Gene therapy with a synthetic adeno-associated viral vector improves audiovestibular phenotypes in Pjvk-mutant mice

Recessive PJVK mutations that cause a deficiency of pejvakin, a protein expressed in both sensory hair cells and first-order neurons of the inner ear, are an important cause of hereditary hearing impairment. Patients with PJVK mutations garner limited benefits from cochlear implantation; thus, alternative biological therapies may be required to address this clinical difficulty. The synthetic adeno-associated viral vector Anc80L65, with its wide tropism and high transduction efficiency in various inner ear cells, may provide a solution. We delivered the PJVK transgene to the inner ear of Pjvk mutant mice using the synthetic Anc80L65 vector. We observed robust exogenous pejvakin expression in the hair cells and neurons of the cochlea and vestibular organs. Subsequent morphologic and audiologic studies demonstrated significant restoration of spiral ganglion neuron density and hair cells in the cochlea, along with partial recovery of sensorineural hearing impairment. In addition, we observed a recovery of vestibular ganglion neurons and balance function to WT levels. Our study demonstrates the utility of Anc80L65-mediated gene delivery in Pjvk mutant mice and provides insights into the potential of gene therapy for PJVK-related inner ear deficits.


Introduction
Sensorineural hearing impairment (SNHI) is the most common inherited sensory defect. Permanent SNHI is estimated to occur in approximately 2 out of 1,000 newborns (1). This disorder, including late-onset SNHI, may affect 2% of school-age children (2,3). Greater than 50% of cases of SNHI in children have genetic causes and are classified as hereditary hearing impairment (HHI) (4). To date, more than 200 genes have been shown to be causally related to HHI (5).
The pathogenesis of SNHI encompasses a wide range of disease mechanisms and can result from damage to any site along the auditory pathway, including hair cells (HCs) or other cochlear cells, spiral ganglion neurons (SGNs), the auditory nerve, or the central auditory pathway. Current treatments for SNHI are mainly tailored according to the severity of the disease. For patients with mild to moderate SNHI, hearing aids are the first-tier treatment. In contrast, for patients with profound SNHI, amplification with hearing aids is usually insufficient because of the severely damaged HCs, and cochlear implantation is the treatment of choice (6). Bypassing the damaged HCs, the cochlear implant directly activates SGNs, transmits auditory signals through auditory nerve fibers and the central neural pathway, and ultimately results in Recessive PJVK mutations that cause a deficiency of pejvakin, a protein expressed in both sensory hair cells and first-order neurons of the inner ear, are an important cause of hereditary hearing impairment. Patients with PJVK mutations garner limited benefits from cochlear implantation; thus, alternative biological therapies may be required to address this clinical difficulty. The synthetic adeno-associated viral vector Anc80L65, with its wide tropism and high transduction efficiency in various inner ear cells, may provide a solution. We delivered the PJVK transgene to the inner ear of Pjvk mutant mice using the synthetic Anc80L65 vector. We observed robust exogenous pejvakin expression in the hair cells and neurons of the cochlea and vestibular organs. Subsequent morphologic and audiologic studies demonstrated significant restoration of spiral ganglion neuron density and hair cells in the cochlea, along with partial recovery of sensorineural hearing impairment. In addition, we observed a recovery of vestibular ganglion neurons and balance function to WT levels. Our study demonstrates the utility of Anc80L65-mediated gene delivery in Pjvk mutant mice and provides insights into the potential of gene therapy for PJVKrelated inner ear deficits. speech understanding in the brain auditory cortex (7). Unfortunately, the outcomes with cochlear implants substantially vary among recipients and are closely related to the underlying pathologies. For instance, our previous study showed that pathogenic variants in the PJVK gene (Gene ID: 494513) were associated with unfavorable performance of the cochlear implant (8).
The PJVK gene, first identified in 2006 to be linked to a recessive form of nonsyndromic HHI DFNB59 (9), contains 7 exons and encodes the 352 aa protein pejvakin. Pejvakin is expressed in inner hair cells (IHCs), outer hair cells (OHCs), supporting cells, and SGNs in the inner ear (9, 10) and plays a crucial role in action potential transmission, intracellular trafficking, and peroxisomal activity (11,12). Previous studies in Pjvk KO mice demonstrated pathologies involving IHCs, OHCs, and SGNs (12,13). To address the molecular mechanisms underlying the unfavorable cochlear implant performance in patients with PJVK variants, we generated a knock-in mouse model with the Pjvk p.G292R (c.874G>A) variant commonly found in the Han Taiwanese population by the CRISPR/Cas9 approach; the Pjvk p.G292R mice showed a significant increase in hearing thresholds and impaired balance compared with WT mice (14). Specifically, degeneration of SGNs in histologic analysis was identified, providing insight into the pathogenetic mechanisms underlying the unfavorable cochlear implant performance associated with the PJVK variant in our previous clinical observation (8).
The pathological involvement of SGNs, as exemplified by patients with pathogenic PJVK variants, represents a limitation of cochlear implantation. Recent advances in gene therapy have suggested therapeutic strategies to address this unmet clinical need. For instance, our previous studies revealed that the synthetic Anc80L65 vector could be used for efficient gene transduction in IHCs, OHCs, and SGNs of neonatal and fetal mice (15,16) and could restore auditory and vestibular functions simultaneously in mice with certain types of HHI (17). Because the tropism of Anc80L65 in the inner ear parallels the pathologies involved in PJVK mutations, in this study, we explored the therapeutic effects of Anc80L65 on hearing and balancing impairments in the Pjvk p.G292R mice.

Results
Anc80L65-CMV-PJVK shows robust transduction in the neonatal mouse inner ear. Anc80L65 is a synthetic capsid facilitating robust transduction efficiency in many cell types in the inner ear without causing significant hearing dysfunction (15,16,18,19). We designed an adeno-associated serotype 2 (AAV2) inverted terminal repeat-flanked (ITR-flanked) construct containing a CMV enhancer and CMV promoter, a fulllength human PJVK coding sequence, an FF2A peptide, and an EGFP sequence and packaged it into an Anc80L65 capsid, yielding Anc80L65-CMV-PJVK ( Figure 1A).
A previous study showed that the Pjvk G292R/G292R phenotypes, including profound hearing loss and IHC bundle loss, were detected from hearing onset (14); therefore, we decided to treat at the neonatal stage (P0-P1) to stop or slow degeneration. After Anc80L65-CMV-PJVK microinjection via the round window membrane (RWM) approach to Pjvk G292R/G292R mice at P0-P1, we first examined the transduction efficiency of the viral vector. The mice were sacrificed 10 days after the injection (P10), and the inner ears were collected for whole-mount or paraffinized section immunofluorescence staining ( Figure 1, B and C) and qPCR ( Figure 1D). The samples were divided into SGNs, vestibular ganglion neurons (VGNs), organs of Corti (OC), and vestibular organs (VO; including the utricle, the saccule, and the anterior, posterior, and horizontal semicircular canals).
Anc80L65-CMV-PJVK rescues Pjvk G292R/G292R SGN density to WT levels. We first evaluated the therapeutic effect of Anc80L65-CMV-PJVK in SGNs. Pjvk G292R/G292R mice were injected at P0-P1 with viral vectors through the RWM approach, and inner ear tissue was collected at P30, P60, P90, and P120 for paraffin sectioning and immunofluorescent staining. Untreated Pjvk G292R/G292R mice and WT samples collected at the same time were used as the negative and positive controls, respectively. The SGN density of the untreated mice was similar to that of the WT mice at P30 but progressively decreased afterward (Figure 2A). Compared with that of the untreated mice, the SGN density of the treated mice was maintained until P120 (untreated Pjvk G292R/G292R versus treated mice, quantified by β-III tubulin + cells per mm 2 9, P < 0.05; and at P120, 473.3 ± 139.7 versus 2,277.0 ± 267.2, P < 0.001; n = 3 untreated mice, n = 4 treated mice at all time points) and showed no significant difference from that of the WT mice at all time points (n = 3 Pjvk WT/WT mice versus n = 4 treated mice, P > 0.05) ( Figure  2B). The results indicated that overexpression of PJVK in Pjvk G292R/G292R mice could significantly rescue the density of SGNs to WT levels. Moreover, this effect could last until at least P120.
Anc80L65-CMV-PJVK enhances long-term survival and bundle number of OHCs. Our previous study showed progressive loss of OHCs and their related hair bundles in Pjvk G292R/G292R mice (14). To test whether Pjvk overexpression could reverse these phenotypes in Pjvk G292R/G292R mice, we collected inner ear tissue at P21, P30, P60, and P90. The HCs were labeled with anti-myosin VIIA and quantified ( Figure 3, A and B). We found a significant difference in HC number compared with that of the untreated mice at P60 and P90 (untreated Pjvk G292R/G292R mice [n = 5] versus treated mice [n = 12], quantified by myosin VIIA + HCs per 100 μm: at P60, 36.0 ± 9.0 versus 41.8 ± 3.0, P < 0.05; at P90, 19.0 ± 4.4 versus 31.3 ± 5.8, P < 0.05). Despite the positive effect observed in the treated Pjvk G292R/G292R mice compared with the untreated mice, the HC number in the treated Pjvk G292R/G292R mice was still lower than that in the WT mice at P60 and P90 (treated Pjvk G292R/G292R mice [n = 4] versus Pjvk WT/WT mice [n = 3], quantified by myosin VIIA + HCs per 100 μm: at P60, 41.8 ± 3.0 versus 49.0 ± 2.6, P < 0.05; at P90, 31.3 ± 5.8 versus 48.0 ± 2.6, P < 0.001). Whole-mount and paraffin-section samples of mouse inner ear organs underwent immunofluorescence staining for β-III tubulin (red) and myosin VIIA (red) to detect neurons and HCs, respectively. Scale bars: 100 μm. (C) The transduction efficiency was quantified in the collected parts of P10-treated mice with GFP expression. For SGN and VGN quantification, cells that were double positive for β-III tubulin and GFP were counted; for vestibular HCs (VHCs) and cochlear HCs (CHCs), cells with myosin VIIA and GFP double positivity were counted (n = 4 or 6 in each collection part). (D) Quantification of human PJVK mRNA expression in P10 WT, untreated Pjvk G292R/G292R , and treated mice. Samples were divided into SGNs, VGNs, VO, and OC and then detected by qPCR (n = 4 or 6 in each collection part). Data are shown as the mean ± SD. ****P < 0.0001, **P < 0.01, *P < 0.05, 1-way ANOVA with Tukey post hoc tests. Missing bundles in the third-row OHCs of Pjvk mutant mice were described in previous studies (13,14). To determine whether exogenous PJVK could rescue the missing third-row bundles, we counted the third-and first-row bundles and determined the ratio at both P21 and P42 (Figure 3, C and D; the white arrows indicate the missing third-row bundles). Fewer missing third-row bundles were observed in the treated mice than in the untreated Pjvk G292R/G292R mice at both time points (untreated Pjvk G292R/G292R versus treated, quantified by the ratio of third-to first-row bundles per OHC: at P21, 58.1 ± 6.3 versus 81.1 ± 10.5, P < 0.01; and at P42, 31.6 ± 6.9 versus 59.6 ± 4.3, P < 0.001; n = 4 untreated mice, n = 4 treated mice at all time points). These results indicated that PJVK delivery to the inner ear of the Pjvk G292R/G292R mice could slow the progressive decrease in the loss of both HC numbers and OHC bundles.
The ABR waveforms in P21 mice were also analyzed. The absolute and interpeak latencies of ABR waveforms in Pjvk G292R/G292R mice, recorded at 100 decibel sound pressure level (dBSPL), were all significantly curtailed after the treatment, indicating the alleviation of both cochlear (e.g., absolute latency of wave I) and retrocochlear (e.g., interpeak latencies I-III and III-IV) lesions after gene therapy (Supplemental Figure 1A and Supplemental Table 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.152941DS1). We also measured the function of OHCs by using distortion-product otoacoustic emissions (DPOAEs). DPOAE amplitudes were increased at high frequencies (18 and 30 kHz) compared with untreated mice (Supplemental Figure 1B). Despite the nature of progressive hearing deterioration determined by ABR, longer term hearing threshold follow-up at P30, P42, P60, until P90 also showed better results in the treated Pjvk G292R/G292R mice than in the untreated mice ( Figure 4C).
To investigate whether overexpression of PJVK in the inner ear of Pjvk G292R/G292R mice would restore vestibule-related balance functions, we performed open-field tests, rotarod tests, and swimming tests on mice at P120 to evaluate balance-related behavior. In the open-field test, we recorded a 3-minute video of the Pjvk WT/WT , Pjvk G292R/G292R , and treated Pjvk G292R/G292R mice to track their circling behavior during walking ( Figure 6A and Supplemental Videos 1-3). Similar to the WT mice, the treated mice circled 0 to 2 times

Discussion
We previously reported that patients with recessive PJVK variants had unfavorable cochlear implant performance (8), which, as shown in our subsequent study in transgenic mice with defective Pjvk, might be attributable to degeneration of SGNs (14). In the present study, we further demonstrated that Anc80L65-mediated gene therapy could improve hearing and rescue balance functions in Pjvk mutant mice.
Our results showed that Anc80L65 could transduce SGNs and VGNs with high efficiency, and Anc80L65-mediated gene therapy could restore the density of SGNs in Pjvk mutant mice. In addition, we also found that myelination of spiral ganglion was disrupted in the Pjvk p.G292R mice, and gene therapy with Anc80L65 restored the myelination status of the spiral ganglion (Supplemental Figure 3). Previous studies demonstrated that the density and myelination of SGNs were positively correlated with the neural response to cochlear implant stimulation (20)(21)(22)(23). Gene therapies with certain neurotrophic factors, including brain-derived neurotrophic factor and neurotrophin-3, effectively improved SGN density in animal models with SNHI (22)(23)(24)(25)(26)(27). In contrast to these nonspecific approaches with neurotrophic therapy for acquired SNHI, the present study demonstrated that the decreased SGN density and myelination defects resulting from specific genetic defects could also be ameliorated by the delivery of the corresponding genes, thus expanding current gene therapy methods for treating SNHI due to different pathogenetic mechanisms. As such, our findings indicate that Anc80L65-mediated gene therapy, when delivered together with the implanted electrode, may serve as a potential modality to augment the function of cochlear implant in patients with defective PJVK expression or similar genetic pathologies.
In addition to SGNs, the pathology of Pjvk-related SNHI involves IHCs and OHCs (13,28). Pejvakin is expressed in both IHCs and OHCs (28) and is crucial for the maintenance of stereocilia of IHCs and OHCs (13). Consistent with these lines of evidence, our previous study in Pjvk mutant mice identified decreased cochlear microphonic amplitude as a physiological feature and substantial stereocilia loss as a histologic feature, both of which are indicative of pathology in HCs (14). As demonstrated in the present study, both the physiological and histologic changes could be partially rescued by Anc80L65-mediated gene therapy.
Different AAV serotypes display tropism to different cell types in the inner ear. Among the natural AAV serotypes, AAV1-4, AAV7, and AAV8 are efficient for transducing the spiral ligament, spiral limbus, and SGNs; AAV5 is efficient for transducing Claudius cells, sulcus cells, and SGNs; and AAV1-3, AAV5, AAV6, and AAV8 are efficient for transducing IHCs (29)(30)(31)(32). Delmaghani et al. (21) reported the use of AAV8 carrying the Pjvk transgene in a DFNB59 mouse model, which was found to have variable degrees of hearing loss with degeneration of HCs and spiral ganglions. These researchers found some degree of electrophysiological recovery of auditory phenotypes at P21 after treatment at P3. Although the natural AAV serotypes show inconsistent transduction of SGNs, IHCs, and OHCs, the synthetic Anc80L65 vector was reported to have ideal transduction efficiency in inner ear cell types such as IHCs, OHCs, and SGNs, as shown in our previous studies (16,18) and in others (15,(17)(18)(19)(33)(34)(35)(36). We found that gene therapy with Anc80L65 restored the morphology of HCs, increased the density of SGNs, and improved hearing. This feature indicates that the Anc80L65 vector is potentially useful for gene therapy in HHI with pathological involvement of both HCs and SGNs, such as HHI caused by pathogenic PJVK variants.
In addition to the auditory system, we observed substantial improvement of vestibular dysfunction in the Pjvk mutant mice after Anc80L65-mediated gene therapy (Figures 5 and 6). Vestibular dysfunction has been reported in patients with pathogenic variants in PJVK (37). We previously ascribed Pjvk-associated vestibular deficits to the degeneration of VGNs because we could not observe pathology in the HCs of VOs in our transgenic mice (14). This hypothesis was substantiated by the findings of the present study, because the restoration of vestibular function concurred with an improved density of VGNs after gene therapy. Notably, vestibular disorders are not uncommon in patients with HHI (38)(39)(40). However, compared with auditory symptoms, vestibular disorders have been relatively overlooked by the scientific community thus far, and the genetic basis of vestibular disorders has just started to be understood (40). In addition to this study, a few prior studies documented recovery of balance function after gene therapy in different transgenic mouse models (17,25,27,41,42), demonstrating the therapeutic potential of gene therapy to address vestibular deficits associated with HHI.
Although PJVK gene therapy restores audiovestibular functions in treated mice, we still observed progression of hearing loss until a later stage (P120). Despite the robust expression of pejvakin after the initial delivery, the transgene tends to subside at a later stage (Supplemental Figure 2) and may not be sufficient to reverse the progression of underlying pathology. The restorative effects shown in the vestibular system may be reflected by the relatively late onset of vestibular defects and indicate a relatively longer therapeutic window. More studies are required to improve the sustainability of PJVK gene therapy either alone or in combination with cochlear implants in the clinical setting.
In this study, we adopted the RWM approach and delivered viral particles directly into the perilymphatic space of the scala tympani in the cochlea. In addition to the RWM approach, 2 other surgical approaches have been applied to deliver therapeutic agents to the inner ear in animal models, namely, the canalostomy approach, whereby the therapeutic agent is injected through a fenestration in the posterior semicircular canal (18,43), and the cochleostomy approach, in which a hole near the round window is drilled through the basal part of the cochlea into the cochlear endolymphatic space (29,44). We observed efficient gene transduction and phenotypic improvement in both the auditory and VOs in Pjvk mutant mice after gene therapy with the RMW approach in the present study. This finding has important clinical implications in using gene therapy alone or as an augmenting strategy for cochlear implantation. The RWM approach is currently the most favorable surgical approach for inserting the electrode array of the cochlear implant into the cochlea, because it incurs less surgical trauma to the inner ear than cochleostomy and has a better chance to preserve residual hearing and intricate cochlear structures in patients with cochlear implants (45).
In conclusion, this study revealed that Anc80L65-mediated gene therapy could improve both hearing and balance function in Pjvk mutant mice. In particular, our results demonstrated that Anc80L65-mediated gene therapy significantly improved the SGN density in Pjvk mutant mice. These results indicate the potential of gene therapy as a standalone treatment as well as in optimizing cochlear implant function in the future.

Methods
Animals. Pjvk G292R/G292R mice were generated on the C57BL/6J background, which was reported previously (14). The homozygous mice were bred together based on their normal fertility. The WT C57BL/6J mice purchased from BioLASCO Taiwan Co., Ltd., were used as controls.
Plasmid construction and AAV vector production. The PJVK-FF2A fragment (1,164 bp) was synthesized and then subcloned into the backbone plasmid pAAV-CMV-RS1-Tag-FF2A-EGFP-WPRE-BGH (provided by the Gene Transfer Vector Core, Massachusetts Eye and Ear Infirmary). Fragment synthesis and subcloning were outsourced and performed by Genscript Biotech Corporation. AAV vectors were produced by using HEK293 cells (Food Industry Research and Development Institute) in hyperflasks with triple plasmid transfection, according to previously described methods (46). Cells were transfected with AAV capsid (Anc80L65), AAV2 ITR-flanked transgene (AAV-CMV-PJVK-FF2A-EGFP-WPRE-bGH), and adenovirus helper plasmids at a ratio of 1:1:2 using polyethylenimine (PEI Max, Polysciences). Three days after transfection, the whole cell culture (i.e., cells and supernatant) was harvested and subjected to lysis, Benzonase treatment, and precipitation in high-salt solution overnight at 4°C. The supernatant was collected after centrifugation and filtration and then concentrated by tangential flow filtration. The full capsids were enriched by iodixanol density-gradient centrifugation. The full capsid-enriched fraction was buffer exchanged and concentrated by running through an Amicon filter with a 100 kDa MW cutoff. The vectors were quantified by qPCR using the TaqMan (Life Technologies) system with primers and probes targeting the promoter sequences of the transgene cassette. Then, the vector purity was assessed by SDS-PAGE. The titer of the virus was 1.51 × 10 12 GC/mL.
Mouse experiments. Neonatal Pjvk G292R/G292R mice (P0-P1) of either sex were used. All surgical procedures were performed in a clean and dedicated place. The instruments were autoclaved and soaked in 75% ethanol before the surgery. The viral vector was initially aliquoted and stored at -80°C to avoid repeated freezing and thawing. Neonatal mice were anesthetized by hyperthermia on ice for 2 minutes and then placed under a microscope for surgery. For RWM microinjection, a postauricular incision was made to expose the anatomic landmarks of the cochlea. After identification of the anatomic landmarks tympanic ring and stapedial artery, the mice were injected with 1 μL of viral vector within 2 minutes (~8 nL/s) per cochlea. The microcapillaries (Drummond) were held by a micromanipulator (WPI) and controlled by a Micro4 microsyringe pump controller (WPI).
Immunofluorescence staining. For whole-mount samples, the mouse inner ears were collected and fixed in 4% paraformaldehyde in PBS at room temperature overnight. After 3 washes with PBS, the inner ears were carefully dissected, and the OCs were divided as previously reported (47). The OCs were permeabilized with 0.25% Triton X-100 and 0.1% paraformaldehyde in PBS at room temperature for 1 hour and then washed 3 times with PBS. After permeabilization, the OCs were blocked by blocking buffer PBT1, which contained 1% BSA, 5% FBS, and 0.1% Triton X-100 in PBS. The OCs were then incubated with the primary Ab anti-myosin VIIA (rabbit, 1:400; Proteus Biosciences) at 4°C overnight. Then, the cells were incubated with Alexa Fluor 568-labeled Abs (1:200; A10042, Invitrogen) at room temperature for 2 hours. After PBS washes, the tissues were mounted using a ProLong Antifade Kit (Molecular Probes). Images were acquired and analyzed by confocal microscopy (Zeiss LSM 880) and ZEN lite software, respectively.
Ganglion neuron counts. The cells in sectioned samples with β-III tubulin + and DAPI + cells were counted. For each sample, 3 nonconsecutive sections were selected. The region for calculation was approximately 1 mm 2 and measured by the open-source image processing software ImageJ (NIH).
Myelination. The cells in sectioned samples were triple stained with MBP, GFP, and β-III tubulin Abs. The quantification method for myelination was described previously (48). Briefly, the MBP + myelin sheath was considered intact if it enclosed more than 80% of the outline of the β-III tubulin + cells. Next, the cells with intact myelin sheath were calculated, and the percentage of β-III tubulin + SGNs with intact MBP + myelin sheaths in each group were quantified. For each sample, 3 nonconsecutive sections were selected. The region for calculation was approximately 1 mm 2 and measured with ImageJ (NIH).
HC counts. OC cells that were myosin VIIA + and DAPI + were counted. For each sample, 5 random confocal images were acquired. There are approximately 520-540 HCs/100 μm OC in WT mice.
OHC bundle count. Scanning electron microscopy images were used for OHC bundle counting. We randomly chose 5 OHCs per cochlea for quantification, and the total number of cochleae sampled for quantification was 4 per ear group: Pjvk WT/WT , treated, and treated Pjvk G292R/G292R . There are 3 rows of bundles per OHC, with 28 to approximately 32 bundles per row, which varies among turns in the cochlea. After calculating the numbers of bundles in the first and third rows of identical OHCs, we calculated the ratio of third and first bundles per OHC to quantify the OHC bundle loss. The detailed quantification method was described previously (13).
ABR and DPOAE measurement. Both ABR and DPOAE tests (Supplemental Figure 1B) were performed in an anechoic chamber to avoid background noise. For ABR measurement, click sounds (4-8 kHz) and 8, 16, and 32 kHz tone bursts at various intensities were generated to evoke ABRs (Smart EP 3.90; Intelligent Hearing Systems). For DPOAE recording, 2 speakers presenting 2 primary tones (f1 and f2) at a frequency ratio of 1.2 were coupled to a microphone and introduced into the ear canal of mice; the pure tone amplitudes were fixed at 65 dBSPL for L1 and 55 dBSPL for L2 (Intelligent Hearing System, SmartOAE Program, Intelligent Hearing Systems Corp.).
Scanning electron microscopy. For observation of bundle morphology, fresh inner ears were collected and prepared for scanning electron microscopy samples as described previously (49). Samples were subjected to critical-point drying using a Samdri PVT-3D and coated with gold-palladium. Imaging was performed using a JEOL JSM-7600F scanning electron microscope. OHCs were selected from the apical to middle regions of the cochlear spiral. Vestibular balance function tests. Balance function tests were performed according to methods described previously (14). Briefly, in the open-field test, the mice moved freely in an enclosure area, and their movement was recorded with a camera for 3 minutes. The number of times each mouse circled during the 3-minute recording was calculated and quantified. In the rotarod test, a rotarod apparatus with an accelerating program (4-40 rpm over 250 seconds) was used. Three trials per test and an appropriate break time between trials were performed during the test day. The latency to fall off the rotarod was recorded. In the swimming test, the mice were placed in a plastic chamber filled with water, which forced them to swim. The swimming behavior of the mice was scored on the basis of the following criteria: swimming normally for more than 15 seconds: score = 0; swimming poorly and drowning immediately without rescue, score = 3.
Availability of data and materials. All data are available in the main text or the supplemental materials.
Statistics. Data are presented as the mean ± SD unless otherwise indicated. All values were analyzed by Student's unpaired t test or 1-way ANOVA with Tukey post hoc tests, depending on the experimental groups. Differences with a P value of less than 0.05 were considered statistically significant. All analyses were performed using GraphPad Prism, version 9.0.0, for Windows (GraphPad Software).
Study approval. All animal experiments were performed according to animal welfare guidelines and approved by the IACUC of the National Taiwan University College of Medicine (approval no. 20200082) and by Taipei Veterans General Hospital (approval no. 2018-148 and 2019-002).