New SNP variants of MARVELD2 (DFNB49) associated with non-syndromic hearing loss in Chinese population

Abstract

Non-Syndromic Hearing Loss (NSHL) is a common defect in humans. Variants of MARVELD2 at the DFNB49 locus have been shown to cause bilateral, moderate to profound NSHL. However, the role of MARVELD2 in NSHL susceptibility in the chinese population has not been studied. Here we conducted a case-control study in an eastern chinese population to profile the spectrum and frequency of MARVELD2 variants, as well as the association of MARVELD2 gene variants with NSHL. Our results showed that variants identified in the chinese population are significantly different from those reported in Slovak, Hungarian, and Czech Roma, as well as Pakistani families. We identified 11 variants in a cohort of 283 NSHL cases. Through Sanger sequencing and bioinformatics analysis, we found that c.730G>A variant has detrimental effects in the eastern chinese population, and may have relatively high correlation with NSHL pathogenicity.

Non-syndromic hearing loss (NSHL) is a common defect in humans. Variants of MARVELD2 at the DFNB49 locus have been shown to cause bilateral, moderate to profound NSHL. However, the role of MARVELD2 in NSHL susceptibility in the Chinese population has not been studied. Here we conducted a case-control study in an eastern Chinese population to profile the spectrum and frequency of MARVELD2 variants, as well as the association of MARVELD2 gene variants with NSHL. Our results showed that variants identified in the Chinese population are significantly different from those reported in Slovak, Hungarian, and Czech Roma, as well as Pakistani families. We identified 11 variants in a cohort of 283 NSHL cases. Through Sanger sequencing and bioinformatics analysis, we found that c.730G>A variant has detrimental effects in the eastern Chinese population, and may have relatively high correlation with NSHL pathogenicity.

NSHL is a common human sensory defect (Smith et al., 2005; Morton and Nance, 2006). Up to now, more than 50 genes with autosomal recessive inheritance and 30 genes with autosomal dominant inheritance have been reported to be associated with NSHL (Dror and Avraham, 2009, 2010; Schraders et al., 2012). MARVELD2, which encodes tricellulin, is located on chromosome 5q13.2 and linked to the DFNB49 locus (Ramzan et al., 2005; Riazuddin et al., 2006; Mašindová et al., 2015). In the human inner ear, there are many fluid-filled compartments of different ionic compositions. The strict compartmentalization plays a significant role in hearing. Tricellulin, together with other tight junction proteins, functions as a seal control lateral ion diffusion to ensure normal hearing (Sterkers et al., 1988; Kitajiri et al., 2004). Thus, it is easy to understand that variants in MARVELD2 may cause hearing loss to varying degrees (Chishti et al., 2008). Up to now, seven single nucleotide polymorphism (SNP) variants of MARVELD2 have been reported to cause human NSHL, including c.1183-1G>A, c.1331+1G>A, c.1331+2T>C, c.1331+2deITGAG, c.1498C>T, c.1543delA, and p.C395-Q501del (Riazuddin et al., 2006; Chishti et al., 2008; Babanejad et al., 2012; Šafka Brožková et al., 2012; Mašindová et al., 2015; Nayak et al., 2015). However, the mutational spectrum and frequency of MARVELD2 in the Chinese NSHL population are still poorly understood. In the present study, a systematic mutational screening of MARVELD2 was performed in a Chinese NSHL cohort.

A total of 283 genetically unrelated NSHL Chinese subjects were recruited. Informed consent, blood samples, and clinical evaluations were obtained from all participants under protocols approved by the Ethics Committee of both Zhejiang University and Wenzhou Medical University, China.

Genomic DNA was isolated from peripheral blood as detailed previously (Zheng et al., 2015). Nine pairs of primers were used to amplify the coding sequence of MARVELD2 (Table S1). Polymerase chain reaction (PCR) products were identified by agarose gel electrophoresis, and the purified fragments were then analyzed by direct sequencing. Subsequently, bidirectional sequencing results were compared with a reference sequence from GenBank (NM_001244734).

A filter was created to analyze candidate variants, which consists of five rounds of screening, including allele frequency, amino acid conservation, amino acid substitution, detrimental prediction, and protein domain and structure analysis. The protein three dimensional (3D) structure and function were predicted by iterative threading assembly refinement (I-TASSER) (Yang et al., 2015).

All the statistical analyses were carried out by SPSS software (Version 21.0, SPSS Inc., Chicago, IL, USA). The chi-square test or Fisher’s exact test was performed to evaluate the detrimental effect of all variants detected in the affected subjects, and P-value of <0.05 was regarded as statistically significant.

In order to study the mutation spectrum of MARVELD2 in the Chinese population, we recruited 283 NSHL subjects from Zhejiang Province in China. This population consisted of 163 males and 120 females, including 191 profound, 47 severe, 37 moderate, and 8 mild hearing loss. Their ages of onset ranged from congenital to 60 years, with an average of 6.8 years.

DNA fragments spanning the entire coding region of MARVELD2 were amplified by PCR. As shown in Table 1, 11 variants localized in exons were identified and all are heterozygous in affected subjects. Except for c.1499G>A, all variants are in exon 2 (Fig. 1a) which is the largest exon in tricellulin and includes the MARVEL domain.

Table 1.

Variants in MARVELD2 gene in 283 Chinese hearing-impaired subjects

Variant	Amino acid substitution	Amino acid property change	Exon	Detrimental effect (SIFT)	Detrimental effect(PolyPhen)	Conservation (%)	Allele frequency in 283 affected subjects (%)	Allele frequency in 252 controls (%)	P-valueb (Fisher’s exact test)	Allele frequency in ExAC (%)
										East Asian	Total
*c.730G>Aa	p.G244R	Non-polar aliphatic to alkaline	2/7	Damaging	Probably damaging	100.00	1.06	0	0.032
*c.772G>A	p.V258M	Non-polar aliphatic to polar neutral	2/7	Damaging	Probably damaging	100.00	0.18	0	1.000	0	0
*c.1006C>T	p.R336W	Alkaline to aromatic	2/7	Damaging	Probably damaging	100.00	0.18	0	1.000	0	0
c.98C>T	p.T33I	Polar to non-polar	2/7	Tolerated	Benign	81.16	34.81	50.42	0	48.62	49.97
c.127G>A	p.A43T	Non-polar aliphatic to polar neutral	2/7	Damaging	Benign	85.19	0.18	0	1.000	0.01	0
c.130G>A	p.D44N	Acidic to polar neutral	2/7	Damaging	Probably damaging	88.89	0.18	0.42	0.604	0.12	0.01
c.230C>T	p.P77L	Non-polar aliphatic to non-polar aliphatic	2/7	Damaging	Probably damaging	100.00	0.18	0	1.000
c.592G>A	p.V198M	Non-polar aliphatic to polar neutral	2/7	Tolerated	Probably damaging	85.88	0.53	0.85	0.713	0.24	0.02
c.949C>G	p.R317G	Alkaline to non-polar aliphatic	2/7	Damaging	Probably damaging	100.00	0.35	0.42	1.000	0.05	0
c.1109C>T	p.A370V	Non-polar aliphatic to non-polar aliphatic	2/7	Tolerated	Benign	85.19	0.18	0	1.000	0.07	0
c.1499G>A	p.R500Q	Alkaline to polar neutral	5/7	Tolerated	Benign	62.96	0.35	0	0.501	0.13	0.01

^aVariants that passed the filter perfectly were denoted by asterisks.

^bStatistical analysis of the variants detected in both controls and affected subjects was carried out by chi-square test; P-value of <0.05 was considered statistically significant. SIFT: scale-invariant feature transform; ExAC: the Exome Aggregation Consortium

Fig. 1.

Schematic diagram of human tricellulin protein and the variants

(a) The proximate locations of 11 selected sense variants are shown; almost all these variants are in exon 2, only c.1499G>A in exon 5. Representative sequencing chromatograms of three variants which passed the filter are listed. (b) Protein structure of human tricellulin. MARVEL domains are shown in purple, and the occludin-ELL domain is in yellow. Red residues indicate the positions of mutations which pass the filter. (c) 3D structure of human tricellulin was predicted using the I-TASSER software. Protein domains are shown in grey, and ligands in green. Red spheres indicate the positions of mutations which pass the filter (Note: for interpretation of the references to color in this figure legend, the reader is referred to the web version of this article)

In order to comprehensively analyze these variants, a sophisticated filter was created. The cut-down thresholds of frequency are less than 0.05 in both population groups (1000 Genomes, as well as the In-House Database provided by BGI China that is composed of 252 unaffected Chinese subjects). Next, we filtered out neutral or quiet mutations. The changes of the amino acid sequence of the expressed protein may not necessarily affect protein structure or function. For example, the replaced amino acid may have very similar chemical properties to the amino acid of the original allele, in which case, the protein may still function normally. In this study, the c.230C>T and c.1109C>T variants caused amino acid change from Pro to Leu at position 77 and from Ala to Val at position 370, respectively. All these amino acids have the non-polar aliphatic side chain. Thus these variants were filtered out (Chasman and Adams, 2001; Teng et al., 2008). In addition, c.127G>A, c.1499G>A, c.1109C>T, c.98C>T, and c.592G>A did not pass the screen of bioinformatic tools (scale-invariant feature transform (SIFT) and PolyPhen-2) (Ng and Henikoff, 2003; Adzhubei et al., 2013). Finally, three variants c.772G>A, c.730G>A, and c.1006C>T passed the filter (Fig. 1b), and could be potential pathogenic variants responsible for NSHL. In addition, only c.730G>A had the P-value <0.05 among the three putative pathogenic variants filtered by our criteria when calculated by Fisher’s exact test. For further confirmation, we used I-TASSER to predict the structure and function of MARVELD2 (Yang and Zhang, 2015; Wang et al., 2017). As shown in Fig. 1c, c.730G>A causes a substitution of 244th glycine to arginine, which is predicted as a ligand binding site. This may lead to an abnormal interactive function of MARVELD2 and eventually cause NSHL.

MARVELD2 encodes tricellulin, which is an important component of tricellular tight junctions and is involved in the formation of tricellular contacts (Ramzan et al., 2005; Riazuddin et al., 2006). In tight junctions, tricellulin minimizes permeability to macromolecules but not to ions, and this minimization is a significant factor in normal hearing (Krug et al., 2009). Raleigh et al. (2010) showed that MARVEL, occludin, and tricellulin have distinct but overlapping functions at the tight junction. Oda et al. (2014) reported that tricellulin also regulates F-actin organization through Cdc42 during cell–cell junction formation. It had been reported that variants in MARVELD2 cause bilateral, moderate to profound NSHL (Riazuddin et al., 2006; Chishti et al., 2008). A gene-targeted knock-in (TricR497X/R497X) mouse was generated by Nayak et al. (2015), which aimed to mimic the pathology of a human MARVELD2 variant. The results showed that deafness appears to be caused either by an increase in K⁺ ion concentration around the basolateral surfaces or by an increase in small molecules such as adenosine triphosphate (ATP) around the hair bundle, leading to cellular dysfunction and degeneration (Higashi et al., 2013). Recently, Nayak et al. (2015) reported that MARVELD2 variants are responsible for about 1.5% of NSHL in Pakistani families.

In this study, we analyzed MARVELD2 variants in the Chinese population and identified an SNP variation (c.730G>A) which may have a correlation with NSHL pathogenicity in that population. The c.730G>A variant showed a significantly higher frequency in the patient population (1.06%) than in controls (0.00%). In addition, c.730G>A caused a substitution for glycine to arginine, which may result in severe change of structure and function in the ligand binding domain and eventually lead to NSHL. Thus, our results indicated c.730G>A as a pathogenic variant of NSHL pathogenicity.

Up to now, seven SNP variants of MARVELD2 causing NSHL have been reported in families of Pakistan, Slovak, Hungarian, and Czech Roma origins (Riazuddin et al., 2006; Chishti et al., 2008; Babanejad et al., 2012; Šafka Brožková et al., 2012; Mašindová et al., 2015; Nayak et al., 2015), as shown in Table 2. However, the MARVELD2 variants identified in this study were not found in any of the previously reported studies. Our work profiles the spectrum and frequency of MARVELD2 variants in the Chinese population, and this expands the MARVELD2 variants pool. In addition, the study increases knowledge on MARVELD2 variants causing NSHL, and this could be important for clinical diagnosis.

Table 2.

Summary of all the pathogenic variants in MARVELD2

Mutation	Amino acids substitution	Origin of family	Reference
c.1183-1G>A		Pakistani	Riazuddin et al., 2006
c.1331+1G>A		Pakistani	Chishti et al., 2008
c.1331+2T>C		Pakistani, Czech Roma, Slovak Roma, Hungarian Roma	Riazuddin et al., 2006; Šafka Brožková et al., 2012; Mašindová et al., 2015
c.1331+2deITGAG		Pakistani	Riazuddin et al., 2006
c.1498C>T	p.R500X	Pakistani	Riazuddin et al., 2006
c.1543delA	p.K517RfsX16	Iranian	Babanejad et al., 2012
Exons 4–5 deletion	p.C395-Q501del	Pakistani	Nayak et al., 2015
c.1006C>T	p.R336W	Chinese	This report
c.730G>A	p.G244R	Chinese	This report
c.772G>A	p.V258M	Chinese	This report
c.949C>G	p.R317G	Chinese	This report

List of electronic supplementary materials

Table S1

Primers used in PCR