All the experiments described in the study were performed in P21–P28 mice of either sex. Vglut3^–/– mice have been described before (Ruel et al., 2008) and the generation of Otof^TDA/TDA will be described in a manuscript (Chen et al.) that is currently under peer review elsewhere. In brief, a CRISPR/Cas9 approach was employed and ribonucleoprotein particles were injected into C57Bl6N oocytes. Correct editing of Otof was validated in founder (F0) mice, which were then crossed with C57Bl6N mice. Germline transmission was confirmed by heterozygosity for the edited allele. F2 mice were born at a Mendelian ratio, and heterozygous breeding was used to generate Otof^TDA/TDA and WT littermates for experiments. Otof^TDA/TDA were found to be deaf by recordings of auditory brainstem responses but seemed otherwise fine according to routine observation. Breeding was done in compliance with the German national animal care guidelines and approved by the local animal welfare committee of the University Medical Center Göttingen and the Max Planck Institute for Multidisciplinary Sciences, as well as the animal welfare office of the state of Lower Saxony, Germany (LAVES, AZ 19/3134).

We dissected the apical coils of the organs of Corti, and patch-clamped IHCs at the tonotopic location of around 6–8 kHz from either pillar or modiolar side. Patch pipettes (Science products, GB150F-8P) were pulled with a P-97 Flaming/Brown micropipette puller (Sutter Instruments). Intracellular solution contained (in mM): 111 Cs-glutamate, 1 MgCl₂, 1 CaCl₂, 10 EGTA, 13 TEA-Cl, 20 HEPES, 4 Mg-ATP, 0.3 Na-GTP, 1 L-glutathione, 0.8 Fluo-4FF (Life Technologies), and 0.01 TAMRA-conjugated CtBP2/RIBEYE binding peptide, pH 7.3, 290 mOsm. The extracellular solution contained the following (in mM): 2.8 KCl, 105 NaCl, 10 HEPES, 1 CsCl₂, 1 MgCl₂, 5 CaCl₂, 35 TEA-Cl, and 2 mg/ml D-glucose, pH 7.2, 300 mOsm. Recordings were performed at room temperature (22–23°C).

Data were acquired with an EPC-10 amplifier controlled by PatchMaster software (HEKA Elektronik). The holding potential of the cell was set to −87 mV. Whole cell Ca²⁺ influx was triggered by applying either voltage step or voltage ramp depolarizations to the cell. Voltage step depolarizations were applied from −82 to 63 mV with 5 mV increments and voltage ramp depolarizations ranged from −87 to 63 mV over the course of 150 ms (1 mV/ms). Recordings were leak corrected using P/n protocol. All voltages were corrected offline for voltage drops across series resistance (R_s) and liquid junction potential, which was calculated to be 17 mV. Recordings were discarded from analysis in case they displayed leak currents beyond −50 pA at the holding potential (−87 mV), R_s above 14 MOhm during the first 3 min after rupturing the cell, or Ca²⁺ current rundown above 25% by the end of the experiment.

Ca²⁺ imaging from IHCs was described previously (Ohn et al., 2016; Cantu-Guerra et al., 2023). Briefly, a spinning disk confocal scanner (CSU22, Yokogawa) mounted on an upright microscope (Axio Examiner, Zeiss) and a 1.0 NA 63x objective (W Plan-Apochromat, Zeiss) were used. Image acquisition was performed with a sCMOS camera (Neo, Andor). The spinning disk speed was set to 2000 rpm.

Z-stacks during the live imaging were acquired using a fast piezoelectric system (Piezosystem). To obtain the morphology of the IHC and identify the positions of the AZs, we first acquired a Z-stack of the cell through imaging TAMRA fluorescence using a 591 nm laser (Cobolt AB). Each plane was exposed for 0.5 s and the step size for Z-scanning was 0.5 μm. Ca²⁺ imaging was performed by exciting Fluo-4FF with 491 nm laser (Cobolt AB). Ca²⁺ imaging was restricted to the planes which contained ribbons. Hotspots of Fluo-4FF fluorescence increases were evoked by voltage ramp depolarizations and simultaneously imaged at a frame rate of 100 Hz at each plane. For each plane, two different voltage ramps were applied, one being 5 ms shifted relative to the other one. This served the purpose of increasing the voltage resolution of Ca²⁺ imaging when later concatenating the two fluorescence traces evoked by the two different voltage ramps.

The samples were fixed in 4% formaldehyde on ice. Time of fixation was 10 min whenever anti-Ca_v1.3 antibody was used; for the rest of the stainings, the samples were fixed for 60 min. The following primary antibodies were used: mouse anti-Ctbp2 (1:200, BD Biosciences, 612044), rabbit anti-Cav1.3 (1:100, Almone Labs, ACC005), and rabbit anti-Myosin 7a (1:200, Enzo Life Sciences, PTS-25-6790-C050). Afterward, the following secondary antibodies were added: Alexa Fluor 488 conjugated anti-rabbit (1:200, Invitrogen, A11008) and Alexa Fluor 633 conjugated anti-mouse (1:500, Invitrogen, A21126). For superresolution STED imaging STAR580 conjugated anti-mouse (1:200, Abberior, 2-0002-005-1) and STAR635 (1:200, Abberior, 2-0012-007-2) conjugated anti-rabbit secondary antibodies were used. In order to prevent the hair cells from flattening due to the weight of the coverslip, 50 μm plastic films were taped between the coverslip and the microscopy slide from both sides of the sample. Imaging was performed in confocal or 2D STED mode using the Abberior Instruments Expert line STED microscope equipped with 488 nm, 561 nm, and 633 nm lasers and a 775 nm STED laser. A 1.4 NA 100x oil immersion objective was used.

Transcriptomic studies of single SGNs in Vglut3^–/– mice revealed an altered molecular subtype specification of SGNs: 80% of the neurons gained type I_a identity, while the proportion of type I_b and I_c neurons was down to 20% (Shrestha et al., 2018; Sun et al., 2018). This has been taken to indicate that postnatal glutamatergic transmission at the afferent IHC synapse is required to maintain the molecular identity of type I_b and I_c SGNs. Here, we asked the question if changes in synaptic transmission and/or altered transsynaptic signaling from SGNs affect the presynaptic heterogeneity of IHCs. To do so, we assessed the functional heterogeneity of IHC AZs by combining IHC patch-clamp and Ca²⁺ imaging at IHC AZs in Vglut3^–/– mice at 3 weeks of age. Whole cell Ca²⁺ currents tended to be slightly larger in Vglut3-deficient IHCs without reaching significance (Vglut3^+/+: −161 ± 7.87 pA, SD = 40.12 pA, n = 27, N = 16 vs. Vglut3^–/–: −170 ± 9.79 pA, SD = 48.96 pA, n = 25, N = 12; Student’s t-test, p = 0.49, Figures 1A, C), while a previous study showed a significantly increased Ca²⁺ influx in Vglut3-deficient IHCs (Ruel et al., 2008). A potential explanation for this discrepancy is the age difference of mice. While an increase in the Ca²⁺ current amplitude was previously reported for p12–p18 mice, we used 3-week-old mice in our study, at which stage a reduced synapse number might mask the increased Ca²⁺ current. The voltage dependence of Ca²⁺ channel activation and its voltage sensitivity, approximated as the slope of the Boltzmann function fit to the fractional activation curve, were not significantly altered, but there was a trend toward activation at lower potentials for Vglut3-deficient IHCs (Figures 1B, D, E).

Next, we imaged synaptic Ca²⁺ signals at single IHC AZs using spinning disk confocal microscopy to analyze the heterogeneity of Ca²⁺ influx among the AZs (Ohn et al., 2016; Jean et al., 2019; Sherrill et al., 2019). IHCs were loaded with a TAMRA-conjugated Ctbp2-binding dimeric peptide that fluorescently labels synaptic ribbons (Zenisek et al., 2004) and the Ca²⁺ indicator Fluo-4FF (K_D ∼10 μM). Scanning the IHC from the basal end to the cuticular plate using a 561 nm laser obtained the IHC morphology and detected the ribbon containing confocal planes, at which we next imaged (491 nm laser) the synaptic Ca²⁺ signals upon voltage ramp depolarization. Employing strong intracellular Ca²⁺ buffering (10 mM EGTA), we interpret the Fluo-4FF fluorescence increase (ΔF/F₀) to approximate the Ca²⁺ influx at the single AZ (Frank et al., 2009). The maximal Ca²⁺ influx at single AZs (ΔF/F_0max) of Vglut3^–/– mice tended to be slightly higher compared in IHCs of Vglut3^+/+ mice without reaching statistical significance (Figures 2A, B; Vglut3^+/+: 1.25 ± 0.083, SD = 0.94, 129 spots, n = 27, N = 16; Vglut3^–/–: 1.45 ± 0.12, SD = 1.18, 98 spots, n = 20, N = 12; Mann-Whitney-Wilcoxon test, p = 0.22; Figures 2C, C_i).

Using 3D reconstruction of the cells we assigned the AZ properties to their positions (see section “2. Materials and methods”). We found mild but significant modiolar-pillar gradient of maximal amplitude of synaptic Ca²⁺ influx (i.e., stronger AZs on the modiolar side of the cell in terms of Ca^{2 +} influx amplitude) in both WT and Vglut3^–/– IHCs (Figure 2D, Supplementary Table 1; Vglut3^+/+: Pearson’s correlation coefficient r = 0.34; Vglut3^–/–: r = 0.21). Binary separation into pillar and modiolar AZs showed a significant difference in the maximum Ca²⁺ influx between pillar and modiolar AZs in Vglut3^–/– IHCs similar to the controls (Vglut3^+/+: pillar: 0.91 ± 0.056, SD = 0.36, 40 AZs; modiolar: 1.4 ± 0.11, SD = 1.08, 89 AZs; Mann-Whitney-Wilcoxon test, p = 0.01; Vglut3^–/–: pillar: 0.98 ± 0.13, SD = 0.64, 24 AZs; modiolar: 1.61 ± 0.16, SD = 1.28, 74 AZs; Mann-Whitney-Wilcoxon test, p = 0.004; Figure 2E).

To test whether the difference in ΔF/F_0max between pillar and modiolar AZs in Vglut3^–/– IHCs was dominated by modiolar AZs with high fluorescence (“winner” AZ, Ohn et al., 2016), we analyzed modiolar-pillar gradient of maximum Ca²⁺ influx without the “winners”. As done before (Ohn et al., 2016), we defined “winner” AZs to have ΔF/F_0max values at least 2.5 higher than the average ΔF/F_0max of the rest (Supplementary Figures 1A, B). Interestingly, the modiolar-pillar gradient of maximum Ca²⁺ influx in Vglut3^–/– IHCs remained without the “winner” spots (Supplementary Figures 1C, D).

Next, we analyzed the voltage dependence of Ca²⁺ channel activation at individual AZs by calculating the fractional activation curves from Ca²⁺ fluorescence–voltage relationships (see section “2. Materials and methods,” Figure 3A). We observed a significant hyperpolarized shift of the voltage of half-maximal Ca²⁺ channel activation (V_h) at the single AZ level in IHCs of Vglut3^–/– mice (Vglut3^+/+: −26.7 ± 0.61 mV, SD = 6.96 mV, 129 AZs, n = 27, N = 16; Vglut3^–/–: −28.7 ± 0.71 mV, SD = 6.7 mV, 89 AZs, n = 20, N = 12; Student’s t-test, p = 0.03; Figures 3B, C) but no change in voltage sensitivity of Ca^{2 +} influx (Vglut3^+/+: 6.77 ± 0.15 mV, SD = 1.74 mV, 129 spots, n = 27, N = 16; Vglut3^–/–: 6.56 ± 0.2 mV, SD = 1.92 mV, 89 spots, n = 25, N = 12; Mann-Whitney-Wilcoxon test, p = 0.26, Figure 3D). Interestingly, a similar hyperpolarized shift was previously observed in Pou4F1 cKO mice (Sherrill et al., 2019). Despite the average hyperpolarized shift, the pillar AZs of Vglut3^–/– IHCs showed a significantly lower V_h than the modiolar AZs, (Vglut3^+/+: pillar: −29.1 ± 1.2 mV, SD = 7.56 mV, 40 AZs; modiolar: −25.6 ± 0.11 mV, SD = 0.68 mV, 89 AZs; Student’s t-test, p = 0.01; Vglut3^–/–: pillar: −32.1 ± 1.26 mV, SD = 5.9 mV, 22 AZs; modiolar: −27.6 ± 0.81 mV, SD = 6.6 mV, 67 AZs; Mann-Whitney-Wilcoxon test, p = 0.005, Figures 3E, F). Such a pillar-modiolar gradient of voltage-dependent AZ activation (i.e., stronger AZs on the pillar side of the cell in terms of V_h) was previously demonstrated for WT IHCs (Ohn et al., 2016; Özçete and Moser, 2021). Correlation analysis of V_h, ΔF/F_0max and synapse position revealed weak but significant positive correlations (Supplementary Table 1). These are consistent with the notion that modiolar synapses of Vglut3^+/+ and Vglut3^–/– IHCs have stronger maximal synaptic Ca^{2 +} influx (larger ΔF/F_0max) yet activating at more depolarized potentials (larger V_h). Together, these results indicate mild shift of AZ activation to lower voltages but intact spatial gradients of the maximal amplitude and the voltage of half maximal activation of synaptic Ca²⁺ influx in Vglut3^–/– IHCs.

While the heterogeneity of Ca²⁺ influx among IHCs was largely preserved in Vglut3^–/– IHCs, diversity of AZ morphology could still be affected. To test this, we first performed 2D STED nanoscopy at AZs which were immunolabeling against Ctbp2/RIBEYE and Ca_v1.3 to examine the morphology of ribbons and Ca_V1.3 clusters (Krinner et al., 2017; Jean et al., 2018, 2019; Neef et al., 2018, Figure 4A). First, we fitted Ctbp2 immunofluorescent spots with a 2D Gaussian function and observed increased full widths at half maxima (FWHM) of both long and short axes in Vglut3^–/– IHCs (Figures 4B, C). This abnormal enlargement or elongation of the ribbons in IHCs of knockout or mutated Vglut3 mice has been observed in previous studies (Seal et al., 2008; Kim et al., 2019; Joshi et al., 2021). Vglut3^+/− animals did not show significant difference in FWHM of long axis of Ctbp2 immunofluorescent spots (Vglut3^+/+: 378 ± 5.7 nm, SD = 55 nm, n = 92, N = 2 vs Vglut3^+/−: 380 ± 9.3 nm, SD = 62 nm, n = 44, N = 3, Student’s t-test, p = 0.87). We found a mild but significant increase in the FWHM of short axis in Vglut3^+/− animals (Vglut3^+/+: 250 ± 3.3 nm, SD = 32 nm, n = 92, N = 2 vs Vglut3^+/−: 261 ± 4.6 nm, SD = 31 nm, n = 44, N = 3, Mann-Whitney-Wilcoxon test, p = 0.015). In total, the area of the ribbons (pi*(FWHM_long_axis/2) *(FWHM_short_axis)) in Vglut3^+/− animals was not significantly changed, consistent with the previous report (Kim et al., 2019). Next, we classified the different morphologies of Ca_v1.3 clusters (Krinner et al., 2017; Jean et al., 2018, 2019; Neef et al., 2018). We did not observe any abnormalities in the proportions of the cluster types in Vglut3^–/– IHCs: the majority of the Ca_v1.3 clusters were line- (around 80%) or spot-like (over 10%) in both WT and Vglut3^–/– IHCs (Figure 4G). By fitting the 2D Gaussian function to the population of line-like Ca_v1.3 clusters of WT and Vglut3^–/– IHCs, we saw a reduction in FWHM of both long and short axes of IHC Ca_v1.3 clusters in Vglut^–/– mice (Figures 4D, F). However, this was not reflected in a reduction of amplitude of the 2D Gaussian fit: instead, Ca_v1.3 clusters of Vglut3^–/– IHCs showed a trend toward an increased amplitude of the 2D Gaussian fit that did not reach significance (Figure 4E). A tighter spatial confinement of a greater number of Ca_v1.3 channels at AZs of Vglut3^–/– IHCs could offer a potential explanation for our findings from synaptic Ca²⁺ imaging and STED nanoscopy.

Next, we tested for spatial gradients of ribbon size, approximated as immunofluorescent intensity in confocal stacks of IHCs (Ohn et al., 2016; Jean et al., 2019; Sherrill et al., 2019) immunostained for Myo7a and Ctbp2/RIBEYE (Figure 5B). For these experiments, we used Vglut3^+/– mice as controls. Overall, we found that the modiolar-pillar gradient of ribbon size (i.e., larger AZs on the modiolar side of the cell in terms of ribbon size) is maintained in Vglut3^–/– IHCs (Vglut3^+/–: pillar: 0.86 ± 0.02, SD = 0.22, 100 spots; modiolar: 0.99 ± 0.02, SD = 0.24, 114 spots; Mann-Whitney-Wilcoxon test, p < 0.0001; Vglut3^–/–: pillar: 0.81 ± 0.03, SD = 0.38, 150 spots; modiolar: 1.1 ± 0.04, SD = 0.51, 209 spots; Mann-Whitney-Wilcoxon test, p < 0.0001; Figures 5A, C–F).

Our analysis of Vglut3^–/– mice showed intact modiolar-pillar gradients of maximal synaptic Ca²⁺ influx and ribbon size in IHCs. Moreover, despite a mild overall hyperpolarized shift of the voltage of half-maximal Ca²⁺ channel activation, its pillar-modiolar gradient was preserved in Vglut3^–/– IHCs. This argues against a strict requirement of glutamatergic signaling for establishing or maintaining IHC synaptic heterogeneity. To test the impact of evoked exocytosis on the AZ heterogeneity and the spatial gradients of AZ properties, we chose to study a novel Otof mutant mouse. These mice harbor a triple aspartate to alanine mutation in the C₂E domain of otoferlin Otof^TDA/TDA —a protein that has been shown to be vital for IHC vesicle fusion (Roux et al., 2006; Pangrsic et al., 2010; Michalski et al., 2017). Exocytosis in IHCs of Otof^TDA/TDA is largely abolished, but, unlike in other Otof mutants (e.g. 50% synapse loss in IHCs of Otof^–/– mice, Roux et al., 2006; Stalmann et al., 2021) the number of the IHC afferent synapses was preserved in 2–3-week-old mice (Chen et al.).

Overall, Otof^TDA/TDA mice allow us to analyze the presynaptic heterogeneity in IHCs largely lacking exocytosis while avoiding any sampling bias due to synaptic degeneration. We performed patch-clamp combined with single AZ Ca²⁺ imaging from the IHCs of P21-P28 Otof^TDA/TDA mice similar to the Vglut3^–/– mice. We observed a non-significant trend toward lower maximal amplitude of synaptic Ca²⁺ influx (WT: 1.22 ± 0.079, SD = 0.92, 138 spots, n = 30, N = 17; Otof^TDA/TDA: 1.00 ± 0.048, SD = 0.53, 123 spots, n = 13, N = 7; Mann-Whitney-Wilcoxon test, p = 0.07; Figures 6A, A_i). The modiolar-pillar gradient of the maximum synaptic Ca²⁺ influx was not affected (pillar: 0.84 ± 0.05, SD = 0.33, 41 spots; modiolar: 1.08 ± 0.07, SD = 0.6, 82 spots; Mann-Whitney-Wilcoxon test, p = 0.04; Figure 6C). Analysis of the voltage dependence of Ca²⁺ channel activation revealed a significant hyperpolarized shift of V_h in Otof^TDA/TDA mice (WT: −26.8 ± 0.49 mV, SD = 5.7 mV, 138 spots, n = 30, N = 17; Otof^TDA/TDA: −30 ± 0.47 mV, SD = 4.78 mV, 102 spots, n = 13, N = 7; Student’s t-test, p < 0.0001; Figures 6B, B_i) but no change was observed in the voltage sensitivity of channel activation (WT: 6.77 ± 0.15 mV, SD = 1.75 mV, 138 spots, n = 30, N = 17; Otof^TDA/TDA: 6.32 ± 0.19 mV, SD = 1.87 mV, 102 spots, n = 13, N = 7; Student’s t-test, p = 0.06; Figure 6B_ii) but no change was observed in the voltage sensitivity of channel activation (WT: 6.77 ± 0.15 mV, SD = 1.75 mV, 138 spots, n = 30, N = 17; Otof^TDA/TDA: 6.32 ± 0.19 mV, SD = 1.87 mV, 102 spots, n = 13, N = 7; Student’s t-test, p = 0.06; Figure 6B_ii). Despite this shift, the pillar-modiolar gradient of V_h was preserved, with pillar AZs activating at lower voltage than modiolar AZs (pillar: −31.8 ± 0.75 mV, SD = 4.38 mV, 34 spots; modiolar: −29.5 ± 0.58 mV, SD = 4.81 mV, 68 spots; Mann-Whitney-Wilcoxon test, p = 0.01; Figure 6D). Interestingly, the distributions of V_h in WT and Otof^TDA/TDA IHCs differ also in the variance (Supplementary Figure 2). Interquartile ranges of the two V_h distributions suggest narrower spread of V_h in Otof^TDA/TDA mice (WT: 8.14 mV vs. Otof^TDA/TDA: 5.38 mV). Consistent with that observation, Levene’s test showed a significant difference between the two distributions, indicating inequality of variances.

We note that all the experiments in this study have been performed at the apical turn of the organ of Corti around the 6–8 kHz region. Therefore, we cannot exclude the possibility that larger effects of impaired transsynaptic signaling could be observed at other tonotopic positions. Furthermore, despite the preserved gradients of IHC AZ properties, Vglut3^–/– animals display abnormal enlargement of synaptic ribbons. These results are inconsistent with the findings in Pou4f1 and Runx1 KOs, where deletion of type I_b and I_c SGN molecular markers led to reduction of average ribbon size. Thus, there is the possibility that direct presynaptic alterations upon Vglut3 deletion mask or overrule potential effects of the altered SGN molecular subtype specification.