J Neurophysiol 115: 226–239, 2016. First published October 28, 2015; doi:10.1152/jn.00559.2015.

Calcium-induced calcium release supports recruitment of synaptic vesicles in auditory hair cells

X Manuel Castellano-Muñoz,1 Michael E. Schnee,1 and Anthony J. Ricci1,2 1Department of Otolaryngology, Stanford University School of Medicine, Stanford, California; and 2Department of Molecular and Cellular Physiology, Stanford University School of Medicine, Stanford, California

Submitted 8 June 2015; accepted in final form 23 October 2015

Castellano-Muñoz M, Schnee ME, Ricci AJ. Calcium-induced ronal functions such as neuronal excitability, gene expression, calcium release supports recruitment of synaptic vesicles in auditory and synaptic plasticity and release (Bouchard et al. 2003). In hair cells. J Neurophysiol 115: 226–239, 2016. First published Octo- central synapses both endoplasmic reticulum (ER) and mito- Downloaded from ber 28, 2015; doi:10.1152/jn.00559.2015.—Hair cells from auditory chondria are well-known intracellular Ca2ϩ stores, and their and vestibular systems transmit continuous sound and balance infor- Ca2ϩ homeostatic modulation alters synaptic transmission pre- mation to the central nervous system through the release of synaptic and postsynaptically (Bardo et al. 2006; Emptage et al. 2001; vesicles at ribbon synapses. The high activity experienced by hair Llano et al. 2000). CICR is also suggested to contribute to cells requires a unique mechanism to sustain recruitment and replen- synaptic transmission at ribbon synapses (Babai et al. 2010; ishment of synaptic vesicles for continuous release. Using pre- and

Lelli et al. 2003). http://jn.physiology.org/ postsynaptic electrophysiological recordings, we explored the poten- tial contribution of calcium-induced calcium release (CICR) in mod- Calcium imaging identified CICR in turtle auditory papilla ulating the recruitment of vesicles to auditory hair cell ribbon syn- hair cells (Tucker and Fettiplace 1995), frog semicircular canal apses. Pharmacological manipulation of CICR with agents targeting (Lelli et al. 2003), P6–P11 mouse inner hair cells (Iosub et al. endoplasmic reticulum calcium stores reduced both spontaneous post- 2015; Kennedy and Meech 2002), and rat and guinea pig outer synaptic multiunit activity and the frequency of excitatory postsyn- hair cells (Evans et al. 2000; Mammano et al. 1999). In aptic currents (EPSCs). Pharmacological treatments had no effect on mammalian outer hair cells, CICR is functionally associated to hair cell resting potential or activation curves for calcium and potas- subsynaptic Ca2ϩ stores in close proximity to efferent termi- sium channels. However, these drugs exerted a reduction in vesicle nals (Lioudyno et al. 2004). In addition, Ca2ϩ can be released release measured by dual-sine capacitance methods. In addition, by inositol triphosphate-gated Ca2ϩ stores at the base of the by 10.220.33.6 on September 23, 2016 calcium substitution by barium reduced release efficacy by delaying outer hair cell hair bundle (Mammano et al. 1999). Although release onset and diminishing vesicle recruitment. Together these results demonstrate a role for calcium stores in hair cell ribbon pharmacological data demonstrate the presence of intracellular stores in hair cells, their physiological role is debatable. Intra- synaptic transmission and suggest a novel contribution of CICR in 2ϩ hair cell vesicle recruitment. We hypothesize that calcium entry via cellular Ca stores have been functionally associated with the calcium channels is tightly regulated to control timing of vesicle control of BK channel activity in inner hair cells (Beurg et al. fusion at the synapse, whereas CICR is used to maintain a tonic 2005; Marcotti et al. 2004), modulation of outer hair cell calcium signal to modulate vesicle trafficking. electromotility (Dallos et al. 1997), homeostatic control of 2ϩ hair cell; dual-sine capacitance; Ca2ϩ-induced Ca2ϩ release; intra- presynaptic Ca levels (Kennedy and Meech 2002; Tucker cellular stores; ribbon synapse; synaptic transmission and Fettiplace 1995), time-dependent segregation of afferent and efferent signaling (Im et al. 2014), and regulation of vesicular trafficking, exocytosis, and synaptic transmission HAIR CELLS, the sensory receptors in the auditory and vestibular (Hendricson and Guth 2002; Lelli et al. 2003). systems, convert mechanical information into synaptic activity Here we performed auditory nerve multiunit and single-unit through the release of neurotransmitter at ribbon synapses. recordings as well as hair cell dual-sine capacitance experi- Each hair cell contains tens of synaptic ribbons (Schnee et al. ments to study the potential contribution of CICR to hair cell 2005, 2011; Sneary 1988), presynaptic specializations sur- synaptic transmission. Pharmacological and divalent cation rounded by synaptic vesicles and associated to active zones and substitution results are consistent with a role for CICR in the ϩ L-type Ca2 channels (Issa and Hudspeth 1994; Roberts et al. recruitment of vesicles to support maintained release in audi- 1990; Tucker and Fettiplace 1995). Similar to other sensory tory hair cell ribbon synapses. synapses, hair cell ribbon synapses operate in a graded fashion, reaching high release rates and exhibiting little fatigue. Both of MATERIALS AND METHODS these properties require rapid vesicle replenishment by a mech- anism that is not well understood. Tissue preparation. The auditory papilla of red-eared sliders (Tra- 2ϩ 2ϩ chemys scripta elegans) was dissected as previously described (Sch- Ca -induced Ca release (CICR) is a mechanism by nee et al. 2005). All animal procedures were approved by the Stanford which the influx of Ca2ϩ through Ca2ϩ channels in the plasma 2ϩ Institutional Animal Care and Use Committee (IACUC) and are in membrane activates Ca release from intracellular stores accord with National Institutes of Health guidelines and standards. (Verkhratsky 2005). CICR is implicated in a number of neu- Turtle half-head preparations were used for multiunit activity mea- surements from the eighth cranial nerve. The turtle head was split in Address for reprint requests and other correspondence: M. Castellano- half and pinned in a Sylgard dissection chamber either with an Muñoz, Inst. of Bioengineering, Miguel Hernández Univ., Avenida de la external solution similar to that used in patch-clamp recordings or Universidad, s/n 03202 Elche, Alicante, Spain (e-mail: [email protected]). with bicarbonate-buffered perilymph containing (in mM) 126 NaCl,

226 0022-3077/16 Copyright © 2016 the American Physiological Society www.jn.org STORED CALCIUM PROMOTES VESICLE RECRUITMENT TO RIBBON SYNAPSES 227

2.5 KCl, 13 NaHCO3, 1.7 NaH2PO4, 1.8 CaCl2, 1 MgCl2, and 5 preparation in the recording dish. A gravity-controlled perfusion glucose (continuously bubbled with 95% O2-5% CO2). The brain was pipette was located ϳ5 mm above the otic capsule and was connected removed and the auditory nerve exposed (Fig. 1A), cutting the con- to a perfusion system with a flow rate of roughly 1 ml/min. nections to posterior ampulla and saccule. The ventral otic membrane For intracellular hair cell recordings, the inner ear was dissected was trimmed to allow access to perfusion prior to mounting of the from the otic capsule in external solution containing (in mM) 128

NaCl, 0.5 KCl, 2.8 CaCl2, 2.2 MgCl2, 10 HEPES, 6 glucose, 2 A Drug Delivery Recording creatine monohydrate, 2 ascorbate, and 2 pyruvate and pH was Pipette adjusted to 7.6 and osmolality to 275 mosmol/kg. The external solution was supplemented with 20 ␮M curare to eliminate efferent activity, and 100 nM apamin was included in some of the experiments to block SK activity. We found no evidence of remaining efferent activity with incubation with curare in both pre- and postsynaptic recordings. To disrupt mechanotransduction channels, after extrane- ous tissue was trimmed and the otoconia removed the papilla was Neck Nose incubated for 15 min in external solution and perfused with 200 ␮lof Downloaded from B 5 mM BAPTA before and after removal of the tectorial membrane 0.4 with a fine insect pin. In some experiments where the tectorial membrane was left intact, cell visualization was impaired but no obvious electrophysiological effects were observed. The basilar pa- 0.2

Vm pilla was transferred to the recording chamber and secured with single strands of dental floss. Cells were imaged with an Axioskop 2 FS plus 0 (Zeiss, Thornwood, NY) with bright field optics using a ϫ60 0.9 NA water objective (LUMPlan Fl/IR, Olympus). Perfusion of bath and http://jn.physiology.org/ caffeine drugs was delivered with a Minipuls 3 pump (Gilson, Middleton, WI). -0.2 Electrophysiology. For multiunit activity we used the turtle half- 50 head preparation, in which the auditory nerve was inserted into an hourglass-shaped suction electrode with a micromanipulator (Na- rishige, East Meadow, NY) and compound action potentials were

s/sekips recorded with a differential AC preamplifier (Grass, P55 Astro-Med, West Warwick, RI). One electrode was inserted into the borosilicate 25 * suction pipette, and the neutral electrode was in contact with the bath.

The signal was band-pass filtered (1 Hz–1 kHz) and amplified 1,000 by 10.220.33.6 on September 23, 2016 times. Compound action potentials were collected through a data acquisition interface (CED Micro 1401 mkII, Cambridge Electronic 0 Design) and analyzed with Spike2 software (Cambridge Electronic 0 1000 2000 Design). Noise levels were identified by blocking afferent activity time (s) with 1 ␮M tetrodotoxin (TTX) or prolonged high potassium concen- C tration, and spike threshold was set to 3 ϫ SD of baseline noise. Drugs were applied via the local perfusion system after a baseline firing rate

/ ) / was established. Control perfusion with the external solution (sans ini 1 ** drug) was used to confirm absence of mechanical artifacts and firing etar ekips-eta etar ** stability throughout recording (96 Ϯ 2% after 20 min, n ϭ 3). ini For hair cell patch-clamp experiments, thick-walled borosilicate etar ekips etar ** 0.5 electrodes of resistance 2.5–3.5 M⍀ were used with internal solution containing (in mM) 110 CsCl, 1 EGTA, 5 creatine phosphate, 3

Na2ATP, 10 HEPES, 3 MgCl2, and 2 ascorbate, pH adjusted to 7.2

r ekips( ** ** and osmolality at 255 mosmol/kg. Stimulus protocols were performed 0 starting 10 min after whole cell configuration to allow solution equilibration and run up stabilization of the Ca2ϩ current (Schnee and Ricci 2003). Hair cells were voltage clamped with an Axopatch 200B (Axon Instruments-Molecular Devices, Sunnyvale, CA) or a VE-2 KynA BHQ TPP+ control DNQXcaffeine amplifier (Alembic Instruments, Montreal, ON, Canada). Data were ryanodineantimycinA collected at 100–200 kHz with an IOTech Daq/3000 acquisition board Fig. 1. Pharmacological disruption of intracellular calcium stores reduces (MC Measurement Computing, Norton, MA) driven by jClamp soft- extracellular spike activity in the 8th cranial nerve. A: low-power view of ware (SciSoft). Voltage was intentionally not corrected for junction half-head preparation with the recording suction electrode and drug application potential or series resistance to match the values used in two-sine pipette labeled. B, top: effect of caffeine application on spontaneous nerve capacitance protocols (Schnee et al. 2011). Dual sinusoidal stimula- activity. Bottom: spontaneous spike rate. Asterisk shows the time point of tion was performed to compensate in-cell stray capacitance at differ- maximum effect after drug delivery selected for rate quantification. C: box ent frequencies before capacitance protocols were run. A dual sinu- plots illustrating spike rate reduction by drugs that interfere with endoplasmic soid with amplitudes and frequencies of 20–30 mV at 3.1–6.2 kHz ␮ reticulum (ER) calcium homeostasis (10 mM caffeine, 50–100 M BHQ, and and 6.2–9.4 kHz was delivered superimposed to the desired voltage 60 ␮M ryanodine). In this and subsequent figures, box plots present raw data (symbols), mean (star), and SD (box). Pharmacological modulation of mito- step. Capacitance measurements were low-pass filtered at 40 Hz, and chondrial calcium homeostasis [100 ␮M tetraphenylphosphonium (TPPϩ) and the onset-offset gating capacitative transients were removed off-line. 10 ␮M antimycin A] reduced spontaneous rate to a lesser extent. Kynurenic Afferent fiber patch recordings were done with solutions similar to acid (KynA) and DNQX, 2 glutamatergic receptor antagonists, were used as those described for hair cell recordings (Schnee et al. 2013). Patch controls. **Paired t-tests show a significant difference at P Ͻ 0.01 level. electrodes were smaller tipped and had resistances of 9–11 M⍀.

J Neurophysiol • doi:10.1152/jn.00559.2015 • www.jn.org 228 STORED CALCIUM PROMOTES VESICLE RECRUITMENT TO RIBBON SYNAPSES

Series resistance was compensated to 70%, resulting in uncompen- dent SK channel, did not vary the spontaneous activity reduc- sated series resistance of 11 Ϯ 3M⍀ (n ϭ 5). tion obtained with caffeine (data not shown), thus ruling out a Data analysis. In hair cell patch-clamp experiments, the initial potential contribution of SK-evoked hyperpolarization due to capacitance of cells located at the center of the papilla was 12.4 Ϯ 1.4 2ϩ ϭ Ca release from ER. Reduction during prolonged caffeine pF (n 47) 1 min after establishment of whole cell configuration. application could alternatively be explained by eventual Ca2ϩ Cells were discarded when the uncompensated series resistance was Ͼ12 M⍀ because of the difficulty in the neutralization of in-cell stray depletion in the ER and a consequent impairment of a potential capacitance. Cells were also discarded when the leak current was Ͼ50 CICR mechanism. pA after 9 min of recording to avoid inactivation due Caffeine reduces postsynaptic activity in auditory fibers. The to slow calcium loading of the cells. Data and graphics show means Ϯ pharmacological effects observed in our multiunit preparation SD and number of experiments (n) unless noted. Statistical analyses cannot, however, distinguish between a presynaptic and a were performed with a two-tailed Student’s t-test assuming normal postsynaptic contribution of stores to auditory synaptic trans- distribution. In the capacitance experiments where multiple release mission (Fitzjohn and Collingridge 2002). To obtain further protocols were evoked, normalized percentage values are related to evidence of a potential role for CICR in synaptic activity, we the value evoked in the first pulse. In figures, the y-axis represents the tested the effect of caffeine on excitatory postsynaptic currents amount of release in a second stimulation protocol normalized by a (EPSCs) measured from individual afferent fibers from the Downloaded from first stimulation protocol (2nd Ϫ 1st pulse/1st pulse); therefore values below 0 represent a reduction in the second pulse, whereas values near auditory nerve (Schnee et al. 2013). Postsynaptic afferent 1 represent twice as much release in the second pulse. patch-clamp recordings were made and spontaneous activity recorded from the neurons (Fig. 2). Application of caffeine (10 RESULTS mM) resulted in a net decrease in EPSC frequency that could be recovered upon washout (Fig. 2A). Change in frequency for Pharmacological manipulation of calcium stores reduces nine fibers is presented in Fig. 2C. Of these, five are whole cell auditory nerve activity. The ability of neuronal ER and mito- recordings and four are cell-attached recordings where spike http://jn.physiology.org/ chondria to store and release Ca2ϩ has been extensively char- rate could be monitored. In all cases the frequency of release acterized (Nicholls 2009; Tang and Zucker 1997; Verkhratsky was reduced. Frequency histograms for EPSC amplitudes were 2005; Wan et al. 2012). To study the potential contribution of also generated; an example is presented in Fig. 2B. The mean these Ca2ϩ stores in hair cell synaptic release, we first tested EPSC amplitude tended to be reduced, perhaps because of a the pharmacological effect of Ca2ϩ store modulators on the loss of synchrony (Schnee et al. 2013); however, this reduction auditory nerve firing rate, using an extracellular multiunit was not statistically significant (Fig. 2D). In three of four fiber preparation (Fig. 1A). Spike activity in control experiments recordings there was a transient increase in EPSC frequency was abolished by TTX (1 ␮M), kynurenic acid (2 mM), or followed by a decrease. Additionally, upon washout in three of by 10.220.33.6 on September 23, 2016 DNQX (1 ␮M) (Fig. 1C), confirming the glutamatergic nature four cells there was an initial overshoot in EPSC frequency of the hair cell ribbon synapse and that neural activity was (data not shown), indicative of an increased permeability to driven by synaptic activity. Caffeine (10–20 mM), a nonspe- calcium, perhaps due to the activation of store-operated cal- cific ER modulator that keeps ryanodine receptors (RyRs) in a cium channels (Lukyanenko et al. 2001). Together these data semiopened state (Zucchi and Ronca-Testoni 1997), reduced support the conclusion that there is a presynaptic role for CICR the spike rate to 28 Ϯ 21% of its initial rate (n ϭ 8, P ϭ in regulating synaptic vesicle release. 0.0001) (Fig. 1, B and C). BHQ, a blocker of the SERCA pump Both multiunit and single postsynaptic bouton recordings in the ER, reduced the spike activity to 44 Ϯ 31% of its initial suggest that pharmacological manipulations of CICR modulate rate (n ϭ 6, P ϭ 0.001). Similarly, ryanodine, which blocks action potential rate by reducing EPSC frequency. The fre- RyRs at high concentrations (60 ␮M), reduced multiunit ac- quency of EPSCs is driven presynaptically as the rate of vesicle tivity to 73 Ϯ 26% of control [n ϭ 6, P ϭ not significant fusion. In turn, fusion can be modulated at multiple levels that (n.s.)]. Ruthenium red (40 ␮M), a nonspecific inhibitor of RyR, are direct or indirect. For example, hyperpolarizing the hair cell also reduced spike activity (n ϭ 1, data not shown). Incubation will reduce release, inactivating Ca2ϩ channels might reduce with drugs known to reduce mitochondrial Ca2ϩ buffering by release directly, and reduction in the Ca2ϩ current might also interfering with mitochondrial membrane potential, such as alter release properties or vesicle trafficking and recycling tetraphenylphosphonium (TPPϩ) and antimycin A, reduced (Grant and Fuchs 2008; Johnson et al. 2008; Lee et al. 2007; spike activity to a lesser extent. Spiking was reduced by TPPϩ Magistretti et al. 2015). The following experiments systemat- (100 ␮M) to 74 Ϯ 4% of control (n ϭ 5, P ϭ 0.002) and by ically evaluate each potential mechanism for reducing synaptic antimycin A (10 ␮M) to 77 Ϯ 22% of control (n ϭ 4, P ϭ n.s.). vesicle fusion. Although pharmacological manipulation suggested a potential Caffeine has no effect on presynaptic electrical properties. contribution of both mitochondrial and CICR Ca2ϩ stores to A presynaptic effect is postulated to underlie the reduction in synaptic activity, we concentrated our attention on the ER, postsynaptic activity after pharmacological manipulation of which provided more robust effects. Ca2ϩ stores (Bouchard et al. 2003). The observed reduction in The contribution of CICR to hair cell synaptic transmission EPSC frequency could be explained by caffeine-driven hair was first observed with vestibular nerve recordings (Hendric- cell hyperpolarization. Caffeine could trigger the release of son and Guth 2002; Lelli et al. 2003; Rossi et al. 2006). In our Ca2ϩ from intracellular stores and activate SK potassium experiments, application of caffeine reduced the spontaneous channels, thus hyperpolarizing the hair cell and reducing re- spiking rate, consistent with the ER depletion effect reported in lease probability. To examine this possibility, we tested the other systems (Albrecht et al. 2002; Alonso et al. 1999; effect of caffeine perfusion on the electrical properties of hair Hongpaisan et al. 2001; Pozzo-Miller et al. 1997). Additional cells (Fig. 3). The responses of hair cells recorded in current application of 100 nM apamin, a blocker of the Ca2ϩ-depen- clamp to 10-pA current injections are presented in Fig. 3A in

J Neurophysiol • doi:10.1152/jn.00559.2015 • www.jn.org STORED CALCIUM PROMOTES VESICLE RECRUITMENT TO RIBBON SYNAPSES 229 AB 25 20 15 10 Events 5 0 0.0 0.2 0.4 0.6 0.8 EPSCs (nA)

CD40 Control 200 pA 30 0.4 100 ms wash 20 0.3 Downloaded from

10 0.2 Frequency (Hz) Frequency 0 0.1 EPSC Peak (nA) EPSC Peak Control 0.0

Fig. 2. Postsynaptic patch recordings from afferent terminals show a reduction in excitatory postsynaptic current (EPSC) frequency and firing rate with caffeine http://jn.physiology.org/ application. A: spontaneous EPSCs from a whole cell recording of an afferent terminal. Red traces are in the presence of 10 mM caffeine, and blue traces are activity during washout of the drug. B: frequency histograms for EPSC amplitudes in the absence (black) and presence (red) of caffeine, where solid lines are Gaussian fits to the histograms. C: EPSC frequency (or extracellular spike frequency) before and after caffeine administration. Paired t-tests show a significant difference at P Ͻ 0.001 level. D: mean EPSC amplitude as measured from the fits to the Gaussian curves in B. Although a trend toward smaller mean amplitudes was found, no statistical difference was observed. the absence and presence of caffeine. We found no change in variability found in hair cells, particularly with repeated mea- resting potential and a variable effect on the electrical reso- sures (Levic et al. 2011; Patel et al. 2012; Quinones et al. 2012; nance response. A summary of the resting potential data is Rutherford and Roberts 2006). Two components of release presented in the box plots of Fig. 3B, where no consistent were observed, an initial linear component, previously demon- by 10.220.33.6 on September 23, 2016 change was observed. In several cells the quality of the reso- strated to be accounted for by vesicles within 0.7 ␮mofthe nance was reduced, but this was not statistically different, nor synapses, and a larger superlinear component that requires did it typically recover upon washout. recruitment of additional vesicles (Fig. 4A) (Schnee et al. 2005, It is also possible that the reduced frequency of release 2011). The first linear release component is proportional to the obtained with caffeine is due to the efflux of calcium from Ca2ϩ load (integral of calcium current) and correlates with the calcium stores, leading to Ca2ϩ channel inactivation at the 2ϩ vesicle population near the synapses (Schnee et al. 2005, synapse (Schnee and Ricci 2003). We compared hair cell Ca 2011). In our hands the traditional pool descriptions of rapidly current amplitudes in response to a depolarizing pulse before and readily released are less valuable in that pool size as and during drug application (caffeine or ryanodine) in the measured by depletion varies within a given cell quite substan- external solution (Fig. 3C). The same protocol (two 3-s depo- larizing pulses separated by 5 min) was performed with drugs tially and is often difficult to observe, suggesting a very in the internal solution (8-BrcAMP or thapsigargin) (Fig. 3D). dynamic population of vesicles with robust recruitment (Sch- Data summarized in Fig. 3, C and D, suggest that Ca2ϩ current nee et al. 2005, 2011). The first component includes these amplitude was not significantly altered by drug application. pools but likely additional vesicles from immediately near the Thus the most parsimonious explanation for the reduction in synapse and potentially some vesicles that are recruited during depolarization. The second superlinear component presents a vesicle release is a synaptically driven mechanism. ϩ We also investigated the macroscopic currents elicited in higher release rate and is independent of Ca2 load. Similar voltage clamp (Fig. 3, E–H). Macroscopic currents elicited superlinear release components are observed in other secretory from both hyperpolarizing and depolarizing voltage steps did cells (Andersson et al. 2011; Seward et al. 1996) as well as not vary significantly in the presence of caffeine. Current- photoreceptors and hair cells by application of trains of short voltage plots elicited from tail currents at the time point stimulation pulses with single-sine capacitance techniques indicated in Fig. 3E and plotted in Fig. 3G were also not (Duncker et al. 2013; Innocenti and Heidelberger 2008; Moser significantly different between control and caffeine treated. and Beutner 2000). At present, we consider the superlinear The voltages of half-activation summarized in Fig. 3H were release to represent the hair cell’s exocytotic activity when also not significantly different. Therefore, these data indicate release sites are maximally filled and releasing. This release that the changes recorded postsynaptically during caffeine component requires significant recruitment of vesicles (Schnee perfusion are likely to be a direct effect on synaptic properties et al. 2005, 2011). This interpretation remains a hypothesis within the hair cell. requiring further investigation. Calcium imaging experiments Time dependence of superlinear release. The two-sine tech- also showed a nonlinear rise in the intracellular Ca2ϩ levels nique (Schnee et al. 2011) was first used to monitor vesicle accompanying the superlinear capacitance change (Schnee et release during cell depolarization to avoid the intercellular al. 2011). One possible interpretation of this nonlinear signal-

J Neurophysiol • doi:10.1152/jn.00559.2015 • www.jn.org 230 STORED CALCIUM PROMOTES VESICLE RECRUITMENT TO RIBBON SYNAPSES

A control B -30 Control 10 mM Caffeine -40 -50 caffeine -60

5 mV Vz (mV) -70

5 ms -80

C 0.0 D 0.0 Fig. 3. Presynaptic excitability is not respon- caffeine sible for the reduction in vesicle release. A: -0.1 -0.1 8Br-cAMP membrane potential responses from a hair Control ryanodine cell elicited by a 10-pA current injection Control thapsigargin about the cell’s resting potential: control -0.2 -0.2 Downloaded from (top) and in the presence of 10 mM caffeine (nA)

(bottom). The slight reduction in quality of -0.3 (nA) -0.3 Ca I resonance is not significantly different. B: Ca I zero-current potential (Vz) in the absence -0.4 -0.4 (black) and presence (red) of caffeine. No statistical difference is noted. C: calcium cur- -0.5 -0.5 rent (ICa) elicited in paired stimulus for cells in the absence and presence of drug (red, caffeine; blue, ryanodine; black, control). No http://jn.physiology.org/ change is identified. D: summary of calcium E 8 F 8 currents elicited from 1st (darker colors) and 2nd (lighter shades) stimuli for drugs that 6 control 6 caffeine were applied internally. No significant differ- ences were found. E and F: voltage-clamp 4 4 responses for step voltage changes between Ϫ104 and ϩ60 in 20-mV increments from a IK (nA) 2 2 Ϫ holding potential of 84 mV for control (E) 0 0 and caffeine (F). G: tail current voltage plots generated at the time point indicated by -2 -2 by 10.220.33.6 on September 23, 2016 dashed line in E. Black, control; red, caf- 100 125 150 175 100 120 140 160 180 feine. No difference was found between plots. H: half-activating voltages (V1/2)ob- Time (ms) Time (ms) tained from data in G. No significant differ- ence was observed. G H 0 1.6 -10 -20 1.2 -30 V1/2 I (nA) 0.8 -40

0.4 -50 -60 -100 -80 -60 -40 -20 0 20 Control caffeine Potential (mV) ing is a potential second internal source of Ca2ϩ, perhaps control hair cells in order to establish a comparator for subse- supporting CICR. quent pharmacological experiments. Control cells were volt- Our postsynaptic results point to a potential role of hair cell age-clamped to Ϫ85 mV, and consecutive 3-s stimulation ER Ca2ϩ homeostasis in the ability to sustain ribbon synaptic pulses to 50% of peak Ca2ϩ current (as estimated from current- transmission. To test whether ER Ca2ϩ homeostasis modulates voltage plots generated for each cell) were tested. These vesicle release, we used the dual-sine capacitance technique to relatively long protocols were selected to ensure observation monitor release properties under control conditions (Fig. 4) and and separation of the two release components, while at the during pharmacological treatment (Figs. 5–7). Previous exper- same time not overtaxing the cell so that it could replenish for iments using single-sine stimuli have demonstrated a presyn- multiple stimulations (Schnee et al. 2005). Also, these stimu- aptic facilitation effect that results in variation in responses to lations are closer to physiological than traditional depolariza- repeated stimuli, which could make interpretation of pharma- tions where potentials are stepped to elicit peak calcium cological manipulations requiring repeated measures difficult currents. (Cho et al. 2011; Schnee et al. 2011). Furthermore, since Figure 4B shows an example in which hair cell calcium repeated stimulation using long depolarizing pulses has not current and capacitance were measured sequentially in re- been described, we first characterized release properties of sponse to a stimulation that elicited a linear and a superlinear

J Neurophysiol • doi:10.1152/jn.00559.2015 • www.jn.org STORED CALCIUM PROMOTES VESICLE RECRUITMENT TO RIBBON SYNAPSES 231

A -30 mV -80 mV

linear 1 pF superlinear Cm

I Ca2+ 100 pA

1 s

B 111min3

Fig. 4. Long interpulse intervals (IPIs) result in Downloaded from an increase in hair cell superlinear release. A: representative example showing linear and su- perlinear components of release using 2-sine

capacitance method. Cm, membrane capaci- tance. B: consecutive 3-s depolarizing pulses to 50% of peak current were delivered in whole cell. Top: release. Bottom: calcium currents.

Note the increase in release for 3-min IPI http://jn.physiology.org/ (scale: 100 fF/10 pA, 1 s). C: traces from A C D total release superimposed. Note that superlinear release on- set (arrowheads) started earlier after longer 1.5 IPIs. D: normalized data showing that release )Fp( 2 eslup 2 )Fp( was enhanced only for 3- to 5-min IPIs (n ϭ 19 1 cells). E and F: box plots showing that super- linear release increased only for 3- to 5-min intervals whereas linear release was un- 0.5 1 min changed. G: superlinear onset shortened only 3 min for 3- to 5-min intervals. **Paired t-tests show by 10.220.33.6 on September 23, 2016 5 min a significant difference at P Ͻ 0.01 level. 0 100 fF 00.51 1.5 1 s pulse 1 (pF)

E linear release F superlinear release G Δ superlinear onset 0.4 ** **

ts1/ ts1/)ts1-d 2 n.s. 0.5 **** 0.2 )ts1-dn2( 1 0 0 -0.5 0

n2(

-0.2 2nd-1st (s) -1.0 -0.4 -1 -1.5 135 135 135 IPI (min) IPI (min) IPI (min) component. In this example, multiple depolarizing pulses sep- that is, until the superlinear onset of the second pulse. For arated by 1 min led to a slight release reduction, consistent with pulses separated by 1 min, release was the same as the initial vesicle depletion. Conversely, release was unexpectedly en- value (n ϭ 5 cells) (Fig. 4D). Both components showed hanced when pulses were separated by 3 min (Fig. 4B). The nonsignificant differences in the second pulse: superlinear amount of release was not significantly modified when two release reduction of 2 Ϯ 17% and linear release reduction of consecutive pulses were separated by 1 min (Fig. 4D). How- 4 Ϯ 7% (Fig. 4, E and F). Similarly, the superlinear release ever, longer interpulse intervals (IPIs) led to a significant onset was constant in the second pulse (0.02 Ϯ 0.1 s, n.s.) (Fig. release increase in all cells studied. Changes in total release as 4G). However, when pulses were separated by 3 min, consec- well as linear and superlinear release were tested by two utive pulses unexpectedly led to total release enhancement (38 consecutive pulses using different IPIs (1, 3, and 5 min) in a Ϯ 25% increase, n ϭ 5 cells, P ϭ 0.02) (Fig. 4D). Enhance- total of 19 cells. The linear component was quantified by ment was also observed for 5-min IPI (52 Ϯ 32%, n ϭ 9 cells, measuring the increase in membrane capacitance from the P ϭ 0.0001). Release enhancement was not accompanied by an onset of depolarization until the appearance of the superlinear increase in Ca2ϩ load (1 Ϯ 2% increase for 3-min IPI, 5 Ϯ 8% component. Since we observed variability in the onset of the for 5-min IPI). The increase in release resulted from an in- superlinear component between pulses (see below), linear crease in the superlinear component for IPIs of 3 min (110 Ϯ release was measured in both pulses until the same time point, 42% increase, P ϭ 0.007) and 5 min (115 Ϯ 39%, P ϭ 0.0001

J Neurophysiol • doi:10.1152/jn.00559.2015 • www.jn.org 232 STORED CALCIUM PROMOTES VESICLE RECRUITMENT TO RIBBON SYNAPSES

for 5 min) (Fig. 4F), whereas the linear component did not were no facilitation the baseline for control would be at 0 (as change significantly (5 Ϯ 11% reduction for 3 min, 4 Ϯ 10% in the linear response) (Fig. 6). The linear release component increase for 5 min) (compare y-axes in Fig. 4, E and F). The increased 4%, whereas superlinear release increased by 115% observed release enhancement was originated by a shortening (Fig. 6, B and C). The increases seen in response to a second in the superlinear release onset time for IPIs of 3 min (0.6 Ϯ pulse are used as the comparator for the pharmacological 0.3 s, P ϭ 0.01) as well as 5 min (0.5 Ϯ 0.2 s, P ϭ 0.0001) manipulations such that the value for total release in control (Fig. 4, C and G) and not by a change in the rate (slope) of the was 152%, whereas the linear release was 104% and superlin- superlinear release. These data show that 1 min is enough for ear release 215% of first pulse. The same depolarization hair cell buffering mechanisms to reach a steady state in protocol was delivered before and after continuous extracellu- cytoplasmic calcium levels after the long depolarizing pulse. lar application of 10 mM caffeine (Fig. 5B)or10␮M ryano- Therefore the observed increase in release after 3–5 min cannot dine. Both pulses were separated in time by 5–10 min to ensure be accounted for by an increase in baseline calcium levels due drug access through the papilla. In the presence of caffeine, to buffer saturation. Release enhancement, however, may be mean total release in the second pulse was 37% of first pulse consistent with an increase in ER Ca2ϩ reuptake leading to an release (Fig. 6A), where linear release reached 65% and super- 2ϩ increase in luminal Ca levels for subsequent stimulation linear release 40% of the control pulse. Ryanodine reduced Downloaded from pulses. Alternatively, it may underlie the activation of Ca2ϩ- total release to 58% of the initial pulse, where mean linear dependent second messengers that could modulate Ca2ϩ stores release reached 85% and superlinear release just 41% of or sensitivity to Ca2ϩ influx or activate store-dependent ex- control values. Similarly, when 30 ␮M 8-BrcADPR, a RyR pression of synaptic (Alkon et al. 1998; Benech et al. antagonist, was included in the internal solution, total release 1999; Sutton et al. 1999). Although we cannot categorically was 46% of control (Fig. 5C). Addition of 0.5–2 ␮M thapsi- interpret the physiological relevance of this release shift, the gargin, a SERCA inhibitor, to the internal solution reduced control data are needed for the pharmacological experiments release to 65% of control, with the linear component maintain- http://jn.physiology.org/ described below. As these data demonstrate that repeated ing 89% of control and the superlinear component reducing to ␮ stimulations alter the response, it is critical to characterize the 56% of control. Inclusion of 1 M xestospongin C, an IP3 change in order to accurately assess pharmacological manipu- receptor blocker, had minimal effects on total release at 85% of lations that require repeated measures. control, while 4 ␮M cyclic ADP ribose (cADPR), a RyR To study whether the shift in the superlinear release onset agonist, also showed limited efficacy, reaching 95% of control was Ca2ϩ dependent and release specific or could also affect release values. The shift in the superlinear release onset ob- other Ca2ϩ-dependent processes taking place distant from the served in controls was only significantly reduced by ryanodine

synaptic ribbon, we tested the threshold for SK channel acti- application (Fig. 6D). by 10.220.33.6 on September 23, 2016 vation, a channel located near efferent terminals (Lioudyno et Total release was significantly reduced by four of the six al. 2004). Hair cells were stimulated and SK was monitored by compounds tested, with those altering RyRs being more effec- removing apamin from the external solution (see examples in tive (Fig. 6A). Separating effects into release components Fig. 5A). In 120 of 124 cells superlinear release was preceded demonstrates that the largest effects were on the superlinear 0.7 Ϯ 0.9 s by SK activation. Superlinear release onset corre- component. Figure 6B shows that only three compounds sig- lated with SK onset (R2 ϭ 0.93), demonstrating that Ca2ϩ nificantly affected the linear component of release. Overall levels reached intracellular locations not necessarily associated linear release in the second pulse was reduced to 84 Ϯ 26% of with the vicinity of the ribbon. In mammalian outer hair cells, control, while the superlinear component was reduced to 53 Ϯ an ER-like cistern opposing every efferent contact is proposed 29% of control. Nineteen of twenty-nine pharmacologically to act as a Ca2ϩ store to amplify Ca2ϩ levels for SK activation treated cells showed a reduction from baseline in superlinear evoked by nicotinic acetylcholine receptor (nAChR) opening release (Fig. 6, B and C) compared with zero of nine control (Fuchs 2014; Lioudyno et al. 2004). In those studies, ryanodine cells. Thus the magnitude of the pharmacologically driven and other store modulators alter SK currents, a result that release reduction was more robust in the superlinear response contrasts with our experiments, where such an obvious effect (compare y-axes in Fig. 6, B and C). Furthermore, the drug was not observed during continuous application of curare, an effects are likely underestimated when compounds are used in inhibitor of nAChRs. The activation of SK channels could be the patch electrode because their effect likely begins prior to directly triggered by the influx of Ca2ϩ through L-type Ca2ϩ the initial measurement. Supporting this contention is the channels, thus circumventing the need for stored Ca2ϩ efflux to reduction in maximal release rate for the first response in cells activate SK channels during depolarization. Moreover, the where drugs are intracellularly perfused (0.5 Ϯ 0.3 pF/s, n ϭ sensitivity to Ca2ϩ might be biochemically modulated and thus 19) compared with control values (0.8 Ϯ 0.6 pF/s, n ϭ 19). underlie changes in timing as has been described in neurons The superlinear component of release could be reduced in (Adelman et al. 2012). several ways: first, the onset of superlinear release could be Pharmacological manipulation of calcium stores reduces lengthened; second, the maximal response could be reduced hair cell release. The effect of Ca2ϩ store modulators in hair (saturation); and third, the rate of release (slope) could be cell release was tested with the dual-sine capacitance technique reduced. Figure 6D demonstrates that the onset of the super- (Figs. 5 and 6). Figure 5A shows the release enhancement linear response was only affected minimally and only statisti- observed in consecutive control stimulations, mainly due to the cally significant for ryanodine. Also, saturation of release did early onset of the superlinear component. Total release in- not appear to be the limiting factor, as depicted in the examples creased by an average of 52% when comparing two pulses of Fig. 5. As also seen in Fig. 5, the maximal release rate separated by 5 min (Fig. 6A). Data are normalized to account (slope) was reduced in the presence of drugs that inhibited for the facilitation observed in control conditions; thus if there CICR. In pharmacologically treated cells where total release

J Neurophysiol • doi:10.1152/jn.00559.2015 • www.jn.org STORED CALCIUM PROMOTES VESICLE RECRUITMENT TO RIBBON SYNAPSES 233 A 0.4 control )An( mI )An( 0.2 0 -0.2 -0.4 min 10 min 15 min 20

)Fp( mC )Fp( 1 Fig. 5. Pharmacological disruption of ryano- dine receptors (RyRs) reduces hair cell re- lease. A: 3 control 3-s depolarizing pulses

were delivered 5 min apart, and current (Im) 0 and Cm were monitored. Consecutive proto- cols showed release enhancement (dashed 024602460246 line). SK and superlinear release onsets were time (s) time (s) time (s) triggered earlier in successive stimulations. Downloaded from Arrowheads show SK activation, and arrows show initial linear release onset. Baseline was BCcaffeine wash subtracted in order to compare the 3 pulses. (min 17) (min 25) 0 B: extracellular caffeine application (10 mM) 0.3 reduced release, whereas SK current behaved -0.1 as in control experiments. Caffeine addition- (nA)

)An( mI )An( 0 2+ ally reduced peak current in 4 of 6 cells. C: -0.2 intracellular application of 8-BrcADPR (30 http://jn.physiology.org/ ICa ␮M) reduced release, with no effect on the -0.3 -0.3 calcium load. Arrowhead indicates superlin- ear onset. SK current was blocked in this cell -0.6 2 8-Br-cADPR by apamin in the external solution. min 13 min 24 min 32 (min 10) 0.8

)F 1

p( Cm (pF) mC 0.4 0 8-Br-cADPR (min 15) by 10.220.33.6 on September 23, 2016 0 1000 2000 3000 4000 5000 041532 6420 6420 6420 time (s) time (s) time (s) time (s) reduction was significant, the mean release rate change (rate 2001; Proks and Ashcroft 1995; Przywara et al. 1993; 2nd pulse Ϫ rate 1st pulse) was reduced by 40 Ϯ 37% Seward et al. 1996; von Ruden et al. 1993). Moreover, Ba2ϩ (reduction observed in 18 of 23 cells). In controls, the mean is not reuptaken into the store through SERCA pumps release rate change was reduced by 9 Ϯ 44% (reduction (Kwan and Putney 1990; Przywara et al. 1993). To study observed in 4 of 9 cells). A reduction in the rate of release can whether Ba2ϩ alters linear or superlinear capacitance be interpreted as a reduction in vesicle trafficking where changes we replaced external Ca2ϩ with equimolar Ba2ϩ trafficking is Ca2ϩ dependent. (Fig. 7). After superlinear release was evoked with a 3-s With the use of dual-sine capacitance recordings, our phar- depolarization to 50% peak current, extracellular Ca2ϩ was macological data confirm previous reports of a potential role of substituted by equimolar Ba2ϩ and release was tested again intracellular Ca2ϩ stores in hair cell synaptic physiology with the same protocol. As expected, Ba2ϩ abolished SK (Beurg et al. 2005; Evans et al. 2000; Hendricson and Guth activation, increased peak currents, and reduced Ca2ϩ chan- 2002; Kennedy and Meech 2002; Lelli et al. 2003; Marcotti et nel inactivation (Fig. 7A) (Schnee and Ricci 2003). Ba2ϩ al. 2004). As opposed to the release enhancement observed in exerted two different effects on release: reduced total re- controls, hair cell pharmacological treatment led to a reduction lease and delayed onset of the initial linear release compo- in both linear and superlinear release after repeated stimula- nent (Fig. 7A). Box plots of total release for each stimulus tion, with the superlinear release component most clearly obtained with external Ca2ϩ or Ba2ϩ are shown in Fig. 7B. affected. This release reduction was not accompanied by a The response in Ca2ϩ shows release enhancement in the reduction in the peak Ca2ϩ current and therefore cannot be second and third responses, whereas the responses in Ba2ϩ accounted for by a pharmacological effect of intracellular Ca2ϩ were unchanged or reduced, despite the larger currents release on Ca2ϩ channel inactivation (Lee et al. 2007). Alto- observed (487 Ϯ 185 pA in Ba2ϩ vs. 408 Ϯ 173 pA in Ca2ϩ, gether, these results suggest that the recruitment of vesicles for P ϭ 0.01). Ba2ϩ reduced total release in a second stimula- release during prolonged stimulation might be physiologically tion (72% of control) and even further in a third pulse (46% linked to a CICR mechanism. of control). Normalizing release output more clearly shows Calcium triggers linear and superlinear release more the enhancement observed in Ca2ϩ that is absent in the efficiently than barium. Ba2ϩ activates CICR with lower presence of Ba2ϩ (Fig. 7C). The time to initial release was efficacy than Ca2ϩ; it slows down exocytosis and reduces reduced in the presence of Ba2ϩ (Fig. 7D), and the delay the number of vesicles available for release (Neves et al. was further increased for a third stimulation. The delay may

J Neurophysiol • doi:10.1152/jn.00559.2015 • www.jn.org 234 STORED CALCIUM PROMOTES VESICLE RECRUITMENT TO RIBBON SYNAPSES

ABtotal release linear release ** ** ** ** * ** *

)ts1/t

)ts1/ts1-dn2(

s1-d

n

2( Fig. 6. Pharmacological disruption of intracel- lular calcium stores reduces hair cell release. A: caffeine, 8-BrcADPR, ryanodine, and thapsi- gargin significantly reduced release. See dashed line for comparison with control. B and C: linear (B) and superlinear (C) release were reduced by drugs that interfere with endoplasmic reticulum control cADPR control cADPR (ER) RyRs (10 mM caffeine, 30 ␮M caffeine caffeine 8BrcADPRryanodine xestosp C 8BrcADPRryanodine xestosp C 8-BrcADPR, 10 ␮M ryanodine). Reduction was thapsigargin thapsigargin Downloaded from more pronounced in the superlinear component CD of release (note y-axis values in B vs. C). Su- superlinear release superlinear onset perlinear release was also reduced significantly by 0.5–2 ␮M thapsigargin and 1 ␮M xesto- ** ** ** ** * ** spongin C. Cyclic ADP ribose (cADPR, 4 ␮M),

)ts1

an IP3 receptor agonist, had no obvious effects ts1-dn2

on release. D: only ryanodine showed a signif- /ts1- icant effect on superlinear onset. Unpaired t- http://jn.physiology.org/ tests show a significant difference: *P Ͻ 0.05,

dn2( **P Ͻ 0.01.

controlcaffeine cADPR controlcaffeine cADPR 8BrcADPRryanodine xestosp C 8BrcADPRryanodine xestosp C by 10.220.33.6 on September 23, 2016 thapsigargin thapsigargin be a reflection of Ba2ϩ being less effective at driving release DISCUSSION 2ϩ mechanisms (Bhalla et al. 2005). Additionally, Ba appli- The existence of CICR has long been observed in conven- cation promoted a merging of the two components. The tional as well as ribbon synapses (Bouchard et al. 2003; linearization of the capacitance response may reflect a single Castellano-Munoz and Ricci 2014), yet the functional signifi- source of divalent ions driving the process as CICR is cance and cellular mechanisms underlying CICR in these cell 2ϩ disabled with Ba application. types are unclear. Here we studied the potential contribution of The maximum slope of the superlinear release component CICR at hair cell ribbon synapses and obtained the following for a third stimulation was reduced by 36 Ϯ 30% during results. First, spontaneous postsynaptic multiunit activity, ϩ Ba2 application, as opposed to controls, which increased along with EPSC frequency, was reduced by pharmacological by 141 Ϯ 119% (P ϭ 0.01) of the initial maximum slope in manipulations that depleted Ca2ϩ stores through RyRs. Sec- the first pulse (Fig. 7F). Unlike the superlinear decrease in ond, no presynaptic changes in excitability, like a change in release, the first component showed no change in release resting potential or sensitivity, could account for the reduction following the delay. To make this comparison we measured in postsynaptic response. Third, pharmacological presynaptic the first component of release in both Ca2ϩ and Ba2ϩ 500 ms effects on CICR resulted in a reduction of the superlinear after the onset of the capacitance change, thus bypassing the capacitance change that was larger and occurred earlier than delay induced by Ba2ϩ (Fig. 7E). Eliminating the onset changes to the first component of release, consistent with the delays demonstrates that the first release component was not hypothesis that CICR is important for modulating vesicle significantly altered in amplitude by the divalent replace- trafficking. Fourth, Ca2ϩ substitution by Ba2ϩ delayed the ment. The delay in the onset of release could be partially onset of release, likely because of its ability to drive synaptic explained by a poor sensitivity for Ba2ϩ in the presynaptic machinery. It also reduced superlinear release more than the Ca2ϩ sensors for vesicle release and recruitment as sug- first component of release, and this effect was greater for larger gested for SNARE-mediated exocytosis (Bhalla et al. 2005). stimulations. However, the role of SNARE proteins in vesicle fusion at Our data suggest a role for RyRs regulating CICR from the hair cell ribbon synapses remains controversial (Nouvian et ER. Identification of the location of the ER responsible for al. 2011). The intensified effect observed with multiple controlling synaptic vesicle populations remains under inves- stimulation protocols, together with the marked effect of tigation. Although mitochondria are highly localized to the Ba2ϩ on the superlinear release component (Fig. 7F), are synaptic region (Graydon et al. 2011; Schnee et al. 2005), the consistent with the existence of a CICR mechanism govern- role of the mitochondria as a Ca2ϩ sink remains suspect, as our ing the recruitment of new vesicles for release. pharmacological data show a mild effect but one that cannot

J Neurophysiol • doi:10.1152/jn.00559.2015 • www.jn.org STORED CALCIUM PROMOTES VESICLE RECRUITMENT TO RIBBON SYNAPSES 235 A B 0.2 2+ total release (s)

)An( mI )An( 1 Ca 0 2.0 control ** Ba2+

-0.2 2 Ba2+ )F 2+ 3 Ba es p( 1.5 -0.4

aeler latot aeler 1 1.0 )Fp( mC )Fp( 1 2 3 0.5

0 Downloaded from 0246 1st pulse 2nd pulse 3rd pulse Time (s)

C normalized total release D time to release (s) ** ** ** http://jn.physiology.org/ 3rd pulse 2nd pulse (3rd-1st)/1st (2nd-1st)/1st

control Ba2+ control Ba2+ control Ba2+ control Ba2+ by 10.220.33.6 on September 23, 2016

EFinitial release normalized max release slope ** )Fp( esa )Fp( 150 *

100

eler laitini eler (3rd-1st)/1st 50 (2nd-1st)/1st

0 2+ 2+ 1st pulse 2nd pulse 3rd pulse control Ba control Ba Fig. 7. Barium delays release and reduces superlinear release component. A: after substitution of calcium by barium in the extracellular solution, linear release component onset was delayed and maximal release slopes reduced. Distinction between linear and superlinear components was blurred with subsequent depolarization protocols. Baseline was subtracted in order to compare the 3 pulses. B: release values elicited by pulses delivered 5–10 min apart in control conditions and after substitution of calcium by barium in the external solution. C: in controls, release enhancement was magnified by subsequent stimulation (black circles). Conversely, barium substitution reduced release (triangles). D: barium substitution delayed initial release. E: initial release measured 500 ms after release onset was not significantly altered by barium. F: maximal release slope increased with successive stimulations in controls but was reduced by barium substitution. Unpaired t-tests show a significant difference: *P Ͻ 0.05,**P Ͻ 0.01. easily be separated from the energy generation role typically izing to potentials that allow less than maximal calcium cur- ascribed to these organelles. Likely the role of mitochondria at rents from hyperpolarized holding potentials (Schnee et al. hair cell ribbons is similar to the Ca2ϩ buffering role described 2011). In turtle hair cells, prepulses to physiological membrane at ribbon synapses in the visual pathway, in which mitochon- potentials augment release and trigger the convergence of dria provide ATP for Ca2ϩ pumps and also sequester Ca2ϩ linear and superlinear release components (Schnee et al. 2011). from cytoplasm (Babai et al. 2010; Cadetti et al. 2006; Krizaj Similarly, in postnatal rat inner hair cells, depolarizing pulses et al. 2003; Suryanarayanan and Slaughter 2006; Wan et al. preceded by holding the cell at physiological resting potentials 2012; Zenisek and Matthews 2000). produced an increase in exocytosis and synaptic strength and a The clear distinction between linear and superlinear release shortening of synaptic latency at hair cell ribbon synapses components can only be experimentally obtained by depolar- compared with holding the cell at more hyperpolarized poten-

J Neurophysiol • doi:10.1152/jn.00559.2015 • www.jn.org 236 STORED CALCIUM PROMOTES VESICLE RECRUITMENT TO RIBBON SYNAPSES tials (Goutman and Glowatzki 2011). We hypothesize that the lease of Ca2ϩ by CICR has a presynaptic role in neuronal superlinear response represents all synaptic release sites being synaptic transmission (Llano et al. 2000; Unni et al. 2004). filled and fusion happening at maximal rates because of the CICR is observed in hair cells of different animals in both enhanced recruitment of synaptic vesicles. We further hypoth- auditory and vestibular organs (Beurg et al. 2005; Evans et al. esize that changes in release rates for the linear component of 2000; Hendricson and Guth 2002; Kennedy and Meech 2002; release may in part represent release from multiple release sites Lelli et al. 2003; Marcotti et al. 2004; Tucker and Fettiplace where the number of filled sites increases with Ca2ϩ load. This 1995). What could the physiological significance of CICR in hypothesis also can explain the variability in obtaining capac- hair cells be? Hair cells must maintain a tight regulation itance responses that show depletion (Schnee et al. 2005, between Ca2ϩ influx through L-type Ca2ϩ channels and vesicle 2011), because if the calcium entry is low enough not to trigger release at ribbon synapses to allow precise control of release recruitment then the pool size will be set by vesicles present at timing. This regulation is achieved in the vicinity of the ribbon any given moment. by controlling Ca2ϩ levels through buffering and extrusion It remains a question as to whether superlinear capacitance mechanisms that locally rapidly remove Ca2ϩ. Given that the changes reflect synaptic vesicle fusion, extrasynaptic vesicle recruitment of vesicles for release during prolonged stimula- 2ϩ fusion, or even fusion of endosomes (Coggins et al. 2007; tion is Ca dependent, how do hair cells manage to bypass the Downloaded from Zenisek et al. 2000). Our pharmacological approach showed a strong Ca2ϩ regulation near the ribbon to achieve Ca2ϩ- postsynaptic reduction in spike and EPSC rates that correlates dependent vesicle recruitment? One possibility is that the Kd of with a presynaptic release reduction. Whereas hair cell linear trafficking is much lower than that for release. Alternatively, release was mildly reduced by drugs interfering with CICR, CICR may serve to amplify and filter the synaptic signal (Fig. superlinear release was strongly reduced. Similarly, substitu- 8). Thus, during continuous stimulation, the opening of L-type tion of Ca2ϩ by Ba2ϩ exerted a stronger effect on the super- Ca2ϩ channels could tightly modulate the fusion of vesicles linear component of release. These results demonstrate that near the ribbon while amplification of this signal away from the http://jn.physiology.org/ both release components are Ca2ϩ dependent and likely inter- synapse through CICR may trigger vesicle recruitment. These related, suggesting that superlinear release might have a phys- calcium-dependent processes may be additionally modulated iological role in synaptic transmission. Whether the release is by the effects of efferent activity on hair cell synaptic calcium synaptic or extrasynaptic requires better resolution than what levels (Im et al. 2014). We hypothesize that CICR could have we have at the moment (Chen et al. 2014). However, distribu- a functional role in the recruitment and replenishment of tions of glutamate receptors as well as a lack of vesicles at synaptic vesicles to guarantee the availability of vesicles for nonsynaptic release sites in hair cells would also support a release during protracted stimulation. synaptic role (Lenzi et al. 1999; Liberman et al. 2011; Schnee Another puzzling finding is the within-cell variability ob- by 10.220.33.6 on September 23, 2016 et al. 2005). Also, given that the postsynaptic response was a served with repeated stimulations. This was observed in turtle dramatic reduction in firing rate and the major presynaptic with the single-sine technique with paired stimulation (Schnee response was a decrease in the superlinear capacitance re- et al. 2011) and often appears as a facilitation of release with sponse, it is plausible that the superlinear response is a reflec- repeated measures, particularly when IPIs are short. Here we tion of robust vesicle trafficking to the synapse. also identify a slower facilitation that enhances release by In neurons, Ca2ϩ regulates exocytosis of synaptic vesicles as shortening the onset time to superlinear release with repeated well as the supply of new vesicles to release sites (Dittman and stimulations that have longer (minute) IPIs. Together these Regehr 1998; Gomis et al. 1999; Stevens and Wesseling 1998; data question measurements where signal averaging is used or Wang and Kaczmarek 1998). Moreover, the intracellular re- where multiple data points need to be obtained via repetitive A 1 2 Vm

Ca2+ Ca2+ Fig. 8. Ca2ϩ-induced Ca2ϩ release (CICR) might sustain the linear release recruitment of synaptic vesicles to the ribbon. Left: a depo- 2 1 larizing pulse triggering an increase in capacitance in response Ca2+Caa22+ Ca2Ca2+2+ to a calcium current. Right: a hair cell presynaptic terminal Cm and the simultaneous postsynaptic responses (EPSCs). A:at 2+ hyperpolarized membrane potential (Vm), a small calcium influx triggers the fusion of vesicles near the synaptic ribbon. I Ca

Prolonged depolarization leads to vesicle depletion and a EPSCs reduction in EPSC frequency. B: larger calcium influx triggers the fusion of vesicles near the synaptic ribbon as well as CICR B from intracellular stores, allowing vesicle recruitment and an superlinear release 1 2 increase in EPSC frequency. The fusion of new recruited linear release vesicles due to CICR is experimentally observed as a super- Ca2+ Ca2+ linear component of release. Note that under physiological 2 conditions hair cells are maintained at depolarized membrane 1 potentials, a situation in which both linear and superlinear Ca2+Caa22+ Ca2+Ca2C22+2+ components of release might merge.

J Neurophysiol • doi:10.1152/jn.00559.2015 • www.jn.org STORED CALCIUM PROMOTES VESICLE RECRUITMENT TO RIBBON SYNAPSES 237 stimuli. More importantly, though, is the question as to the GRANTS physiological relevance of the variability. The variability ap- This work was funded by National Institute on Deafness and Other pears to be biological, in that no other measured biophysical Communication Disorders Grant DC-009913 to A. J. Ricci and by core grant parameters are changing over this time frame. This enhance- P30 44992. M. Castellano-Muñoz was supported by a Dean’s Postdoctoral ment could simply be due to changes in calcium levels or Fellowship from Stanford School of Medicine and a Cajamadrid Foundation Fellowship. reflect a modulation in the sensitivity triggered by a biochem- ical modification. In our experiments the onset of SK channel activation paralleled superlinear release onset. Both phenom- DISCLOSURES ena are unaffected by CICR pharmacological treatment, thus No conflicts of interest, financial or otherwise, are declared by the author(s). ruling out the possibility that release enhancement is simply due to calcium baseline modulation. Moreover, the calcium AUTHOR CONTRIBUTIONS sensitivity of SK2 and SK3 channels depends on the phosphor- Author contributions: M.C.-M. and A.J.R. conception and design of re- ylation state of SK-bound (Adelman et al. 2012), search; M.C.-M., M.E.S., and A.J.R. performed experiments; M.C.-M. and pointing to a phosphorylation modulation as the origin of the A.J.R. analyzed data; M.C.-M., M.E.S., and A.J.R. interpreted results of release and SK onset shift. experiments; M.C.-M. and A.J.R. prepared figures; M.C.-M. and A.J.R. drafted Downloaded from One possibility consistent with the present data is that the manuscript; M.C.-M., M.E.S., and A.J.R. edited and revised manuscript; Ca2ϩ stores are incompletely filled under our recording con- M.C.-M. and A.J.R. approved final version of manuscript. ditions and initial stimulations serve to fill this pool, which can then be more efficient at driving vesicles to release sites. The REFERENCES stores might be considered in a dynamic equilibrium with Adelman JP, Maylie J, Sah P. Small-conductance Ca2ϩ-activated Kϩ chan- ϩ cytosolic Ca2 where it can act as both sink and source nels: form and function. Annu Rev Physiol 74: 245–269, 2012. Ϫ Albrecht MA, Colegrove SL, Friel DD. Differential regulation of ER Ca2ϩ http://jn.physiology.org/ depending on excitation level. Thus holding a cell at 80 mV 2ϩ 2ϩ 2ϩ uptake and release rates accounts for multiple modes of Ca -induced Ca and dialyzing with ATP and a Ca buffer moves the equilib- release. J Gen Physiol 119: 211–233, 2002. rium toward store depletion so that Ca2ϩ is driven into stores ϩ Alkon DL, Nelson TJ, Zhao W, Cavallaro S. Time domains of neuronal upon stimulation. Thus this Ca2 is available for release upon Ca2ϩ signaling and associative memory: steps through a calexcitin, ryano- further stimulation. Consistent with this idea is the finding that dine receptor, Kϩ channel cascade. Trends Neurosci 21: 529–537, 1998. Alonso MT, Barrero MJ, Michelena P, Carnicero E, Cuchillo I, Garcia using holding potentials more depolarized leads to more robust ϩ ϩ AG, Garcia-Sancho J, Montero M, Alvarez J. Ca2 -induced Ca2 release release (Schnee et al. 2011). in chromaffin cells seen from inside the ER with targeted aequorin. J Cell Another confusing component of these data is the variability Biol 144: 241–254, 1999. in responsiveness to the drug applications, even those where Andersson SA, Pedersen MG, Vikman J, Eliasson L. Glucose-dependent by 10.220.33.6 on September 23, 2016 drugs were included within the patch pipette. Not all cells docking and SNARE -mediated exocytosis in mouse pancreatic alpha-cell. Pflügers Arch 462: 443–454, 2011. responded, and those that did showed a larger than expected Babai N, Morgans CW, Thoreson WB. Calcium-induced calcium release variance. Again, the variation appears to be biologically driven contributes to synaptic release from mouse rod photoreceptors. Neurosci- and not a function of the biophysical status of the hair cells. ence 165: 1447–1456, 2010. One possibility relates to the previous discussion that drug Bardo S, Cavazzini MG, Emptage N. The role of the endoplasmic reticulum 2ϩ efficacy will be directly determined by the state of the store Ca store in the plasticity of central neurons. Trends Pharmacol Sci 27: 78–84, 2006. (how filled it is) at the time of drug administration. Another Benech JC, Crispino M, Kaplan BB, Giuditta A. Protein synthesis in 2ϩ confounding point is that turtle hair cells have large Ca presynaptic endings from squid brain: modulation by calcium ions. J currents and it is likely that large depolarizations that lead to Neurosci Res 55: 776–781, 1999. major influx via Ca2ϩ channels can mask the effects of Ca2ϩ Beurg M, Hafidi A, Skinner LJ, Ruel J, Nouvian R, Henaff M, Puel JL, Aran JM, Dulon D. Ryanodine receptors and BK channels act as a stores. Alternatively, there may be additional biochemical presynaptic depressor of neurotransmission in cochlear inner hair cells. Eur modifications that we are not controlling, for example, CaM J Neurosci 22: 1109–1119, 2005. activity. Bhalla A, Tucker WC, Chapman ER. isoforms couple In summary, the data presented identify a presynaptic role distinct ranges of Ca2ϩ,Ba2ϩ, and Sr2ϩ concentration to SNARE-mediated for CICR to modulate hair cell vesicle release. Both pre- and membrane fusion. Mol Biol Cell 16: 4755–4764, 2005. Bouchard R, Pattarini R, Geiger JD. Presence and functional significance of postsynaptic recordings support the argument that CICR may presynaptic ryanodine receptors. Prog Neurobiol 69: 391–418, 2003. regulate vesicle trafficking. Data demonstrate the involvement Cadetti L, Bryson EJ, Ciccone CA, Rabl K, Thoreson WB. Calcium- of RyRs in this pathway. Capacitance measurements target the induced calcium release in rod photoreceptor terminals boosts synaptic major effect to a reduction in the superlinear component of transmission during maintained depolarization. Eur J Neurosci 23: 2983– 2990, 2006. release, previously argued to require vesicle recruitment to the Castellano-Munoz M, Ricci AJ. Role of intracellular calcium stores in synapses. These data indirectly support the contention that both hair-cell ribbon synapse. Front Cell Neurosci 8: 162, 2014. linear and superlinear capacitance changes might be synaptic in Chen M, Krizaj D, Thoreson WB. Intracellular calcium stores drive slow nature. Furthermore, a novel hypothesis is put forth suggesting non-ribbon vesicle release from rod photoreceptors. Front Cell Neurosci 8: that CICR is needed to amplify and filter the tightly controlled 20, 2014. 2ϩ Cho S, Li GL, von Gersdorff H. Recovery from short-term depression and synaptic Ca signal to ensure continuous and reliable vesicle facilitation is ultrafast and Ca2ϩ dependent at auditory hair cell synapses. J trafficking required to maintain high and sustainable afferent Neurosci 31: 5682–5692, 2011. firing rates. Coggins MR, Grabner CP, Almers W, Zenisek D. Stimulated exocytosis of endosomes in goldfish retinal bipolar neurons. J Physiol 584: 853–865, ACKNOWLEDGMENTS 2007. Dallos P, He DZ, Lin X, Sziklai I, Mehta S, Evans BN. Acetylcholine, outer We thank the members of our research group for providing helpful com- hair cell electromotility, and the cochlear amplifier. J Neurosci 17: 2212– ments on the project and the manuscript. 2226, 1997.

J Neurophysiol • doi:10.1152/jn.00559.2015 • www.jn.org 238 STORED CALCIUM PROMOTES VESICLE RECRUITMENT TO RIBBON SYNAPSES

Dittman JS, Regehr WG. Calcium dependence and recovery kinetics of Liberman LD, Wang H, Liberman MC. Opposing gradients of ribbon size presynaptic depression at the climbing fiber to Purkinje cell synapse. J and AMPA receptor expression underlie sensitivity differences among Neurosci 18: 6147–6162, 1998. cochlear-nerve/hair-cell synapses. J Neurosci 31: 801–808, 2011. Duncker SV, Franz C, Kuhn S, Schulte U, Campanelli D, Brandt N, Hirt Lioudyno M, Hiel H, Kong JH, Katz E, Waldman E, Parameshwaran-Iyer B, Fakler B, Blin N, Ruth P, Engel J, Marcotti W, Zimmermann U, S, Glowatzki E, Fuchs PA. A “synaptoplasmic cistern” mediates rapid Knipper M. Otoferlin couples to clathrin-mediated endocytosis in mature inhibition of cochlear hair cells. J Neurosci 24: 11160–11164, 2004. cochlear inner hair cells. J Neurosci 33: 9508–9519, 2013. Llano I, Gonzalez J, Caputo C, Lai FA, Blayney LM, Tan YP, Marty A. Emptage NJ, Reid CA, Fine A. Calcium stores in hippocampal synaptic Presynaptic calcium stores underlie large-amplitude miniature IPSCs and boutons mediate short-term plasticity, store-operated Ca2ϩ entry, and spon- spontaneous calcium transients. Nat Neurosci 3: 1256–1265, 2000. taneous transmitter release. Neuron 29: 197–208, 2001. Lukyanenko V, Viatchenko-Karpinski S, Smirnov A, Wiesner TF, Gyorke Evans MG, Lagostena L, Darbon P, Mammano F. Cholinergic control of S. Dynamic regulation of sarcoplasmic reticulum Ca2ϩ content and release membrane conductance and intracellular free Ca2ϩ in outer hair cells of the by luminal Ca2ϩ-sensitive leak in rat ventricular myocytes. Biophys J 81: guinea pig cochlea. Cell Calcium 28: 195–203, 2000. 785–798, 2001. Fitzjohn SM, Collingridge GL. Calcium stores and synaptic plasticity. Cell Magistretti J, Spaiardi P, Johnson SL, Masetto S. Elementary properties of Calcium 32: 405–411, 2002. Ca2ϩ channels and their influence on multivesicular release and phase- Fuchs PA. A “calcium capacitor” shapes cholinergic inhibition of cochlear locking at auditory hair cell ribbon synapses. Front Cell Neurosci 9: 123, hair cells. J Physiol 592: 3393–3401, 2014. 2015. Gomis A, Burrone J, Lagnado L. Two actions of calcium regulate the supply Mammano F, Frolenkov GI, Lagostena L, Belyantseva IA, Kurc M, Downloaded from of releasable vesicles at the ribbon synapse of retinal bipolar cells. J Dodane V, Colavita A, Kachar B. ATP-induced Ca2ϩ release in cochlear Neurosci 19: 6309–6317, 1999. outer hair cells: localization of an inositol triphosphate-gated Ca2ϩ store to Goutman JD, Glowatzki E. Short-term facilitation modulates size and timing the base of the sensory hair bundle. J Neurosci 19: 6918–6929, 1999. of the synaptic response at the inner hair cell ribbon synapse. J Neurosci 31: Marcotti W, Johnson SL, Kros CJ. Effects of intracellular stores and 7974–7981, 2011. extracellular Ca2ϩ on Ca2ϩ-activated Kϩ currents in mature mouse inner Grant L, Fuchs P. Calcium- and calmodulin-dependent inactivation of cal- hair cells. J Physiol 557: 613–633, 2004. cium channels in inner hair cells of the rat cochlea. J Neurophysiol 99: Moser T, Beutner D. Kinetics of exocytosis and endocytosis at the cochlear 2183–2193, 2008. inner hair cell afferent synapse of the mouse. Proc Natl Acad Sci USA 97: http://jn.physiology.org/ Graydon CW, Cho S, Li GL, Kachar B, von Gersdorff H. Sharp Ca2ϩ 883–888, 2000. nanodomains beneath the ribbon promote highly synchronous multivesicular Neves G, Neef A, Lagnado L. The actions of barium and strontium on release at hair cell synapses. J Neurosci 31: 16637–16650, 2011. exocytosis and endocytosis in the synaptic terminal of goldfish bipolar cells. Hendricson AW, Guth PS. Transmitter release from Rana pipiens vestibular J Physiol 535: 809–824, 2001. hair cells via mGluRs: a role for intracellular Caϩϩ release. Hear Res 172: Nicholls DG. Mitochondrial calcium function and dysfunction in the central 99–109, 2002. nervous system. Biochim Biophys Acta 1787: 1416–1424, 2009. Hongpaisan J, Pivovarova NB, Colegrove SL, Leapman RD, Friel DD, Nouvian R, Neef J, Bulankina AV, Reisinger E, Pangrsic T, Frank T, Andrews SB. Multiple modes of calcium-induced calcium release in sym- Sikorra S, Brose N, Binz T, Moser T. Exocytosis at the hair cell ribbon 2ϩ pathetic neurons II: a [Ca ]i- and location-dependent transition from synapse apparently operates without neuronal SNARE proteins. Nat Neu-

endoplasmic reticulum Ca accumulation to net Ca release. J Gen Physiol rosci 14: 411–413, 2011. by 10.220.33.6 on September 23, 2016 118: 101–112, 2001. Patel SH, Salvi JD, Ó Maoiléidigh D, Hudspeth AJ. Frequency-selective Im GJ, Moskowitz HS, Lehar M, Hiel H, Fuchs PA. Synaptic calcium exocytosis by ribbon synapses of hair cells in the bullfrog’s amphibian regulation in hair cells of the chicken basilar papilla. J Neurosci 34: papilla. J Neurosci 32: 13433–13438, 2012. 16688–16697, 2014. Pozzo-Miller LD, Pivovarova NB, Leapman RD, Buchanan RA, Reese TS, Innocenti B, Heidelberger R. Mechanisms contributing to tonic release at the Andrews SB. Activity-dependent calcium sequestration in dendrites of cone photoreceptor ribbon synapse. J Neurophysiol 99: 25–36, 2008. hippocampal neurons in brain slices. J Neurosci 17: 8729–8738, 1997. Iosub R, Avitabile D, Grant L, Tsaneva-Atanasova K, Kennedy HJ. Proks P, Ashcroft FM. Effects of divalent cations on exocytosis and endo- Calcium-induced calcium release during action potential firing in develop- cytosis from single mouse pancreatic beta-cells. J Physiol 487: 465–477, ing inner hair cells. Biophys J 108: 1003–1012, 2015. 1995. Issa NP, Hudspeth AJ. Clustering of Ca2ϩ channels and Ca2ϩ-activated Kϩ Przywara DA, Chowdhury PS, Bhave SV, Wakade TD, Wakade AR. channels at fluorescently labeled presynaptic active zones of hair cells. Proc Barium-induced exocytosis is due to internal calcium release and block of Natl Acad Sci USA 91: 7578–7582, 1994. calcium efflux. Proc Natl Acad Sci USA 90: 557–561, 1993. Johnson SL, Forge A, Knipper M, Munkner S, Marcotti W. Tonotopic Quinones PM, Luu C, Schweizer FE, Narins PM. Exocytosis in the frog variation in the calcium dependence of neurotransmitter release and vesicle amphibian papilla. J Assoc Res Otolaryngol 13: 39–54, 2012. pool replenishment at mammalian auditory ribbon synapses. J Neurosci 28: Roberts WM, Jacobs RA, Hudspeth AJ. Colocalization of ion channels 7670–7678, 2008. involved in frequency selectivity and synaptic transmission at presynaptic Kennedy HJ, Meech RW. Fast Ca2ϩ signals at mouse inner hair cell synapse: active zones of hair cells. J Neurosci 10: 3664–3684, 1990. a role for Ca2ϩ-induced Ca2ϩ release. J Physiol 539: 15–23, 2002. Rossi ML, Prigioni I, Gioglio L, Rubbini G, Russo G, Martini M, Farinelli

Krizaj D, Lai FA, Copenhagen DR. Ryanodine stores and calcium regulation F, Rispoli G, Fesce R. IP3 receptor in the hair cells of frog semicircular in the inner segments of salamander rods and cones. J Physiol 547: 761–774, canal and its possible functional role. Eur J Neurosci 23: 1775–1783, 2006. 2003. Rutherford MA, Roberts WM. Frequency selectivity of synaptic exocytosis Kwan CY, Putney JW Jr. Uptake and intracellular sequestration of divalent in frog saccular hair cells. Proc Natl Acad Sci USA 103: 2898–2903, 2006. cations in resting and methacholine-stimulated mouse lacrimal acinar cells. Schnee ME, Castellano-Munoz M, Ricci AJ. Response properties from turtle Dissociation by Sr2ϩ and Ba2ϩ of agonist-stimulated divalent cation entry auditory hair cell afferent fibers suggest spike generation is driven by from the refilling of the agonist-sensitive intracellular pool. J Biol Chem synchronized release both between and within synapses. J Neurophysiol 265: 678–684, 1990. 110: 204–220, 2013. Lee S, Briklin O, Hiel H, Fuchs P. Calcium-dependent inactivation of Schnee ME, Lawton DM, Furness DN, Benke TA, Ricci AJ. Auditory hair calcium channels in cochlear hair cells of the chicken. J Physiol 583: cell-afferent fiber synapses are specialized to operate at their best frequen- 909–922, 2007. cies. Neuron 47: 243–254, 2005. Lelli A, Perin P, Martini M, Ciubotaru CD, Prigioni I, Valli P, Rossi ML, Schnee ME, Ricci AJ. Biophysical and pharmacological characterization of Mammano F. Presynaptic calcium stores modulate afferent release in voltage-gated calcium currents in turtle auditory hair cells. J Physiol 549: vestibular hair cells. J Neurosci 23: 6894–6903, 2003. 697–717, 2003. Lenzi D, Runyeon JW, Crum J, Ellisman MH, Roberts WM. Synaptic Schnee ME, Santos-Sacchi J, Castellano-Munoz M, Kong JH, Ricci AJ. vesicle populations in saccular hair cells reconstructed by electron tomog- Calcium-dependent synaptic vesicle trafficking underlies indefatigable re- raphy. J Neurosci 19: 119–132, 1999. lease at the hair cell afferent fiber synapse. Neuron 70: 326–338, 2011. Levic S, Bouleau Y, Dulon D. Developmental acquisition of a rapid calcium- Seward EP, Chernevskaya NI, Nowycky MC. Ba2ϩ ions evoke two kinet- regulated vesicle supply allows sustained high rates of exocytosis in audi- ically distinct patterns of exocytosis in chromaffin cells, but not in neuro- tory hair cells. PLoS One 6: e25714, 2011. hypophysial nerve terminals. J Neurosci 16: 1370–1379, 1996.

J Neurophysiol • doi:10.1152/jn.00559.2015 • www.jn.org STORED CALCIUM PROMOTES VESICLE RECRUITMENT TO RIBBON SYNAPSES 239

Sneary MG. Auditory receptor of the red-eared turtle: II. Afferent and efferent long-term depression at hippocampal CA3-CA3 pyramidal neuron synapses. synapses and innervation patterns. J Comp Neurol 276: 588–606, 1988. J Neurosci 24: 9612–9622, 2004. Stevens CF, Wesseling JF. Activity-dependent modulation of the rate at Verkhratsky A. Physiology and pathophysiology of the calcium store in the which synaptic vesicles become available to undergo exocytosis. Neuron 21: endoplasmic reticulum of neurons. Physiol Rev 85: 201–279, 2005. 415–424, 1998. von Ruden L, Garcia AG, Lopez MG. The mechanism of Ba2ϩ-induced Suryanarayanan A, Slaughter MM. Synaptic transmission mediated by exocytosis from single chromaffin cells. FEBS Lett 336: 48–52, 1993. internal calcium stores in rod photoreceptors. J Neurosci 26: 1759–1766, Wan QF, Nixon E, Heidelberger R. Regulation of presynaptic calcium in a 2006. mammalian synaptic terminal. J Neurophysiol 108: 3059–3067, 2012. Sutton KG, McRory JE, Guthrie H, Murphy TH, Snutch TP. P/Q-type Wang LY, Kaczmarek LK. High-frequency firing helps replenish the readily calcium channels mediate the activity-dependent feedback of syntaxin-1A. releasable pool of synaptic vesicles. Nature 394: 384–388, 1998. Nature 401: 800–804, 1999. Zenisek D, Matthews G. The role of mitochondria in presynaptic calcium Tang Y, Zucker RS. Mitochondrial involvement in post-tetanic potentiation handling at a ribbon synapse. Neuron 25: 229–237, 2000. of synaptic transmission. Neuron 18: 483–491, 1997. Zenisek D, Steyer JA, Almers W. Transport, capture and exocytosis of single Tucker T, Fettiplace R. Confocal imaging of calcium microdomains and synaptic vesicles at active zones. Nature 406: 849–854, 2000. calcium extrusion in turtle hair cells. Neuron 15: 1323–1335, 1995. Zucchi R, Ronca-Testoni S. The sarcoplasmic reticulum Ca2ϩ channel/ Unni VK, Zakharenko SS, Zablow L, DeCostanzo AJ, Siegelbaum SA. : modulation by endogenous effectors, drugs and disease Calcium release from presynaptic ryanodine-sensitive stores is required for states. Pharmacol Rev 49: 1–51, 1997. Downloaded from http://jn.physiology.org/ by 10.220.33.6 on September 23, 2016

J Neurophysiol • doi:10.1152/jn.00559.2015 • www.jn.org