To encode continuous sound stimuli, the inner hair cell (IHC) ribbon synapses utilize calcium-binding proteins (CaBPs), which reduce the inactivation of their CaV1.3 calcium channels. Mutations in the CABP2 gene underlie non-syndromic autosomal recessive hearing loss DFNB93. Besides CaBP2, the structurally related CaBP1 is highly abundant in the IHCs. Here, we investigated how the two CaBPs cooperatively regulate IHC synaptic function. In Cabp1/2 double-knockout mice, we find strongly enhanced CaV1.3 inactivation, slowed recovery from inactivation and impaired sustained exocytosis. Already mild IHC activation further reduces the availability of channels to trigger synaptic transmission and may effectively silence synapses. Spontaneous and sound-evoked responses of spiral ganglion neurons in vivo are strikingly reduced and strongly depend on stimulation rates. Transgenic expression of CaBP2 leads to substantial recovery of IHC synaptic function and hearing sensitivity. We conclude that CaBP1 and 2 act together to suppress voltage- and calcium-dependent inactivation of IHC CaV1.3 channels in order to support sufficient rate of exocytosis and enable fast, temporally precise and indefatigable sound encoding.
This fundamental study reveals the major role of calcium-binding proteins (CaBP1 and CaBP2) in sustained exocytosis from mouse inner hair cell ribbon synapses. Compelling data and analysis from CaBP1/2 double-knockout mice show enhanced calcium channel (CaV1.3) inactivation, slowed recovery from inactivation, and reduced synaptic vesicle exocytosis as assayed by membrane capacitance measurements, as well as greatly reduced in vivo spontaneous and sound-evoked spikes from the postsynaptic spiral ganglion neurons. Importantly, transgenic expression of CaBP2 led to the rescue of hearing capabilities. The continuous transmission of sound-evoked signals from auditory hair cells thus depends on the expression of both CaBP1 and CaBP2 and their suppression of CaV1.3 inactivation.
Calcium influx through voltage-gated calcium channels (CaVs) is a pivotal initiator of neurotransmission. These channels typically undergo considerable Ca2+- and voltage-dependent inactivation (CDI and VDI), which limits the amount of incoming Ca2+ ions and thus possible excitotoxicity. In sensory receptor cells such as sensory inner hair cells and photoreceptors, however, the extent of inactivation is significantly reduced to allow encoding of graded and sustained membrane potential fluctuations in response to ongoing sound or visual (t.i. “darkness”) stimuli [reviewed in 1]. The attenuation of channel inactivation may be largely attributed to the action of calcium-binding proteins, CaBPs 2, which primarily abate the calmodulin (CaM)-mediated CDI of CaV channels 2–7. At the inner hair cell (IHC) ribbon synapses, the Ca2+ current is carried through the L-type CaV1.3 channels 8,9. Their biophysical properties may be regulated by several CaBP family members 10,11, but most prominently by CaBP2 12. Mutations in the CaBP2 gene cause moderate to severe autosomal recessive hearing impairment DFNB93 [see e.g. 12–16]. Interestingly, the loss of CaBP2 seems to significantly affect VDI, but not CDI of the IHC Ca2+ currents 12. Furthermore, isolated disruption of CaBP1 or CaBP4 in mice leads to at most modest enhancement of IHC CaV1.3 inactivation 10,11. Moreover, the automodulatory C-terminal domain of the long CaV1.3 isoform, abundant at the IHC ribbon synapses 17,18, does not seem critical for CDI suppression either, as its disruption does not lead to hearing impairment 18. The question of which mechanisms are primarily responsible for suppressing IHC CaV1.3 CDI thus remains unresolved. Since CaBP2, followed by the structurally similar CaBP1, are the most abundant members of the CaBP family in the cochlea 19, we set out to investigate the phenotype of Cabp1/2 double-knockout (DKO) animals at the systems and cellular levels. Severe hearing impairment that we observed in these animals reveals that CaBP1 and 2 act together to attenuate the inactivation of the IHC CaV1.3 channels. Furthermore, we demonstrate that the combined absence of CaBP1 and 2 almost abolishes sustained, and, upon increased hair cell synaptic activity, the fast component of exocytosis. Such phenotype was observed neither in Cabp1 nor in Cabp2 single KO animals 11,12. To obtain a more comprehensive understanding of the roles of the two calcium-binding proteins, we attempted to rescue the auditory function of the CaBP1/2-depleted auditory system by viral delivery of a Cabp2 coding sequence in the inner ear. These experiments revealed a substantial recovery of hearing, which may be related to the successful restoration of Ca2+ currents and IHC exocytosis upon transgenic expression of CaBP2. Our findings indicate that the functions of CaBP1 and 2 partially overlap at the IHC ribbon synapse. Together, they provide sufficient attenuation of channel inactivation to sustain Ca2+ currents and exocytosis required for temporally precise, indefatigable neurotransmission at this synapse.
Pronounced VDI and CDI of CaV1.3 channels in CaBP1/2-deficient IHCs
CaBP1 and 2 show high expression levels in the organ of Corti 19, where they may modulate voltage-gated CaV1.3 channels at the IHC ribbon synapses (Fig. 1A). Deletion of CaBP2 results in an early-onset, progressive hearing impairment in the KO mouse model and moderate to severe hearing impairment, DFNB93, in affected human patients 12–14. Loss of CaBP1 in mice, on the other hand, leaves hearing unperturbed in the first few months after birth 11. Since these proteins show a high degree of homology 2, it is plausible that they possess similar features and are partially redundant, at least initially. Therefore, we were interested in understanding their cumulative impact on the function of hair cells and hearing. For this, we analyzed the phenotype of the Cabp1/2-DKO animals and tested the extent of functional rescue following the transgenic expression of CaBP2 (Fig. S1A) in the ears of the DKO mice.
To investigate possible synergistic functions of the two major IHC CaBPs, we first measured Ca2+ currents in the IHCs of 3-4-week-old Cabp1/2-DKO and WT animals in the presence of 1.3 mM extracellular [Ca2+]. Employing 500 ms-long depolarization steps to maximal Ca2+-current potential (-25 to -15 mV) revealed pronounced inactivation of CaV1.3 channels in the DKO IHCs (Fig. 1B). This was significantly stronger when compared to the inactivation observed in the IHCs of Cabp1 or Cabp2 single KOs in similar conditions (Fig. 1C) 11,12,20, and even when comparing to data from 5-week-old Cabp2 KO mice obtained at physiological temperature (Fig. S1D), further substantiating the evidence of severity and early onset of the DKO phenotype.
The “non-inactivating” nature of the IHC CaV1.3 channels was partially restored in the Cabp1/2-DKOs by transgenic expression of WT CaBP2 (Student’s t-test: p < 0.0001; Fig. 1B-C; see also Fig. S1). The lack of complete recovery suggests either an inadequate expression of transgenic material or irreversible adaptive changes in response to the lack of major CaBPs that may occur prior to gene delivery. Similarly, Ca2+ currents were greatly reduced upon application of short train stimuli and rescued efficiently by the introduction of the transgenic CaBP2 (Fig. 1D). However, it has to be noted that the cells in which train stimuli were applied showed a considerably lower calcium current inactivation upon a 500-ms step depolarization than the average (0.36 (n=5) vs 0.47 (n=21)). Short stimuli at different potentials revealed maximal Ca2+-current amplitudes in the IHCs of DKO mice that were comparable to controls (Fig. 1E). We observed a slight shift in the voltage-dependence of CaV1.3 activation towards more hyperpolarized values (Fig. 1F; potential of half-activation – V0.5 of -33.6 ± 0.5 vs -30.9 ± 0.5 mV in Cabp1/2 DKO and WT IHCs, respectively; p < 0.0001, Student’s t-test), and an increased slope of the Boltzmann fit function (7.6 ± 0.1 vs 7.1 ± 0.1, p < 0.0001, Student’s t-test). This finding is somewhat surprising as the opposite would be expected based on data from heterologous expression systems 4,12,21. At present we can not exclude potential changes in the relative amounts of different CaV1.3 splice variants, auxiliary subunits or other interaction partners that may in the absence of CaBPs lead to these changes in the IHCs.
To further investigate the nature of the enhanced CaV1.3 inactivation, we recorded CaV1.3 currents in the presence of 2 mM Ca2+ or Ba2+ in the extracellular solution (Fig. 2), and analyzed the two types of inactivation as described previously 22. The lack of CaBP1 and 2 significantly enhanced both, VDI and CDI of IHC CaV1.3 channels, which were approx. two- and six-times larger than those in WT IHCs (Fig. 2B; VDI: 0.30 ± 0.02 vs 0.13 ± 0.02 and CDI: 0.65 ± 0.01 vs 0.11 ± 0.02 in Cabp1/2 DKO and WT IHCs, respectively; p < 0.0001, Student’s t-test).
Impaired synaptic transmission in the IHCs lacking CaBP1 and 2
Since Ca2+ influx through CaV1.3 channels triggers IHC neurotransmission we next investigated how the prominent inactivation affects exocytosis in the CaBP1/2-deficient IHCs. Exocytosis was probed by recordings of membrane capacitance (Fig. 3). In response to short stimuli [primarily recruiting the readily-releasable pool (RRP) of vesicles], increases in capacitance were comparable, while long stimuli (reporting sustained exocytosis) evoked significantly smaller responses in the CaBP1/2-deficient IHCs (Fig. 3A-B). These results are in line with the observation of comparable peak Ca2+-current amplitudes but reduced calcium charge transfer (QCa) for stimulus durations above 50 ms. However, calculating exocytosis efficiency revealed that the reduction in sustained exocytosis cannot completely be explained by the diminished Ca2+ influx (Fig. 3C). This suggests an additional impairment in exo- or endocytic processes, the underlying mechanism of which is currently unknown. Transgenic expression of CaBP2 largely restored the efficiency of exocytosis, indicating that the impairment as observed in the IHCs of DKOs is related to the absence of the CaBPs. Such exocytosis defect was not observed in Cabp2-KOs, suggesting that when IHCs are held at hyperpolarized potential at RT, CaBP1 provides sufficient compensation to obscure the synaptic phenotype12,20.
Next, we have reasoned that holding a cell at a relatively hyperpolarized potential of -85 mV as typically utilized in patch-clamp recordings may not reveal the synaptic phenotype in its severity. We thus performed additional sets of experiments (Fig. 3D-E), in which the cells were held at either -65 or -55 mV, closer to the suggested resting potential of IHCs 23–25. Raising the holding voltage to -55 mV resulted in a significant reduction in the maximal Ca2+ currents, which was more pronounced in the CaBP1/2-deficient IHCs as compared to controls (60 ± 4 % vs 25 ± 4 % reduction in DKO and WT IHCs, respectively; p < 0.005, Student’s t-test). Consequently, further discrepancy between the QCa in IHCs of WT and DKO animals was observed (p < 0.001), which in DKO animals resulted in a complete loss of sustained exocytosis (Fig. 3D; p < 0.005). The latter was well restored upon AAV-mediated transgenic expression of CaBP2 (p < 0.05), which also reduced the amount of calcium channel inactivation (38 ± 8 % reduction in the maximal Ca2+ current as well as a partial recovery of QCa as compared to Cabp1/2-DKO IHCs; p < 0.05). Rapid voltage oscillations during capacitance recordings may additionally activate calcium channels prior to the depolarizing step. We thus reduced the sine wave amplitude, which resulted in an intermediate phenotype, indicating pronounced activity-dependent deterioration of synaptic function in the absence of CaBPs (Fig. 3E). We propose that in vivo in the absence of CaBP1 and 2 a large fraction of IHC CaV1.3 channels are in a state of prolonged inactivation. This does not support sufficient RRP exocytosis (Fig. 3D) to sustain sensory coding during ongoing stimulation of IHCs.
Lastly, we asked whether the deletion of CaBP1 and 2 affects the speed of recovery from CaV1.3 inactivation (Fig. 4). We applied an initial 5-ms long pulse followed by a conditioning stimulus and a series of 5-ms long pulses to test the kinetics of recovery of the Ca2+-current amplitude. Upon 500-ms depolarization step (Fig. 4B), the speed of recovery from inactivation was fast both in DKO as well as WT IHCs where a much smaller fraction of channels underwent inactivation (time constants of 0.62 and 0.75 s in DKO and WT IHCs, respectively; Table 1, left). As expected, transgenic expression of CaBP2 reduced the initial inactivation (Ca2+-current amplitude 10 ms after the conditioning pulse, expressed as the fraction of initial current – I10ms_post/Ipre, WT: 0.73 ± 0.01; DKO: 0.36 ± 0.03; DKO+PHP.eB-Cabp2: 0.64 ± 0.02; p < 0.001 for DKO vs DKO+PHP.eB-Cabp2). However, the recovery from the step-evoked inactivation was slowest in the Cabp2-injected DKO IHCs. We hypothesize that this may be related to the speed at which the intracellular [Ca2+] after depolarization is restored to the basal level allowing CaV1.3 recovery from CDI (see discussion). A 2-min long sine wave (± 15 mV) from -55 mV to mimic a milder but prolonged hair cell stimulation (Fig. 4C) elicited smaller inactivation of CaV1.3 channels as compared to a 500-ms depolarization step (I10ms_post/Ipre of 0.73 ± 0.01 in WT; 0.58 ± 0.03 in DKO; 0.68 ± 0.03 in DKO+PHP.eB-Cabp2; p < 0.05 for DKO vs DKO+PHP.eB-Cabp2). In these conditions, however, the recovery from inactivation was slowest in the DKO IHCs (Table 1, right). This suggests that in the absence of CaBP1 and 2 even mild but prolonged stimulation of IHCs in vivo may drive a significant fraction of CaV1.3 channels into “deep-state” of inactivation with very slow recovery, as was described for NaV, KV and CaV channels 26–29, not allowing efficient sound encoding.
Altered synaptic morphology in Cabp1/2-DKO mice
Immunolabeling of the organs of Corti indicated normal presynaptic CaV1.3 clustering with no apparent change in size or shape of the clusters (Fig. S2A), and comparable abundance of ribbon synapses in the apical IHCs of WT and DKO animals (12.9 ± 0.2 vs 12.8 ± 0.8; respectively) at 3 weeks of age (Fig. S2B-C’). However, the intensities of pre-(RibeyeA/CtBP2) and postsynaptic (Homer1) immunofluorescent puncta were increased in the CaBP1/2-deficient IHCs as compared to age-matched controls, processed in parallel (Fig. S2C’’). A similar alteration in the synaptic size was observed in some other synaptic mutants as well as upon noise trauma 30,31, but not in Cabp2 single KOs 12, which show significantly less impaired presynaptic Ca2+ signalling. This data is in line with compensatory postsynaptic changes in conditions of no or very little postsynaptic activity, possibly governed through adaptive changes in the postsynaptic elements.
As earlier studies of CaV1.3 channelopathies demonstrated a lag or stop in IHC development 30,32,33 we next examined the expression of small-(SK) and large-conductance Ca2+-activated potassium channels (BK) in the IHCs at different ages. As expected for normal maturation, the SK2 puncta were mostly absent from DKO IHCs by 3 weeks of age and the BK puncta underwent a typical transition from a spotty appearance throughout the cell membrane to the patches at the IHC necks (Fig. S2D-G). However, by 3 weeks and more obviously later on, the BK channel clusters appeared smaller in the DKO IHCs as compared to controls, which could be partially reversed by virally-mediated expression of CaBP2 (Fig. S2F-G). We hypothesize that CaBPs may directly (e.g. as transcription factors) regulate BK expression, a notion that needs to be tested in the future. Alternatively, the effects may be more indirect, as they are common in mutants with impaired intracellular Ca2+ signaling 30,32,33.
Severe hearing impairment in mice deficient in CaBP1 and CaBP2
Hearing was assessed by measuring auditory brainstem responses (ABRs; Fig. 5, Fig. S3). Already in young (3-4-week-old) Cabp1/2-DKO animals, ABR thresholds were highly elevated, with tone bursts up to a maximum intensity of 90 dB barely eliciting any ABR response; click responses were better preserved but greatly reduced in amplitude (Fig. S3A-B). At 7-13 weeks of age, further worsening of the click thresholds was observed. Upon 80-dB click stimulation no sizable ABR wave I was detected in average traces of DKO animals at either age tested (Fig. S3B), suggesting reduced and/or desynchronized spiking of spiral ganglion neurons (SGNs). Like in other mouse mutants with defects in the IHC synapse 34,35, the later ABR peaks were better preserved, indicating partially preserved activation of neurons of the auditory brainstem. To assess cochlear amplification by the outer hair cells (OHCs), we recorded distortion product otoacoustic emissions (DPOAEs). While OAEs were so far only detectable in some of the younger DFNB93 patients 12,36, mouse work corroborates the notion of at least initially intact cochlear amplification in the absence of CaBP2 11,12,20. Our results suggest that cochlear amplification is largely preserved also in DKO animals (Fig. S3C-F). In about half of the DKO animals in both age groups, the DPOAEs were normal. In the other half, they were reduced in amplitude but still detectable (Fig. S3E-F), which may suggest partially hampered cochlear amplification in a subset of DKO animals.
Transgenic expression of CaBP2 led to significant improvement of the IHC CaV1.3 function and synaptic exocytosis (Figs. 1-4), we thus next asked, how this may influence the hearing of the animals. As previously observed 20, unilateral injection of the PHP.eB-Cabp2-eGFP resulted in strong eGFP immunofluorescence signal in the vast majority of hair cells and elsewhere (Fig. S4A-B). Transgenic expression of CaBP2 in DKO animals led to significantly improved hearing with lower ABR thresholds, increased click-evoked ABR amplitudes, and shortened wave I latency (Fig. 5, Fig. S4C-D).
Finally, we studied sound encoding by inspecting sound-induced spiking responses of SGNs (Fig. 6). In WT mice, we recorded responses in 39 neurons, which were classified as SGNs based on their location, response patterns, spike waveforms and recording stability. Their spiking characteristics resembled typical WT data 12,37–40. In response to 50- or 500-ms tone burst at the characteristic frequency 30 dB above threshold (i.e. where the sigmoidal rate-intensity functions have saturated spike rates) and at a stimulus repetition rate of 5 Hz, they showed the SGN-typical very high onset firing rates, gradually adapting to reach steady-state rates between 100 and 400 Hz (Fig. 6A-B). In contrast, only one out of the 20 neurons we recorded and analyzed from DKO animals showed a normal response in our automated tuning curve algorithm with a threshold of 35 dB at 7.8 kHz, while all others displayed thresholds above 80 dB and extremely low spike rates (Fig. 6A-B). Since in the majority of neurons frequency tuning was very broad, we used subjectively chosen best frequency tone burst or white noise burst stimulation of at least 80 dB for subsequent characterization. 16 out of 19 DKO units responded with a relatively high onset rate but strong adaptation to a minimal sustained spiking (Fig. 6A-B), which prevented unit classification based on the response patterns. Judging from other criteria described above, we assume that approximately 13 of these represent SGNs and 3 represent bushy cells or multipolar cells of the anteroventral cochlear nucleus. Three additional units (two resembling a pauser/buildup- and one displaying a clear chopper neuron-type response) were excluded from further analysis. Overall, in DKO neurons, peak and adapted rates were greatly reduced as compared to WT controls (Fig. 6C), which came along with an increased latency (12.6 ± 2.8 ms vs WT 4.8 ± 0.2 ms; p < 0.001, Wilcoxon rank-sum test) and increased jitter (18.2 ± 5.0 vs WT 1.5 ± 0.4; p < 0.00001, Wilcoxon rank-sum test) of the first spike that followed the spike onset.
DKO responses greatly improved when we lowered the stimulus repetition rate to 0.5 Hz, and plateaued at 0.2 Hz (Fig. 6C-D). We hypothesize that in DKO SGNs, recovery of CaV1.3 channels after a silent interval in the range of 1-2 seconds is the limiting factor for determining the maximal SGN spiking rate at the sound onset. This may be further contributed by potentially slowed re-supply of synaptic vesicles, as suggested by reduced efficiency of sustained exocytosis in the CaBP1/2-deficient IHCs (see Fig. 3). In contrast, in WT SGNs, the maximal spike rates are mostly recovered after a silent interval of 150 ms, which may be mostly limited by vesicle replenishment at the IHC ribbon synapse 35,41. During a 5-Hz sound stimulation, the strength of adaptation was greatly enhanced (peak to adapted ratio 12.5 ± 3.0 vs WT 3.4 ± 0.1; p < 0.00001, Wilcoxon rank-sum test) whereas the time constant of adaptation was unchanged (7.9 ± 2.0 ms in DKO compared to 7.0 ± 0.4 ms in WT (p = 0.5, Wilcoxon rank-sum test). However, for the longer 500-ms tone bursts used in the 0.5 Hz stimulation dataset, it becomes obvious that despite the strong adaptation, DKO neurons are still able to sustain spiking at a low rate throughout the stimulation (Fig. 6B). DKOs also showed reduced spontaneous spiking activity (Fig. 6E), which supports the hypothesis that already in the absence of sound stimulation over tens of seconds a considerable fraction of calcium channels may be in steady-state inactivation.
The reduction in spike rates and their strong dependence on the strength of stimulation in the preceding seconds introduces experimental bias to our data. In the units of the DKO animals, it was not feasible to faithfully assess sound thresholds or dynamic ranges. Furthermore, the overall experimental yield was generally much lower in DKO animals. On average we hit 0.3 sound-responsive neurons per hour of recording, compared to 2.5 in WT. Very likely many DKO neurons failed to generate spontaneous and evoked action potentials, thus we were not able to identify them. Based on the DPOAE data we assume that cochlear mechanics and active amplification and thus the IHC receptor potential thresholds are mostly normal in the DKOs. We cannot exclude that IHC receptor potentials are decreased in amplitude, which however should rather reduce SGN adaptation. We thus interpret the observed increase in adaptation and delay of recovery of SGN spiking to be the effects of changed synaptic CaV1.3 channel kinetics as described above.
In summary, sound responses of DKO SGNs and cochlear nucleus neurons showed a drastic reduction in spontaneous and evoked action potential rates, which was associated with increased sound thresholds and reduced temporal precision of action potential generation. Together, these factors explain well the poor ABR thresholds (Fig. 5, Fig. S4A).
In this study, we examined the effects of genetic disruption of the two major CaBPs expressed in the cochlea. In the IHCs of CaBP1/2-deficient animals, we found largely inactivated Ca2+ currents with prolonged recovery from inactivation upon mild oscillatory stimulation, which did not support sufficient neurotransmission for proper sound encoding. The extent of channel inactivation and impairment of IHC synaptic exocytosis aggravated with the extent of IHC stimulation. Similarly, SGN action potential rates in vivo declined at increased sound stimulation rates and displayed very strong adaptation upon stimulation, going along with impaired sustained IHC exocytosis. The impairment was significantly stronger as compared to the phenotype in the absence of CaBP2 alone. This suggests that under in vivo conditions the roles of CaBP1 and 2 are partially overlapping.
Function and expression profile of CaBPs in the cochlea
The murine cochlea shows strong expression of CaBP1 and 2, while the levels of CaBP4 and 5 seem negligible 19,42,43. Among the three putative splice isoforms of CaBP2 differing in the N-terminal aminoacid sequence, the alternative long isoform, termed CaBP2-alt, may be the most abundant splice variant in the IHCs, as suggested by in situ hybridization 19, while SGNs seem to express none. So far our gene-rescue experiments were conducted with the conventional L-isoform only, which however shows strong inhibition of CaV1.3 inactivation in the HEK cells as well as IHCs 12,20. Alternative splicing also generates three CaBP1 variants 19. The alternative long variant caldendrin is primarily expressed in the SGNs or the surrounding glia cells, while weak expression of CaBP1-S/-L (as judged by the mRNA production) was detected in hair cells 19. Our DKO model lacks all splice isoforms of both CaBPs. Since Cabp1 KO animals show normal hearing several weeks after birth, and CaBP2 is absent from SGNs, we interpret the hearing impairment of the Cabp1/2 DKO animals to be primarily due to defects of hair cell function and judge contribution by SGN dysfunction as rather unlikely, at least in the early postnatal weeks.
The ABR thresholds of Cabp1 KO animals are normal at 4 weeks of age, but deteriorate with age, starting with high-frequency hearing loss at 9 weeks. Conversely, the suprathreshold ABRs to tone burst stimulation show increased amplitudes and reduced latencies during the first two months after birth. This suggested enhanced SGN activity at first, followed by slowly progressing functional decline, possibly due to overexcitability-related toxicity 11. Due to early strong impairment of hair cell synaptic function, the latter effects are not expected to occur in the DKO animals. However, ABR thresholds for clicks were further increased and an additional reduction in the summating potential was observed in the DKO animals at two as compared to one month of age despite relatively well-preserved DPOAEs (Fig. S4). It thus remains to be clarified whether the absence of CaBP1 and 2 may affect (IHC) mechanotransduction at more mature stage. Unfortunately, this is difficult to assess due to the increased fragility of hair bundles at mature age. Intriguingly, CaBPs share similarities with calcium and integrin-binding protein 2 (CIB2), associated with hair bundle defects 44–46. When we overexpressed CaBP2 in the DKO animals, interestingly, suprathreshold responses of ABR wave I were very large (Fig. 5) and the summated wave I-IV responses were not distinguishable from WT responses. However, tone burst and click thresholds did not completely recover to WT levels. The difference in the extent of rescue of threshold and suprathreshold responses may be due to possibly distinct requirements for CaBPs at synapses providing inputs to low- or high-threshold SGN fibers. Apart from a small shift in voltage activation and abundance of the channels 47, not much is known about the possible differences in the CaV1.3 properties and interaction partners at distinct ribbons. Neither is it known whether the composition of the channel complex [t.i. proportions of alpha subunit splice variants 18,48,49, and different beta subunits 50] differs systematically between ribbon-AZs partnered by different classes of SGN fibers. Investigation of intracellular Ca2+ signals from individual AZs may offer some answers in the future. Different amounts of rescue of threshold vs suprathreshold responses could be further due to confounding effects of CaBP1 and 2, e.g. augmented SGN response in Cabp2-injected DKO animals still lacking CaBP1 (reminiscent of the phenotype of young Cabp1 KOs) in combination with “only” partial rescue of CaBP2 function. Possible adverse effects of exogenous expression of CaBP2, e.g. off-target toxicity and uncontrolled overexpression of the transgene, have been discussed previously 20, and further experiments need to be performed in the future to achieve (close to) complete rescue of hearing function in these mutants.
Our results reveal that the amount of CaV1.3 inactivation increases dramatically when IHCs lack both major CaBPs. This suggests that the sole loss of either CaBP1 or 2, can be largely compensated by the other of the two CaBP members. Yet, the loss of CaBP2 alone still results in progressive hearing impairment in mice and DFNB93 in humans. The inability of CaBP1 to fully replace CaBP2 in the CaBP2-deficient IHCs may stem from either too low concentration of CaBP1 (and other CaBPs), or from distinct properties of the two proteins (enabling CaBP2 to exert more potent suppression of CaV1.3 inactivation, e.g. VDI, or potentially performing additional, yet unidentified cellular function). Interestingly, even the overexpression of CaBP2 did not completely rescue the hearing phenotype of DKO animals. Future experiments involving overexpression of CaBP1 (and possibly CaBP1-2 molecular chimeras) in Cabp2 SKO and/or Cabp1/2 DKO animals together with further biophysical interaction studies are required to gain a better understanding of the potentially distinct functions performed by the two major CaBPs in the hair cells.
CaBPs as the regulators of IHC synaptic transmission
Previously, patch-clamp recordings surprisingly revealed normal whole-cell exocytosis in CaBP2-deficient IHCs despite enhanced inactivation of calcium channels 12, and no obvious defects in the CaBP1-deficient IHCs 11. In contrast, IHCs deficient in both, CaBP1 and 2, displayed strongly impaired exocytosis. Transgenic overexpression of CaBP2 in IHCs resulted in a partial recovery of the Ca2+ influx and an even better recovery of exocytosis. Together, these results corroborate the hypothesis of the role of CaBPs in preventing extensive steady-state inactivation of CaV1.3 channels to support sufficient synaptic exocytosis for proper sound encoding 12. The lack of CaBP1 and 2 seems to impair sustained exocytosis and efficiency of exocytosis beyond the levels that could be attributed to reduced QCa. This became particularly apparent when the IHCs were challenged by higher holding potentials in combination with large sine wave amplitudes during capacitance measurements, which we assume to partially deplete the pool of RRP vesicles. At present, it is not clear how CaBPs may directly affect the exocytosis, e.g. by influencing the coupling of the CaV1.3 channels and synaptic vesicles or potentially affecting the expression levels of other synaptic proteins that mediate synaptic vesicle replenishment. In this respect, it is worth noting that CaBP1 and 2 have been reported to modulate CaMKII activity 2, which in turn may through phosphorylation modify the Ca2+ affinity of otoferlin C2 domains 51. In the future, it is thus essential to investigate if CaBPs regulate the function of otoferlin and thus synaptic processes. Furthermore, other members of the CaBP/CaM family have been suggested to interact with synaptic proteins and/or be involved in the regulation of synaptic vesicle cycle: CaM in endocytosis 52, CaBP5 in priming via interaction with Munc18-1 53, CaBP1 via inhibition of IP3 receptors 54,55. An important challenge for future studies is to identify further CaBP(2)-interaction partners and possible functions beyond modulation of calcium channels.
The recovery of CaV1.3 channels from inactivation upon prolonged milder stimulation (2 min sine wave) was following a single exponential course in WT IHCs, but additional slower component was detected in the IHCs deficient in CaBP1 and 2. In the IHCs of injected DKO animals, this second component appeared smaller and faster, suggesting that CaBPs speed-up recovery of CaV1.3 channels from inactivation. Upon a comparably shorter (500 ms) but stronger depolarization, the recovery of CaV1.3 from inactivation was very fast in DKO IHCs, in fact faster than in (WT or) Cabp2-injected DKO IHCs. We hypothesize this to be due to faster Ca2+ clearance from the DKO IHC synapses as the intracellular [Ca2+] levels are expected to rise significantly less in CaBP1/2-deficient IHCs due to fast and pronounced CaV1.3 inactivation. In contrast, higher [Ca2+] may buildup at the synapses of WT and “rescued” DKO IHCs; consequently, CDI may be prolonged. Another plausible explanation for the observation is, if we consider that CaPB2 may act as a slow mobile calcium buffer 12. In the presence of excessive “overexpressed” CaBP2 in injected IHCs of DKO animals, the depolarization-evoked intracellular [Ca2+] transients may become biexponential as has been described for parvalbumin alpha 56, which may prolong the time constant of Ca2+ clearance and determine the kinetics of channel recovery from inactivation. Our results of the recovery upon 500-ms depolarization may thus mostly reflect the kinetics of the Ca2+ clearance from the cytosol, while the experiments with prolonged sine wave stimulation (when e.g. the function of calcium pumps and other extrusion systems and intracellular [Ca2+] levels may be at equilibrium), may report the speed of calcium channel recovery from a slow inactivation. Slow inactivation with very slow recovery was described for NaV, KV and CaV channels 26–29. It has to be noted that the time constant of recovery of CaV1.2 from slow VDI was comparably shorter (4 s) as observed in the present study 29. Possibly different conditions (e.g. stimulus parameters) or channel subtype may underlie this difference. Future experiments combining patch-clamp recordings with calcium imaging should be employed to address this question further.
From our in vitro recordings at higher holding potential or upon prolonged stimulation with small voltage oscillations we predict that in vivo a noteworthy proportion of CaV1.3 channels may be effectively inactivated and not available for mediating IHC synaptic transmission. This could explain the low experimental yield, the low “spontaneous” spiking rates of the SGNs and lower sound-evoked spike rates with higher thresholds. Similarly, thresholds may be high for tone bursts in the ABR measurements due to high proportion of inactivated channels upon stimuli at high repetition rate used for these recordings (20-40 Hz repetition rate for click and tone burst stimulation, respectively). We believe that any sound stimulation and potentially even spontaneous activity may effectively silence DKO synapses. As expected, longer interstimulus intervals may allow a larger proportion of channels to recover between stimuli, increasing the onset rates of SGN spiking while decreasing the first spike latency and jitter. Still, high-intensity stimulation had to be used in the DKO animals to evoke an increase in SGN spiking. It is possible that spontaneous activity and/or our routines used to determine the SGN characteristic frequency inactivate a majority of IHC CaV1.3 channels thus not allowing reliable assessment of the characteristic frequency. In addition to CaV1.3 channel inactivation, other factors may contribute to low hearing sensitivity, such as possible impairment of synaptic vesicle replenishment, or IHC mechanotransduction, and, in some cases, impaired cochlear amplification. A complete lack of BK channels was shown to affect the IHC resting potential, receptor potential amplitude and timing, and consequently the timing and frequency of spiking in the SGNs 57. Lower expression of BK channels is not expected to significantly affect hearing thresholds, at least initially 58–60, and thus likely does not contribute to the hearing defect of the Cabp1/2 DKO animals. However, at advanced ages, the effects described above may combine with the lack of CaBPs to additionally desynchronize already scarce spikes in the auditory nerve fibers.
Clinical implications for the diagnosis and treatment of DFNB93
The SGN sound encoding phenotype in Cabp2-KO mice was surprisingly mild compared to the deficits observed in DFNB93 patients e.g. 14,12: SGN thresholds and evoked spike rates were only minimally affected 12. Given the strong recovery phenotype in the DKO we now consider the possibility that synaptic activity in response to everyday sounds could already inactivate calcium channels and reduce SGN spiking. In this case, communication in daily life would be more strongly impaired than audiometric data obtained in a soundproof chamber, especially when short stimuli separated by silent intervals are used. In mild cases, a gap detection test may be used to assess auditory fatigue. Based on our data we further predict that patients that have a stronger channel inactivation phenotype might not benefit much from sound amplification by hearing aids. Instead, they should use technologies that improve the signal to noise ratio (e.g. noise cancellation, directional microphones, wireless transmitters). An alternative or additional future treatment approach may be gene therapy. In our hands, re-expression of CaBP2 by genetic treatment partially restored CaV1.3 properties, IHC exocytosis and consequently ABR wave amplitudes and thresholds. This successful viral rescue supports the view that a reversible inactivation of IHC CaV1.3 currents is the main mechanism of hearing loss in CaBP-related deafness, which can be targeted by gene therapy in the future.
Materials and Methods
For the generation of the Cabp1/Cabp2-double-knockout (Cabp1/2-DKO) line, CaBP2LacZ/LacZ mice, carrying LacZ trapping cassette in place of the exons 3 and 4 12, were crossbred with Cabp1 KO mice 61 and maintained on a C57/Bl6 background. Mice from 3-13 weeks were used for electro-physiological and systems physiology recordings. All experiments complied with the national animal care guidelines and were approved by the University of Göttingen Board for animal welfare and the animal welfare office of the state of Lower Saxony.
Systems physiology recordings: ABR and DPOAE
Prior to the experiments mice were weighed and anesthetized via intraperitoneal injections of ketamine (125 µg/g) and xylazine (2.5 µg/g). Recordings were obtained in a custom-made soundproof box with a Tucker-Davis (TDT) BioSig System II/BioSig software (Tucker-Davis Technologies, Alachua, FL). Body temperature was maintained using a heating pad (37°C) and ECG was monitored continuously. For ABR recordings, subcutaneous needles were placed at the vertex, the ipsilateral pinna and at the lower back. Using a JBL2402 speaker in free-field configuration, tone bursts (10-ms plateau, 1-ms cos2 rise/fall) or clicks of 0.03 ms were ipsilaterally presented at 40 Hz for tone bursts or 20 Hz for clicks. ABR traces were amplified 50000-fold, band-pass filtered between 400 Hz and 4 kHz (NeuroAmp) and averaged from 1300 repetitions sampled at 50 kHz for 20ms. Thresholds were defined in stacked waveforms as the lowest stimulus intensity (dB SPL) at which a reproducible waveform could be visually detected with a 10 dB precision. Wave I amplitudes were measured peak-to-trough. DPOAEs were measured using a custom-written MATLAB (MathWorks) routine. Two primary tones f1 and f2 (ratio f1/f2 = 1.2, level difference f2 = f1 + 10 dB) were presented via two speakers (MF-1, TDT) and a custom-made ear probe containing an MKE-2 microphone (Sennheiser). Stimulus duration was 16 ms and f2 levels varied from 10 to 70 dB SPL in 10 dB steps. The microphone signal was amplified (DMX 6Fire; Terratec (young animals), or UAC-2; Zoom (older animals)) and digitized (TDT System III). DPOAE amplitude was analyzed at 2×f2-f1 using custom-written MATLAB software (MathWorks) with fast Fourier transformation. DPOAE thresholds were determined as responses that exceeded -18 dB SPL in young and -4 dB SPL in older animals.
Auditory nerve fiber recordings
Mice were anesthetized by injection of xylazine (5 mg/kg), urethane (1.32 mg/kg) and buprenorphine (0.1 mg/kg) i.p., tracheotomized, and placed in a stereotactic apparatus. The left occipital bone and the cerebellum were partially removed to visualize the left superior semicircular canal. The latter served as a landmark to navigate a glass microelectrode through the cochlear nucleus, aiming at the location where the auditory nerve leaves the internal auditory canal and enters the cochlear nucleus. 85 or 90 dB broadband noise bursts (ScanSpeak, Avisoft Bioacoustics) were used as search stimuli to identify neurons that displayed sound-driven spiking activity. They were characterized by spontaneous rate and frequency tuning and the responses to tone bursts at the characteristic frequency, 30 dB above the threshold, as previously described 35. In the majority of sound-responsive DKO neurons, thresholds were so high and tuning so broad that neither the automated tuning curve algorithm nor systematic or manual changes in sound frequency elicited frequency-specific responses. In these cases, we either chose a subjective “best frequency” or used white noise burst stimulation at 85 dB.
For patch-clamp recordings, the apical coils of 3-4-week-old animals were dissected in HEPES-Hanks solution containing (in mM): 10 HEPES, 5.35 KCl, 141.7 NaCl, 0.5 MgSO4, 1 MgCl2, 11.1 D-glucose and 3.42 L-glutamine. pH was adjusted to 7.2. IHCs of the ∼6 kHz tonotopical region were investigated using the perforated patch configuration. The pipette solution contained (in mM): 130 Cs-gluconate, 10 TEA-Cl, 10 4-AP, 1 MgCl2, 10 HEPES, 300 mg/mL amphotericin B, pH 7.2 ∼290 mOsm. The bath solution was composed of (in mM): 111 NaCl, 35 TEA-Cl, 2.8 KCl, 1 CsCl, 1 MgCl2, 10 NaOH-Hepes, 11.3 D-glucose, and 1.3 or 2 CaCl2 or BaCl2 (pH 7.2, 300–310 mOsm). When the bath solution contained 2 mM BaCl2/CaCl2, NaCl was adjusted to 110 mM. The bath solution was supplemented by the SK2 channel blocker apamin (100 nM; Peptanova, Germany) freshly on the day of the experiment. Recordings were performed at room temperature using an EPC-9 amplifier (Heka-Germany) controlled by Pulse software. Currents were leak-corrected using a p/10 protocol, sampled at 50 kHz and corrected offline for the liquid junction potential (-15 mV). IHCs were patched at a holding potential of -85 mV (in a subset of experiments, -65 and -55 mV) and stimulated with depolarizations to the peak Ca2+-current potential. Changes in membrane capacitance were monitored by applying a 2-kHz-sinewave with an amplitude of 35 mV. In a subset of experiments, the amplitude was reduced to 15 mV to diminish possible stimulation of the cells caused by the sinewave. VDI and CDI were calculated as follows: VDI = 1 - IBa_500/IBa_initial and CDI = 1 − (ICa_500/I Ca_initial)/(IBa_500/IBa_initial), where IBa_500 and ICa_500 represent whole cell Ba2+ and Ca2+ currents 500 ms after beginning of depolarization, and IBa_initial and ICa_initial correspond to the initial, peak currents with the respective divalent cations 22.
Apical coils (6-8 kHz tonotopical region) of the organ of Corti were fixed with 4% (vol/vol) formal-dehyde in PBS on ice. After incubation for 1 h at room temperature in goat serum dilution buffer [16% (vol/vol) normal goat serum, 450 mM NaCl, 0.6% Triton X-100, 20 mM phosphate buffer (pH 7.4)], the explants were incubated in primary antibodies over night at 4 °C. The following antibodies were used: chicken anti-Homer1 (1:200, Synaptic Systems), rabbit anti-Myosin6 (1:200, Proteus), guinea pig anti-Synapsin1/2 (1:200, Synaptic Systems), rabbit anti-RibeyeA (1:200, Synaptic Systems), guinea pig anti–parvalbumin-α (1:200; Synaptic Systems), rabbit anti-BK (1:100; Alomone), rabbit anti-SK2 (1:200; Alomone), mouse IgG1 anti-CtBP2 (1:200; BD Biosciences), rabbit anti-GluA3 (1:200; Santa Cruz), rabbit–anti-CaV1.3 (1:100; Alomone), chicken anti-GFP (1:200, Abcam). The explants were then incubated with secondary AlexaFluor-labeled antibodies (1:200; Molecular Probes) for 1 h at room temperature, washed 3 times in PBS and mounted. Confocal images were acquired with an Abberrior Instruments Expert Line STED microscope using a 1.4 NA 100x oil immersion objective. ImageJ software (NIH) was employed to analyze 2D confocal stacks and generate maximum projections. Volumetric synapse analysis ran on Imaris (Bitplane, Zurich, Switzerland) paired with a custom-written Matlab software 62.
Molecular biology and AAV vector production
The construct containing the long isoform of mouse CaBP2 (CaBP2-L; NCBI Reference Sequence: NP_038906.2) and eGFP is depicted in Fig. S1A. AAV9-PHP.eB particles were generated using our standard AAV purification procedure previously described in more detail in ref. 63. In brief, triple transfection of HEK-293T cells was performed using pHelper plasmid (TaKaRa/Clontech), the trans-plasmid providing viral capsid PHP.eB (a generous gift from Viviana Gradinaru, Addgene plasmid #103005) and the cis-plasmid providing hCMV/HBA_wtCaBP2-P2A-eGFP. PHP.eB viral particles were harvested from the medium 72 h after, and from cells and the medium 120 h after transfection. Precipitation was done with 40% polyethylene glycol 8000 (Acros Organics, Germany) in 500 mM NaCl for 2 h at 4°C. Both, precipitate and cells were lysed in high salt buffer (500 mM NaCl, 2 mM MgCl2, 40 mM Tris-HCl, pH 8,0) and non-viral DNA was degraded using salt-activated nuclease (SAN, Arcticzymes, USA). The cell lysates were clarified by centrifugation at 2000×g for 10 min and then purified over iodixanol (Optiprep, Axis Shield, Norway) step gradients (15, 25, 40, and 60%) at 350000×g for 2.25 h 64,65. Finally, viral particles were concentrated using Amicon filters (EMD, UFC910024) and formulated in sterile PBS supplemented with 0.001% Pluronic F-68 (Gibco, Germany). Virus titer was 8.2×1012 genome copies/ml measured by determining the number of DNase I resistant genome copies using qPCR (StepOne, Applied Biosystems) and AAV titration kit (TaKaRa/Clontech). The purity of viruses was routinely checked by silver staining (Pierce, Germany) after gel electrophoresis (NovexTM 4-12% Tris-Glycine, Thermo Fisher Scientific).
Mice were injected at postnatal day 6 using the round window approach as described in earlier studies 66. In brief, anesthesia was established with isoflurane (5% for induction, 2-3% for maintenance, frequent testing of the absence of hind-limb withdrawal reflex). For analgesia, buprenorphine (0.1 mg/kg, 30 min before surgery) and carprofen (5 mg/kg, during and 1 day after surgery) were applied subcutaneously and Xylocain (10 mg spray) locally. Body temperature was maintained by a custom-built heating blanket. Following a retro-auricular approach, the facial nerve was exposed to determine where to puncture the cartilaginous bulla with the injection pipette and target the scala tympani where virus suspension (1-1.5 μl, corresponding to 0.82-1.23×1010 AAV particles) was injected. Following injection, the endogenous tissue was relocated and the surgical situs closed by suturing the skin. 2-3 weeks later, the injected mice were used for hearing tests, in vitro electrophysiology and immunohistochemistry of the cochleae.
Data were analyzed using Igor Pro Software (Wavemetrics), GraphPad PRISM (Graphpad Software, Inc.), Matlab (Mathworks) and Imaris (Bitplane). All data are presented as mean ± SEM. For the analysis of half-max Ca2+ current (Vhalf) and activation constant (kact), current-voltage traces were transformed into an activation function (ICa = gmax(V-Vrev)), which was subsequently fitted by a Boltzmann function: 1/(1 + exp((V0.5−V)/kact)). Gmax describes the maximum conductance, V the holding potential and Vrev the reversal potential. Data were tested for randomness, normality (Jaque-Bera test), equality of variances (F-test) and finally compared for statistical significance by performing Student’s t-test for normally distributed data of equal variance or else by the Wilcoxon rank sum test. Systems physiology data were tested for significance using ANOVA with Tukey’s or Ŝidak’s multicomparisons tests.
The authors would like to thank Christiane Senger-Freitag for the virus injections, Daniela Gerke for help with virus production, and Sandra Gerke, Ina Preuss, and Sina Langer for excellent technical support. The authors would further like to thank Amy Lee (University of Iowa) for providing the Cabp1 KO animals, and Mohona Mukhopadhyay and Tobias Moser for critical reading of the paper, useful comments and suggestions. This work was supported by the German Research Foundation: DFG Priority Program 1608 “Ultrafast and temporally precise information processing: Normal and dysfunctional hearing” [to TP (PA 2769/1-1)], the Clinician Scientist Program “Cell Dynamics in Disease and Therapy” at the University Medical Center Göttingen (to DO; project number 413501650), CRC889 (subproject A06; to NS, and B09; to TP); and by the Multiscale Bioimaging Cluster of Excellence (EXC 2067/1-390729940; to TP).
- 1.Voltage-Gated Calcium Channels: Key Players in Sensory Coding in the Retina and the Inner EarPhysiol. Rev 98:2063–2096
- 2.Five members of a novel Ca(2+)-binding protein (CABP) subfamily with similarity to calmodulinJ. Biol. Chem 275:1247–1260
- 3.Structural basis for the differential effects of CaBP1 and calmodulin on Ca(V)1.2 calcium-dependent inactivationStruct. Lond. Engl. 1993 18:1617–1631
- 4.Essential role of Ca2+-binding protein 4, a Cav1.4 channel regulator, in photoreceptor synaptic functionNat. Neurosci 7:1079–1087
- 5.Differential modulation of Ca(v)2.1 channels by calmodulin and Ca2+-binding protein 1Nat. Neurosci 5:210–217
- 6.Switching of Ca2+-dependent inactivation of Ca(v)1.3 channels by calcium binding proteins of auditory hair cellsJ. Neurosci 26:10677–10689
- 7.Allostery in Ca2+ channel modulation by calcium binding proteinsNat. Chem. Biol 10:231–238
- 8.Loss of Ca(v)1.3 (CACNA1D) function in a human channelopathy with brad-ycardia and congenital deafnessNat. Neurosci 14:77–84
- 9.Congenital deafness and sinoatrial node dysfunction in mice lacking class D L-type Ca2+channelsCell 102:89–97
- 10.Ca2+-binding proteins tune Ca2+-feedback to Cav1. 3 channels in mouse auditory hair cellsJ. Physiol 585:791–803
- 11.Functions of CaBP1 and CaBP2 in the peripheral auditory systemHear. Res 364:48–58
- 12.Ca2+-binding protein 2 inhibits Ca2+-channel inactivation in mouse inner hair cellsProc. Natl. Acad. Sci 114:E1717–E1726
- 13.A Novel Pathogenic Variant in the CABP2 Gene Causes Severe Nonsyn-dromic Hearing Loss in a Consanguineous Iranian FamilyAudiol. Neurotol 24:258–263
- 14.A Mutation in CABP2, Expressed in Cochlear Hair Cells, Causes Autoso-mal-Recessive Hearing ImpairmentAm. J. Hum. Genet 91:636–645
- 15.First reported CABP2 ‐related non‐syndromic hearing loss in Northern EuropeMol. Genet. Genomic Med 9
- 16.DFNB93, a novel locus for autosomal recessive moderate-to-severe hearing impairmentClin. Genet 79:594–598
- 17.Functional Properties of a Newly Identified C-terminal Splice Variant of Cav1.3 L-type Ca2+ ChannelsJ. Biol. Chem 286:42736–42748
- 18.Cell-type-specific tuning of Cav1.3 Ca(2+)-channels by a C-terminal auto-modulatory domainFront. Cell. Neurosci 9
- 19.Expression and Localization of CaBP Ca2+ Binding Proteins in the Mouse Coch-leaPLoS ONE 11
- 20.Cabp2-Gene Therapy Restores Inner Hair Cell Calcium Currents and Improves Hearing in a DFNB93 Mouse ModelFront. Mol. Neurosci 14
- 21.Complex regulation of voltage-dependent activation and inactivation properties of retinal voltage-gated Cav1.4 L-type Ca2+ channels by Ca2+-binding protein 4 (CaBP4)J. Biol. Chem 287:36312–36321
- 22.Molecular endpoints of Ca2+/calmodulin- and voltage-dependent inactivation of Ca(v)1.3 channelsJ. Gen. Physiol 135:197–215
- 23.Low-frequency characteristics of intracellularly recorded receptor potentials in guinea-pig cochlear hair cellsJ. Physiol 338:179–206
- 24.Optimised tuning of auditory inner hair cells to encode complex sound through synergistic activity of six independent K+ current entitiesMol. Cell in press
- 25.Resting potential and submembrane calcium concentration of inner hair cells in the isolated mouse cochlea are set by KCNQ-type potassium channelsJ. Neurosci. Off. J. Soc. Neurosci 23:2141–2149
- 26.Distribution and kinetics of membrane dielectric polarization. 1. Long-term inactivation of gating currentsJ. Gen. Physiol 79:21–40
- 27.Crystal structure of a voltage-gated sodium channel in two potentially inactivated statesNature 486:135–139
- 28.Correlation between Charge Movement and Ionic Current during Slow Inactivation in Shaker K+ ChannelsJ. Gen. Physiol 110:579–589
- 29.Two components of voltage-dependent inactivation in Ca(v)1.2 channels revealed by its gating currentsBiophys. J 84:3662–3678
- 30.Cochlea-Specific Deletion of Cav1.3 Calcium Channels Arrests Inner Hair Cell Differentiation and Unravels Pitfalls of Conditional Mouse ModelsFront. Cell. Neurosci 13
- 31.Vesicular Glutamatergic Transmission in Noise-induced Loss and Repair of Cochlear Ribbon SynapsesJ. Neurosci 39:2228–18
- 32.Ca V 1.3 Channels Are Essential for Development and Presynaptic Activity of Cochlear Inner Hair CellsJ. Neurosci 23:10832–10840
- 33.The Ca2+Channel Subunit beta2 Regulates Ca2+Channel Abundance and Function in Inner Hair Cells and Is Required for HearingJ. Neurosci 29
- 34.Hair cell synaptic ribbons are essential for synchronous auditory signallingNature 434:889–894
- 35.Hair cell synaptic dysfunction, auditory fatigue and thermal sensitivity in otoferlin Ile515Thr mutantsEMBO J 35
- 36.Cochlear neuropathy and the coding of supra-threshold soundFront. Syst. Neurosci 8
- 37.The synaptic ribbon is critical for sound encoding at high rates and with temporal precisioneLife 7
- 38.Disruption of the presynaptic cytomatrix protein bassoon degrades ribbon anchorage, multiquantal release, and sound encoding at the hair cell afferent synapseJ. Neurosci 33:4456–4467
- 39.Response Properties of Single Auditory Nerve Fibers in the MouseJ. Neurophysiol 93:557–569
- 40.The BEACH protein LRBA is required for hair bundle maintenance in cochlear hair cells and for hearingEMBO Rep 18:2015–2029
- 41.Hearing requires otoferlin-dependent efficient replenishment of synaptic vesicles in hair cellsNat. Neurosci 13:869–876
- 42.Single-cell transcriptomic profiling of the mouse cochlea: An atlas for targeted therapiesProc. Natl. Acad. Sci 120
- 43.Profiling mouse cochlear cell maturation using 10× Genomics single-cell transcriptomicsFront. Cell. Neurosci 16
- 44.CIB2 interacts with TMC1 and TMC2 and is essential for mechanotransduction in auditory hair cellsNat Commun 8
- 45.Mutations in CIB2, a calcium and integrin binding protein, cause Usher syndrome type 1J and nonsyndromic deafness DFNB48Nat. Genet 44:1265–1271
- 46.Loss of CIB2 Causes Profound Hearing Loss and Abolishes Mechanoelectrical Transduction in MiceFront. Mol. Neurosci 10
- 47.Hair cells use active zones with different voltage dependence of Ca2+influx to decompose sounds into complementary neural codesProc. Natl. Acad. Sci 113:E4716–E4725
- 48.Modulation of Voltage- and Ca2+-dependent Gating of CaV1.3 L-type Calcium Channels by Alternative Splicing of a C-terminal Regulatory DomainJ. Biol. Chem 283:20733–20744
- 49.Different CaV1.3 Channel Isoforms Control Distinct Components of the Synaptic Vesicle Cycle in Auditory Inner Hair CellsJ. Neurosci. Off. J. Soc. Neurosci 37:2960–2975
- 50.RBP2 stabilizes slow Cav1.3 Ca2+ channel inactivation properties of coch-lear inner hair cellsPflüg. Arch. - Eur. J. Physiol 472:3–25
- 51.Activity-Dependent Phosphorylation by CaMKIIΔ Alters the Ca2+Affinity of the Multi-C2-Domain Protein OtoferlinFront. Synaptic Neurosci 9
- 52.Ca2+ and calmodulin initiate all forms of endocytosis during depolarization at a nerve terminalNat. Neurosci 12:1003–1010
- 53.Insight into the Role of Ca 2+-Binding Protein 5 in Vesicle ExocytosisInvestig. Opthalmology Vis. Sci 52
- 54.Calcium-binding Protein 1 Is an Inhibitor of Agonist-evoked, Inositol 1,4,5-Trisphosphate-mediated Calcium SignalingJ. Biol. Chem 279:547–555
- 55.Regulation of InsP3 receptor activity by neuronal Ca2+-binding proteinsEMBO J 23:312–321
- 56.Kinetics of Ca2+ binding to parvalbumin in bovine chro-maffin cells: implications for [Ca2+] transients of neuronal dendritesJ. Physiol 525:419–432
- 57.The role of BKCa channels in electrical signal encoding in the mammalian auditory peripheryJ. Neurosci. Off. J. Soc. Neurosci 26:6181–6189
- 58.Deletion of the Ca2+-activated potassium (BK) alpha-subunit but not the BKbeta1-subunit leads to progressive hearing lossProc. Natl. Acad. Sci. U. S. A 101:12922–12927
- 59.Olivocochlear suppression of outer hair cells in vivo: evidence for combined action of BK and SK2 channels throughout the cochleaJ. Neurophysiol 109:1525–1534
- 60.Cochlear Function in Mice Lacking the BK Channel α, β1, or β4 SubunitsJ. Biol. Chem 282:3312–3324
- 61.Localization and expression of CaBP1/caldendrin in the mouse brainNeuroscience 268:33–47
- 62.Quantitative optical nanophysiology of Ca2+signaling at inner hair cell active zonesNat. Commun 9
- 63.Application of Targeting-Optimized Chronos for Stimulation of the Auditory Pathwayin Methods in Molecular Biology (Clifton, N.J 2191:261–285
- 64.Production and characterization of adeno-associated viral vectorsNat. Protoc 1:1412–1428
- 65.Recombinant adeno-associated virus purification using novel methods improves infectious titer and yieldGene Ther 6:973–985
- 66.Overloaded Adeno-Associated Virus as a Novel Gene Therapeutic Tool for Otoferlin-Related DeafnessFront. Mol. Neurosci 13