A Novel Missense WFS1 Variant: Expanding the Mutational Spectrum Associated with Nonsyndromic Low-Frequency Sensorineural Hearing Loss

Background Nonsyndromic low-frequency sensorineural hearing loss (LFSNHL) is an uncommon form of hearing loss (HL) that typically affects frequencies at 2000 Hz and below. Heterozygous variants in the WFS1 gene at the DFNA6/14/38 locus are considered a common cause of LFSNHL. To date, 34 different pathogenic genetic variants have been reported to cause LFSNHL with seven of these variants identified in the Chinese population. However, limited reports are available on the association between WFS1 gene and LFSNHL. Here, we report a five-generation Chinese family with an autosomal dominant inheritance pattern of postlingual and progressive LFSNHL. Methods Routine clinical and audiological examinations were performed on 16 affected and 7 healthy members in this family. The targeted next-generation sequencing of 127 known deafness genes was performed to identify variants in affected individuals. Sanger sequencing were further employed to confirm the pathogenic variant identified. Results A novel heterozygous pathogenic genetic variant c.2530G > T (p.Ala844Ser) was identified in the WFS1 gene in all patients of this family. The mutated Ala residue is evolutionarily conserved and cosegregated with HL. The variant was predicted to be deleterious by MutationTaster, PolyPhen-2, LRT, and Fathmm software. Conservation analysis and 3D protein structure model indicated that the variant caused a structural change in the protein. Conclusions Our present study identifies a novel heterozygous WFS1 variant associated with LFSNHL in a Chinese family.


Introduction
Hearing loss (HL) is the most common sensory impairment, severely affecting people's lives [1]. According to the World Health Organization report, more than 1.5 billion people worldwide have varying degrees of HL, which is expected to increase to more than 2.5 billion by 2050 (https://www .who.int/zh/news-room/fact-sheets/detail/deafness-andhearing-loss), generating heavy social and economic burdens on many individuals and families as well as the society. There are numerous factors that cause deafness, such as aging, ototoxic drugs, environmental factors, genetic factors, inflammation, and other unknown causes [2]. However, it is worth noting that more than 60% of severe HL cases in neo-nates or early childhood are related to genetic factors [1]. Depending on its association with other organ abnormalities, the cases of hereditary deafness are generally grouped into two categories, i.e., syndromic HL and nonsyndromic hearing loss (NSHL), accounting for 30% and 70% of the total cases of hereditary deafness, respectively [3]. The NSHL inheritance patterns include autosomal recessive (80%), autosomal dominant (15-20%), sex chromosome chain disorders (1%), and mitochondrial DNA inheritance (1%) [4]. To date, a total of 161 genetic loci (68 dominant and 93 recessive) and 134 genes (45 dominant, 74 recessive, and 5 X-linked) have been reported to be associated with NSHL based on the Hereditary Hearing Loss database (https://hereditaryhearingloss.org/; accessed in August 2021).
The low-frequency sensorineural hearing loss (LFSNHL) is a rare type of HL that affects low frequencies of 2000 Hz and lower, showing a characteristic audiogram configuration that is mostly ascending. Studies have shown that a group of four genes, including DIAPH1, MYO7A, CCDC50, and WFS1, are associated with the nonsyndromic LFSNHL [5][6][7]. The variants in WFS1 are considered common causes of LFSNHL [8,9]. For example, Fukuoka et al. reported that variants in the WFS1 gene were detected in one-third autosomal dominant LFSNHL Japanese families during the screening of 206 Japanese autosomal dominant and 64 autosomal recessive (sporadic) nonsyndromic HL probands [8].
Hearing screening combined with genetic diagnosis can help us detect more pathogenic genetic variants. In this study, we performed mutational analysis of a fivegeneration Chinese family using both deafness genes NGS panel and Sanger sequencing techniques. The missense variant WFS1 c.2530G > T (p.Ala844Ser) was identified as the cause of this nonsyndromic LFSNHL in this family. Four pathogenicity predicting tools were used to indicate that the variant was deleterious, and 3D protein structure model predicted showed that the novel variant in WFS1 may alter the normal structure of the wolframin protein. Our findings would be a significant addition to the collection of WFS1 variants and would make the genetic counseling available for this family.

Materials and Methods
2.1. Study Subjects and Controls. The family with LFSNHL comprised a total of 94 individuals in five generations, including 26 affected members. With the informed consent obtained from each individual, a total of 23 family members (16 affected and 7 unaffected) participated in a clinical evaluation study. A total of 200 individuals with normal hearing were selected as controls. The written informed consents of all participants were obtained before their inclusion in the study, which was approved by the Ethics Committee of Shandong University (approval number 014).
2.4. Targeted Deafness Gene Capture and Next-Generation Sequencing. Targeted genomic capturing and nextgeneration sequencing analyses were performed on proband (III-10) by BGI Inc. (Shenzhen, China). All exons, splicing sites, and immediate flanking intron sequences of a total of 127 deafness genes (Supplementary Table 1) were sequenced. The E210 DNA-shearing instrument (CovarisS2, Massachusetts, USA) randomly fragmented genomic DNA samples into fragments of 200-300 base pairs. Then, the end-repair, adenylation, and adapter ligation were performed for library preparation following the standard Illumina protocols. Capture of targeted DNA fragments was conducted by hybridization with capture arrays. Highthroughput sequencing for captured library was performed on Illumina HiSeq2000 Analyzers. Three-step filtering analyses were performed on the raw data to generate "clean reads" for further analysis. First, the indexed primers were used to identify different reads from different samples in the primary data (i.e., reads that were perfectly matched to the theoretically indexed sequences and reads that were matched with the theoretical primer indexed sequences with a maximum of three mismatches). Second, the reads containing partial adapter sequences were removed. Third, the low-quality reads (i.e., reads containing more than 10% Ns in the read length as well as 50% reads with a quality value of less than 5 and with an average quality of less than 10 and adapter sequences including indexed sequence) were removed [10]. The high-quality clean reads were aligned to the human reference genome (hg19) using the BWA (Burrows-Wheeler Aligner) MultiVision software package. SOAPsnp software and GATK Indel Genotyper were used to identify SNPs and indels, respectively. The novelty of variant was determined by screening of 1000 Genomes, HapMap, dbSNP, gnomAD, and the BGI in-house databases. The pathogenic features of missense variant were predicted by several different computer algorithms (i.e., MutationTaster, PolyPhen-2, LRT, and Fathmm). We excluded variants with minor allele frequencies >0.05 in the 1000 Genomes Project and PVFD database (https://github .com/BGI-flexlab/PVFD). Pathogenicity of the variant was classified based on the guidelines of the American College of Medical Genetics and Genomics (ACMG) 2015 [11].

Clinical Features of the Participants. A five-generation
Chinese family with a total of 94 members from Shandong Province, China, participated in this study with the pedigree constructed according to the verbal descriptions of the participants (Figure 1(a)). The affected members of this Han Chinese family were presented with dominant NSHL, showing bilateral, symmetric, and progressive hearing impairment. The PTA results showed that it was predominantly low-frequency impaired (Figure 1(b)). In addition, according to the age distribution map, the age at onset of hearing impairment was varied from 1 to 30 years in this family, and most of these patients mainly developed HL around 10 years old (Figure 2). The proband (III-10) developed severe sensorineural HL involving low frequencies; the average auditory thresholds on PTA for the right and left ears were 67 and 63 dB HL, respectively (Figure 1(b)). Only wave V was observed in both ears under 96 dB HL stimulation on ABR testing, and DPOAE were absent bilaterally.
Results of MRI, vestibular function tests, electrocochleography, and utricle and balloon function tests revealed no abnormality of otology-related malformations or vestibular dysfunction in the participants. The acoustic immittance tests were all A-type tympanograms, and the external auditory canal volume, tympanic pressure, and compliance values were all normal. All family members appeared normoglycemic. Assessment of optic nerve electroretinograms and fundus images revealed no abnormalities. A comprehensive clinical examination and medical histories revealed no other clinical syndrome phenotypes. The main clinical features of this family are described in Table 1.

Detection and Analysis of Variants.
In order to identify the pathogenic variant in this family, the targeted gene  (Figure 3(a)). The variant c.2530G > T (p.Ala844Ser) was not listed in the public databases of HGMD, dbSNP, gno-mAD, 1000 Genomes, and HapMap. Conservation analysis (Figure 3(b)) showed that the amino acid residue of this missense variant is conserved among 8 species (i.e., Homo sapiens, Pan troglodytes, Bos taurus, Canis familiaris, Rattus norvegicus, Mus musculus, Gallus gallus, and Danio rerio). Finally, the mutation p.Ala844Ser was predicted to be deleterious by MutationTaster, PolyPhen-2, LRT, and Fathmm software ( Table 2). Based on the amino acid sequence of WFS1, the threedimensional structures of WT and mutant WFS1 proteins were simulated. This missense variant caused a change from a    (Figure 4(a)). The substitution of WFS1 p.Ala844Ser could lead to the formation of two hydrogen bonds between the hydroxyl functional group    (Figure 4(b)). These changes in amino acid interactions are predicted to prompt the formation of protein alpha helical structures.

Subcellular Localization of WFS1 Protein in HL Patients.
In order to identify the intracellular location of the dominant variant, we constructed both the WT and mutant (p.Ala844Ser) WFS1 expression plasmids fused with Flag tags (Supplementary Figure 1). Then, we immunostained both HEK-293 cells and HEI-OC1 cells that transiently expressed WT and dominant variant p.Ala844Ser, respectively. The results showed that similar to WT, the wolframin protein bearing this variant was localized to the endoplasmic reticulum (ER) in both cell lines ( Figure 5).

Discussion
WFS1 is located in the 4p16.1, containing a total of eight exons, with exon 1 as a noncoding region and exon 8 encoding the large portion of the transmembrane regions and the carboxy terminals of the wolframin protein, which is a type of glycoprotein with nine transmembrane domains [12][13][14]. Pathogenic variants in the WFS1 gene are generally associated with Wolfram syndrome (WS), Wolfram-like syndrome (WLS) and LFSNHL [15]. WS, also known as DIDMOAD, is an uncommon monogenetic spectrum disorder characterized by juvenile onset diabetes, diabetes 7 BioMed Research International insipidus, optic atrophy, and HL and often accompanied by other progressive neurological abnormalities [16]. The WSinduced deafness is recognized as HL at middle and high frequencies, and variants in WFS1 are generally distributed throughout its coding region [17]. In contrast, the LFSNHL caused by WFS1 variants at the DFNA6/14/38 locus is characterized by showing the bilateral symmetrical low frequency HL with the variants mainly located at the amino acids in the C-terminal domain of wolframin proteins [15].
In the present study, we identified the genetic variant responsible for the deafness phenotype in the pedigree of a five-generation Chinese family. The next-generation sequencing and sanger sequencing revealed a WFS1 c.2530G > T missense variant that was cosegregating with HL in this family. Fukuoka et al.'s study showed that WFS1-associated LFSNHL was noticed mainly between 5 and 14 years old (average of 10 years old) and thereafter gradually progressed over a long period [8]. Thorpe et al.'s study further showed that WFS1associated LFSNHL progressed to high-frequency HL, reaching at least a severe HL (>70 dB HL) at high frequencies by age 70 years [18]. Our findings in this study further supported these findings. The distribution plot of age at onset in this family ( Figure 2) showed that the age of onset in WFS1 variant carriers was mainly concentrated between 6 and 15 years old. The PTA revealed a progressive HL in this Han family, in which only mild low-frequency HL was observed in 5- year-old patients, while the 70-year-old affected person developed with severe full-frequency HL. Notably, the result of Sanger sequencing showed that there was a heterozygous mutation in the 10-year-old child (V-8) of this family, but the PTA detection was normal (Supplementary Figure 2). We speculated that the child did not have a deafness phenotype because he had not reached the age of onset. In addition, none of the family members showed symptoms of tinnitus and vertigo, which was consistent with the results previously reported [15].
The results of spatial structure model analysis showed that the variant of p.Ala844Ser promoted the formation of hydrogen bonds between serine residue and multiple amino acid residues, resulting in the misformation of the α-helix structure at the C-terminal of wolframin protein. Hydrogen bond, as an intermolecular force, plays an important role in maintaining the spatial conformation and stability of proteins. Therefore, the formation of local abnormal conformation may interfere with the normal function of wolframin protein, e.g., disrupting the interaction between wolframin protein and other molecules or proteins. Alterations in the spatial structure and function of wolframin protein may be responsible for HL in this Chinese family. To date, there have been seven reported cases of LFSNHL in Chinese pedigrees caused by missense variants in WFS1 (Table 3)   BioMed Research International are predominantly located in exon 8, except for two missense variants in exon 5 ( Figure 6(a)), and most of the known variants in LFSNHL are concentrated in the Cterminal domain of wolframin protein (Figure 6(b)), indicating that the C-terminal domain of wolframin protein plays an essential role for the normal function of wolframin protein [35]. The wolframin protein mainly located in the ER widely exists in brain tissue, pancreatic β-cell, heart, lung, and placenta [36,37]. Regarding the protein expression in the inner ear, studies reported that the expression of wolframin protein was observed in different cell types of mice and adult marmoset, including the organ of Corti, spiral ganglion neurons, and supporting cells of the cochlea [38,39]. Morikawa [40], whereas the variant found in this study did not induce significant changes in wolframin protein levels in cells (Supplementary Figure 1). In addition, the WFS1 variants (i.e., p.N325_ I328del, p.Q194X, and p.L543R) induced constitutive ER stress and cell apoptosis in HEK-293 T cells [40]. However, the mechanism of LFSNHL caused by WFS1 variants is still unclear. Because the HEI-OC1 cells themselves could highly express wolframin protein, considering that if the mutant plasmid is directly transfected into HEI-OC1 cells, the endogenous wolframin protein will compensate the molecular changes caused by the mutant plasmid. Therefore, to further explore the effect of wolframin protein bearing this variant on ER function in inner ear hair cells, the constructions of cell and transgenic mouse models carrying this variant are necessary. The inheritance pattern of WFS1 variants as well as the type of variants inherited could affect both the onset and the severity of the main clinical features of WS [41]. The WFS1 variants are either homozygous or compound heterozygous in autosomal recessive hereditary WS. Studies have reported that the earlier-onset diabetes and optic atrophy occur in patients who are homozygous or compound heterozygous for two inactivating variants [42,43]. Furthermore, compound heterozygosity for missense variants may lead to a mild phenotype [43]. However, patients who are compound heterozygous are at a higher risk of psychiatric disorders, DM, and HL [44]. The WFS1 variants are heterozygous in autosomal dominant LFSNHL, and younger patients are often asymptomatic or exhibiting mild HL only at low frequencies. The novel missense variant c.2530G > T (p.Ala844-Ser) identified in this study caused only nonsyndromic HL in this family. Interestingly, the variant with the substitution of A described in another independent family of nonsyndromic LFSNHL by Noguchi et al. [31] is not only in the same codon but also in the same genomic position as the variant with the substitution of T detected in our study, evidently suggesting that the Ala844 residue of WFS1 may play an important role in NSHL. To date, neither the type nor location of the pathogenic variant could predict the phenotype [45]. Therefore, the establishment of the genotypephenotype correlations is extremely difficult due to the molecular complexity of WFS1, its diverse clinical features, and the small size of the patient cohorts.
In summary, we identify a novel heterozygous c.2530G > T WFS1 variant cosegregating with HL in a five-generation Chinese family by targeted capture sequencing, which expands the mutational spectrum of WFS1. Our data provide additional molecular and clinical information for establishing a strong genotype-phenotype correlation for LFSNHL.

Data Availability
The data used to support the findings of this study are available from the corresponding author upon reasonable request.

Consent
The patients and relatives signed written informed consent for publication of this article.

Conflicts of Interest
The authors declare that they have no conflicts of interest.