Burst activity and ultrafast activation kinetics of CaV1.3 Ca2+ channels support presynaptic activity in adult gerbil hair cell ribbon synapses

Auditory information transfer to afferent neurons relies on precise triggering of neurotransmitter release at the inner hair cell (IHC) ribbon synapses by Ca2+ entry through CaV1.3 Ca2+ channels. Despite the crucial role of CaV1.3 Ca2+ channels in governing synaptic vesicle fusion, their elementary properties in adult mammals remain unknown. Using near-physiological recording conditions we investigated Ca2+ channel activity in adult gerbil IHCs. We found that Ca2+ channels are partially active at the IHC resting membrane potential (−60 mV). At −20 mV, the large majority (>70%) of Ca2+ channel first openings occurred with an estimated delay of about 50 μs in physiological conditions, with a mean open time of 0.5 ms. Similar to other ribbon synapses, Ca2+ channels in IHCs showed a low mean open probability (0.21 at −20 mV), but this increased significantly (up to 0.91) when Ca2+ channel activity switched to a bursting modality. We propose that IHC Ca2+ channels are sufficiently rapid to transmit fast signals of sound onset and support phase-locking. Short-latency Ca2+ channel opening coupled to multivesicular release would ensure precise and reliable signal transmission at the IHC ribbon synapse.


Introduction
The mammalian auditory system relies on temporally precise high-fidelity neurotransmitter release at inner hair cell (IHC) ribbon synapses (Fuchs, 2005). IHC neurotransmitter release is triggered by Ca 2+ entry in response to cell depolarization during sound-induced hair bundle deflection. The IHC Ca 2+ current is carried almost exclusively by Ca V 1.3 Ca 2+ channels (>90%: Platzer et al. 2000;Brandt et al. 2003), which are clustered at the presynaptic active zones and colocalized with readily releasable vesicles (Brandt et al. 2005;Graydon et al. 2011). However, the properties of Ca V 1.3 Ca 2+ channels in adult mammalian cells remain unknown and it is not clear whether their activation kinetics are sufficiently rapid to sustain phase locking to sound (Palmer & Russell, 1986).
In adult IHCs, a single ribbon synapse signals to an auditory afferent fibre, highlighting the importance of accurate neurotransmission at these synapses (Fuchs, 2005). In addition to ensuring sustained, high rates of vesicle release (Moser et al. 2006), hair cell ribbon synapses are able to synchronize the release of multiple vesicles to produce large AMPA-mediated excitatory postsynaptic currents (EPSCs; Glowatzki & Fuchs, 2002). The underlying mechanism for multivesicular release at ribbon synapses is unknown. Hair cell depolarization has been shown to increase the frequency and amplitude of EPSCs in lower vertebrates (Li et al. 2009). In mammals, only the frequency seems to be affected by IHC depolarization (Glowatzki & Fuchs, 2002;Grant et al. 2010). While there is no current explanation for the absence of EPSC amplitude increase with IHC depolarization, which is normally seen in other synapses (e.g. Christie & Jahr, 2006), their increase in frequency is thought to depend upon the incremental recruitment of Ca 2+ channels per synapse with depolarization, with each new channel opening producing an additional vesicle fusion event (Brandt et al. 2005). The problem with this interpretation is that, classically, membrane depolarization is expected to increase the open probability of each Ca 2+ channel and not the number of available Ca 2+ channels. Therefore, upon depolarization each channel will be open for a longer time, which increases the probability of having overlapping channel openings. Here we analysed the unitary Ca V 1.3 currents in adult mammalian IHCs to determine how they are likely to influence vesicle fusion at the presynaptic release site.

Ethics statement
In the UK, all animal studies were licensed by the Home Office under the Animals (Scientific Procedures) Act 1986 and were approved by the University of Sheffield Ethical Review Committee. In Germany, care and use of the animals and the experimental protocol were reviewed and approved by the animal welfare commissioner and the regional board for scientific animal experiments in Tübingen.

Immunocytochemistry
Gerbil cochleae (P20) were fixed (2% paraformaldehyde), decalcified and cryosectioned as previously described (Zampini et al. 2010). Primary antibodies to Ca V 1.3 (rabbit, Alomone Laboratories, Jerusalem, Israel, diluted 1:50) and Ribeye/CtBP2 (mouse, BD Transduction Laboratories, Oxford, UK, diluted 1:50) were detected with Cy3-conjugated (Jackson ImmunoResearch Laboratories, West Grove, USA) or Alexa Fluor 488-conjugated (Life Technologies, Darmstadt, Germany) secondary antibodies. Images were acquired using a CCD camera and analysed with cellSense Dimension software (OSIS GmbH, Munster, Germany). The distribution of Ca 2+ channels and ribbons were imaged over a distance of several micrometres with the coverage of the entire IHC synaptic region in an image-stack along the z-axis (z-stack) followed by three-dimensional deconvolution as previously described (Zampini et al. 2010). Immunolabelling was repeated at least three times in cells from different animals.

Single Ca 2+ channel data analysis
Single Ca 2+ channel analysis was performed using Clampfit as previously described (Zampini et al. 2010). Briefly, leak and uncompensated capacitive currents were corrected by subtracting average episodes without channel activity (null sweeps) from the active sweeps. Event detection was performed with the 50% threshold detection method with each transition inspected before being accepted. Idealized traces were used to calculate channel amplitude distribution (event duration >0.34 ms), open probability (P o ) and open and closed time histograms. Amplitude distributions were fitted with a Gaussian function. P o was calculated as the fraction of time spent open vs. the total recording time. Sweeps containing two or more Ca 2+ channels were excluded from the analysis. The total number of Ca 2+ channels per IHC was estimated using: where N is the number of channels, I Ca is the peak macroscopic Ca 2+ current, i is the single-channel current size and P o the channel open probability. To analyse the single channel open and closed times (Table 1), data from IHCs were pooled to obtain a distribution of dwell times on a log scale (12 bins decade −1 ) with normalization of the number of observations for bin amplitude. The plots obtained were interpolated, using the maximum likelihood method, with the sum of n (two or three) exponential functions (Zampini et al. 2010). The first latency distribution was investigated by measuring the time interval between the last point of the capacitative transient and the first opening. The distribution of the first latency was analysed as for the open and closed times. When fitting the dwell-time distributions, events of less than 0.34 ms in duration were ignored because they were under-represented due to low-pass filtering, which caused an underestimation of the fastest component of the first-latency distribution. Statistical comparisons of means were made using Student's two-tailed t test. Unless otherwise specified, mean values are quoted ±SEM, where P < 0.05 indicates statistical significance.

Ca 2+ channel distribution in IHCs along the adult gerbil cochlea
In adult gerbil IHCs, Ca 2+ channel clusters were only detected at the presynaptic region ( Fig. 1A), which agrees with previous findings in apical-coil mouse IHCs (Brandt et al. 2005;Zampini et al. 2010). The average number of immuno-positive Ca V 1.3 spots in basal IHCs was 14 ± 2 (n = 6), which were all colocalized with synaptic ribbons (CtBP2) (Fig. 1B). Despite performing cell-attached recordings from the bottom-half of IHCs, which contains the 14 presynaptic regions, the number of successful patches with stable Ca 2+ channel activity was extremely low (∼4%).

Figure 1. Distribution of Ca V 1.3 and CtBP2/RIBEYE in adult gerbil IHCs
A, basal-coil IHC from an adult (P20) gerbil cochlea immunostained for the Ca V 1.3 Ca 2+ channel (red) and ribbon marker CtBP2/RIBEYE (green). Colocalization is shown in the merge image in the right column. White dotted lines delineate IHCs. Images represent the maximum intensity projection over all layers of the z-stack. Nuclei were stained with DAPI (blue). Scale bar, 10 μm. B, total number of immunopositive spots for Ca V 1.3 (red bar), total number of CtBP2/RIBEYE (green bar) and colocalized (yellow bar). Number of IHCs analysed for cochlear region is indicated above the bars.
J Physiol 591.16

Unitary current of Ca V 1.3 Ca 2+ channels in adult IHCs
Voltage-dependent L-type Ca 2+ channels are encoded by four different pore-forming α1 subunit genes (α1C or Ca V 1.2, α1D or Ca V 1.3, α1S or Ca V 1.1, α1F or Ca V 1.4) and are sensitive to 1,4-dihydropyridines, such as the antagonist nifedipine and the agonist BayK 8644.
In the mammalian cochlea, IHCs almost exclusively express the Ca V 1.3 isoform (Platzer et al. 2000), and are therefore ideally suited to investigate the properties of this Ca 2+ channel in isolation. Single Ca V 1.3 Ca 2+ channel recordings were performed from IHCs in acutely isolated cochleae maintained at 34-37 • C, using 5 mM extracellular Ca 2+ and 5 μM BayK 8644. The use of BayK 8644 was essential when working at 34-37 • C since in its absence the majority of single-channel openings were not resolved and the apparent sub-conductive open states became very frequent. Although BayK 8644 is known to produce longer openings of L-type Ca 2+ channel (Hess et al. 1984;Nowycky et al. 1985;Markwardt & Nilius 1988;Ceña et al. 1989), it does not affect the first latency (Hess et al. 1984), or its elementary Ca 2+ conductance and voltage sensitivity (Zampini et al. 2010). As a result, at macroscopic level the impact of BayK 8644 is to increase the peak Ca 2+ current about 3-fold with no change in activation kinetics (Zampini et al. 2010). Initially, experiments were performed in a high-K + extracellular solution, which, by bringing the IHCs' resting membrane potential near to 0 mV, allowed for control over transmembrane potential in the recorded patches (Zampini et al. 2010). Under these conditions, unitary Ca 2+ channel openings became more frequent and longer lasting with membrane depolarization ( Fig. 2A). Moreover, irrespective of the membrane potential (V m ), Ca 2+ channels exhibited two distinct opening modes: one characterized by brief and rather infrequent openings (arrows: Fig. 2A) and the other by long-lasting clusters or bursts of long and brief openings separated by brief closures (arrowheads: Fig. 2A). These two modes are reminiscent of gating 'mode 1' (brief) and 'mode 2' (long), previously reported for L-type Ca 2+ channels (Hess et al. 1984;Nowycky et al. 1985). Although mode 2 gating is favoured by BayK 8644 (Hess et al. 1984;Nowycky et al. 1985), it is a characteristic behaviour of L-type Ca 2+ channels. This is also indicated by the observation that clusters of brief and long openings (bursts) can be seen in the absence of BayK 8644, which increase the duration of all openings, in immature mouse IHCs (Zampini et al. 2010). Often, a specific gating mode was largely predominant in one or a group of successive sweeps, indicating that the two gating modes were not randomly distributed among sweeps, which is consistent with the idea that they are controlled by intracellular modulators (Nowycky et al. 1985;Kamp & Hell, 2000;Carabelli et al. 2001). The single-channel current-voltage (I-V ) relation was linear in the voltage range investigated with an average slope conductance of 15 pS ( Fig. 2B and C). This is similar to that measured in immature mouse IHCs (Zampini et al. 2010) and in a cell culture system (Bock et al. 2011), but larger than that proposed for frog hair cells (3.5 pS: room temperature, 2 mM Ca 2+ , Graydon et al. 2011). We found that single Ca 2+ channel activity was present at the resting V m for adult IHCs (−60 mV: Johnson et al. 2011), and fell within the activation range of the macroscopic I Ca ).

Ca V 1.3 Ca 2+ channel open probability in Na + -based extracellular solution
We investigated single Ca 2+ channel properties while maintaining IHCs at their physiological V m using a Na + -based extracellular solution. The Na + -based solution prevented us from directly determining the IHC resting V m in cell-attached recordings. Therefore, the patch transmembrane voltage is indicated as the unknown IHC V m plus the voltage step delivered to the patch pipette (e.g. V m + 20 mV: 20 mV depolarization from V m ). The actual patch transmembrane voltage was estimated using the amplitude of the elementary Ca 2+ current and extrapolating it from the I-V curves obtained in high-K + solution ( Fig. 2B and C), assuming identical single-channel conductance between the two recording conditions (Zampini et al. 2010). Calcium channel recordings obtained by applying 500 ms step depolarizations to V m + 20 mV and V m + 50 mV in a Na + -based solution are shown in Fig. 3A. The estimated transmembrane voltage applied to IHCs was about −50 mV for V m + 20 mV and −20 mV for V m + 50 mV (Fig. 2C). Calcium channel gating 'mode 1' and 'mode 2' (Fig. 2A) were also seen in the Na + -based solution ( Fig. 3B; see also Supplemental Fig. 1, available online only). The percentage of null-sweeps in adult IHCs was 46% near −20 mV. When only sweeps containing channel openings (500 ms duration) were considered, the Ca 2+ channel mean open probability (P o ) increased with depolarization from 0.01 at about −50 mV to 0.21 at about −20 mV. We found that the maximal P o varied significantly among sweeps, from the lowest value of 0.014 when the channel opened rarely and briefly (gating mode 1) to 0.91 in the presence of prolonged periods of opening (mode 2). The total number of Ca 2+ channels present in adult IHCs (see eqn (1): I Ca = −197 pA; i = −0.34 pA; P o = 0.21) is likely to be in the order of 2800. The macroscopic I Ca (−197 pA) was measured in adult gerbils using experimental conditions similar to those used for single-channel recordings (see Methods). A higher P o of ∼0.8 and smaller elementary conductance has previously been estimated using fluctuation analysis by calculating the variance and mean of whole-cell tail Ca 2+ currents measured at −62 mV from pre-step depolarization to +58 mV (Brandt et al. 2005). However, strong membrane depolarization (+58 mV), as opposed to voltage levels within a physiological range (−20 mV), has been shown to produce an increased frequency of long-duration (mode 2) Ca 2+ channel openings (Josephson et al. 2002), which will result in P o overestimation. Despite the presence of BayK 8644 in our recording conditions, a similar or slightly higher Ca 2+ channel P o was also found in bullfrog hair cells (Graydon et al. 2011) and in the retina (Doering et al. 2005), indicating that the low P o is likely to be a characteristic of Ca 2+ channels at ribbon synapses. Ca V 1.3 Ca 2+ channel splice variants with very low P o have also been described in cell culture systems (Bock et al. 2011).

Kinetic properties of the Ca 2+ current
The activation and inactivation time constants of the ensemble-average current ( Fig. 3C: τ activation1 = 0.33 ms, τ activation2 = 6.08 ms, τ inactivation = 92 ms; 300 sweeps from 8 IHCs) from single-channel recordings (Supplemental Fig. 1) were in the range of those obtained with whole-cell recordings (Fig. 3D: τ activation1 = 0.50 ± 0.03 ms, τ activation2 = 3.33 ± 0.56 ms, τ inactivation = 151 ± 15 ms, n = 5) using similar experimental conditions (5 mM Ca 2+ , BayK 8644). The similarity between the kinetics of the ensemble-average current and that recorded in whole-cell suggests that the cell-attached configuration, and the mechanical perturbation it could induce upon the patched membrane, is unlikely to significantly alter the kinetic behaviour of the Ca 2+ channels. Using whole-cell recordings, we also found that the activation kinetics of I Ca in 5 mM Ca 2+ (0.74 ± 0.22 ms near −10 mV, n = 5) were about 4 times slower compared to those in 1.3 mM Ca 2+ (0.16 ± 0.02 ms, n = 7), which is most likely caused by surface screening effects (Byerly et al. 1985;Smith et al. 1993). Time constants (τ) and the relative contributions (W, %) were obtained from the exponential fits of the latency of the first opening (A), open (B) and closed (C) time distributions at one or two different membrane voltages. Open (B) and closed (C) time constants obtained from fitting the dwell time distributions were grouped as follows: τ 1 below 1 ms, τ 2 between 1 ms and 10 ms, τ 3 greater than 10 ms.
The distribution of first latencies, the delay between the stimulus onset and the first observed Ca 2+ channel opening, was well defined by the sum of three exponentials (Fig. 3E). The fastest component showed a sub-millisecond time constant near the peak of the macroscopic I Ca (−20 mV), the weight of which was much larger than the other two components ( Table 1). The fast component's relative weight is likely to be greater than our estimate, since in several sweeps (28%) containing early-onset channel openings the first latencies could not be measured, due to the residual capacitive transient (see Methods). The similarity between the first two time constants of the first latency distribution (τ 1 = 0.18 ms; τ 2 = 6.3 ms in 5 mM Ca 2+ ; Table 1) and those of the macroscopic I Ca activation, indicates that they are the main determinant of current activation in response to membrane depolarization. Since in the presence of 1.3 mM extracellular Ca 2+ the activation kinetics of macroscopic I Ca became about 4 times faster (see above), the first latency time constants can also be expected to become faster. We estimate that τ 1 , which represents 73% of the total first latency distribution (Table 1), would decrease from 0.18 ms to about 50 μs.
Fitting the dwell time distributions (data not shown) revealed two or three open (τ o1 , τ o2, τ o3 ) and three closed (τ c1 , τ c2 , τ c3 ) time constants (Table 1). We found that depolarization induced an overall increase in the relative weight of τ o2 , the appearance of a longer time constant (τ o3 ) and an increase in the weight of the shortest close time constant (τ c1 ). A very slow exponential component, with a time constant (τ c3 ) of about 95 ms, was also present in closed-time distributions. Although τ c3 was probably underestimated, due to the high probability of long closure events being interrupted at the end of the 500 ms depolarization, it was 12 times greater than the 'intermediate' time constant τ c2 . Moreover, the relative weight of the slowest component was only 2%. Therefore, the average number of 'short' closures per sweep exceeded that of 'long' closures, indicating that single Ca 2+ channel openings had a relatively high probability of being separated by short closings. This implies that Ca 2+ channel activity was largely organized in bursts, consisting of sequences of openings separated by short closings, and interrupted by prolonged closures (Fig. 3A and B). The mean burst duration, defined as any cluster of openings occurring without superimpositions and separated from the previous and/or following openings by an interval of at least 15 ms (i.e. twice the value of τ c2 ), was 81 ± 72 ms (136 bursts from 101 sweeps: 8 IHCs). Bursting activity greatly increased P o in a sweep. As seen with the high-K + extracellular solution, bursts of channel openings often appeared in successive sweeps (Fig. 4A), indicating a shift of the Ca 2+ channel gating mode towards bursts with depolarization. Moreover, burst onsets were concentrated at the very beginning of the sweep (Fig. 4B), consistent with the short Ca 2+ channel first latency.

Discussion
In this study, using near-physiological experimental conditions, we determined that in adult IHCs the first Ca 2+ channel opening latency following membrane depolarization is likely to be about 50 μs. We also found that most Ca 2+ channel openings are rare and very brief (∼0.5 ms). Despite the low mean P o , Ca 2+ influx into IHCs through a Ca 2+ channel can be maximized by burst activity. We propose that brief single Ca 2+ channel openings are sufficient to trigger vesicle release, and the short-latency would ensure reliable and precise signal transmission at the IHC ribbon synapse during high-frequency activity.

Ca 2+ channel short latency allows high-frequency tuning
The coding of auditory stimuli in mammals requires temporally precise transfer at IHC ribbon synapses (Fuchs, 2005). However, the activation kinetics of IHC Ca 2+ channels were deemed to be too slow (several milliseconds in adult hair cells: reviewed by Moser et al. 2006) to follow the exact timing of sound stimuli. We found that the fastest time constant of Ca 2+ channel first latency (τ 1 ) was ∼0.18 ms in adult gerbil IHCs (at −20 mV; 5 mM Ca 2+ ; 34-37 • C). This value is much faster than that recorded in immature mouse IHCs (τ 1 , 0.70 ms: Zampini et al. 2010), indicating that the kinetic properties of Ca 2+ channel change with development similar to the macroscopic I Ca . Moreover, the fast component of the total first latency distribution was predominant, contributing about 73% of it (Table 1). Finally, the comparison with whole-cell recordings indicated that in the presence of 1.3 mM Ca 2+ , real τ 1 values can be assumed to be even smaller at any given potential, with an expected value of 50 μs at −20 mV (see Results). These findings show that Ca 2+ channel activation rates are sufficiently rapid to support phase-locking to sound (Palmer & Russell, 1986).

Single Ca 2+ channel openings sustain release at rest
IHCs release glutamate tonically, modulating the rate of release as stimulus intensity changes. The resting V m of adult IHCs has been estimated in vitro to be around −60 mV using physiological conditions (Johnson et al. 2011). Since we observed Ca 2+ channel activity at membrane potentials as negative as −70 mV (Fig. 1), a fraction of these channels is likely to be active at rest. The resting Ca 2+ channel activity would elicit 'spontaneous' neurotransmitter release at IHC ribbon synapses and drive the background firing activity observed in auditory afferent fibres (Robertson & Paki, 2002). We calculated that there are about 2800 Ca 2+ channels present in adult IHCs, which is similar to that found in immature mouse IHCs (Zampini et al. 2010) and adult bullfrog hair cells (Rodriguez-Contreras & Yamoah, 2001;Graydon et al. 2011). The similar elementary conductance of Ca 2+ channels recorded from different patches in gerbil IHCs is consistent with previous findings showing that in these cells the Ca 2+ current is almost exclusively carried by Ca V 1.3 channels (Platzer et al. 2000;Brandt et al. 2003). Recent fast confocal Ca 2+ imaging studies have hypothesized that IHCs could adjust the number and the gating of Ca V 1.3 channels at their active zones to diversify their transmitter release rates ). This is consistent with our observation that the gating (mode) of Ca V 1.3 channels varies significantly in the same patch, presumably as a consequence of intracellular modulation. The presence of different Ca V 1.3 splice variants and/or intracellular modulators among different synapses could allow IHCs to regulate neurotransmitter release at distinct active zones.
Considering that high-frequency adult gerbil IHCs contain ∼14 active zones, and assuming that in these cells ∼2800 Ca 2+ channels are expressed, ∼10% of which seem to be extra-synaptic (Meyer et al. 2009), then there would be ∼180 Ca 2+ channels in each presynaptic active zone (∼0.25 μm 2 : Lenzi & von Gersdorff, 2001;Meyer et al. 2009). This high channel density agrees with that reported for Ca V 2.1 channels in hippocampal glutamatergic terminals (Holderith et al. 2012) and for Na + channels in rapidly conducting systems such as the rat node of Ranvier (Neumcke & Stämpfli, 1982). Given the mean single-channel P o of 0.01 at −50 mV, on average two Ca 2+ channels per active zone would be simultaneously open near the IHC resting V m . However, considering that Ca 2+ channel bursting (mode 2) is the main contributor to the P o , and that in the absence of BayK 8644 the channel P o is expected to be even lower, it is likely that a single-channel opening will provide enough Ca 2+ to trigger a vesicle fusion event, which is in partial agreement with previous indirect observations (Brandt et al. 2005;Li et al. 2009).
An interesting feature of hair cell ribbon synapses is that vesicle release evokes short-lived EPSCs of variable amplitude (Glowatzki & Fuchs 2002;Li et al. 2009;Grant et al. 2010), the majority (>70%) of which show a rapid monophasic rise time (<1 ms: Li et al. 2009;Grant et al. 2010). Large monophasic EPSCs are thought to originate from a highly synchronized and very rapid presynaptic fusion of multiple vesicles (up to 20). Currently, the mechanism underlying multivesicular release in IHCs is unknown. Given the very low Ca 2+ channel resting P o in IHCs and other ribbon synapses (Graydon et al. 2011;Doering et al. 2005), it is extremely unlikely that large monophasic EPSCs originate from simultaneous opening of distinct Ca 2+ channels (estimated probability for two simultaneous openings is ∼0.3% or even lower without BayK 8644). Thus, our gerbil IHC data indicate that a single Ca 2+ channel opening is likely to be able to trigger the simultaneous fusion of multiple vesicles, which supports a similar proposal for bullfrog hair cells (Graydon et al. 2011). At retinal bipolar cell ribbon synapses, it has been proposed that multivesicular release could originate from the compound fusion of multiple vesicles (Sterling & Matthews, 2008). In hair cells, evidence for such a mechanism remains elusive. Recent findings showed that the amplitude of EPSCs is independent of presynaptic Ca 2+ influx (Glowatzki & Fuchs, 2002;Grant et al. 2010), indicating that postsynaptic mechanisms could also contribute to the highly variable EPSC amplitude, as proposed for central glutamate synapses (Franks et al. 2003). Indeed, the density of postsynaptic AMPA receptors has been shown to vary at a single IHC ribbon synapse (Ottersen et al. 1998;Meyer et al. 2009) as well as among different afferent terminals (Liberman et al. 2011).

Single Ca 2+ channel openings during membrane depolarization
During sound-induced stimulation, the IHC receptor potential is driven by the mechanotransducer current. We found that membrane depolarization greatly increased the open probability of Ca 2+ channels: P o changed from 0.01 at rest to 0.21 at −20 mV. This agrees with the observation that the frequency of EPSCs increases with IHC depolarization (Glowatzki & Fuchs, 2002). However, there are two unusual features of EPSC recordings that are difficult to reconcile with the single Ca 2+ channel properties: (1) EPSC amplitude does not increase with IHC depolarization, which is normally seen in other synapses (e.g. Christie & Jahr, 2006), indicating that their size is independent of the amount of Ca 2+ influx into the cell (Glowatzki & Fuchs, 2002;Goutman & Glowatzki, 2007); (2) monophasic EPSCs remain more frequent than multiphasic EPSCs with IHC depolarization (Li et al. 2009;Grant et al. 2010), which contradicts what we would anticipate considering that the increased Ca 2+ channel P o with depolarization is expected to produce mostly random, non-synchronized, overlapping channel openings.
Pharmacological manipulation of the macroscopic Ca 2+ influx into IHCs led to the hypothesis that large EPSCs could originate from the incremental recruitment of single Ca 2+ channels with depolarization (Brandt et al. 2005). However, adult IHCs express a homogeneous population of Ca V 1.3 Ca 2+ channels, with analogous voltage dependency, and depolarization only increased the P o of each Ca 2+ channel, and not the number of available Ca 2+ channels: in a single patch, Ca 2+ channel P o varied from <0.01 at −50 mV to >0.9 at −20 mV with no sign of overlapping Ca 2+ channels. This means that all Ca 2+ channels controlling vesicle fusion, which are presumably equally sensitive to voltage change, will, on average, be open for longer with membrane depolarization. Since Ca 2+ channel P o is largely determined by gating mode 2 (bursting), we propose instead that depolarization increases the chance that Ca 2+ channels opening in mode 1 (largely silent) switch to mode 2 (bursting), thus increasing the probability of vesicle-fusion events. This is likely to be true even in the absence of BayK 8644 since Ca V 1.3 Ca 2+ channels show bursting behaviour even without the agonist (Zampini et al. 2010). Macroscopically, this would appear as an apparent increase in the number of active Ca 2+ channels (Brandt et al. 2005). This model is consistent with the observation that the frequency of EPSCs of varying amplitude increases with IHC depolarization (Glowatzki & Fuchs, 2002). On the other hand, the presence of the low incidence of multiphasic EPSCs (about 30%: Grant et al. 2010) could result from Ca 2+ -induced Ca 2+ release from intracellular stores or perhaps be due to Ca 2+ occasionally escaping from the nanodomain and diffusing to additional release sites. This 'spillover' of Ca 2+ ions could be a consequence of prolonged Ca 2+ channel openings during bursts of activity and the saturation of the Ca 2+ sensor(s) at the presynaptic site, or saturation of intracellular Ca 2+ buffers, since its volume seems to be restricted by the presence of the ribbon (Graydon et al. 2011).