Novel digenic inheritance of PCDH15 and USH1G underlies profound non-syndromic hearing impairment

Over the last decade, genetic studies have taught us that there is a continuous spectrum of genetic influences between monogenic and oligogenic diseases. The simplest model of multifactorial inheritance is digenic, where in its original definition, two loci are necessary to express or extremely modify the severity of a phenotype. Compared to monogenic disease inheritance, digenic inheritance does not follow Mendelian segregation and is probably underdiagnosed due to the difficulty in verifying true digenic effects. However, several convincing cases of digenic inheritance have been found in genetically heterogeneous disorders including hearing impairment (HI) [1‐4]. These findings encouraged researchers when analyzing exome and genome sequence data to consider variants in related genes or similar pathways that fit a digenic disease model as candidates, which has led to additional promising reports [5‐9].

Several putative digenic recessive interactions causing non-syndromic (NS) HI and syndromic HI, e.g. Usher and Pendred syndromes have been described [1‐9]. Digenic GJB2 (Cx26) and GJB6 (Cx30) heterozygous variants are an often observed cause of HI in humans [4, 5]. A 309-kb deletion, also referred to as del (GJB6-D13S1830), which involves GJB6, causes HI in the homozygous state, or in the compound heterozygous state with a large variety of GJB2 mutations. However, this example should be considered as monogenic GJB2 autosomal recessive NSHI and not truly digenic in its underlying molecular nature, since the GJB6 deletion inactivates GJB2 [10, 11], which is its neighboring gene on chromosome 13.

There are several examples of true digenic inheritance for HI. For example, digenic inheritance of CDH23 and PCDH15 is well established [1], and has been shown to cause age-related HI in mice, and Usher Syndrome Type I in humans. Both proteins interact closely and are crucial for the normal organization of the stereocilia bundle. Digenic heterozygous mice showed degeneration of the stereocilia and a base-apex loss of hair cells and spiral ganglion cells [1]. Other described digenic cases include SLC26A4 and FOXI [2], which causes Pendred syndrome or HI associated with enlarged vestibular aqueducts (EVA) in humans or EVA in the mouse mutant, and SLC26A4 and KCNJ10 [3], which have been observed to cause HI and EVA in humans. In addition, some putative digenic inheritances have been suggested but still require further evidence or need to be replicated, such as GJB2 and TMPRSS3 [7] and MYO7A and PCDH15 [8], amongst others.

For HI, dominant ‘digenic’ additive effects of two genes have also been described, which leads to a more severe hearing loss than the effect of a single variant. For example, for a Swedish family, an additive effect of linked loci DFNA2 and DFNA11, resulted in a more severe phenotype for which the causative variants and genes have yet to be identified [12].

Digenic inheritance can refer to different scenarios [13, 14], and there is currently no clear consensus regarding the definition of digenic inheritance. The most commonly used definition, requires two loci for expression or extreme modification of the severity of a similar phenotype. There is a thin line between the digenic modification definition and genetic modifiers, as both are often used in a similar context.

The Digenic Diseases Database (DIDA) [13] classifies digenic cases into two classes which are simplifications of the original definitions provided by Schäffer [15]: 1) The first class is referred to as the ‘true digenic’ class, i.e. variants at both loci are required for expression of the disease, and neither variant alone displays a phenotype. 2) The second class is a composite class as it includes different possibilities, such as Mendelian variants plus modifiers that vary the phenotype, or dual molecular diagnoses, wherein Mendelian variants at each of the two loci segregate independently and results in a combination of both phenotypes [13]. However, there are a spectrum of scenarios possible that can blur these defined borders [14]. In OMIM (Online Mendelian Inheritance in Man), digenic inheritance is classified into two categories: Digenic dominant inheritance is defined as heterozygous mutations in two genes, while digenic recessive inheritance signifies a homozygous or compound heterozygous mutation in one gene and a heterozygous mutation in a second gene.

The digenic inheritance described in this article entails a true digenic model, in which two trans heterozygous mutations in two genes (on different chromosomes) whose protein products function closely together at the stereocilia tips in the hair cells (PCDH15 and USH1G) are required for the expression of a phenotype.

The study was approved by the Institutional Review Boards of the Quaid-i-Azam University and the Baylor College of Medicine and Affiliated Hospitals (H-17566). Written informed consent was obtained from all participating members.

DNA samples were collected from five family members of a consanguineous family with hereditary non-syndromic hearing loss (Family 4667; Fig. 1a) from the Khyber Pakhtunkhwa province in Pakistan. These samples include DNA from two affected siblings (IV:3 and IV:4), two unaffected siblings (IV:1 and IV:2) and their mother (III:2) (Fig. 1a).

Genomic DNA was extracted from peripheral blood using a phenol chloroform procedure [16]. Exomic libraries were prepared from one affected individual (IV:4) with the Roche NimbleGen SeqCap EZ Human Exome Library v.2.0 (~ 37 Mb target), following the manufacturer’s protocol. Sequencing was performed by 70 bp paired-end sequencing on a HiSeq2500/4000 instrument (Illumina Inc., San Diego, CA, USA). Reads were aligned to the Human genome (Hg19/GRC37) using the Burrows-Wheeler transform (BWA-MEM), PCR duplicates were removed with Picard MarkDuplicates, and indel realignment was performed (GATK IndelRealigner). Single nucleotide polymorphisms (SNP) s and small insertions/deletions (Indels) variants were recalibrated with BaseRecalibrator and called jointly with HaplotypeCaller (GATK), annotated with dbNSFP and ANNOVAR for further filtering and interpretation [17]. Copy number variants (CNVs) were called using CoNIFER [18] and XHMM [19].

Variants were further filtered based on location (coding region and splice region +/− 12 bp), and frequency [minor allele frequency (MAF) Genome Aggregation Database (gnomAD) < 0.005 in all populations]. Variants with a predicted damaging functional effect were identified (e.g., splice-site, non-synonymous, nonsense, etc.), and conservation scores (e.g., PhastCons, GERP), and the Combined Annotation Dependent Depletion (CADD) score were evaluated prior to testing for segregation within the pedigree. We selected both heterozygous and homozygous variants for segregation testing in the pedigree, assuming several modes of inheritance possible in this pedigree: autosomal recessive (homozygous or compound heterozygous), X-linked, germline mosaicism or parental mosaicism, and digenic.

Sanger sequencing was used to validate variants and verify segregation with the HI phenotype in the family. Primers surrounding region of interest were designed using primer3 software [20]. PCR amplified products were treated with ExoSAP-IT™ PCR Product Cleanup Reagent (ThermoFisher Scientific, Sugerland, TX) and sequenced using the BigDye terminator v3.1 cycle sequencing kit (Applied Biosystems, Foster City, CA) on an ABI 3130 Genetic Analyzer (Applied Biosystems, Foster City, CA).

Hair cells of the inner ear are mechanosensors for the detection of sound and balance/movement. At the apical surface of each hair cell is its mechanically sensitive organelle, the hair bundle, which consists of dozens of stereocilia. Mechanotransduction channels are located near stereociliary tips and open or close on deflection of the stereocilia. Tip-links stretch from the tips of stereocilia in the short and middle rows to the sides of neighboring, taller stereocilia. These Tip-links on stereocilia are made of cdh23 and pcdh15 [27]. In the Ames waltzer mice, recessive mutations of Pcdh15 cause deafness due to disorganized stereocilia bundles and degeneration of inner ear neuroepithelia [28].

Sans, the protein coded by Ush1g, interacts with the cytoplasmic domains of cdh23 and pcdh15 in vitro and is absent from the hair bundle in mice defective for either of the two cadherins [27]. Sans (Ush1g) localizes mainly to the tips of short- and middle-row stereocilia in vivo, and plays a critical role in the maintenance of molecular complex at the lower end of the tip-link [27]. Thus, Sans locates at stereocilia tips, near the location of Pcdh15. In Ush1g^−/− mice, the cohesion of stereocilia is also disrupted, and both the amplitude and the sensitivity of the transduction currents are reduced [27]. Interaction between USH1G and PCDH15 is further demonstrated in digenic heterozygous mice. +/Pcdh15^av-3J +/Ush1g^js double heterozygous mice display hearing loss, with highly significant elevated auditory brainstem response (ABR) thresholds at 3–4 months [29], suggesting Pcdh15-Ush1g epistasis [29].

In the traditional definition, epistasis describes the interaction of two or more genetic loci, which can substantially modify disease severity or result in an entirely new phenotype. In the literature within and between different fields, there are contradictions in the definitions and interpretations of epistasis [30]. Adopting the original definition of epistasis, a non-linear interaction, we describe a family where the hearing impaired members carry trans heterozygous variants in PCDH15 and USH1G and have profound HI and single variant carriers have normal hearing (Fig. 1). We cannot, confirm epistasis in vitro, i.e. biochemical epistasis [31]. We hypothesize that the biochemical function of their network is severely affected by these two variants and results in a profound HI, because both proteins function together at the stereocilia tips in the hair cells and are necessary for proper mechanotransduction. Since each gene separately is known to cause autosomal recessive HI, reduced activity/functioning of both proteins in the same close interacting network is a likely disease model.

In this study, we suggest epistasis between PCDH15 and USH1G in humans, through the study of a consanguineous family with profound hereditary HI, segregating a heterozygous and predicted damaging mutation in both PCDH15 and USH1G (Fig. 1). Digenic inheritance of hearing impairment in mice and humans suggest that the proteins interact or perform co-dependent functions in hair cells. The study of digenic diseases can help us understand more about the complex interaction within the inner ear and is an initial step towards the understanding of more complex oligogenic diseases, such as age-related hearing loss.