Waardenburg syndrome (WS) is a rare auditory-pigmentary disorder characterized by various combinations of sensorineural hearing loss and pigmentation anomalies. It is estimated to affect approximately 1 in 42,000 live births, representing 1–3% of all congenital deafness patients (Read and Newton, 1997). Four subtypes have been clinically defined based on the presence or absence of additional symptoms, with related causal genes reported: WS1 (OMIM # 193500) or WS3 (OMIM # 148820), resulting from pathogenic variants in PAX3 (Foy et al., 1990; Tassabehji et al., 1992), marked by dystopia canthorum; WS2 (OMIM # 193510, 608890, 611584), caused by mutations in SOX10 (Bondurand et al., 2007), KITLG (Ogawa et al., 2017), MITF (Hughes et al., 1994; Tassabehji et al., 1994), or SNAI2 (Sánchez-Martín et al., 2002); and WS4 (OMIM # 277580, 613265, 613266), arising from mutations in SOX10 (Pingault et al., 1998), EDN3 (Edery et al., 1996), or EDNRB (Attié et al., 1995; Van Camp et al., 1995), often accompanied by Hirschsprung disease (HSCR).

Current diagnoses of these genes explain approximately 61%−71% of WS cases (Ideura et al., 2019; Batissoco et al., 2022; Wang et al., 2022). Within WS patients who undergo molecular diagnoses, the detection of pathogenic variants in the SOX10 gene constitutes a significant proportion, approximately 1/3. While most studies have focused on mutations in the coding region of the genome, non-coding variations may also play an important role. Bondurand et al. (2012) and Lecerf et al. (2014) observed that deletions in the SOX10 enhancers could contribute to the development of WS4 (Bondurand et al., 2012) or HSCR (Lecerf et al., 2014). In vitro experiments demonstrated a significant reduction in SOX10 transactivation capacity with the enhancer mutant construct compared to the wild-type (Lecerf et al., 2014).

Predicting the effects of non-coding variants poses challenges. Given that non-coding regions comprise about 98% of the genome, pinpointing functionally significant genomic areas becomes particularly daunting (Spielmann and Mundlos, 2016). Non-coding variants influence gene expression levels and cellular processes by modulating regulatory elements like promoters, enhancers, and non-coding RNAs. Previous investigations have examined and compared SOX10 enhancers across various species, including humans, mouse, and zebrafish, revealing 12 enhancers located upstream of SOX10 and 2 downstream of the gene (Werner et al., 2007; Antonellis et al., 2008; Betancur et al., 2010). These identified SOX10-related regulatory elements facilitate comprehension of variant pathogenicity within these regions.

Approximately 50% of the non-coding region of the human genome consists of repetitive elements such as LINE-1 and Alu sequences, which exhibit high homology and may disrupt genomic stability, leading to CNV formation through mechanisms like non-allelic homologous recombination (NAHR) or non-homologous end joining (NHEJ). The deletion involved these repeat elements and NAHR of TMC1 has been reported in non-syndromic hearing loss (NSHL) patients (Shearer et al., 2014).

Evaluation of gene dosage sensitivity is crucial for classifying associated CNVs. Heterozygous deletions of SOX10 detected in WS2 and WS4 patients, along with mutations resulting in mRNA reduction via nonsense-mediated decay (NMD) pathway reported in WS4 patients, confirm the haploinsufficiency (HI) of SOX10 in relation to WS (Pingault et al., 1998; Southard-Smith et al., 1999; Inoue et al., 2004).

In this study, we utilized WGS data from 59 WS patients to investigate CNVs affecting upstream regulatory elements of the SOX10 gene. We identified five non-coding CNVs in enhancer and promoter regions, further demonstrating the pathogenicity of non-coding CNVs in WS and proving to be indispensable for WGS-based patient reporting.

A total of 59 WS patients from 51 Chinese families, previously undiagnosed by known WS genes, were recruited for this study. These patients were selected from a sub cohort of 290 WS patients in the CDGC who met the clinical diagnostic criteria. WGS was employed to detect CNVs within regulator regions spanning from 65kb upstream to 4.5kb downstream of the SOX10 gene, which includes the promoter, 5'UTR, and enhancers (U1/MCS7, U2/MCS5, U3/MCS4, U4, U5/MCS2, D6, D7, MCS1, MCS3, MCS6, MCS8, and MCS9, as illustrated in Figure 2A). The enhancer elements MCS1-9 and U1-5, D6, and D7 were all derived from previous studies, where these enhancers were identified in the genomes of mouse and zebrafish, respectively. We aligned these elements with the human genome using sequence conservation.

By combining the outcomes of three analysis software tools (Manta, CNVnator, and CNMOPS), we detected 7 heterozygous deletions in 10 WS patients from 7 families. Validation experiments were conducted to verify the reliability of WGS BAM files, gap-PCR, and co-segregation Sanger sequencing in the families. Subsequent analysis revealed that two CNVs were false positives in the WGS BAM files. Therefore, only 5 deletions (Table 1 and Figure 2A) underwent further scrutiny for pathogenicity, following the ACMG & ClinGen technical standards.

Among the identified CNVs, CNV1; chr22:37988513-38058868 in Family 1 and CNV2; chr22:37997030-38062216 in Family 2, were identified, spanning regions containing four homologous enhancers (U1, U2/MCS5, U3, and U4). These CNVs showed complete overlap with reported pathogenic CNVs within the SOX10 enhancer subregions. Moreover, CNV3; chr22:37984413-37984860 in Family 3, CNV4; chr22:37983938-37986779 in Family 4, and CNV5; chr22:37983877-37987126 in Family 5 were identified. These variants were located within the 5'UTR, spanning exon 1 and part of exon 2, as well as the promoter region of the SOX10 gene. The breakpoints associated with these copy number losses were confirmed via gap-PCR, except for CNV5, which could only be demonstrated by BAM file owing to DNA sample depletion (Figure 2B). Importantly, these CNV events were absent in both the gnomAD (CNVs v4.0) and DGV databases.

All families were followed up to confirm the presence of relatives with WS-related phenotypes and to collect blood samples from parents, siblings, or other related individuals. In Family 1, three consecutive generations exhibited autosomal dominant inheritance of WS, with CNV1 inherited from the proband's grandmother. Notably, none of the parents in Families 2-5 displayed WS-related phenotypes. Additionally, CNV2 in Family 2 was confirmed as de novo through gap-PCR, while parental samples from Families 3-5 were unavailable. Based on the ACMG guidelines modifier, we classified 5 CNVs as pathogenic/likely pathogenic, with detailed evidence and final pathogenicity classification in Table 2.

Analysis of sequences surrounding each CNV breakpoints using Repeat Masker (http://www.repeatmasker.org/) revealed significant similarity only for CNV1 breakpoints, corresponding to Alu repeats. These Alu repeats belong to distinct Alu families: AluSx and AluSq2. Further analysis revealed an 85.8% similarity between AluSx and AluSq2 (see Figure 2C), suggesting that the CNV was initially generated by Alu-mediated NAHR.

Considering the phenotypic heterogeneity of WS2 and its potential to manifest as NSHL, we expanded our analysis to include CNVs in the non-coding regions of SOX10 among 8,674 hearing loss patients with WGS. Among these patients, only one heterozygous deletion was detected overlapping with SOX10 enhancers. Specifically, CNV6 encompassing genomic region 22q13.1 (chr22: 37992678-38003872) was identified in two unrelated NSHL patients. However, this CNV did not segregate with the disease in either family. In Family S1, exhibiting an autosomal dominant inheritance pattern, the proband did not inherit the CNV from the affected mother, while in Family S2, the proband inherited the variant from an unaffected mother. These data suggest CNV6 is a likely benign variant (see Supplementary Figure 1).

There was phenotypic heterogeneity associated with CNVs identified in this study (Figures 3A–C). In Family 1, the proband (F1: III-1) displayed bilateral profound hearing loss and bilateral blue irises, while the father (F1: II-2) presented with sensorineural hearing loss in the right ear and a blue iris in the right eye. The grandmother (F1: I-1) only displayed a blue iris in the right eye with normal hearing levels. Upon examination via computed tomography (CT) scan, the proband revealed an enlarged vestibular pool, with a narrow and insufficiently developed lumen of the lateral semicircular canal (Figure 3D). Patients in Families 2 and 5 both manifested bilateral profound hearing loss and bilateral blue irises. In Family 3, siblings displayed premature graying during their teenage years, along with bilateral profound hearing loss and blue iris pigmentation. The proband of Family 4 exhibited unilateral sensorineural deafness and unilateral blue iris. Notably, none of these patients exhibited congenital megacolon.

In this study, we conducted WGS on 59 patients with WS and examined CNVs within the regulatory region of the SOX10 gene. Previously, the CDGC cohort collected 290 clinically diagnosed WS patients from 228 families. Among them, 77.6% (177/228) were diagnosed with pathogenic variants in the coding regions of SOX10, PAX3, MITF, or EDNRB genes through TGS (these data will be presented in separate article). Here, our investigation identified five pathogenic deletions within the non-coding region of the SOX10 gene, including the pathogenic CNVs in the promoter and 5'UTR reported for the first time. This led to a 2.2% (5/228) increase in molecular diagnostic rates, underscoring the importance of rare non-coding variations in gene identification. Therefore, these findings indicate that non-coding CNVs play an important role in WS and should be considered in WGS-based diagnostics.

Evaluating the pathogenicity of variants in non-coding regions presents a significant challenge in clinical genetic testing. Guidelines provided by ACMG/AMP and ClinGen offer valuable insights for this purpose. According to ACMG guidelines (Riggs et al., 2020), copy number loss, as outlined in Section 2, pertains to “Overlap with Established/Predicted HI or Established Benign Genes/Genomic Regions.” Bondurand et al. (2012) and Lecerf et al. (2014) reported clinical patients involving SOX10 enhancer deletions (Werner et al., 2007; Ellingford et al., 2022). CNV; chr22:38013680-38051826 resulted in HSCR, while CNV; chr22:38008019-38064588 led to WS4. Both deletions were identified de novo through comparative genomic hybridization (CGH)-array analysis. Based on these findings, the minimal overlap region between all pathogenic CNVs upstream of SOX10 gene should be considered an established HI genomic region. CNV1 and CNV2 in this study contain extensions of established SOX10 enhancers that completely overlap with the previously identified HI region; therefore, we applied criterion 2A to CNV1 and CNV2. CNVs 3-5 occur within the SOX10 promoter and 5' UTR region. Given the critical role of both the promoter and UTR in transcriptional regulation, and the emphasis on promoter and UTR variants in the clinical interpretation of non-coding region variants, as highlighted by Ellingford et al. (2022), we used criterion 2C-2, which specifies “Partial overlap with the 5' end of an established HI gene and only the 5' UTR is involved.” CNV6, which occurs within the upstream enhancer region of SOX10 and meets criterion 1A, was detected in two unrelated NSHL patients but did not segregate in two separate pedigrees (in accordance with 5E), leading to its classification as likely benign. This suggests that heterozygous deletions involving only enhancers MCS3 (zebrafish) and U4 (mouse) may be tolerated.

In our study, we identified CNV1, which resulted from Alu-mediated NAHR. Alu elements are transposable elements that constitute approximately 10% of the human genome (Kazazian and Moran, 2017). Due to their abundance and sequence homogeneity, Alu elements frequently participate in genomic rearrangements. These rearrangements can have significant implications for human health as ~0.1% of human genetic disorders are linked to Alu-mediated genomic rearrangements (Deininger and Batzer, 1999).

Understanding the pathogenicity of non-coding variants is crucial for identifying potential disease-causing mechanisms, improving disease risk prediction, and developing targeted therapies (Mattick et al., 2018). However, interpreting the pathogenicity of non-coding variants poses several challenges. Firstly, their wide-ranging and intricate nature makes precise annotation of regulatory elements difficult. Regulatory elements often display cell-type-specific activity, resulting in varying effects across different tissues and cell types. Analyzing non-coding variants in relevant cell types is essential for accurately assessing their functional impact. Secondly, limited data on non-coding CNV allele frequency pose challenges to achieving statistical power. Accumulating sufficient data for the analysis of rare variants requires collaboration and large-scale genomic projects. Thirdly, experimental validation of the functional impact of non-coding variants is still immature and technically challenging. In vivo experiments on the impact of non-coding region variants on gene function are even more challenging.

This study identifies five SOX10 non-coding CNVs in WS patients, lacking direct SOX10 expression data in human cochlea or skin tissue. Subsequent investigations should utilize cell models or organoid models to explore the impact of SOX10 non-coding CNVs on gene function. For the remaining 51 undiagnosed WS patients, will continue to analyze the pathogenicity of CNVs and other types of variants (such as SNVs and InDels) in non-coding regions of other known WS genes and explore reliable experimental validation methods. Additionally, ongoing work includes identifying new genes causing WS.

The CNVs we identified align with previously documented regions but manifest distinct phenotypes. Specifically, while all five CNVs identified in our study correlate with WS2, previously reported patients exhibited a megacolon phenotype, leading to diagnoses of WS4 or HSCR. The underlying mechanism behind this phenotypic variation remains poorly understood.

In summary, our study unveiled five novel non-coding deletions within the SOX10 gene, spanning the 5'UTR, promoter, and enhancer regions. These findings aid in diagnosing more WS patients and provide valuable guidance for their genetic counseling. Moreover, our discoveries broaden the spectrum of pathogenic variants within the non-coding regions of the SOX10 gene and established a framework for interpreting non-coding regions in other genes.

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding authors.

The studies involving humans were approved by Ethics Committee on Biomedical Research, West China Hospital of Sichuan University. The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation in this study was provided by the participants' legal guardians/next of kin. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

YH: Conceptualization, Formal analysis, Writing—original draft, Writing—review & editing. JG: Writing—original draft, Writing—review & editing. YLo: Resources, Writing—review & editing. WX: Writing—review & editing, Resources. LK: Data curation, Writing—review & editing. MC: Writing—review & editing, Software. TT: Writing—review & editing. MZ: Writing—review & editing. FB: Formal analysis, Writing—review & editing. YLu: Resources, Writing—review & editing. JC: Conceptualization, Supervision, Writing—original draft, Writing—review & editing. HY: Supervision, Writing—review & editing.