Genotype-phenotype correlation analysis of MYO15A variants in autosomal recessive non-syndromic hearing loss

Sensorineural hearing loss (SNHL) is one of the most common sensory defects, affecting approximately 466 million people worldwide, 34 million of whom are children (http://​www.​who.​int/​). More than 50% of congenital hearing impairment cases are attributable to genetic factors, with more than 100 syndromic and non-syndromic genes known to date. Non-syndromic hearing loss (NSHL) is further categorized as autosomal dominant, autosomal recessive, X-linked, Y-linked or mitochondrial hearing loss. Autosomal recessive non-syndromic hearing loss (ARNSHL) accounts for up to 70% of NSHL. As of Jan 2019, more than 70 genes have been reported to be responsible for ARNSHL (http://​hereditaryhearin​gloss.​org/​). Variants of GJB2, SLC26A4, MYO15A, OTOF, CDH23, MYO7A, TMC1, HGF and CIB are worldwide major contributors to ARNSHL [1], and myosin 15A (MYO15A, MIM: 602666) gene variants are the third or fourth most frequent causes of ARNSHL [2]. According to previous reports, the prevalence of MYO15A variants was 3~9.9% (19/557) among Turkish ARNSHL patients, 5.7% among Pakistani patients [1], 5.7% among Iranian patients [3], 2.71% among Korean patients, 1.67% among Japanese patients (10/600) and 2.6% among Chinese patients.

Encoded by MYO15A, myosin XVa is a member of the unconventional myosin superfamily and plays an indispensable role in the graded elongation of stereocilia and actin organization in hair cells of the inner ear, which are essential for normal hearing function [4]. As a result, MYO15A variants are responsible for non-syndromic autosomal recessive deafness (DFNB3, OMIM 600316) in humans, whereas variants of homologous Myo15 genes lead to deafness and vestibular dysfunction in mice (shaker-2) [5‐8].

Initially, the majority of MYO15A variants were documented in patients with severe to profound congenital sensorineural deafness phenotypes from consanguineous families in the Middle East and Southeast Asia. However, with the recent development of next-generation sequencing (NGS), the reported number of MYO15A variants has sharply increased in other countries and regions, such as East Asia and Europe. Contrary to previous reported severe auditory phenotypes, partial deafness with residual hearing at low frequencies was reported, always associated with pathologic variants in exon 2 affecting the encoding of the N-terminal domain. Interestingly, recent studies have reported that some variants of MYO15A affecting the function of other domains, which were expected to bring about congenital severe to profound hearing loss, also were observed to result in various milder auditory phenotypes [9, 10].

In the present study, we reported 15 Chinese MYO15A variant cases from 14 unrelated ARNSHL families and identified 16 novel variants and 12 MYO15A variants in total by targeted gene capture and high-throughput sequencing. Additionally, we analysed all reported MYO15A genotype-phenotype relationships worldwide through a review of the literature. Our work extended the MYO15A variant spectrum, enriched our knowledge of auditory phenotypes, and tried to explore the genotype-phenotype correlation in different populations in order to investigate the cause of the complex MYO15A genotype-phenotype correlation.

In 1995, the DFNB3 locus was first mapped to chromosome 17p-17q12 by linkage analysis in two large multi-generational non-consanguineous families from a remote Indonesian village, Bengkala, where 2.2% of the population (47/2185) had severe-to-profound hearing loss. The locus was subsequently further refined to chromosome 17p11.2 [70, 71]. Since 1998, when three homozygous variants of MYO15A were initially identified in three multigenerational ARNSHL families from Indonesia and India [5], more than 200 variants have been reported to date. Our results expanded the spectrum of sequence variants in the MYO15A gene. In the present study, we identified 16 novel possible pathogenic variants, including one nonsense variant, two ±1 or 2 splice site variants, three frameshift variants with evidence of pathogenicity and 9 missense variants (p.Arg1248Trp, Ala1556Thr, p.Ser1583Pro, p.Ala1608Glu, p.Leu1836Pro, p.Pro2160Leu, p.Asp2608Asn, Arg2924His, p.Leu3501Glu).

The MYO15A gene contains 67 exons and encodes several alternatively spliced transcripts, and the longest mRNA transcript contains 3530 amino acids. It includes a long N-terminal extension encoded by giant exon 2, an ATPase motor domain, two light chain binding IQ motifs, and a tail region containing two myosin-tail homology 4 (MyTH4) domains, two band 4.1 superfamily (ezrin, radixin and moesin) (FERM), a Src-homology-3 (SH3) domain and a C terminal class I PDZ-ligand domain. Additionally, based on the presence or lack of giant exon 2, there are two alternatively spliced transcripts, isoform classes II and I. Myosin XVa is a member of the unconventional myosin superfamily, which is indispensable in the graded elongation of stereocilia of cochlear and vestibular hair cells. Whirlin is a scaffolding protein that is essential for maintaining normal human hearing (DFNB31, OMIM #607084) and sight (USH2D, OMIM #611383). EPS8 is an actin capping protein and is similarly essential for human hearing (DFNB102, OMIM #615974). Myosin XVa can transport whirlin and Eps8 to the tip of stereocilia and interacts with them to form a stereocilia tip complex, which can facilitate the extremely important transformation of microvilli into mature stereocilia [72]. Shaker-2 (Myo15sh2) mice exhibit profound hearing loss and abnormal vestibular function caused by short stereocilia and by losing the normal staircase structure of stereocilia in hair cells [7, 8].

In our literature review, the number of MYO15A variants affecting the motor domain is the largest of all domains (Fig. 2c). The motor domain consists of ATP- and actin-binding sites, which can generate force and move actin filaments; therefore, it is not surprising that motor domain dysfunction affected by the MYO15A variant will lead to shorter stereocilia with an ectopic staircase structure of stereocilia associated with a severe deafness phenotype [73]. Earlier studies showed that the motor and tail regions of myosin XVa were critical for normal auditory structure and function, whereas the exact biological function of the large N-terminal extension initially remained obscure [73]. However, recent study demonstrated that the 133-kDa N-terminal domain enables myosin 15 to maintain mechanotransducing stereocilia and is essential for hearing [74]. In 2007, two mutant alleles encoding the N-terminal domain were identified that provided evidence that the class I isoform of myosin XVa with an N-terminal extension is essential for normal hearing in humans [22]. More particularly, compared with the previous phenotype involving profound hearing loss at all frequencies, a milder phenotype with residual hearing at low frequencies (250 Hz, 500 Hz) was shown for the first time. In addition, abnormally short stereocilia could be restored to a wild type-like staircase architecture by simply transfecting myosin XVa Isoform II in shaker-2 mice. Therefore, it was hypothesized that a variant in the N-terminal extension domain might be associated with functional problems or minor structural defects in the staircase architecture of hair cell stereocilia [22]. Further, the results from the later mouse model experiments suggested that the pathogenic mechanisms of the two isoforms were different. Isoform II with no N-terminal domain can transport whirlin and EPS8 to the tips of hair cell stereocilia and drive elongation of the core actin cytoskeleton. Alternatively, Isoform I, with an N-terminal domain, is responsible for stabilizing the actin cytoskeleton and preventing the disassembly of mature mechano-transduction sites [2] . This finding suggests that there may be a genotype-phenotype correlation. Variants in giant exon 2, which may seldom result in drastic changes in the protein structure and function of the inner ear, may be involved in a variety of milder hearing loss phenotypes, such as residual hearing at low frequencies, moderate-to-severe hearing loss and progressive post-lingual deafness. In our study, we reported one case (1607545) with two compound heterozygous variants (p.Pro286Serfs*15, p.Ser1176Valfs*14) in 2 exons encoding the N-terminal domain who showed residual hearing at low frequencies. It is rather remarkable that this homozygous MYO15A variant (p.Ser1176Valfs*14) had recently been identified in a non-consanguineous Chinese family. The audiograms of two affected siblings also showed moderate residual hearing at identical low frequencies, which seems to provide additional evidence of an association between the mutant allele p.Ser1176Valfs*14 and a less severe auditory phenotype, especially in Chinese families [13]. In addition, our literature review illustrated that variants in the N-terminal domain were more frequently related to less sever hearing impairment than variants in the other domains.

A genotype-phenotype correlation of MYO15A variants should predict hearing loss tendency according to affected domain, which will inform the choice of optimal hearing rehabilitation methods for individuals [43]. For instance, on the basis of this hypothetical genotype-phenotype correlation, cases with variants affecting the N-terminal extension seem to need cochlear implantation with low frequency hearing retention or hearing aids due to their milder hearing loss. However, upon review of the relevant literature, we found that the genotype-phenotype correlation of MYO15A was much more complicated. First, recent studies showed that some pathologic variants affecting domains (motor, IQ, MyTH4 1, FERM 1, SH3, MyTH4 2, FERM 2) that would have been expected to cause a severe hearing phenotype, actually resulted in a milder auditory phenotype. The above relationship was further confirmed by our finding that case (1507382) showed residual low-frequency hearing and compound heterozygosity for MYO15A (c.4666G > A, c.6177 + 1G > T), affecting the motor and MyTH4 1 domains. Conversely, some variants associated with dysfunction of the N-terminal domain cause congenital severe to profound hearing loss instead of a milder phenotype. In addition, we reviewed 12 MYO15A variants associated with both severe and milder phenotypes (Table 3). We found that affected members of the same family expressed different auditory phenotypes, even though they carried the same MYO15A variants [17, 60]. For instance, Cengiz et al. [17] reported a Turkey ARNSHL family carried a homozygous MYO15A variant (p.Tyr289*) affecting the N-terminal domain. The audiograms of three affected siblings revealed congenital severe to profound sensorineural hearing loss at all frequencies, while their mother showed progressive severe sensorineural hearing loss of a milder form, which suggested the presence of a modifier. To our knowledge, in general, autosomal dominant non-syndromic hearing loss (ADNSHL) were associated with post-lingual hearing impairment, whereas ARNSHL were associated with congenital or pre-lingual hearing impairment. However, according to the results of our literature review, there were 8 MYO15A variant cases from Japan and Korea with unconventional post-lingual hearing impairment (Table 2). It should be noted that unlike the majority of MYO15A variant cases with a stable auditory phenotype, progressive hearing loss, especially at low frequencies, was observed in both affected individuals with a less severe auditory phenotype in our study based on 3-year auditory follow-ups, which further demonstrated the phenotypic diversity of the MYO15A variant.

Genetic modifiers or environmental factors might contribute to the complex genotype-phenotype correlations of MYO15A variants [9]. In addition, it is noteworthy that in contrast to conventional genetic tests such as linkage analysis, the more cost-effective and highly efficient technology of next-generation sequencing (NGS), which was commonly employed after 2012, saw a dramatic increase in the number of pathogenic variants identified. With the use of NGS, more sporadic cases with milder auditory phenotypes from non-consanguineous families have the opportunity to be involved in genetic studies to identify pathogenic gene variants. Therefore, more MYO15A compound heterozygous variants have been identified from various ethnic groups, without being limited to consanguineous families of the Middle East. This enriches the pool of known genotypes and phenotypes of MYO15A variants. It is also worth noting that there were several cases of progressive hearing loss both in recent reports and in our present study, therefore it is necessary to focus not only on age of onset but also age of auditory testing when performing genotype-phenotype correlation analyses. Gathering sufficient historical data and long-term follow-up related to auditory testing are fundamental to establishing specific genotype-phenotype correlations.

In summary, we identified 16 novel variants and 12 reported pathogenic MYO15A variants in 15 NSHL patients. Additionally, we described two distinct auditory phenotypes, a severe phenotype and a milder phenotype, in the present study. Our work extends the spectrum of MYO15A pathogenic variants, enriches our knowledge of auditory phenotypes, and deepens our understanding of the MYO15A genotype-phenotype correlation. In the future, additional functional studies including more cases with phenotypes based on accurate assessments of hearing loss phenotypes are necessary to ascertain the MYO15A genotype-phenotype correlation and provide optimal rehabilitation methods for affected individuals through precisely targeted genetic counselling.