Hereditary hearing impairment (HI) is a heterogenous disease that can vary in degree of disability, time of onset, accompanying symptoms, mode of inheritance and underlying molecular pathomechanism. The most common form is newborn non-syndromic HI that is mostly recessively inherited and in Caucasians is caused by alterations in the gap junction beta 2 (GJB2
) gene in approximately 50% of cases [1
]. Dominant forms of genetic HI amount to almost 20% of cases. In the latter group, HI usually develops at different stages of life and gradually progresses over time. Unlike congenital recessive forms, no common causative genes have been identified for dominant late-onset progressive HI, making gene selection for genetic screening of dominant cases particularly challenging.
In recent years, the application of massively parallel sequencing (MPS) has greatly accelerated and improved the diagnosis of Mendelian diseases [2
]. To reduce the costs of analysis and sequencing data loads, targeted capture of known HI genes and whole-exome sequencing (WES) have been successfully applied to identify causative genes and alterations in HI cohorts from numerous populations [3
Hearing rehabilitation in affected patients is achieved by hearing aid adaptation or, in severe and profound cases, by means of cochlear implantation (CI). This procedure is highly standardized and successful in congenital HI cases in young children. In recent years, the indication spectrum for CI has expanded to postlingual deaf adult individuals, although auditory performance with the bionic ear differs more widely in this patient group and is less predictable [6
]. Several studies have addressed CI performance predictability in both pediatric and adult CI recipients and some genetic factors were found to be associated with either good [7
] or poor [10
] auditory performance, leading to speculation that particular patients with altered gene expression in retrocochlear neuronal structures could be less suitable candidates for implantation [11
]. A common finding in many patients with HI is that high frequencies are affected earlier in time and more severely, while low and mid frequencies are still preserved or less affected. Electroacoustic stimulation (EAS) cochlear implants take advantage of a combinatory approach for hearing rehabilitation using a short cochlear implant electrode to stimulate only the basal turns that transduce high frequency signals while leaving apical cochlear regions intact to allow natural acoustic stimulation [12
In this study, we applied WES to investigate the genetic cause of disease in four generations of an Austrian family suffering from dominant very-late-onset progressive HI. We also demonstrate by a case report of one of the family members how the molecular diagnostics helped to predict cochlear implant performance and aided in the clinical decision to perform bilateral EAS cochlear implant surgery in the patient.
Patients and methods
Patients and clinical test battery
The patients participating in this study were recruited at the department of Otorhinolaryngology at the Medical University of Vienna, Austria. The study met the World Medical Association (WMA) Helsinki Declaration criteria and was approved by the local Ethics Committee of the Medical University of Vienna (ECS 198/2004, last extension: January 2017). After obtaining informed consent, clinical and audiometric examinations were performed on all participants. The test battery included ear inspection, interrogation of full medical history (family medical history, age of onset, accompanying symptoms, history of noise trauma or exposure to ototoxic drugs) and pure tone audiometry. The Freiburger monosyllabic speech test was used to determine word recognition scores. The Freiburger speech audiometry test consists of two independent test series that present a list of two digit numbers and a list of monosyllabic words to the patient. To measure speech intelligibility a) the intensity level needed to understand 50% of two-digit numbers correctly (speech recognition threshold; SRT), b) the percentage of monosyllabic words understood correctly at 65 dB (65 dB phenome score) and c) at 95 dB intensities (95 dB phenome score) are obtained. In normal hearing individuals, the SRT is 18.4 dB and a 100% score of correctly identified monosyllables is achieved at 50 dB. Additionally, in the patient receiving a CI, high-resolution temporal bone computed tomography (CT) and magnetic resonance imaging (MRI) of the brain and cochlea was performed prior to surgery. The index patient family suffering from late-onset progressive HI was termed family AD1.
Patient whole blood was drawn from a peripheral vein and genomic DNA was then extracted with a commercial DNA isolation kit (Invisorb blood universal kit 1000, STRATEC Molecular, Berlin, Germany). Initially, all patient samples were prescreened for variants in the GJB2
coding region as described previously [13
]. An index patient from family AD1 (member III/7) was then selected for WES.
Whole-exome libraries were created with a commercial capture kit (SureSelectXT All Exon, V5, Agilent Technologies, Santa Clara, CA) according to the manufacturer’s instructions and enriched samples then underwent paired-end sequencing on a HiSeq 2000 device (Illumina Biotechnology, San Diego, CA). A cochlin variant identified in the index patient by WES was then validated in the remaining family members for co-segregation and in single index patients from a cohort of 18 Austrian families suffering from dominant bilateral progressive HI with Sanger-based sequencing. For validation PCR, forward 5’-CAGAGGCTTGGACATCAGGA-3’ and reverse 5’-ACGTCTGCATTTCTCTCCCA-3’ primers were used.
Read mapping to the human reference genome (version hg19) and variant calling were performed using the Burrows-Wheeler read aligner [14
] and the Genome Analysis Tool Kit [15
], respectively. All single nucleotide variants, deletion and insertion variants in coding regions and splice sites along with 5 bp upstream and downstream of the adjacent intronic sequences were uploaded to an online MPS analysis platform (Genomatix GeneGrid, Genomatix, Munich, Germany) for further investigation. Filters were applied to exclude any variants with an allele frequency higher than 0.01 in the gnomAD database [16
] and only heterozygous variants matching the dominant inheritance in family AD1, were included.
Bilateral cochlear implantation was performed with a MedEl Synchrony® device (MedEl Medical Electronics, Innsbruck, Austria) coupled to a 20 mm flex EAS electrode inserted into the cochlea via a round window approach on the left and a promontorial cochleostomy on the right side.
In the present study, a family with progressive non-syndromic autosomal dominant HI of unknown genetic etiology was analyzed for the causative mutation by means of WES and a known pathogenic c.151C>T missense transversion within the COCH gene was identified. Molecular diagnostics in patients with autosomal dominant forms of HI are particularly challenging because no major, frequent causative genes have been identified to date in this patient group; however, WES allowed an effective and rapid diagnosis in family AD1. The strength of WES as a comprehensive screening and effective diagnostic tool in dominant HI patients is therefore emphasized by our findings.
This is the first family reported in Austria with a COCH mutation causing non-syndromic dominant HI. Our findings imply that COCH should be included in the genetic screening for autosomal dominant HI in Austria, though extending the screening for c.151C>T to a cohort of index patients from 18 families with autosomal dominant sensorineural HI did not identify this mutation in any of the patients under study.
Cochlin is a secretory protein that is encoded by the COCH
gene located at 14q11.2-q13. The full length protein is transcribed from 12 exons and comprises 550 amino acids that form a signal peptide, an LCCL (limulus factor C, cochlin, late gestation lung protein Lgl1) domain, which is thought to bind lipopolysaccharides and to possibly play a role in innate host defence mechanisms [22
] and two von Willebrand factor type A (vWFA) homology domains, which are ligand-binding domains often involved in collagen interactions and predominantly found in secreted proteins [24
]. Numerous isoforms of cochlin are expressed, which emerge from alternative splicing and posttranslational processing. The full length protein is uniquely and highly expressed in the inner ear [25
], particularly in the habenula perforata, in fibrocytes of the spiral limbus, spiral ligament, modiolus and beneath the sensory epithelium of the cristae ampullaris, in the maculae and in vestibular nerve fiber channels [28
-associated hearing phenotype (autosomal-dominant deafness type 9; DFNA9) is linked to cochlear abnormalities that are characterized by the accumulation of large amounts of cochlin-containing, acidophilic deposits in the area of the spiral limbus, spiral ligament, spiral osseous lamina, the stroma of cristae and maculae and vestibular nerve channels [28
]. Cochlin null mice exhibit normal hearing, which is suggestive of a dominant-negative or gain-of-function disease mechanism [31
]. It is conceivable that the deposits seen in DFNA9 disturb inner ear function and integrity leading to gradual cell degeneration and progressive hearing deterioration.
The c.151C>T (NM_001135058) mutation found in the presented family leads to the replacement of a highly conserved proline with a serine. Nuclear magnetic resonance spectroscopy (NMR) has revealed that Pro51 is located on the surface of the wild-type protein. The c.151C>T mutation plays a direct role in protein folding in vitro and prevents the generation of 3D structures characteristic of the wild-type LCCL domain [32
]. These results suggest that pathological effects from the p.Pro51Ser mutation may stem from a misfolded protein phenotype rather than from protein instability or altered interactions with hypothetical binding partners. Transfection of human cell cultures with transiently expressed p.Pro51Ser mutant proteins did not result in intracellular accumulation and extracellular distribution patterns appeared normal compared to wild-type proteins; however, the proteins might behave differently within the native ECM of the inner ear or over longer periods of time [33
]. It has been shown that mouse mutant proteins corresponding to human p.Pro51Ser form stable, secreted homo-dimers and even homo-oligomers with wild-type cochlin.; however, early dimers are not detectable under reducing conditions that dissolve disulfide bonds. Thus, abnormal disulfide pairing has been proposed to lead to dimerization. Injection of these dimeric cochlin complexes into the mouse inner ear has confirmed damaging effects [35
]. Not only could these interactions account for a dominant-negative effect, it is also conceivable that the misfolded oligomers accumulate to form the deposits characteristic of DFNA9.
The p.Pro51Ser mutation had already been detected in Dutch, Belgian and American families and is believed to result from a founder effect. The reported phenotypes are consistent as to the age of onset (mostly in the fifth decade of life), progressive nature starting at high frequencies and the occurrence of vestibular manifestations with incomplete penetrance [17
]. Initial high frequency hearing loss observed in the present patient and previously described DFNA9 patients and the very late onset in the fifth decade of life resemble the findings in patients with age-related hearing loss. It is therefore possible that genetic variability in the Cochlin gene may play a role in this condition as well. To date, 21 pathogenic mutations in COCH
have been reported [36
]. Analysis of non-Finnish European allele frequencies in the gnomAD database [16
] revealed the presence of p.Pro51Ser (frequency; 0.000009), p.Pro89His (0.00011) and p.Met512Thr (0.000009). A combined mutated allele frequency of 0.000128 indicates that 2.56 subjects in a population of 10,000 bear a heterozygous pathogenic COCH
mutation in Europe.
The full length cochlin isoform is exclusively expressed in the inner ear and patients with mutations in COCH
do not show further phenotypes despite inner ear pathologies. Furthermore, patients with COCH
mutations have been previously reported to have good auditory outcomes after cochlear implantation [9
]. The molecular diagnosis in family AD1 allowed an estimation of expected auditory performance after cochlear implantation in the index patient III/7. This was important in the clinical decision making in the patient because adult onset patients differ widely in outcomes after CI. In fact, patient III/7 showed excellent bilateral speech recognition thresholds and speech discrimination values 6 months postoperatively. In addition, residual hearing could be well preserved in the patient. Although our findings only report the outcome data of a single patient, this case report illustrates how an improved personalized molecular diagnostic could routinely influence therapeutic decisions in personalized medicine in the near future.