WES is a powerful technique that can be used to discover causative genes in human diseases. Although WES has been integral in identifying more than 1,000 novel genes in Mendelian disorders
, there is still a need for increased efficiency of gene discovery using WES data. In this regard, we analyzed WES data from a family with a history of NSHL by focusing on three categories of genetic information: CNVs, linkage analysis, and SNVs. Utilizing these data, we undertook a stepwise multiphasic approach to identify disease-causing variations in the family.
8p23.1, which contains a beta-defensin cluster, was detected as a region with high copy number (II-9) and low copy number (II-3 and II-7) (Figure
2). The defensin cluster, containing both alpha- and beta-defensins, was previously studied as a dynamic genomic region with varying copy numbers ranging from one to twelve
. The parents had normal copy numbers, which was in contrast to the low copy numbers seen in two children and high copy number observed in one child. A total of four haplotypes of 8p23.1 may have been inherited in this family, and each parent may have had both under- and over-amplified alleles of 8p23.1. The overall copy number of a parent can appear to be normal due to compensation of copy number from over- to under-amplified alleles
. In the family in this study, GSTM1 and UGT2B17, genes with frequently reported deletions
, as well as BNTL3 and LILRB2, exhibited CNVs. We used Fisher’s exact test on the affected and unaffected family members after validating this method for 8p23.1 and GSTM1 groups to determine the amplification or deletion of multiple exons that matched the co-segregation pattern of the disease. Multiple statistically significant peaks at 8p23.1 and GSTM1 were identified, and were identical to plots from the first approach. However, there was only one statistically significant peak identified by testing the two groups that segregated with the disease, and this peak did not correlate with disease status. Thus, while WES may provide a method to identify CNV regions with highly similar sequences, determining accurate copy numbers can prove difficult.
Linkage analysis was performed to narrow down the number of candidates based on WES data. Importantly, linkage analysis with a relatively small number of markers still provides useful information. Fewer markers from WES data are available and can be obtained from a SNP microarray, and the markers that are identified may not be evenly distributed. Given these limitations, it is necessary to consider the potential disadvantages of this approach. Because we analyzed only exonic SNPs (~1% of genome-wide SNPs), we may have lost critical information located outside of exons. In addition, potential genotyping errors in linkage analyses can reduce statistical power for detecting linkage peaks or result in false positive linkage peaks
. Even so, the results obtained from the different data sets in this study confirmed the validity of our approach. Linkage analysis requires a large number of subjects to help identify putative loci. Unless a proper number of subjects are available, an informative result is difficult to obtain.
After applying linkage analysis results, the co-segregated variants were all found to be located in the loci of high LOD scores. However, linkage analysis can decrease the number of candidate variants, particularly in instances where candidate variants are widely distributed. Additional linkage analysis of WES data demonstrated a similar performance to that of SNP microarray data and simultaneously generated results during variant calling. Considering that CNVs could be also detected using this approach, the multiphasic analysis of WES data efficiently narrowed and identified candidate variants and was advantageous compared with established methods such as initial aCGH, variant calling according to WES data alone, or linkage analysis based on SNP microarray data.
Actin is a highly conserved cytoskeletal protein that plays important roles in eukaryotic cell processes such as cell division, migration, endocytosis, and contractility. Actin isoforms are classified into two groups based on expression patterns. ACTA1, ACTA2, ACTC, and ACTG2 are “muscle” actins, predominantly expressed in striated or smooth muscle, whereas ACTB and ACTG1 are cytoplasmic “non-muscle” actins
. Autosomal dominant progressive sensorineural hearing loss, DFNA20/26 (MIM: 604717), is caused by a mutation in the gamma-actin gene on chromosome 17 at q25.3. Some ACTG1 mutations are associated with Baraitser-Winter syndrome, which is characterized by developmental delay, facial dysmorphologies, brain malformations, colobomas, and variable hearing loss. The constellation of these abnormalities is suggested as the most severe phenotype of ACTG1 mutations
[24, 25]. A genome-wide screen of DFNA20 localized candidates to 17q25.3
 and mapped the causative missense mutations to highly conserved actin domains of the gamma-actin gene (ACTG1)
[27, 28]. In vivo and in vitro studies of ACTG1 indicate that it is required for reinforcement and long-term stability of actin filamentous structures of stereocilia, but not for auditory hair cell development, which is in line with the progressive nature of hearing loss related to ACTG1 mutations in humans
[29, 30]. Further, missense mutations in either ACTB or ACTG1 have recently been reported to cause Baraitser-Winter syndrome. Interestingly, of the 11 mutations that cause DFNA20
[27, 28, 31–34] and 6 mutations that cause Baraitser-Winter syndrome (see OMIM entry - *102560) that have been reported, are all missense mutations. The predicted interaction between Met305 and ATP in bovine beta-actin, a protein with a 99% identity to ACTG1, implies that the mutation of Met305 may influence ATP binding of ACTG1, which is essential for polymerization of G-actin to F-actin.
ACTG1 is predominantly expressed in intestinal epithelial and auditory hair cells
. Detection of exclusively missense mutations in this gene may imply that truncating mutations have more severe effects and might cause embryonic lethality. The hearing impaired subjects in this study (II-2, II-3, and II-5) did not report any gastrointestinal complaints. The subjects in this study required cochlear implants, recapitulating what has previously been reported regarding the management of patients with mutations in ACTG1 and resultant NSHL
. The severe phenotype and rapid progression of hearing loss to a profound level within one or two decades associated with mutations in ACTG1 necessitates an early molecular genetic diagnosis and timely auditory rehabilitation.