ADNSHL is diagnosed in 10 individuals from a Chinese pedigree

We ascertained a pedigree of Chinese Han origin, with bilateral, postlingual ADNSHL, without tinnitus or vestibular symptoms. In affected individuals, hearing loss started at middle to high frequencies by 6–7 years of age, and subsequently progressed to all frequencies by the first or second decade (Figs 1A and EV1). For example, during our 6 years of clinical follow‐up, the pure tone threshold of the proband (IV:2) was elevated by about 30 dB on average (Fig 1B). Otoscopic examinations revealed a normal external auditory canal and tympanic membranes with normal tympanometry. Auditory steady‐state response (ASSR), temporal bone high‐resolution computed tomography (HRCT), and magnetic resonance imaging (MRI) for the proband (at 13 years of age) were normal, excluding auditory neuropathy spectrum disorders and inner ear malformations as potential causes of hearing loss. Excessive noise exposure, ototoxic medication history, or complaints of other systemic abnormalities were absent from the clinical reports of the affected individuals. A summary of the auditory phenotype is shown in Table 1.

A CGN variant co‐segregates with the ADNSHL

To identify the pathogenic variant, exome sequencing (ES) was carried out on seven affected and five unaffected individuals of the family. Genome‐wide linkage analysis of the ES results identified two candidate regions on chromosome 1 and one candidate region on chromosome 6, all with a maximum logarithm of the odds (LOD) score of 2.1 (Appendix Fig S1A and B). Further analysis revealed variants in a total of seven candidate genes (ITGA10, CGN, IL6R, MST1L, SERPINB6, SSR1, and TUBB2B/PSMG4) that co‐segregated with the disease phenotype (Appendix Fig S1C). However, except for the CGN variant, all other variants were common single‐nucleotide polymorphisms (SNPs) and, therefore, excluded from further analyses.

Sanger sequencing on the seven affected and five unaffected individuals, confirmed that the CGN variant co‐segregating with hearing loss was c.3330delG located in exon 20 of the human gene (NM_020770) (Fig 1C and D). The variant resulted in a frameshift and premature truncation of the normal C‐terminal tail of the CGN protein at amino acid Leu1110 (p.Leu1110Leufs*17), which is highly conserved among the mammalian species (Fig 1E and F). The variant was predicted to be deleterious by the Mutation Taster, PolyPhen‐2, and SIFT computational tools and was not present in the dbSNP, 1000 Genomes, ESP6500, nci60, GnomAD databases or in 10,588 ethically matched control individuals (Cao et al, 2020). This variant also was not detected in 150 additional sporadic cases of postlingual NSHL patients.

We then enrolled in the study a selection of 96 probands of independent familiar cases with hereditary hearing loss compatible with an autosomal dominant inheritance pattern. These samples have been previously subjected to genetic screening using the custom gene panel, OTO‐NGS, version 1 and 2, designed in our laboratory (Morin et al, 2020), with no causative deafness mutations being identified. CGN exon screening in this Spanish cohort only revealed benign or likely benign genetic variations, all missense in nature with some detected in two probands of independent familial cases (Appendix Table S1). These results indicate that the CGN‐associated hearing loss is likely one of the rarer forms of genetic deafness, similar to those linked to HOMER2 or CCDC50 genes (Modamio‐Hoybjor et al, 2007; Lachgar et al, 2021).

CGN is expressed in the mouse cochlea

To explore potential roles of CGN, we first examined its expression pattern in different mouse tissues, including the cochlea by RT‐qPCR analysis. Cgn mRNA is widely expressed in mouse tissues, including cochlea, kidney, and liver, with higher expression levels in lung, gonad, and intestine (Fig 2A). Western blot analysis demonstrated that the 140‐kDa CGN protein is expressed in the cochlea and the other tissues examined (Fig 2B). The specificity of the three CGN antibodies used in this study was validated by western blot and immunofluorescence analyses of cell lines transfected with CGN plasmids (Appendix Fig S2). In the postnatal mouse cochlea, Cgn expression is detected at P0, and increases from P7 to P21 (Fig 2C and D). Based on previously published cell‐type‐specific RNA‐seq results (Liu et al, 2018a; Data ref: Liu et al, 2018b), Cgn is widely expressed in P28‐35 cochlear sensory epithelial cells, with enriched expression in both inner (IHCs) and outer (OHCs) hair cells (Fig 2E). Consistently, immunofluorescence analysis of whole mount cochleae showed that CGN protein is mainly expressed at the cellular junctions in the organ of Corti, with enriched localization at the cuticular plates and circumferential belts of both cochlear IHCs and OHCs (Fig 2F). The presence of CGN at the cuticular plates of OHCs was further confirmed by its co‐immunostaining with LMO7 (Fig 2G), a cuticular plate marker protein (Du et al, 2019). Similar to the RT‐qPCR and Western blot analyses (Fig 2C and D), expression levels of CGN at the cuticular plates of OHCs increase gradually from P3 to P21 (Fig 2G).

The CGN variant results in abnormal expression of CGN protein

In order to evaluate the effect of the mutation on CGN expression, we constructed both wild‐type (WT) and mutant human CGN (p.L1110Lfs*17) plasmids with a C‐terminal FLAG‐tag (Fig 3A), and transfected them into Madin–Darby canine kidney (MDCK) cells. Compared to the WT CGN, mutant CGN was expressed at significantly lower levels by either immunofluorescence (Fig 3B) or Western blot (Fig 3C). Reduced protein expression of the mutant CGN was not due to alterations in steady‐state RNA as shown by comparable levels of the WT and mutant CGN mRNA (Fig 3D). Furthermore, transfected WT CGN protein showed preferential localizations at the cell periphery with sheet‐like or filamentous accumulations in the cytoplasm (Fig 3E), similar to previous observations (Citi et al, 2009). In contrast, expression of the residual mutant CGN protein was mainly distributed in the cytoplasm as puncta and failed to localize to the cell periphery (Fig 3E). Similar results were also observed in human colon epithelial cancer (CACO2) cells, human embryonic kidney (HEK) 293T, and CV‐1 in Origin Simian‐7 (COS‐7) cells (Appendix Fig S3). To exclude possible interference of the C‐terminal Flag‐tag on CGN expression, N‐terminal EGFP‐tagged WT and mutant human CGN plasmids were also constructed and transfected into MDCK or HEK293T cells (Fig EV2). The N‐terminal EGFP‐tagged mutant CGN exhibited the same reduced protein expression and abnormal subcellular localization as the C‐terminal Flag‐tag protein (Fig EV2). These results suggest that the expression and subcellular localization of human CGN are significantly altered by the candidate deafness mutation.

Deletion of Cgn in hair cells induces hearing impairment

As wildtype CGN is highly expressed at the apical surfaces of hair cells, we generated a hair cell‐specific Cgn conditional knockout mice (Cgn‐cKO) by crossing Cgn ^fl/fl mice with the inner ear hair cell‐specific CreER line, Pou4f3 ^EGFP‐creER mice (Du et al, 2020) (Fig EV3A and B). Following tamoxifen injection at P3‐P6, Cgn‐cKO mice lost substantial CGN immunoreactivity in both IHCs and OHCs at 2 months of age (Fig EV3C), and the average Cre recombination efficiency was about 60% (Fig EV3D).

We then assessed the auditory function in 2‐month‐old Cgn‐cKO mice by measuring distortion product otoacoustic emissions (DPOAE) which reflect the activity of OHCs, and auditory brainstem responses (ABR), which represent the sound‐evoked neural activity in the auditory nerve and brainstem. Compared to wild‐type littermates, Cgn‐cKO mice had significant elevations in DPOAE and ABR thresholds at 32 kHz (Fig 4A and B). In addition, ABR peak 1 (P1) amplitudes, the summed activity of the cochlear nerve, were also reduced at 8–32 kHz (Fig 4C–I). Histological analyses demonstrated significant loss of OHCs in Cgn‐cKO mice at the 32 kHz cochlear region (Fig 4J and K), consistent with observed ABR and DPOAE threshold elevations. These data indicate that Cgn deletion in cochlear hair cells leads to high‐frequency OHC loss and hearing impairment.

Cgn ^delG knockin mice display dose‐dependent progressive hearing loss

To determine the pathophysiological roles of the CGN variant (c.3330delG) identified in the affected individuals, we generated Cgn knockin mice (Cgn ^delG mice) harboring the exact G‐base deletion using CRISPR‐Cas9 technology (Fig EV3E–G). Cgn mRNA expression in cochlea of homozygous Cgn knockin (CGN ^delG/delG ) mice was significantly decreased compared with wild‐type (Cgn ^+/+ ) and heterozygous Cgn knock in (Cgn ^delG/+ ) mice (Fig 5A), likely due to nonsense mediated mRNA decay (NMD) of the frameshifted mutant Cgn mRNA. Western blot results confirmed that CGN protein was expressed as a truncated form and that expression was significantly reduced in cochlea and lung tissues from both heterozygous and homozygous Cgn ^delG mice (Figs 5B and EV4A). Whole mount immunofluorescence of Cgn ^delG mice cochlea showed similar results (Figs 5C and EV4B).

Similar to the Cgn‐cKO mouse, DPOAE and ABR thresholds were elevated in Cgn ^delG/delG mice at 2 and 6 months of age (Fig 5D–G). At 2 months of age, Cgn ^delG/delG mice exhibited significant decreases in ABR P1 amplitudes (32 kHz) (Fig 5H). Although hearing thresholds in the heterozygous Cgn ^delG/+ mice were normal, the ABR P1 amplitudes (32 kHz) of the Cgn ^delG/+ mice were significantly decreased at 6 months (Fig 5I), but not at 2 months (Fig 5H). These results indicated that the Cgn delG mutation results in a dose‐dependent and progressive hearing impairment in the knockin mice. Similar to the Cgn‐cKO mice, OHCs were also lost at high frequency in the Cgn ^delG/delG mice (Fig 5J and K). Together, these data show that the reduced expression of CGN in the Cgn ^delG/delG mice results in hearing impairment and loss of OHCs. However, OHCs of the heterozygous knockin mice were not lost, suggesting that other hair cell pathologies may also contribute to the high‐frequency hearing loss in Cgn ^delG mice.

Interestingly, CGN is also highly expressed in the apical junctions of the utricular sensory epithelia of wild‐type mice (Fig EV5A and B) but reduced in Cgn ^delG mice (Fig EV5C). However, balance function behavior assessed by rotarod tests is normal in the mutant mice (Fig EV5D). These results suggest that although CGN is widely expressed in sensory epithelia of cochlea and utricle, it is essential for normal auditory function but dispensable for vestibular function, consistent with a lack of overt vestibular symptoms in human patients.

Cgn ^delG knockin mice have increased sensitivity to noise exposure

To evaluate whether carrying the Cgn ^delG mutation alters the sensitivity to environmental insults, 2‐month‐old mice were exposed to broadband noise at 100 dB SPL for 2 h (Fig 6A). This noise exposure results in temporary shifts of auditory thresholds, which typically recover to baseline after 1 week in wild‐type mice (Cgn ^+/+ ). In contrast, DPOAE thresholds, ABR thresholds, and ABR P1 amplitudes (32 kHz) of the Cgn ^delG/delG mice exhibited larger impairment after noise exposure (Fig 6B–D). Remarkably, the heterozygous Cgn ^delG/+ mice also showed significant hearing impairment after noise exposure in all three auditory measures (Fig 6B–D). Consistently, OHCs were lost at high frequency (32 kHz) in Cgn ^delG/+ and Cgn ^delG/delG mice after noise exposure (Fig 6E and F). These data indicate that both Cgn ^delG/+ and Cgn ^delG/delG mice are more sensitive to moderate noise insults than wild types, and that noise exposure exacerbates overall hearing impairment in the heterozygous Cgn ^delG mice.

Cgn ^delG mice show normal synaptic density, tight junction, and microtubule structures

Since the reduced ABR P1 amplitudes that we observed in Cgn ^delG mice are often associated with loss of synapses formed between IHCs and SGN fibers, a.k.a. cochlear synaptopathy (Kohrman et al, 2020), we compared the number of ribbon synapses in wild‐type and Cgn mutants (Appendix Fig S4A). The synaptic ribbon density in Cgn ^delG mice was comparable to that in wildtype mice (Appendix Fig S4B), indicating that the hearing impairment of the Cgn ^delG mice may not be due to synaptic loss.

Cingulin is specifically localized at the apical junctions of cells (Citi et al, 2012) (Fig 3E) and the circumferential belts in the organ of Corti (Fig 2F and G). It has been suggested that CGN might stabilize tight junction structures by regulating the activity of Rho family GTPases (Gonzalez‐Mariscal et al, 2014). Therefore, we speculated that hearing impairment in Cgn mutant mice might be related to the disruption of its regulatory function in tight junctions. Therefore, we evaluated the expression of several markers associated with tight junction formation in MDCK cells overexpressing wildtype and mutant human CGN plasmids. Overexpression of the mutant human CGN had no significant effect on mRNA and protein expression of tight junction markers, Occludin and ZO‐1 (Appendix Fig S5A and B). mRNA expression levels of other tight junction‐related markers (Claudin‐1, Claudin‐3, Claudin‐5, ZO‐2, ZO‐3, GEF‐H1, RhoA, and ZONAB) were also unaffected (Guillemot & Citi, 2006) (Appendix Fig S5B). We next used shRNAs reported in a previous study to knock down (KD) endogenous Cgn in MDCK cells (Guillemot & Citi, 2006) (Appendix Fig S5C), and examined whether CGN deficiency would affect the formation of tight junctions (Appendix Fig S5D and E). Again, Cgn KD in MDCK cells did not affect expression of the tight junction markers (Appendix Fig S5F). Lastly, we examined the expression of ZO‐1, a marker of tight junction formation and maintenance, in cochleae of the Cgn ^delG and Cgn‐cKO mice. The expression patterns of ZO‐1 in cochleae of the two mutant mice were also similar to the wild type (Appendix Fig S5G and H), suggesting that the auditory phenotypes of the Cgn mutant mice may not be caused by tight junction abnormalities.

Cingulin has been reported to regulate the localization of microtubules at tight junctions (Yano et al, 2018) and microtubules are widely expressed in the organ of Corti (Steyger et al, 1989). During inner ear development, each hair cell has a single kinocilium, which is a microtubule‐based organelle. Kinocilia spontaneously degenerate postnatally and are thought to play an important role in the proper orientation of the actin filament‐based stereocilia bundles in mature hair cells (Bieniussa et al, 2023). In mature hair cells, microtubules are mainly found in the apical and lateral membrane regions of the cell (Steyger et al, 1989). We examined microtubule expression in mature cochlea from 2‐month‐old Cgn‐cKO and 7‐month‐old Cgn ^delG mice. However, both types of Cgn mutant mice exhibited normal microtubule structures (Appendix Fig S6A and B). We also examined kinocilium position in immature cochlea of P0 Cgn ^delG mice and found no significant difference between the wild‐type and Cgn ^delG mice (Appendix Fig S6C). These data indicate that abnormal expression of CGN had no obvious effects on microtubule expression patterns, and that the hearing impairment in Cgn mutant mice thus may not be related to microtubule alterations in the cochlea.

Cgn ^delG mice display abnormal morphology of the hair cell cuticular plates

Cingulin is considered a classical actin‐binding protein and its N‐terminal globular head domain has high binding affinity for actin filaments, thus serving as an important contributor to epithelial structural formation (Cordenonsi et al, 1999; Yano et al, 2018). Therefore, we examined the effects of wild‐type and mutant CGN on actin dynamics using the SRF‐RE luciferase reporter system (Miralles et al, 2003) (Fig 7A). Expression of wild‐type CGN significantly enhanced actin polymerization. Moreover, co‐expression of wild‐type CGN and RhoA, a member of Rho GTPase family, synergistically promote actin polymerization (Fig 7B). In contrast, the mutant CGN (p.L1110Lfs*17) showed significantly reduced actin polymerization activity (Fig 7C), likely due to its reduced expression levels and abnormal subcellular localization.

We also used fluorescent labeled‐Phalloidin to visualize F‐actin structures present in the cuticular plates and hair cell stereocilia in wild‐type and mutant mice. Interestingly, in addition to the loss of the OHCs (visualized as loss of F‐actin signals), the cuticular plates of the surviving OHCs appeared abnormal in the Cgn ^delG cochleae (Figs 5J and EV4B). To further interrogate this morphological feature, we immunostained the organ of Corti for two cuticular plate‐specific markers, LMO7 and Spectrin β II (Du et al, 2019; Liu et al, 2019). The areas of both cuticular plates and circumferential belts from three rows of OHCs were also quantified (Fig 7D). Bases on the LMO7 immunostaining, the morphology of the cuticular plates in Cgn ^delG/delG mice was significantly altered compared to the wild‐type mice at both basal (32 kHz, Fig 7E and F) and middle (16 kHz, Appendix Fig S7A) cochlear regions. Quantitative results showed that the cuticular plate areas of Cgn ^delG/delG mice decreased (Fig 7G and Appendix Fig S7B), while the circumferential belt areas increased (Fig 7H and Appendix Fig S7C) relative to those of wild‐type mice. Importantly, the relative areas of the cuticular plates were significantly reduced (Fig 7I and Appendix Fig S7D). Similar results were obtained with the Spectrin β II antibody (Fig 7J–N and Appendix Fig S7E–H). Together, these data suggest that the Cgn mutant reduces actin polymerization and alters the morphology of the actin‐rich cuticular plates of the OHCs.

Cgn ^delG mice display abnormal hair cell bundle morphology

As the actin filaments of stereocilia bundles are rooted in the cuticular plate, we predicted that the altered cuticular plates in Cgn mutant mice might affect the morphology of their associated bundles. Scanning electron microscopy (SEM) revealed a sporadic OHC bundle loss at 32 kHz cochlear region in 2‐month‐old Cgn ^delG/delG but not Cgn ^+/+ mice (Fig 8A), consistent with loss of OHCs in these mice observed previously (Fig 5). The morphology of the remaining hair bundles in Cgn ^delG/delG cochlea is largely indistinguishable from that in Cgn ^+/+ cochlea, with neither orientation deficits nor stereocilia degeneration detected in either IHCs or OHCs (Fig 8A–C). Interestingly, the demarcation of outer hair cell apical surfaces appeared to be more rounded in the Cgn ^delG/delG mice (Fig 8C), supporting the notion that cuticular plate morphology was altered based on LMO7 and Spectrin β II immunostainings (Fig 7).

In Cgn ^+/+ mice, OHC hair bundles were organized into the normal V‐shaped morphology, with an increase in bundle opening angles from apical (8 kHz) to basal (32 kHz) cochlear regions (Fig 8C, left). In contrast, the opening angles of the V‐shaped OHC bundles were significantly reduced in Cgn ^delG/delG mice (Fig 8C, right). To quantify the opening angles of the OHC bundles, the bundle width and the height of the bundle arc were measured for individual OHCs, and the width index calculated (bundle width/height of bundle arc) (Fig 8D). The width indices of bundles from all three rows of OHCs were significantly smaller in the Cgn ^delG/delG mice (Fig 8C, right) than those of Cgn ^+/+ mice at 8 kHz (Fig 8E), 16 kHz (Fig 8F), or 32 kHz (Fig 8G). Together, the data suggest that the altered cuticular plates affect the morphology of hair bundles in Cgn mutant OHCs, which may contribute to the hearing loss observed in Cgn mutant mice and the affected human patients.

Subjects and clinical evaluations

We recruited a three‐generation Chinese Han family with postlingual autosomal‐dominant non‐syndromic hearing loss. Twelve members (seven affected and five unaffected) participated in the present study, informed consent was signed by all participating subjects. The experiments conformed to the principles set out in the WMA Declaration of Helsinki and the Department of Health and Human Services Belmont Report. This study was approved by the ethics committee of Nanjing Drum Tower Hospital, the Affiliated Hospital of Nanjing University Medical School (2019‐170 and 2021‐122‐02).

Hearing levels of all participating members were measured by pure tone audiometry. For affected members, a complete medical history and physical examination were performed to exclude the possibility of environmental causes or syndromic hearing loss. For the proband, additional auditory evaluations were performed including distortion product otoacoustic emission (DPOAE), auditory steady‐state response (ASSR), auditory brainstem response (ABR), temporal bone high‐resolution computed tomography scanning (HRCT), and internal acoustic meatus magnetic resonance imaging (MRI).

Exome sequencing (ES) and linkage analysis of the Chinese family

All the 12 participating members were included in the ES study. Genomic DNA samples were extracted from whole blood samples of the participating members.

The exomes and flanking intronic regions from whole blood DNA samples were captured by Agilent SureSelect Human All Exon Kit (Agilent Technologies). The captured DNA was sequenced on Illumina HiSeq 4000 sequencing platform (Illumina). Bioinformatics were aligned to NCBI build37/hg19 assembly using the BWA (0.7.12) software. Each sample was covered to an average sequencing depth of at least 100x. SNPs and indels were identified using GATK HaplotypeCaller software. Candidate pathogenic variants were defined as nonsense, missense, splice‐site, and indel variants with allele frequencies of 0.001 or less in public variant databases dbSNP, 1000 Genomes, ESP6500, nci60, GnomAD and in disease database of COSMIC, Clinvar, OMIM, GWAS. Genotypes distributed in every 0.3 cM of genomic region were chosen for calculation of the logarithm of odds (LOD) scores using the Merlin v. 1.1.2 parametric linkage analysis package.

ES of selected Spanish subjects

Exome sequencing was performed on the 96 Spanish index cases using the Agilent SureSelect V6 exome capture‐based method in a NovaSeq6000 platform (Illumina, 150 PE, 18Gb/sample). The sequence data were mapped against the human genome sequence (build GRCh37/hg19), and data analysis was performed using the DNANEXUS' software that enables the single nucleotide variations (SNVs) and the copy number variation (CNV) analysis of the targeted exonic sequences. The analysis of the variant calling files (VCF) was restricted to the cingulin gene (CGN) and variant prioritization was carried out using a custom filtering strategy (Morin et al, 2020).

Sanger sequencing

Possible pathogenic variants from the ES analysis were evaluated by computational tools, including Mutation Taster, PolyPhen‐2, and SIFT. Candidate pathogenic variants were further genotyped for the 12 family members by Sanger sequencing.

Mouse models and genotyping

Cgn flox mice, named Cgn ^fl , were generated by CRISPR‐Cas9 genome‐editing technology on C57BL/6J background, by inserting LoxP fragments at both ends of the exon 6 and exon 21 of mouse CGN (ENSMUSG00000068876) (Gempharmatech Inc, China). Genotypes of Cgn ^fl mice were identified by a primer pair targeting the LoxP insert (forward primer: 5′‐CTG GGC TGG GCT CTT CTA TGT AG‐3′, reverse primer: 5′‐CTA GGG ATT GAA CCC AGA ACT TGA C‐3′).

Cgn disease mutation knockin mice, named Cgn ^delG , were generated by CRISPR‐Cas9 genome‐editing technology on C57BL/6J background. A repair DNA harboring the G‐base deleted in the identified human patients was used as donor template. The mutant allele was amplified using a specific forward primer (5′‐ACT CTG CTT AGC TAA CAT TAC GT‐3′) paired with the reverse primer (5′‐GCA GGG CTG GCT GGG ATC CT‐3′). The wildtype‐ allele was genotyped in a similar manner using the same reverse primer combined with a specific forward primer for the wildtype‐ allele (5′‐ACT CTG CTT AGC TAA CCC TGA GG‐3′).

Pou4f3 ^EGFP‐creER knockin mice were generated as described previously (Du et al, 2020). Genotypes of Pou4f3 ^EGFP‐creER mice were identified by three primers (forward primer: 5′‐GTG GGG GAG AGG GGA GGC AG‐3′, the wildtype allele‐specific reverse primer: 5′‐GAG AGA GCG CGG GGG AGA CA‐3′, the mutant allele‐specific reverse primer: 5′‐CGG TTC ACC AGG GTG TCG CC‐3′). Cgn hair cell‐specific conditional knockout mice (Cgn ^fl/fl ;Pou4f3 ^creER/+ ), named Cgn‐cKO, were generated by mating Cgn ^fl/fl with Pou4f3 ^EGFP‐creER mice. For Cre recombinase activation, pups at P3 to P6 were intraperitoneally injected daily with 33 mg/kg tamoxifen (T5648, Sigma) dissolved in corn oil.

C57BL/6J were purchased from Gempharmatech Inc, China. Both male and female mice were used in this study. Animals were assigned to control or experimental groups based on their genotypes. For noise exposure experiments, animals of the same genotype were randomly assigned to control or noise exposure groups. Genotypes of the animals were blinded to the experimenters. All animal procedures were approved by the Institutional Animal Care and Use Committee of Model Animal Research Center of Nanjing University, China with protocol approval number #WGQ04.

Plasmid constructions

Human CGN (NM_020770.3) and p.Leu1110Leufs*17 variant (C‐terminal Flag‐tagged or N‐terminal EGFP‐tagged) was cloned into pcDNA3.1 vector using EcoRV (R3195V, New England Biolabs) site. Mouse Cgn (NM_020770.3) (C‐terminal Flag‐tagged) was also cloned into pcDNA3.1 vector using EcoRV site. All mammalian expression plasmids were amplified in DH5α bacteria. All constructs were verified by sequencing.

Cell culture and transfection

MDCK (ATCC: CCL‐34) and COS‐7 (ATCC: CRL‐1651) cells were cultured in Dulbecco's Modified Eagle Medium (DMEM) (12800017, Gibco) supplemented with 10% fetal bovine serum (FBS) (40130ES76, Yeasen) and penicillin–streptomycin (P/S) (E607011‐0100, Sangon Biotech). CACO2 (ATCC: HTB‐37) cell was cultured in DMEM supplemented with 20% FBS and P/S. HEK293T (ATCC: CRL‐3216) cell was cultured in DMEM supplemented with 10% FBS, GlutaMAX (35050079, Gibco), non‐essential amino acid (NEAA) (11140050, Gibco), sodium pyruvate (11360070, Gibco), and P/S. Cell lines were obtained from ATCC and were tested mycoplasma‐negative prior to experiments. Cell lines were transfected with indicated plasmids using Hieff Trans™ Liposomal Transfection Reagent (40802ES03, Yeasen) following the manufacturer's instructions.

Reverse transcription quantitative polymerase chain reaction (RT‐qPCR) analyses

Cultured cells or mouse tissues were homogenized and the total RNA extracted by RNAiso Plus kit (9109, Takara). The quantity and purity of samples were examined by determining absorbance at 260/280 nm by Nanodrop (Thermo Scientific). Reverse transcription reactions were performed by HiScript III RT SuperMix for qPCR (+gDNA wiper) (R323‐01, Vazyme). QPCR was performed with AceQ qPCR SYBR Green Master Mix (Q111‐02, Vazyme) on a LightCycler® 96 Instrument (Roche). QPCR reactions were performed with the primers listed in Appendix Table S2. Expression of target genes was normalized to the internal reference genes (GAPDH for human and mouse cells/tissues, Actb/β‐actin for the canine MDCK cells). Fold changes of expression were normalized to control treatments or wildtype CGN control samples.

Western blot analysis

Tissues were freshly isolated after sacrificing the mice and grinded into powders in liquid nitrogen. Cultured cells were carefully rinsed three times with phosphate‐buffered solution (PBS). Tissue powders or cells were extracted and homogenized with the lysis buffer containing 0.2% SDS (cells, lung, liver, brain, intestine, kidney) or 1% SDS (cochlea), 50 mM Tris–HCl pH 7.4, 150 mM NaCl, 1 mM ethylene diamine tetra‐acetic acid (EDTA) pH 8.0, 1% Triton X‐100 (A110694‐0500, Sangon Biotech) and 1 mM phenylmethylsulfonyl fluoride (PMSF) (ST005, Beyotime) containing the cOmplete Protease Inhibitor Cocktail (11697498001, Roche) at 4°C for 10 min.

Proteins were subjected to SDS–PAGE and transfer to PVDF membrane and analyzed by following antibodies: rabbit anti‐Cingulin (1:1,000, 21369‐1‐AP, Proteintech), mouse anti‐Cingulin (1:1,000, sc‐365264, Santa Cruz), mouse anti‐Flag (1:2,000, 30503ES60, Yeasen), rabbit anti‐EGFP (1:2,000, 31002ES60, Yeasen), rabbit anti‐ZO‐1 (1:1,000, 21773‐1‐AP, Proteintech), rabbit anti‐Occludin (1:1,000, 13409‐1‐AP, Proteintech), mouse anti‐β‐actin (1:20,000, 4970, Cell Signaling Technology) antibodies followed by incubation in HRP conjugated anti‐Rabbit (1:5,000, BS13278, Bioworld) or anti‐mouse (1:5,000, BS12478, Bioworld) secondary antibody for 2 h at room temperature (RT). The signals were visualized using ECL substrate (180‐5001, Tanon) on an automatic chemiluminescence/fluorescence image analysis system (Tanon 4600, Tanon).

Cgn shRNA lentiviral vectors and lentivirus packaging

For Cgn gene silencing, previously described target sequences for silencing canine CGN (Guillemot & Citi, 2006) were cloned into pLKO1, a lentiviral vector with puromycin resistance screening label, using EcoRI (R3101V, New England Biolabs) and AgeI (R3552S, New England Biolabs) sites. The lentiviral particles were produced by co‐transfecting each pLKO1 vector with other two lentiviral packaging helper plasmids (pSPAX2 package and pMD2.G envelope plasmids) in HEK293T cell. The supernatant was collected and concentrated by ultracentrifugation. The Cgn shRNA lentivirus were used to infect MDCK cells and puromycin was used to select infected cells to obtain stable shRNA‐expressing cell lines. Cgn knockdown efficiency in MDCK cell was determined by RT‐qPCR, Western blot and immunofluorescence.

SRF‐RE dual‐luciferase reporter assays

The pGL4.34[luc2P/SRF‐RE/Hygro] vector contains a serum response factor (SRF) response element (SRF‐RE) that drives transcription of the luciferase reporter gene luc2P in response to the activation of serum response factors through multiple signaling pathways, including RhoA GTPase activation. HEK293T cells in 24‐well plates were transfected with plasmids overnight as follows: pGL4.34[luc2P/SRF‐RE/Hygro] (Promega), pRL‐TK, and CGN (WT or mutant), with or without co‐transfecting the constitutively active RhoA. Transfected cells were serum starved for 6 at 18 h after transfection. Starved cells were lysed and the supernatants collected. The dual‐luciferase assay kit (RG089M, Beyotime) was used to detect SRF activity, an indirectly representation of actin polymerization levels.

Immunofluorescence

Isolated mouse cochleae were fixed in 4% paraformaldehyde (PFA) in PBS for 2 h at RT. Samples were then decalcified with 5% (w/v) EDTA for 4 days at RT. Decalcified cochleae can be stored in PBS at 4°C. After micro‐dissection of the cochleae, samples were permeabilized by freeze‐thawing in 30% sucrose and blocked in 5% normal horse serum (NHS) (008‐000‐121, Jackson ImmunoResearch) with 0.3% Triton X‐100 in PBS for 1 h. Then, the samples were incubated with primary antibodies (diluted in 1% NHS with 0.3% Triton X‐100 in PBS) overnight at 4°C. Next day, samples were incubated with corresponding secondary antibodies for 2 h at RT and mounted to cover slides.

Transfected cells were fixed on ice in 2% PFA for 15 min and then blocked in 5% NHS with 0.3% Triton X‐100 in PBS for 1 h at RT. Then, samples were incubated with primary antibodies (diluted in 1% NHS with 0.3% Triton X‐100 in PBS) overnight at 4°C. Next day, samples were incubated with corresponding secondary antibodies for 2 h at RT.

In this study, the primary antibodies used were rabbit anti‐Cingulin (1:200, 21369‐1‐AP, Proteintech; 1:200, A15489, ABclonal), mouse anti‐Cingulin (1:200, sc‐365264, Santa Cruz), rabbit anti‐ZO‐1 (1:200, 21773‐1‐AP, Proteintech), mouse anti‐Parvalbumin (1:500, sab4200545, Sigma), rabbit anti‐Myosin 7a (1:500, 25‐6790, Proteus Biosciences), rabbit anti‐Ctbp2 (1:200, 612044, Millipore), mouse anti‐LMO7 (1:50, sc‐376807, Santa Cruz), mouse anti‐Spectrin β II (1:100, sc‐136074, Santa Cruz), mouse anti‐Pou4f3 (1:200, sc‐81980, Santa Cruz), mouse anti‐Pan tubulin (1:1,000, 3873, Cell Signaling Technology), mouse anti‐Acetylated tubulin (1:1,000, T6793, Sigma), mouse anti‐Flag (1:1,000, 30503ES60, Yeasen), and rabbit anti‐EGFP (1:400, 31002ES60, Yeasen). The secondary antibodies used were Alexa Fluor® 488 AffiniPure Goat Anti‐Rabbit IgG (H+L) (1:500, 111‐545‐003, Jackson ImmunoResearch), Alexa Fluor® 594 AffiniPure Goat Anti‐Rabbit IgG (H + L) (1:500, 111‐585‐003, Jackson ImmunoResearch), Alexa Fluor® 647 AffiniPure Goat Anti‐Rabbit IgG (H+L) (1:500, 111‐605‐003, Jackson ImmunoResearch), Alexa Fluor® 488 AffiniPure Goat Anti‐Mouse IgG (H+L) (1:500, 115‐545‐003, Jackson ImmunoResearch), Alexa Fluor® 594 AffiniPure Goat Anti‐Mouse IgG (H+L) (1:500, 115‐585‐003, Jackson ImmunoResearch), Alexa Fluor® 647 AffiniPure Goat Anti‐Mouse IgG (H+L) (1:500, 115‐605‐003, Jackson ImmunoResearch), Alexa Fluor™ 488‐conjugated goat anti‐mouse (IgG2b) (diluted 1:500; A21141, Life Technologies), Alexa Fluor™ 568‐conjugated goat anti‐mouse (IgG2b) (diluted 1:500; A21144, Life Technologies), and Alexa Fluor™ 568‐conjugated goat anti‐mouse (IgG1) (1:500; A21124, Life Technologies). F‐actin and nucleus were visualized with Rhodamine‐conjugated Phalloidin (1:1,000, PHDR1, Cytoskeleton), Actin‐Tracker Green (1:500, C1033, Beyotime Biotechnology), and DAPI (1:1,000, 28718‐90‐3, Roche), respectively.

ImageJ (https://imagej.net/software/fiji) measure line plugin was used to map cochlear length to cochlear best frequencies. Samples were examined using a Leica SP5 confocal microscope (Leica TCS SP5, Leica Microsystems), Zeiss LSM 800 confocal microscope (ZEN 2.1, Carl Zeiss) or inverted fluorescence microscope (XD, SOPTOP). ImageJ was used for image processing.

Scanning electron microscopy (SEM)

Temporal bones of 2‐month‐old mice were fixed with 2.5% glutaraldehyde in 0.1 M phosphate buffer at 4°C overnight. The cochleae were dissected out and postfixed with 1% osmium tetroxide in 0.1 M phosphate buffer at 4°C for 2 h. The samples were then dehydrated in ethanol, followed by critically point drying using a Leica EM CPD300 (Leica, Germany). After sputter coating with platinum to a thickness of 15 nm using a Cressington 108 sputter coater (Cressington, United Kingdom), the samples were imaged with a Quanta250 field‐emission scanning electron microscope (FEI, The Netherlands). The bundle width is defined as the distance between the two ends of each hair bundle. The height of the bundle arc is defined as the distance from the vertex to the line connecting the two ends of each hair bundle. Width index of each hair bundle is calculated as the bundle width divided by the height of the bundle arc.

DPOAE and ABR measurements

Mice were anesthetized by intraperitoneal injections of 375 mg/kg Avertin (Sigma) before ABR and DPOAE tests.

For ABR measurements, three needle electrodes, A, B, and G, were placed under the skin: A at the dorsal midline between the two ear flaps, B behind the right pinna, and G at the base of the tail. ABR potentials were evoked with 5‐ms tone pips (0.5‐ms rise‐fall, with a cos2 envelope, at 33/s) delivered to the eardrum at log‐spaced frequencies at 5.6, 8, 11.3, 16, 22.6, and 32 kHz. The responses were amplified (10,000×) and filtered (0.3–3 kHz) with an analog‐to‐digital board in a PC‐based data‐acquisition system. Sound pressure levels (SPLs) were raised at 5 dB‐step from 10 to 80 decibels (dB). At each SPL, 1,024 responses were averaged (with alternating polarity) after “artifact rejection”. DPOAEs were measured immediately following ABR tests. The DPOAE signal in response to primary and secondary tones with frequencies f1 and f2 respectively was recorded at the third frequency (2 × f1)−f2, with f2/f1 = 1.2, and the f2 level 10 dB lower than the f1 level. SPLs at the ear canal was amplified and digitally sampled at 4‐ms intervals. DPOAE thresholds were defined as the f1 level required to evoke a response at −10 dB SPL. Both DPOAE and ABR recordings were carried out using EPL cochlear function test suite software (Mass Eye and Ear, Boston, MA, USA). ABR peak 1 amplitudes were analyzed with ABR peak analysis software (Mass Eye and Ear).

Noise exposure

Both Cgn wild‐type and mutant mice were subjected to noise exposure at 2 months of age. Individual mouse was housed within small cells in a subdivided cage on a rotation station, which is suspended in a reverberant chamber. Noise exposure experiment was carried out using LabState software (AniLab Software & Instruments, China) to produce 2–20 kHz broadband noise at 100 dB for 2 h.

Rotarod tests

Vestibular function of 2‐month‐old wildtype and mutant mice was performed using a rotarod system (ZB‐200, Chengdu Techman Software, China). Mice were trained each day for 3 consecutive days at 15 round per minute (rpm) for 240 s prior to the experiment. Mice were subjected to three different rotarod test protocols: (i) 12 rpm fixed speed for 240 s; (ii) 20 rpm fixed speed for 240 s; and (iii) 0–44 rpm accelerating speed with 1 rpm/s and maintained at 44 rpm from 44 to 60 s. All three test protocols were performed for three times with 30–60 m resting time in between. The latencies to fall from the rotarod were then recorded and presented as averages of the three tests for each test protocol.

Statistical analysis

Statistical tests were performed using GraphPad Prism 9. Specific statistical tests used in each experiment and replicates were described in figure legends. Distribution of the data was assumed to be normal, but this was not formally tested. No statistical methods were used predetermine sample sizes, but our sample sizes are similar to those reported in our previous publications (Wan et al, 2014; Wan & Corfas, 2017; Zhu et al, 2020). The genotypes of the control and experimental animals were blinded to the investigators until the completion of data acquisition and analysis. For noise exposure experiments, mice with the same genotype were randomly assigned to control or noise exposure conditions.