Developmental Acquisition of a Rapid Calcium-Regulated Vesicle Supply Allows Sustained High Rates of Exocytosis in Auditory Hair Cells

Auditory hair cells (HCs) have the remarkable property to indefinitely sustain high rates of synaptic vesicle release during ongoing sound stimulation. The mechanisms of vesicle supply that allow such indefatigable exocytosis at the ribbon active zone remain largely unknown. To address this issue, we characterized the kinetics of vesicle recruitment and release in developing chick auditory HCs. Experiments were done using the intact chick basilar papilla from E10 (embryonic day 10) to P2 (two days post-hatch) by monitoring changes in membrane capacitance and Ca2+ currents during various voltage stimulations. Compared to immature pre-hearing HCs (E10-E12), mature post-hearing HCs (E18-P2) can steadily mobilize a larger readily releasable pool (RRP) of vesicles with faster kinetics and higher Ca2+ efficiency. As assessed by varying the inter-pulse interval of a 100 ms paired-pulse depolarization protocol, the kinetics of RRP replenishment were found much faster in mature HCs. Unlike mature HCs, exocytosis in immature HCs showed large depression during repetitive stimulations. Remarkably, when the intracellular concentration of EGTA was raised from 0.5 to 2 mM, the paired-pulse depression level remained unchanged in immature HCs but was drastically increased in mature HCs, indicating that the Ca2+ sensitivity of the vesicle replenishment process increases during maturation. Concomitantly, the immunoreactivity of the calcium sensor otoferlin and the number of ribbons at the HC plasma membrane largely increased, reaching a maximum level at E18-P2. Our results suggest that the efficient Ca2+-dependent vesicle release and supply in mature HCs essentially rely on the concomitant engagement of synaptic ribbons and otoferlin at the plasma membrane.


Introduction
The ribbon synapse of cochlear hair cells (HCs) encodes sound information by tightly controlling the number (discharge rate) and precise timing (temporal coding) of postsynaptic spikes. Remarkably, this synapse can drive postsynaptic auditory nerve fibers at extremely high instantaneous discharge rates (over several thousand spikes/s at stimulus onset) and, after rapid adaptation, can support sustained discharge rates over several hundred spikes/ s during ongoing sound stimulation [1], [2]. To sustain such high rates of synaptic exocytosis, auditory HCs must have efficient mechanisms to rapidly and constantly replenish the pool of synaptic vesicles. While it is well established that the rate of vesicle fusion is tightly controlled by Ca 2+ ions flowing through nearby voltage-gated Ca 2+ channels [3], little is known about the mechanisms regulating the kinetics of vesicle supply at the ribbon active zone. Recently, it was hypothesized that otoferlin, a multi-C2 Ca 2+ sensor that directly regulates SNARE-membrane fusion in vitro [4] and is required for HC vesicle exocytosis [5], also controls the supply of synaptic vesicles at the active zones [6]. Surprisingly, during their early developmental period, immature auditory HCs transiently express several Ca 2+ -dependent synap-totagmins and do not require otoferlin to control phasic transmitter release driven by spontaneous action potentials [7]. With cochlear maturation, the Ca 2+ efficiency and kinetics of exocytosis in HCs largely increase [8], [9], [10], as well as the expression level of otoferlin in HCs [5], [7]. However, the precise mechanisms of how otoferlin is engaged to produce fast vesicle supply and release in developing HCs remains to be elucidated.
Notably, the first sound-evoked responses of immature auditory nerve fibers in the developing cochlea display high thresholds of rhythmic bursting activity and are unable to maintain a sustained steady-state response to long duration tone bursts [11]. The present study tests the hypothesis that neurotransmitter availability is an important factor that limits sustained postsynaptic firing activity in developing auditory afferent neurons. By studying when the ability to produce sustained high rate of exocytosis is acquired by HCs during development, we attempt to gain insights into the mechanisms of synaptic vesicle replenishment. In chick embryo, the first indication of sound-evoked electrical responses from the inner ear have been reported from the 11th day of incubation (E11) [12], [13], [14]. The auditory thresholds then show continuous maturation between E15 and the first post-hatching day (P1) to attain adult values. In the present study, we took advantage of the slow maturation of the auditory chick basilar papilla to characterize the progressive changes occurring in exocytosis and vesicle supply at the HC ribbon synapse.

Preparation of semi-intact chicken basilar papilla
The present investigation was performed in accordance with the guidelines of the Animal Care and Use Committee of the European Communities Council Directive of November 24 th , 1986 (86/609/ EEC) and the University of Bordeaux (ethics committee: Direction Régionale de l'Alimentation, de l'Agriculture et de la Forêt d'Aquitaine (DRAAF Aquitaine) permit number B 33075, approved this study). The study included chickens at different stages of embryonic development ranging from E10-E21 as well as two-day post-hatched chickens. Fertilized eggs were incubated at 37uC in a Marsh automatic incubator (Lyon Electric, Chula Vista, CA). Chicken embryos were sacrificed and staged according to their number of somites, and additionally for the late stages as following: from E8-E12 based on visceral arches, feather gems and eyelids: after E12 based on the length of the beak [15]. Basilar papillae were isolated as described previously [16]. The preparations were dissected in oxygenated chicken saline containing (in mM) 155 NaCl, 6 KCl, 4 CaCl 2 , 2 MgCl 2 , 5 Hepes, and 3 glucose, pH 7.4. The tegmentum vasculosum and the tectorial membrane were removed without any prior enzymatic treatment using a fine minutia needle. Chicken basilar papillae were stored in a 37uC incubator in Minimum Essential Medium (Invitrogen) before recordings from HC in situ. All experiments were performed at room temperature (21-23uC) within 5-45 min of isolation. All reagents were obtained from Sigma Chemicals, unless otherwise specified. Recordings were done in neural (tall) HCs along the basilar papilla. These HCs which are mainly innervated by the afferent fibers correspond to the inner hair cells in mammals. The tonotopic location of HCs was divided in three parts. From the proximal narrow end of the papilla, the first 1/ 3 part was considered as the high frequency coding region of basal HCs. Low frequency apical HCs were recorded at the top 1/3 part of the wide end of the basilar papilla.

Electrophysiology
Calcium currents were recorded in whole-cell voltage-clamp configuration using 3-5 MV resistance pipettes. Currents were recorded with an EPC 10 amplifier (Heka Electronik, Lambrecht/ Pfalz, Germany) and filtered at a frequency of 2-5 kHz through a low-pass Bessel filter. The sampling frequency was determined by the protocol used. No online leak current subtraction was made.
Only recordings with holding current less than 20 pA were accepted for analyses.
Real-time changes in membrane capacitance (DC m ) were recorded using the EPC 10 amplifier. A 2 kHz sine wave of 10 mV was applied to the cells from a holding potential of -90 mV. Capacitance (Cm) signals were low-pass filtered at 80 Hz. Changes in membrane capacitance were measured 0.05-0.5 s after the end of the depolarizing pulse and averaged over a period of 0.2-2 s. Membrane and series resistance (R m and R s ) were monitored during the course of the experiment. Only recordings with stable R m and R s were considered for further analysis. The study included ,247 cells with R s within the 5-20 MV range. Holding membrane potentials were corrected for liquid junction potentials. Extracellular solution for measuring Ca 2+ currents contained (in mM) NaCl/CholineCl 125, KCl 6, CaCl 2 5, 25 TEA, 5 4-AP, D-glucose 10, MgCl 2 1, HEPES 10, pH 7.3, 310 mOsm. TEA and 4-AP were present in the external medium to block residual inward rectifying K + -currents that may contaminate the measured inward Ca 2+ current. Intracellular solution contained (in mM) NMG 75, CsCl 70, Na 2 ATP 5, MgCl 2 2, HEPES 10, EGTA 0.5-10, and glucose 10; pH 7.3, 300 mOsm.

Immunohistochemistry
Tissues were fixed with 4% paraformaldehyde in PBS (PBS) for ,3 hrs, then rinsed and immunostained with a polyclonal antibody directed against mouse CtBP2 (1:200, Sigma) and a monoclonal HCS1-antibody (1:250, a gift from Dr Jeffrey Corwin, University of Virginia, [17]. Immunostaining was visualized with anti-goat secondary antibody conjugated to Alexa 488 (green, CtBP2) and anti-mouse secondary antibody conjugated Alexa 546 (Red, HCS1). Omission of the primary antibodies eliminated staining in all preparations examined. HC actin was counterstained with phalloidin conjugated to Alexa Fluor 647 (1:200, Molecular Probes (Invitrogen, Carlsbud, CA). Fluorescent images were collected and analyzed with a confocal laser scanning upright microscope (Leica DMR TCS SP2 AOBS, Bordeaux Imaging Center). Images of ribbons were taken in the basal synaptic area of the HCs (step size 0.4 mm).

Data analysis
The number of cells (n) is given with each data set. Data were analyzed using pClamp10 (Axon Instruments) and Origin7.0 (Microcal Software). Pooled data were presented as mean 6 SD. Significant difference between groups of cells or between different embryonic stages of development was evaluated using a two-tailed Student's t test; p values are presented in the text and figure to indicate statistical significance. Time constants (t s ) were obtained from fits using Origin software. Time constants were obtained by fitting multiple exponential equations to the activation decay of the current. The equation was of the form: Where I 0 is the initial current magnitude, t 1 , t 2 ...t n are the time constants, and A 1 , A 2 ...A n , are the proportionality constants. Synaptic transfer functions relating Ca 2+ current (I Ca ) and DCm, or Q Ca 2+ and DC m were calculated using an integral of total I Ca , including the tail currents. The data was fitted using first-order power functions: where s = slope factor (fF/pA or fF/pC), and N = power index. The % RRP refilling was calculated as: where DC m = DC m measured using the first, control pulse, and DC m, test = DC m measured using test pulse. The % of I recovered was calculated as: Where I Ca, control, = I Ca measured using the first, control pulse, and I Ca, test = I Ca, test measured using test pulse.

Kinetics and Ca 2+ -efficiency of RRP exocytosis increase with cochlear maturation
The efficiency of Ca 2+ -evoked exocytosis was characterized at four developmental periods of cochlear synaptogenesis: embryonic stages (in ovo) E10-11, E12-14, E16-18 and 2 days post-hatching P2. The first embryonic period, E10-11, corresponds to an early stage of synaptogenesis when the first presynaptic specializations (synaptic bodies or ribbons) can be detected in HCs [18] and when the afferent fibers first contact their base [19]. At stage E11-E14 low frequency hearing starts in the chick embryo [12], [13], [14]. Stage E16-E18 (5-3 days before hatching) corresponds to the final step of synaptogenesis and HC maturation. Finally, P2 corresponds to nearly adult hearing values [20].
At all developmental stages from E10 to P2, rapidly activating inward current (I Ca ) and a concomitant increase in membrane capacitance (DC m ) was recorded when HCs were voltage-stepped from -90 mV to varying depolarized potentials ( Fig. 1A-1B). The voltage-activation curve of DC m displayed a bell shape that followed the I Ca activation curve (maximum amplitude near -10 mV), a behavior consistent with DC m being activated consecutive to Ca 2+ influx. Indeed, complete blockage of I Ca by 250 mM Cd 2+ eliminated DC m (data not shown), confirming that DC m is sensitive to Ca 2+ entry via VGCC, in agreement with [21]. For a 100 ms-depolarization, which is considered to entirely release the readily releasable pool (RRP) [22], the amplitude of DC m responses increased with HC maturation (Fig 1A-B). In low frequency apical HCs, DC m responses were as follows: (in fF at -10 mV) E10, 1062 (n = 10); E12, 2866 (n = 13); E16, 3767 (n = 11); and P2, 4566 (n = 9). High frequency HCs recorded at the base of the basilar papilla were also found to undergo a similar increase in DC m responses with maturation ( Fig. 1A-B and table 1). Notably, at embryonic stages earlier than E10, while activating significant I Ca (17.563.1 pA at -10 mV, n = 7, E7-E8), 100-ms voltage-step depolarization did not produce significant DC m responses in all HCs tested (below background level of 3.861.3 fF; data not shown). The synaptic transfer function relating DC m as a function of charge entry (Q Ca as time integral of I Ca ) was compared at different developmental stages by stepping the cells to various potentials from -60 to -10 mV for a constant 100 ms duration (Fig.1C). With maturation, data points both in apical and basal HCs were fitted by first-order power functions with decreasing power index (N; cooperative index) and increasing slope factors (Ca 2+ efficiency, fF/pC) (Fig. 1D-E; table 1). These results indicated that, similarly to mouse cochlear HCs [8], [9], maturation of the chick HC synapse is associated with a better coupling between Ca 2+ influx and vesicular release.
Changes in release rate were then compared by stepping HCs to constant potential (from -90 to -10 mV) for different durations from 20 to 3000 ms (Fig. 2). Data points were best fitted by two exponential functions that likely described a fast release of a readily releasable pool (RRP) of vesicles (up to 100 ms) and a secondary, slowly releasable pool (SRP) as previously described [9], [21], [23]. Kinetics of RRP release largely increased with maturation from E10 to P2, with respective time constants of RRP release corresponding to 8967 ms (n = 10) and 3964 ms (n = 9), (p,0.001, Fig.2B). Accordingly, the RRP release rate (vesicles/s) increased from 3,1586698 at E10 to 30,54064122 at P2 in apical HCs (p,0.001; table 1). The SRP amplitude also increased nearly 3-fold from E10 to P2 (Fig. 2B). By varying the concentration of the Ca 2+ buffer EGTA from 0.5 (estimated buffer space of , 1000 nm) to 2 mM (estimated buffer space of , 200 nm from the point of Ca 2+ source) [22], the kinetics of RRP release were unaffected both in immature E12 HCs and in mature P2 HCs (Fig. 2D). However, as expected for the recruitment of vesicle located farther than 200 nm from Ca 2+ entry, the SRP was largely reduced by 2 mM EGTA (Fig. 2C).
The rate and Ca 2+ sensitivity of vesicle supply to the RRP increase during development Auditory HCs must be able to quickly replenish the RRP to sustain neurotransmitter release [22], [24]. To address this issue, we examined the developmental changes associated with the rate of vesicle supply to the RRP in apical and basal HCs using a 100 ms paired-pulse depolarization protocol (Fig. 3). In both basal and apical immature HCs, exocytosis showed marked paired-pulse depression that decreased with maturation from 25% in E12 to 5% in P2 HCs (Fig. 3A-3B). When varying the inter-pulse interval, approximately 95% of the RRP was restored within , 6 s at E12, and , 0.7 s at E16 (e.g. time constants (s) E12: best fit with dual exponential function with t 1 = 0.860.1 and t 2 = 6.161.4 (n = 5); E16: best fit with a single exponential function t = 0.7 60.1 (n = 5), p,0.05; Fig. 3C). Notably, marked paired-pulse depression or inactivation was also observed for I Ca in immature E12 HCs as compared to mature P2 HCs. The kinetics of RRP recovery in E12 HCs paralleled the time course of I Ca recovery (Fig. 3D), indicating that DC m paired-pulse depression was mainly due to I Ca inactivation in immature HCs. By contrast, mature HCs reconstituted , 95% of the RRP within less than 200 ms and showed almost no I Ca inactivation (e.g. P2: 9562%, (n = 4), p,0.001; table 2). We did not refine the kinetics of paired-pulse recovery below 200 ms because capacitance measurements are altered by Ca 2+ tail currents below this time frame.
Next, we compared the rate of vesicle supply to the RRP when using an intracellular recording solution containing 2 mM EGTA instead of 0.5 mM (Fig. 4). Surprisingly, at E12 using either 0.5 mM or 2 mM EGTA, the RRP showed a similar level of paired-pulse depression, indicating that the refilling rate was poorly Ca 2+ -sensitive (Fig. 4A, 4C, 4D). By contrast, reduced intracellular Ca 2+ availability with high EGTA largely increased paired-pulse depression in mature HCs (e.g. at P2, (mM EGTA, % RRP recovery 200 ms after the first 100 ms-pulse): 0.5, 9562 (n = 4); 2, 1062, (n = 4), p,0.001), Fig. 4B, 4C, 4D). These results obtained in mature chick HCs are in agreement with previous findings [22] and suggest that the supply of vesicle to the RRP is Ca 2+ -sensitive. The novelty of our findings is that this vesicle supply process in immature HCs and mature HCs shows a different Ca 2+ sensitivity.

The efficiency of vesicular recruitment increases during development and allows sustained release
The RRP replenishment was also studied by stimulating the HCs with a train of consecutive brief stimuli consisting of 100 ms depolarizing steps from -90 mV to -10 mV separated by 200 ms ( Fig. 5). Cumulative DC m responses showed marked depression in immature developing HCs, as indicated by a progressive decrease in DC m responses during the repetitive stimuli: In basal high frequency HCs, DC m decreased from a mean of 2764 fF after the first stimulus to 1063 fF after the 20 th one at E12 (n = 5; p,0.001; Fig. 5G). Similar depression was observed in E12 low frequency HCs (Fig. 5D). In these immature HCs, the Ca 2+ efficiency of vesicle release (DC m /Q Ca ) from RRP remained constant during the repetitive stimulations (Fig. 5E, 5H), indicating that the depression of exocytosis mainly arose from the marked inactivation of the Ca 2+ current (Fig. 5A). By contrast, mature P2 HCs from base or apex showed no depression of the RRP, as indicated by a near linear increase in cumulative DC m (Fig. 5B-5F). Mature P2 HCs showed constant high Ca 2+ efficiency in exocytosis (Fig. 5E, 5H) and absence of I Ca inactivation (Fig. 5B).

Otoferlin expression and number of synaptic ribbons per HC increase with maturation
Several factors could explain the increase in kinetics and Ca 2+ efficiency of exocytosis in HCs during development: a reduction in the distance between the Ca 2+ channels and the sites of release at the active zone; a change in the affinity of the Ca 2+ sensor that controls membrane fusion; or a reduction in the diffusion barrier of Ca 2+ ions at the site of release. We found that the average number of ribbons per HC largely increased with development from 0.360.5 (n = 50) at E8, 2.161.6 (n = 50) at E12, 3.161.4 (n = 50) at E16 and 9.362.2 (n = 50) at P2 (Fig. 6A and table 2). These results indicate a positive correlation between Ca 2+ efficiency in exocytosis and the number of synaptic ribbons in HCs. Our observations are in agreement with the recent finding that the synaptic ribbons contribute largely to synaptic neurotransmission by facilitating high rates of exocytosis, while their absence significantly compromise the temporal resolving power of the auditory system [25], [26].
We then used the recently characterized monoclonal antibody HCS-1 [17] to explore the expression of the calcium sensor otoferlin during development of chick basilar papillae. Otoferlin was weakly expressed in HCs at embryonic stages earlier than E10 and then increased with development to reach a maximum level at E18-P2 (Fig 6B). At these late stages of development, HCS-1 immunolabeling was largely distributed at the plasma membrane from the apical part of the HCs (below the cuticular plate) to the lower end of the HCs (synaptic area). It is to be mentioned that, in absence of a true control as in mouse knock out for the otoferlin gene [5], [7], we cannot ascertain that the HCS-1 labeling is entirely specific. However, it is to be noted that both the plasma membrane labeling and the developmental increase of HCS-1 labeling matched very well the recent results obtained in mouse HCs [7].

Discussion
This report characterizes the functional changes occurring during progressive maturation of the HC synaptic machinery in a precocial post-hearing vertebrate, the chick, where sound-evoked cochlear nuclei activity can be measured as early as E11 in ovo [12], [13]. Concomitantly to an increased expression of ribbons and otoferlin, exocytosis of chick HCs progressively displayed faster kinetics and higher Ca 2+ efficiency with maturation. Similar changes have been shown in HCs of pre-hearing animals such as mouse and gerbil [9], [23]. Our study demonstrates for the first time that vesicle supply and RRP release undergo a parallel maturation to allow mature HCs to sustain high rates of exocytosis. In addition, we show that vesicle recruitment is highly Ca 2+ -dependent in mature chick HCs, in agreement with previous findings [22]. Notably, a constant vesicle trafficking from a reserve pool has also been recently proposed to be Ca 2+ -dependent in turtle auditory HCs [27].
Remarkably, immature chick HCs displayed significant depression in exocytosis during repetitive brief stimuli or paired-pulse stimulation, while mature HCs showed little RRP depression. This exocytotic depression in immature HCs is likely due in part to the rapid inactivating property of the Ca 2+ current at this developmental stage. Indeed, the Ca 2+ current and the RRP showed similar kinetics of recovery during paired-pulse stimulations. Notably, while the Ca 2+ current of mature chick HCs is mainly driven by non-inactivating dihydropyridinesensitive L-type Ca 2+ channels [28], [29], immature chick HCs (in addition to L-type channels) transiently express fast inactivating T-type Ca 2+ channels, [16] and unpublished data. Furthermore, L-type Ca 2+ currents of immature HCs display strong calmodulin-mediated calcium-dependent inactivation [30]. Therefore, Ca 2+ current inactivation leading to RRP depression could partially explain the transient rhythmic temporal discharge pattern of the auditory nerve fibers observed in young kittens [11]   and in chicken embryos [14], [31]. These immature animals show high threshold low frequency hearing and are unable to maintain a sustained steady-state response to long duration tone bursts [32]. In agreement with [22], we found that RRP replenishment, but not initial release, was diminished by using 2 mM EGTA instead of 0.5 mM in mature chick HCs. These results suggest that the release sites are less than 200 nm from Ca 2+ entry, while the reloading sites extend farther than 200 nm. Similar Ca 2+ regulation of vesicle replenishment has been shown at the cone ribbon synapses of the retina [33]. In HCs of the amphibian papilla, recovery from paired-pulse depression has recently shown to be ultrafast and also dependent on Ca 2+ [34]. The most intriguing result of our study was the observation that the RRP recovery of immature HCs, unlike mature chick HCs, was not sensitive to 2 mM intracellular EGTA. This may reflect differences in the distance of the stock of vesicular supply from the release sites and Ca 2+ entry during development. A larger and wider extrasynaptic distribution of Ca 2+ channels at the early stage of development, as shown in immature mouse HCs [35], could place Ca 2+ entry closer to the refilling machinery (reserve pool of vesicles) in immature HCs and in turn make vesicle replenishment less sensitive to EGTA. Notably, we found a positive correlation between the increasing number of ribbons with maturation and the efficiency of vesicle supply and release. Our results obtained in developing HCs are in good agreement with those obtained in a recent study using transgenic mice lacking the presynaptic scaffold protein bassoon, an essential element to dock the ribbon to the active zone [26]. The latter study concluded that the ribbon is essential for organizing Ca 2+ channels and vesicles in the synaptic active zone in order to promote efficient vesicle replenishment.
In addition to an increased number of ribbons during maturation, the organization of a different Ca 2+ -dependent vesicle supply may also progressively take place in mature HCs. Otoferlin, which is considered to be a high affinity Ca 2+ sensor that directly triggers SNARE-membrane fusion in vitro [4] and regulates Ca 2+evoked membrane fusion at the ribbon synapse of both cochlear [5] and vestibular HCs [36], could also regulate a Ca 2+ -dependent vesicle supply at a large distance from Ca 2+ entry. Indeed, we found that the expression of otoferlin increases at the right period of cochlear maturation when vesicle replenishment becomes efficient, suggesting a progressive engagement of this Ca 2+ sensor in the ribbon active zone. By progressively replacing other Ca 2+ sensors such as synaptotagmins during development [7], otoferlin may facilitate vesicle supply and release at the mature HC ribbon synapse.
Contrary to mature auditory HCs, the ribbon synapses of the retina do not express otoferlin [17], [37] and display pronounced paired-pulse depression that is attributable to a limiting slow replenishment of vesicles [38], [39]. Indeed, RRP recovery in retinal bipolar neurons displays rather slow kinetics (t , 4 to 8 s) spanning the range of what we found in immature HCs. Our present study, showing a concomitant developmental onset of a fast Ca 2+ -sensitive vesicle supply and otoferlin expression, suggests that this multi-C2 protein may also act as a Ca 2+ sensor for the recruitment of vesicles located far from the release sites and Ca 2+ channels, and probably farther than 200 nm as suggested by its sensitivity to 2 mM EGTA. This hypothesis is also reinforced by the phenotype of the pachanga mouse model, which carries a missense mutation in the C2F domain of otoferlin, and where the replenishment process of synaptic vesicles is affected independently of RRP fusion [6].