Strong founder effect of p.P240L in CDH23 in Koreans and its significant contribution to severe-to-profound nonsyndromic hearing loss in a Korean pediatric population

This study was approved by the Institutional Review Boards (IRBs) at the Seoul National University Bundang Hospital (SNUBH) (IRB-B-1007-105-402) and the Seoul National University Hospital (SNUH) (IRBY-H-0905-041-281). Written informed consent was obtained from all participating subjects. In case of the children, written informed consent was obtained from their parents or guardians on their behalf.

A total of 438 subjects who visited hereditary hearing loss clinics of tertiary referral centers: (SNUH and SNUBH) for their SNHL from September 2010 to October 2014 were initially enrolled in this study. Clinical evaluations were performed in this cohort, including gender, date of birth, medical history, physical examination, pure tone audiometry and when available, imaging studies (temporal bone computed tomography or magnetic resonance imaging).

All of the enrolled subjects underwent an audiologic evaluation. The pure-tone thresholds were recorded at 0.25, 0.5, 1, 2, 4 and 8 kHz. However, some infants could be recorded at only limited frequencies due to poor cooperation. The hearing threshold was calculated by averaging the thresholds of 0.5, 1, 2 and 4 kHz, and was classified as subtle (16–25 dB), mild (26–40 dB), moderate (41–70 dB), severe (71–95 dB), or profound (>95 dB).

Based on the results of the above clinical and audiologic evaluations, pediatric subjects with prelingual onset, severe-to-profound SNHL segregating either in a sporadic or an autosomal recessive fashion were selected for this study. Individuals with unilateral or phenotypic markers in the inner ear structure, such as an incomplete partition type III, were excluded. However, subjects with an enlarged vestibular aqueduct (EVA) were included to compare the CDH23 mutation frequency with those of mutations in SLC26A4 and other deafness genes in the same cohort (Fig. 1). Consequently, 128 unrelated children with seemingly nonsyndromic severe-to-profound sporadic, or arSNHL without other notable abnormality were finally included in this study. All recruited subjects were Korean.

Molecular genetic studies were conducted in 128 probands and their family members. Genomic DNA was extracted from peripheral blood using standard protocols (Gentra Puregene Blood Kit, Qiagen, cat. 158389; Venlo, Limburg, Netherlands). The nonsyndromic prelingual arSNHL subjects with an enlarged vestibular aqueduct (EVA) underwent direct PCR-based Sanger sequencing of SLC26A4. Sanger sequencing of GJB2, 3 and 6 was performed on subjects without noticeable phenotypic markers, as described previously [20] (Fig. 1).

Non-EVA samples that were also negative for mutations in connexins GJB2, GJB3, and GJB6 were subject to Sanger sequencing of p.P240L, which was the most frequent mutant allele of CDH23 in Korean cochlear implantees, as reported in our previous study [17] (Fig. 1). The primer sequence for Sanger sequencing to screen p.P240L of CDH23 was CDH23-8F(5′-CCACCATCACTCAACCTAAG-3′) and CDH23-8R (5′-ACTAACAGGGCCCATCCCTA-3). Next, targeted resequencing of 80, 129 (in collaboration with Otogenetics (Norcross, GA, USA) or 200 known deafness genes [17] including CDH23, was performed on the remaining non-EVA samples without a conclusive molecular diagnosis after sequencing of GJB2, GJB3, and GJB6, as described previously [17, 21], with the exception of those samples in which homozygous p.P240L mutations were detected at the screening of p.P240L by Sanger sequencing (Fig. 1). An additional file shows the supplementary lists of the genes included in the three targeted panels (see Additional file 1). This process identified DFNB12 children carrying two probable pathogenic CDH23 variants in a trans configuration.

The haplotype of the p.P240L allele of CDH23 in Koreans was determined to ascertain whether the mutant allele was a common founder. For this purpose, we chose four STR markers flanking the CDH23 gene (Fig. 2a) which included two markers (D10S1650 and D10S1694) previously reported to be hypervariable and informative [5, 11]. Primers for four STR markers flanking the CDH23 gene were generated based on the UCSC Genome Browser as described [21]. The sequences of the microsatellite markers were: D10S584 (5′-TCAATGGGAATGGATACC-3′) (5′-GCAGATCCGAACATGG-3′), D10S1650 (5′-GAAGCCTGTGGTCTAATGAG-3′) (5′-TTCTGGCCTCTGCAGC-3′), D10S606 (5′-TTTGAACCTGGGAGACG-3′) (5′-CATGGACATTCTGCTGC-3′), D10S1694 (5′-CCTGTCTGGCCCAGGTA-3′) (5′-AGTAGGGGTGCTGCTTGA-3′) [5, 11]. Loci were amplified using AmpliTaq DNA polymerase (Invitrogen Life Technologies, Carlsbad, CA, USA) in a Perkin Elmer 9700 thermal cycler (Perkin Elmer, Waltham, MA, USA). The genotypes of these loci on 10 chromosomes were defined using an AmpliTaq gold (Applied Biosystems, Waltham, MA, USA) touchdown protocol under the following conditions: 10 min at 95 °C; 10 cycles of 30 s at 94 °C, 30 s at 65 °C, and 30 s at 72 °C; and a further 20 cycles under the same conditions but with an annealing temperature lowered from 0.5 to 55 °C, 15 cycles of annealing at 55 °C, and a 72 °C final extension for 10 min. STR marker genotyping was performed as described previously [22].

One multiplex family (SB116) with co-segregating post-lingual onset of SNHL and a heterozygous p.P240L mutant allele of CDH23 was also included in this analysis. STR marker analyses were conducted in 11 subjects from five SNHL families (four prelingual and one post-lingual) in total, segregating p.P240L of CDH23, and also in 40 adult Korean subjects with documented normal hearing thresholds.

The allele frequency of p.P240L was compared between prelingually onset, severe-to-profound sporadic/autosomal recessive SNHL in a pediatric population and 272 normal hearing controls (control group II). Differences in STR genotypes and the distribution of haplotypes between mutant chromosomes carrying p.P240L and 80 wild-type Korean chromosomes from 40 normal-hearing subjects (control group I) were analyzed by Fisher’s exact test. One postlingual adult DFNB12 subject (SB116-208) was additionally included for the STR genotyping, while the subject was excluded for calculation of the allele frequency of p.P240L among pediatric prelingual severe to profound SNHL. Contributions of some genotypes were compared among subjects, but not between chromosomes, due to the lack of parental samples, as described previously [17, 23]. Especially, haplotypes of normal controls were constructed in a conservative way that was toward minimizing the statistical difference of haplotypes between p.P240L carriers with SNHL and normal controls.

Among 128 children with prelingual onset, nonsyndromic, sporadic or autosomal recessive, severe-to-profound SNHL, 22 (17.2 %) had two mutant alleles of GJB2 (DFNB1) and 23 (18.0 %) carried two mutant alleles of SLC26A4 (DFNB4). The detection frequency of CDH23 mutations (SH 59-133 [17], SH 164-359 [17], SB56-103 and SH 97-211) in our pediatric cohort was 4/128 (3.1 %), which was slightly higher than the 3/128 (2.3 %) and 3/128 (2.3 %) of MYO15A and MYO7A, respectively, in the same cohort. Considering the fact that we did not screen CDH23 from 22 subjects with two mutant alleles of GJB2, the actual frequency of CDH23 might be even slightly high than this figure. Our result showed that CDH23 was one of the leading causes of prelingual-onset profound nonsyndromic arSNHL in children and was a component of the suite of genes that includes SLC26A4 and GJB2, in the Korean pediatric cohort (Fig. 1). The mean age at detection of hearing loss was 2.25 years, with a range of 1–4 years. All four subjects with CDH23 mutations showed a similar phenotype and prelingual profound hearing loss with minimal residual hearing (Fig. 3). None of the four CDH23 carriers exhibited symptoms of nyctalopia, and ophthalmologic examinations revealed no definite abnormalities.

Interestingly, all four subjects carried p.P240L as either a homozygote or compound heterozygotes, suggesting that this mutant allele was either a strong common founder or a mutational hotspot in this population (Table 1) (Fig. 3).

To ascertain whether p.P240L was a causative variant in these subjects, the carrier frequency of the p.P240L allele was calculated among 272 adult controls with allegedly normal hearing thresholds (control group II). None of the normal control subjects carried the mutant allele, which further confirmed its pathogenic potential (6/256 vs 0/544, p = .001, by Fischer’s exact test) (Table 2).

p.P240L was assumed to act as a common founder mutation or a mutational hot spot. To ascertain whether p.P240L was a common founder, p.P240L-linked haplotypes of four STR markers (D10S584, D10S1650, D10S606, and D10S1694) were constructed and analyzed in five unrelated probands from five families (four prelingual [SH 59, SH164, SB56, and SH97] and one postlingual [SB116]), and in 40 adult Korean normal-hearing subjects (control group I) (Table 3; Fig. 2b). For two pediatric probands (SH164-359 and SH59-133) and one adult proband (SB116-208), haplotypes could definitely be assigned, based on information from the parents and affected siblings (Fig. 2b). For the remaining two pediatric probands (SH56-103 and SH97-211), precise haplotypes could not be assigned due to a lack of parental or sibling samples. Instead, their most likely probable haplotype was inferred from the other p.P240L-carrying subjects in our cohort (Fig. 2b). In detail, our three p.P240L-carrying probands (SH164-359, SH59-133, and SB116-208) with definitive STR haplotypes shared a common single haplotype (shaded box in Fig. 2b) in their all four alleles linked to p.P240L. Based on this, the other two subjects (SB56-103 and SH97-211) were also assumed to have this common haplotype (Fig. 2b). However, these two subjects (SB56-103 and SH97-211) might have a different haplotype linked to p.P240L against our assumption. Nonetheless, at least four (57.1 %) (shaded box in Fig. 2b) (two alleles from SH164-359, and each allele from SH59-133 and SB116-208) of the seven p.P240L-linked alleles (shaded or open box in Fig. 2b) showed identical haplotypes (Fig. 2b). In contrast, the normal controls showed diverse haplotypes for these STR markers, and the identical haplotype present in the p.P240L carriers could not be assigned to any of the normal controls (0/80) (Table 3). Consequently, p.P240L was significantly associated with a single haplotype in our cohort even under a conservative setting where SH56-103 and SH97-211 were assumed to have a different haplotype (57.1 % [4/7] vs. 0 % [0/80], p < .0001 by Fisher’s exact test). However, the identical haplotype could be inferred also in two subjects (SH56-103 and SH97-211) without any additional samples from their families by assuming the most likely haplotype, making the proportion of shared identical haplotypes 85.7 % (6/7). Conversely, when only three STR markers were considered (D10S584, D10S1650, and D10S606), seven alleles from the normal controls (8.75 %, 7/80) could potentially be matched to an identical haplotype that was strongly associated with p.P240L. Even under this assumption, a highly significant statistical association was detected between p.P240L and a single haplotype (p < .001 by Fisher’s exact test) (Table 3).