- Email: [email protected]
Evidence for presynaptic facilitation in primary cochlear afferent neurons
Hearing Research, 29 (1987) 169-171 Elsevier
HRR
169
00951
Evidence for presynaptic facilitation in primary cochlear afferent neurons J.H. Siegel and E.M. Relkin * Auditory Physiology Lnboratory, Depts. of Communrcation Scrences and Disorders and Neurobiology and Physioloa, Northwestern University, Evanston, U.S.A. (Received
17 November
1986; accepted
2 April 1987)
Evidence for presynaptic facilitation was sought in the discharge patterns of single units in the chinchilla cochlear nerve. Pairs of acoustic clicks, separated by a variable interval, were delivered and spike discharge times stored for offline analysis. By choosing an appropriate binwidth (5 ms) and collecting data only from units with high characteristic frequencies (CF) the response to each click was contained in a single bin. The ratio of spike counts in the bins containing the responses to the two clicks was computed. For units with low spontaneous rates (SR) of discharge (SR < 8/s), an enhancement of the response to the second click was seen for low stimulus levels. As the stimulus level was raised, the response to the second click became reduced, presumably because of adaptation to the first click. Units with high (SR > 15 spikes/s) seldom exhibited enhancement of the response to the second click. The results are explained with a conceptual model in which two processes, depletion and facilitation decay exponentially following a stimulus. Since the two processes have opposite influences on the rate of transmitter release, the magnitudes of both processes may be underestimated by observing their net effect. Facilitation;
Auditory
nerve; Synaptic
transmission
Introduction Presynaptic facilitation is the increase in transmitter release by a chemically-transmitting cell brought about by prior stimulation. It is usually studied in isolated synapses by delivering pairs of shocks, separated by a variable time interval, to the presynaptic cell. While facilitation has been demonstrated in a variety of preparations (Del Castillo and Katz, 1954; Dude1 and Kuffler, 1961; Katz and Miledi, 1968; Rahamimoff, 1968; Magleby, 1973; Charlton and Bittner, 1978; Pamas, et al., 1982), it has not been conclusively demonstrated in sensory hair cell synapses, despite the widespread belief that the mechanism of trans-
Correspondence to: J.H. Siegel, Auditory Physiology Laboratory, Departments of Communication Sciences and Disorders and Neurobiology and Physiology, Northwestern University, Evanston, IL 60201, U.S.A. * Present address: Institute for Sensory Research, Syracuse University. Syracuse, NY 13244-5290, U.S.A. 0378-5955/87/$03.50
6 1987 Elsevier Science Publishers
mitter release by hair cells is similar to that of other synapses. There has been only indirect evidence from auditory psychophysics and population neural studies for facilitation-like phenomena (reviewed by Cacace and Smith, 1986). Depletion of readily-releasable transmitter quanta also controls the amount of transmitter released by the presynaptic cell during repetitive stimulation (Liley and North, 1953; Hubbard, 1963; Elmqvist and Quastel, 1965; Collins et al., 1984). Considering individual cycles of a tone-burst input as separate stimuli, Furukawa and Matsuura (1978) found that the amplitudes of successive excitatory postsynaptic potentials (EPSP) declined monotonically following the onset of the tone burst. This adaptation was successfully modelled as the depletion of presynaptic neurotransmitter quanta over the timecourse of the stimulus (Schroeder and Hall, 1974; Furukawa et al., 1978; Schwid and Geisler, 1982; Smith and Bra&man, 1982; Ross, 1982). The same phenomenon apparently underlies the adaptation in the discharge rate seen in responses of the sensory neurons of the goldfish
B.V. (Biomedical
Division)
170
saceulus (Kuno, 1983) and the mammalian cochlea (Smith, 1977). Since facilitation and adaptation simultaneously affect transmitter release (in opposite directions) it is possible that facilitation has not been apparent in hair cell afferent synapses because adaptation may dominate for stimuli commonly used to study afferent discharge patterns. Thus ~thou~ both processes may influence the response of the postsynaptic neuron, a net adaptation is usually produced. To test this hypothesis, we attempted to find stimulus conditions in which facilitation dominates. Pairs of acoustic clicks were chosen as stimuli, since this is most directly analogous to the inventions stimulus (i.e., pairs of electrical pulses delivered to the presynaptic cell) used in investigations of facilitation. Since we recorded spike discharges and not transmitter release directly, we could not infer the amount of transmitter released by each click, but only if enough transmitter was released to elicit a postsynaptic action potential. We reasoned that facilitated transmitter output should be expressed as an increased probability of discharge when the click level was such that the probability that either click elicited a spike was less than one. Under some stimulus conditions, units with low spontaneous rates (SR) of discharge did, in fact, respond with greater probability to the second click. A preliminary account of this work has been presented (Siegel and Relkin, 1984). Materials
and Mdods
We used a conventional approach to the eighth nerve through the posterior fossa of adult chinchillas. Data were collected from five animals. Glass ~cropi~t~s filled with 3 M KC1 were placed under visual observation on the cochlear nerve and advanced remotely with a stepping microdrive (Transvertex AB., Sweden). Characteristic frequencies (CF) were measured using an automated threshold tracking paradigm similar to that developed by Kiang and colleagues (I970). Single units with CF above 3 kI-Iz were selected because spikes elicited by low-level clicks fell within a 5 ms interval immediately following the stimulus. In no case did we accept data which did not meet this criterion. We developed a paired-click paradigm
analogous to the paired-pulse method used in other systems, replacing the usual presynaptic electrical stimuli with pairs of 100 ps acoustic clicks separated by a variable time interval (AT). The click level was usually adjusted so that the first click elicited a response in roughly half the presentations. This adjustment was made online by displaying each histogram with 5 ms bins along with the total spike count. Under these conditions, each of the two clicks presented during a given trial typically (i.e. 95% of all trials) elicited no more than one spike. With a fixed number of stimulus repetitions (usually 200~400) it was relatively easy to compute the number of evoked spikes per stimulus. The click level was then held constant, while click separation was varied. It was often possible to repeat the sequence at additional stimulus levels before contact with the fiber was lost. As the stimulus level was raised, an increasing number of trials resulted in multiple responses to a given click. The levels of the two clicks were always equal. During offline data analysis, PST histograms were formed with 5 ms bins such that the response to each click was contained within a single bin (Fig. 1). The relative frequency of response to each of the two clicks ( fi and _&) was simply taken as the spike count in the appropriate bin (n, and nz) divided by the number of stimulus presentations (N). The relative response measure was the ratio of the relative frequencies of response to the two clicks:
rW’3 =fdfi = w% We also examined the effect of a response to the first click per se on the relative frequency of response to the second click. Trials were selected from the total (N) in which there was no response to the first click (N’). For these selected trials, the conditional relative frequency of response to the second click (&‘) was calculated by dividing the number of response counts (ni) by the number of selected trials (N’). In particular, we wished to compare the relative frequencies of response to the two clicks when neither was immediately (i.e. within one stimulus repetition period) preceded by a response to a stimulus. Since no stimulus (and thus no response) immediately preceded the first click, the appropriate reference is fi. The condi-
171
12’ -AT-
I
Fig. 1. ~xpe~mental paradigm. Two sented at the indicated times separated construction of the peristimulus-time bins positioned such that the response pletely in one bin. The spike counts in and a2.
ciick stimuli were preby interval AT. Offline histograms used 5 ms to each click fell comthese bins are noted rzr
tional relative response is then:
r’@ll”) =f;/fi In the absence of a stimulus, some counts in the two 5 ms analysis intervals would be expected from spontaneous activity. For low-SR units the expected number of spontaneous spikes is a small fraction of the stimulus-evoked spikes (Fig. l), however, this may not be true for high-SR units. Because we studied primarily low-SR units, no correction was made for spontaneous activity in computing the response ratios. If spontaneous activity contributes equally to the two response windows, its effect would be to reduce the ratio between unequal click responses. This may explain why we found less compel~g evidence for facilitation in high-SR units than for low-SR units. Results
reliability of the measurements. Considerable variability from trial to trial was observed in both the total number of spikes and the ratio of responses to the two clicks. However, when a ratio substantially greater than one (i.e. f2/f2 > 1.2) was observed, it seldom became less than one on any subsequent trial. To demonstrate the statistical significance of individual data points would require a sufficient number of independent samples to characterize the variance of the underlying processes, since it may be dangerous to assume a particular stochastic model, particularly for very short counting intervals (Teich and Khanna, 1985; Young and Barta, 1986; Relkin, Unpubl. Obser.). When the separation between clicks (AT) was varied, the ratio usually remained greater than one for a considerable range of click spacing. A simple nonparametric sign test was used to establish the likelihood that such observations were simply due to chance.
Low-SR units For our sample of units with SR of 7.6 spikes/s or less, the response to the second click was usually larger than that to the first click at low stimulus levels, but this relations~p reversed as the stimulus level was raised (Fig. 2). For the lowest level stimulus condition in Fig. 2, all 9 click separations less than 100 ms resulted in click
1.51
I
5O
For most click levels, the response to the first click was greater than the response to the second click, resulting in ratios less than one. However, enhancement of the second click response could be seen in low-SR units at low stimulus levels. Repeated measurements were sometimes made with a given stimulus configuration to assess the
. 20
40 AT
1
60
80
(msec)
Fig. 2. Click response ratios obtained from a unit with low-SR (CF 5.9 kHz, SR 0 spikes/s). At the lowest click level ( - 60 dB Attn.), the ratio was greater than one for AT< 100 ms As click level was raised (-40 dB), ratios less (0 -•). than one were obtained for short click separations (W------ W. At the highest click level ( - 30 dB), the ratios were uniformly less than one (A4
ratios greater than one. Using a nonparametric sign test, the probability that this result is simply due to chance is only 0.2%. Contact with 25 units in this category was maintained long enough to present at least two runs of the paired click stimulus. Click response ratios of 1.2 or greater were seen in 15 (60%) of these units for at least one stimulus condition. Response ratios between 1.1 and 1.2 were seen in 9 (36%) of these units, but only one exhibited a maximum ratio less than 1.1. Enhancement of the response to the second click was seen in low-SR units for click separations as large as 75 ms and as short as 7.5 ms. In some units the ratio could exceed 1.5 (Fig. 3). Using the sign test, the probability that chance produced ratios greater than one for all stimulus conditions was calculated to be 0.8% and 1.6% for the -39 dB and - 34 dB stimuli, respectively. The ratio was usually maximal with click separations from 10 to 50 ms and close to one for click separations of 100 ms or more. The fact that the effects seen with short click separations disappeared as the click interval became larger also supports our claim that the patterns seen with shorter click separations were not due to chance. High-SR units Units with high SR (SR > 15 spikes/s) were deliberately selected against in this study. The spontaneous activity made it difficult to use audiovisual cues to initially set the click level when searching for the 50% response criterion.
1.5
0
20
40 60 AT (msecl
60
100
Fig. 3. Large response ratios obtained from a unit with low-SR -39 dB (CF 8.0 kHz, SR 3.0 spikes/s). Attenuation: (0 -0); - 34 dB (Dv.
.5
0
’
20
40 60 AT (msecl
.
60
11
100
Fig. 4. Click ratios from four units with high-SR (e0) CF 4.7 kHz, SR 30/s; (Dn) CF 3.7 kHz, SR 123/s; CF 16.8 kHz, SR 27/s; (A ---A) CF 5.7 (+ -4) kHz, SR 21/s. Ratios less than one were often seen at short click separations, but no pronounced trend was seen at longer separations.
Only seven units with SR of 18 spikes/s or more were studied. These units often showed ratios smaller than one at click intervals less than 10 ms (Fig. 4). The ratio was usually smallest at the shortest intervals between clicks. Although enhancement between 1.1 and 1.2 was seen in 5 of the 7 cases, the large variability between runs gave us little confidence in the significance of this result. Discussion Our measure of discharge probability with low-level clicks is not sensitive to changes in the synchrony of discharge within the analysis interval, since the 5 ms bins completely contained the response. The short duration of the click response of high-CF units minimizes the number of spontaneous spikes counted in the response window and allows click intervals as short as 7.5 ms to be evaluated. Studies of the transient response of the basilar membrane in the basal turn suggest that there should be little overlap between the responses to two clicks with separations of 7.5 ms or more (Robles et al., 1976). Although there have been no published reports of the intracellular responses of mammalian cochlear hair cells in the basal turn to acoustic clicks, it is probably a depolarizing pulse with a waveform similar to the
173
(rectified) envelope of the transient response of the basilar membrane. Judging from the absence of adaptation in the intracellular receptor potentials (Russell and Sellick, 1978; Goodman et al., 1982), the receptor potentials produced by the two clicks should be nearly identical. While most studies of presynaptic facilitation employ pairs of electrical shocks which initiate pairs of action potentials in the presynaptic neuron, our paradigm is directly analogous to the delivery of pairs of depolarizing current pulses to the presynaptic terminal in the absence of action potentials (Dude1 et al., 1983). We have recently reported intracellular recordings from afferent terminals presumably innervating IHCs (Siegel and Dallos, 1986; Siegel, 1986). Records of spontaneous activity show EPSPs which nearly always elicit action potentials. Since we have not yet observed EPSPs with pharmacolo~c~ blockage of the action potential, it has not been possible to directly assess the quanta1 content of these synaptic potentials. However, an analysis of the onset slopes of the EPSPs suggests that they are actually composed of integer multiples (up to three) of a unit quantum (Siegel, 1986). Even though this analysis has only been performed on spontaneous activity, it seems likely that stimuli such as low-level acoustic clicks should evoke coordinated quantal transmission. Furthermore, spatial and temporal integration of independently released quanta1 synaptic potentials by the postsynaptic neuron appears to play a relatively small role in this system. Since a spike may be initiated nearly every time the hair cell releases transmitter quanta, the observation of action potentials when recording from a single unit probably gives a reasonable measure of when release occurs. However, it remains to be seen if low-level stimuli can evoke subthreshold postsynaptic potentials. Thus, the absence of an action potential following a low-level click stimulus does not necessarily imply that transmitter was not released. In spite of these uncerta~ties, the ratios of click responses are likely to provide a reasonable estimate of the relative probability of transmitter release in response to the two clicks. However, since we cannot directly assess quanta1 release by recording action potentials, the accuracy of the ratios as a measure of relative transmitter
release must be questioned. On the other hand, this measure seems a reasonable indicator of the existence of differences in release, since the probability of discharge should be directly related to the amount of trans~tter released (Runo, 1983). Accordingly, we conclude that si~ificant enhancement of the response to the second of two clicks is evidence for the facilitation of transmitter release. This conclusion is strengthened by our observation that enhancement of the response to the second click was present when there was no response to the first click (Fig. 5). Axons have been shown to exhibit a period of reduced threshold and increased propagation velocity (supernormality) following an electrically evoked action potential (reviewed by Swadlow et al., 1980). If the enhancement we have observed were due to this postsynaptic mechanism, then the conditional response ratio should not exhibit explement. In fact, the enhancement was greater for the conditioned measure than for the unconditioned measure. Since we could only detect propagating action potentials, it is possible that subthreshold transmitter release can also lead to postsynaptic supernormality, though there is no evidence for this from other systems. Presynaptic facilitation of trans~tter release
I
I
1.5.
5.I
20 a
40 I
AT
60 1
.
80 I
100 .I
(msec)
Fig. 5. Effect of a response to the first click on the response ratio. The data from the -60 dB condition of Fig. 2 are replotted (O---0) along with the ratio of the conditioned frequency of response to the second click and the unconditioned frequency of response to the first click (o- -0). The conditioned ratio was greater than the unconditioned ratio for short click separations.
has been shown in other systems to be maximal at short interstimulus intervals, where the ratio of postsynaptic potentials is typically 1.5-2 : 1 (Del Castillo and Katz, 1954; Dude1 and Kuffler, 1961; Katz and Miledi, 1968; Rahamimoff, 1968; Magleby, 1973; Charlton and Bittner, 1978; Pamas, et al., 1982; Charlton et al., 1982). Facilitation declines monotonically over tens to hundreds of milliseconds as the interval between stimuli is seen in increased. Similarly, the adaptation cochlear afferent neurons, presumably resulting from transmitter depletion, has its maximal effect at short intervals following the adapting (in this case a forward masking) stimulus (Smith, 1977; Harris and Dallos, 1979). The recovery from adaptation also exhibits time constants in the range of 15-40 ms. Since these two factors have opposite effects on transmitter release, the magnitudes of both phenomena are likely to be underestimated by any measure of transmitter output. A simple model can be used to illustrate this point. The recovery from both depletion and facilitation are represented as exponentially decaying functions of opposite polarity: Facilitation :
f(t)
=&pm
Depletion :
d(t)
= A, . &r/r,)
Net response:
r(r) = f(t) -d(t)
Where A, and Tr represent the initial amplitude and decay time constant, respectively, of facilitation. Similarly, A, and T, represent the initial amplitude and recovery time constant, respectively, of depletion. The range of behavior seen in our experiments can be qualitatively described by adjusting the weighting and time constants of the functions (Fig. 6). Two points may be emphasized: (1) the general shape of our data is best reproduced when the amplitude and time course of the two functions are similar; (2) the initial magnitude of either function may be substantially greater than the resultant at any time. Although both factors are simultaneously present, one or the other process may dominate at a given time. The observation that facilitation was greater when there was no response to the first click (Fig. 5) suggests that there was less transmitter released in response to the first click, and hence less depletion, in those selected trials.
Fig. 6. Conceptual model of the interaction between two exponentially decaying functions representing the recovery from depletion (d(t)) and the decay of facilitation (f(t)) following a stimulus. The net response (r(r)) depends on the initial amplitudes and time constants of the two processes.
With appropriate click level and separation, facilitation was clearly evident in low-SR units. At short interclick intervals, dominant adaptation was commonly seen, but facilitation was apparent for intervals as long as 75 ms. As the stimulus level was raised, adaptation became increasingly dominant. We may thus suggest, based on our two-click paradigm, that depletion of transmitter stores is not the only presynaptic factor influencing the dependence of afferent discharge on prior history. It is reasonable to assume that facilitation also influences the response of the cochlear afferent neurons to other stimuli such as tone bursts, particularly at stimulus onset. Just how significant this effect may be under different stimulus conditions remains to be established, but both additional experimental and theoretical treatment of this question is warranted. Though we have neglected refractoriness in this discussion, this postsynaptic phenomenon undoubtedly influenced our results, especially for click separations less than 7.5 ms (Gaumond et al., 1982). However, since refractoriness should affect the ratio of click responses in the same direction as depletion, ne-
175
glecting refractoriness underestimates the relative importance of facilitation when compared to depletion. The observation that high-SR units did not consistently exhibit facilitation may be explained if the number of transmitter quanta immediately available for release at the IHC receptoneural junction is small and, as a result, easily depleted. Facilitated transmitter release in response to the second click is presumably not expressed if the readily-releasable store of transmitter was substantially depleted by the first click. Thus, the fact that facilitation was not consistently seen in highSR units may indicate that the synaptic junctions supplying these fibers are in a relatively depleted state. This result is analogous to the rapid decline in postsynaptic potentials in nerve-muscle preparations when a tetanic stimulus train elicits a large release of transmitter (Liley and North, 1953; Hubbard, 1963; Elmqvist and Quastel, 1965; Collins et al., 1984). Afferent inputs to motoneurons in the cat have been shown to exhibit facilitation or depression depending on the initial amplitude of the EPSP {Collins et al., 1984). Thus, the idea that synapses with a small, readily depleted presynaptic store of trans~tter will express facilitation only under restrictive conditions may be generalized. Since we deliberately sampled primarily low-SR units and as a result collected considerably less data from high-SR units, a larger sample may have shown more prominent facilitation. However, we found that the compound neural response also exhibited a decrement, but never enhancement (not shown) with paired click stimuli identical to those which produced facilitation in low-SR units. Since high-SR neurons constitute the vast majority of spiral ganglion cells, they are expected to dominate the population response. As mentioned in Materials and Methods, spontaneous activity may also affect our measure of facilitation in high-SR units. Presynaptic facilitation has a variety of possible origins, but seems best explained by the so called ‘residual calcium’ hypothesis (Katz and Miledi, 1968). It has been established that calcium entering the presynaptic cell following a depolarization of the nerve membrane leads to transmitter release (Katz and Miledi, 1967; Llinas et al., 1981b). The action of calcium is ter~nated by its removal
from the cytoplasm by cytoplasmic calcium buffers and, ultimately, by active uptake by intracellular organelles or extrusion across the cell membrane. Since this clearing process taken time, some calcium may remain in the presynaptic cytoplasm when a second stimulated calcium influx occurs. The second transient rise in calcium concentration may thus exceed the first even when the amplitude of the successive depolarizations is the same. As transmitter release is directly related to presynaptic calcium concentration (Llinas et al., 1981a, b), more transmitter is liberated during the second stimulus. One extension of this hypothesis is that the response to the second stimulus may be facilitated even if calcium entry following the first stimulus does not evoke transmitter release. Our observation of facilitation in the absence of a response to the first click is consistent with this prediction. Most models of presynaptic facilitation assume implicitly that the store of available transmitter is infinite, so that depletion effects are ignored (Katz and Miledi, 1968; Pamas and Segel, 1980, 1981; Barton et al., 1983; Zucker and Stockbridge, 1983; Fogelson and Zucker, 1985; Cohen and Van der Kloot, 1986). This may be quite well justified in preparations like the neuromuscular junction where the store of presynaptic transmitter is large. Under the conditions of high concentrations of magnesium and low calcium commonly used to study quanta1 transmission in that preparation, the fraction of the available transmitter quanta released by a single electrical stimulus (typically less than ten) is several orders of magnitude smaller than the number of presumed release sites (Heuser et al., 1979; Walrond and Reese, 1985; Ko and Propst, 1986). On the other hand, models of synaptic transmission in the cochlea have focused on adaptation and thus emphasize depletion of transmitter stores (Schroeder and Hall, 1974; Furukawa et al., 1978; Schwid and Geisler, 1982; Smith and Brachman, 1982; Ross, 1982; but see Eggermont, 1985). It appears likely that models which include calcium entry and removal as well as depletion and restocking of transmitter stores will need to be developed to enhance the predictive power of such models as aids to more refined experiments. Even though such models would be quite complex, they
176
may provide a means of assessing the relative contributions of facilitation and depletion over a wide range of stimulus conditions. Similar models would, no doubt, also be useful in exploring other synapses where the presynaptic transmitter store is limited. Acknowledgements The authors wish to thank P. Dallos, M.A. Cheatham and K. Ohlemiller for their helpful comments on this manuscript. This work was supported by NSF grant BNS-8217273 and NINCDS grant NS-08635. References Barton, S.B., Cohen, I.S. and Van der KIoot, W. (1983): The calcium dependence of spontaneous and evoked quantal release at the frog neuromuscular junction. J. Physiol. (London) 337, 735-751. Cacace, A.T. and Smith, R.L. (1986); Some poststimulatory effects on the whole nerve action potential. Hear. Res. 23. 223-232. Charlton, M.P. and Bittner, G.D. (1978): Facilitation of transmitter release at squid synapses. J. Gen. Physiol. 72, 471-486. Charlton, M.P., Smith, S.J. and Zucker, R.S. (1982): Role of presynaptic calcium ions and channels in synaptic facilitation and depression at the squid giant synapse. J. Physiol. (London) 323, 173-193. Cohen, I.S. and Van der KIoot, W. (1986): Facilitation and delayed release at single frog neuromuscular junctions. J. Neurosci. 6,2366-2370. Collins, W.F. III, Honig M.G. and Mendell, L.M. (1984): Heterogeneity of group Ia synapses on homonymous amotoneurons as revealed by high-frequency stimulation of Ia afferent fibers. J. Neurophysiol. 52, 980-993. Del Castillo, J. and Katz, B. (1954): Quantal components of the end-plate potential. J. Physiol. (London) 124, 560-573. Dudel, J. and Kuffler, SW. (1961): The quantal nature of transmission and spontaneous miniature potentials at the crayfish neuromuscular junction. J. Physiol. (London) 155, 514-529. Dudel, J., Pamas, I. and Pamas, H. (1983): Neurotransmitter release and its facilitation in crayfish muscle. VI. Release determined by both, intracellular calcium concentration and depolarization of the nerve terminal. Pflligers Arch. 399, l-10. Eggermont, J.J. (1985): Peripheral adaptation and fatigue: A model oriented review. Hear. Res. 18, 57-71. Elmqvist, D. and Quastel, D.M.D. (1965): A quantitative study of end-plate potentials in isolated human muscle. J. Physiol. (London) 178, 508-529. Evans, E.F. (1975): CochIear nerve and co&ear nucleus. In:
Handbook of Sensory Physiology, pp. l-98. Editors: W.D. Keidel and W.D. Neff. Springer-Verlag, Berlin, Heidelberg, New York. Fogelson, A.L. and Zucker, R.S. (1985): Presynaptic calcium diffusion from various arrays of single channels. Biophys. J. 48, 1003-1017. Furukawa, T. and Matsuura, S. (1978): Adaptive rundown of excitatory post-synaptic potentials at synapses between hair cells and eighth nerve fibers in the goldfish. J. Physiol. (London) 276, 193-209. Furukawa, T., Hayashida, Y. and Matsuura, S. (1978): Quanta1 analysis of the size of excitatory post-synaptic potentials at synapses between hair cells and afferent nerve fibers in goldfish. J. Physiol. (London) 276, 211-226. Gaumond, R.P., Mohmr, C.E. and Kim, D.O. (1982): Stimulus and recovery dependence of cat ccchlear nerve fiber spike discharge probability. J. Neurophysiol. 48, 856-873. Goodman, D.A., Smith. R.L. and Chamberlain. SC. (1982): Intracellular and extracellular responses in the organ of Corti of the gerbil. Hear. Res. 7. 161-169. Harris, D.M. and Dallos, P. (1979): Foreward masking of auditory nerve fiber responses. J. Neurophysiol. 42. 1083-1107. Heuser, J.E., Reese, T.S.. Dennis, M.J., Jan, Y.. Jan, L. and Evans, L. (1979) Synaptic vesicle exocytosis captured by quick freezing and correlated with quantal transmitter release. J. Cell Biol. 81, 275-300. Hubbard, J.I. (1963): Repetitive stimulation at the mammalian neuromuscular junction and the mobilization of transmitter. J. Physiol. (London} 169, 641-662. Katz, B. and Miledi, R. (1967): The timing of calcium action during neuromuscular transmission. J. Physiol. (London) 189, 535-544. Katz, B. and Miledi, R. (1968): The role of calcium in neuromuscular facilitation. J. Physiol. (London) 195, 481-492. Kiang, N.Y.S., Moxon, E.C. and Levine, R.A. (1970): Auditory-nerve activity in cats with normal and abnormal co&leas. In: Sensorineural Hearing Loss, Ciba Symposium, pp. 241-273. Editors: G.E.W. Wolstenholme and J. Knight. J. and A. Churchill, London. Ko, C.-P. and Propst, J.W. (1986): Freeze-fracture of physiologically identified neuromuscular junctions from single frog muscle fibers. J. Elect. Micr. Tech. 4, 147-156. Kuno, M. (1983): Adaptive changes in firing rates in goldfish auditory fibers as related to changes in mean amplitude of excitatory postsynaptic potentials. J. Neurophysiol. 50. 573-581. Liley, A.W. and North, K.A.K. (1953): An investigation of effects of repetitive stimulation of the mammalian neuromuscular junction. J. Neurophysiol. 16, 509-527. Llinas, R., Steinberg, I.Z. and Walton, K. (1981a): Presynaptic calcium currents in squid giant synapse. Biophys. J. 33, 289-322. Llinas, R., Steinberg, I.Z. and Walton, K. (1981b): Relationship between presynaptic calcium current and postsynaptic potential in squid giant synapse. Biophys. J. 33. 323-352. Magleby, K.L. (1973): The effects of repetitive stimulation on facilitation of transmitter release at the frog neuromuscular junction. J. Physiol. (London) 234. 327-352.
Magleby,
K.L. (1979): Facilitation,
augmentation,
and potenti-
ation of transmitter release. Prog. Brain Res. 49, 175-182. Parnas, H. and Segel, L.A. (1980): A theoretical explanation for some effects of calcium on the facilitation of neurotransmitter release. J. Theor. Biol. 84, 3-29. Pamas, H. and Segel, L.A. (1981): A theoretical study of calcium entry in nerve terminals, with application to neurotransmitter release. J. Theor. Biol. 91, 125-169. Pamas , H., Dudel, J. and Pamas, 1. (1982): Neurotransmitter release and its facilitation in crayfish. 1. Saturation kinetics of release, and of entry and removal of calcium. Pfliigers Arch. 393, l-14. Rahamimoff, R. (1968): A dual effect of calcium ions on neuromuscular facilitation. J. Physiol. (London) 195, 471-480. Robles, L., Rhode, W.S. and Geisler, C.D. (1976): Transient response of the basilar membrane measured in squirrel monkeys using the Mlissbauer effect. J. Acoust. Sot. Am. 59, 926-939. Ross. S. (1982): A model of the hair-cell-primary fiber complex. J. Acoust. Sot. Am. 71, 926-941. Russell. 1.J. and Sellick, P.M. (1978): Intracellular studies of hair cells in the mammalian cochlea. J. Physiol. (London) 284, 261-290. Schroeder, M.R. and Hall, J.L. (1974): Model for mechanical to neural transduction in the auditory receptor. J. Acoust. Sot. Am. 55, 1055-1060. Schwid, H.A. and Geisler, C.D. (1982): Multiple reservoir model of neurotransmitter release by a co&fear inner hair cell. J. Acoust. Sot. Am. 72, 1435-1440. Siegel, J.H. (1986): Analysis of synaptic potentials in the mammalian organ of Corti. Assoc. Res. Otolaryngol. Abstr. 9, 60.
Siegel, J.H. and Dallas, P. (1986): Spike activity recorded from the organ of Corti. Hear. Res. 22, 245-248. Siegel. J.H. and Relkin, E.M. (1984): Facilitation in afferent synapses of cochlear hair ceils. Assoc. Res. Otolaryngol. Abstr. 7, 36. Smith, R.L. (1977): Short-term adaptation in single auditory nerve fibers: Some poststimulatory effects. J. Neurophysiol. 40, 1098-1112. Smith, R.L. and Bra&man, M.L. (1982): Adaptation in auditory-nerve fibers: A revised model. Biol. Cyber. 44,107-120. Smith, S.J. and Zucker, R.S. (1980): Aequorin response facilitation and intracellular calcium accumulation in molluscan neurones. J. Physiol. (London) 300, 167-196. Stockbridge, N. and Moore, J.W. (1984): Dynamics of intracellular calcium and its possible relationship to phasic transmitter release and facilitation at the frog neuromuscular junction. J. Neurosci. 4, 803-811. Swadlow, H.A., Kocsis. J.D. and Waxman, S.G. (1980): Modulation of impulse conduction along the axonal tree. Annu. Rev. Biophys. 9, 143-179. Teich, M.C. and Khanna, SM. (1985): Pulse-number distribution for the neural spike train in the cat’s auditory nerve. J. Acoust. Sot. Am. 77. 1110-1128. Walrond, J.P. and Reese, T.S. (1985): Structure of axon terminals and active zones at synapses on lizard twitch and tonic muscle fibers. J. Neurosci. 5, 1178-1131. Young, E.D. and Barta, P.E. (1986): Rate responses of auditory nerve fibers to tones in noise near masked threshold. J. Acoust. Sot. Am. 79, 4266442. Zucker, R.S. and Stockbridge, N. (1983): Presynaptic calcium diffusion and the time courses of transmitter release and synaptic facilitation at the squid giant synapse. J. Neurosci. 3. 1263-1269.
HRR
169
00951
Evidence for presynaptic facilitation in primary cochlear afferent neurons J.H. Siegel and E.M. Relkin * Auditory Physiology Lnboratory, Depts. of Communrcation Scrences and Disorders and Neurobiology and Physioloa, Northwestern University, Evanston, U.S.A. (Received
17 November
1986; accepted
2 April 1987)
Evidence for presynaptic facilitation was sought in the discharge patterns of single units in the chinchilla cochlear nerve. Pairs of acoustic clicks, separated by a variable interval, were delivered and spike discharge times stored for offline analysis. By choosing an appropriate binwidth (5 ms) and collecting data only from units with high characteristic frequencies (CF) the response to each click was contained in a single bin. The ratio of spike counts in the bins containing the responses to the two clicks was computed. For units with low spontaneous rates (SR) of discharge (SR < 8/s), an enhancement of the response to the second click was seen for low stimulus levels. As the stimulus level was raised, the response to the second click became reduced, presumably because of adaptation to the first click. Units with high (SR > 15 spikes/s) seldom exhibited enhancement of the response to the second click. The results are explained with a conceptual model in which two processes, depletion and facilitation decay exponentially following a stimulus. Since the two processes have opposite influences on the rate of transmitter release, the magnitudes of both processes may be underestimated by observing their net effect. Facilitation;
Auditory
nerve; Synaptic
transmission
Introduction Presynaptic facilitation is the increase in transmitter release by a chemically-transmitting cell brought about by prior stimulation. It is usually studied in isolated synapses by delivering pairs of shocks, separated by a variable time interval, to the presynaptic cell. While facilitation has been demonstrated in a variety of preparations (Del Castillo and Katz, 1954; Dude1 and Kuffler, 1961; Katz and Miledi, 1968; Rahamimoff, 1968; Magleby, 1973; Charlton and Bittner, 1978; Pamas, et al., 1982), it has not been conclusively demonstrated in sensory hair cell synapses, despite the widespread belief that the mechanism of trans-
Correspondence to: J.H. Siegel, Auditory Physiology Laboratory, Departments of Communication Sciences and Disorders and Neurobiology and Physiology, Northwestern University, Evanston, IL 60201, U.S.A. * Present address: Institute for Sensory Research, Syracuse University. Syracuse, NY 13244-5290, U.S.A. 0378-5955/87/$03.50
6 1987 Elsevier Science Publishers
mitter release by hair cells is similar to that of other synapses. There has been only indirect evidence from auditory psychophysics and population neural studies for facilitation-like phenomena (reviewed by Cacace and Smith, 1986). Depletion of readily-releasable transmitter quanta also controls the amount of transmitter released by the presynaptic cell during repetitive stimulation (Liley and North, 1953; Hubbard, 1963; Elmqvist and Quastel, 1965; Collins et al., 1984). Considering individual cycles of a tone-burst input as separate stimuli, Furukawa and Matsuura (1978) found that the amplitudes of successive excitatory postsynaptic potentials (EPSP) declined monotonically following the onset of the tone burst. This adaptation was successfully modelled as the depletion of presynaptic neurotransmitter quanta over the timecourse of the stimulus (Schroeder and Hall, 1974; Furukawa et al., 1978; Schwid and Geisler, 1982; Smith and Bra&man, 1982; Ross, 1982). The same phenomenon apparently underlies the adaptation in the discharge rate seen in responses of the sensory neurons of the goldfish
B.V. (Biomedical
Division)
170
saceulus (Kuno, 1983) and the mammalian cochlea (Smith, 1977). Since facilitation and adaptation simultaneously affect transmitter release (in opposite directions) it is possible that facilitation has not been apparent in hair cell afferent synapses because adaptation may dominate for stimuli commonly used to study afferent discharge patterns. Thus ~thou~ both processes may influence the response of the postsynaptic neuron, a net adaptation is usually produced. To test this hypothesis, we attempted to find stimulus conditions in which facilitation dominates. Pairs of acoustic clicks were chosen as stimuli, since this is most directly analogous to the inventions stimulus (i.e., pairs of electrical pulses delivered to the presynaptic cell) used in investigations of facilitation. Since we recorded spike discharges and not transmitter release directly, we could not infer the amount of transmitter released by each click, but only if enough transmitter was released to elicit a postsynaptic action potential. We reasoned that facilitated transmitter output should be expressed as an increased probability of discharge when the click level was such that the probability that either click elicited a spike was less than one. Under some stimulus conditions, units with low spontaneous rates (SR) of discharge did, in fact, respond with greater probability to the second click. A preliminary account of this work has been presented (Siegel and Relkin, 1984). Materials
and Mdods
We used a conventional approach to the eighth nerve through the posterior fossa of adult chinchillas. Data were collected from five animals. Glass ~cropi~t~s filled with 3 M KC1 were placed under visual observation on the cochlear nerve and advanced remotely with a stepping microdrive (Transvertex AB., Sweden). Characteristic frequencies (CF) were measured using an automated threshold tracking paradigm similar to that developed by Kiang and colleagues (I970). Single units with CF above 3 kI-Iz were selected because spikes elicited by low-level clicks fell within a 5 ms interval immediately following the stimulus. In no case did we accept data which did not meet this criterion. We developed a paired-click paradigm
analogous to the paired-pulse method used in other systems, replacing the usual presynaptic electrical stimuli with pairs of 100 ps acoustic clicks separated by a variable time interval (AT). The click level was usually adjusted so that the first click elicited a response in roughly half the presentations. This adjustment was made online by displaying each histogram with 5 ms bins along with the total spike count. Under these conditions, each of the two clicks presented during a given trial typically (i.e. 95% of all trials) elicited no more than one spike. With a fixed number of stimulus repetitions (usually 200~400) it was relatively easy to compute the number of evoked spikes per stimulus. The click level was then held constant, while click separation was varied. It was often possible to repeat the sequence at additional stimulus levels before contact with the fiber was lost. As the stimulus level was raised, an increasing number of trials resulted in multiple responses to a given click. The levels of the two clicks were always equal. During offline data analysis, PST histograms were formed with 5 ms bins such that the response to each click was contained within a single bin (Fig. 1). The relative frequency of response to each of the two clicks ( fi and _&) was simply taken as the spike count in the appropriate bin (n, and nz) divided by the number of stimulus presentations (N). The relative response measure was the ratio of the relative frequencies of response to the two clicks:
rW’3 =fdfi = w% We also examined the effect of a response to the first click per se on the relative frequency of response to the second click. Trials were selected from the total (N) in which there was no response to the first click (N’). For these selected trials, the conditional relative frequency of response to the second click (&‘) was calculated by dividing the number of response counts (ni) by the number of selected trials (N’). In particular, we wished to compare the relative frequencies of response to the two clicks when neither was immediately (i.e. within one stimulus repetition period) preceded by a response to a stimulus. Since no stimulus (and thus no response) immediately preceded the first click, the appropriate reference is fi. The condi-
171
12’ -AT-
I
Fig. 1. ~xpe~mental paradigm. Two sented at the indicated times separated construction of the peristimulus-time bins positioned such that the response pletely in one bin. The spike counts in and a2.
ciick stimuli were preby interval AT. Offline histograms used 5 ms to each click fell comthese bins are noted rzr
tional relative response is then:
r’@ll”) =f;/fi In the absence of a stimulus, some counts in the two 5 ms analysis intervals would be expected from spontaneous activity. For low-SR units the expected number of spontaneous spikes is a small fraction of the stimulus-evoked spikes (Fig. l), however, this may not be true for high-SR units. Because we studied primarily low-SR units, no correction was made for spontaneous activity in computing the response ratios. If spontaneous activity contributes equally to the two response windows, its effect would be to reduce the ratio between unequal click responses. This may explain why we found less compel~g evidence for facilitation in high-SR units than for low-SR units. Results
reliability of the measurements. Considerable variability from trial to trial was observed in both the total number of spikes and the ratio of responses to the two clicks. However, when a ratio substantially greater than one (i.e. f2/f2 > 1.2) was observed, it seldom became less than one on any subsequent trial. To demonstrate the statistical significance of individual data points would require a sufficient number of independent samples to characterize the variance of the underlying processes, since it may be dangerous to assume a particular stochastic model, particularly for very short counting intervals (Teich and Khanna, 1985; Young and Barta, 1986; Relkin, Unpubl. Obser.). When the separation between clicks (AT) was varied, the ratio usually remained greater than one for a considerable range of click spacing. A simple nonparametric sign test was used to establish the likelihood that such observations were simply due to chance.
Low-SR units For our sample of units with SR of 7.6 spikes/s or less, the response to the second click was usually larger than that to the first click at low stimulus levels, but this relations~p reversed as the stimulus level was raised (Fig. 2). For the lowest level stimulus condition in Fig. 2, all 9 click separations less than 100 ms resulted in click
1.51
I
5O
For most click levels, the response to the first click was greater than the response to the second click, resulting in ratios less than one. However, enhancement of the second click response could be seen in low-SR units at low stimulus levels. Repeated measurements were sometimes made with a given stimulus configuration to assess the
. 20
40 AT
1
60
80
(msec)
Fig. 2. Click response ratios obtained from a unit with low-SR (CF 5.9 kHz, SR 0 spikes/s). At the lowest click level ( - 60 dB Attn.), the ratio was greater than one for AT< 100 ms As click level was raised (-40 dB), ratios less (0 -•). than one were obtained for short click separations (W------ W. At the highest click level ( - 30 dB), the ratios were uniformly less than one (A4
ratios greater than one. Using a nonparametric sign test, the probability that this result is simply due to chance is only 0.2%. Contact with 25 units in this category was maintained long enough to present at least two runs of the paired click stimulus. Click response ratios of 1.2 or greater were seen in 15 (60%) of these units for at least one stimulus condition. Response ratios between 1.1 and 1.2 were seen in 9 (36%) of these units, but only one exhibited a maximum ratio less than 1.1. Enhancement of the response to the second click was seen in low-SR units for click separations as large as 75 ms and as short as 7.5 ms. In some units the ratio could exceed 1.5 (Fig. 3). Using the sign test, the probability that chance produced ratios greater than one for all stimulus conditions was calculated to be 0.8% and 1.6% for the -39 dB and - 34 dB stimuli, respectively. The ratio was usually maximal with click separations from 10 to 50 ms and close to one for click separations of 100 ms or more. The fact that the effects seen with short click separations disappeared as the click interval became larger also supports our claim that the patterns seen with shorter click separations were not due to chance. High-SR units Units with high SR (SR > 15 spikes/s) were deliberately selected against in this study. The spontaneous activity made it difficult to use audiovisual cues to initially set the click level when searching for the 50% response criterion.
1.5
0
20
40 60 AT (msecl
60
100
Fig. 3. Large response ratios obtained from a unit with low-SR -39 dB (CF 8.0 kHz, SR 3.0 spikes/s). Attenuation: (0 -0); - 34 dB (Dv.
.5
0
’
20
40 60 AT (msecl
.
60
11
100
Fig. 4. Click ratios from four units with high-SR (e0) CF 4.7 kHz, SR 30/s; (Dn) CF 3.7 kHz, SR 123/s; CF 16.8 kHz, SR 27/s; (A ---A) CF 5.7 (+ -4) kHz, SR 21/s. Ratios less than one were often seen at short click separations, but no pronounced trend was seen at longer separations.
Only seven units with SR of 18 spikes/s or more were studied. These units often showed ratios smaller than one at click intervals less than 10 ms (Fig. 4). The ratio was usually smallest at the shortest intervals between clicks. Although enhancement between 1.1 and 1.2 was seen in 5 of the 7 cases, the large variability between runs gave us little confidence in the significance of this result. Discussion Our measure of discharge probability with low-level clicks is not sensitive to changes in the synchrony of discharge within the analysis interval, since the 5 ms bins completely contained the response. The short duration of the click response of high-CF units minimizes the number of spontaneous spikes counted in the response window and allows click intervals as short as 7.5 ms to be evaluated. Studies of the transient response of the basilar membrane in the basal turn suggest that there should be little overlap between the responses to two clicks with separations of 7.5 ms or more (Robles et al., 1976). Although there have been no published reports of the intracellular responses of mammalian cochlear hair cells in the basal turn to acoustic clicks, it is probably a depolarizing pulse with a waveform similar to the
173
(rectified) envelope of the transient response of the basilar membrane. Judging from the absence of adaptation in the intracellular receptor potentials (Russell and Sellick, 1978; Goodman et al., 1982), the receptor potentials produced by the two clicks should be nearly identical. While most studies of presynaptic facilitation employ pairs of electrical shocks which initiate pairs of action potentials in the presynaptic neuron, our paradigm is directly analogous to the delivery of pairs of depolarizing current pulses to the presynaptic terminal in the absence of action potentials (Dude1 et al., 1983). We have recently reported intracellular recordings from afferent terminals presumably innervating IHCs (Siegel and Dallos, 1986; Siegel, 1986). Records of spontaneous activity show EPSPs which nearly always elicit action potentials. Since we have not yet observed EPSPs with pharmacolo~c~ blockage of the action potential, it has not been possible to directly assess the quanta1 content of these synaptic potentials. However, an analysis of the onset slopes of the EPSPs suggests that they are actually composed of integer multiples (up to three) of a unit quantum (Siegel, 1986). Even though this analysis has only been performed on spontaneous activity, it seems likely that stimuli such as low-level acoustic clicks should evoke coordinated quantal transmission. Furthermore, spatial and temporal integration of independently released quanta1 synaptic potentials by the postsynaptic neuron appears to play a relatively small role in this system. Since a spike may be initiated nearly every time the hair cell releases transmitter quanta, the observation of action potentials when recording from a single unit probably gives a reasonable measure of when release occurs. However, it remains to be seen if low-level stimuli can evoke subthreshold postsynaptic potentials. Thus, the absence of an action potential following a low-level click stimulus does not necessarily imply that transmitter was not released. In spite of these uncerta~ties, the ratios of click responses are likely to provide a reasonable estimate of the relative probability of transmitter release in response to the two clicks. However, since we cannot directly assess quanta1 release by recording action potentials, the accuracy of the ratios as a measure of relative transmitter
release must be questioned. On the other hand, this measure seems a reasonable indicator of the existence of differences in release, since the probability of discharge should be directly related to the amount of trans~tter released (Runo, 1983). Accordingly, we conclude that si~ificant enhancement of the response to the second of two clicks is evidence for the facilitation of transmitter release. This conclusion is strengthened by our observation that enhancement of the response to the second click was present when there was no response to the first click (Fig. 5). Axons have been shown to exhibit a period of reduced threshold and increased propagation velocity (supernormality) following an electrically evoked action potential (reviewed by Swadlow et al., 1980). If the enhancement we have observed were due to this postsynaptic mechanism, then the conditional response ratio should not exhibit explement. In fact, the enhancement was greater for the conditioned measure than for the unconditioned measure. Since we could only detect propagating action potentials, it is possible that subthreshold transmitter release can also lead to postsynaptic supernormality, though there is no evidence for this from other systems. Presynaptic facilitation of trans~tter release
I
I
1.5.
5.I
20 a
40 I
AT
60 1
.
80 I
100 .I
(msec)
Fig. 5. Effect of a response to the first click on the response ratio. The data from the -60 dB condition of Fig. 2 are replotted (O---0) along with the ratio of the conditioned frequency of response to the second click and the unconditioned frequency of response to the first click (o- -0). The conditioned ratio was greater than the unconditioned ratio for short click separations.
has been shown in other systems to be maximal at short interstimulus intervals, where the ratio of postsynaptic potentials is typically 1.5-2 : 1 (Del Castillo and Katz, 1954; Dude1 and Kuffler, 1961; Katz and Miledi, 1968; Rahamimoff, 1968; Magleby, 1973; Charlton and Bittner, 1978; Pamas, et al., 1982; Charlton et al., 1982). Facilitation declines monotonically over tens to hundreds of milliseconds as the interval between stimuli is seen in increased. Similarly, the adaptation cochlear afferent neurons, presumably resulting from transmitter depletion, has its maximal effect at short intervals following the adapting (in this case a forward masking) stimulus (Smith, 1977; Harris and Dallos, 1979). The recovery from adaptation also exhibits time constants in the range of 15-40 ms. Since these two factors have opposite effects on transmitter release, the magnitudes of both phenomena are likely to be underestimated by any measure of transmitter output. A simple model can be used to illustrate this point. The recovery from both depletion and facilitation are represented as exponentially decaying functions of opposite polarity: Facilitation :
f(t)
=&pm
Depletion :
d(t)
= A, . &r/r,)
Net response:
r(r) = f(t) -d(t)
Where A, and Tr represent the initial amplitude and decay time constant, respectively, of facilitation. Similarly, A, and T, represent the initial amplitude and recovery time constant, respectively, of depletion. The range of behavior seen in our experiments can be qualitatively described by adjusting the weighting and time constants of the functions (Fig. 6). Two points may be emphasized: (1) the general shape of our data is best reproduced when the amplitude and time course of the two functions are similar; (2) the initial magnitude of either function may be substantially greater than the resultant at any time. Although both factors are simultaneously present, one or the other process may dominate at a given time. The observation that facilitation was greater when there was no response to the first click (Fig. 5) suggests that there was less transmitter released in response to the first click, and hence less depletion, in those selected trials.
Fig. 6. Conceptual model of the interaction between two exponentially decaying functions representing the recovery from depletion (d(t)) and the decay of facilitation (f(t)) following a stimulus. The net response (r(r)) depends on the initial amplitudes and time constants of the two processes.
With appropriate click level and separation, facilitation was clearly evident in low-SR units. At short interclick intervals, dominant adaptation was commonly seen, but facilitation was apparent for intervals as long as 75 ms. As the stimulus level was raised, adaptation became increasingly dominant. We may thus suggest, based on our two-click paradigm, that depletion of transmitter stores is not the only presynaptic factor influencing the dependence of afferent discharge on prior history. It is reasonable to assume that facilitation also influences the response of the cochlear afferent neurons to other stimuli such as tone bursts, particularly at stimulus onset. Just how significant this effect may be under different stimulus conditions remains to be established, but both additional experimental and theoretical treatment of this question is warranted. Though we have neglected refractoriness in this discussion, this postsynaptic phenomenon undoubtedly influenced our results, especially for click separations less than 7.5 ms (Gaumond et al., 1982). However, since refractoriness should affect the ratio of click responses in the same direction as depletion, ne-
175
glecting refractoriness underestimates the relative importance of facilitation when compared to depletion. The observation that high-SR units did not consistently exhibit facilitation may be explained if the number of transmitter quanta immediately available for release at the IHC receptoneural junction is small and, as a result, easily depleted. Facilitated transmitter release in response to the second click is presumably not expressed if the readily-releasable store of transmitter was substantially depleted by the first click. Thus, the fact that facilitation was not consistently seen in highSR units may indicate that the synaptic junctions supplying these fibers are in a relatively depleted state. This result is analogous to the rapid decline in postsynaptic potentials in nerve-muscle preparations when a tetanic stimulus train elicits a large release of transmitter (Liley and North, 1953; Hubbard, 1963; Elmqvist and Quastel, 1965; Collins et al., 1984). Afferent inputs to motoneurons in the cat have been shown to exhibit facilitation or depression depending on the initial amplitude of the EPSP {Collins et al., 1984). Thus, the idea that synapses with a small, readily depleted presynaptic store of trans~tter will express facilitation only under restrictive conditions may be generalized. Since we deliberately sampled primarily low-SR units and as a result collected considerably less data from high-SR units, a larger sample may have shown more prominent facilitation. However, we found that the compound neural response also exhibited a decrement, but never enhancement (not shown) with paired click stimuli identical to those which produced facilitation in low-SR units. Since high-SR neurons constitute the vast majority of spiral ganglion cells, they are expected to dominate the population response. As mentioned in Materials and Methods, spontaneous activity may also affect our measure of facilitation in high-SR units. Presynaptic facilitation has a variety of possible origins, but seems best explained by the so called ‘residual calcium’ hypothesis (Katz and Miledi, 1968). It has been established that calcium entering the presynaptic cell following a depolarization of the nerve membrane leads to transmitter release (Katz and Miledi, 1967; Llinas et al., 1981b). The action of calcium is ter~nated by its removal
from the cytoplasm by cytoplasmic calcium buffers and, ultimately, by active uptake by intracellular organelles or extrusion across the cell membrane. Since this clearing process taken time, some calcium may remain in the presynaptic cytoplasm when a second stimulated calcium influx occurs. The second transient rise in calcium concentration may thus exceed the first even when the amplitude of the successive depolarizations is the same. As transmitter release is directly related to presynaptic calcium concentration (Llinas et al., 1981a, b), more transmitter is liberated during the second stimulus. One extension of this hypothesis is that the response to the second stimulus may be facilitated even if calcium entry following the first stimulus does not evoke transmitter release. Our observation of facilitation in the absence of a response to the first click is consistent with this prediction. Most models of presynaptic facilitation assume implicitly that the store of available transmitter is infinite, so that depletion effects are ignored (Katz and Miledi, 1968; Pamas and Segel, 1980, 1981; Barton et al., 1983; Zucker and Stockbridge, 1983; Fogelson and Zucker, 1985; Cohen and Van der Kloot, 1986). This may be quite well justified in preparations like the neuromuscular junction where the store of presynaptic transmitter is large. Under the conditions of high concentrations of magnesium and low calcium commonly used to study quanta1 transmission in that preparation, the fraction of the available transmitter quanta released by a single electrical stimulus (typically less than ten) is several orders of magnitude smaller than the number of presumed release sites (Heuser et al., 1979; Walrond and Reese, 1985; Ko and Propst, 1986). On the other hand, models of synaptic transmission in the cochlea have focused on adaptation and thus emphasize depletion of transmitter stores (Schroeder and Hall, 1974; Furukawa et al., 1978; Schwid and Geisler, 1982; Smith and Brachman, 1982; Ross, 1982; but see Eggermont, 1985). It appears likely that models which include calcium entry and removal as well as depletion and restocking of transmitter stores will need to be developed to enhance the predictive power of such models as aids to more refined experiments. Even though such models would be quite complex, they
176
may provide a means of assessing the relative contributions of facilitation and depletion over a wide range of stimulus conditions. Similar models would, no doubt, also be useful in exploring other synapses where the presynaptic transmitter store is limited. Acknowledgements The authors wish to thank P. Dallos, M.A. Cheatham and K. Ohlemiller for their helpful comments on this manuscript. This work was supported by NSF grant BNS-8217273 and NINCDS grant NS-08635. References Barton, S.B., Cohen, I.S. and Van der KIoot, W. (1983): The calcium dependence of spontaneous and evoked quantal release at the frog neuromuscular junction. J. Physiol. (London) 337, 735-751. Cacace, A.T. and Smith, R.L. (1986); Some poststimulatory effects on the whole nerve action potential. Hear. Res. 23. 223-232. Charlton, M.P. and Bittner, G.D. (1978): Facilitation of transmitter release at squid synapses. J. Gen. Physiol. 72, 471-486. Charlton, M.P., Smith, S.J. and Zucker, R.S. (1982): Role of presynaptic calcium ions and channels in synaptic facilitation and depression at the squid giant synapse. J. Physiol. (London) 323, 173-193. Cohen, I.S. and Van der KIoot, W. (1986): Facilitation and delayed release at single frog neuromuscular junctions. J. Neurosci. 6,2366-2370. Collins, W.F. III, Honig M.G. and Mendell, L.M. (1984): Heterogeneity of group Ia synapses on homonymous amotoneurons as revealed by high-frequency stimulation of Ia afferent fibers. J. Neurophysiol. 52, 980-993. Del Castillo, J. and Katz, B. (1954): Quantal components of the end-plate potential. J. Physiol. (London) 124, 560-573. Dudel, J. and Kuffler, SW. (1961): The quantal nature of transmission and spontaneous miniature potentials at the crayfish neuromuscular junction. J. Physiol. (London) 155, 514-529. Dudel, J., Pamas, I. and Pamas, H. (1983): Neurotransmitter release and its facilitation in crayfish muscle. VI. Release determined by both, intracellular calcium concentration and depolarization of the nerve terminal. Pflligers Arch. 399, l-10. Eggermont, J.J. (1985): Peripheral adaptation and fatigue: A model oriented review. Hear. Res. 18, 57-71. Elmqvist, D. and Quastel, D.M.D. (1965): A quantitative study of end-plate potentials in isolated human muscle. J. Physiol. (London) 178, 508-529. Evans, E.F. (1975): CochIear nerve and co&ear nucleus. In:
Handbook of Sensory Physiology, pp. l-98. Editors: W.D. Keidel and W.D. Neff. Springer-Verlag, Berlin, Heidelberg, New York. Fogelson, A.L. and Zucker, R.S. (1985): Presynaptic calcium diffusion from various arrays of single channels. Biophys. J. 48, 1003-1017. Furukawa, T. and Matsuura, S. (1978): Adaptive rundown of excitatory post-synaptic potentials at synapses between hair cells and eighth nerve fibers in the goldfish. J. Physiol. (London) 276, 193-209. Furukawa, T., Hayashida, Y. and Matsuura, S. (1978): Quanta1 analysis of the size of excitatory post-synaptic potentials at synapses between hair cells and afferent nerve fibers in goldfish. J. Physiol. (London) 276, 211-226. Gaumond, R.P., Mohmr, C.E. and Kim, D.O. (1982): Stimulus and recovery dependence of cat ccchlear nerve fiber spike discharge probability. J. Neurophysiol. 48, 856-873. Goodman, D.A., Smith. R.L. and Chamberlain. SC. (1982): Intracellular and extracellular responses in the organ of Corti of the gerbil. Hear. Res. 7. 161-169. Harris, D.M. and Dallos, P. (1979): Foreward masking of auditory nerve fiber responses. J. Neurophysiol. 42. 1083-1107. Heuser, J.E., Reese, T.S.. Dennis, M.J., Jan, Y.. Jan, L. and Evans, L. (1979) Synaptic vesicle exocytosis captured by quick freezing and correlated with quantal transmitter release. J. Cell Biol. 81, 275-300. Hubbard, J.I. (1963): Repetitive stimulation at the mammalian neuromuscular junction and the mobilization of transmitter. J. Physiol. (London} 169, 641-662. Katz, B. and Miledi, R. (1967): The timing of calcium action during neuromuscular transmission. J. Physiol. (London) 189, 535-544. Katz, B. and Miledi, R. (1968): The role of calcium in neuromuscular facilitation. J. Physiol. (London) 195, 481-492. Kiang, N.Y.S., Moxon, E.C. and Levine, R.A. (1970): Auditory-nerve activity in cats with normal and abnormal co&leas. In: Sensorineural Hearing Loss, Ciba Symposium, pp. 241-273. Editors: G.E.W. Wolstenholme and J. Knight. J. and A. Churchill, London. Ko, C.-P. and Propst, J.W. (1986): Freeze-fracture of physiologically identified neuromuscular junctions from single frog muscle fibers. J. Elect. Micr. Tech. 4, 147-156. Kuno, M. (1983): Adaptive changes in firing rates in goldfish auditory fibers as related to changes in mean amplitude of excitatory postsynaptic potentials. J. Neurophysiol. 50. 573-581. Liley, A.W. and North, K.A.K. (1953): An investigation of effects of repetitive stimulation of the mammalian neuromuscular junction. J. Neurophysiol. 16, 509-527. Llinas, R., Steinberg, I.Z. and Walton, K. (1981a): Presynaptic calcium currents in squid giant synapse. Biophys. J. 33, 289-322. Llinas, R., Steinberg, I.Z. and Walton, K. (1981b): Relationship between presynaptic calcium current and postsynaptic potential in squid giant synapse. Biophys. J. 33. 323-352. Magleby, K.L. (1973): The effects of repetitive stimulation on facilitation of transmitter release at the frog neuromuscular junction. J. Physiol. (London) 234. 327-352.
Magleby,
K.L. (1979): Facilitation,
augmentation,
and potenti-
ation of transmitter release. Prog. Brain Res. 49, 175-182. Parnas, H. and Segel, L.A. (1980): A theoretical explanation for some effects of calcium on the facilitation of neurotransmitter release. J. Theor. Biol. 84, 3-29. Pamas, H. and Segel, L.A. (1981): A theoretical study of calcium entry in nerve terminals, with application to neurotransmitter release. J. Theor. Biol. 91, 125-169. Pamas , H., Dudel, J. and Pamas, 1. (1982): Neurotransmitter release and its facilitation in crayfish. 1. Saturation kinetics of release, and of entry and removal of calcium. Pfliigers Arch. 393, l-14. Rahamimoff, R. (1968): A dual effect of calcium ions on neuromuscular facilitation. J. Physiol. (London) 195, 471-480. Robles, L., Rhode, W.S. and Geisler, C.D. (1976): Transient response of the basilar membrane measured in squirrel monkeys using the Mlissbauer effect. J. Acoust. Sot. Am. 59, 926-939. Ross. S. (1982): A model of the hair-cell-primary fiber complex. J. Acoust. Sot. Am. 71, 926-941. Russell. 1.J. and Sellick, P.M. (1978): Intracellular studies of hair cells in the mammalian cochlea. J. Physiol. (London) 284, 261-290. Schroeder, M.R. and Hall, J.L. (1974): Model for mechanical to neural transduction in the auditory receptor. J. Acoust. Sot. Am. 55, 1055-1060. Schwid, H.A. and Geisler, C.D. (1982): Multiple reservoir model of neurotransmitter release by a co&fear inner hair cell. J. Acoust. Sot. Am. 72, 1435-1440. Siegel, J.H. (1986): Analysis of synaptic potentials in the mammalian organ of Corti. Assoc. Res. Otolaryngol. Abstr. 9, 60.
Siegel, J.H. and Dallas, P. (1986): Spike activity recorded from the organ of Corti. Hear. Res. 22, 245-248. Siegel. J.H. and Relkin, E.M. (1984): Facilitation in afferent synapses of cochlear hair ceils. Assoc. Res. Otolaryngol. Abstr. 7, 36. Smith, R.L. (1977): Short-term adaptation in single auditory nerve fibers: Some poststimulatory effects. J. Neurophysiol. 40, 1098-1112. Smith, R.L. and Bra&man, M.L. (1982): Adaptation in auditory-nerve fibers: A revised model. Biol. Cyber. 44,107-120. Smith, S.J. and Zucker, R.S. (1980): Aequorin response facilitation and intracellular calcium accumulation in molluscan neurones. J. Physiol. (London) 300, 167-196. Stockbridge, N. and Moore, J.W. (1984): Dynamics of intracellular calcium and its possible relationship to phasic transmitter release and facilitation at the frog neuromuscular junction. J. Neurosci. 4, 803-811. Swadlow, H.A., Kocsis. J.D. and Waxman, S.G. (1980): Modulation of impulse conduction along the axonal tree. Annu. Rev. Biophys. 9, 143-179. Teich, M.C. and Khanna, SM. (1985): Pulse-number distribution for the neural spike train in the cat’s auditory nerve. J. Acoust. Sot. Am. 77. 1110-1128. Walrond, J.P. and Reese, T.S. (1985): Structure of axon terminals and active zones at synapses on lizard twitch and tonic muscle fibers. J. Neurosci. 5, 1178-1131. Young, E.D. and Barta, P.E. (1986): Rate responses of auditory nerve fibers to tones in noise near masked threshold. J. Acoust. Sot. Am. 79, 4266442. Zucker, R.S. and Stockbridge, N. (1983): Presynaptic calcium diffusion and the time courses of transmitter release and synaptic facilitation at the squid giant synapse. J. Neurosci. 3. 1263-1269.