Background
Hearing loss is a multifactorial disease with nearly 50% of cases being heritable and attributable to genetic defects [
1]. The annual incidence of congenital deafness is estimated to be 1:1000 newborns [
2], hence, identification of the heritable causal genes is pivotal to reduce the incidence of childhood deafness. Pendred syndrome (PDS, MIM #274600) is among the most common types of syndromic hearing impairment, and accounts for approximately 10% of hereditary deafness [
3]. PDS is clinically characterized by sensorineural deafness, enlargement of the vestibular aqueduct (EVA), goiter, and incomplete iodide organification [
3,
4]. Even though PDS could be clinically confirmed with perchlorate discharge test in combination with temporal bone scan of the ear architecture, misdiagnosis with other deafness associated disease remains as the key challenge [
5]. Late onset of goiter manifestation which usually develops after the age of 10 years and the presentation of only mild hypothyroidism have made definitive diagnosis of PDS difficult [
6,
7]. Furthermore, the degree of hearing loss could vary from mild to profound, either contributed by physical malformation or genetic defects. The limitations to getting an accurate diagnosis will prevent early treatment and may lead to mental retardation which is preventable by thyroxine replacement therapy [
8]. As such, identification of genes contributed to PDS is desirable to pave the way towards early detection of PDS as well as for carrier testing.
PDS is a complex genetic disease which may be inherited monogenically or digenically [
4,
9‐
11]. It has been well documented that biallelic mutations in
SLC26A4 (MIM #605646) is the hallmark of PDS, with a frequency of 25% [
4,
9]. Clinically,
SLC26A4 mutation has been used as genetic test to differentiate between PDS and non-syndromic familial EVA, which otherwise would not be possible to clinically distinguish, even with perchlorate discharge test [
6,
12]. However, nearly 50% probands harboured only monoallelic mutation in
SLC26A4, and for some patients, PDS is not due to S
LC26A4 gene mutations [
4]. The discovery of the involvement of other deafness genes, including
FOXI1 (MIM #601093),
KCNJ10 (MIM # 602208) and
GJB2 (MIM #121011) [
9‐
11] in combination with
SLC26A4 monoallelic mutation has proposed the existence of digenic inheritance pattern in PDS and EVA. The complexity of the genetic defects attributed to PDS suggests that a comprehensive mutational screening is warranted to identify the disease causal genes.
In the past, limitations in genomic sequencing technologies have only permitted the identification of disease-causing mutations through the candidate gene screening approach. Now, with the advent of next generation sequencing technologies, genome wide screening can now be performed in a cost-effective manner. Among these, whole exome sequencing (WES) is preferable as it focuses only on coding regions in which ~85% disease-causing mutations are located [
13]. WES has also successfully discovered genes for many rare diseases [
14]. Given that the genetic makeup of PDS remains largely unknown and complex, we performed WES to identify the genes responsible for PDS in a family with 2 affected siblings and their unaffected parents. This study will enhance our understanding about the genetic aetiology underpinning PDS, and to identify candidate genes which may be useful for precise molecular diagnosis and to guide family planning for better management of heritable deafness.
Methods
Subjects
Two siblings diagnosed to have PDS were referred for molecular evaluation and confirmation of diagnosis. These sisters were the only children of a pair non-consanguineous parents. They were 15 and 9 years old respectively at the time of referral.
Elder sister
The elder sister first presented to her local doctor at the age of 10 months with progressively enlarging goiter. Investigation then showed hypothyroidism and L-thyroxin was started. At 3 years of age, her parents noted profound hearing impairment for which she required hearing aid. There was no other significant past medical or surgical history and she was not on any long term medications. Her parents reported that she had normal speech development and was able to attend normal school. On examination, she had a large, diffused multinodular goiter. Ultrasound of the thyroid gland revealed multiple complex cysts. Tc99m thyroid scan showed a hyperfunctioning multinodular goiter with increased total iodine uptake of 34.6%. These findings were consistent with dyshormonogenesis (goitrous hypothyroidism with increased radioactive iodine uptake by the thyroid). Unfortunately, perchlorate discharge test could not be done. Hearing test confirmed bilateral profound sensorineural hearing loss > 60 dB, worse for high frequency sounds. Magnetic resonance image (MRI) of the inner ear and temporal bones showed bilateral dilated vestibular aqueduct and presence of only 1 ½ turns of the cochlear (normal: 2 ¾ turns) consistent with cochlear hypoplasia. In view of the large goiter size, patient underwent total thyroidectomy at 13 years of age. Currently she is receiving full dose of thyroxin i.e. 100 mcg daily.
Younger sister
The younger sister presented with profound hearing loss at 10 months of age associated with delayed speech development. At 16 months, she was wearing hearing aid and receiving speech therapy. She underwent cochlear implantation at 5 years of age. Her parents also noticed progressively enlarging goiter then. She had delayed speech development. She used sign language for communication and attended special school. On examination, there was a large, diffused multinodular goiter. Ultrasound revealed multiple complex cysts within the thyroid gland. Tc99m thyroid scan showed a hyperfunctioning multinodular goiter with increased total iodine uptake of 30.4%. Hearing test showed bilateral profound sensorineural hearing loss > 60 dB, worse for high frequency sounds. MRI of the inner ear and temporal bones showed bilateral dilated vestibular aqueduct and presence of only 1 ½ turns of the cochlear (normal 2 ¾ turns) consistent with cochlear hypoplasia. Total thyroidectomy was anticipated.
Exome library construction and sequencing
Peripheral blood samples were collected from all the individuals included in this study with written informed consent. Genomic DNA was extracted from peripheral blood using salt extraction method and the DNA quality was assessed using agarose gel electrophoresis. The DNA samples were of good quality (A260/A280 > 2.0; A260/A230 > 2.0) as assessed by Nanodrop (Thermo Fisher Scientific, USA). The DNA concentration was measured using Qubit dsDNA BR Assay Kit (Thermo Fisher Scientific, USA). The DNA libraries were prepared employing the Ion AmpliSeq™ Exome RDY Kit (Thermo Fisher Scientific, USA) and were then sequenced by the Ion Proton™ System (Thermo Fisher Scientific, USA), according to the manufacturer’s protocol.
Read mapping and variant calling were performed by the Ion TorrentSuite™ v4.4.2 software (Thermo Fisher Scientific, USA) using default parameters setting. The reads were aligned to human reference genome hg19, followed by variant calling using TorrentSuite™ Variant Caller v4.4.2.1. Next, the variants with SNP quality scores ≤ 30 were filtered out using SnpSift [
15], followed by annotation with ANNOVAR [
16]. Only non-synonymous variants in the coding regions (exonic, splicing) with a read depth greater than 5X were retained for further analysis. Polymorphisms with allele frequencies > 0.01 reported in 1000 Genomes Project, NHLBI Exome Sequencing Project, and Maximum Population Frequency were filtered out. Subsequently, we identify the candidate disease causing mutation by comparing the variants detected in affected sisters with their parents based on monogenic (autosomal recessive), followed by digenic and polygenic inheritance traits. Variants which fulfilled the above criteria were manually inspected using Integrative Genomics Viewer to filter out false positive variants [
17,
18]. The effect of the variants was assessed using several in silico prediction tools, including SIFT [
19], Polyphen2 [
20], MutationTaster [
21], FATHMM [
22], CADD [
23], PROVEAN [
24], and DANN [
25]. Candidate mutations which predicted deleterious by one of the above tools were further studied by searching literature database.
Sanger validation
A total of 5 predicted pathogenic candidate mutations, i.e.
SLC26A4 (ENST00000265715:c.1343C > T, p.Ser448Leu),
GJB2 (ENST00000382844:c.368C > A, p.Thr123Asn),
SCARB2 (ENST00000264896:c.914C > T, p.Thr305Met),
DUOX2 (ENST00000603300:c.1588A > T, p.Lys530*), and
DUOX2 (ENST00000603300:c.3329G > A, p.Arg1110Gln) were selected for validation by Sanger sequencing. The primers were designed using Primer3 (Additional file
1: Table S4). The regions were amplified by PCR using AmpliTaq Gold Polymerase (Thermo Fisher Scientific, USA), and the amplified products were purified using PCR Purification Kit (Qiagen, Germany), and sequenced using ABI BigDye Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher Scientific, USA). The chromatograms were visualized using BioEdit software.
Discussion
Clinically, PDS is characterized by the manifestation of a combination of severe to profound sensorineural hearing loss, inner ear anomalies such as Mondini’s dysplasia, EVA or vestibular anomalies, and goiter [
26‐
28]. Also, deafness in PDS generally profound (>60 dB) with prelingual onset [
29], and sometimes a fluctuating but worsening course [
30‐
32], consistent with a progressive lesion of the sensory organ. In this case study, clinical diagnosis confirmed both sisters were PDS: (1) MRI examination of the inner ear confirmed both sisters had EVA, an essential prerequisite for the diagnosis of PDS [
33,
34]; (2) both sisters had bilateral sensorineural hearing loss, with frequency > 60 dB; (3) both sisters are euthyroid and diagnosed with hypothyroidism at age of 1 year old; (4) the disease is potentially heritable via autosomal recessive or digenic/polygenic traits as both sisters are affected whilst their parents were unaffected.
It has been long considered that PDS is a monogenic disease attributed to
SLC46A4 biallelic mutations [
35,
36] or a digenic disease attributed to a combination of
SLC46A4 and
KCNJ10,
FOXI1 or
GJB2 [
9‐
11]. Notably, our analysis did not detect homozygous or compound heterozygous in the known PDS genes (i.e.
SLC26A4,
KCNJ10, FOX1,
GJB2) based on monogenic autosomal recessive trait, hence suggesting PDS in this family could be a more complex digenic or polygenic disorder. Interestingly, both sisters were found inherited
SLC26A4 and
GJB2 monoallellic mutation from their unaffected father. Loss of function in both
SLC26A4 and
GJB2 have been implicated in syndromic and non-syndromic hearing loss [
10,
37,
38]. Whilst
SLC26A4 defects mainly attributed to syndromic hearing loss,
GJB2 mutations accounts for up to 50% of all recessive non-syndromic hearing loss based on ethnic background [
39]. Essentially, S
LC26A4 involves in maintaining the endocochlear potential [
35,
36], whereas
GJB2 play role in auditory transduction by recycling potassium ions back to the endolymph of the cochlear duct [
40]. Given that both genes play pivotal roles in maintaining normal hearing function, we postulated that the
SLC26A4 and
GJB2 missense mutations are among the PDS driver mutations in this family. In addition, in contrast with earlier studies which have shown biallelic mutation of
SLC26A4 to be correlated with bilateral EVA, while monoallelic mutation or zero mutation of
SLC26A4 correlated with unilateral EVA [
35,
39,
41], we did not observe the association of this monoallelic
SLC26A4 mutation and the severity of cochlea anomalies. Both sisters with
SLC26A4 monoallelic mutation had incomplete partition type II abnormalities and presented with bilateral hearing loss at the age of 13 and 8 years old respectively.
As the evidence showing
SLC26A4 and/or
GJB2 monoallelic mutation was not sufficient to cause PDS in this family, we explored the implication of other possible causal mutations. Our analysis discovered pathogenic heterozygous mutation in another deafness associated gene,
SCARB2 (MIM #602257), in both siblings and mother.
SCARB2 encodes for lysosomal integral membrane protein type 2, which is involved in membrane transportation and the reorganization of endosomal and lysosomal compartment. An earlier study has shown loss of function in
SCARB2 being implicated in hearing loss, whereby the
SCARB2 knockout mice manifested cochlear deafness, which is associated with massive spiral ganglion neuron losses, concomitant with loss of the inner and outer hair cells and a strongly impaired capacity to generate an endocochlear potential [
42]. Beyond that, mutational analysis also identified that
SCARB2 mutation was associated with hearing impairment [
43,
44]. Given that both sisters inherited similar
SCARB2, SLC26A4 and
GJB2 mutations from their unaffected parents, our data supports the notion that a combination of these 3 heterozygous mutations may led to bilateral hearing loss in these 2 sisters.
In addition, we detected a compound heterozygous mutation in
DUOX2 (p.Lys530* & p.Arg1110Gln) in both siblings.
DUOX2 encodes for a key enzyme required to generate hydrogen peroxide (H
2O
2) which is essential for thyroid hormone synthesis and normal thyroid function [
45,
46]. It has been well documented that mutations in
DUOX2 are associated with congenital hypothyroidism [
47‐
51]. For instance, biallelic and triallelic mutations in
DUOX2 are associated with permanent congenital hypothyroidism, whilst mononoallelic mutation caused transient congenital hypothyroidism [
49,
51]. Mutation p.Lys530* and p.Arg1110Gln in
DUOX2 were found in patients with transient congenital hypothyroidism [
49]. Earlier studies also have shown that p.Arg1110Glu in
DUOX2 reduced H
2O
2 production (5–9%,
P < 0.01), hence contributed towards transient congenital hypothyroidism [
48,
52]. Our analysis suggested that the
DUOX2 compound heterozygous mutations in both sisters may be involved in permanent congenital hypothyroidism, and correlated with significant goiter manifestation at a young age.
DUOX2 mutational screening may be useful to detect thyroid dysfunction as compared to perchlorate discharge test, and to differentiate between PDS and other hearing loss diseases.
Taken together, our analysis suggested that PDS in this family could be a complex polygenic disorder which attributed to a combination of 3 heterozygous mutations implicated in deafness-related genes (SLC26A4:p.Ser448Leu; GJB2:p.Thr123Asn; SCARB2:p.Thr305Met), as well as a compound heterozygous mutation implicated in gene associated with thyroid function (DUOX2:p.Lys530* & p.Arg1110Gln).