, relative to the first, in hyperpolarized terminals might have reduced transmitter release and caused the decrease in facilitation compared to that found in normal terminals. For example, during conditioning hyperpolarization (Table IV), the peak voltage of the second action potential was reduced from 77 to 73 mV compared to the peak voltage reached by the first action potential; this reduction was associated with a facilitation of 0.09 compared to 0.81 at the same terminal without conditioning hyperpolarization. However, changes in the peak voltage of action potentials in either normal or artificially hyperpolarized terminals were generally only about 2-5 mV and, as shown earlier, such changes probably had little effect on transmitter release (see previous section on naturally occurring changes in peak voltage). In any event, it should be noted that a decrease of total amplitude of the testing action potential during conditioning hyperpolarization was still associated with some facilitation (0.09). This observation again suggests that increases in total spike amplitude in testing action potential were not fully responsible for facilitation. There was one other obvious difference between action potentials in artificially hyperpolarized terminals and those in nonhype rpolarized terminals. Following an action potential in artificially hyperpolarized terminals-? - the membrane potential took longer to return to its previous level than in normal terminals (Figs. 12-14). At 15°C, the membrane potential regained its maximum level exponentially with a time constant of 17 msec after an action potential reached an artificially hyperpolarized terminal (Figs. 12- 14). In the non-hyperpolarized condition, the resting potential was regained exponentially with a time constant of 5-6 msec. Due to the slow return of hyperpolarization, successive action potentials in artificially hyperpolarized terminals began at progressively less hyperpolarized voltages (Figs. 12-14). Since the amount of transmitter release was a function of the level of hyperpolarization (Takeuchi and Takeuchi, 1962; Miledi and Slater, 1966), I attempted to determine by a graphical method whether the loss of hyperpolarization in the second or successive action potentials in an artificially hyperpolarized terminal could be responsible for the decrease in facilitation in these terminals. The effect of conditioning hyperpolarization of transmitter release was determined at a few different levels of artificial hyperpolarizations measured from records of single oscilloscope sweeps (Figs. 12, 13) or computer averages of 20-40 pulses. Both methods gave similar results. From the relationship between conditioning hyperpolarization and psp amplitude shown in Figure 16 (data taken from Fig. 13), I estimated an expected value for the second psp. This value was obtained by finding (on Fig. 16) the psp amplitude which would have been produced by an action potential in which the voltage at the foot was the same as that of the second action potential of the hyperpolarized pair. Facilitation was then calculated as defined in equation 1 except that V ecualled the amplitude of n the observed second psp and V equalled the amplitude of the expected o psp. When facilitation was calculated in this manner in four different preparations, hyperpolarized action potentials were shown to produce between 60% and 100% of the facilitation produced by nonhyperpolarized action potentials (Fig. 15). Furthermore, most of the deficit in facilitation during short trains of stimuli in hyperpolarized terminals could also be accounted for by this calculation (Figs. 15A, 15B). Therefore it appears that much of the decrease in facilitation seen in artificially hyperpolarized terminals could be explained by the decrease in total amplitude of the testing pulses relative to the first pulse. (The increase in psp amplitude at artificially hyperpolarized terminals may also cause some of the decrease in facilitation.) The Effect of Artificial Changes in Peak Voltage on Facilitation It is known that transmitter release can be initiated by presynaptic depolarizations which are much smaller than those of normal action potentials (Bloedel et al., 1966; Katz and Miledi, 1967). Since it was of interest to determine what aspects the mechanism of facilitation had in common with the basic mechanism of transmitter release, I attempted to determine whether facilitation could also be initiated with small depolarizations. One method of reducing the amplitude of action potentials was to artificially depolarize the membrane by a few millivolts prior to the initiation of an action potential (Takeuchi and Takeuchi, 1962; Miledi and Slater, 1966). Figure 10 shows that facilitation was increased when the peak voltage of both action potentials of a pair was reduced about 10 mV by a small maintained artificial depolarization. In another preparation, the peak voltage of the action potentials was reduced from 76 mV above resting potential to 60 mV above resting potential by the application of a conditioning depolarization. In this experiment, the amplitude of the first psp of a pair given 7 msec apart was reduced from 1.27 mV to 0.72 mV but facilitation was unchanged. In another kind of experiment, pairs of depolarizing pulses were delivered to a terminal poisoned by tetrodotoxin to eliminate action potentials (Fig. 17). Facilitation was produced by pairs of artificial depolarizations of as little as 21 mV above resting potential measured at the recording electrode. While I do not know the value for the space constant at this terminal, it seems likely that much of the terminal was depolarized to a voltage less than that found during a normal action potential. Hence, these data suggest that facilitation can be produced by depolarizations that are smaller than those generated by normal action potentials. Facilitation, although reduced, was substantial when the total amplitude of action-potentials was greatly increased by depolarizing pulses (Table IV) . All of these data indicate that action potentials which are much smaller or larger than normal can still produce facilitation. The Effect of Transmitter Release on Facilitation During twin-pulse stimulation at low levels of transmitter release, facilitation is not affected by failure or success of the first action potential to release transmitter at frog neuromuscular synapses (del Castillo and Katz, 1954). Using this same preparation, but somew 7 hat different experimental paradigms, Mallart and Martin (1968) and Rahamimoff (1968) showed that facilitation decreased when quantal content of psp’s was increased by increases in [Ca ++ ] o or decreases in [Mg‘ + ] o . This apparent paradox was studied further in the squid giant synapse. As reported earlier (Chapter 1) in experiments on squid synapses (Table II), increases in [Ca ] o or decreases in [Mn ] o led to increases in psp amplitude of up to twelvefold and were associated with increases in the amount of facilitation produced by a conditioning pulse (four out of six experiments). These experiments were performed after synapses had been equilibrated to changed ionic concentrations. In other experiments, I tried a slightly different approach by testing for facilitation at particular twin-pulse intervals as psp amplitude decreased while the synapse equilibrated to reduced [Ca ++ ] o . When facilitation was measured in this way in ten synapses at twin-pulse intervals of 6, 7 and 10 msec, four synapses showed decreases, four synapses showed increases and two synapses showed no change in facilitation with decreases in psp amplitude. These and all previously published experiments except those of del Castillo and Katz (1954) have involved the effects of changes in both divalent cation concentration and psp amplitude on facilitation. In a different kind of experiment, I have now studied the effect on facilitation of systematic variations in psp amplitude caused by manipulation of pre-synaptic depolarization levels of conditioning pulses at a constant divalent cation concentration. In one preliminary experiment (Table IV) facilitation decreased when presynaptic depolarization and psp amplitude were increased. The results of another kind of experiment are detailed in Figures 18 and 19. In this experiment a variable amplitude pulse of depolarizing current was injected into a pre-synaptic terminal 8 msec before a pair of normal action potentials (intra-pair interval = 6 msec) was elicited by extracellular stimulation. This stimulus paradigm thus produced three psp’s in which the first psp varied in amplitude according to the amplitude of the conditioning depolarizing pulse. The variable depolarizing pulse was considered to be a variable conditioning pulse, while the two action potentials were standard test pulses. I then compared the amplitudes of the psp’s evoked by the two action potentials with the amplitude of a control psp elicited by a single action potential. Facilitation at each test psp was measured as test psp (mV) psp (mV)“ 1 ’ The facilitation detected by the second test pulse (B in Fig. 19) consisted of the facilitation remaining from that produced by the variable conditioning pulse and the facilitation produced by the first test pulse. The data show that facilitation detected by the first test pulse (A in Fig. 19) increased with increases in conditioning psp amplitude from 0.4 to 6 mV but declined as conditioning psp amplitude approached 12 mV. Furthermore, changes in the facilitation at the second pulse seemed to parallel changes in the first pulse. It is remarkable that facilitation was virtually identical at conditioning psp amplitudes of 0.4 mV and 12 mV (thirtyfold difference) . While the interpretation of this experiment is difficult since both conditioning depolarization and psp amplitude varied together, it is apparent that facilitation can be produced when very small amounts of transmitter are released (the psp of 0.4 mV was probably only 1/90 of the amplitude of psp’s which would be produced at normal [Ca ++ ] o ). Discussion The exact configuration of pre-synaptic action potentials has a great influence on the amount of transmitter released by a presynaptic terminal. Since increases in peak voltage, total amplitude or duration of pre-synaptic action potentials all act to increase the amount of transmitter released by a pre-synaptic terminal (Takeuchi and Takeuchi, 1962; Miledi and Slater, 1966; Bloedel et al., 1966; Katz and Miledi, 1967), changes in these parameters have been suspected to cause facilitation (Takeuchi and Takeuchi, 1962; Dudel and Kuffler, 1961; Dudel, 1971). Other theories for the mechanism of facilitation involve increasing degrees of terminal invasion by action potentials (Dudel and Kuffler, 1961; Bittner, 1968) or slow depolarization of pre-synaptic terminals (Lang and Atwood, 1973). The results of these experiments indicate that changes in nerve terminal potentials cannot entirely explain the phenomenon of facilitation and that there must be some other mechanism or mechanisms responsible for facilitation. During twin-pulse facilitation, the second of a pair of action potentials usually had a larger total amplitude than the first. This difference in amplitude was largest at short intervals and at lower temperatures when facilitation was greatest. The increase in amplitude was due to the after-hyperpolarization remaining from the first action potential and an increase in peak voltage of the second action potential. I will now consider the likelihood that either of these changes are involved in twin-pulse facilitation or in the growth of facilitation during trains. Naturally Occurring After-Hyperpolarizations It has been shown quantitatively (by application of artificial hyperpolarization in experiments such as those in Figures 13 and 14) that the naturally occurring hyperpolarization at the foot of the second action potential was much smaller than that w T hich would be required to cause the observed facilitation. This conclusion can be firmly stated since, in contrast to the naturally occuring hyperpolarization which could only act for a few milliseconds, the large artificial hyperpolarizations in these experiments were begun 10 sec before action potentials were elicited. Hyperpolarization acts with a very slow time course to increase transmitter release (Miledi and Slater, 1966; Dudel, 1971), so the effect of the artificially applied hyperpolarization would have been maximal while the effect of the naturally occurring hyperpolarization could not have been maximal. Furthermore, due to spatial decrement of current, the average amount of hyperpolarization was considerably greater than that which is indicated in Figures 13 and 14. The result of this inaccuracy is that the amount of artificial hyperpolarization required to increase transmitter release by a given amount would be greater than indicated in Figures 12, 13 and 16 and there would be less likelihood that the naturally occurring hyperpolarizations could account for much of the observed facilitation. During trains of action potentials (Fig. 10) there was a decrease in the hyperpolarization at the foot of each action potential —a change which would only be expected to reduce transmitter release. Nevertheless, it could be argued that the average hyperpolarization during a train could increase transmitter release by a time-dependent mechanism and thus produce facilitation (Dudel, 1971). This hypothesis was not substantiated in my experiments. The largest facilitated psp in normal terminals exceeded the amplitude of a psp produced by the same terminals when they were artificially hyperpolarized over 20 mV for 10 sec (Fig. 11). Thus, a greater-thannormal level of hyperpolarization applied for a longer-than-normal time did not cause an increase in transmitter release as great as that which occurred in the normal terminal during repetitive stimulation. These data show that the naturally occurring hyperpolarization during a train of action potentials could not have produced the observed facilitation. Changes in Peak Voltage of Pre-synaptic Action Potentials By artificially increasing the peak voltage of pre-synaptic action potentials, I have shown that the small increase in peak voltage in the second of a pair of action potentials cannot explain the naturally occurring facilitation (Table IV) . This conclusion should be quite reliable since most of the terminal was actually depolarized to a greater extent than shown in Table IV. Furthermore, progressive increases in peak voltage or duration did not occur during trains of pre-synaptic action potentials. In fact, in some experiments there were small decreases in peak voltage during a train. Increases in peak voltage are obviously not involved in facilitation during trains. Combined Effects of Naturally Occurring Hyperpolarizations and Increases in Peak Voltage When the effects of both naturally occurring hyperpolarization and increases in peak voltage were combined, I calculated that the sum of both effects acting together could not produce the observed facilitation at the second pulse in a pair. This is a quantitative demonstration that changes in action potentials during twinpulse facilitation cannot account for the observed facilitation. These results are consistent with the demonstration that twin-pulse facilitation in the squid giant synapse (Miledi and Slater, 1966) and in the chick ciliary ganglion (Martin and Pilar, 1964) is not necessarily accompanied by action potential changes which would tend to increase transmitter release. However, Miledi and Slater (1966) used normal [Ca ++ ] o salines and rapid repetitive stimulation to depress transmitter release to subthreshold levels. It is possible that Miledi and Slater (1966) did not study facilitation but rather measured recovery from depression during the twin-pulse intervals they used. This explanation might account for the fact that the time course of the facilitation reported by Miledi and Slater (1966) differed greatly from the time course of the facilitation which I and others have reported (Table III). In contrast to the work of Miledi and Slater (1966), my experiments employed conditions which should have enhanced the detection of facilitation (Mallart and Martin, 1967; Kusano and Landau, 1975), and there is no doubt that the conclusions reached here actually apply to the phenomenon of facilitation. I have show with intracellular techniques, that the growth of psp’s during short trains of stimuli is not due to progressive increases in duration, peak voltage, level of hyperpolarization or total amplitude of spikes during trains of pre-synaptic action potentials.' In fact, the total action potential amplitude and inter-spike hyperpolarization declined as psp's increased. Many studies in other preparations have indicated that increases in psp amplitude and facilitation during trains of stimuli are not associated with changes in the amplitude or duration of extracellularly recorded pre-synaptic action potentials (Hubbard and Schmidt, 1963; Katz and Miledi, 1965; Braun and Schmidt, 1966; Linder, 1973; Zucker, 1974 b). However, the interpretation of extracellularly recorded action potentials is difficult since these potentials do not indicate slow changes in membrane resting potentials or changes in peak voltage of action potentials. According to Katz and Miledi (1965 a extracellular potentials only register local membrane current proportional to the second derivative of intracellular action potentials. My direct recordings of intracellular action potentials during repetitive stimulation circumvent the difficulties associated with extracellular records and leave no doubt that there are no progressive changes in action potentials which could cause facilitation during repetitive stimulation. The Effect of Artificial Hyperpolarization on Facilitation When terminals were artificially hyperpolarized, facilitation of psp’s was drastically reduced. 2 Most of the decrease in facilitation in artificially hyperpolarized terminals could be explained by decreases in peak voltage and by decreases in the hyperpolarization at the foot of the second or successive action potentials. However, not all the loss of transmitter at the second psp is explained by calculations such as those from the data of Figure 16 Some inaccuracies in this calculation are to be expected since the entire terminal was not iso-potential during the conditioning hyperpolarization experiments. Thus, the values plotted on Figure 16 are representative only of the voltages at the recording electrode while the average hyperpolarization in the whole terminal would have been larger than indicated. Therefore the curve in Figure 16 should actually be shifted to the right (higher levels of hyperpolarization for a given psp amplitude) and the differences between the observed and expected second psp’s in Figure 16 increased. This would have the effect of increasing the amount of facilitation calculated to occur at the second artificially hyperpolarized action potential. The increase in transmitter release due to hyperpolarization takes several seconds to develop fully (Miledi and Slater, 1966; Dudel, 1971) and the effect might be partially abolished by the occurrence of an action potential. If this were true, not only would the amount of hyperpolarization be less at the second and successive action potentials but also its effect could be less. This factor would also decrease the transmitter released by the second and successive action potentials and explain why the calculated loss of transmitter output was too little in some cases. I therefore wish to emphasize that calculations made in a similar way as those in Figure 16 can almost restore the twin-pulse facilitation curve of an artificially hyperpolarized cell to its normal parameters (Fig. 15). This result indicates that neither the amount of facilitation produced nor its decay with time were drastically affected by artificial hyperpolarization. Furthermore, summation of facilitation was not drastically affected by artificial hyperpolarization (Figs. 15A, 15B). It thus appears that conditioning hyperpolarization may reduce the expression of facilitation but does not necessarily affect the mechanisms which produce facilitation. Other Manipulations Facilitation also appeared to be unaffected by maintained artificial depolarization of terminals (Fig. 10). It is thus unlikely that facilitation is directly related to the increase in sodium conductance or to the influx of sodium that occurs during action potentials since sodium conductance is partially inactivated by conditioning depolarization (Hodgkin and Huxley, 1952). This conclusion is strengthened by the fact that I (Fig. 17) and others (Bloedel et al., 1966; Katz and Miledi, 1967) have found that facilitation can still occur in the presence of tetrodotoxin, a poison which eliminates the voltage sensitive sodium conductance. It is also evident that action potentials need not attain their normal peak voltage or total amplitude to produce facilitation (Fig. 10) and that even small depolarizations can elicit facilitation (Fig. 17). I was not able to determine if facilitation varied continuously with the peak voltage or total amplitude of the first presynaptic action potential of a pair. The results of one experiment (Figs. 18 and 19) certainly do not rule out the possibility that the production of facilitation is a graded event dependent on the amplitude of pre-synaptic depolarization in the conditioning pulse. Mechanism of Facilitation In the first part of this chapter it was shown that voltage changes in pre-synaptic action potentials were neither necessary nor sufficient for the production of the observed facilitation of transmitter release. While the data clearly indicate that voltage changes do not cause all the facilitation, they do not rule out the possibility that changes in certain ionic currents could produce facilitation. A change in a particular membrane current which was small in relation to other membrane currents would not be expected to affect the shape or amplitude of action potentials (Katz and Miledi, 1969 b). If facilitation is produced by changes in a particular ionic current in successive action potentials, this current must have the following attributes. First, the proposed facilitation current must not be activated by hyperpolarization since this treatment does not increase facilitation (Fig. 15) . Second, the proposed facilitation current must not be inactivated by maintained depolarization since this does not reduce facilitation (Fig. 10). Third, the facilitation current might not have a voltage threshold since facilitation can be produced by very small artificial depolarizations (Fig. 17) and by action potentials which are much smaller than normal (Fig. 10). Fourth, the facilitation current is not sensitive to tetrodotoxin since facilitation can be elicited in synapses poisoned by this chemical. Of all the voltage sensitive conductances known to exist in squid pre-synaptic terminals only the calcium conductance has all of the above attributes (Katz and Miledi, 1971; Llinas and Nicholson, 1975). Thus, the data are consistent with a hypothesis for the mechanism of facilitation in which calcium conductance is increased in successive action potentials and the resultant increase in calcium entry causes an increase in transmitter release (Stinnakre and Taue, 1973; Zucker, 1974 a) . This hypothesis is the result of the observation by Stinnakre and Taue (1973) that successive action potentials in Aplysia neuron cell bodies admit increasing amounts of calcium. While none of the data provide a critical test of the calcium conductance hypothesis for facilitation, some of the data are supportive of this hypothesis. At present there are no data which relate the shape of the twin-pulse facilitation decay curve (Fig. 1) or the summation of facilitation curves to the calcium conductance mechanism for facilitation. This would be a fruitful area for future study since information of this kind would suggest a causal relation ship between increases in calcium conductance and facilitation. inaccuracy due to spatial decrement would have been much larger for depolarizing current than for hyperpolarizing current, since the former was applied when membrane conductance was high (during an action potential), while the latter actually tended to reduce membrane conductance and hence reduce spatial decrement (Katz and Miledi, 1967; personal observations). 2 ln contrast to the results of Miledi and Slater (1966), the second psp of a pair produced by a hyperpolarized terminal was never smaller than the first. This disparity between these data may again be explained by the differences in experimental paradigms noted earlier. I used low [Ca ++ ] o salines to reduce transmitter release to subthreshold levels while Miledi and Slater (1966) used repetitive stimu lation at normal [Ca ++ ] o to depress transmitter release. Since transmitter stores may have been partially exhausted in Miledi and Slater’s terminals, there may not have been sufficient transmitter available to allow any facilitation when the amplitude of the first psp was increased by conditioning hyperpolarization. AP Conditioning Hyperpolarization (mV) Total Amplitude (mV) Peak Voltage Above Resting Potential (mV) PSP (mV) f nAPl 0 73 73 1.2 0.81 nAP2 0 84 76 2.2 dAPl 0 83 83 2.3 0.34 dAP2 0 90 82 3.1 dAPl 0 101 101 2.6 0.30 dAP2 0 109 103 3.4 hAPl 24 100 77 2.6 0.09 hAP2 94 73 2.9 hAPl 24 100 76 2.7 0.08 hAP2 98 77 2.8 Table IV: Changes in action potentials, post-synaptic potentials and facilitation. (nAP), normal action potentials; (dAP), action potentials with increased peak voltage due to superimposed depolarizing current pulses; (hAP) , action potentials elicited during 24 mV of conditioning hyperpolarization. Twin-pulse interval = 7 msec. Data from same preparation as shown in Figures 10, 13 and 16, but taken from averaging computer records obtained after Figures 10, 13 and 16. SUMMARY The occurrence of facilitation is almost as widespread in chemical synapses as is the quantal release mechanism (Table III). Facilitation thus is likely to represent a basic property of synaptic transmitter release. I have attempted to test two major theories for the mechanism of facilitation. In one theory, some of the calcium which enters the pre-synaptic terminal during an action potential is available to increase the probability of transmitter release by a second potential. I attempted to test this residual calcium theory by changing the extracellular calcium concentration and measuring changes in facilitation. The results of these experiments were inconclusive and neither supported nor refuted the residual calcium theory. Indeed, as noted in the discussion in Chapter 1, experiments of this nature can give varying results in synapses of a single species and contradictory results in synapses of other species. It appears that either changing extracellular calcium concentration does not always affect the amount of residual calcium or that residual calcium is not the major mechanism responsible for facilitation. It must also be pointed out that experiments involving changes in [Ca ]o cannot actually demonstrate that residual calcium exists or that residual calcium enhances transmitter release. Indeed, as Katz and Miledi (1965 b have noted, it is extremely difficult to design a test for the residual calcium hypothesis. My data do not substantiate models for facilitation in which a single activating compound or process declines with a dual exponential time course following an action potential. Such a model cannot account for the delayed increases in facilitation seen in the twin-pulse decay curves. There must be more than one process respon sible for facilitation. A second major class of theories invokes changes in presynaptic action potentials to account for facilitation. According to these theories/ increases in peak voltage, total amplitude, hyperpolarization or duration of pre-synaptic action potentials would produce facilitation. I have tested these theories in two ways. First, I recorded pre-synaptic action potentials during facilitation. Although some changes in pre-synaptic action potentials occurred during twin-pulse facilitation, there were no progressive changes in pre-synaptic action potentials which would have increased transmitter release during repetitive stimulation when psp amplitude increased. Second, by manipulation of membrane potentials, I was able to show that action potential changes which did occur were not sufficient to account for facilitation. Since changes in pre-synaptic action potentials were neither necessary nor sufficient to account for facilitation, some other mechanism must have produced facilitation. Since the natural behavior of pre-synaptic action potentials provided little positive information on the mechanism of facilitation, I tried to perturb the facilitation system. I found that conditioning hyperpolarization reduced the expression of facilitation but not its production. In other experiments, I showed that maintained depolarization did not reduce facilitation at particular twin-pulse intervals. Finally, I was able to demonstrate that facilitation could be produced by action potentials which did not reach their normal peak voltages and also that facilitation could be produced by small depolarizations in the presence of tetrodotoxin. These data provide the first supportive information for a theory of facilitation involving increases in calcium conductance in successive action potentials. This theory arose from the observation (Stinnakre and Taue, 1973) that successive action potentials in a nerve cell body could admit increasing quantities of calcium. Similar data concerning calcium fluxes in nerve terminals are not available but our data are consistent with this theory. In synapses which do not transmit in a one-to-one manner,, the post-synaptic cell integrates post-synaptic currents. If all psp’s produced by a pre-synaptic cell were of the same amplitude, the relation between pre-synaptic firing frequency and post-synaptic depolarization would be linear. However, if facilitation occurred and psp amplitude increased with pre-synaptic firing frequency, the same pre-synaptic cell could control a greater range of post-synaptic depolarizations than it could without facilitation. The adaptive significance of facilitation in the squid giant synapse is not clear since this synapse normally transmits in a one-to-one manner and the time constant of decay of facilitation is about two orders of magnitude smaller than the interval between action potentials in a giant fiber of a swimming squid. FIGURES Figure 1. Facilitation in a giant synapse of L. brevis. The decay of facilitation (f) is plotted using semi-logarithmic coordinates against the interval between conditioning and test stimuli (t) . In this case the decay of the first component of facilitation is described by f = 1.3 exp (-t/f) where 1.3 = (the intercept with the ordinate) and 5 = (the time constant of decay). Insert shows a plot of a computer average of 20 post-synaptic potentials with an interval of about 10 msec. Calibration pulse: 2 mV, 2 msec. [Ca ++ ] o =4 mM, 18®C. Figure 2. (A) Decay of facilitation at 15°C (triangles, filled circles) and 12°C (open circles) in a giant synapse of pealei. Experimental sequence was 15° 12° -> 15° (filled circles). (B) Summation of facilitation (triangles) during trains of stimuli at intervals of 6 msec (At = 6) and 10 msec (At = 10) during the first 15°C experiment above. Solid lines represent growth of facilitation predicted by the linear summation model (equation 5) using from (A). (A) and (B) from same preparation. [Ca ++ ]o = 5 mM, [Mn ++ ]o = 4 mM. Figure 3. Decay of facilitation in an L. pealei giant synapse at 20°C (filled circles) and 15°C (open circles). At 20°C, F 1 = 1.5, Tg = 4 msec. At 15°C, Fg = 0.54, Tg = 15 msec. Psp amplitude at both temperatures was 1.8 mV. [Ca ++ ]o=s mM, [Mn ++ ]o = 4 mM. Figure 4. Decay of facilitation in three synapses of the same pre-synaptic giant cell in L. pealei at two different calcium concentrations (A, B) . Psp’s were recorded simultaneously in synaptic regions of the seventh (circles), fourth (triangles) and third (squares) post-synaptic giant cells (gf 7, gf 4, 3) . (A) [Ca ++ ]o = 6 mM, [Mn ++ ]o = 2 mM, Fp = 2.7, Tp = 3.66, 19.6°C (B) [Ca ++ ]o = 4 mM, [Mn ++ ]o = 2 mM, Fp =l, Tp = 5.63, 19.6°C gf 7 gf 4 psp amplitude (A) 6.9 mV 1.5 mV psp amplitude (B) 1.6 mV 0.83 mV Figure 5. Computer average of 40 pairs of action potentials (interval = 10 msec) in a L. pealei pre-synaptic terminal at 15°C and 12°C. The first action potential at 15°C has been aligned so that its resting potential and initial rise are coincident with those of the first action potential at 12 °C. The line labelled Er indicates the original resting potential; the two lines at the top of the figure have been aligned with the peaks of the first action potentials at 12°C and 15°C. The peaks of the first and second action potentials at 12°C were 71.7 mV and 75.2 mV (difference - 3.5 mV) above Er while the peaks at 15°C were 66.8 mV and 68 mV above Er (difference = 1.2 mV). The first psp was 2.2 mV at 15°C and 1.5 mV at 12°C. Same preparation as Figure 2. SA = Stimulus artifact; F = beginning (foot) of second action potentials. Calibration = 2 mV, 2 msec. Figure 6. Decay (A) and summation (B) of facilitation in a L. brevis non-giant synapse in 4mM Ca ++ , 5mM Mn (filled circles) and 4mM Ca ++ , OmM Mn ++ (open circles). Stimulus interval in (B) was 5 msec (At = 5). Solid lines in (B) are the predicted curves from the linear summation model (equation 5) using Fp = 0.69, Tp = 10.1 at 5 mM [Mn ++ ] o = 0.51, Tp = 10.1 at OmM [Mn ++ ] o as found in (A). (A) and (B) from same synapse. 20°C. Figure 7. Decay (A) and summation (3) of facilitation in a L. brevis non-giant synapse. Solid lines in (B) are the predicted curves from the linear summation model (eguation 5) using F 1 = 0.51, T x = 20 found in (A) for stimulus intervals of 10 msec (lower) and 5 msec (upper). (A) and (B) from same synapse. [Ca ++ ]o=3 mM. 20°C. Figure 8. Summation of facilitation in two terminals of a pre-synaptic giant cell of L. pealei in (A) 6 mM Ca ++ , 2mM Mn ++ and (B) 4mM Ca ++ , 2mM Mn ++ . Post-synaptic potentials were recorded from the seventh (gf 7) and fourth (gf 4) post-synaptic giant cells simultaneously. Same preparation as Figure 4. Solid lines indicate facilitation predicted by the linear summation model (equation 5) using parameters found in the experiments in Figure 4. (A) [Ca ++ ]o = 6 mM, [Mn ++ ]o = 2 mM, = 2.7, = 3.66, 19.6°C (B) [Ca ++ ]o = 4 mM, [Mn ++ ]o = 2 mM, Fj = 1.0, T T = 5.63, 19.6°C Figure 9. Schematic diagram of modified Howland constant current pump. The voltage at the input controlled the output current passing through the microelectrode. Output current was constant despite changes in microelectrode resistance. Modifications of the original circuit consisted of the addition of variable resistor and variable capacitor These additional components formed a variable positive feedback. network which enabled the operator to adjust the frequency response of the amplifier and pass constant current pulses through the microelectrode. R 2 was trimmed to minimize current fluctuations with changes in load resistance while adjusted the relationship between control voltage ) and output current (I ou t)• The variable capacitor provided positive feedback and was adjusted for optimum frequency response, although the setting of determined to a great extent the amount of capacitance required in The power supply for the operational amplifier (Teledyne-Philbrick model 1022) was offset to allow larger current voltage swings for depolarizing current than for hyperpolarizing current. Figure 10. (A) Oscillographic records of a pair of normal action potentials (N,N’) and a pair of reduced-amplitude action potentials (R,R‘) produced during a conditioning depolarization. (B) Pairs of psp's resulting from the action potentials in (A). Notice that the peak voltage of the second normal action potential (N 1 ) was somewhat higher than that of the first normal action potential (N) and that the foot of N’ was at a more negative voltage than the foot of N. After the normal action potentials were recorded, the terminal was depolarized 5-6 mV by the injection of steady depolarizing current into the terminal. When the depolarization had reached a steady level, two-action potentials (R,R*) were elicited by extracellular stimulation. The peak voltages of these action potentials were reduced about 10 mV (compared to the normal action potentials) and the foot of the second action potential (R’) was not hyperpolarized with respect to the foot of the first (R) . Psp’s produced by the reduced action potentials (R, R’) were smaller than those produced by normal action potentials (N,N‘), but facilitation was, if anything, increased. Calibration pulse preceding all pairs represents 2 mV, 2 msec. Twin-pulse interval =lO msec. [Ca ++ ]o = 5 mM, [Mn ++ ]o=4 mM, 15°C. Figure 11. Intracellularly recorded pre-synaptic action potentials (A, C) and psp’s (B, D) produced by repetitive stimulation (10 msec interval) in a normal terminal (A, B), and in the same terminal during the of 24 mV of conditioning hyperpolarization (C, D). The short dashed line in (A) and (C) represents the normal res-ting membrane potential. In (C) the membrane potential was hyperpolarized by 24 mV for 10 sec prior to the arrival of the first action potential. The hyperpolarization was maintained for more than 5 min while forty trains were elicited and averaged in a computer (the hyperpolarized level is represented by a long dashed line). All records represent the average responses to forty trains of stimuli and were traced from a chart recording of the computer output. Notice in both trains that psp amplitude continued to increase after the second spike, even though the total amplitude of action potentials decreased and the peak voltage and spike durations remained unchanged. Also note that the first psp produced by the hyperpolarized terminal (D) was smaller than the last psp produced by the normal terminal (B). Each train was preceded by a calibration pulse of 2 mV, 2 msec. Pulse intervals are 10 msec. [Ca ] o = 5 mM, [Mn ++ ]o=6 mM, 15°C. Figure 12. Action potentials (A) and psp’s (B) produced by a normal (N,N ! ) terminal and by the same terminal when artificially hyperpolarized by 13 mV H Hyperpolarization was applied at least 10 sec prior to the time when action potentials were elicited. Note that the foot of the second normal action potential (N‘) was hyperpolarized 4 mV with respect to the foot of the first normal action potential (N) and that the peak voltage of the second normal action potential (N*) was somewhat higher than that of the first normal action potential (N) . During conditioning hyperpolarization, the peak voltage of the first action potential was somewhat higher than that of the normal action potential (N) but this increase in peak voltage was not sustained at the second hyperpolarized action potential The hyperpolarization at the foot of action potentials and was less than at the foot of In (B) the peaks of the psp’s produced by the action potentials are identified with the same symbols as the action potentials. Note in (B) that psp was about the same amplitude as psp N’, and that facilitation decreased at greater levels of artificial hyperpolarization. Following the pair of action potentials, the membrane potential returned to its previous level at a much slower rate when artificially hyperpolarized H l7 ’) than when not artificially hyperpolarized. Figure 12, continued: Same preparation as in Figure 3. Calibration pulses at 2 mV, 2 msec precede each pair of potentials. Twin-pulse interval -- 7 msec. [Ca ++ ]o = 5 mM, [Mn ++ ]o = 6 mM, 15°C. Figure 13. Pairs of action potentials (A) and psp’s (B) elicited before (N, N’) and during H ll’' H lB' H 18*) con- ditioning artificial hyperpolarization of 11 mV and 18 mV . The stimulus paradigm was identical to that of Figure 4, except that the twin-pulse interval was 10 msec.. Note that the foot of action potentials H ’ and H ’ was less hyperpolarized ± 1 J_ o than the foot of action potentials H andH . As in Figure 4, 11 18 psp amplitude increased with conditioning hyperpolarization but facilitation decreased. Each pair of potentials is preceded by a calibration pulse of 2 mV, 2 msec, [Ca ++ ] o = 5 mM. [Mn ++ ] o = 4 mM, 15°C. Figure 14. Facilitation during a very large conditioning hyperpolarization. Pairs of action potentials (A) and psp's (B) were elicited before (N, N') and during (H^ Q , con ~ ditioning artificial hyperpolarization of 60 mV. Facilitation still occurred despite the high level of conditioning hyperpolarization. Notice that the conduction velocity of ths action potentials following conditioning hyperpolarization was slower than the velocity in normal terminals. Same preparation as in Figure 5. Twin-pulse interval = 10 msec. Calibration pulse = 2 mV, 2 msec, [Ca ++ ] 0 = 5 mM, [Mn ++ ] o =4 mM r 15°C. Figure 15. Decay (A) and summation (B, C) of facilitation before and during conditioning artificial hyperpolarization. In (A) the decay of facilitation (f) is plotted versus twinpulse interval. The solid dots represent the decay of facilitation in a normal terminal and the solid line represents the regression line through those points at twin-pulse intervals less than 20 msec. The squares in (A) represent the decay of facilitation in the same terminal during application of 20 mV of conditioning hyperpolarization as explained in the text. The interrupted line is the regression line through the squares at intervals less than 20 msec. The dotted line represents the decay of facilitation calculated to occur when allowance is made for loss of hyperpolarization at the second action potential of a pair (according to the method outlined in Fig. 16) . Note that the apparent loss of facilitation in the hyperpolarized terminal (A) was greater at short intra-pair stimulus intervals than at longer intervals. (B) and (C) represent summation of facilitation at stimulus intervals of 7 and 10 msec in the normal terminal (solid circles and triangles) and in the same terminal during 20 mV of conditioning hyperpolarization (squares) . The solid lines in (B) and (C) represent the growth of facilitation predicted by the linear summation theory assuming that and are described by the solid line in A. The open circles in (B) and (C) represent the summation Figure 15, continued: of facilitation during 20 mV of conditioning hyperpolarization when compensation was made for loss of hyperpolarization at the foot of action potentials as explained in text. All data represent computer averages of forty stimulus presentations. Same preparation as used, in Figure 4 [Ca ++ ] o = 5 mM, [Mn ++ ]o = 6 mM, 15°C. Figure 16. Effect of conditioning pre-synaptic hyperpolarization on psp amplitude and facilitation. Data are from Figure 5. The solid circles represent the amplitude of the psp’s produced by the first of a pair of normal action potentials (N, 0 mV hyperpolarization) and those produced by the first of a pair of action potentials given during conditioning hyperpolarization of 11 mV (H-q) and 18 mV (H^g). The solid curve was drawn through these points (solid circles) by eye and represents the probable relation between pre-synaptic hyperpolarization and psp amplitude for this synapse. The triangles represent the observed amplitude of the second psp in the pair (10 msec interval) produced by the normal terminal (N*) and by the same terminal during conditioning hyperpolarization of 11 mV and 18 mV The triangles have been placed over the level of hyperpolarization which was actually measured at the foot of the second action potential. For instance, at the foot of the second normal action potential (N 1 ) there was 2 mV of hyperpolarization (H ) while there was 9 mV (H ) of hyperpolarization at the foot of the second action potential in the terminal which had been conditioned with 11 mV of hyperpolarization. When the terminal was conditioned with 18 mV of hyperpolarization there was only 14 mV of hyperpolarization at the foot of the second action potential in the pair. The numbers beside the open circles represent the level of pre-synaptic hyperpolarization measured at the foot of the Figure 16, continued: second action potential in the terminal conditioned with 0 mV , 11 mV (Hg) , and 18 mV of hyperpolarization. These points (open circles) were used to predict the expected psp amplitude which would have been produced by a single (nonfacilitated) action potential if the terminal had been hyperpolarized by 2,9, or 14 mV. Facilitation was then calculated as the ratio (minus one) of the amplitudes of the actual second psp's (N’, to am P-itude of the (non- facilitated) psp's which would have been produced had the presynaptic hyperpolarization been that which actually occurred at the second action potentials (Hg, Hs, . For example the vertical distance between and represents the measured difference in transmitter release between the first (H, o ) and the second (H ’) psp's when the terminal had been conditioned with 18 mV of hyperpolarization. Facilitation was (H^ g ’ "1 = (4.7/4.5)-1 = 0.05. The vertical distance between H ' and H 1 o 14 represents the difference between the second psp in the terminal which had been conditioned with 18 mV of hyperpolarization and the expected (non-facilitated) psp amplitude had the terminal been hyperpolarized by only 14 mV which was the level of hyperpolarization which actually occurred at the foot of the second potential. The facilitation at the second pulse was then calculated as (H.'/H.. .)-l or (4.7/3.6)-1 = 0.32. io 14 Figure 16, continued: Similarly, when the terminal was conditioned with 11 mV of hyperpolarization, the first and second psp’s were 3.3 and 4.3 mV respectively and the facilitation was (4.3/3.3)-1 = 0.03. However, the hyperpolarization at the foot of the second action potential was only 9 mV and when facilitation was calculated as the ratio (minus one) between the second psp and the non-facilitated psp amplitude which would have been expected had the terminal been hyperpolarized only 9 mV, it was found that facilitation was: (Hll'/Hg)-! = (4.3/2.9)-l = 0.46. When no hyperpolarization was used, facilitation was: (N'/N)-l = (3.4/2.4)-l = 0.42. There was 2mV of hyperpolarization at the foot of the second action potential and when the increase in transmitter release due to this hyperpolarization was considered, the remaining facilitation was: (N'/H 2 )-l = (3.4/2.5)-l = 0.35. The data are taken from Figure 5. Data from other preparations yielded similar results. [Ca ++ ]o = 5 mM, [Mn ++ ] o = 4 mM, 15°C. Figure 17. Facilitation produced by small artificial depolarizations in the presence of tetrodotoxin (0.5 pgm/ml). Upper: pre-synaptic depolarizing pulses. Lower: psp’s produced by depolarizing pulses. SA = stimulus artifact. Calibration pulse = 2 mV, 2 msec. [Ca ++ ]o = 5 mM, [Mn ++ ]o = 0 mM, 16°C. Figure 18. Facilitation following variable pre-synaptic pulses. Explanation of traces: (a) the foot and peaks of psp’s are shown by arrows while other deflections represent stimulus artifacts and crosstalk between pre- and post-synaptic electrodes; (b) pre-synaptic depolarization and action potentials; (c) current injected into pre-synaptic terminal; (d) extracellular stimulation of pre-synaptic axon (twin-pulse interval = 6 msec). Traces (a) and (b) represent computer averages of forty responses and are traced from a chart recording of the /* computer output. Traces (c) and (d) have been added for clarity. In A and B the first psp was produced by a variableamplitude pre-synaptic depolarization while the second and third psp’s were produced by action potentials elicited by extracellular stimulation. In A, the first psp was about ImV and the pre-synaptic depolarization was slightly smaller than an action potential. In B, the first psp was about 3.8 mV and the pre-synaptic depolarization was larger than an action potential. In C, two psp’s were elicited by extracellular twin-pulse stimulation of the pre-synaptic axon (6 msec interval).. Several trials, similar to A and B, were performed using various amplitudes for the first pre-synaptic depolarization and for the first psp. The extracellularly evoked Figure 18, continued: action potentials were invariant and were considered to be standard testing pulses which tested for facilitation remaining after the variable ”conditioning”, depolarization pulse. In these trials, the facilitation which followed the first depolarization was determined by finding the ratio (minus one) of the amplitude of the second (or third) psp to the amplitude of the first (non-facilitated) psp in C. The results are plotted in Figure 11. Note that the second and third psp's in A are virtually the same amplitude as the second and third psp’s in B despite the fact that the first psp was much larger in B than in A. Calibration: 2 mV, 2 msec. [Ca +-H ] o = 5 mM, [Mn ++ ] o = 4 mM, 20°C. Figure 19. Facilitation following variable pre-synaptic depoL arizations. Results of the experiment in Figure 18. The amplitude of a variable "conditioning" psp is plotted versus the amount of facilitation which was detected by a standard, testing, action potential which arrived at the terminal 8 msec after the variable depolarization which produced the variable psp. The solid circles (A) represent the facilitation at the second pulse in trials similar to A and B in Figure 10 while the open circles (B) represent the facilitation at the third pulse. Facilitation was as defined in Figure 18. [Ca ++ ] o = 5 mM, [Mn ++ ] o = 4 mM, 20°C. Figure 1. Figure 2. Figure 3. Figure 4. Figure 5. Figure 6. Figure 7. Figure 8. Figure 9. Figure 10. Figure 11. Figure 12. Figure 13. Figure 14. Figure 15. Figure 16. Figure 17. Figure 18. Figure 19. BIBLIOGRAPHY Arnold, J. M< et al. 1974. A guide to laboratory use of the squid Loligo pealei. Published by the Marine Biological Laboratory, Woods Hole, Massachusetts 02543, U.S.A. Baker, P. F. et al. 1971. Depolarization and calcium entry in squid axons. J. Physiol., 218, p. 709-755. Balnave, R. J. and P. W. Gage. 1970. Temperature sensitivity of the time course of facilitation of transmitter release. Brain Res., 21, p. 297-300. Balnave, R. J. and P. W. Gage. 1973. The inhibitory effect of manganese on transmitter release at the neuromuscular junction of the toad. Br. J. Pharm., 47, p. 339-352. Balnave, R. J. and P. W. Gage. 1974. On facilitation of transmitter release at the toad neuromuscular junction. J. Physiol., 239, p. 657-675. Bittner, G. D. 1968. Differentiation of nerve terminals in the crayfish opener muscle and its functional significance. J. Gen. Physiol., 51, p. 731-758. Bittner, G. D. and J. Harrison. 1970. A reconsideration of the Poisson Hypothesis for transmitter release at the crayfish neuromuscular junction. J. Physiol., 206, p. 1-30. Bittner, G. D. and L. Sewell. 1976. Facilitation at crayfish neuromuscular junctions. In preparation. Bloedel, J. et al. 1966. Transmitter release at the squid giant synapse in the presence of tetrodotoxin. Nature, 212, p. 49-50. Braun, M. and R. F. Schmidt. 1966. Potential changes recorded from the frog motor nerve terminal during its activation. Pflugers Archiv., 287, p. 56-80. Charlton, M. P. and G. D. Bittner. 1974. Facilitation of transmitter release at the squid giant synapse. Biol. Bull., J 147, p. 471-472. del Castillo, J. and B. Katz. 1954. Statistical factors involved in neuromuscular facilitation and depression. J. Physiol., 124, p. 574-585. Dudel, J. 1971. The effect of polarizing current on action potential and transmitter release in crayfish motor nerve terminals. Pflugers Archiv., 324, p. 227-248. Dudel, J. and S. W. Kuffler. 1961. Mechanism of facilitation at the crayfish neuromuscular junction. J. Physiol., 155, p. 530-542. Endo, M. , M. Tanaka, and Y. Ogawa. 1970. Calcium induced release of calcium from the sarcoplasmic reticulum of skinned skeletal muscle fiber. Nature, 228, p. 34-36. Ferguson, T. S. 1967. Mathematical statistics: A decisive theoretic approach. Academic Press, N.Y. Ford, L. E. and R. J. Podolsky. 1970. Regenerative calcium release within muscle cells. Science, 167, p. 58-59. Giese, A. C. 1962. Cell Physiology. 2nd edition. W. B. Saunders Co., Philadelphia, Pa. Hendricks, J. et al. 1974. Biochemical aspects of the visual process. XXV. Light induced calcium movements in isolated frog outer segments. Biochim. Biophys. Acta, 345, p. 468-473. Hodgkin, A. L. and A. F. Huxley. 1952. The dual effect of membrane potential on sodium conductance in the giant axon of Loligo. J. Physiol., 116, p. 497-506. Hubbard, J. 1., S. F. Jones and E. M. Landau. 1971. The effect of temperature change upon transmitter release, facilitation and post-tetanic potentiation. J. Physiol., 216, p. 591-609 Hubbard, J. I. and R. F. Schmidt. 1963. An electrophysiological investigation of mammalian motor nerve terminals. J. Physiol., 166, p. 145-165. Hubbard, J. I. and W. D. Willis. 1962. Hyperpolarizat.'lon of mammalian motor nerve terminals. J. Physiol., 163, p. 115-137. Hubbard, J. I. and W. D. Willis. 1968. The effects of depolarization of motor nerve terminals upon the release of transmitter by nerve impulses. J. Physiol., 194, p. 381-407. Katz, B. 1969. The release of neural transmitter substances. Charles C. Thomas, Publisher. Springfield, Illinois, U.S.A. Katz, B. and R. Miledi. 1965 a. Propagation of electric activity in motor nerve terminals. Proc. Roy. Soc. 8., 161, p. 453-482. Katz, B. and R. Miledi. 1965 b. The effect of calcium on acetylcholine release from motor nerve terminals. Proc. R. Soc. 8., 161, p. 496-503. Katz, B. and R. Miledi. 1967. A study of synaptic transmission in the absence of nerve impulses. J. Physiol., 192, p. 407-436. Katz, B. and R. Miledi. 1968. The role of calcium in neuromuscular facilitation. J. Physiol., 195, p. 481-492. •. t Katz, B. and R. Miledi. 1969 a. The effect of divalent cations on transmission in the squid giant synapse. Publ. Staz. Zool. Napoli, 37, p. 303-310. Katz, B. and R. Miledi. 1969 b. Tetrodotoxin-resistant electric activity in pre-synaptic terminals. J. Physiol., 203, p. 459-487. Katz, B. and R. Miledi. 1970. Further study of the role of calcium in synaptic transmission. J. Physiol., 207, p. 789-801. Katz, B. and R. Miledi. 1971. The effect of prolonged depolarization on synaptic transfer in the stellate ganglion of the squid. J. Physiol., 216, p. 503-512. Kuno, M. 1964. Mechanism of facilitation and depression of the excitatory synaptic potential in spinal motoneurones. J. Physiol., 175, p. 100-112. Kuno, M. and J. N. Weakley. 1972 a. Facilitation of monosynaptic excitatory synaptic potentials in spinal motoneurones evoked by internuncial impulses. J. Physiol., 224, p. 271-286. Kuno, M. and J. N. Weakley. 1972 b. Quantai components of the inhibitory synaptic potential in spinal motoneurones of the cat. J. Physiol., 224, p. 287-303. Kusano, K. and E. M. Landau. 1972. Facilitation and depression of transmitter release in squid synapse. Biol. Bull., 143, p. 467. Kusano, K. and E. M. Landau. 1975. Depression and recovery of transmission at the squid giant synapse. J. Physiol., 245, p. 13-32. Landis, D. M. D. et al. 1973. Similar arrays of plasma membrane particles at subsurface cisterns in striated muscle and neurons. J. Cell Biol., 59, p. 184 a. Lang, F. and H. L. Atwood. 1973. Crustacean neuromuscular mechanisms: Functional morphology of nerve terminals and the mechanism of facilitation. Am. Zool., 13, p. 337-355. Lester, H. A. 1970. Transmitter release by pre-synaptic impulses in the squid stellate ganglion. Nature, 227, p. 493-496. Liley, A. W. 1956. The quantal components of the mammalian endplate potential. J. Physiol., 133, p. 571-587. Linder, T. M. 1973. Calcium and facilitation at two classes of crustacean neuromuscular synapse. J. Gen. Physiol., 61, p. 56-73. Linder, T. M. 1974. The accumulative properties of facilitation at crayfish neuromuscular synapses. J. Physiol., 238, p. 223-234. Llinas, R. and C. Nicholson. 1975. Calcium role in depolarizationsecretion coupling: an aequorin study in squid giant synapse. Proc. Nat. Acad. Sci., 72, p. 187-190. Magleby, K. L. 1973 a. The effect of repetitive stimulation on facilitation of transmitter release at the frog neuromuscular junction. J. Physiol., 234, p. 327-352. Magleby, K. L. 1973 b. The effect of tetanic and post-tetanic potentiation on facilitation of transmitter release at the frog neuromuscular junction. J. Physiol., 234, p. 353-371 Mallart, A. and A. R. Martin. 1967. An analysis of facilitation of transmitter release at the neuromuscular junction of the frog. J. Physiol., 193, p. 679-694. Mallart, A. and A. R. Martin. 1968. The relationship between quantum content and facilitation at the neuromuscular junction of the frog. J. Physiol., 196, p. 593-604. Martin, A. R. 1955. A further study of the statistical composition of the end-plate potential. J. Physiol., 130, p. 114-122. Martin, A. R. and G. Pilar. 1964. Pre-synaptic and post-synaptic events during post-tetanic potentiation and facilitation in the avian ciliary ganglion. J. Physiol., 175, p. 17-30. McLachlan, E. M. 1975. An analysis of the release of acetylcholine from preganglionic nerve terminals. J. Physiol., 245, p. 447-466. Meiri, U. and R. Rahamimoff, R. 1972. Neuromuscular transmission: Inhibition by manganese ions. Science, 176, p. 308-309. Miledi, R. 1967. Spontaneous synaptic/potentials and quantal release of transmitter in the stellate ganglion of the squid. J. Physiol., 192, p. 379-406. Miledi, R. 1973. Transmitter release induced by injection of calcium ions into nerve terminals. Proc. R. Soc. Bond. 8., 183, p. 421-429. Miledi, R. and C. R. Slater. 1966. The action of calcium on neuronal synapses in the squid. J. Physiol., 184, p. 473-498. Miledi, R. and R. Thies. 1971. Tetanic and post-tetanic rise in frequency of miniature end-plate potentials in low calcium solutions. J. Physiol., 212, p. 245-257. Muir, R. B. and R. Porter. 1973. The effect of preceding stimulus on temporal facilitation at corticomotoneural synapses. J. Physiol., 228, p. 749-763. Rahamimoff, R. 1968. A dual effect of calcium ions on neuromuscular facilitation. J. Physiol., 195, p. 471-480. Rahamimoff, R. 1973. Modulation of transmitter release at the neuromuscular junction. The Nejurosciences, Third Study Program, Schmitt and Worden. The Mir" Pres s, Cambridge, Massachusetts. Rahamimoff, R. and Y. Yaari. 1973. Delayed release of transmitter at the frog J. Physiol., 228, p. 241-257. Stirmakre, J. and L. Taue. 1973. Calcium influx in active Aplysia neurones detected by injected aequorin. Nature, New Biol., 242, p. 113-115. Takeuchi, A. and N. Takeuchi. 1962. Electrical changes in pre- and post-synaptic axons of the giant synapse of Loligo. J. Gen. Physiol., 45, p. 1181-1193. Ward, D. et al. 1972. Effect of temperature at the neuromuscular junction. Am. J. of Physiol., 222, No. 1, p. 216-219. Young, J. Z. 1939. Fused neurons and synaptic contacts in the giant nerve fibres of cephalopods. Phil. Trans. Roy. Soc. 8., 229, p. 465-503. Young, J. Z. 1973. The giant fiber synapse of Loligo. Brain Res., 57, p. 457-460. Younkin, S. G. 1974. An analysis of the role of calcium in facilitation at the frog neuromuscular junction. J. Physiol., 237, ,p. 1-14. Zucker, R. S. 1974 a. Characteristics of crayfish neuromuscular facilitation and their calcium dependence. J. Physiol., 241, p. 91-110. Zucker, R. S. 1974 b. Crayfish neuromuscular facilitation activated by constant pre-synaptic action potentials and depolarizing pulses. J. Physiol., 241, p. 111-126. The vita has been removed from the digitized version of this document.