Two Novel Compound Heterozygous CDH23 Mutations Underlying Non-Syndromic Hearing Loss Authors: Pengcheng Xu, Qi Liao, Yun Lin, Ning Zhao, Lin Li Published: 2026 ABSTRACT AbstractPurposeHereditary hearing loss is a genetically heterogeneous disorder, with more than 120 genes implicated in the non-syndromic forms. The genetic cause of non-syndromic hearing loss (NSHL) in a Chinese Han family was investigated, and the function of the identified mutation was characterized.MethodsThe family members were clinically evaluated. Targeted next-generation sequencing of 414 deafness-related genes was performed on the proband. The identified variants were validated in the family members of the proband using Sanger sequencing and PCR. The functional impacts of the variants on splicing were predicted using SpliceAI and MaxEntScan.ResultsTwo compound heterozygous CDH23 mutations (c.7054+1G>A and c.4209+794_4825 del) were identified in a thirteen-month-old girl with NSHL. Sanger sequencing confirmed recessive inheritance, with each unaffected parent carrying a single heterozygous variant. The c.7054+1G>A variant is a canonical splice donor variant predicted to severely disrupt splicing (SpliceAI donor loss score: 0.99; MaxEntScan score reduced from 10.28 in the wild-type to 2.10 in the mutant, a 79.57% reduction). The c.4209+794_4825 del is a 6.45 kb heterozygous deletion encompassing exons 35–38, predicted to resulted in the in-frame skipping of exons 35–38. AlphaFold-based structural modeling revealed marked conformational changes in the mutant extracellular domain, supporting the hypothesis that these variants compromised CDH23-mediated tip-link formation and mechanotransduction.ConclusionWe identified two novel compound heterozygous CDH23 mutations (c.7054+1G>A and c.4209+794_4825 del) in a Chinese Han family, expanding the known spectrum of CDH23 mutation and adding to the understanding of the genotype–phenotype correlations in NSHL.Keywords: non-syndromic hearing loss, CDH23, autosomal recessive, compound heterozygous mutation, Chinese Han Introduction Hearing loss (HL) is the most prevalent sensory disorder globally, affecting nearly 1 in 100 newborns.1 Hereditary HL exhibits considerable genetic heterogeneous, encompassing autosomal dominant, autosomal recessive, X-linked, and mitochondrial inheritance,2 with autosomal recessive transmission accounting for nearly 80% of inherited cases.3 Among these, non-syndromic hearing loss (NSHL) constitutes the most inherited hearing impairment cases. Given that over 120 causative genes have been implicated, routine sequencing-based approaches remain time-consuming and costly. NSHL presents a formidable clinical challenge not primarily in treatment, but in molecular diagnosis. The CDH23 gene, located on chromosome 10q21-q22, encodes cadherin-related 23, a critical component of stereocilia tip links in inner ear hair cells.4 CDH23 mutations impair the formation of tip links and alter the hair-bundle architecture, resulting in dysfunctional mechanotransduction: transduction channels fail to remain open under resting conditions.5 Pathogenic variants in CDH23 variants can lead to two distinct clinical entities: syndromic hearing loss (Usher syndrome type 1D (USH1D)) and the non-syndromic hearing loss phenotype (DFNB12).6,7 Intriguingly, USH1D is typically caused by biallelic truncating mutations that result in complete protein loss, whereas DFNB12 is more frequently associated with missense variants that preserve residual protein function.8 Variants in the CDH23 gene are known to cause a broad spectrum of hearing loss phenotypes,5 and considerable cases originate from East Asia. Distinct from Korean patients carrying the prevalent founder variant p.P240L,9 the mutational spectrum is highly diverse among Chinese Han patients and lacks predominant hotspot variants. Thus, comprehensive screening of the entire CDH23 coding region is necessary for molecular diagnosis in Chinese Han patients. In this study, we investigated a Chinese Han family with suspected hereditary NSHL by integrating CNV validation (quantitative PCR and long-range PCR), breakpoint delineation via Sanger sequencing, and advanced splice-site prediction algorithms (MaxEntScan and SpliceAI). Our objective was to identify the compound heterozygous mutations causing hearing loss in this pedigree. Collectively, our findings may not only expand the mutational landscape of CDH23 in the Chinese population but also provide tangible evidence to refine precision genetic counseling and clinical management for NSHL. Methods Patients and Clinical Evaluations This study included a Han Chinese family with recessive hereditary hearing loss. The affected family comprised three members across two generations (Figure 1). The hearing of patient (II-1) was evaluated using auditory brainstem response testing. The hearing abilities of the parents (I-1 and I-2) were assessed using pure-tone audiometry (PTA). PTA was calculated as the average of the hearing threshold of patients at 500, 1000, 2000, and 4000 Hz. All participating family members were clinically assessed through a detailed medical history interview and a comprehensive physical examination with otoscopic inspection to exclude causes of hearing loss acquired via infection, trauma, or other external factors. Tympanometry was used to assess middle ear function, and outer hair cell activity in the cochlea was measured using distortion product otoacoustic emissions. Temporal bone computed tomography (CT) scans were captured to evaluate the middle and inner ear anatomy. Patient II-1 (13 months old) did not undergo vestibular and fundus examinations because of her inability to cooperate due to age. This study was approved by the Ethics Committee of Shanghai Ninth People’s Hospital, Shanghai Jiao Tong University School of Medicine. All procedures were conducted in accordance with the principles of the Declaration of Helsinki. Written informed consent was obtained from all adult participants and the parents of the child participant. Pedigrees and genotypes of the affected family. The black filled circle denotes the affected female proband (II-1). Roman numerals I and II represent family generations, and the arrow points to the proband. The proband (II-1) carries both c.7054+1G>A and c.4209+794_4825del (compound heterozygous). The father (I-2) carries the c.7054+1G>A variant only. The mother (I-1) carries the c.4209+794_4825del variant only. Mutation Identification Genomic DNA was extracted from whole-blood samples of all participants using a commercial Blood DNA kit (TIANGEN BIOTECH, Beijing, China) according to the standard protocol provided by the manufacturer. Targeted next-generation sequencing (NGS) was then conducted for 414 known deafness-related genes (Table S1). NGS was performed using a capture panel covering approximately 400 candidate genes, with a total target region length of 1.87 Mb. The average sequencing depth of the deleted region was 530.92, with target coverage of 99.52% at 10× and 99.29% at 20×. Targets were enriched by capturing the coding exons and adjacent splice regions of the genes of interest using a designed chip library. The enriched DNA libraries were qualitatively assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies, CA, USA) and subjected to circular sequencing on a BGISEQ-500 platform. The raw data were then aligned to the human reference genome (GRCh38/hg38) using the Burrows-Wheeler Aligner (BWA). Variants were called to identify all discrepant nucleotide sites. PCR amplification and Sanger sequencing were used to validate the putative mutations. The impact of the identified variants on pre-mRNA splicing was assessed using SpliceAI (v1.3.1); the strength of splice sites was predicted using MaxEntScan.10 To model the splice acceptor strength of the novel junction created by the c.4209+794_4825del deletion, the 23-nucleotide sequence immediately upstream of exon 34 (comprising the native 3′ splice site of intron 34) and the 23-nucleotide sequence immediately downstream of exon 39 (comprising the native 3′ splice site of intron 38) were concatenated in silico to generate the artificial junction sequence. This sequence was then submitted to MaxEntScan to evaluate the strength of the putative novel acceptor site formed after the deletion. The prediction results are provided as numerical scores, with higher scores indicating a higher likelihood of the site functioning as a canonical splice site. Results Clinical Characterization Proband II-1 was a thirteen-month-old girl with profound sensorineural hearing loss. The two parents (I-1, 28 years old and I-2, 30 years old) had normal hearing. The temporal bones of the proband were structurally normal upon CT. Pure-tone audiometry indicated that the parents had bilateral PTA thresholds above 20 dB nHL, along with type A tympanograms in both ears. Distortion product otoacoustic emissions were present in both ears, consistent with normal cochlear function. The tympanograms of the proband were bilaterally type A, indicating normal middle ear function. No waveforms were detectable on ABR testing up to a stimulation intensity of 100 dB nHL. Auditory steady-state response testing revealed average hearing thresholds across frequencies of 500, 1000, 2000, and 4000 Hz at approximately 100 dB nHL. The proband was diagnosed as having profound sensorineural hearing loss based on these findings. Given that CDH23 mutations may cause both DFNB12 and USH1D, we performed a follow-up. During a recent telephone follow-up, the proband’s parents reported no observable abnormalities in visual function. It is acknowledged that the onset of retinitis pigmentosa in USH1D is typically progressive and may not manifest until the second decade of life. The long-term regular ophthalmological and vestibular follow-up is warranted. Mutation Analysis The 414 known deafness genes were subjected to targeted NGS for proband II-1. We identified 16 candidate variants (Table S2). Compound heterozygous mutations in CDH23 (c.4209+794_4825 del and c.7054+1G>A) have been identified as putative pathogenic variants according to the American College of Medical Genetics and Genomics guidelines (ACMG).11 The splice site variant c.7054+1G>A was classified as Likely Pathogenic (PVS1, PM2, and PP3). The large deletion c.4209+794_4825del was similarly classified as Likely Pathogenic (PVS1 and PM2) (Table S3). The intrafamilial cosegregation of the variants was confirmed via Sanger sequencing and long-range PCR in all family members. We identified a variant, c.7054+1G>A, at the donor splice site of exon 50 in CDH23. The candidate mutations were consistent with recessive inheritance, as the unaffected parents were heterozygous carriers of a single mutation. Figure 2 shows that the c.7054+1G>A variant was heterozygous in the proband (II-1) and her father (I-2) but absent in the mother (I-1). The Sanger sequencing results of c.7054+1G>A mutation in CDH23. The red downward arrows indicate the position of the c.7054+1G>A variant. Chromatograms of family members II-1, I-2 and I-1 are displayed from top to bottom. SpliceAI predicted that the c.7054+1G>A variant would highly disrupt splicing, with a donor loss (score of 0.99), indicating a severely disrupted canonical splice site. The MaxEntScan results indicated that the variant substantially reduced the strength of the canonical splice donor site, with the score decreasing from 10.28 in the wild type to 2.10 in the mutant. This reduction of 79.57% suggested that the original splicing mechanism was severely impaired. Furthermore, SpliceAI predicted the de novo creation of a novel donor site (Donor Gain score = 0.69) 30 nucleotides upstream of the variant (–30 bp). These aberrant splicing events may lead to the generation of an abnormal protein product or a complete loss of protein expression due to nonsense-mediated mRNA decay (NMD). The NGS results indicated a copy number variation in CDH23, delineating an approximate region for the mutation. Furthermore, the qPCR results validated a heterozygous deletion designated c.4209+794_4825 del. c.4209+794_4825 del is a heterozygous deletion spanning exons 35–38 of CDH23. We mapped the breakpoint of the suspected heterozygous genomic deletion by designing a series of probes (G-P1, G-P2, and P1-P12) in the surrounding genomic region. The probes were designed for PCR amplification in patient II-1 and the parents (Figure 3A). The primer sequences used are listed in Table S4. On the 5’ side, primers P3, P5, and P11 were used to narrow down the upstream breakpoint. On the 3’ side, primers P4, P6, P7, P8, P9, P10, and P12 were employed to define the downstream breakpoint. CNV validation was performed using primers P1/P2 and G-P1/G-P2. Candidate breakpoint intervals were defined as the interval between adjacent probes with aberrant copy numbers (Figure 3B). One PCR amplification was performed if both breakpoints of the deletion were mapped within an interval of < 300 bp (Figure 3C). Identification of a novel c.4209+794_4825 del in the CDH23. (A) Probes designed for mapping the breakpoint of the heterozygous c.4209+794_4825 del in CDH23. Arrows represent qPCR probe locations. The 5’ breakpoint was mapped using primers P3, P5, and P11, while the 3’ breakpoint was defined using primers P4, P6, P7, P8, P9, P10, and P12. Primers P1/P2 and G-P1/G-P2 were used for validation. Yellow dashed box indicates the candidate deletion range, and green dashed lines mark the final refined deletion boundaries. (B) Relative copy number results for proband II-1 (blue), father I-2 (green), and mother I-1 (purple). (C) The position of CDH23 gene deletion on Chromosome 10. The black arrow indicates the fusion breakpoint of the deleted sequence. Red dashed boxes indicate the bases on the genomic breakpoints. Black bar indicates alignment position of the deletion. The heterozygous variant (c.4209+794_4825 del) was present in the proband (II-1) and mother (I-1) and absent in the father (I-2) (Figure 3B). The identified c.4209+794_4825 del variant constitutes a complex genomic rearrangement on chromosome 10, involving a 6.45 kb heterozygous deletion. This deletion spanned the genomic region from chr10:73,495,209 to chr10: 73,501,658 (GRCh38/hg38), resulting in the complete loss of exons 35–38 of the CDH23 transcript (NM_022124.6; Figure 3C). Table 1 shows that the native 3’ splice site for introns 34–38 exhibited high scores ranging from 7.95 to 14.82, confirming their robustness. However, the novel junction sequence formed between exons 34 and 39 after the large deletion yielded a low score of –4.37, which predicted a non-functional splice site. This finding provided computational evidence that the deletion predicted the in-frame skipping of exons 35–38 in the mature transcript, resulting in the protein variant p.1404_1605 del. We further employed AlphaFold (https://deepmind.google/science/alphafold/) to predict the three-dimensional (3D) structures of both the wild-type and mutant proteins, focusing on comparing the structural conformation of the mutant EC (extracellular cadherin) domain with that of the native counterpart. The results suggested that the c.4209+794_4825 del resulted in the elimination of four critical ECs repeats (EC35–EC38) within CDH23 (Figure 4). Table 1MaxEntScan Scores for Predicted 3’ Splice Site Strength of Gene IntronsIntron IDSequence Context (5′ to 3′)MaxEntScan ScoreRelative Strength38TGACCCACTCTTTTCCACAG GCC10.27Intermediate37CTCTCCTGCTCTCCTCCCAG AAC11.09Intermediate36AGTTGCCCTCCTCCTTGCAG GTT11.70Strong35ACCCTCTCTTCTCCCCACAG GTG14.82Very Strong34CTGACCACGTTCACCCTCAG GTC7.95Weak33TGCAATGGGAGTGGGTCCAG GGC−4.37Very Weak MaxEntScan Scores for Predicted 3’ Splice Site Strength of Gene Introns The changes of three-dimensional (3D) structures in the CDH23 protein domains induced by the c.4209+794_4825del variant. Discussion Hereditary hearing loss is a major cause of auditory impairment worldwide, accounting for 50–60% of all hearing loss cases. NSHL is the most prevalent form of hereditary hearing loss and is characterized by wide genetic heterogeneity, with over 120 genes implicated in NSHL. CDH23 is a gene associated with NSHL, contributing to syndromic (Usher syndrome type 1D) and non-syndromic (DFNB12) hearing loss, largely determined by the nature and combination of the variants. DFNB12 is usually linked to missense mutations that retain partial protein function, causing non-progressive or slowly progressive sensorineural hearing loss, without retinal or vestibular effects.12 USH1D is mainly caused by truncating variants or large deletions that fully disrupt CDH23 function, leading to congenital hearing loss, vestibular dysfunction, and retinitis pigmentosa with onset in the second decade of life.13,14 Although additional CDH23 variants are being identified;15,16 however, the complete mutational spectrum of CDH23 remains unclear. As such, novel CDH23 variants must be molecularly and functionally characterized. In our previous study, we identified pathogenic compound heterozygous variants in the TMC1 and MYO15A genes as causes of NSHL.17 The present study extends our ongoing investigation of hereditary hearing loss by examining an independent family and identifying two novel CDH23 variants. These findings further expand the mutational spectrum of CDH23 and provide additional evidence for genotype–phenotype correlations in CDH23-associated hearing loss. We used multiple molecular techniques, including using targeted NGS, long-range PCR, and Sanger sequencing, to validate the complex structural variants and splice-site mutations of CDH23. NGS technologies, such as targeted sequencing and whole-genome sequencing, have been developed, providing powerful and efficient approaches for identifying disease-associated genetic variants.18,19 Large deletions may not be captured with basic tests genetically complex disorders such as NSHL. NGS offers guidance for diagnosing and clinically managing hereditary hearing loss.2 An integrated strategy must be adopted for accurately detecting and validating large genomic deletions and canonical splice-site variants, such as those associated with NSHL.20–22 We identified two compound heterozygous CDH23 mutations (c.7054+1G>A and c.4209+794_4825 del) in a proband of a Chinese Han family with profound sensorineural hearing loss. CDH23 encodes a cadherin family protein that is essential for forming and maintaining stereociliary tip links in the inner ear hair cells. These structures are critical for mechanotransduction.23 CDH23 mutations disrupt stereociliary tip-link structures in the inner ear, leading to dysfunctional auditory transduction. Liao et al identified two CDH23 variants (c.2398–6G > A and c.6068C > A) that cause NSHL.24 Sun et al reported two heterozygous mutations in CDH23 (c.2651 A>G and c.2113 G>T) as causing deafness.25 More than 200 pathogenic variants of CDH23 are associated with USH1D and DFNB12;26 however, the functional consequences of splice-site mutations and large genomic deletions remain poorly understood. The c.7054+1G>A variant disrupts the canonical donor splice site of exon 50, as supported by the high SpliceAI and MaxEntScan scores in this study, leading to aberrant splicing, NMD, or an abnormal protein product. A CDH23 variant, c.2398–6G > A, disrupts normal mRNA splicing, leading to an abnormal splice product with a four-nucleotide (GCAG) insertion between exons 22 and 23. This variant leads to a loss-of-function CDH23 protein, comprising 804 amino acids rather than the complete 3354-amino-acid protein.24 In our study, the c.4209+794_4825 del results in an in-frame deletion of exons 35–38, which encode cadherin essential for protein–protein interactions and structural integrity. This genomic alteration was predicted to eliminate four critical extracellular cadherin (EC) repeats (EC35–EC38) within CDH23. The loss of these EC repeats likely has severe functional consequences because these EC repeats are essential for mediating precise homophilic protein–protein interactions and forming the rigid, rod-like structure of the protein. We postulate that this truncated protein is defective in molecular adhesive function, which is crucial for hair cell stereocilia tip-link formation, and mechanical integrity, leading to a destabilized structure that improperly transmits mechanical forces. Boehler et al27 similarly reported that the deletion of exons 11 and 12 in the murine Cdh23 gene introduced a premature stop codon in the corresponding EC domain. This genetic lesion produced a truncated protein and caused profound hearing and vestibular dysfunction, showing that disruptions in the integrity of the EC domains are responsible for the loss of CDH23 function in vivo. Li et al28 reported that deletion of exon 68 in Cdh23 primarily impairs tip-link stability, resulting in progressive hearing loss with largely preserved vestibular function. Collectively, deletion variation affecting different exons may give rise to phenotypic heterogeneity. The NM_022124.6: c.6068C > A CDH23 variant was predicted to change the three-dimensional structure, alter the conformation of the CDH23 protein, and interrupt protein stability and binding ability,24 which was similar with our AlphaFold finding of four ECs repeats elimination induced by c.4209+794_4825 del. Our finding of a deletion spanning exons 35–38 supported the notion that disruptions in this region were highly pathogenic and likely to cause severe auditory impairment. The compound heterozygosity of the c.7054+1G>A and c.4209+794_4825 del variants aligned with an autosomal recessive inheritance pattern, which is consistent with the genetic profile of DFNB12 hearing loss owing to CDH23 mutation. The unaffected parents were heterozygous carriers of one variant, further supporting their pathogenicity. The absence of these variants in the normal control cohorts of the ESP6500, ExAC, and 1000 Genomes databases underscores their rarity and pathogenicity. The splice-site variant likely induces aberrant splicing and NMD, whereas the multiexon deletion disrupts critical extracellular cadherin repeats, which are essential for mechanotransduction. Despite the absence of current retinal or vestibular abnormalities in the 13-month-old patient (II-1), the potential for USH1D syndrome, which presents prior to adolescence as visual impairment and retinal pigmentation, warrants further follow-up. The profound congenital hearing loss in the proband can be explained by the synergistic effects of the two variants. The c.7054+1G>A splice-site variant is predicted to trigger nonsense-mediated decay (NMD), leading to a complete loss of protein expression from this allele. Concurrently, the c.4209+794_4825del removes four critical EC repeats (EC35–EC38), likely producing a truncated and misfolded protein that is either degraded or functionally inert. The combination of a null allele and a structurally defective allele is predicted to result in a severe depletion of functional CDH23 protein in hair cell stereocilia. This may disrupt calcium-dependent homophilic adhesion at tip links, impair mechanoelectrical transduction, and ultimately cause the early-onset, profound hearing loss.28 Given that USH1D-associated retinitis pigmentosa usually presents after the first decade of life,29 the absence of retinal symptoms in this 13-month-old patient does not exclude future manifestation. Thus, long-term ophthalmological and vestibular follow-up is warranted. There are some limitations in this study. First, although we attempted to validate the splicing effect of the c.7054+1G>A variant by RT-PCR using RNA extracted from the proband’s peripheral blood, we were unable to obtain a specific CDH23 amplicon. This is likely due to the highly tissue-specific expression of CDH23, that predominantly expressed in the inner ear and retina. We therefore relied on multiple in silico prediction tools (SpliceAI, MaxEntScan and AlphaFold), which consistently predicted complete disruption of the canonical donor splice site, supporting the pathogenic classification. Future functional validation studies employing minigene assays or cDNA analysis from inner-ear-derived cells would be valuable to confirm our findings. Second, comprehensive ophthalmologic examinations were not performed at the time of diagnosis, as the patient was only 13 months old and exhibited no obvious visual symptoms. Future longitudinal follow-up with full ophthalmologic and vestibular evaluations is essential to monitor for potential syndromic manifestations. Third, our findings are derived from a single family, and thus the observed genotype–phenotype correlation may not be generalizable to other populations. Thus, functional validation in larger cohorts is warranted to confirm the pathogenic roles of these variants. In conclusion, we report two novel compound heterozygous mutations in CDH23, namely c.7054+1G>A (a canonical splice donor variant) and c.4209+794_4825del (a multi-exon deletion involving exons 35–38) in a Chinese Han family with NSHL. Both variants were classified as “Likely Pathogenic” according to ACMG/AMP guidelines (PVS1, PM2, PM3, PP5). Although the current phenotype is limited to hearing loss, the patient remains at risk for USH1D. The lacking of RNA-level functional validation was a limitation in our study. The combined in silico predictions, ACMG criteria, and AlphaFold-based structural modeling may support their pathogenicity. Our findings may broaden the known mutations of CDH23 and provide functional insights into the pathogenicity of these variants, supporting the essential role of CDH23 in auditory function. Functional validation in large cohorts are needed in the near future. Funding Statement This study was supported by grants from Wuxi Medical Center, Nanjing Medical University (WMCG202321) and Wuxi People’s Hospital (YZ-QNBJ-XPC-2025). The funder had no role in the design, data collection, data analysis, and reporting of this study. Data Sharing Statement All data generated or analyzed during this study are included in this article and its supplementary material files. Further enquiries can be directed to the corresponding author. Ethical Statement This study protocol was reviewed and approved by the Ethics Committee of Shanghai Ninth People’s Hospital, approval number SH9H-2019-T225-1. Written informed consent was obtained from all adult participants and the parents of the child participant for participation in the study and publication of the family case history and genetic findings. Author Contributions All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work. Disclosure The authors have no conflicts of interest to declare. References Associated Data Data Availability Statement All data generated or analyzed during this study are included in this article and its supplementary material files. Further enquiries can be directed to the corresponding author.