View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

Neuron, Vol. 20, 797–807, April, 1998, Copyright 1998 by Cell Press Inactivation of Presynaptic Current Contributes to Synaptic Depression at a Fast Central

Ian D. Forsythe,*‡ Tetsuhiro Tsujimoto,† Presynaptic Ca2ϩ currents have been directly re- Margaret Barnes-Davies,* Matthew F. Cuttle,* corded and characterized in several nonmammalian and Tomoyuki Takahashi† preparations. At squid stellate ganglia, the presynaptic * Group Ca2ϩ current activates rapidly, showing little inactivation Department of Cell Physiology and Pharmacology (Katz and Miledi, 1971; Llinas et al., 1981), and was University of Leicester pharmacologically classified as P-type (Llinas et al., Leicester LE1 9HN 1989). In the chick cilliary ganglion calyx, the current is United Kingdom principally generated by N-type Ca2ϩ channels (Stanley † Department of Neurophysiology and Goping, 1991; Yawo and Momiyama, 1993) and also Faculty of Medicine shows little inactivationwith short (Stan- University of Tokyo ley and Goping, 1991; but see Yawo and Momiyama, Tokyo 113 1993). The Ca2ϩ current at the synaptic terminal of the Japan goldfish retinal bipolar cell is L-type (Tachibana et al., 1993) and has a very slow inactivation (von Gersdorff and Matthews, 1996), whereas in rat neurohypophyseal Summary nerve terminals, slowly inactivating L-type Ca2ϩ current coexists with a rapidly inactivating N-type current Voltage-gated calcium channels are well character- (Lemos and Nowicky, 1989). Although presynaptic Ca2ϩ ized at neuronal somata but less thoroughly under- currents of fast in the vertebrate central ner- stood at the presynaptic terminal where they trigger vous system are pharmacologically identified as P-, Q-, transmitter release. In order to elucidate how the in- and N-types (Takahashi and Momiyama, 1993; Luebke trinsic properties of presynaptic calcium channels in- et al., 1993; Regehr and Mintz, 1994; Wheeler et al., fluence synaptic function, we have made direct re- 1994), information about their basic characteristics in

cordings of the presynaptic calcium current (IpCa)ina the nerve terminal is lacking. This knowledge is central giant synapse called the calyx of Held. The to our understanding of CNS function, since theoretically current was pharmacologically classified as P-type the presynaptic Ca2ϩ channel properties should be an and exhibited marked inactivation. The inactivation important determinant of synaptic efficacy. In addition, was largely dependent upon the inward calcium cur- the molecular mechanism underlying these characteris- rent magnitude rather than the membrane potential, tics could be a potential target for activity-dependent displayed little selectivity between divalent charge . -carriers (Ca2؉,Ba2؉and Sr2؉), and exhibited slow recov- To address this question, we have recorded Ca2ϩ cur ery. Simultaneous pre- and postsynaptic whole-cell rents from a giant presynaptic terminal (called the calyx

recording revealed that IpCa inactivation predominantly of Held) in the rat brainstem using the patch-clamp contributes to posttetanic depression of EPSCs. Thus, method (Forsythe, 1994). This synapse is located in the

because of its slow recovery, IpCa inactivation underlies medial nucleus of the trapezoid body (MNTB) and gener- this short-term synaptic plasticity. ates large excitatory postsynaptic currents (EPSCs), principally mediated by ␣-amino-3-hydroxy-5-methyl-4- Introduction isoxazole proprionic acid (AMPA) receptors (Forsythe and Barnes-Davies, 1993; Barnes-Davies and Forsythe, Voltage-gated Ca2ϩ channels are clustered at the pre- 1995). We have found that the presynaptic Ca2ϩ current synaptic terminal (Robitaille et al., 1990; Llinas et al., rapidly inactivates through a charge carrier–dependent 1992), and Ca2ϩ influx through these channels triggers mechanism and that the recovery from inactivation is exocytotic transmitter release (Katz, 1969). The Ca2ϩ relatively slow. The functional significance of the Ca2ϩ channel is also the target of presynaptic modulation current inactivation was revealed by simultaneous pre- through –coupled receptors such as metabo- and postsynaptic recordings (Takahashi et al., 1996), tropic glutamate receptors (Takahashi et al., 1996) or which demonstrated that the slow recovery allows inac-

GABAB receptors (Wu and Saggau, 1995), and even a tivation to accumulate, thereby contributing to postte- small change in the presynaptic calcium current can tanic depression at this excitatory synapse. cause a significant change in synaptic efficacy, due to 2ϩ the power relationship between Ca influx and transmit- Results ter release (Dodge and Rahamimoff, 1967). Biophysical 2ϩ properties of Ca channels such as activation, deacti- Pharmacological Identification of Presynaptic vation, and inactivation have been documented for vari- Calcium Channels 2ϩ ous subtypes of voltage-gated Ca channels at neu- In order to identify the Ca2ϩ channel subtypes involved ronal somata. Although these inherent properties will in transmitter release at the calyx of Held, we first tested 2ϩ affect overall Ca influx, their physiological significance the effect of type-specific antagonists on excitatory post- at presynaptic terminals is not known. synaptic currents (EPSCs) evoked in the principal neu- ron of the MNTB (Barnes-Davies and Forsythe, 1995). ‡ To whom correspondence should be addressed. As shown in Figure 1A, the P/Q-type channel blocker Neuron 798

Figure 1. Block of EPSCs and Presynaptic Calcium Current by ␻-Agatoxin IVA (A) Effect of Ca2ϩ channel blockers on EPSCs. Nimodipine (10 ␮M), ␻-conotoxin GVIA (2 ␮M), and ␻-agatoxin IVA (200 nM) were ap- plied during theperiods indicated by thebars. Sample traces are from the points indicated in the time plot (i–v). An MNTB neuron was voltage clamped at a holding potential of Ϫ70 mV, and EPSCs were evoked at 0.3 Hz. Each point is the average of seven responses.

(B) Block of IpCa by 5 nM ␻-agatoxin IVA. IpCa was evoked by a from a Ϫ80 mV holding potential to Ϫ10 mV at 0.1 Hz (i).

5nM␻-agatoxin IVA blocked IpCa by 86% in 30 min (ii). Strong depolarizations (to ϩ120 mV, 60 ms, 40 times at 4 Hz, applied at the

arrow) unblocked the current and IpCa recov- ered to 78% of control (iii). Subsequent appli- cation of 200 nM ␻-agatoxin IVA rapidly

blocked IpCa by 97%, with the remainder being abolished by 50 ␮M cadmium.

␻-agatoxin IVA (200 nM) completely blocked the EPSC 1997, Soc. Neurosci., abstract). The ␻-conotoxin–sensi- (n ϭ 10), whereas neither the L-type channel blocker tive component gradually diminished as animals ma- nimodipine or nicardipine (2–10 ␮M, n ϭ 10) nor the tured, suggesting that the presynaptic Ca2ϩ channel N-type channel blocker ␻-conotoxin GVIA (2 ␮M, n ϭ subtypes are developmentally regulated (Iwasaki and 10) had such an effect. These results obtained from Takahashi, unpublished data). 10- to 18-day-old rats are consistent with our previous One pharmacological criterion that distinguishes P-type observations on the presynaptic Ca2ϩ current at this from Q-type channels is the dose of ␻-agatoxin IVA synapse (Takahashi et al., 1996) and also similar to those required to block the Ca2ϩ current by 50% (half-inhibi-

100ف reported for the squid stellate ganglion (Llinas et al., tory dose, IC50); this is 1–3 nM for P-type and 1989) and adult mouse (Katz et nM for Q-type channels (Mintz et al., 1992; Randall and al., 1997), but they contrast with observations at other Tsien, 1995). We tested the effect of low doses of 2ϩ mammalian central synapses (Takahashi and Momi- ␻-agatoxin IVA on the presynaptic Ca current (IpCa) yama, 1993; Regehr and Mintz, 1994; Wheeler et al., recorded from the calyceal presynaptic terminal. As il- 1994) and also with those of the avian brainstem auditory lustrated in Figure 1B, 5 nM ␻-agatoxin IVA gradually synapse (Sivaramakrishnan and Laurent, 1995). How- and irreversibly (data not shown) blocked IpCa to 25% Ϯ ever, in younger (6- to 9-day-old) rats, ␻-conotoxin could 4.6% (mean Ϯ SEM, n ϭ 3) in 30 min. Much of this partially block EPSCs at the calyx–MNTB synapse (Iwa- reduction was distinct from Ca2ϩ current rundown, since saki and Takahashi, unpublished data; see also Wu et al., strong depolarizations (60 ms steps to ϩ120 mV, 40 Presynaptic Calcium Current Inactivation 799

Figure 2. Voltage Dependence of IpCa

(A) Current–voltage relationship of IpCa. (B) Steady-state inactivation (closed circles) assessed by applying conditioning prepulses for 500 ms followed by a test depolarization

to Ϫ10 mV (inset). Inactivation of IpCa followed calcium current generation during the condi- tioning pulse (above Ϫ40 mV) as shown by

the open symbols. Test IpCa amplitude was normalized to that evoked after the Ϫ130 mV conditioning pulse (n ϭ 8) and plotted against the conditioning voltage. (C) Activation curve (closed circles) obtained from tail currents (inset) evoked in a calyx following a 10 ms step depolarization from a Ϫ60 mV holding potential. Normalized data from four to eight calyces are plotted against voltage. A solid line fitted to the data points

is a Boltzmann function: I ϭ 1/(Imax ϩ

exp(V1/2 Ϫ V)/k). The half-activation voltage

(V1/2) and slope factor (k) were Ϫ21 mV and 4.4 mV, respectively.

times at 4 Hz) of the calyx resulted in unblock of the mV (n ϭ 8). The kinetics were slower when repolarized channels (Mintz et al., 1992) to 82% Ϯ 1.9% of control to less negative potentials (Ϫ60 mV, 0.20 Ϯ 0.02 ms, levels (n ϭ 3). An identical protocol applied prior to toxin n ϭ 5; to Ϫ40 mV, 0.41 Ϯ 0.03 ms, n ϭ 6). The activation application had no effect on IpCa amplitude (data not curve was measured from the tail current amplitude and shown). Subsequent application of 200 nM ␻-agatoxin could be fitted to a Boltzmann function with a half-

IVA rapidly blocked IpCa, with a small toxin-resistant com- activation voltage of Ϫ21 mV and a slope factor of ϩ4 ponent (Ͻ5%) being abolished by Cd2ϩ (50 ␮M). Since mV (Figure 2C).

5nM␻-agatoxin blocked IpCa by more than 50%, IC50 One of the typical features of IpCa was a lack of steady- must be less than 5 nM. These results suggest that state inactivation at potentials more negative than Ϫ40 transmitter release at the calyx–MNTB synapse is medi- mV (Ͻ10% between Ϫ130 and Ϫ40 mV; Figure 2B, ated by a relatively homogeneous population of P-type closed circles). In contrast, a marked inactivation was 2ϩ Ca channels. observed once IpCa was activated (Figure 2B, open cir- cles). The presence of inactivation in these P-type Ca2ϩ Inactivation of Presynaptic Calcium Current currents contrasts with the observed biophysical prop- 2ϩ IpCa was activated when a presynaptic terminal was de- erties of somatic P-type Ca currents (Randall and polarized beyond Ϫ40 mV and showed a maximal inward Tsien, 1995), which do not exhibit inactivation (Mintz et current at around Ϫ10 mV (Figure 2A). Inactivation of IpCa al., 1992; Usowicz et al., 1992), but is similar to Q-type was always observed during maintained depolarizations Ca2ϩ currents, which do show inactivation (Randall and positive to Ϫ40 mV (Figure 2A), and on repolarization it Tsien, 1995). In the present study, we have classified rapidly deactivated (Figure 2C, inset). The deactivation IpCa as P-type by adopting the pharmacological criterion time constant measured from the tail current was 0.15 Ϯ rather than the biophysical one.

0.01 ms when the nerve terminal was repolarized to Ϫ80 The properties of IpCa inactivation were examined by Neuron 800

Divalent Cation Dependence of IpCa Inactivation

We next examined whether the IpCa inactivation is specifi- cally dependent on Ca2ϩ influx. Inactivation developed with a biphasic time course having time constants of 43.6 Ϯ 6.4 ms and 477 Ϯ 51 ms (n ϭ 7), with the slow component contributing 57% Ϯ 3% of the peak Ca2ϩ

current (Figure 4A). When IpCa was recorded using an internal patch solution containing 16 mM BAPTA instead of 0.5 mM EGTA, the inactivation was still present but its time course was slightly slowed (time constants, 52.0 Ϯ 6.0 and 2155 Ϯ 298 ms, n ϭ 4; Figure 4B). When 2ϩ 2ϩ 2ϩ [Ca ]o was replaced by Sr or Ba , the currents also inactivated with a slower time course: Ca2ϩ Ͻ Sr2ϩ Ͻ Ba2ϩ (decay time constants, 48.8 Ϯ 2.9 and 576 Ϯ 49 ms for Sr2ϩ and 88.5 Ϯ 8.6 and 673 Ϯ 46 ms for Ba2ϩ, n ϭ 7; Figure 4C). These results indicate that the underly- ing mechanismis not specific for Ca2ϩ and may therefore differ from the inactivation mechanism at L-type Ca2ϩ channels (Ashcroft and Stanfield, 1981; Eckert and Chad, 1984; von Gersdorff and Matthews, 1996). Inter- estingly, the magnitudes of the Ba2ϩ and Sr2ϩ currents were comparable with those of the Ca2ϩ current (117%Ϯ 13% and 101% Ϯ 8%, respectively), in contrast to other high voltage–activated currents in which Ba2ϩ is more permeable than Ca2ϩ (Hille, 1992). In skeletal muscle, a time-dependent depletion of 2ϩ 2ϩ [Ca ]o in transverse tubules can cause a decline of Ca current during depolarization (Almers et al., 1981). A similar depletion mechanism in the synaptic cleft might

conceivably account for the IpCa inactivation here. But if this were the case, the rate of current decline should be related to the amount of Ca2ϩ influx. We have tested Figure 3. A Bell-Shaped Voltage Profile of IpCa Inactivation this possibility by reducing the number of functional (A) Voltage protocols generating IpCa inactivation in a test pulse are 2ϩ shown above sample records. Conditioning depolarizations (1.9 s Ca channels by applying 20 nM ␻-agatoxin IVA (Figure duration) of various magnitudes were followed 100 ms later by a 4D). Although the amplitude of IpCa was reduced to about test pulse to Ϫ20 mV. Holding potential was Ϫ80 mV. half control levels, IpCa declined with a similar time 2ϩ (B) The peak amplitudes of Ca currents evoked by the conditioning course, as shown by normalizing and overlaying the pulse (triangles) and that observed during the test pulse (circles) control and test data (Figure 4D). This result indicates from five presynaptic terminals are plotted against the membrane 2ϩ that [Ca ]o depletion cannot account for IpCa inacti- potential. IpCa amplitude was normalized to that of the conditioning

IpCa amplitude at Ϫ20 mV. Inactivation increased as the conditioning vation. potential activated more inward current, it was maximal when the It is also possible that the apparent current depen- conditioning inward current was maximal, and it declined as the dency of IpCa inactivation is caused by a complex voltage conditioning potential approached ECa2ϩ. dependence of the inactivation mechanism as proposed by a model for N-type Ca2ϩ current inactivation (Jones and Marks, 1989). This model was supported by the result that the inactivation rate was independent of the a double-pulse protocol. A long (1.9 s) conditioning de- external concentration of the charge carrier Ba2ϩ (Jones polarization of variable magnitude was followed by a and Marks, 1989). We have tested this possibility by 2ϩ 2ϩ brief (50 ms) test pulse of constant magnitude at an reducing external Ca concentration. As [Ca ]o was interval of 100 ms. As shown in Figure 3 (circles), inacti- reduced, IpCa amplitude was gradually diminished. Con- 2ϩ vation of the test IpCa increased concomitantly with the comitantly, the rate of Ca current inactivation became amplitude of the conditioning IpCa, with maximal inactiva- slower. When normalized to peak amplitude, the rate of 2ϩ tion being observed following a conditioning pulse to inactivation was clearly slower with less [Ca ]o (Figure around Ϫ10 mV, which is also the voltage generating 4E). These results suggest that the IpCa inactivation is the peak inward Ca2ϩ current (triangles). Inactivation of caused by Ca2ϩ itself, rather than by membrane po- the test current then declined as the conditioning volt- tential. age approached the apparent at more To further examine whether a divalent-dependent positive potentials. A plot of inactivation against condi- mechanism was responsible for IpCa inactivation, we tioning voltage thus formed an inverse bell-shaped adopted the method of making the Ca2ϩ channel perme- curve, which mirrored the I–V relationship (Figure 3B). able to Naϩ, as reported for N- and L-type Ca2ϩ channels This result suggests the possibility that the inactivation (Konnerth et al., 1987; Cox and Dunlap, 1994). After 2ϩ 2ϩ is Ca -dependent. [Ca ]o was lowered to submicromolar concentrations Presynaptic Calcium Current Inactivation 801

Figure 4. Ionic Dependence of IpCa Inactivation

(A) IpCa inactivation followed a dual exponential time course. The time constants are 40 ms and 778 ms, respectively, with the slow component contributing 61% of the peak amplitude.

(B) With 16 mM BAPTA in the patch electrode, IpCa inactivation was slower; in this case, the time constants were 90 ms and 2100 ms, respectively, with the slow component contributing 67% of the peak amplitude. See text for mean values. 2ϩ 2ϩ 2ϩ 2ϩ (C) Replacement of [Ca ]o by Sr or Ba slowed inactivation rate but did not block it. Presynaptic Ca channel currents carried by either Ca2ϩ,Ba2ϩ,orSr2ϩ (1 mM) are shown superimposed without normalization. Note that the amplitude is hardly changed for different charge carriers. Data were obtained from the same calyx.

(D) IpCa before and after application of 20 nM ␻-agatoxin IVA. Although the amplitude of the current was depressed, the rate of inactivation was unchanged as shown below by the complete overlap of normalized data. 2ϩ (E) Reducing [Ca ]o from 1.0 to 0.1 mM gradually reduced the calcium current amplitude and concomitantly slowed the rate of inactivation. 2ϩ Sample data are normalized and superimposed below. Data were recorded with 0 mM [Mg ]o. The I–V relationship did not show a significant 2ϩ shift during the [Ca ]o reduction (n ϭ 3; data not shown). by buffering with EGTA, large presynaptic inward cur- 8), with the fast component comprising 62% Ϯ 6% of rents with little inactivation were evoked on depolariza- the peak amplitude (Figure 6A). This slow recovery from tion (Figure 5). These currents were readily abolished inactivation should therefore allow accumulation of in- by ␻-agatoxin IVA (200 nM), indicating that the current activation during repetitive stimulation with short depo- ϩ 2ϩ was carried by Na through P-type Ca channels (Fig- larizations. When IpCa was evoked repetitively, using brief ure 5A). Using the same dual-pulse protocol employed (1 ms) depolarizing voltage steps, the Ca2ϩ current was above (with calcium as the charge carrier; Figure 3A), initially enhanced but then declined (Figure 6B). This the inactivation of this Naϩ current was no longer bell initial facilitation was surprising and is currently under shaped and displayed only a weak voltage dependence investigation. Continued repetitive activation of IpCa at (Figure 5B) similar to that observed for the steady-state these frequencies (50–200 Hz) showed that the decline in inactivation (see Figure 2B). These results strongly sug- IpCa amplitude with higher frequencies was the dominant gest that under physiological conditions, IpCa is inacti- feature (Figure 6B) and suggests that accumulation of vated by its own charge carrier, namely calcium ions. IpCa inactivation may have physiological consequences.

Recovery of IpCa from Inactivation

An additional important feature of the IpCa inactivation Physiological Significance of IpCa Inactivation was its slow rate of recovery. Once inactivation had To examine whether the IpCa inactivation might play a been induced with a long depolarization (1.9 s to Ϫ10 physiological role in modulating synaptic transmission, mV), complete recovery took more than 1 min. The re- we made simultaneous pre- and postsynaptic recordings covery had a biexponential time course with time con- in which EPSCs were induced with Ca2ϩ currents gener- stants of 7.5 Ϯ 0.7 s and 52.9 Ϯ 6.3 s, respectively (n ϭ ated by a brief depolarizing pulse (Takahashi et al., Neuron 802

Figure 5. Inactivation When Naϩ is the Charge Carrier through Presynaptic Ca2ϩ Channels ϩ ϩ (A) Presynaptic inward currents carried by Na in divalent-free ACSF (with 1 mM EGTA). [Na ]o was reduced to 90 mM (substituted by tetraethylammonium chloride). Inward currents (top) were completely blocked by 200 nM ␻-agatoxin IVA, indicating that the Naϩ is passing through the IpCa channel. The time course of the block is shown below. (B) Inactivation of the Naϩ current measured using the same protocol as in Figure 3. (Top) Sample current traces evoked by three different conditioning pulses. (Bottom) The peak amplitude of currents evoked by conditioning pulses (triangles) and by the test pulse (circles) plotted against conditioning voltage (n ϭ 4). Data were normalized to the conditioning current amplitude at Ϫ20 mV as in Figure 3. Only a minor voltage-dependent inactivation component remains.

1996). Generation of tetanic stimulation (100 Hz) caused current from inactivation determines the length of the EPSCs to undergo a depression in spite of the initial posttetanic depression period. transient facilitation of IpCa (Figures 7A and 7B; see also Figure 6B). When the stimulation frequency was re- Discussion turned to 0.2 Hz after a 10 s tetanus, the EPSC quickly recovered by 50% within 5 s, while I remained inacti- pCa The ability to recordfrom the calyx of Held nerve terminal ف vated. Thus, up to this point, the I and resultant EPSCs pCa is an important development, allowing examination of exhibited some degree of divergent behavior, sug- presynaptic conductances and their influence on trans- gesting the involvement of other mechanisms (see Dis- mitter release in the mammalian CNS. We have shown cussion). Thereafter, however, EPSCs recovered in par- that the presynaptic Ca2ϩ current is pharmacologically allel with I (Figure 7A). The time course of recovery pCa classified as a P-type Ca2ϩ current, which exhibits a was dependent upon the duration of the tetanus (Figure marked divalent cation-dependent inactivation. This in- 7C); after a 1 s tetanus, recovery was complete within activation was found to contribute to the posttetanic 5 s (data not shown), but it took more than 1 min after depression of excitatory transmission. Thus, in addition a 10 s tetanus. To estimate the contribution of I inacti- pCa to triggering exocytosis, these presynaptic Ca2ϩ chan- vation to the EPSC depression during this recovery pe- nels also have an important role in short-term plasticity riod, we compared the I –EPSC relationship under two pCa at this central glutamatergic synapse. conditions at the same synapse (Takahashi et al., 1996): first during recovery from posttetanic depression and 2ϩ 2؉ second during the reduction of [Ca ]o. As illustrated Inactivating P-Type Ca Channels Mediate in Figure 7D, the input–ouput (IpCa–EPSC) relationship Synaptic Transmission during the posttetanic recovery period (closed circles) At many central synapses, transmitter release is medi- 2ϩ 2ϩ was indistinguishable from that during [Ca ]o reduction ated by multiple Ca channel subtypes (Takahashi and (open circles) at all four synapses examined. Thus, the Momiyama, 1993; Luebke et al., 1993; Regehr and Mintz, depression of the EPSC during the posttetanic recovery 1994; Wheeler et al., 1994). In contrast, at the brainstem period can be largely accounted for by a reduction in calyx–MNTB synapse, excitatory transmission was me- the presynaptic Ca2ϩ current caused by inactivation. diated almost exclusively by P-type Ca2ϩ channels. This Hence, the slow rate of recovery of presynaptic Ca2ϩ presynaptic P-type Ca2ϩ current shares many properties Presynaptic Calcium Current Inactivation 803

Figure 6. Time Course of Recovery from Inactivation and Effect of Repetitive IpCa Activation (A) The time course of recovery from inactivation induced by a 1.9 s depolarization to Ϫ10 mV (voltage protocol shown above) and monitored by a 10 ms depolarizing pulse to Ϫ20 mV, applied at 0.2 Hz. Inset shows superimposed IpCa traces during (i) control, (ii) inactivation, and (iii) recovery. The fast and slow time constants of recovery were 4.6 s and 30.8 s, respectively (holding potential, Ϫ80 mV), with the fast component contributing 68% of the peak amplitude. See text for mean values.

(B) Summation of IpCa inactivation by repetitive activation. IpCa tail currents were generated bya1msdepolarization to ϩ30 mV from a holding potential of Ϫ80 mV. An initial transient facilitation was followed by a depression during high frequency activation (50, 100, and 200 Hz). Each point represents average values of four sequential events. The sample records in the inset are averaged IpCa tail currents recorded at different 2ϩ 2ϩ times (i–iii) during the 200 Hz activation. Recorded in ACSF containing 1 mM [Ca ]o and 1 mM [Mg ]o.

2ϩ with the P-type current recorded from cerebellar Pur- can be blocked by chelating [Ca ]i with BAPTA. How- kinje cell somata or proximal dendrites, with respect to ever, the inactivation of P-type Ca2ϩ channels observed the rapid deactivation time (Mintz et al., 1992) and chan- here had little specificity between divalent cations, and nel kinetics (Usowicz et al., 1992), but was distinguished intracellular loading of high concentrations of BAPTA from the soma-dendritic channel by having a clear inac- (16 mM) slowed but failed to block inactivation at the tivation. Aninactivating and agatoxin-sensitive Ca2ϩ cur- calyx of Held. The physical structure of presynaptic ter- rent recorded from neuronal somata has been classified minals, such as higher surface-to-volume ratio com- as Q-type (Wheeler et al., 1994). A distinction between pared with somata, may influence inactivation charac- P- and Q-type currents can be made pharmacologically teristics. However, it is unlikely that the relatively small on the basis of their dose-dependent block by ␻-aga- effect of BAPTA on inactivation was due to poor diffu- times sion inside the calyx, since synaptic transmission is 100ف toxin IVA, with the P-type channels being more sensitive (Randall and Tsien, 1995). From this crite- strongly reduced by BAPTA at 1 mM (Borst et al., 1995) rion, the presynaptic Ca2ϩ channels at the calyx of Held and completely blocked after 10 min of whole-cell re- can be classified as P-type and are likely to be part cording with 5 mM (unpublished data). Inactivation of N-type Ca2ϩ channels can be via volt- of the ␣1A calcium channel family. This difference in inactivation might represent a novel presynaptic ␣1 sub- age-dependent (Jones and Marks, 1989) or divalent cat- ion current-dependent mechanisms (Cox and Dunlap, unit, or it could be caused by different regional expres- 1994). Jones and Marks (1989) favored a voltage-depen- sion of Ca2ϩ channel ␤ subunits (Ludwig et al., 1997), dent rather than a current-dependent mechanism, based which are known to influence inactivation kinetics (Isom on the result that the inactivation rate was independent et al., 1994; Olsece et al., 1994; De Waard and Campbell of the external Ba2ϩ concentration. Our data clearly show 1995). 2ϩ that the inactivation rate was slower with lower [Ca ]o and that inactivation was largely abolished when Naϩ Mechanism of Inactivation was the sole charge carrier. Therefore, we conclude that Mechanisms of inactivation have been extensively stud- inactivation of the presynaptic P-type Ca2ϩ channel is ied for high voltage–activated Ca2ϩ channels. A Ca2ϩ- current dependent, but that the mechanism has little dependent inactivation has been documented for L-type specificity between different divalent cations. In agree- Ca2ϩ channels in a variety of cells (Eckert and Chad, ment with our results for P-type channels, it has recently 1984), including goldfish retinal bipolar cell synaptic ter- been demonstrated that Ba2ϩ current through recombi- minals (von Gersdorff and Matthews, 1995). This type nant L-type calcium channels inactivates by a current- of inactivation is generally specific for Ca2ϩ over other dependent rather than a voltage-dependent mechanism 2ϩ divalents and is dependent upon raised [Ca ]i, which (Ferriera et al., 1997). Neuron 804

Figure 7. Posttetanic Depression of EPSCs

Follows IpCa Inactivation

IpCa and EPSCs were recorded simultaneously

at a calyx–MNTB synapse. IpCa (pre) was gen- erated by a 1 ms depolarizing pulse from a holding potential of Ϫ70 mV to Ϫ20 mV at 0.2 Hz.

(A) Time plot of IpCa (circles) and EPSCs (trian-

gles). Tetanic activation of IpCa at 100 Hz was made for 15 s at the bar. After the recovery

of IpCa and EPSCs, the perfusate was switched 2ϩ 2ϩ to low [Ca ]o solution (0 mM Ca and 5 mM Mg2ϩ with 0.1 mM EGTA) for three trials (only the first trial is shown and indicated by the bar).

(B) Sample traces of IpCa and corresponding EPSCs recorded at the points indicated in the time plot (i–vi).

(C) Recovery time course of IpCa and EPSCs following 100 Hz stimulation was slower with increasing length (4–10 s) of tetanus.

(D) Relationship between amplitudes of IpCa and EPSCs: one during recovery from depres- sion following a tetanic stimulation (closed circles) and the other during the reduction of 2ϩ [Ca ]o (open circles). In this experiment, L-2- amino-4-phosphono-butyric acid (L-AP4) and adenosine (100 ␮M) were included in the per- fusate to saturate the presynaptic mGluR and adenosine receptors. The results obtained in three other experiments were essentially the same.

A simple explanation for the inactivation of presynap- after tetanic stimulation, recovery of EPSCs from de- 2ϩ tic P-type Ca channels is that the divalent ion may bind pression was largely explicable by the recovery of IpCa to a site within the channel pore (blocking permeation) or from inactivation. However, during and just after tetanic perhaps to another site on the ␣ subunit (Zhou et al., stimulation, the changes in EPSC amplitude did not fol-

1997). Since under physiological conditions the normal low the changes in IpCa amplitude. Specifically, a tran- 2ϩ 2ϩ charge carrier is Ca , P-type Ca channel inactivation sient initial facilitation of IpCa was not accompanied by can be described as “Ca2ϩ dependent” in this respect. an increase in EPSC amplitude, and a rapid recovery of EPSC amplitude within 5 s of terminating the tetanus 2؉ Ca Current Inactivation Underlies Posttetanic was not associated with a recovery of IpCa. In addition Synaptic Depression to IpCa inactivation, two other principal factors might be

One of the particular features of the IpCa inactivation responsible for synaptic depression. First, massive re- is its slow recovery time. Compared with the calcium lease of transmitter during tetanic stimulation will desen- current inactivation in neurohypophyseal nerve termi- sitize postsynaptic receptors, thereby depressing EPSCs nals (Branchaw et al., 1997), the recovery from inactiva- (Otis et al., 1995). The recovery time course of AMPA tion of IpCa at the calyx–MNTB synapse was one to two receptor desensitization is in the range of 10–300 ms orders of magnitude slower. This implies that IpCa inacti- (Colquhoun et al., 1992; Trussell et al., 1993; Otis et al., vation can accumulate during repetitive stimulation. We 1996), whereas recovery of EPSCs after the tetanus took observed that the presynaptic calcium current produced more than 1 min. Therefore, postsynaptic receptor de- by a brief pulse declined during tetanic stimulation and sensitization maycontribute to synaptic depression dur- that its recovery was slower following a longer tetanus. ing tetanic stimulation but not much during the post-

Since IpCa triggers transmitter release, these features are tetanic recovery period. A second factor that can cause likely to influence the postsynaptic response. Indeed, synaptic depression is depletion of synaptic vesicles Presynaptic Calcium Current Inactivation 805

(Kusano and Landau, 1975). In hippocampal tissue cul- PTD to progress into a long-term depression of synaptic tures, recycling time is estimated to transmission. s (Stevens and Tsujimoto, 1995; Dobrunz and 10ف be Stevens, 1997). Hence, during the tetanus, in addition Experimental Procedures to the IpCa inactivation, receptor desensitization and ve- sicular depletion as well as other factors such as adapta- Transverse brainstem slices (150 ␮m thick) were prepared from 8- to 18-day-old Wistar or Lister Hooded rats killed by decapitation, as tion of exocytosis (Hsu et al., 1996) might contribute to described previously (Edwards et al., 1989; Forsythe, 1994; Taka- synaptic depression. After the tetanus, however, our hashi et al., 1996). Most experiments were conducted in slices ob- results indicate that the prolonged depression of EPSCs tained from animals over 10 days old. One slice was placed in a recording chamber on a fixed stage microscope fitted with Nomarski was largely due to the slow recovery of IpCa from inactiva- optics; the preparation was viewed with a CCD camera using a tion. Had factors other than IpCa inactivation significantly 63ϫ water immersion objective (Axioskop, Zeiss). Whole-cell patch contributed to posttetanic depression of EPSCs, then recordings were made from visually identified MNTB neurons and/ the input–output relationship obtained during the recov- or from a calyx surrounding one MNTB neuron. ery period would have deviated from that obtained by EPSCs were evoked by stimulating presynaptic axons at the mid- 2ϩ reducing [Ca ]o. Since the input–output relations were line of a slicewith a bipolar platinum electrode (Forsythe andBarnes- Davies, 1993) or by evoking a presynaptic Ca2ϩ current (Takahashi similar, we conclude that IpCa inactivation contributes almost exclusively to the posttetanic depression of et al., 1996; Borst and Sakmann, 1996). Presynaptic terminals were identified as thin flattened processes around single MNTB neurons EPSCs at this mammalian fast synapse. and could be confirmed by including 0.2–2 mg/ml Lucifer yellow in Bushy neurons in the anteroventral cochlear nucleus, the internal solution. The standard artificial cerebrospinal fluid which give rise to the calyx, fire spontaneously at rates (ACSF) contained (inmM): 120 NaCl, 2.5 KCl, 26NaHCO3, 10 glucose, of up to 10 Hz in vivo (Friauf and Oswald, 1988), and in 1.25 Na2HPO4, 3.0 myo-inositol, 2.0 sodium pyruvate, 0.5 ascorbic brain slice preparations both bushy and MNTB neurons acid, and 4.0 lactic acid with 2.0 CaCl2 and 1.0 MgCl2 (pH 7.4; gassed with 5% CO2 and 95% O2; temperature, 22ЊC–26ЊC; perfusion rate, will follow synaptic stimuli at frequencies of over 460 1–2 ml/min). For calcium current recordings, tetraethylammonium Hz (Oertel, 1985; Wu and Kelly, 1993). Hence, the stimuli chloride (TEA-Cl, 10 mM) and tetrodotoxin (TTX, l ␮M) were added utilized in this study are within physiological ranges of to the ACSF. Postsynaptic patch pipettes contained (in mM): 97.5 transmission in the binaural auditory pathway. This path- K gluconate, 32.5 KCl, 10 HEPES, 5.0 EGTA, and 1.0 MgCl2 (pH 7.3). way is involved in computing sound source localization N-(2,6-diethylphenylcarbamoylmethyl)-triethyl-ammonium bromide (QX314; 5 mM; Research Biochemicals) was added to suppress in auditory space by detecting interaural timing and generation in some cases. The presynaptic patch level/volume differences (ITD and ILD) of sound de- solution contained (in mM): 110 CsCl, 40 HEPES, 0.5 EGTA, 1.0 tected at both cochlea. These complementary mecha- MgCl2, 2.0 Mg-ATP, 0.5 Na2-GTP, 12 Na2 phosphocreatinine, and 10 nisms rely on very precise transmission and preserva- mM TEA-Cl (pH adjusted to 7.3with CsOH). 10 mM Cs glutamate was tion of the timing information of the auditory signal (in substituted for equimolar CsCl in paired recordings. For experiments where Ca2ϩ was replaced by Naϩ, 1 mM EGTA was added to the microsecond timescales) from both sides of the head 2ϩ ϩ ACSF, Mg was omitted, and [Na ]o was reduced to 90 mM by (Carr, 1993; Trussell, 1997). In mammals, ITD and ILD substitution with TEA-Cl. detection occurs in the medial and lateral superior ol- Pre- and postsynaptic recordings were made using borosilicate ives, respectively, and each of these nuclei receives an glass pipettes (Clark Electromedical) and had resistances of 4–7 inhibitory input from the MNTB. The large size of the M⍀. The series resistance of presynaptic recordings was typically calyx of Held and the EPSC that it generates are adapta- 12–24 M⍀ and was compensated by Ͼ70%. Leak subtraction was performed by either a P/4 method or subtraction of appropriately tions that contribute to the fidelity of transmission by scaled responses to small averaged voltage steps.Data were filtered minimizing latency fluctuations and maximizing the at 1–10 kHz and sampled at 2–27 kHz using a CED1401 interface and safety factor in generating an action potential in the analyzed using “Patch” software (Cambridge Electronic Design). postsynaptic neuron (Forsythe et al., 1995; Oertel, 1997; Junction potentials were not corrected. Data were plotted as means Ϯ SEM. ␻-conotoxin GVIA (Peptide Institute, Japan; Alomone Trussell, 1997). Given this large safety factor, IpCa inacti- Labs, Israel) and ␻-agatoxin IVA (Peptide Institute, Japan) were vation would also contribute to the security of transmis- dissolved in ACSF containing 0.1–1.0 mg/ml cytochrome c. Applica- sion by reducing the probability of transmitter release tion of drugs and changes in the ACSF composition were achieved during high frequency stimulation. This would increase by switching perfusion between different lines at the inflow to the the average interval between exocytotic events within experimental chamber. the population of release sites at an individual calyx Acknowledgments (Ͼ200; Borst and Sakmann, 1996), so promoting recov- ery of the postsynaptic receptors from desensitization We thank Peter Stanfield, Toshiya Manabe, Katsunori Kobayashi, and slowing or preventing presynaptic vesicle depletion. and John Chad for comments on early drafts of the manuscript and Posttetanic potentiation and posttetanic depression David Jones, Antony Smith, Kyoko Matsuyama, and John Rowley (PTD) of synaptic transmission are short-term forms of for technical assistance. This work was supported by a Grant-in- synaptic plasticity that are differentially observed at par- Aid for Scientific Research from the Ministry of Education of Japan, the Wellcome Trust, and a Travel Grant from Japan Society for the ticular synapses (Eccles, 1964). Synaptic vesicular deple- Promotion of Science. I. D. F. is a Wellcome Senior Research Fellow tion has previouslybeen thought to be the general mech- in Basic Biomedical Sciences. anism underlying PTD (Kusano and Landau, 1975), but here we demonstrate that inactivation of presynaptic Received November 17, 1997; revised January 20, 1998. calcium currents largely contributes to PTD at a mam- malian fast synapse. The molecular mechanism underly- References ing the recovery of IpCa from inactivation remains to be Almers, W., Fink, R., and Palade, P.T. (1981). Calcium depletion in elucidated, but any modulatory factor that acted to re- frog muscle tubules: the decline of calcium current under maintained tard the rate of recovery might be expected to enable depolarization. J. Physiol. (Lond.) 312, 177–207. Neuron 806

Ashcroft, F.M., and Stanfield, P.R. (1981). Calcium dependence of Katz, B., and Miledi, R. (1971). The effect of prolonged depolarization the inactivation of calcium currents in skeletal muscle fibers of an on synaptic transfer in the stellate ganglion of the squid. J. Physiol. insect. Science 213, 224–226. (Lond.) 216, 503–512. Barnes-Davies, M., and Forsythe, I.D. (1995). Pre- and postsynaptic Katz, E., Ferro, P.A., Weisz, G., and Uchitel, O.D. (1997). Calcium glutamate receptors at a giant excitatory synapse in rat auditory channels involved in synaptic transmission at the mature and regen- brainstem slices. J. Physiol. (Lond.) 488, 387–406. erating mouse neuromuscular junction. J. Physiol. (Lond.) 497, Borst, J.G.G., and Sakmann, B. (1996). Calcium influx andtransmitter 687–697. release in a fast CNS synapse. Nature 383, 431–434. Konnerth, A., Lux, H.D., and Morad, M. (1987). Proton-induced trans- Borst, J.G.G., Helmchen, F., and Sakmann, B (1995). Pre- and post- formation of calcium channel in chick dorsal root ganglion cells. J. synaptic whole-cell recordings in the medial nucleus of trapezoid Physiol. (Lond.) 386, 603–633. body of the rat. J. Physiol. (Lond.) 489, 825–840. Kusano, K., and Landau, E.M. (1975). Depression and recovery of Branchaw, J.L., Banks, M.I., and Jackson, M.B. (1997). Ca2ϩ- and transmission at the squid giant synapse. J. Physiol. (Lond.) 245, voltage-dependent inactivation of Ca2ϩ channels in nerve terminals 13–32. of the neurohypophysis. J. Neurosci. 17, 5772–5781. Lemos, J.R., and Nowicky, M.C. (1989). Two types of calcium chan- Carr, C.E. (1993). Processing of temporal information in the brain. nels coexist in peptide-releasing vertebrate nerve terminals. Neuron Annu. Rev. Neurosci. 16, 223–243. 2, 1419–1426. Colquhoun, D., Jonas, P., and Sakmann, B. (1992). Action of brief Llinas, R., Steinberg, I.Z., and Walton, K. (1981). Presynaptic calcium pulses of glutamate on AMPA/kainate receptors in patches from currents in squid giant synapse. Biophys. J. 33, 289–322. different neurons of rat hippocampal slices. J. Physiol. (Lond.) 458, Llinas, R., Sugimori, M., Lin, J.-W., and Cherksey, B. (1989). Blocking 261–287. and isolation of a calcium channel from neurons in mammals and Cox, D.H., and Dunlap, K. (1994). Inactivation of N-type calcium cephalopods utilizing a toxin fraction (FTX) from funnel-web spider current in chick sensory neurons: calcium and voltage dependence. poison. Proc. Natl. Acad. Sci. USA 86, 1689–1693. J. Gen. Physiol. 104, 311–336. Llinas, R., Sugimori, R., and Silver, R.B. (1992). Microdomains of De Waard, M., and Campbell, K.P. (1995). Subunit regulation of high calcium concentration in a presynaptic terminal. Science 256, 2ϩ the neuronal ␣1A Ca channel expressed in Xenopus oocytes. J. 677–679. Physiol. (Lond.) 485, 619–634. Ludwig, A., Flockerzi, V., and Hofmann, F. (1997). Regional expres- Dobrunz, L.E., and Stevens, C.F. (1997). Heterogeneity of release sion and cellular localization of the ␣1 and ␤ subunit of high voltage– probability, facilitation, and depletion at central synapses. Neuron activated calcium channels in rat brain. J. Neurosci. 17, 1339–1349. 18, 995–1008. Luebke, J.I., Dunlap, K., and Turner, T.J. (1993). Multiple calcium Dodge, F.A., and Rahamimoff, R. (1967). Co-operative action of channel types control glutamatergic synaptic transmission in the calcium ions in transmitter release at the neuromuscular junction. hippocampus. Neuron 11, 895–902. J. Physiol. (Lond.) 193, 419–432. Mintz, I.M., Adams, M.E., and Bean, B.P. (1992). P-type calcium Eccles, J.C. (1964). The Physiology of Synapses (Berlin: Springer- channels in rat central and peripheral neurons. Neuron 9, 85–95. Verlag). Oertel, D. (1985). Use of brain slices in the study of the auditory 2ϩ Eckert, R., and Chad, J.E. (1984). Inactivation of Ca channels. system: spatial and temporal summation of synaptic inputs in cells Prog. Biophys. Mol. Biol. 44, 215–267. of the anterioventral cochlear nucleus of the mouse. J. Acoust. Soc. Edwards, F.A., Konnerth, A., Sakmann, B., and Takahashi, T. (1989). Am. 78, 328–333. A thin slice preparation for patch clamp recordings from neurones Oertel, D. (1997). Encoding of timing in the brain stem auditory nuclei of the mammalian . Pflugers Arch. 414, ¨ of vertebrates. Neuron 19, 959–962. 600–612. Olcese, R., Qin, N., Schneider, T., Neely, A., Wei, X., Stefani, E., and Ferreira, G., Yi, J., Rios, E., and Shirokov, R. (1997). Ion-dependent Birnbaumer, L. (1994). The amino terminus of a calcium channel inactivation of barium current through L-type calcium channels. J. subunit sets rates of channel inactivation independently of the sub- Gen. Physiol. 109, 449–461. unit’s effect on activation. Neuron 13, 1433–1438. Forsythe, I.D. (1994). Direct patch recording from identified presyn- Otis, T.S., Raman, I.M., and Trussell, L.O. (1995). AMPA receptors aptic terminals mediating glutamatergic EPSCs in the rat CNS, in with high Ca2ϩ permeability mediate synaptic transmission in the vitro. J. Physiol. (Lond.) 479, 381–387. avian auditory pathway. J. Physiol. (Lond.) 482, 309–315. Forsythe, I.D., and Barnes-Davies, M. (1993). The binaural auditory Otis, T., Zhang, S., and Trussell, L.O. (1996). Direct measurement of pathway: excitatory amino acid receptors mediate dual time course AMPA receptor desensitization induced by glutamatergic synaptic excitatory postsynaptic currents in the rat medial nucleus of the transmission. J. Neurosci. 16, 7496–7504. trapezoid body. Proc. R. Soc. Lond. B Biol. Sci. 251, 151–157. Randall, A., and Tsien, R.W. (1995). Pharmacological dissection of Forsythe, I.D., Barnes-Davies, M., and Brew, H.M. (1995). The calyx multiple types of Ca2ϩ channel currents in rat cerebellar granule of Held: a model for transmission at mammlian glutamatergic syn- apses. In Excitatory Amino Acids and Synaptic Transmission, Sec- neurons. J. Neurosci. 15, 2995–3012. ond Edition, H. Wheal and A. Thomson, eds. (London: Academic Regehr, W.G., and Mintz, I.M. (1994). Participation of multiple cal- Press), pp. 133–144. cium channel types in transmission at single to Pur- Friauf, E., andOstwald, J. (1988). Divergent projections of physiolog- kinje cell synapses. Neuron 12, 605–613. ically characterised rat ventral cochlear nucleus neurons as shown Robitaille, R., Adler, E.M., and Charlton, M.P. (1990). Strategic loca- by intra-axonal injection of horseradish peroxidase. Exp. Brain Res. tion of calcium channels at transmitter release sites of frog neuro- 73, 263–284. muscular synapses. Neuron 5, 773–779. Hille, B. (1992). Ion Channels of Excitable Membranes, Second Edi- Sivaramakrishnan, S., and Laurent, G. (1995). Pharmacological char- tion (Sunderland, MA: Sinauer). acterization of presynaptic calcium currents underlying glutama- Hsu, S.-F., Augustine, G.J., and Jackson, M.B. (1996). Adaptation tergic transmission in the avian auditory brainstem. J. Neurosci. 15, of Ca2ϩ-triggered exocytosis in presynaptic terminals. Neuron 17, 6576–6585. 501–512. Stanley, E.F., and Goping, G. (1991). Characterization of a calcium Isom, L.L., De Jongh, K.S., and Catterall, W.A. (1994). Auxiliary sub- current in a vertebrate cholinergic presynaptic nerve terminal. J. units of voltage-gated ion channels. Neuron 12, 1183–1194. Neurosci. 11, 985–993. Jones, S.W., and Marks, T.N. (1989). Calcium currents in bullfrog Stevens, C.F., and Tsujimoto, T. (1995). Estimates for the pool size sympathetic neurons: II. Inactivation. J. Gen. Physiol. 94, 169–182. of releasable quanta at a single synapse and for the time required Katz, B. (1969). The Release of Neural Transmitter Substances (Liv- to refill the pool. Proc. Natl. Acad. Sci. USA 92, 846–849. erpook, UK: Liverpool University Press). Tachibana, M., Okada, T., Arimura, T., Kobayashi, K., and Piccolino, Presynaptic Calcium Current Inactivation 807

M. (1993). Dihydropyridine-sensitve calcium current mediates neu- rotransmitter release from bipolar cells of the goldfish retina. J. Neurosci. 13, 2898–2909. Takahashi, T., and Momiyama, A. (1993). Different types of calcium channels mediate central synaptic transmission. Nature 366, 156– 158. Takahashi, T., Forsythe, I.D., Tsujimoto, T., Barnes-Davies, M., and Onodera, K. (1996). Presynaptic calcium current modulation by a metabotropic glutamate receptor. Science 274, 594–597. Trussell, L.O. (1997). Cellular mechanisms for preservation of timing in central auditory pathways. Curr. Opin. Neurobiol. 7, 487–492. Trussell, L.O., Zhang, S., and Raman, I.M. (1993). Desensitization of AMPA receptors upon multiquantal release. Neu- ron 10, 1185–1196. Usowicz, M.M., Sugimori, M., Cherksey, B., and Llinas, R. (1992). P-type calcium channels in the somata and dendrites of adult cere- bellar Purkinje cells. Neuron 9, 1185–1199. von Gersdorff, H., and Matthews, G. (1996). Calcium-dependent inactivation of calcium current in synaptic terminals of retinal bipolar neurons. J. Neurosci. 16, 115–122. Wheeler, D.B., Randall, A., and Tsien, R.W. (1994). Roles of N-type and Q-type Ca2ϩ channels in supporting hippocampal synaptic transmission. Science 264, 107–111.

Wu, L.-G., and Saggau, P. (1995). GABAB receptor–mediated presyn- aptic inhibition in guinea pig hippocampus is caused by reduction of presynaptic Ca2ϩ influx. J. Physiol. (Lond.) 485, 649–657. Wu, S.H., and Kelly, J.B. (1993). Response of neurons in the lateral superior olive and medial nucleus of the trapezoid body to repetitive stimulation: intracellular and extracellular recordings from mouse brain slice. Hearing Res. 68, 189–201. Yawo, H., and Momiyama, A. (1993). Re-evaluation of calcium cur- rents in pre- and postsynaptic neurones of the chick ciliary ganglion. J. Physiol. (Lond.) 460, 153–172. Zhou, J., Olcese, R., Qin, N., Noceti, F., Birnbaumer, L., and Stefani, E. (1997). Feedback inhibition of Ca2ϩ channels by Ca2ϩ depends on a short sequence of the C terminus that does not include the Ca2ϩ-binding function of a motif with similarity to Ca2ϩ-binding do- mains. Proc. Natl. Acad. Sci. USA 94, 2301–2305.

Note Added in Proof

The data referred to as Iwasakiand Takahashi, unpublished data, are now in press: Iwasaki, S., and Takahashi, T. (1998). Developmental changes in calcium channel types mediating synaptic transmission in rat auditory brainstem. J. Physiol. (Lond.), in press.