Background

Hearing loss (HL) is one of the most common human pathologies that significantly affects the quality of life [1]. About 60% of congenital HL is caused by genetic factors [2, 3]. Non-syndromic sensorineural hearing loss (NSHL) is considered a major cause of HL. To date, mutations in 124 genes have been identified in individuals affected with NSHL, among which mutations in 78 genes were related to autosomal recessive non-syndromic sensorineural hearing loss (ARNSHL), mutations in 51 autosomal dominant genes and 5 X-linked genes were correlated with NSHL (Hereditary Hearing Loss Homepage, http://hereditaryhearingloss.org, updated on 30 August 2021). The most common variations that were found in ARNSHL were in the genes GJB2, SLC26A4, CDH23, MYO15A and OTOF [4, 5]. Genetic variations in MYO15A were considered the third most common cause of ARNSHL in Iran due to prevalent consanguineous marriages [5, 6]. Whereas in the cohort of Korean ARNSHL patients, MYO15A mutations were recognized as the fourth most important deafness gene variants after those detected in other genes like GJB2, SLC26A4 and CDH23 [7, 8].

MYO15A (OMIM #602666) is a 71 kb long gene that contains 66 exons. It is localized on chromosome 17p11.2 (chr17:18012020–18083116; hg19 assembly) and encodes the myosin-XV protein with 35,390 amino acids [9]. Myosin proteins are a large family of actin-based molecular motors that bind actin filaments to produce force and motion, thus contributing to the hydrolysis of ATP.

The MYO15A protein contains an N-terminal domain (amino acids (AA) 1–1223), a motor domain (AA 1224–1899), three light-chain binding IQ motifs (AA1909–1942), two myosin-tail homologies 4 domains (MyTH4, AA 2066–2174 and 3051–3161), two band F, ezrin, radixin, myosin domains (FERM, AA 2687–2867 and 3217–3497), an Src-homology-3 domain (SH3, AA 2865– 2959) and a C-terminal PDZ ligand motif [6, 10, 11].

It is reported that MYO15A mutations cause sensorineural HL in human autosomal recessive deafness 3 (DFNB3, OMIM #600316) [7]. The DFNB3 locus was discovered in patients from a remote village in Indonesia, where 2.2% (47/2185) of the population was affected by hearing loss [12, 13]. So far, more than 200 MYO15A variants have been reported in more than 20 countries and regions, such as Algeria, Arab, Brazil, China, France, Germany, India, Iran, Israel, Japan, Mexico, the Netherlands, Oman, Pakistan, Palestine, Qatar, South Korea, Spain, Tunisia, Turkey and the United States. However, due to the large size of the gene and its many exons, simple techniques for detecting variants are discordant with it. Therefore, the clinical characteristics of MYO15A related to NSHL hearing level, age of onset, the degree of progression, associated symptoms and hotspot mutations were not clearly identified. So far, MYO15A had been reported sporadically in China. In this study, 81 cases from 74 families identified with at least one MYO15A pathogenic or likely pathogenic variants, or uncertain significant variants, diagnosed by next-generation sequencing (NGS) from 2263 Chinese cases with NSHL, were enrolled to analyze the correlation between the MYO15A genomic variants and NSHL pathological phenotype. Co-segregation of variants was confirmed in probands and healthy parents, as well as more family members if available, via NGS and Sanger sequencing. This study is by far the largest focused on MYO15A variants and their implication in the outcome of NSHL. As well as we were able to detect the gene frequency and the recurrent variant of the MYO15A in Chinese patients with NSHL. The association of MYO15A variants with hereditary deafness patients, their severity, progression and age of onset was further conducted.

Methods

Purpose of test

The performed test aimed to examine the role and the genotype–phenotype correlation of MYO15A variants in a cohort of Chinese NSHL patients.

Subjects and clinical evaluation

There were 2263 participants from 1842 families with NSHL from the Genetic Testing Center for Deafness at the College of Otolaryngology Head and Neck Surgery, Chinese PLA General Hospital enrolled in the study, from June 2015 to September 2021. Trio WES was performed in 95 cases and their parents, trio/quadro NGS in 2009 cases and their family members, and singleton NGS in 159 cases.

And 81 cases from 74 families with detected MYO15A variants, related to NSHL, were analyzed for the assessment of the correlation between the MYO15A genotype and the NSHL phenotype. Detailed interviews were conducted with probands and their families to obtain their medical and familial histories.

All underwent testing that included physical examination, otoscopy, pure tone audiometry (PTA), tympanometry, assessment of auditory brainstem responses (ABR), distortion product otoacoustic emission (DPOAE), multiple auditory steady-state evoked responses (ASSR), temporal bone computerized tomography scans, and magnetic resonance of the brain. The definition for the severity of hearing impairment, according to pure-tone audiometry (PTA) of the better ear, was made based on the average hearing threshold level at four frequencies (500, 1000, 2000 and 4000 Hz) of air conduction. 26–40 dB HL were considered to be mild hearing loss; 41–55 dB HL, moderate hearing loss; 56–70 dB, moderately severe hearing loss; 71–90 dB HL, severe hearing loss; > 90 dB HL, profound hearing loss. The occurrence of hearing loss was categorized as prelingual (≤ 3 years) or post-lingual (> 3 years). Asymmetric hearing loss (AHL) was defined as greater than 15 dB between the ears at 0.5, 1, and 2 kHz or greater than 20 dB at 4 kHz on the audiogram (American Academy Otolaryngology-Head Neck Surgery 1997) [14] as reported previously [15].

Peripheral blood samples were collected from all cases, their parents and siblings (if any). All cases obtained informed consent for the performed molecular genetic analysis and their clinical data publication. The study was approved by the Ethics Committee of the Chinese PLA General Hospital (reference number S2016-120–02). Written informed consent was obtained from the participants and in the case of young cases from their parents.

Targeted deafness gene capture and NGS

Targeted deafness gene capture and NGS were performed as previously reported [16]. DNA samples of 64 cases from 58 families were subjected to targeted NGS, 35 cases of them conducted trio (proband and parents) targeted NGS and 29 cases conducted quarto (proband, parents and sibling) targeted NGS. The proband received the panel test containing 168 deafness-related genes (Additional files 1: Table S1). All coding exons, along with 100-bp flanking regions were sequenced on the Illumina HiSeq 2000 (Illumina, San Diego, CA, USA) using the MyGenostics gene enrichment system (MyGenostics, Boston, MA, USA).

Whole-exome sequencing (WES)

Illumina NovaSeq6000 sequencing platform was used to conduct the WES by MyGenostics (Beijing, China) (detailed procedures shown in Additional files 3). DNA samples from 17 MYO15A-related cases and their parents were subjected to trio WES and subsequently validated by Sanger sequencing. The nomenclature of the mutation described in Table 1 is based on MYO15A cDNA and protein accession numbers NM_016239.3 and NP_057323.3, respectively. We used the genomic coordinates from GRCH37/hg19 constructed from the human genome.

Bioinformatics

After sequencing the targeted region, quality control was performed to ensure the accuracy of the data. Low-quality data were filtered out to obtain clean sequencing data. Burrows-Wheeler alignment was used to align the clean sequence to the human reference genome hg19. Genome Analysis Toolkit (GATK) was used to detect single-nucleotide and insertion/deletion polymorphisms (indel). The NCBI ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/, last accessed date 16 December 2021), the Human Gene Mutation Database (HGMD, http://www.hgmd.cf.ac.uk/ac/, last accessed date 16 December 2021), the Genome Aggregation Database (gnomAD, https://gnomad.broadinstitute.org, last accessed date 16 December 2021) and the Deafness Variation Database (DVD v8, https://deafnessvariationdatabase.org) were used to obtain the variants information, including gene information, variant consequence, minor allele frequency (MAF), altered protein function, and related disease information. The predictive score of pathogenicity of the variation was calculated, and the effect of amino acid substitution on protein structure and function was evaluated by Polyphen2 (http://genetics.bwh.harvard.edu/pph2/) and SIFT (http://provean.jcvi.org/). Pathogenicity was assessed according to the expert specification of the American Society for Medical Genetics and Genomics/ Association of Medical Pathology (ACMG/AMP) guidelines for genetic HL[17].

Sanger sequencing

Presumed pathogenic or likely pathogenic variants and variants of uncertain significance detected by WES and deafness gene panel in the probands were subsequently validated by a polymerase chain reaction (PCR) amplification and Sanger sequencing. Segregation analysis was performed on the probands and their family members. The primer sets are listed in Additional files 2: Table S2.

Results

Detected variants

Clinical features and genotypes of the pathogenic, likely pathogenic and uncertain MYO15A variants are summarized in Table 1. In particular, 4 cases were found to carry homozygous variants, 77 carried compound heterozygous variants.

Table 1 Summary of the MYO15A variants identified in this study
Full size table

In our study we have found 102 MYO15A variants, among which the most recurrent variants were c.10245_10247delCTC (0.27%, 12/4526), followed by c.10419_10423delCAGCT (0.24%, 11/4526), c.10251_10253delCTT (0.15%, 7/4526), c.4441 T > C (0.09%, 4/4526), c.4898 T > C (0.09%, 4/4526), c.3524dupA (0.07%, 3/4526), c.5964 + 3G > A (0.07%, 3/4526), c.6177 + 1G > T(0.07%, 3/4526), c.8827insT (0.07%, 3/4526) and c.9690 + 1G > A (0.07%, 3/4526). Other variants appeared only once or twice (Table 1).

Our analysis showed that the most common type of MYO15A variants was missense (44/102, 43.14%), followed by frameshift (27/102, 26.47%), nonsense (14/102, 13.72%), splice site (10/102, 9.80%), in frame (4/102, 3.92%), non-coding (2/102, 1.96%) and synonymous (1/102, 0.98%) (Fig. 1). The variants showed the various degree of HL, although the cases with the same variant type showed different phenotypes. In frame and splice variants showed more possibilities to cause profound HL, and frameshift and missense variants related to various degrees of HL (Fig. 1).

Fig. 1
figure 1

The degree of HL and the types of detected variants in the identified MYO15A variations. *The Multiple column represented the cases with the same variations showed different degrees of HL

Full size image

The variants were located in 41 of the 66 protein-coding exons of the MYO15A gene (Table 1) and identified in all domains in this study. Seventy-eight novel and 24 reported variants were identified, and all of them were confirmed by Sanger sequencing. (Fig. 2).

Fig. 2
figure 2

The locations of the detected 102 MYO15A variants. The figure shows the locations of 102 MYO15A variants correlated with NSHL found in this study. The previously reported ones are shown at the bottom. Pathogenic variants were expressed in red words, likely pathogenic variants in green words, and VUS in black words

Full size image

According to the guidelines of the ACMG/AMP on hereditary hearing loss, the variations in the MYO15A were manually classified [17, 18]. Based on the ACMG/AMP rating, ClinVar, HGVS and DVD database, respectively, the pathogenicity of the 102 MYO15A variants identified in this study included 40 pathogenic (P), 24 likely pathogenic (LP) and 38 variants with uncertain significance (VUS). (Table 2) We identified 36 cases with bi-allelic MYO15A pathogenic or likely pathogenic variants. The others with VUS in one of the alleles (LP/VUS, P/VUS and VUS/VUS) were also included in the study that classified as the best candidate of DFNB3. We also compared the severity of HL by the pathogenicity of variants. The results were inconclusive, and even the cases with the same variations showed various phenotypes (Table 3).

Table 2 Summary of the genotype–phenotype of patients in the MYO15A identified in this study
Full size table
Table 3 The severity of HL with different pathogenicity of variants
Full size table

Variants with HIGH impact (e.g., frameshift variants, splice variants, stop gain variants, etc.) were counted as protein-truncating variants (PTVs) [19]. The 81 cases were divided into groups according to the number of PTVs: 2 truncating (34 cases); 1 truncating and 1 non-truncating (37 cases); 2 non-truncating variants (13 cases) (Table 4). We compared the severity of HL among the groups. The 2 non-truncating variants group had the least number of cases (12/81), and most of them (10/12) were with profound NSHL. Thus, we suggested that cases with the monoallelic or biallelic truncating MYO15A variant may increase the risk of HL.

Table 4 The severity of HL cases with different numbers of truncating variants
Full size table

Although synonymous variation is generally considered as non-pathogenic, the variant c.8340G > A(p.Thr2780Thr) identified in the case M488 (Fig. 3) was considered to be pathogenic (PVS1_Very Strong, PM2_Moderate, PP5_Supporting) based on the ACMG/AMP classification in our cohort. In the NCBI ClinVar database, it was shown that the c.8340G > A (p.Thr2780Thr) predicted loss of exon 45 and led to a stop codon. (National Center for Biotechnology Information. ClinVar; [VCV000236038.1], https://www.ncbi.nlm.nih.gov/clinvar/variation/VCV000236038.1 (accessed Sept. 20, 2021).)

Fig. 3
figure 3

The audiograms and the pedigree of case M488. a Pedigree of case M488 and her family members. b All case M488 and her parents had the same profound HL of total deafness type

Full size image

Clinical findings

Among the 2263 cases from 1842 families with NSHL included in this study, including 1215 males and 1048 females. Age ranged from a few days after birth to 65 years with a mean age of 15.01 ± 13.67 years and the median age of 7.92 years. In our cohort, 1654 cases had prelingual HL and 609 had postlingual HL; 71 cases were mild, 238 were moderate, 179 were moderately severe, 512 were severe and 1263 were profound HL.

There were 81 (3.58%, 81/2263) cases from 74 families identified with at least one MYO15A pathogenic or likely pathogenic variant, or uncertain significant variant. Among them, 45 were males and 36 females, aged from 3 months to 43 years, with an average age of 10.41 ± 10.32 years. The ethnic distribution among the cases was as follows: one case was belonged to Korean ethnic group, one of Manchu, one of Tujia, while the others were all Han. None of the participants had a history of using aminoglycoside antibiotics.

Most of the audiological assessments and clinical history of the affected members showed a prelingual (92.59%, 75/81), symmetrical (97.53%, 79/81), bilateral (100%, 81/81), non-syndromic (100%, 81/81), sensorineural (100%, 81/81) HL (Fig. 4). Only a few showed a postlingual (7.41%, 6/81) and asymmetrical (2.47%, 2/81) HL. Analysis of the high-resolution CT scan of the temporal bone in the affected members showed a normal middle and inner ear structure.

Fig. 4
figure 4

Audiological phenotype of MYO15A-related HL

Full size image

The cases showed large variations in the degree of HL. The degree of HL was profound in 61 cases (75.30%, 61/81), severe in 12 (14.81%, 12/81), moderately severe in 4 (4.94%, 4/81), moderate in 3 (3.70%, 3/81) and mild in 1 (1.22%, 1/81). The last had the right ear with a mild HL and the left ear with a profound HL. Audiogram forms showed 6 cases with a flat type, 50 cases with total deafness, 10 cases with a descending type, whereas 24 remained undefined.

The age of onset among cases ranged from a few days after birth to 41 years. The hearing loss in 79.01% (64/81) of the cases appeared at birth, in 13.58% (11/81) was detected during the first 1–3 years, in 6.17% (5/81) HL arose around the age of 4–10 years, in 1.23% (1/81) was reported after 18 years (with severe deafness in the left ear and moderate deafness in the right ear, especially at the age of 41). (Fig. 5).

Fig. 5
figure 5

Age of onset of MYO15A-related HL

Full size image

In our study, it was found that the genotype–phenotype correlation between the variants in the MYO15A gene and the HL in some cases was different from that of the others. For example, the cases M80 and M646 with an asymmetric unilateral severe deafness bore the compound heterozygous variants c.2957delC(p.Thr986fs)/c.9478C < T(p.Leu3160Phe) and c.1179insC(p.Glu396Argfs*36)/c.1261C > T (p.Pro421Ser), respectively. (Fig. 6).

Fig. 6
figure 6

The audiograms and the pedigree of case M80 (a, b) and M646 (c, d)

Full size image

Affected subjects also showed progression with the different onset of HL. Case M623 with c.10251_10253delCTT homozygous variants was found in this study, who passed the hearing screening at birth, but was diagnosed with HL at the age of 3 and the HL demonstrated a progressive trait. His brother carrying the same c.10251_10253delCTT homozygous variant showed severe bilateral sensorineural HL at the age of 3 years.

Through the telephone follow-up of 56 MYO15A-related cases, the effect of using hearing aids and cochlea implants was satisfactory in most of the participants.

Discussion

Mutations in MYO15A were initially identified in HL individuals of consanguineous families from Bengkala, Bali in 1995 [12, 13]. Screening for the reported variants in the MYO15A gene with 66 exons was a very difficult and expensive task at that time. Therefore, the MYO15A gene was rarely sequenced in familial segregated deafness unless significant genetic linkage data implicated the presence of the DFNB3 locus. Instead, efforts were invested in screening for variations in smaller genes that have been identified as important contributors to inherited HL in humans, such as GJB2, which has only one protein-coding exon. The widespread contribution of MYO15A mutations on human HL was not recognized until the NGS became cost-effective and widely adopted around the world [20]. Now, mutations in MYO15A are a widely recognized cause of recessively inherited NSHL globally. More than 200 MYO15A variants have previously been reported ranging along with the domains and motifs of the encoded by MYO15A protein myosin XVA (Table 5) [8, 11, 13, 20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68].

Table 5 Overview of published variants of the MYO15A in NSHL patients
Full size table

Many studies analyzed the mutations in the GJB2 and SLC26A4 genes among cases with NSHL in different parts of the world. The obtained results demonstrated that the prevalence of the variants in GJB2 and SLC26A4 in HL accounted for about 15% to 25% and 2% to 12.6%, respectively, all dependent on the region localized [69]. The reported frequency of MYO15A variations in HL was 1.1% to 28% in respect to the different regions [70]. Besides Farjami et al. [70] reported that the MYO15A variant frequency in NSHL was 4.9% considering the variant rate of the GJB2 gene of 20%. In our study, the estimated prevalence of MYO15A variants in NSHL was 3.58%, which was similar to Farjami’s report. Moreover, Farjami et al. [70] proposed a total of 192 recessive MYO15A variants related to HL. The evaluated proportions of the various types of variants detected by him were similar to those noticed in our study. The composition of the detected variant types was similar in the different intensities of the HL (see Fig. 1). The c.10245_ 10247delCTC variant was identified as the most recurrent HL variant in our cohort. According to the MAF of 0.000016 in the Exome Aggregation Consortium (ExAC) database, 0.000389 in East Asian population and 0.000281 in total population by gnomAD, the c.10245_10247delCTC had been previously reported pathogenic, causing ARNSHL in the Japanese, Korean and Chinese individuals [7, 21, 71]. Therefore, we suggest that this variant is the hotspot of the MYO15A-related NSHL variant in East-Asian populations.

In the past two decades, scholars worldwide have gradually made a progress in the understanding of the correlation between the genotype and the resultant phenotype of MYO15A variants. During the first decade, it was thought that the hearing phenotype of ARSNHL was congenital, bilateral, full-frequency, severe to profound sensorineural hearing loss (SNHL). In 2007, Nal et al. [22] reported for the first time that an N-terminal variant (p.Glu1112fs*1124) in the exon 2 of the MYO15A gene resulted in a mild hearing loss with residual hearing at low frequency. At that time, it was considered that the phenotype of the hearing loss in cases with MYO15A variants was closely related to the region where this gene variant was located. However, subsequent studies showed that the correlation between the genotype and phenotype of MYO15A seemed to be more complex. Notably, the congenital non-progressive NSHL was investigated as the main consequence of the MYO15A variants. Interestingly, in families with ARNSHL with the same MYO15A pathogenic variant, the degree of the hearing phenotype was different [23, 24]. Different hearing phenotypes of non-congenital binaural severe SNHL were reported. Except for the residual hearing in the low-frequency region [25], it also included congenital moderate and severe SNHL with descending hearing curve [22, 23, 26, 27], all-frequency moderate and severe SNHL [28], progressive high-frequency descending severe SNHL [29], delayed and progressive moderate and severe SNHL [7, 30]. Allelic heterogeneity is common in hearing loss and is associated with clinical phenotype heterogeneity [72]. The variability of phenotypes makes clinical diagnosis and variant interpretation in genetic hearing loss diagnosis and maintenance [17]. And in our study, we found that the MYO15A variants-related hearing phenotype of SNHL in China was similar to the previous reports.

Nevertheless, some reports showed that MYO15A pathogenic variants cause moderate-to-severe HL, although they previously had been presented to cause profound HL [7, 31]. We found three cases in our cohort with MYO15A variants in the N-terminal, motor and MyTH domains that were diagnosed with a subtle HL. The hypothesis indicated that the predicted amino acid substitutions of the intrinsically disordered N-terminal domain were structurally less menacing, leading to a subtler HL. Based on these results, we believe that MYO15A variants may be the cause leading to the postlingual onset of partial deafness, the molecular mechanism of which requires further investigation. The occurrence of this non-severe hearing phenotype may be related to the following factors: the weak pathogenicity of MYO15A alleles, the existence of modified genes to reduce the degree of HL, and the influence of environmental factors. In addition, the progress of technologies for genetic diagnosis recently has further enriched the phenotypic spectrum of MYO15A. In the past, linkage analysis was often used in the study of inbreeding hereditary ear families. Those cases with severe hearing phenotypes caused by homozygous variants were always given priority to be included in the relevant genetic research. However, with the use of the WES technology and Molecular Genetics techniques, sporadic and medium-sized families around the world started to be increasingly diagnosed, and more cases with compound heterozygous variants with different phenotypes were identified, which allowed the MYO15A variants to show more diverse phenotypic characteristics.

We have detected a synonymous variant in MYO15A which was considered as a pathogenic variant. Generally, synonymous variants are considered to be non-pathogenic and are not expected to change the function of proteins. In recent years, this paradigm has been challenged with the evidence that the changes in the codon usage affected the efficiency and speed of translation, which in turn modified the folding and function of proteins [73]. Furthermore, the possible pathogenic mechanism of the abnormal splice site caused by a single nucleotide substitution at the codon wobble site and its implication in the phenotypes of HL was often ignored. Its pathogenicity was suggested by both NCBI ClinVar and DVD databases. NCBI ClinVar database, c.8340G > A (p.Thr2780Thr) predicted loss of exon 45 (116 bp), leading to a stop codon 2803 of 3531, and was the only synonymous variant considered as pathogenic. The other synonymous variants were classified as benign, likely benign, uncertainly significant, and to some extent conflicting interpretations of pathogenicity (National Center for Biotechnology Information. ClinVar; [VCV000236038.1], https://www.ncbi.nlm.nih.gov/clinvar/variation/VCV000236038.1 (accessed Sept. 20, 2021).) Danial-Farran N et al. [32] reported that c.8340G > A (p.Thr2780Thr), in the last nucleotide of exon 46 eliminated the full exon inclusion isoform, indicating that this variant impaired splicing of exon 46. Therefore, c.8340G > A (p.Thr2780Thr) was also classified as PTV.

There was a limited understanding about the impact of MYO15A PTV across multiple phenotypes. In this study, the cases with biallelic non-truncating MYO15A variants commonly related with profound HL, and the cases with one or two truncating variants tended to show more prone to HL. Therefore, it suggested a correlation between genotype and phenotype in MYO15A-related NSHL.

Consistent with previous genetic studies, MYO15A variants are considered to play an important role in the pathogenesis of HL in China. There were several limitations of this study. First, the approach yet could not detect variants in the promoter or enhancer region and copy number variants. In addition, the follow-up time varies, some cases lack long-term follow-up results and objective evaluation, particularly the cochlear implant cases.

Conclusion

In summary, we found that a total of 3.58% of the Chinese population with NSHL were related to MYO15A variants. MYO15A variants associated with NSHL were proven by NGS and validated by Sanger sequencing. Here, we report 78 novel and 24 reported MYO15A variants, which further enriched the MYO15A variant spectrum regarding the NSHL. Auditory features of the affected individuals were consistent with that previously reported for the recessive variants in the MYO15A gene. The hearing loss in most affected individuals was severe to profound, but in a few cases showed mild to moderate deafness. We suggest that the detected large variations in the phenotype of MYO15A-related NSHL might be correlated with the epigenetics and other factors that require further investigation. Noteworthy, screening for MYO15A variants in NSHL patients is of high necessity for efficient genetic diagnosis, patients’ counseling and clinical intervention.