Functional implications of Cav2.3 R‐type voltage‐gated calcium channels in the murine auditory system – novel vistas from brainstem‐evoked response audiometry

Voltage‐gated Ca2+ channels (VGCCs) are considered to play a key role in auditory perception and information processing within the murine inner ear and brainstem. In the past, Cav1.3 L‐type VGCCs gathered most attention as their ablation causes congenital deafness. However, isolated patch‐clamp investigation and localization studies repetitively suggested that Cav2.3 R‐type VGCCs are also expressed in the cochlea and further components of the ascending auditory tract, pointing to a potential functional role of Cav2.3 in hearing physiology. Thus, we performed auditory profiling of Cav2.3+/+ controls, heterozygous Cav2.3+/− mice and Cav2.3 null mutants (Cav2.3−/−) using brainstem‐evoked response audiometry. Interestingly, click‐evoked auditory brainstem responses (ABRs) revealed increased hearing thresholds in Cav2.3+/− mice from both genders, whereas no alterations were observed in Cav2.3−/− mice. Similar observations were made for tone burst‐related ABRs in both genders. However, Cav2.3 ablation seemed to prevent mutant mice from total hearing loss particularly in the higher frequency range (36–42 kHz). Amplitude growth function analysis revealed, i.a., significant reduction in ABR wave WI and WIII amplitude in mutant animals. In addition, alterations in WI‐WIV interwave interval were observed in female Cav2.3+/− mice whereas absolute latencies remained unchanged. In summary, our results demonstrate that Cav2.3 VGCCs are mandatory for physiological auditory information processing in the ascending auditory tract.


| 1585
LUNDT eT aL. active zone of HCs. As expected, Ca v 1.3 −/− IHCs exhibited only marginal exocytosis, lacked early Ca 2+ -dependent action potentials and exhibited a complex developmental failure (Brandt et al., 2003). Similar to the IHCs, VGCCs also seem to be mandatory for the maturation of OHCs as the latter degenerate in Ca v 1.3 −/− mice shortly after the time point of normal physiological onset of hearing (Glueckert et al., 2003;Michna et al., 2003). Whereas Ca v 1.3 L-type Ca 2+ channels have been in the focus of interest, the low resting potentials of OHCs and their slight depolarization upon sound stimuli suggest that LVA Ca 2+ channels may also contribute to intracellular Ca 2+ regulation (Inagaki, Ugawa, Yamamura, Murakami, & Shimada, 2008). L-type Ca 2+ channels are likely to play a role in phasic neurotransmitter release (Dou et al., 2004), and the function of other VGCC entities may be obscured by their baseline activity and minimal contribution to Ca 2+ influx in hair cells (HCs) (Moser & Beutner, 2000;Spassova, Eisen, Saunders, & Parsons, 2001). Indeed, Dou et al. (2004) early suggested that other VGCCs contribute to the remaining dihydropyridine (DHP)-insensitive Ca 2+ current in HCs (Su, Jiang, Gu, & Yang, 1995;Platzer et al., 2000;Martini et al., 2000;Rodriguez-Contreras & Yamoah, 2001).
Ca v 2.3 VGCCs could serve as one of these candidates. From P2 to P10, Ca v 2.3 VGCCs seem to be expressed in the outer rather than the inner spiral bundle efferent endings and in medial efferent fibres. Astonishingly, Ca v 2.3 expression vanished around P14 but was observed later at P19 in the basal poles of the OHC membranes again (Waka et al., 2003). In addition, electrophysiological studies, in situ hybridization and RT-PCR also point to a functional expression of Ca v 2.3 in the ascending auditory tract (Parajuli et al., 2012;Soong et al., 1993;Williams et al., 1994). Functionally, Ca v 2.3 and Ca v 1.3 VGCCs share essential physiological properties. Ca v 1.3 was reported to be mid voltage-activated (MVA) to LVA instead of being a classical HVA Ca 2+ channel (Koschak et al., 2001;Michna et al., 2003). The same holds true for Ca v 2.3, as demonstrated by recent studies showing that Ca v 2.3 Ca 2+ channels can exhibit MVA to LVA properties depending on the presence or absence of divalent heavy metal ions in the brain (Shcheglovitov et al., 2012).
Additionally, low micromolar concentrations of DHPs cannot be used to reliably discriminate between L-type from Non-L-type HVA channels and Ca v 2.3 can clearly underlie a low DHP-sensitive Ca 2+ current component (Lu et al., 2004;Stephens, Page, Burley, Berrow, & Dolphin, 1997;Weiergraber, Kamp, et al., 2006b). Considering that Ca v 1.3 and Ca v 2.3 VGCCs are coexpressed in many regions, it becomes obvious that both channels might functionally contribute to a low-to mid voltage-activated and low DHP-sensitive Ca 2+ current component in the auditory tract (Perez-Reyes, 2003;Shcheglovitov et al., 2012;Weiergraber, Kamp, et al., 2006b).
Based on these findings, we performed auditory profiling of Ca v 2.3 +/− and Ca v 2.3 −/− mice using brainstem-evoked response audiometry. Our results demonstrate complex alterations in click and tone burst-related hearing thresholds and amplitude growth function in Ca v 2.3 +/− and Ca v 2.3 −/− mice with a potential gene dose-dependent effect. This is the first report of altered auditory information processing in Ca v 2.3 mutant animals.

| Experimental animals
Ca v 2.3 +/− embryos (kindly provided by Richard J. Miller; Department of Neurobiology Pharmacology, and Physiology; The University of Chicago; Chicago) were re-derived with C57BL/6J mice and maintained with random intra-strain mating obtaining all genotypes (Wilson et al., 2000). The mutant line was originally generated by the use of homologous recombination in which the S4-S6 region of domain II was replaced with a neomycin/URA3 selection cassette. Removal of the pore-lining and its neighbouring transmembrane regions resulted in a null allele of Cacna1e with no detectable Ca v 2.3 transcript in Northern blot analysis and no detectable Ca v 2.3 protein in Western blot analysis in Ca v 2.3 knockouts (Wilson et al., 2000). The resultant Ca v 2.3 −/− mice represent a constitutive knockout.
All mice were housed in groups of 2-5 in clear Makrolon cages type II with ad libitum access to drinking water and standard food pellets. Using ventilated cabinets (Model 9AV125PYN, Tecniplast, Germany; UniProtect, Zoonlab, Germany) as a noise-protected environment, mice were maintained at a temperature of 21 ± 2°C, 50%-60% relative humidity, and on a conventional 12-hr light/dark cycle with a light onset at 5:00 a.m. Prior to experimentation, the animals were strictly adapted to this circadian pattern for 14 days (Lundt, Seidel, et al., 2019;. All animal experimentation was carried out according to the guidelines of the German Council on Animal Care, and all protocols were approved by the local institutional and national committee on animal care (LANUV). The authors further certify that all animal experimentation was carried out in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals    (Lundt, Seidel, et al., 2019;. Specific effort was made to minimize the number of animals used and their suffering (3R strategy).
snap-frozen in liquid nitrogen. Subsequently, 1 ml of lysis buffer containing 5 mM Tris-HCl, 2 mM EDTA and proteinase inhibitors (cOmplete Protease Inhibitor Cocktail Tablet) (pH 7.4; all components obtained from Sigma-Aldrich) was added to the frozen tissue followed by homogenization using a rotor-stator homogenizer (TissueRuptor, Qiagen) for 20 s. Cortical samples were then centrifuged for 15 min at 500 × g at 4°C (Centrifuge 5417R; Eppendorf), and the supernatant was kept on ice. Homogenization and centrifugation of the remaining pellet were repeated with another 0.5 ml of lysis buffer, and both supernatants from each animal were finally merged. Subsequently, the merged supernatants were centrifuged at 100,000 × g for 40 min at 4°C (Ultracentrifuge Optima L-80XP, Beckman Coulter) and the resulting pellet was solubilized in 250 µl resuspension buffer (containing 75 mM Tris, 12.5 mM MgCl 2 , 5 mM EDTA and protease inhibitors (cOmplete Protease Inhibitor Cocktail Tablet) (all components obtained from Sigma-Aldrich). Protein concentration was determined using NanoDrop (NanoDrop 1,000 Spectrophotometer; Thermo Fisher), and appropriate microsomal aliquots were stored at −20°C.
For SDS-PAGE and Western blotting, 50 and 75 µg probes of cortical microsomes from each genotype were mixed with pre-heated 2 × Lämmli buffer (Bio-Rad) and loaded to a precast gel (7.5% Mini-PROTEAN TGX Precast Protein Gel, Bio-Rad). SDS-PAGE was carried out in a Mini-PROTEAN Tetra Cell (Bio-Rad) filled with TGS buffer (25 mM Tris, 192 mM glycine, 0,1% SDS, pH 8.3, Bio-Rad). Prior to blotting, the PVDF membrane was activated for 5 min in pure methanol (Sigma-Aldrich). The blotting sandwich made up of sponges, filter papers, membrane and SDS gel was assembled and inserted into a Mini Trans-Blot Cell (Bio-Rad). The individual components were pre-wetted, and the buffer tank was filled with TG buffer w/o methanol (25 mM Tris, 192 mM glycine, pH 8.3, Bio-Rad). A cooling unit was used to prevent the system from over-heating. Microsomal proteins were blotted for 1 hr at 100 V followed by overnight blotting at 30 V at 4°C to allow the transfer of high molecular weight proteins. Following transfer, the PVDF membrane air-dried for 4 hr to enhance protein fixation and was subsequently blocked for 2 hr in TBS-T (Bio-Rad), containing 5% milk powder and 5% goat serum. The membrane was stained with Ponceau S to check for proper protein transfer. In addition, the SDS gel was analysed for remaining proteins by Coomassie Blue staining. After rinsing with TBS-T, the PVDF membrane was separated into two parts (below and above 70kDa) and incubated with the 1st antibody overnight, at 4°C. The upper PVDF membrane, containing proteins larger than 70 kDa, was either incubated with a polyclonal Ca v 2.3 C-Term antibody (host: rabbit, mouse reactivity, diluted 1:1,000 in TBS-T; No. ABIN350140, antibodies-online.com, Germany) or with a polyclonal Ca v 2.3 II-III loop antibody (host: rabbit, mouse reactivity, corresponding to amino acid residues 892-907 of rat Ca v 2.3, diluted 1:200 in TBS-T; No. PA5-77300, Thermo Fisher). The lower PVDF membrane, containing proteins <70 kDa, was incubated with the control monoclonal antibody ß-actin (No. ab179467, diluted 1:5,000 in TBS-T, Abcam). Prior and post incubation with the secondary HRP-conjugated antibody (goat-anti rabbit HRP; 1:5,000; Abcam) for 1 hr at RT, the membrane slips were washed 3 times for 10 min in TBS-T using an orbital shaker (SI500, Stuart). Membranes were incubated for 1 min using Super Signal West Pico Plus Chemiluminescent Substrate (Thermo Fisher), and blot exposure was carried out using ChemiDoc Touch (Bio-Rad).

| ABR recording procedure
Prior to ABR recordings, animals were anesthetized by intraperitoneal (i.p.) injection of ketamine (100 mg/kg body weight, Ketanest ® S, 25 mg/ml, Pfizer) and xylazine (10 mg/ kg body weight, Rompun ® 2%, Bayer Health Care) and placed inside a sound-attenuating cubicle (ENV-018V, Med Association Inc.) lined with an acoustic foam ( Figure S1a). Additional technical/experimental details of this ABR approach such as electrical shielding, temperature support for anesthetized animals and protection from corneal desiccation were described in detail previously (Lundt, Seidel, et al., 2019;. For recording of monaural bioelectrical auditory potentials, subdermal stainless steel electrodes (27GA 12 mm, Rochester Electro-Medical) were inserted at the vertex, axial the pinnae (positive (+) electrode) and ventrolateral of the right pinna (negative (−) electrode) ( Figure S1c). The ground electrode was positioned at the hip of the animal (Lundt, Seidel, et al., 2019;. For details on impedance measurement of the electrodes, verification of proper electrode placement/ conductivity, loudspeaker positioning under free field conditions, and programming of stimulus protocols for click and tone bursts including the software used, see Lundt, Seidel, et al. (2019),   (Figure S1b). ABR data were sampled at 24.4 kHz, and signals were bandpass filtered (high pass 300 Hz, low pass 5 kHz) using a 6-pole Butterworth filter. The individual ABR data acquisition time was 25 ms consisting of a 5-ms baseline period prior to the individual acoustic stimulus onset (pre-ABR baseline) and exceeding the 10-ms ABR section by another 10-ms baseline (post-ABR baseline, Figure S2a) (Lundt, Seidel, et al., 2019;. Click stimuli were used to determine click thresholds, ABR wave I-IV amplitudes and wave I-IV latencies. Tone burst stimuli were utilized to identify frequency-specific hearing thresholds in the individual mouse lines in the frequency range of 1-42 kHz in 6 kHz steps. For averaging, the acoustic stimuli were applied 300 times at a rate of 20 Hz. ABR threshold recordings were carried out in the increasing SPL mode, that is in 5 dB steps for clicks and 10 dB steps for tone bursts, ranging from 0 to 90 dB. For further details concerning calibration of the ABR setup and online confirmation of spectral characteristics of sound stimuli using fast Fourier transformation (FFT), please refer to Lundt, Seidel, et al. (2019), .

| Analysis of hearing thresholds
To characterize the click and tone burst-derived thresholds of ABR recordings, three distinct time windows (TWs) were defined to calculate the signal-to-noise ratio (SNR): TW 1 (0-5 ms), TW 2 (5-15 ms) and TW 3 (15-25 ms). For the calculation of noise standard deviation of the baseline, ABR trace resetting and definition of ABR hearing thresholds, see Lundt, Seidel, et al. (2019),  and Figure S2a.

| ABR wave amplitude and wave latency analysis
For determination of positive (p) waves (peaks, see intercept points of red-grey lines with ABR trace) and negative (n) waves (pits, see intercept points of blue-orange lines with ABR trace, Figure S2b), a wavelet-based approach was carried out utilizing the "Mexican hat" wavelet which uses a default wavelet by the continuous wavelet transform (CWT)based pattern-matching algorithm (Du et al., 2006) related to the following equation (Daubechies, 1992): where s(t) is the signal, a is the scale, b is the translation, (t) is the mother wavelet, a,b (t) is the scaled and translated wavelet and C is the 2D matrix of wavelet coefficients.
A detailed description of this automated tool for ABR analysis is given in Lundt, Seidel, et al. (2019), . It allows for amplitude growth function analysis and latency comparison of all waves (W I-IV ), identifying maximum amplitudes ( Figure S2b, green crosses) and mean latencies ( Figure S2b, red-grey lines) of each of the four p-peaks within the time frame of the related n-peaks. Note that all results based on the self-programmed automatic wavelet tool were visually checked afterwards. In rare cases, individual ABR runs were excluded from statistics due to, for example, noise contamination (Lundt, Seidel, et al., 2019;.

| Real-time PCR of Ca v 2.3 mutant mouse cochlea
qPCR was carried out in male and female Ca v 2.3 +/+ , Ca v 2.3 +/− and Ca v 2.3 −/− mice to identify potential alterations in cochlear transcript levels of other VGCCs (i.e. HVA L-type Ca v 1.2 and Ca v 1.3, LVA T-type Ca v 3.1, Ca v 3.2 and Ca v 3.3) that were previously reported to be expressed within the cochlea and the auditory tract. For each genotype, the following subgroup was used for analysis: Ca v 2.3 +/+ : ♂, n = 8, 21.23 ± 0.16 weeks; ♀, n = 8, 21.54 ± 0.32 weeks; Ca v 2.3 ± : ♂, n = 8, 20.71 ± 0.14 weeks; ♀, n = 8, 22.25 ± 0.61 weeks; Ca v 2.3 −/− : ♂, n = 8, 20.98 ± 0.25 weeks; ♀, n = 6, 21.91 ± 0.50 weeks. Notably, experimental animals for cochlear qPCR analysis were not used in ABR experiments before. Both cochleae of each individual animal were dissected in an RNase-free environment (RNAlater stabilization reagent, Qiagen) and snap-frozen in liquid nitrogen. Total RNA from both mouse cochleae was extracted using Direct-zol RNA Micro Kit (Zymo Research, Freiburg i.Br.) followed by an additional step of DNase digest (Turbo DNA-free Kit, Ambion TM , Thermo Fisher Scientific). Quality and quantity of total RNA were evaluated using the NanoDrop standard procedures (NanoDrop1000, Thermo Fisher Scientific). cDNA synthesis was carried out using a two-step RT-PCR approach using both random hexamer and anchored-oligo(dt) 18 primers with 250 ng of total cochlea RNA from each animal for the final 50 μl first-strand cDNA mix (Transcriptor First-Strand cDNA synthesis Kit, Roche). cDNA (2 μl) served as template for qPCR (see below), and signal detection was based on SYBR Green I Master (Roche). qPCR experiments were performed using a LightCycler 480 System (Roche) with the following protocol (per cycle) being applied for all primer pairs (Table 1): 95°C (10 min, pre-incubation step); 95°C (10 s, denaturation step); 60°C (20 s, annealing step); and 72°C (30 s, extension step). In total, 40 cycles were performed.
All cochlea samples were tested in triplicates, and two negative controls in duplicates (no template; no RT) were added to the qPCR 96-well-plate (Roche). Furthermore, cochlea cDNA derived from C57BL/6J mice served as positive control and calibrator cDNA (again in triplicates in every plate) to avoid inter-run variations and guarantee statistical comparability among the plates. Amplification specificity was verified by melting curve analysis (LightCycler 480 System Software, Roche). Deionized, nuclease-free water (no cDNA) and total RNA samples (without RT) were used as controls and HPRT served as internal reference gene. The LightCycler 480 System software (Roche) was used to calculate the Ct-values (cycle threshold) (Lundt, Seidel, et al., 2019;.
Considering the individual primer efficiency, analysis and qPCR statistics were carried out using qBase + qPCR analysis software (Biogazelle) which is based on a delta-Cq quantification model with PCR efficiency correction, reference gene normalization and inter-run calibration (Hellemans, Mortier, Paepe, Speleman, & Vandesompele, 2007). The results were depicted as CNRQ (Calibrated Normalized Relative Quantity) and statistically analysed using the Mann-Whitney test (Lundt, Seidel, et al., 2019;.

| Statistical analysis
All results in this study are presented as group means ± SEM using GraphPad Prism 6 software (V6.07 GraphPad Software, Inc.). Both genders were analysed separately. Statistical differences were compared with an ordinary one-way ANOVA for click-evoked hearing thresholds analysis ( Figure 4) and differences in W I-IV interwave intervals (IWI, Figure 7) by Tukey's multiple comparisons test. Two-way repeated-measure ANOVA followed by Tukey's adjustment for multiple comparisons was performed to evaluate differences in tone burst-evoked hearing thresholds (Figure 5a,b) and to calculate amplitude growth function differences ( Figure 6). To test statistical significances, we used α-level = 0.05 and p-values defined as *p < .05; **p < .01; ***p < .001; ****p < .0001. Note that asterisks indicate significant differences between controls and mutant (Ca v 2.3 +/− or Ca v 2.3 −/− ) mice, whereas " + " icons represent significant differences between heterozygous and knockout animals.

| Click-and tone-evoked ABRs in control, Ca v 2.3 +/− and Ca v 2.3 −/− mice
To get a closer insight into the functional involvement of Ca v 2.3 VGCCs in auditory information processing, we performed click-and tone burst-evoked ABR recordings and analysis of hearing thresholds, amplitude growth functions and latencies in controls, Ca v 2.3 +/− and Ca v 2.3 −/− mice. Special attention was payed to gender-specific differences, as gender is of major influence in auditory profiling in both men (Murphy & Gates, 1997;Pearson et al., 1995) and mice (Henry, 2004;Ison, Allen, & O'Neill, 2007). ABRs to free

| Tone burst-related hearing thresholds in control, Ca v 2.3 +/− and Ca v 2.3 −/− mice
To determine potential alterations in ABR threshold levels evoked by different tone burst frequencies (1-42 kHz, Figure  5a,b), we performed repeated two-way ANOVA followed by a Tukey multiple comparisons test. Significant interaction was obtained regarding genotypes and stimulus frequencies (♀: F 14,189 = 3.478, p = .0001; ♂: F 14,161 = 2.725, p = .001), whereas there was no significant effect of the genotype on threshold levels. Multiple comparison revealed several significant alterations for individual stimulus frequencies with heterozygous Ca v 2.3 +/− mice exhibiting increased ABR thresholds compared with controls, particularly in the range of 6-18 kHz (Figure 5a,b). The percentage of mice with a detectable hearing threshold for the individual frequencies is displayed in Figure 5c,d. The binary response variable "hearing" (yes/no) was analysed with a generalized linear mixed effects model using a logit link (generalized logistic regression), accounting for fixed effects "frequency" (continuous), "group" (Ca v 2.3 +/+ , Ca v 2.3 +/− , Ca v 2.3 −/− ) and "sex" (male and female) and a random effect "animal." There was no significant gender effect (OR = 1.19; p = .6). In addition, no group-specific differences were detected for Ca v 2.

| Click-evoked ABR amplitude growth function analysis
In response to moderate to high-intense clicks, there may occur up to six ABR peaks (W I -W VI ) in mice which are assumed to be related to the following neuroanatomical structures: W I , auditory nerve (distal portion, within the inner ear); W II , cochlear nucleus (proximal portion of the auditory nerve, brainstem termination); W III , superior olivary complex (SOC); W IV , lateral lemniscus (LL); W V , termination of the lateral lemniscus (LL) within the inferior colliculus (IC) on the contralateral side; W VI , thalamus (medial geniculate body) (Kallstrand, Lewander, Baghdassarian, & Nielzen, 2014;Knipper, Dijk, Nunes, Ruttiger, & Zimmermann, 2013). Notably, the exact association of ABR-related waves II-IV and potential underlying neuroanatomical structures of the ascending auditory pathway is to some extend still a matter of debate.
In 19% of all click-evoked ABR recordings, automated wavelet analysis detected six distinct positive waves. Five distinct positive waves were observed within 45% of all clickevoked ABR recordings and a minimum of four distinct positive waves in 36% of all recordings within the first 10 ms at an SPL of 55 dB. Based on these findings, we focussed our final analysis on W I-IV .
Waves I-IV were determined based on their latencies, for example W I appeared 1.70 ± 0.16 ms and 1.58 ± 0.14 ms after the acoustic stimulus in female and male controls, respectively; W II after 2.51 ± 0.16 ms in females and 2.39 ± 0.17 ms in males; W III after 3.27 ± 0.16 ms in females and 3.16 ± 0.15 ms in males; and W IV after 4.49 ± 0.20 ms in females and 4.29 ± 0.22 ms in males at an SPL of 55 dB in Ca v 2.3 +/+ mice aged 140-142 days (see also Figure 7). ABR amplitude growth function was analysed for W I-IV , and results are depicted in Figure 6. Maximum wave F I G U R E 4 Increased ABR click-evoked hearing thresholds in female and male Ca v 2.3 +/− mice. Click-evoked hearing thresholds of female (a) and male (b) Ca v 2.3 +/+ (♀, n = 9; ♂, n = 9), Ca v 2.3 +/− (♀, n = 10; ♂, n = 9) and Ca v 2.3 −/− (♀, n = 11; ♂, n = 10) mice aged 140 -142 days. Hearing thresholds were obtained as described from raw ABR traces (see representative ABR recordings for females and males in Figure 2). Oneway ANOVA followed by Tukey multiple comparisons test revealed significant increase in hearing threshold for Ca v 2.3 +/− female (F 2,27 = 3.508, p = .04) and male (F 2,25 = 4.317; p = .02) mice compared with Ca v 2.3 +/+ female and male animals. Data are presented as scatter plots including mean ± SEM amplitudes were plotted against SPL levels tested to unravel potential alterations in wave amplitude growth function over stimulus intensity. Due to the nonexistence or rare appearance of deflections (waves) for low SPL (0-25 dB), wavelet analysis detected no or only limited confirmed accordance of waves in this SPL range. For higher SPL (30-90 dB), wavelet analysis identified mostly all waves (W I-IV ) in all experimental animals. For W I , regular two-way RM ANOVA revealed significant effects of the genotype for male mice (F 2,25 = 4.236, p < .026) and significant interaction between genotype and SPL (♀, F 24,324 = 2.417, p = .0003; ♂, F 24,300 = 3.564, p < .0001). Tukey multiple comparisons test revealed significant lower amplitudes for Ca v 2.3 +/− female mice (SPL 45, 50, 80, 90 dB) compared with Ca v 2.3 +/+ control females and significantly lower amplitude values for SPL 80-90 dB in Ca v 2.3 +/− compared with Ca v 2.3 −/− female mice (Figure 6a). Male Ca v 2.3 +/− W I amplitudes turned out to be significantly lower compared with Ca v 2.3 −/− amplitude levels between SPL 50 and 80 dB and Ca v 2.3 −/− male mice displayed a significantly higher amplitude for 60 dB SPL compared with Ca v 2.3 +/+ controls using Tukey multiple comparisons test (Figure 6b).
Amplitude growth function for W III was significantly affected by genotype (♀, F 2,27 = 8.479, p = .001; ♂, F 2,25 = 5.931, p = .008) as well as the interaction of the genotype and the stimulation SPL [dB] (♀, F 24,324 = 5.255, p < .0001; ♂, F 24,300 = 2.578, p = .0001) as determined by two-way RM ANOVA (Figure 6e,f). Ca v 2.3 +/− and Ca v 2.3 −/− female and male mice display significantly lower amplitude growth and overall amplitude levels compared with Ca v 2.3 +/+ mice in the range of 45-90 dB SPL as revealed by Tukey multiple comparisons test (Figure 6e,f). W IV two-way RM ANOVA analysis elicited a significant effect of the genotype on Ca v 2.3 male mice (F 2,25 = 3.720, p = .04, Figure 6h) and significant interaction of the genotype and stimulus SPL on Ca v 2.3 female mice (F 24,324 = 4.151, p < .0001, Figure 6g). Significant effects of the SPL on amplitude growth function of Ca v 2.3 mutant mice (both ♀ and ♂, p < .0001) were observed for all waves (W I-IV ) by two-way RM ANOVA (Figure 6a-h). Tukey multiple comparisons test for W IV amplitude shows significantly higher amplitudes between 65 and 75 dB SPL for Ca v 2.3 −/− and Ca v 2.3 +/− female mice compared with Ca v 2.3 +/+ female mice ( Figure  6g). Significantly different amplitude values were detected between Ca v 2.3 +/− and Ca v 2.3 −/− male mice between 45 and 65 dB SPL as well as 90 dB SPL using Tukey multiple comparisons test (Figure 6h).

| Click-evoked ABR waveform latency analysis
In order to investigate the role of Ca v 2.3 Ca 2+ channels on the temporal aspects of auditory information processing within the inner ear and brainstem, we analysed clickevoked wave latencies by measuring the processing time of each ABR wave (W I -W IV ). We also analysed the W I-IV interwave interval (IWI) which reflects the conduction time from cranial nerve VIII (as due to W I ) to the lateral lemniscus (W IV ) (Burkard, Eggermont, & Manuel, 2007). Latency analysis was carried out at 55 dB SPL as resultant ABRs provided best fit using the automated complex "Mexican hat"-based wavelet approach.
In addition, latency analysis was carried out for specific sensation levels, that is 10 and 20 dB above the individual hearing threshold of the experimental animals (data not shown). No statistical alterations were observed under these settings.

Ca v 2.3 mutant mice
Various VGCCs are expressed in the murine cochlea and ascending auditory pathway including the HVA Ca v 1.2 and Ca v 1.3 L-type channels, and the LVA T-type channels Ca v 3.1-3.3. qPCR was carried out to reveal potential compensatory changes in these channel entities upon monoallelic or complete Ca v 2.3 gene inactivation. Analysis in males revealed no transcriptional changes in these VGCCs in the cochlea of Ca v 2.3 +/− and Ca v 2.3 −/− mice that could be directly attributed to the observed alterations in clickand tone burst-related hearing thresholds, W I-IV amplitude growth function and W I-IV latencies (Figure 8, see also fold changes and statistics in Table S1). In females however, a significant alteration in Ca v 3.1 transcripts between Ca v 2.3 +/− and Ca v 2.3 −/− mice was detected (HT/KO fold change: −1.572, p = .03, Figure 9c, Table S1). In addition, gender differences were observed in Ca v 2.3 −/− mice for F I G U R E 6 Click-evoked ABR amplitude growth function analysis of Waves I-IV for female (left) and male (right) Ca v 2.3 mutant mice.
Wave I-IV amplitude (µV) plotted against increasing SPL (dB) for click-evoked ABR wave analysis for Ca v 2.3 +/+ (♀, n = 9; ♂, n = 9; black line representing the approximated control curve including the 95% confidence interval in grey), Ca v 2.3 +/− (♀, n = 10; ♂, n = 9, ■) and Ca v 2.3 −/− mice (♀, n = 11; ♂, n = 10, ○) aged 140-142 days. Ca v 2.3 +/− female and male mice exhibit significant delayed increase in amplitude growth as well as lower maximum amplitudes across the increasing SPL compared with Ca v 2.3 +/+ mice (a, d, e, f). Significant differences in amplitude growth and maximum amplitude were also found between Ca v 2.3 +/− and Ca v 2.3 −/− female and male mice (a, b, h). Ca v 2.3 −/− animals displayed significantly higher amplitudes compared with Ca v 2.3 +/+ (b, g) but also significantly lower amplitude (d, e, f). Data are presented as mean ± SEM. Asterisks (*) indicate significant differences between mutant mice (Ca v 2.3 +/− , Ca v 2.3 −/− ) and Ca v 2.3 +/+ control animals, and "+" icons indicate significant alterations between Ca v 2. Females Males 3 VGCCs in the inner ear, whereas amplitude alterations in W III might originate from the superior olivary complex. As latency analysis of identical sensation levels (10 and 20 dB above the individual hearing thresholds) did not reveal mouse line-specific differences, alterations, particularly in F I G U R E 7 Click-evoked ABR latency and interwave interval W I-IV analysis for female and male Ca v 2.3 +/+ , Ca v 2.3 +/− and Ca v 2.3 −/− mice.

| Paradoxic genotype-phenotype correlation in Ca v 2.3 +/− and Ca v 2.3 −/− mice
In our study, we did not observe a typical gene dose-dependent auditory phenotype in Ca v 2.3 +/− and Ca v 2.3 −/− mice. There is often a strong bias in statistics on genotype-phenotype correlation in genetically modified mice due to variable depth of scientific investigation, potential publication restrictions of negative results, etc. (Barbaric, Miller, & Dear, 2007). In about 10%-15% of knockouts generated so far, no overt phenotype could be detected and mutant mice do not seem to exhibit pathophysiological alterations, although one might have expected a severe phenotype based on the reported function of the gene and its expression pattern (Barbaric et al., 2007). In terms of auditory profile, Ca v 2.3 −/− mice seem to exhibit strong phenotypic and genetic robustness which could be due to compensatory alterations in transcriptional profiles affecting ion channel physiology, signal transduction cascades, and neuronal degeneration and apoptosis, and which might counteract the deletion of the cacna1e target gene. The mechanisms of such robustness could be dichotomous, that is, by activation of alternative pathways for auditory processing (genetic buffering), or by functional complementation, in which genes are redundant in function to a variable extent (Gu et al., 2003). A lack of a prominent knockout phenotype, as observed in the Ca v 2.3 −/− auditory profile, could be related to paralogous genetic redundancy (Barbaric et al., 2007) in a complete or partial fashion (Thomas, 1993 in the Organ of Corti (Waka et al., 2003), spiral ganglion neurons (SGNs) (Peng et al., 2004), the cochlear nucleus (Bal & Oertel, 2007;Kim & Trussell, 2007;Parajuli et al., 2012), the pontine nuclei, inferior olive, lateral superior olive and the nucleus of the solitary tract (Parajuli et al., 2012;Soong et al., 1993;Williams et al., 1994). Besides Ca v 2.3, numerous electrophysiological studies already suggested an important role of low-to mid voltage-activated Ca 2+ currents in these structures, including Ca v 1.3 L-type and Ca v 3 T-type VGCCs: Analysis of Ca v 1.3 −/− mice revealed cardiac arrhythmia and deafness (Platzer et al., 2000), secretory and developmental deficits in IHCs and OHCs and alterations in the functional interference with an armamentarium of other voltage-and ligand-gated ion channels, for example, Ca 2+ -activated K + channels (BK, SK), acetylcholine receptors (AChR), Ca v 1.2 L-type, Ca v 2.1 P/Q and Ca v 2.2 N-type VGCCs (Beutner, Voets, Neher, & Moser, 2001;Frank, Khimich, Neef, & Moser, 2009;Glueckert et al., 2003;Goutman & Glowatzki, 2007;Johnson & Marcotti, 2008;Johnson, Marcotti, & Kros, 2005;Kim, Li, & von Gersdorff, 2013;Marcotti, Johnson, Holley, & Kros, 2003;Michna et al., 2003;Moser & Beutner, 2000;Nemzou, Bulankina, Khimich, Giese, & Moser, 2006;Zorrilla de San, Pyott, Ballestero, & Katz, 2010). Electrophysiologically, Ca v 1.3 VGCCs were proven to exhibit low-to mid-voltage-activated kinetics in hair cells (Inagaki & Lee, 2013;Zampini et al., 2010). In addition, classical LVA T-type Ca 2+ channels, such as Ca v 3.1 and Ca v 3.2, were reported to play an important role in auditory information processing as well (Inagaki et al., 2008;Lei et al., 2011;Lundt, Seidel, et al., 2019;Nie et al., 2008;Shen et al., 2007). Ca v 2.3 VGCCs have exceptional electrophysiological characteristics (Soong et al., 1993;Weiergraber, Kamp, et al., 2006b;Williams et al., 1994) and have attracted specific attention due to their functional involvement in neurotransmitter release (Gasparini, Kasyanov, Pietrobon, Voronin, & Cherubini, 2001;Wu, Westenbroek, Borst, Catterall, & Sakmann, 1999) and synaptic plasticity (Yasuda, Sabatini, & Svoboda, 2003). Thus, the functional implications of Ca v 2.3 VGCC in the auditory system are complex. In cellular electrophysiology, Ca v 2.3 Ca 2+ channels can serve as sophisticated tuning elements, acting as low-to mid voltage-activated ion channels capable of triggering or regulating complex cellular firing patterns. The latter includes transition of tonic firing to oscillatory burst like activity and vice versa or modulation of neuronal afterhyperpolarization (Shcheglovitov et al., 2012;Weiergraber, Kamp, et al., 2006b). Both simple and complex action potential (spike) patterns and afterhyperpolarizations in auditory structures require Ca v 3 T-type and Ca v 2.3 R-type Ca 2+ channels in addition to BK and SK channels (Kim & Trussell, 2007). For example, the firing rate of principal neurons in the LSO is a linear function of differences in interaural sound intensity. It has been hypothesized that this linear response results from the functional integration of excitatory ipsilateral and inhibitory contralateral inputs. In the LSO, Ca v 3.2 and Ca v 2.3 VGCCs were detected and reported to be highly sensitive to Ni 2+ (Kang et al., 2006) and both might contribute to the complex firing pattern of LSO cells (Jurkovicova-Tarabova et al., 2012). Importantly, Ca v 2.3 seems to partially compensate Ca v 1.3 ablation in LSO neurons (Jurkovicova-Tarabova et al., 2012). Ca v 1.3 VGCCs, which are known to be of central importance in IHCs, display fundamental electrophysiological properties similar to those of typical Ca v 3 LVA channels, such as rapid activation kinetics (Inagaki & Lee, 2013;Koschak et al., 2001;Xu & Lipscombe, 2001;Zampini et al., 2013Zampini et al., , 2010. The latter are relevant for the temporal characteristics of sound coding and the ability to accurately trigger auditory nerve firing to reflect sound frequency in terms of phase locking. In immature IHCs, Ca v 1.3 VGCCs activate at relatively negative potentials (~-70 mV) which is a basic electrophysiological property of LVA channels as well (Koschak et al., 2001;Xu & Lipscombe, 2001;Zampini et al., 2010). Therefore, given a resting membrane potential (RMP) of ~-60 mV in these cells (Marcotti et al., 2003), Ca v 1.3 and Ca v 2.3 Ca 2+ channels may support tonic neurotransmitter release at rest and effectively link increased sound pressure levels with higher rates of transmitter release. Mechanistically, Ca v 2.3 might contribute to these processes by involvement in the complex spatiotemporal interdependence of intracellular Ca 2+ levels and Ca 2+ -activated K + currents in HCs in membranaceus nanodomains (Bloodgood & Sabatini, 2007Joiner & Lee, 2015;Zaman et al., 2011). In the SGN, inhibition of Ca 2+ currents resulted in attenuated spontaneous activity and different subtypes of Ca 2+ currents activated resting outward conductances. Consequently, blockage of these Ca 2+ currents caused depolarization of the RMP (Lv et al., 2012;Peng et al., 2004). Similarly, in glycinergic interneurons (Cartwheel cells) of the dorsal cochlear nucleus, early complex spike firing patterns were based on Ca v 2.3 R-type Ca 2+ channels together with BK and SK channels (Kim & Trussell, 2007).
Recent expression studies and electrophysiological analysis carried out by Chen et al. (2011) elicited that VGCCs are relevant for neuronal responsiveness in both the highand low-frequency ranges. This tonotopic specialization is characterized by neurons with rapid kinetic features coding for high-frequency auditory signals and other neurons with slower kinetic features coding for low-frequency auditory signals. Developmental variations in activation and inactivation kinetics along the tonotopic axis enable VGCCs to shape the firing pattern and modulate the unique functional specialization of auditory neurons . Several VGCCs are expressed in the inner ear and auditory tract. However, Ca v 2.3 exhibits the most heterogenous and extraordinary functional expression compared to all other VGCCs . Besides expression in SGNs, Ca v 2.3 VGCCs were also detected in satellite cells, putative myelinating Schwann cells and compact myelin  and the density of Ca v 2.3 expression was highly variable in these structures. Thus, functional integration of Ca v 2.3 Ca 2+ channels regarding the tonotopic specialization and action potential propagation is most complex and potentially much more sophisticated than for any other VGCC reported so far. Importantly, Ca v 3.1 VGCCs exhibited a similar expression compared to Ca v 2.3 channels . This is of high relevance as our qPCR results suggest reduced Ca v 3.1 transcript levels in heterozygous Ca v 2.3 +/− mice, whereas Ca v 3.1 levels in Ca v 2.3 −/− female animals remain normal. This points to a potential compensatory mechanism in knockout mice that is not effective in Ca v 2.3 +/− animals.
Finally, our observations of unaltered click-evoked hearing thresholds and increased percentage of hearing animals in Ca v 2.3-deficient mice could also indicate an overlapping effect of Ca v 2.3 ablation on both functional auditory information processing on the one hand and neurodegenerative processes on the other hand. Ca v 2.3 VGCCs are involved in excitotoxicity and neurodegeneration, and Ca v 2.3-mediated Ca 2+ influx can trigger neuronal cell death under specific circumstances (Suzuki et al., 2004;Weiergraber et al., 2007). While ablation of Ca v 2.3 might thus be critical for proper HC function and synaptic processing in the auditory tract, its ablation might be preservative or neuro-/otoprotective in terms of age-related degeneration of HCs and further structures of the auditory tract and underlines the Janus-like behaviour of Ca v 2.3.

| Perspectives
Future qualitative and quantitative immunohistochemical studies on cochlear hair cells and SGN could prove a potential otoprotective effect of Ca v 2.3 ablation in the auditory tract. Assuming that both neuroprotective effects and age-related hearing loss are negligible at early age, ABR studies in young mutant mice might help to further disentangle the complex functional properties of Ca v 2.3 in the auditory tract. Finally, cellular electrophysiology will be necessary to characterize the exact cellular mechanistic role of Ca v 2.3 VGCCs in the physiology and pathophysiology of the inner ear and peripheral auditory tract. Given the complex findings presented here, Ca v 2.3 VGCCs might serve as an important candidate for pharmaceutical interference in the auditory tract in the future.