Hereditary etiology of non-syndromic sensorineural hearing loss in the Republic of North Ossetia–Alania

Characteristic of NSNHL patients

The cohort of patients with hearing loss from RNO–Alania consisted of 109 individuals from 93 unrelated families: 83 Ossetian: 12 Russian: three Ingush: two Avar; nine people of other ethnicities. The ethnicity of the participants was determined upon completion of a questionnaire that gathered information about three generations of direct relatives from both parents.

All of the studied patients had prelingual bilateral sensorineural hearing loss of degrees 2–4 or complete deafness as determined by an audiologist. There were no known external environmental risk factors for the development of hearing loss in the anamneses of the patients. The age of the patients ranged from 0.5 to 68 years (mean age 19.87 ± 17.10 years); the male/female sex ratio was 58/53. The diagnosis was made on the basis of the clinical picture in the hearing center of Vladikavkaz, RNO–Alania. The diagnosis was established upon birth (in ~73% cases) or within the 1^st year of life. Patients were also examined by a geneticist at the Research Centre for Medical Genetics (Moscow) to exclude syndromic forms of sensorineural hearing loss.

Characteristics of the cohort of healthy individuals

The cohort of healthy individuals consisted of 349 North Ossetians living in the city of Vladikavkaz as well as in the Prigorodny, Alagirsky, Ardonsky, Digorsky, and Irafsky Districts. Ethnic origin to the third generation was determined when compiling the questionnaire of informed consent to participate in the study.

DNA isolation

DNA from peripheral blood was isolated using the Wizard Genomic DNA Purification Kit (Promega, Madison, WI, USA) in accordance with the manufacturer’s recommendations.

Search for point mutations in the GJB2, GJB6, and GJB3 genes

The presence of the c.35delG variant in the GJB2 gene was determined by amplified fragments length analysis (AFLP). Fragments containing the first and second exons of the GJB2 gene as well as coding exons of the GJB6 and GJB3 genes were subjected to Sanger sequencing.

In the cohort of healthy individuals, the c.35delG and c.358_360delGAG variants in the GJB2 gene were screened by AFLP using the primer sequences shown in Table 1.

Search for large genome rearrangements associated with NSNHL

Copy number variations (CNVs) in causative genome loci were analyzed by multiplex ligase-dependent probe amplification (MLPA) analysis using the SALSA MLPA P163-GJB-WFS1-POU3F4 kit (MRC-Holland, Amsterdam, The Netherlands). The MLPA results were analyzed using the Coffalyser.Net program software (MRC-Holland). The set of probes for the MLPA analysis P163-GJB-WFS1-POU3F4 contains probes for each of the genes: GJB3 (1p34.3); WFS1 (4p16.1); GJB2 (13q12.11); GJB6 (13q12.11); POU3F4 (Xq21.1). The CNVs are named according to the ISNC-2020 recommendations using the GRCh37/hg19 human genome reference and the LiftOver tool (UCSC Genome Browser; https://genome.ucsc.edu/).

Whole-exome sequencing

The DNA samples from two patients were subjected to whole-exome sequencing (WES) with an BGISEQ-500 instrument using pair-end readings with a length of 2 × 100 bp and average on-target coverage of 75× with the MGIEasy Exome Capture V4 reagents (MGI, Shenzhen, China) for library preparation by Genomed Ltd, Moscow, Russia. Bioinformatic analysis was performed using an in-house software pipeline designed to detect both single-nucleotide variants (SNVs) and copy number variations (CNVs) as described earlier (Marakhonov et al., 2020). The results were further filtered for functional consequences and population frequencies (gnomAD AF <0.5% and <0.1% for recessive and dominant genes, respectively) as well as for clinical relevance according to the Human Phenotype Ontology database (Köhler et al., 2019). The pathogenicity status of the identified DNA sequence variants was established based on the recommendations of the American College of Medical Genetics and the Association of Molecular Pathology (Richards et al., 2015).

Analysis of variants in the MYO15A gene

The SNVs in the MYO15A gene detected during WES were analyzed by bidirectional Sanger sequencing of two fragments of the first coding exon and a fragment containing exon 18. Specific primers were developed (Table 2) for amplification and sequencing of fragments of the MYO15A gene (isoform NM_016239.4).

Statistical analysis

The frequency of identified alleles was calculated according to the formula: p_i = n_i/n, where n_i is the number of i-th alleles and n is the sample size (Zhivotovsky, 1991). The population frequencies of variants in different cohorts were compared using Fisher’s exact test according to a generally accepted methodology (Zhivotovsky, 1991).

Spectrum of GJB2 variants associated with hereditary hearing loss in RNO–Alania

In the total patient cohort, six different pathogenic variants of the GJB2 gene were identified (Tables 3 and 5). The most frequent mutation causing NSNHL in many European populations, GJB2:c.35delG, accounted for a significant proportion of the identified alleles in the total patient cohort (~53%, 38/71): in Russian patients (~83%, 10/12); in Ossetian patients (~38%, 18/48). Hence, in Ossetian patients the frequency of the GJB2 variant c.35delG was significantly lower than in the Russian patients (p = 0.0076) (Table 5).

The variant GJB2:c.358_360delGAG was the second most frequent in the total patient cohort (~38%; 27/71) and it was the most frequent variant in Ossetian patients (~54%; 26/48) of the identified alleles (Table 5).

The in-frame deletion variant GJB2:c.358_360delGAG, p.(Glu120del), which does not change the reading frame, leads to the loss of one amino acid. A functional study showed that GJB2:p.Glu120del cannot form homotypic slit channels, which leads to the loss of their conductivity (Bruzzone et al., 2003). This variant has been described in at least 20 patients with hearing loss including 7 homozygotes and 12 compound heterozygotes (Mani et al., 2009; The National Center for Biotechnology Information, 2022). This variant was classified as pathogenic (ClinVar Variation ID: 17006) (The National Center for Biotechnology Information, 2022).

According to Bliznetz and coauthors (Bliznetz et al., 2017), the variant GJB2:c.358_360delGAG was the sixth most common variant of the GJB2 gene (~0.9%) in NSNHL patients from the Russian Federation, but the ethnicity of patients carrying this variant was not presented in that report. The GJB2:c.358_360delGAG variant was found in healthy Chechen individuals from the Republic of Chechnya and the Republic of Ingushetia (Bliznetz et al., 2017). In the study of NSNHL in the Karachay–Cherkess Republic, the GJB2:c.358_360delGAG variant was found in four patients from three families (Circassians and Kabardians) (Petrina et al., 2017). The geographical proximity of the studied regions and the high level of migration in the historical past suggest a common source of origin of this variant among the North Caucasian peoples.

Comparing the frequencies of the GJB2:c.35delG and GJB2:c.358_360delGAG variants, we found no statistically significant difference between Ossetians and the studied populations of the North Caucasian and Volga–Ural regions (Table 4).

Two variants, GJB2:c.358_360delGAG and GJB2:c.35delG, accounted for a large proportion of GJB2-associated NSNHL in Ossetians. Their total share was ~92% of the identified alleles (Table 5). Therefore, to evaluate the frequency of NSNHL associated with the GJB2 gene in the Ossetian population, it was sufficient to test the variants c.358_360deGAG and c.35delG. A cohort of 349 healthy Ossetians was tested for the two common pathogenic GJB2 variants (Table 4). Allele frequency of the GJB2:c.35delG variant was 0.0072, that of the GJB2:c.358_360delGAG was 0.0129, so their summed frequency is 0.0201. According to the Hardy–Weinberg equilibrium, the frequency of all pathogenic alleles of the GJB2 gene (q) should be 0.0219 ((0.0201 × 100%)/91.67%); the frequency of hearing loss due to the variants of the GJB2 gene (q²) in Ossetians should be 0.0004796 (1/2,085 of population).

GJB2-negative hearing loss in RNO–Alania

For ~70% of NSNHL patients from the total patient cohort (78/111), in ~75% (63/84) of Ossetians and half (6/12) of Russians the molecular cause of hearing loss was not related to the GJB2 gene (Table 3). In five patients (four of them Ossetians), only one heterozygous pathogenic variant in the GJB2 gene was detected, two patients had the GJB2:c.313_326delAAGTTCATCAAGGG variant, two had the GJB2:c.358_360delGAG variant, and one Avar patient carried the GJB2:c.35delG variant. Hearing loss in those patients was apparently not associated with the GJB2 gene, and the heterozygous carrier state was due to the relatively high population frequency of GJB2 variants. Indeed, for two of those five patients it was possible to identify the molecular cause of hearing loss: in one, a carrier of the variant GJB2:c.358_360delGAG, the cause was found when analyzing CNVs; in the second, a carrier of the GJB2:c.313_326delAAGTTCATCAAGGG variant, the cause was found through whole exome sequencing. In addition, the latter patient’s affected sibling was found not to have the GJB2:c.313_326delAAGTTCATCAAGGG variant.

Large genome rearrangements causing X-linked deafness

The analysis of large genome rearrangements revealed pathogenic CNVs associated with Х-linked deafness in two unrelated male patients. They were Ossetians and represented sporadic cases of hearing impairment.

In a male patient, a heterozygous carrier of the GJB2:c.358_360delGAG variant, hearing loss was found to be caused by hemizygous deletion rsa[GRCh37] Xq21.1(81789851× 1,81841838_81842142×0,81979450×1) found by MLPA analysis. The patient’s healthy mother was a heterozygous carrier of both of the variants carried by the proband, the rearrangement and the GJB2:c.358_360delGAG variant. A healthy sister was a heterozygous carrier of the GJB2:c.358_360delGAG variant.

A deletion found by the same MLPA probes was identified in another male patient in whom no pathogenic variant was found in the GJB2, GJB3 and GJB6 genes. Thus, X-linked hearing loss was found in two male Ossetian patients, and this accounted for ~6% of the cases (2/33).

The abovementioned chromosome deletion located in the Xq21.1 region removed a distal cis-regulatory region ~920 kb upstream from the POU3F4 gene, and it did not affect the coding sequence. Exact chromosome breakpoints were not determined. Size could vary from 300 bp to 200,000 bp. Previously, two other deletions in the same area (~8 and ~200 kb in size) were identified in patients from Europe. They were associated with non-syndromic hearing loss (OMIM #304400; DFNX2) (The National Center for Biotechnology Information, 2022). Observations of similar chromosome region deletions in patients from different populations from West Europe and the North Caucasus indicate that the Xq21 region with its multiple conserved non-coding sequences is a hotspot for chromosome breaks. Earlier, Petrina and coauthors (Petrina et al., 2020) revealed the variant c.907C>T in the POU3F4 gene as a cause of X-linked hearing loss in a Nogai family from the Karachay–Cherkess Republic. These findings are the reason to include Xq21 deletion screening as well as target Sanger sequencing of the POU3F4 gene in the protocols of routine molecular genetic analysis of inherited deafness in the Russian Federation.

Search for disability-causing variants by NGS analysis

Due to the relatively low proportion of patients for whom a molecular cause was established in the first stage, whole-exome DNA sequencing (WES) was performed for two NSNHL patients.

One of these patients was from a family with one affected sibling and healthy parents, and they are ethnic Iron-Ossetians who were born of a non-consanguineous marriage. Three single nucleotide variants had been identified in the MYO15A gene, each in a heterozygous state. The first two were in exon 1 – c.823G>C, p.(Gly275Arg) and c.3576G>A, p.(Trp1192Ter). The third was in exon 18 – c.5192T>C, p.(Phe1731Ser). Of the three identified variants, only the c.823G>C (rs183969516) variant was registered in the gnomAD database with frequency 0.0014384 (The Genome Aggregation Database, 2022). It had been found in one study of NSNHL patients from the United States and was regarded as variant of unknown clinical significance (Shearer et al., 2013). The two other pathogenic variants were identified for the first time. The results of the pathogenicity prediction for the new variants are included in Table S1.

The NM_016239.4(MYO15A):c.3576G>A, p.(Trp1192Ter) variant leads to creation of a premature stop codon and can be regarded as probably pathogenic. The NM_016239.4(MYO15A):c.5192T>C, p.(Phe1731Ser) variant is a missense mutation, and the gnomAD database contained two missense variants that result from substitutions in neighboring nucleotides: c.5191T>C (p.Phe1731Leu; rs1459406061; found on one of 248,934 alleles with frequency 0.000004017) and c.5193C>T (p.Phe1731Phe; rs767426819; found on 18 of 280,172 alleles with frequency 0.00006424), the pathogenicity of the latter variant being considered contradictory (The National Center for Biotechnology Information, 2022).

Mutations of myosin XVA encoded by the MYO15A gene are the cause of severe congenital deafness type DFNB3 (The National Center for Biotechnology Information, 2022). The full-size transcripts of myosin XVA contain 66 exons >12 thousand bp long and encode a 365-kDa protein that differs from other myosins by the presence of a very long 1,200-a.a. N-terminal region preceding the conservative motor domain. The tail regions of myosin XVA protein contain two MyTH4 domains, two regions similar to the membrane-coupled FERM domain, and a putative SH3 domain. Northern-blot analysis showed that myosin XVA is expressed in the pituitary gland in both humans and mice. In in situ hybridization experiments, myosin XVA transcripts were observed in areas corresponding to the sensory epithelium of the cochlea and the vestibular system in the developing inner ear of mice. Immuno-staining of the organ of Corti in adult mouses showed that myosin XVA protein is concentrated in the cuticle plate and stereocilia of the cochlear sensory hair cells. These results indicate a probable role of myosin XVA in the formation or maintenance of unique, actin-rich structures of sensory hair cells of the inner ear (Liang et al., 1999).

To confirm the diagnosis, we performed segregation analysis of all three identified MYO15A variants in the proband’s family (Table 6). All three variants were confirmed by Sanger sequencing in the proband and were found in his affected sibling, each variant being in the heterozygous state, and the healthy mother was found to be a heterozygous carrier of the two variants, c.823G>C and c.5192T>C, i.e., those variants making up a single complex allele of the MYO15A gene.

Next, we searched for the three described MYO15A variants in 54 patients with NSNHL in whom the molecular cause of the disease had not been identified. The c.823G>C variant in heterozygous state was detected in four patients with frequency 0.03704 (4/108). Variants c.3576G>A and c.5192T>C were not found. With a high degree of probability, the c.823G>C variant could not be considered as pathogenic since its frequency in the population exceeds 1% according to the gnomAD database (The Genome Aggregation Database, 2022). Thus, in the affected family, c.3576G>A and c.5192T>C variants should be considered as causative.

In a Russian family with two affected patients, WES was performed for one of the patients. The variant NM_024729.3(MYH14):c.500G>A (rs776632941), p.(Arg167His) in heterozygous state was identified. Sanger sequencing of the MYH14 gene fragment confirmed the heterozygosity for the variant in the proband and his affected father. This variant was present in the gnomAD database, and it was found on five alleles from 280,642 with frequency 0.0000178 among healthy people (The Genome Aggregation Database, 2022), but this was not mentioned in the ClinVar dataset as a pathogenic mutation (The National Center for Biotechnology Information, 2022). The p.Arg167His substitution is located in a functionally significant domain of the myosin head of the protein. The results of pathogenicity prediction for this variant are included in Table S1. This variant should be regarded as having uncertain clinical significance with a level of significance PM2 (moderate piece of evidence for pathogenicity), PP3 (Multiple lines of computational evidence support a deleterious effect on the gene or gene product), and PP4 (phenotype or family history is highly specific for a disease with a single genetic etiology) according to the American College of Medical Genetics and Genomics (ACMG) criteria.

The MYH14 gene encodes one of the heavy chains of class II non-muscle myosin (NMIIC), a member of the myosin superfamily, and it is a causative gene for autosomal dominant sensorineural hearing loss (AD NSNHL, OMIM #600652, DFNA4). It is widely expressed in the inner ear, including the organ of Corti. MYH14-associated hearing loss is rare, the currently available information regarding the variant spectrum and clinical characteristics being limited. There have been reports of 43 MYH14 variants causing AD NSNHL, most of which were missense mutations (Donaudy et al., 2004; Hiramatsu et al., 2021).