According to the World Health Organization, ∼466 M people worldwide are affected by disabling hearing loss, which can have a big functional, emotional and socio-economic impact on their life, and society. Besides factors like aging or noise, a large proportion of hearing impairment is caused by genetic factors, and has mostly monogenic origins (reviewed in Korver et al., 2017; Wrobel et al., 2021). At present, the treatment of sensorineural hearing loss is limited to the use of conventional hearing aids or, in cases of severe and profound deafness, cochlear implants. Despite their phenomenal success, cochlear implants have their drawbacks. The spread of electric current limits their frequency resolution, which together with a low dynamic range hampers music perception and speech intelligibility in noisy environment. Furthermore, cochlear implants can typically not prevent or stop degeneration of spiral ganglion neurons (SGNs) that is often observed in malfunctioning ears. Thus, novel methods for improved hearing restoration are currently being developed, including gene replacement, and editing or silencing as well as optogenetics (reviewed in Kleinlogel et al., 2020). In the recent years, the potential of a therapeutic approach using AAV-mediated gene transfer has been tested in a few animal deafness models (e.g., Akil et al., 2012, 2019; Jung et al., 2015; Pan et al., 2017; Al-Moyed et al., 2019; Nist-Lund et al., 2019; Rankovic et al., 2021; Taiber et al., 2021).

So far, to our knowledge, no gene-therapy study has targeted the Cabp2 gene, which encodes calcium binding protein 2, a potent modulator of voltage-gated Ca_V1.3 channels in inner hair cells (IHCs). Pathologic mutations in CABP2 have been identified in Iranian, Turkish, Italian, Czech, and Pakistani families (Tabatabaiefar et al., 2011; Schrauwen et al., 2012; Bademci et al., 2016; Marková et al., 2016; Picher et al., 2017; Koohiyan et al., 2019; Park et al., 2020; Safka Brozkova et al., 2020) that lead to autosomal recessive moderate to severe non-syndromic hearing impairment DFNB93. Audiological assessment in those patients revealed an auditory synaptopathy phenotype with often intact otoacoustic emissions signifying outer hair cell (OHC) function being present at least in some patients (Picher et al., 2017). The underlying disease mechanism was recently studied in a knock-out mouse model that suggested enhanced steady-state inactivation of Ca_V1.3 channels at the IHC ribbon synapse (Picher et al., 2017). Consequently, fewer Ca_V1.3 channels are available to trigger exocytosis and support IHC neurotransmission, which results in lower temporal precision of sound encoding, and elevated hearing thresholds (Picher et al., 2017; Yang et al., 2018). In contrast to several other deafness mouse models, hair cells of these mice appear to have normal mechanotransduction and synaptic development and show no signs of early secondary degeneration (Picher et al., 2017; the current study; but see Yang et al., 2018), thus suggesting an extended window for gene therapy interventions.

In this study, we tested the potency of two AAV vectors [AAV2/1 and AAV-PHP.eB, abbreviated to PHP.eB in the following, (Chan et al., 2017)] to rescue hearing in Cabp2^–/– mice. AAV containing the coding sequence of wild-type Cabp2 was injected through a round window into the cochlea of early postnatal mice. Gene transfer by either AAV capsid improved hearing in a majority of 5–8-week-old Cabp2^–/– animals. Furthermore, AAV-mediated expression of Cabp2 restored the typical “non-inactivating” IHC Ca_V1.3 current properties. This likely underlies partial rescue of hearing and further supports the hypothesis of increased calcium channel inactivation as the key disease mechanism of DFNB93. Collectively, our data suggests that AAV-mediated gene therapy might be a well-suited option to treat DFNB93-related hearing impairment. Normal hair cell development and retained overall hair cell morphology in DFNB93 underline the potential of this approach, and make it particularly interesting for translational approaches in humans.

Hearing restoration using gene therapy offers great hope for causative therapy of genetic forms of severe hearing impairment. Here, we asked whether AAV-mediated gene therapy could be applicable in a knock-out mouse model of non-syndromic hearing impairment DFNB93 (Picher et al., 2017), characterized by moderate to severe hearing loss in patients (Tabatabaiefar et al., 2011; Schrauwen et al., 2012; Bademci et al., 2016; Marková et al., 2016; Picher et al., 2017; Koohiyan et al., 2019; Park et al., 2020; Safka Brozkova et al., 2020). Plasmids encoding for bicistronic expression of Cabp2 and eGFP via the P2A peptide driven by the hCMV/hBA promoter and followed by the WPRE and bGH posttranscriptional elements were packaged into AAV2/1, and PHP.eB capsids (Figure 1A and Supplementary Figure 1). Mice received unilateral round window injections at the age of 5–7 days after birth and were tested for hearing, hair cell physiology or cochlear morphology 4–7 weeks afterward (Figure 1B).

We first tested the potency of the AAV2/1 vector carrying the Cabp2 coding sequence (titer: 2.0 × 10¹³ genome copies/ml) for restoration of hearing in Cabp2^–/– mice. Examination of native eGFP fluorescence in acute cochlear preparations or eGFP immunofluorescence in fixed organs of Corti revealed high transduction efficiency of the AAV2/1 vector (Figures 1C,D and Supplementary Figure 2A) with the large majority of IHCs being eGFP positive. Qualitative inspection of the injected cochleae showed no malformations nor hair cell degeneration, suggesting no gross adverse effects of the viral gene delivery. To evaluate the extent of hearing restoration, we examined auditory brainstem responses (ABRs) to tone bursts and clicks in 5–8-week-old mice (Figure 1E). Hearing thresholds of the injected and non-injected contralateral ears were compared using two-way repeated measures (RM) ANOVA and Šidák’s multiple comparisons test. Over all, the injected ears showed lower thresholds for tone burst frequencies (p = 0.002-0.03; N = 24 animals). In 67% of the AAV2/1-Cabp2-injected animals (16 out of 24 animals), the improvement reached 20 dB SPL or more at least for one of the tested frequencies (p < 0.0001 for each tone burst frequency; N = 16). Moreover, injected ears showed increased amplitudes of click-evoked ABRs (Figures 1F–H and Supplementary Figure 2; RM ANOVA and Šidák’s multiple comparisons test; p < 0.001 for summated as well as individual ABR wave I-IV amplitudes; wave I amplitudes of 1.3 ± 0.2 vs. 0.8 ± 0.1 μV in injected vs. contralateral non-injected Cabp2^–/– ears at 80 dB SPL). Improvement of hearing at least at lower frequencies was also observed when compared to control, non-injected Cabp2^–/– animals, where any potential spread of virus from contralateral injected ears could be excluded (Supplementary Figure 2B; two-way ANOVA, and Šidák’s multiple comparisons test; p < 0.03 and 0.001 for 6 and 12 kHz, respectively). Overexpression of Cabp2 in Cabp2^+/– mice did not alter hearing thresholds, suggesting no adverse effects of viral transduction (Figure 1E).

To evaluate the impact of the chosen AAV capsids, we then performed the same experiments using the PHP.eB vector (titer: 8.2 × 10¹² genome copies/ml) that transduces hair cells with high efficiency (Rankovic et al., 2021). Cochleae of injected animals were harvested at 5–6 weeks of age. Preparations were immunostained and eGFP fluorescence levels were analyzed as a proxy of successful transduction (Figure 2). As with the AAV2/1, strong eGFP immunofluorescent signal was detected but not limited to the IHCs (Figure 2A). On average, 98% of IHCs in the injected ears were eGFP positive (Figures 2B,C). Injection of PHP.eB additionally led to transduction of the contralateral non-injected ears albeit with lower efficiency of IHC transduction, declining from the apex toward the cochlear base. ABR recordings revealed lower average hearing thresholds in injected Cabp2^–/– mice when compared to Cabp2^–/– mice that received no viral injections (Figure 3; two-way ANOVA). Partial hearing restoration was found in both, injected and contralateral non-injected ears (Figure 3A, N = 9, see Table 1A for statistical analysis). The latter is consistent with the observation of significant hair cell transduction of the contralateral ears when using the PHP.eB vector (Figure 2). Seven out of nine injected animals (78%) displayed ABR thresholds (at least for one tone burst frequency) that were at the border or lower than the mean – 2xSD of the thresholds in the non-injected knock-out control animals (Figure 3B and Table 1B). Furthermore, PHP.eB-mediated expression of Cabp2 increased the amplitudes of click-evoked ABRs (Supplementary Figure 3; Tukey’s multi-comparisons test, p < 0.0001; Table 2). This was particularly pronounced in Cabp2^–/– injected mice (Figures 3C–E and Table 2) with improved thresholds (Figure 3B). Partial viral rescue of the wave I amplitude supports the hypothesis of improved auditory function at the level of IHCs/SGNs [2.2 (and 2.0) ± 0.3 vs. 1.2 ± 0.3 μV in injected (and non-injected contralateral) vs. control Cabp2^–/– ears at 80 dB SPL; Figures 3C–E]. To assess cochlear amplification, we recorded distortion product otoacoustic emissions (DPOAEs). These showed comparable thresholds in injected Cabp2^–/– and wild-type control animals, suggesting unaltered OHC function (Figure 3F). Finally, we tested the effects of Cabp2 overexpression in wild-type animals. Using the PHP.eB vector, we observed normal ABR amplitudes (Figures 3C–E, Table 2, and Supplementary Figure 3; 5.4 ± 0.7 and 5.5 ± 0.5 μV in injected, and control Cabp2^+/+ animals at 80 dB SPL), but, for unknown reasons, elevated thresholds of tone burst-evoked ABRs compared to non-injected wild-type controls (Figure 3B). The latter finding contrasts our results with the AAV2/1-Cabp2 in heterozygous animals. Given the lack of antibodies suitable for Cabp2 immunohistochemistry (Picher et al., 2017), we could not directly compare the expression levels of Cabp2 in these animals. However, we suspect that high concentration of Cabp2, which likely occurs in the presence of two functional Cabp2 alleles and with high virus-mediated Cabp2 expression by the PHP.eB vector, may hamper IHC function, e.g., by excessive Ca²⁺-buffering. DPOAE recordings revealed no systematic effect of viral injection in the wild-type animals (Figure 3F). Five out of eight PHP.eB-injected wild-type animals displayed normal DPOAE thresholds (Supplementary Figure 3H). We note that DPOAEs were strongly impaired in three animals despite well preserved ABR click thresholds of 30–40 dB. We can thus not exclude a possibility of a low-mid-frequency OHC defect in a selection of wild-type injected animals. This may be due to mis-expression of (too high concentrations of) Cabp2 in the OHCs, where eGFP was also detected (Figure 2A), or potential complications of the viral injection itself.

In the last set of experiments, we addressed the function of Cabp2 in modulating IHC Ca_V1.3 channels by suppressing calcium- and voltage-dependent channel inactivation (Cui et al., 2007; Schrauwen et al., 2012; Picher et al., 2017). We tested whether AAV-mediated delivery of Cabp2 restores Ca_V1.3 channel function in the IHCs of Cabp2^–/– animals. For this, we dissected the organs of Corti of the PHP.eB-Cabp2-injected Cabp2^–/– animals at 5–6 weeks of age and investigated IHC function using the perforated patch-clamp technique (Figure 4). In order to closely match physiological conditions, recordings were performed at 33.0 ± 0.1°C, at an extracellular calcium concentration of 1.3 mM and holding potential of −65 mV. Using short depolarizations to increasing membrane potentials, we first examined calcium current-voltage relationships in IHCs of different test groups (Figure 4B). In line with a previous study (Picher et al., 2017), calcium currents in Cabp2-deficient IHCs had maximal amplitudes comparable to age-matched controls (Figure 4B), but showed pronounced inactivation upon 500-ms long step depolarizations (Figures 4C–E; p = 0.0001 when testing residual currents using Wilcoxon rank test). In IHCs from injected ears of Cabp2^–/– mice, the inactivation of calcium currents was significantly attenuated and residual currents restored to close to wild-type levels (I₅₀₀ values in wild-type IHCs: 76 ± 2%, Cabp2^–/– IHCs: 58 ± 2%, and PHP.eB-Cabp2-injected Cabp2^–/– IHCs: 72 ± 1% of initial current; p < 10^–5 and 0.04 for IHCs of PHP.eB-Cabp2-injected Cabp2^–/– vs. Cabp2^–/– mice or vs. wild-type mice, respectively; Figure 4E). Finally, we probed exocytosis of Cabp2^–/– IHCs following viral rescue using membrane capacitance recordings. 10- and 100-ms long depolarization steps to maximal calcium current potentials were applied to inspect fast and slow components of exocytosis (example trace in Figure 4F). We found comparable IHC exocytosis in all tested groups (Figure 4G). This implies that the rescue of Cabp2 expression via viral delivery did not interfere with the synaptic physiology of IHCs. However, IHCs of injected wild-type animals showed slightly reduced calcium current amplitudes (Supplementary Figures 4A–C) and consequently a tendency toward decreased calcium charge transfer (Figure 4G). This could be explained by potential harmful effects of Cabp2 overexpression that may reduce the viability of the system. Furthermore, slight adverse effects of viral injections, or of the presence and overexpression of eGFP cannot be excluded. In this respect, it has to be noted that upon inspection of immunohistochemical preparations no hair cell loss was apparent (Supplementary Figure 4D), suggesting that partially elevated ABR thresholds in wild-type injected animals are not due to extensive hair cell death.

In this study, we investigated the potential of an AAV-mediated gene therapeutic approach for DFNB93 hearing impairment. We demonstrate efficient gene delivery with partial rescue of hearing in the Cabp2^–/– mouse model of DFNB93 using two viral capsids, AAV2/1, and synthetic PHP.eB. Both capsids very efficiently transduced IHCs and showed partial improvement of hearing thresholds as well as increased ABR amplitudes in high proportions (ca. 70%) of injected animals. Upon injection of PHP.eB significant transduction along with partial rescue of hearing was also detected in contralateral, non-injected ears. AAV-mediated expression of Cabp2 in the IHCs of Cabp2^–/– animals efficiently inhibited IHC Ca_V1.3 channel inactivation. Hearing impairment in DFNB93 was suggested to be caused by steady-state inactivation of these channels resulting in hampered synaptic transmission at the first auditory synapse (Schrauwen et al., 2012; Picher et al., 2017). Our data suggests that inhibition of calcium channel inactivation underlies hearing improvement in the injected Cabp2^–/– animals, thereby corroborating the proposed DFNB93 disease mechanism, and provides preclinical proof of concept for future gene therapy of DFNB93.

Both AAV variants partially restored hearing thresholds as well as click-evoked ABR amplitudes in Cabp2^–/– animals. Here, improvement of the ABR wave I amplitudes, which reflect synchronized activity of the SGNs, is of particular importance (see Figures 1F-G, Figures 3C-D and Supplementary Figure 3C). To elucidate the mechanisms underlying heterogeneous effectiveness of these AAV-based rescue approaches, in particular with respect to different frequency ranges, a direct correlation of Cabp2 expression level with IHC physiology, and exact tonotopic position will be required in the future. The results of the current analysis of PHP.eB transduction efficiency and ABR analysis suggest that gene therapy matching the native Cabp2 concentration may be most critical toward the cochlear base. In situ hybridization mRNA probes suggested no major concentration differences along the tonotopic axis (Yang et al., 2016), however proper estimation of the Cabp2 protein concentrations in the IHCs in general and along the tonotopic axis more specifically, is still missing. This is relevant, since Cabp2, as other members of the EF-hand protein family, binds calcium ions and may thus act as a calcium buffer (Pangrsic et al., 2015), in particular when present at high concentrations (Picher et al., 2017). Previous work has shown similar endogenous buffer concentration in the apical and basal IHCs of gerbils and rats (Hackney et al., 2005; Johnson et al., 2008), but different coupling of calcium channels to synaptic vesicle fusion (Johnson et al., 2017). Due to looser coupling, exocytosis of the high-frequency basal gerbil IHCs may be more sensitive to calcium buffering as compared to the apical IHCs (Johnson et al., 2017). Hence, basal IHCs may be more susceptible to potentially altered calcium buffering by overexpression of Cabp2. Here, development of promoters and regulatory elements to more closely meet endogenous expression levels of Cabp2 might be of great benefit for the future gene therapeutic studies. Connected to that, our current analysis of the Cabp2 expression was limited to the observation of the fluorescence of eGFP, which is not covalently linked to the protein of interest to not interfere with its functionality. In the future, improved labeling of the Cabp2 (e.g., via sensitive and specific Cabp2 antibodies) is required to support further development of the DFNB93 gene therapy.

This is further motivated by the observation that wild-type animals injected with the same viral construct as the Cabp2^–/– animals showed modest elevation of hearing thresholds as well as somewhat reduced IHC calcium currents as compared to non-injected controls. The underlying cause is currently unknown, but may include potential harmful effects of the viral vector, overexpression of a large tracer protein eGFP (Eckrich et al., 2019; but see also Askew et al., 2015; György et al., 2017; Landegger et al., 2017; Suzuki et al., 2017; Dulon et al., 2018; Lee et al., 2020), and the injection procedure itself (Chien et al., 2016) or overexpression of the Cabp2 in either IHCs or other cochlear cell types (e.g., SGNs). The latter would most strongly affect IHCs in injected wild-type animals. For example, in the presence of CaBP2, the Ca_V1.3 calcium current density in HEK cells may be decreased (e.g., Yang et al., 2006; but see Picher et al., 2017), suggesting a potential role of Cabp2 in regulating Ca_V1.3 expression levels. Although maximal current amplitudes in IHCs from Cabp2^–/– animals were comparable to wild-type controls, the consequences of Cabp2 overexpression may be stronger. Furthermore, whereas no hair cell loss and no OHC defect was observed in injected Cabp2^–/– animals, DPOAEs were defective in a fraction of injected wild-type animals. Potential reasons include mis-expression of Cabp2 in OHCs and manipulation of the ear during viral injection itself, and could have contributed to elevated tone burst ABR thresholds in these animals. Finally, potential harmful effects of Cabp2 on SGNs or supporting cells have not yet been studied. Further development of more selective promoters for a targeted expression of Cabp2 in IHCs but not OHCs or other cell types may be crucial for the future therapy development.

Re-introduction of Cabp2 in Cabp2^–/– IHCs prevented strong inactivation of Ca_V1.3 channels typically observed in untreated Cabp2-deficient IHCs. As hypothesized previously, enhanced steady-state Ca_V1.3 channel inactivation likely hampers IHC synaptic transmission (Picher et al., 2017). We were, however unable to observe reduced exocytic responses in Cabp2-deficient IHCs. As previously (Picher et al., 2017), we hypothesize this could be due to (i) potentially increased extra-synaptic exocytosis due to the loss of Cabp2-dependent calcium-binding capacity as observed in IHCs deficient in mobile calcium buffers (Pangrsic et al., 2015) or to (ii) a failure to mimic in vivo conditions (with likely higher IHC activity evoking steady-state inactivation) in the isolated organs of Corti (where calcium channels can recover from inactivation between consecutive stimuli).

The results of the current study, demonstrating that hearing rescue in the Cabp2-deficient animals is possible, are encouraging. However, future studies with improved specificity of viral targeting as well as better control of expression levels of the transgene(s) are required to further improve the extent of hearing rescue. Moreover, a previous study suggested the alternative splice variant Cabp2-alt as the most abundant isoform of Cabp2 in the IHCs (Yang et al., 2016). So far, it is not known whether the two long splice variants (Cabp2-L, used in this study; and Cabp2-alt) differ in their biochemical properties, interactions, or function. A future study should address this question and test whether the choice of the splice variant can affect the extent of hearing rescue. Importantly, Cabp2^–/– animals show normal cochlear development and no early degeneration of stereocilia, ribbon synapses or SGNs, nor hair cell loss (Picher et al., 2017), which suggests a prolonged temporal window for therapeutic intervention in mice and humans may exists, a notion that should be tested in the future. If so, once improved for specificity and with a better dose control, gene therapy holds great promise for success in adult DFNB93 patients.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was reviewed and approved by the board for animal welfare of the University Medical Center Göttingen and the animal welfare office of the state of Lower Saxony (LAVES; Animal protocol numbers 14.1391, 19.3133, and 19.3134).

TP, DO, MP, and TM designed the study. DO performed experiments (cell electrophysiology, immunohistochemistry, and part of ABR/DPOAE recordings) and analyzed the data. MP and VR designed the plasmids. VR generated PhP.eB viral vectors. MP and TP performed initial experiments. MP performed molecular cloning. TP analyzed the data. TM established inner ear injections. TP and DO wrote the manuscript. All authors revised the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.