The purpose of this study was to identify the genetic basis of hereditary hearing loss using next-generation sequencing technology in small families who could not be analyzed using the current genetic approaches. Eight small families with autosomal dominant, non-syndromic, sensorineural hearing loss were selected, and 80 target genes associated with hearing loss were screened using target capture and massively parallel sequencing methods. In current study, 5 non-synonymous mutations were confirmed in 5 of the 8 families. The causative genes underlying the hearing loss in the 3 other families are still awaiting discovery. There are 2 possible explanations for the hearing loss in these families: (1) pathogenic mutations exist in 1 of the 80 candidate genes studied but in an exon that was not covered by our sequencing (approximately 6-10%) or within intronic regulatory sequences, (2) their causative mutation is in an as-yet-unidentified hearing loss gene. However, 5 distinct causative mutations were identified in the other 5 families, including 4 missense mutations and 1 nonsense mutation. None of these mutations, except the p.I502V mutation in MYO1F, has been identified in previous studies of hereditary hearing loss.
ACTG1 (NM_001199954), which is responsible for DFNA20/26 hearing loss, encodes γ-actin, one of the non-muscle cytoskeletal proteins. This protein is predominantly expressed in cochlear hair cells and contributes to the structural maintenance of stereocilia, cuticular plates and adherens junctions
. γ-actin consists of 4 sub-domains (sub-domains 1-4), and the novel missense mutation p.D187H that is caused by a c.559 G > C transversion results in the substitution of a basic amino acid (histidine) for an acidic amino acid (aspartic acid) in sub-domain 4. Otterbein et al. found that even minor changes in this domain may lead to major effects on the structural stability of the actin polymer
. Using 2D gel electrophoresis, Verrills et al. identified that p.D187H in ACTG1 leads to the expression of more basic gamma-actin in leukemia cells, and they reported that Asp187 is a surface residue in close proximity to the ATP-binding cleft of the protein. The substitution of the histidine for aspartic acid in ACTG1 changes the charge from negative to positive, which results in reduced hydrophobicity and electrostatic interactions in this region. According to their research, mutant gamma-actin expressing leukemia cells display resistance to anti-microtubule drugs. The mutant gamma-actin expressing cells intactly retained their morphology, whereas the cells expressing wild-type gamma-actin were contracted and destroyed. These data suggest that the Asp187 residue in gamma-actin contributes to the interaction with microtubules, and the p.D187H variant inhibits depolymerization of tubulin in leukemia cells. This result suggests that the p.D187H variant could collapse the polymerization-depolymerization balance of microtubules, which leads to the destruction of cellular homeostasis in normal hair cells
. These conclusions from previous biochemical studies provide convincing evidence that p.D187H is a novel mutation that has pathogenic effects on the normal functions of γ-actin in the hair cells.
Mutation p.S288X in EYA4 (NM_004100.4) was the only nonsense mutation identified in this study. The protein encoded by EYA4 (DFNA10) is a member of the vertebrate Eya family of transcriptional activators, and it consists of two functional domains: the C-terminal EYA homolog domain and the N-terminal transactivation domain
. The EYA homolog domain and SIX family transcription factors interact to form transcriptional complexes that regulate the expression of target genes that are required for the development and maturation of the organ of Corti
. Most of the reported EYA4 mutations produce truncated proteins missing a part of the EYA homolog domain
. The novel nonsense mutation p.S288X changed Ser288 to a stop codon, which produced a truncated protein lacking the entire EYA homolog domain. It suggests that this nonsense mutation may inhibit normal development and maintenance of the organ of Corti and cause sensorineural hearing loss.
MYO1F (NM_012335.3) has been frequently proposed as a candidate hearing loss gene, because several myosin genes have been demonstrated as causative genes of non-syndromic hearing loss and MYO1F is expressed in cochlea. Recently, Zadro et al. reported that MYO1F mutations were identified in hearing loss patients, and one of the reported mutations, p.I502V, was detected in this study
. According to their study, the mutated residue, Ile502, is located near the actin-binding site in the motor domain of myosin-1 f. Through homology modeling, it was predicted that the Ile502 residue contributes to the structural stabilization of the protein by forming hydrophobic interactions with the Val444, Leu447 and Ile448 residues. Additionally, there is an ATP-binding site near Ile502. Therefore, the authors concluded that the substitution of isoleucine for valine may interrupt the hydrophobic interaction with other residues, resulting in structural instability of the protein and disturbance of ATP binding. Based on these predictions, it can be hypothesized that the p.I502V mutation likely has a pathogenic effect on the cellular function of myosin-1 F
Human diaphanous 1 protein, encoded by DIAPH1 (NM_005219.4), belongs to the formin protein family, which regulates various cellular mechanisms such as cytoskeleton remodeling and the maintenance of cell polarity in hair cells
[16, 26–29]. The formins have several functional domains, including formin homology-1 and 2 domains (FH1, FH2), which play a key role in the polymerization of unbranched actin filaments by interacting with profilin
[26, 30, 31]. The FH1 domain is characterized by consecutive proline residues. Profilin-actin complexes bind to the poly-L-proline stretch of the FH1 domain and are assembled into unbranched actin filaments allowing barbed end elongation
. The novel mutation p.P678S is located in the poly-L-proline stretch of the FH1 domain. This mutation changes a hydrophobic non-polar residue (proline) to a hydrophilic polar residue (serine) in the poly-L-proline stretch, and it will likely adversely affect actin polymerization at the barbed end. Therefore, it can be hypothesized that this mutation may disturb the interactions with partner proteins, which interrupts actin polymerization and collapses cellular polarity in hair cells of cochlea.
Transcription factors bind directly to DNA and regulate expression of target genes. Brn-3.1, encoded by the POU4F3 gene (NM_002700.2), is a well-known transcription factor that contributes to the differentiation and survival of hair cells in cochlea
[33, 34]. This protein is composed of two functional DNA-binding domains: a POU-specific domain and a POU-homeodomain. In the DNA binding domains, the molecular features of amino acids such as electric property and acidity significantly influence the maintenance of the structural conformation of the protein
[34, 35]. The p.E232K mutation detected in this study is caused by a single nucleotide substitution of adenine for guanine at nucleotide position 694, resulting in the replacement of a negatively charged glutamic acid with a positively charged lysine in an α-helix (α3) in the POU-specific domain. According to 3D modeling by Collin et al., among the 4 α-helix structures in the POU-specific domain, the α3 helix where the variant p.E232K occurs has the most direct interaction with target DNA
. Thus, altered electric properties of the Glu232 residue would directly reduce the structural stability of the domain or the strength of the interaction with the target DNA. It suggests that this missense mutation might have pathogenic effects on the maturation and survival of hair cells by failing to regulate the expression of downstream genes.
Although the pedigrees have insufficient genetic information, various probable pathogenic mutations were successfully detected by NGS technique and molecular genetic analysis in current study. Moreover, it will completely overcome the weakness of this study ensuring pathogenicity of the mutations when segregation study for the detected mutations can be performed with all of the members in the family.
According to a number of previous genetic studies, hereditary hearing loss shows distinct spectrums and prevalence of mutations in different ethnic groups
[36–40]. For example, the GJB2 gene accounts for approximately 20-40% of genetic hearing loss in Caucasian populations but only 10% of genetic hearing loss in Korean populations
[37, 38, 41–43]. In addition, the most frequent GJB2 mutation also varies among populations: c.35delG, c.235delC and c.167delT are the most common variants in Caucasian, Asian and Jewish populations, respectively
[38, 44–46]. However, this feature may not be applicable for autosomal dominant hearing loss because major hearing loss genes such as GJB2 and SLC26A4 predominantly cause autosomal recessive hearing loss. To date, no major genes have been described that cause a significant proportion of dominant hearing loss in any ethnic population. In this study, 5 mutations were detected in 5 different genes in 5 Korean families with autosomal dominant, non-syndromic, sensorineural hearing loss. Additionally, we provide the first evidence of pathogenic mutations in the ACTG1, EYA4, DIAPH1 and MYO1F genes in a Korean population. These results, as well as previous genetic studies performed by our group, suggest that there are no mutational hot spots for dominant hearing loss in the Korean population
[33, 47–50]. This conclusion is consistent with the idea that the genetic causes of autosomal dominant hearing loss are more heterogeneous than those of autosomal recessive hearing loss in most ethnic groups.
Because of these characteristics, it is difficult to diagnose and establish the exact causes of hereditary hearing loss, although it accounts for approximately 50% of all hearing loss
. Currently, several simple DNA tests are performed in medical institutions in many countries to diagnose hereditary hearing loss. However, the test evaluates only few major genes, such as GJB2, SLC26A4 and mitochondrial genes, and is unable to detect other genetic causes of hereditary hearing loss. Considering that the ultimate goal of genetic disease research is to establish basic information and genetic databases for clinical diagnosis and treatment, the use of population genetic studies to accurately understand the genetic background of diseases is essential. Although current genetic technologies, such as linkage analysis and Sanger sequencing, are very reliable methods for identifying genomic variations associated with genetic disorders, they are not well suited for the analysis of heterogeneous diseases. Next-generation sequencing (NGS) is accelerating the qualitative improvement of mutational studies for numerous heterogeneous disorders due to its ability to perform simultaneous and massively parallel sequencing. Although the current NGS technique is too expensive to be widely used, the cost of NGS is gradually decreasing, which will lead to the increased applicability of this technology. Two recent population genetic studies on hereditary hearing loss have been performed in American and Jewish populations
[52, 53]. In these studies, various genetic mutations and a founder mutation were detected using next-generation sequencing. Our study is the first report of dominant hearing loss causative gene mutations identified by the targeted-sequencing of affected individuals in an East Asian population. Our successful identification of several pathogenic mutations using target-capture and massively parallel sequencing demonstrates that gene targeted sequencing is a highly effective and powerful tool for clinical and population genetic studies of heterogeneous disorders.