This study included a large Han Chinese pedigree with sensorineural hearing loss, consisting of 72 family members. Each member provided demographic information, age of onset, illness development, obstetric history, noise exposure, ototoxic medication use, head trauma, infectious infections, family history, and other pertinent clinical symptoms. Audiological assessments and otolaryngological examinations were conducted on probands, including pure-tone audiometry, impedance testing, distortion product otoacoustic emissions, auditory brainstem response, and temporal bone CT scans. Pure-tone audiometry was performed to assess the hearing of other family members using frequencies of 0.125 kHz, 0.25 kHz, 0.5 kHz, 1 kHz, 2 kHz, 4 kHz, and 8 kHz via air conduction bilaterally. The average values of the thresholds of air conduction were determined at 500–4,000 Hz to determine the degree of hearing loss in the family. The study protocol was approved by the Institutional Review Board of West China Hospital [2021 Audit (190)]. All family members provided written informed consent before participating in this study.

Approximately 5 mL of peripheral blood was collected from the proband and family members. DNA was isolated using the AxyPrep-96 Blood Genomic DNA Kit (Axygen BioScience, Union City, CA, USA). Whole-genome sequencing (WGS) sequencing was performed on the proband (III-2) and another family member (VI-4) using DNBSEQ-T7 sequencing instruments (MGI, Shenzhen, China), with paired-end reads of 100 base pairs. The sequenced reads were aligned to the human reference genome (GRCh38/hg38) using the Burrows-Wheeler Aligner (BWA v0.7.10). Subsequent processing, including duplicate marking, local realignment, and base quality score recalibration, was performed using the Genome Analysis Toolkit (GATK v4.0). Variant calling was conducted with GATK HaplotypeCaller. The identified variants were filtered using the following criteria: (1) population frequency < 1% in the gnomAD and an in-house Chinese database; (2) presence in a custom panel of 189 genes known to be associated with hearing loss (Supplementary material); and (3) compatible with an autosomal dominant inheritance model. The functional impact of prioritized variants was predicted using PolyPhen-2, MutationTaster, and SIFT.

Sanger sequencing was used for variants validation. Primer3 online (https://www.bioinformatics.nl/cgi-bin/primer3plus/primer3plus.cgi) was used to design primers for the GSDME gene variants. A pair of exon 8 primers (forward, 5′-CCGTCAGTGAAATGTAGCC-3′ and reverse, 5′-TTCCACAGTTACCACCTCTG-3′) across the intron-exon boundaries were used to span the sequences. One hundred ng Genomic DNA (gDNA) was used as a template in a total reaction volume of 15 μL (2 × PrimeSTAR Max Premix 7.5 μl, primer forward 10 μmol/L 0.25 μL, primer reverse 10 μmol/L 0.25 μL, and dH2O to 15 μL). PCR was performed with an initial denaturation at 98 °C for 1 min, followed by 35 cycles of 98 °C for 10 s, 60 °C for 5 s, and 72 °C for 20 s, and a final extension at 72 °C for 5 min. Direct sequencing of the DNA in both directions was performed by Sangon Biotech (Chengdu, China). The reference sequence of GSDME used was GenBank NM_001127453.2.

The pSPL3 vector, pre-purchased by our research group, was utilized to construct plasmids for in vitro splicing validation. This vector has 6031 base pairs and comprises an SV40 promoter, as well as two exons, SD6 and SA2. Specific primers with XhoI and BamHI restriction enzyme sites (F5′-CCCTCGAGCCGTCAGTGAAATGTAGCC3′ and R5′-CGGGATCCTTCCACAGTTACCACCTCTG3′) were utilized to amplify a 471 bp fragment that encompasses exon 8 of GSDME. The PCR products and plasmid underwent digestion with XhoI (Takara, 1094S) and BamHI (Takara, 1010S) restriction enzymes, respectively. Subsequently, T4 ligase (Takara, 2011A) was used for ligation. The ligated products were then transformed into E.coli DH5α Competent Cells (Takara, 9057) for amplification. Finally, single colonies were selected for sequencing to confirm the accuracy of the mutation.

The mini-gene vector was transfected into COS-7 cells using Lipofectamine 3000 (Invitrogen, L3000015). After 48 h, cells were collected and RNA was extracted using the Trizol (Invitrogen, 15596018). The extracted RNA was reverse transcribed into cDNA with a reverse transcription kit (Takara, RR047A). Amplification was performed using exon-specific primers (SD6-F: TCTGAGTCACCTGGACAACC and SA2-R: ATCTCAGTGGTATTTGTGAGC) from pSPL3. PCR was performed with an initial denaturation at 95 °C for 3 min, followed by 35 cycles of 95 °C for 15 s, 60 °C for 15 s, and 72 °C for 15 s, and a final extension at 72 °C for 5 min. The PCR products were electrophoresed on a 2% agarose gel. Direct sequencing of the DNA in both directions was performed by Sangon Biotech (Chengdu, China). Three controls were included in each transfection experiment: (1) the wild-type (WT) construct (negative control for aberrant splicing); (2) a construct carrying a known pathogenic splice-site variant (c.991-21_991-19delTTCinsT) (positive control for exon 8 skipping); and (3) the empty pSPL3 vector (background control).

A thorough evaluation was undertaken on pathogenic variations of GSDME previously identified in persons. PubMed was used to conduct literature searches using the keywords (GSDME OR DFNA5) AND (hearing loss OR deafness). The most recent search occurred on March 22, 2024. Information was collected on the location of the variant, age of onset, audiometric manifestations, associated symptoms, and other systemic abnormalities documented in the literature in affected individuals.

The MANE SELECT transcript was used to convert all filtered pathogenic variants. dbscSNV, MaxEntScan, and Splice AI were used to predict the effect of mutations on splicing. In addition, ESEfinder 3.0 (https://esefinder.ahc.umn.edu) was used for splicing prediction.

The pedigree(Figure 1A) shows a typical autosomal dominant inheritance pattern of hereditary hearing loss, covering five generations and including a total of 55 family members. There is no gender disparity among affected individuals in the pedigree. Hearing loss typically appears during the first or second decades of life. It is characterized by post-lingual, bilateral, symmetric, and progressive sensorineural deafness. Initially, it affects high frequencies and gradually extends to involve all frequencies (Figure 1B). Notably, individuals IV-1 and IV-3 present with asymmetric hearing loss due to otitis media. Most patients report intermittent tinnitus without symptoms of vertigo or balance disturbances and with normal intellectual and cognitive functions. None of the patients had a history of aminoglycoside exposure or noise exposure before presentation, and physical exams were normal. High-resolution CT scan of the temporal bone showed no abnormality. Thirteen patients showed moderate to severe bilateral sensorineural hearing loss during auditory testing. The audiograms showed that the high frequencies were affected at the beginning and that the middle and low frequencies were gradually affected with increasing age (Table 1). Although V-8 currently has average hearing thresholds within the normal range, the audiogram showed a significant decline in high-frequency hearing (Figure 1B).

Whole-genome sequencing of the proband (III-2) and an unaffected relative (IV-4) initially yielded multiple variants in hearing loss-associated genes. After filtering against population databases and enforcing an autosomal dominant inheritance model, a single candidate variant emerged: a heterozygous c.1123G>T transition in exon 8 of GSDME (NM_001127453.2), predicted to introduce a premature termination codon (p.Glu375Ter). Sanger sequencing confirmed this variant co-segregated perfectly with the hearing loss phenotype in the family (Figure 1C) and was absent from gnomAD, 1,000 Genomes, and our in-house controls.”

“A second heterozygous variant in MITF (c.1156G>C, p.Val386Leu) was also identified in both sequenced individuals. However, it was excluded as causative due to phenotype mismatch (no features of Waardenburg syndrome) and lack of co-segregation within the pedigree.

The in vitro minigene assay was employed to determine the effect of c.1123G>T on splicing. RT-PCR analysis of RNA from transfected COS-7 cells yielded a single 456-bp product for the wild-type construct, which sequencing confirmed contained the vector-derived SD6 exon, GSDME exon 8, and the SA2 exon (Figures 2B, C). In contrast, the mutant construct produced two products: the expected 456-bp wild-type band and a predominant aberrant band of 263 bp (Figure 2B). Sequencing of the smaller band confirmed the complete absence of GSDME exon 8, indicating direct splicing from SD6 to SA2 (Figures 2C, D). This exon skipping creates a frameshift, introducing a premature termination codon at position 372 and predicting a truncated protein lacking the C-terminal 126 amino acids.”

“Bioinformatic analysis using ESEfinder 3.0 suggested that the c.1123G>T substitution disrupts a binding motif for the splicing factor SF2/ASF, potentially reducing exonic splicing enhancer activity and contributing to exon 8 skipping (Figure 2E).

A review of the literature identified 20 previously reported pathogenic GSDME variants, predominantly clustered around exon 8 (Figure 3A, Table 2). All experimentally tested variants have been shown to cause exon 8 skipping. The associated phenotypes consistently involve progressive, high-frequency sensorineural hearing loss with an onset ranging from 5 to 50 years, often accompanied by tinnitus. Considerable intra- and inter-familial phenotypic variability was noted, even for the same mutation.

In silico splice prediction tools (SpliceAI, dbscSNV, MaxEntScan) were applied to these variants. Their performance varied, with MaxEntScan achieving 100% accuracy for variants within canonical splice regions, while SpliceAI's accuracy was lower (66.7%) despite its ability to assess all genomic regions (Table 2).