Next‐generation sequencing reveals the mutational landscape of clinically diagnosed Usher syndrome: copy number variations, phenocopies, a predominant target for translational read‐through, and PEX26 mutated in Heimler syndrome

Abstract Background Combined retinal degeneration and sensorineural hearing impairment is mostly due to autosomal recessive Usher syndrome (USH1: congenital deafness, early retinitis pigmentosa (RP); USH2: progressive hearing impairment, RP). Methods Sanger sequencing and NGS of 112 genes (Usher syndrome, nonsyndromic deafness, overlapping conditions), MLPA, and array‐CGH were conducted in 138 patients clinically diagnosed with Usher syndrome. Results A molecular diagnosis was achieved in 97% of both USH1 and USH2 patients, with biallelic mutations in 97% (USH1) and 90% (USH2), respectively. Quantitative readout reliably detected CNVs (confirmed by MLPA or array‐CGH), qualifying targeted NGS as one tool for detecting point mutations and CNVs. CNVs accounted for 10% of identified USH2A alleles, often in trans to seemingly monoallelic point mutations. We demonstrate PTC124‐induced read‐through of the common p.Trp3955* nonsense mutation (13% of detected USH2A alleles), a potential therapy target. Usher gene mutations were found in most patients with atypical Usher syndrome, but the diagnosis was adjusted in case of double homozygosity for mutations in OTOA and NR2E3, genes implicated in isolated deafness and RP. Two patients with additional enamel dysplasia had biallelic PEX26 mutations, for the first time linking this gene to Heimler syndrome. Conclusion Targeted NGS not restricted to Usher genes proved beneficial in uncovering conditions mimicking Usher syndrome.


Introduction
The co-occurrence of bilateral hearing impairment (here comprehensively termed "deafness") and visual impairment, if due to retinal degeneration, is of genetic origin in most cases in industrial countries. Usher syndrome mutations account for approximately 11% of deaf and hard of hearing children, and the population prevalence was estimated to be 1/6000 (Kimberling et al. 2010). The by far most prevalent causes are mutations in the 11 genes (MYO7A, OMIM *276903; USH1C, OMIM *605242; CDH23, OMIM *605516; PCDH15, OMIM *605514; USH1G, OMIM *607696; CIB2, OMIM *605564; USH2A, OMIM *608400; ADGRV1, OMIM *602851; DFNB31/WHRN, OMIM *607928; CLRN1, OMIM *606397; PDZD7, OMIM *612971) associated with Usher syndrome (Besnard et al. 2014), an autosomal recessive trait characterized by congenital deafness and RP in the first decade (in type 1, USH1; about 35% of cases (Petit 2001)) or by progressive hearing loss and RP of later onset in USH2 (about two thirds of patients). Symptoms apart from deaf-blindness, however, may indicate other (genetic) diagnoses (e.g., disease related to mutations in PEX1, OMIM *602136, or PEX6, OMIM *601498). Especially in consanguineous families, simultaneous presence of two non-syndromic sensory deficits must be taken into account. In our comprehensive analysis of a large cohort of deaf-blindness patients clinically diagnosed as Usher syndrome, we therefore conducted both conventional Sanger and next-generation sequencing (NGS) of a large gene panel not only comprising the Usher genes but also the known genes for non-syndromic deafness and for syndromes that may comprise both sensory deficits. We efficiently established identification of CNVs from NGS data, highlighting targeted NGS as a tool for diagnosing both point mutations and copy number alterations. Simultaneous homozygosity for mutations in genes associated with isolated retinal degeneration and hearing loss (OTOA, OMIM *607038 and NR2E3, OMIM *604485), and mutations in PEX26 (OMIM *608666) in patients with additional enamel dysplasia demonstrate how rare, genetically distinct entities may mimic Usher syndrome. but did neither correspond to USH1 nor to USH2 (either because of clinical course or "plus symptoms" that were unusual for Usher syndrome) were categorized as "atypical Usher syndrome". In nine pediatric or adolescent patients with apparently non-syndromic deafness who had been referred for genetic testing of hearing loss genes (including the most important syndrome genes like those for Usher syndrome), the diagnosis was reversed (to a syndrome with RP to develop in the future) due to the genetic findings. For clarity, and although these patients had not been referred as Usher syndrome patients, they were grouped retrospectively under the clinical subtype that is usually associated with the respective gene ( Table 1).

Workflow of genetic analysis and determination of diagnostic yield
The analytic workflow depended on the assumed diagnosis and the request of the physician in charge of the patient. If the clinical diagnosis was USH2, Sanger sequencing (and possibly MLPA) of the USH2A exons was the initial step of genetic testing in most cases because of the high probability to identify the causative mutation with this approach, followed by NGS for patients without USH2A mutations. For most patients who were categorized as USH1 or atypical Usher syndrome, NGS was carried out without other precedent tests. In P135, whose symptoms indicated a peroxisome biogenesis disorder (PBD), Sanger sequencing of PEX1 and PEX6 was carried out, followed by NGS. MLPA or array-CGH analysis was conducted to verify CNVs that were indicated by quantitative analysis of NGS data (see below). In a few cases, genotyping of Usher locus-specific polymorphic microsatellite markers or genome-wide linkage analysis (as reported previously (Zaki et al. 2016)) preceded gene analysis.
When calculating the diagnostic yield, we considered patients with monoallelic mutations in a gene compatible with the respective clinical subtype as "resolved", assuming that the secondary mutations had escaped detection due to atypical extra-exonic localizations (deep-intronic, non-coding regulatory regions).
ligated to barcoded adaptors for multiplexing. Exons were targeted by an in-solution customized sequence capture library (NimbleGen). Amplified enriched DNA was subjected to NGS. Reads were mapped against the hg19 human reference genome using BWA (Li and Durbin 2009) (Reese et al. 1997), and Net-Gene2 (Brunak et al. 1991;Hebsgaard et al. 1996). SeqPilot SeqNext module (v4.0.1, JSI medical systems) was used for visualization and final assessment of SNVs. Verification of all point mutations identified by NGS was carried out by Sanger sequencing. If samples from other family members were available, segregation analyses were carried out to confirm biallelic constellationsin particular in case of compound-heterozygous mutations, but also in case of apparent homozygosity to rule out large deletions in trans to point mutations. Because the identified mutations were clearly pathogenic in almost all cases, biallelic situations are very likely true also in cases where segregation analyses were not possible.

Copy number variation analysis from NGS data
We performed copy number variation (CNV) analysis on highly covered samples sequenced on the Illumina Hise-q1500 TM system. Potential copy number alterations (CNA) were initially identified with the tools copy number and copyCaller from VarScan v2.3.6 (Koboldt et al. 2012) on mapped reads with a maximum segment size of 300. All other parameters were used with standard settings. Thereby, coverage of every target region of the sample of interest was internally normalized and compared versus normalized control data of other samples of the same run. CNVs were annotated using RefSeq gene file from UCSC (ftp://hgdownload.cse.ucsc.edu/golden-Path/ hg19/database/refGene.txt.gz). CNVs were initially taken into account if indicated by VarScan against at least 85% of the control patients and if the log 2 threshold was ≥0.6 (in case of an amplification) or ≤À0.6 (in case of a deletion).

MLPA and array-CGH
Results from CNV analysis were verified by MLPA (multiplex ligation-dependent probe amplification) analysis or, if corresponding MLPA kits were not available, by array-CGH. The following SALSA MLPA probe mixes (MRC-Holland, Amsterdam, The Netherlands) were applied: P361-A1 and P362-A1 for USH2A, and P292-A2 for PCDH15 (USH1F). In every MLPA analysis, six samples without CNVs in the investigated locus were used as negative controls.
Molecular karyotyping (array-CGH) was performed using Agilent Human Genome CGH 244A (Agilent Technologies, Santa Clara, CA, USA) according to the manufacturer's instructions. Genomic positions were defined using NCBI37/hg19. CNVs were considered if at least five contiguous oligonucleotides presented with an abnormal log 2 ratio.
Translational read-through approach for p.Trp3955* USH2A HEK293T cells (cultured at 37°C and 5% CO 2 in Dulbecco's Modified Eagle Medium with GlutaMax TM , with 10% fetal bovine serum; Invitrogen, Karlsruhe, Germany) were transiently transfected (Lipofectamine TM with PLUS TM reagent; Invitrogen, Karlsruhe, Germany) with cDNAs coding for the FN3 domains 24 and 35, the transmembrane domain, and the cytoplasmic tail (residues p.3955-4175 fused to residues p.4926-5202) of wild-type and mutant USH2A (USH2-A_p.Trp3955*), respectively. A cDNA fragment from c.12250-15996 of USH2A isoform b, encoding protein residues p.3955-5202, was amplified and inserted into the pDest SP S/F-C-Tag vector with a C-terminal Flag tag. The region of c.12910-14988 was deleted using the restriction enzymes BlpI and PmII (NEB, Frankfurt am Main, Germany). The reading frame was recovered by insertion of the bases G and C at position c.12907 of the wild-type sequence using the QuickChange Lightning Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA). The p.Trp3955* mutation was generated using the QuickChange Lightning Site-Directed Mutagenesis Kit.
After 6 h, PTC124 (Selleckchem, Houston, USA; dissolved in DMSO; Sigma-Aldrich, Deisenhofen, Germany) was applied to the culture media for 48 hours. Readthrough of the nonsense mutation was validated by indirect immunofluorescence using antibodies against Flag (Sigma-Aldrich) on methanol-fixed HEK293T cells as previously described (Goldmann et al. 2012). The amount of restored USH2A protein expression was calculated as the ratio of Flag-positive cells in PTC124-treated p.Trp3955*-transfected cells, normalized to the total amount of analyzed cells.
Cell cultures were grown on sterile cover slips and fixed using cold methanol. PBS-washed cover slips were blocked with blocking solution (0.5% cold water fish gelatin, 0.1% ovalbumin in PBS) for 30 min, followed by incubation with primary antibodies overnight at 4°C. Cover slips were incubated with secondary antibodies conjugated to Alexa 488 (Molecular Probes, Leiden, Netherlands) and DAPI (4 0 ,6-diamidino-2-phenylindole, Sigma-Aldrich) for staining of the nuclear DNA for 1 h at room temperature. PBS-washed cover slips were mounted in Mowiol 4.88 (Hoechst, Frankfurt, Germany).

High diagnostic yield with predominance of Usher syndrome mutations
Biallelic mutations (in trans constellation was either proven by segregation analysis or very likely, see Methods) were identified in the vast majority of patients (97% of USH1, 90% of USH2, and 92% of atypical Usher syndrome). When considering patients with monoallelic mutations (USH1: none; USH2: six; atypical Usher: one) as resolved, the diagnostic yield was 97% for both USH1 and USH2, and 92% for atypical Usher syndrome. In one USH1 patient and in three USH2 patients, no mutation was identified despite NGS of the aforementioned extended gene panel. The genetic diagnosis was made before onset of RP in 10 young patients with apparently isolated hearing impairment: nine with Usher syndrome due to mutations in MYO7A and USH2A, and one with a peroxisome biogenesis disorder (PBD) due to compound heterozygous PEX26 mutations. Overall, 83 alleles carried a novel mutation, several of which were observed more than once. This was often the case in patients from the KSA and other MENA countries, then often in homozygous state. The "rarest" Usher syndrome genes with mutations were as follows: USH1G (1 patient), CLRN1 (1 patient), PCDH15 (2 patients), and USH1C (4x).

CNVs account for a significant proportion of Usher syndrome mutations
We have established quantitative analysis of NGS data to detect CNVs such as deletions or duplications of one or several exons. We have previously shown that this bioinformatic tool effectively uncovers such structural mutations which escape detection in conventional approaches (PCR and Sanger sequencing of exons) if present in heterozygous state (Eisenberger et al. 2013). In our cohort, CNVs in USH genes significantly contribute to the mutational load. Compatible with its prevalence, but probably also due to its large size, USH2A is most often affected (Fig. 2). In patients with biallelic USH2A mutations, CNVs account for 10% (16/157 alleles; Table 1, Fig. 4A). Some CNVs were observed more than once and likely represent regional founder alleles. For example, a deletion of PCDH15 exons 1-3 was found in two families from Syria.
In two (not knowingly related) Saudi patients with Usher syndrome and hyperinsulinism (P96 and P97), we identified a homozygous deletion of the largest part of the USH1C gene (exons 3-27). One patient, P97, had a family history with likewise affected members and a deletion involving USH1C and ABCC8. Accordingly, high-resolution array-CGH revealed a homozygous microdeletion of approximately 123 kb on chromosome an 11p15.1 between genomic positions 17,439,772 and 17,546,526 bp (Fig. 3), defined by 14 contiguous oligomers (eight in ABCC8, MIM #600509; three in USH1C; three between ABCC8 and USH1C). The deletion breakpoints are located in intron 22 of ABCC8 and in intron 2 of USH1C, corresponding to the previously reported 11p15-p14 deletion syndrome (MIM #606528 (Bitner-Glindzicz et al. 2000)). In addition to Usher syndrome, patients with this condition present with congenital hyperinsulinism, severe enteropathy, and renal tubulopathy, and they may develop non-autoimmune diabetes in adolescence (Hussain et al. 2004;Al Mutair et al. 2013). In P96, the USH1C/ABCC8 deletion was primarily detected by NGS and confirmed by array-CGH.
p.Thr3977* USH2A : highly prevalent in USH2 and rectifiable by read-through drugs The USH2A mutation c.2299delG (p.Glu767Serfs*21) is the most prevalent USH mutation in several populations (Liu et al. 1999;Leroy et al. 2001;Pennings et al. 2004;Aller et al. 2006Aller et al. , 2010aDreyer et al. 2008). Unexpectedly, we found that the USH2A nonsense mutation, c.11864G>A (p.Trp3955*), previously reported in several studies (van Wijk et al. 2004;Le Quesne Stabej et al. 2012;Lenarduzzi et al. 2015), was even more common in our cohort, accounting for 13% of determined USH2A alleles (compared to 11% for c.2299delG; Fig. 4A). Both, c.2299delG and p.Trp3955*, have been annotated in dbSNP (rs80338903 and rs111033364, respectively), and c.2299delG has a higher minor allele frequency (MAF) than p.Trp3955* (0.07915 compared to 0.01071%; ExAC database), with no homozygotes annotated in the healthy population. Although our cohort consists of patients from diverse geographic regions and ethnic backgrounds, the largest group consists of patients of German descent (65%). The high prevalence of c.2299delG and p.Trp3955* is in accordance with the results of a recent large-scale study on Usher syndrome (Bonnet et al. 2016). In contrast to that study, however, we found predominance of the p.Trp3955* mutation in German patients where it exceeds the prevalence of c.2299delG.

Read-through of p.Trp3955* USH2A
The most prevalent mutation in our study, p.Trp3955* mutation in USH2A, alters the TGG codon at position 11864 of the cDNA sequence into a premature UAG stop codon. Targeting of such nonsense mutations by small molecules such as PTC124 (Welch et al. 2007), known as translational read-through-inducing drugs (TRIDs), has become an important therapy approach (Fig. 4B; reviewed in Nagel-Wolfrum et al. (2016)). To test this approach for the USH2A p.Trp3955* nonsense mutation, we transfected HEK293T cells with Flag-tagged wild-type (USH2A-WT) and mutant USH2A plasmids (USH2A-Trp3955*), and determined USH2A expression by indirect immunofluorescence using anti-Flag antibodies. In contrast to USH2A-WT cells (Fig. 4C), a low number of Flag-positive cells was detected in DMSO-treated USH2A-Trp3955* cells (Fig. 4C 0 ), most probably resulting from Figure 3. Contiguous gene syndrome due to a deletion of USH1C and ABCC8. (A) NGS indicated a homozygous deletion of USH1C exons 3-27 in two not knowingly related USH1 patients from Saudi Arabia, P96 and P97. (B) Array-CGH revealed that the deletion also comprises the neighboring ABCC8 gene. Thus, the alteration corresponds to a contiguous gene syndrome previously described in the USH1C gene identification study (Bitner-Glindzicz et al. 2000). The replication of this mutation in our study indicates that this is a founder mutation from the Arabian Peninsula. Figure 2. CNVs in USH genes detected by quantitative analysis of NGS data. The coverage plots illustrate the statistical readout, with the absolute coverage deduced from unique read count and as calculated by the CNV analysis mode in SeqNext (JSI Medical Systems). The coverage of affected and neighboring exons of patients (red) and controls (green) from the same NGS runs is shown in overlay schemes for comparison. While most patients harbor heterozygous deletions, reflected by approximately 50% reduction in coverage, patients P61 and P39 (the heterozygous father is shown for comparison) have homozygous deletions, reflected by virtually no coverage in the respective plot. Patient P103 had a homozygous duplication of nine CDH23 exons (19-27; also see Fig. S1) the heterozygous father is depicted for comparison. spontaneous read-through of the p.Trp3955* mutation. Application of PTC124 to the Trp3955* cells resulted in a 3.3-fold increase of USH2A expression (Fig. 4C 00 ) compared to DMSO-treated Trp3955* cells.

Simultaneous homozygosity of mutations in non-syndromic genes and a novel Heimler syndrome gene
In three patients with deaf-blindness, disease was found to be due to "non-Usher" gene mutations: Quantification of NGS reads in a Saudi patient from a consanguineous family, apparently affected by USH1, revealed a homozygous deletion of OTOA, a gene known to be associated with autosomal recessively inherited deafness, DFNB22. Because the patient's retinal phenotype (deep pigment deposits along the vascular arcades, subretinal fibrosis; delayed, depressed, and simplified scotopic flash response in the ERG) appeared compatible with a recessive NR2E3-related dystrophy (Khan et al. 2007(Khan et al. , 2010, this gene was sequenced. Indeed, a homozygous NR2E3 missense mutation, p.Arg311Gln, previously reported as a pathogenic mutation (Haider et al. 2000;Kanda and Swaroop 2009;von Alpen et al. 2015), was identified (Fig. 5).
Very recently, Heimler syndrome, characterized by the association of an "Usher-like" presentation (retinal degeneration and hearing loss) with enamel dysplasia and nail abnormalities (Heimler et al. 1991), has been found to be caused by recessive mutations in two genes, PEX1 and PEX6, known to be associated with peroxisome biogenesis disorders, PBD (Ratbi et al. 2015;Smith et al. 2016;Zaki et al. 2016). PBD-associated genes have therefore been considered in our analysis. Here, we identified compound-heterozygous mutations in another PBD-related gene, PEX26, in two patients from two families (Fig. 6A): A 14-year-old boy (P135), diagnosed with "Usher syndrome with additional abnormalities", carries two missense mutations affecting evolutionarily highly conserved residues, p.Asp43His and p.Arg98Trp (Fig. 6B,C). After birth, lack of reaction to noise was noted. At 22 months, profound hearing loss (80 dB), hepatosplenomegaly, and elevation of liver enzymes (which persisted) were diagnosed. Retinitis punctata albescens with macular involvement and significant visual loss was diagnosed at 5 1 / 2 years. Opacities of deciduous teeth indicated thin enamel, and permanent teeth showed severe enamel dysplasia in terms of amelogenesis imperfecta combined with gingival hyperplasia and progressive preeruptive crown resorption (Fig. 6D,E). The other patient (P136), a 4-year-old girl with apparently non-syndromic hearing loss, was found to carry a translation initiation codon mutation (p.Met1?) and, as P135, p.Arg98Trp. Subsequent detailed inspection of the deciduous teeth revealed enamel defects. Although development of retinal degeneration in this patient seems very likely, stressful in-depth ophthalmological investigations were not carried out now.

Discussion
Deaf-blindness is mostly of genetic origin in developed countries, and biallelic mutations in the genes for Usher syndrome, an autosomal recessive disorder, are the predominant cause. With 11 known, mostly very large genes whose mutations explain the majority of cases, Usher syndrome is a prime condition for targeted NGS. Several reports have accordingly shown that the method has matured into a powerful diagnostic tool Besnard et al. 2014;Bujakowska et al. 2014;Krawitz et al. 2014;Jiang et al. 2015;Bonnet et al. 2016).
The diagnostic yield in our study is very high and similar to a recent large-scale study on Usher syndrome patients (Bonnet et al. 2016). Many mutations were novel, confirming that the genetic basis is often "private" and not detectable by screens that focus on previously reported mutations (Cremers et al. 2007). Repeatedly observed mutations, however, were common, not confined to but particularly in patients from MENA countries where the rate of parental consanguinity is high (Table 2). This indicates regional founder mutations that may guide the genetic diagnostic approach. Among the recurrent mutations are several large CNVs, including the 11p15-p14 deletion syndrome that involves USH1C and ABCC8 in two KSA families. Hyperinsulinism is therefore a symptom to be considered in deaf(-blind) patients from the Arabian Peninsula.
Here, we aimed at a one-method approach, based on targeted NGS. To achieve this, we a) established quantitative readout of NGS data to detect CNVs, with conventional methods like MLPA and array-CGH only being used for verification of NGS-based CNV detection; b) included genes mutated in clinically overlapping conditions like Heimler syndrome; and c) simultaneously enriched genes implicated in isolated deafness to recognize cases with co-occuring non-syndromic defects mimicking a single syndrome.
Because all CNVs detected by quantitative readout of NGS data could be confirmed by MLPA or array-CGH and the majority of patients were found to carry two (either proven or very likely) biallelic mutations, we assume that probably no CNV escaped detection in our study. This eliminates a major dead corner in diagnostic testing of deaf-blindness genes and enables highly efficient testing by a single technique, targeted NGS comprising all exons of genes for Usher syndrome, clinically overlapping conditions, and the repeatedly reported deep-intronic c.7595-2144A>G mutation in USH2A. Deep-intronic mutations apart from c.7595-2144A>G USH2A (which accounted for only four alleles in our study) have been reported (Liquori et al. 2016) and would still escape detection in our non-genomic approach butgiven the small proportion of patients with monoallelic or no mutationsdo not seem to play a significant role. Very recently, however, by whole-genome sequencing, we found evidence that a deep intronic founder mutation in CLRN1 may significantly contribute to USH1 on the Arabian Peninsula (Khan et al. 2017). Such recurrent "hidden" splice mutations should be considered at least in patients with the respective ethnic background.
The clinical presentation of most patients with Usher syndrome corresponded to either USH1 or USH2. All genetically resolved patients from these two categories had mutations in Usher syndrome genes. The same applied to the nine patients diagnosed with "atypical Usher syndrome" in whom course and/or age of onset of sensorineural hearing impairment and RP did not allow for clear-cut assignment to USH1 or USH2 and who had no additional abnormalities: They were found to have atypical expressions of USH1B (MYO7A), USH1D (CDH23), USH2A, and USH2C (ADGRV1). The diagnosis had to be reversed in the three remaining patients with apparent "atypical Usher syndrome": They were found to have clinically similar conditions, a peroxisome biogenesis disorder (PBD; PEX26 mutations), or simultaneous presence of two non-syndromic conditions. Retrospectively, these patients had additional abnormalities (in case of PEX26-associated PBD) or a distinct, NR2E3-typical form of retinopathy. Detailed clinical characterization before genetic testing could have largely ruled out Usher syndrome beforehand. However, patients undergoing genetic testing for deaf-blindness usually represent a heterogeneous cohort and range from poor to excellent clinically characterized. Although most will have mutations in Usher syndrome genes, it is the geneticists' role to anticipate this problem and to equally take rare differential diagnoses into account.
In patient 93, the specific retinopathy due to NR2E3 mutation homozygosity could have indicated a diagnosis distinct from Usher syndrome, but congenital deafness resulting from the homozygous deletion of OTOA is indistinguishable from the hearing impairment in USH1. If both components, deafness and RP, present as in Usher syndrome, as we have previously reported for patients with double homozygosity for mutations in DFNB59 (deafness) and MERTK (RP) (Ebermann et al. 2007), and if the co-occurence of the two non-syndromic conditions in the index patient is not uncovered by their division in siblings, the genetic diagnosis is essential. Although the disentanglement of the genetic basis of disease in patient 93 does not change his medical follow-up, it makes a major difference in genetic counseling for the parents whose family planning was ongoing: Instead of a 25% recurrence risk for Usher syndrome in further children, the actual risk is 25% for non-syndromic deafness, 25% for non-syndromic retinopathy, and 6.25% for deaf-blindness. With the increasing diagnostic application of largescale panel NGS, whole-exome and whole-genome sequencing, it has become apparent that the co-occurence of two (or more) monogenic conditions in one patient is not so uncommon (Boycott and Innes 2017)and hard to recognize if it resembles a syndrome. Given the relatively high prevalence of carriers for (mostly recessively inherited) monogenic retinal dystrophies (with about 20% of the general population assumed to be carriers (Nishiguchi and Rivolta 2012)) and hearing impairment (assuming a monogenic cause in about 80% of the affected newborns (1 in 500) (Shearer and Smith 2012)), mimicking of Usher syndrome by both conditions should be a recurrent scenario. Although this is more likely in offspring from consanguineous parents (as in case of patient 93), migration from regions with high rates of consanguinity, like the Middle East and North Africa (MENA regions), this phenomenon will increase, for example, in Central Europe. In summary, our results demonstrate the potential of extended NGS panels that include non-Usher genes to resolve difficult genetic constellations.
Inherited retinal dystrophies are a major cause of blindness worldwide. There has been remarkable progress in different therapeutic approaches (Scholl et al. 2016), such as gene therapy and optogenetics. Gene addition with adeno-associated viral (AAV) vectors has shown to be effective in case of RPE65 in patients with Leber's congenital amaurosis (LCA) type 2. In Usher syndrome, the retinopathy component would be the primary target of gene therapy, especially in patients who still have earlystage RP or who are still non-syndromic ("only" deaf). The identification of the causative gene will therefore be of utmost importance if gene therapy should become routinely available. However, the enormous size of many genes, including USH2A, represents a major hurdle for the packaging capacity of AAV vectors. Therefore, alternative gene-based strategies for therapy or slowdown of retinal degeneration are necessary. Translational readthrough represents a promising alternative for patients with nonsense mutations (Nagel-Wolfrum et al. 2014a,b). Here, we show that PTC124 effectively induces translational read-through of the predominant p.Trp3955* USH2A nonsense mutation which accounted for 13% of determined mutant USH2A alleles in our cohort. Because 35% of Usher syndrome patients with determined mutations in our study carry nonsense mutations on at least one allele (48/138), translational read-though could be a promising therapeutic strategy for Usher syndrome patients of all genetic subtypes.
Pinpointing the molecular diagnosis can be crucial for specific prophylaxisas in case of patients P135 and P136 who both carry biallelic PEX26 mutations. While P135 expresses the full picture of Heimler syndrome (except specific nail abnormalities which do not seem to be obligate part of the syndrome (Ratbi et al. 2016;Witters et al. 2016)), the only "Heimler sign" so far in patient P136 was enamel dysplasia, probably because of her young age (3 years). This patient can benefit, in terms of early prophylaxis, from the molecular diagnosis by being monitored for signs of hepatopathy, elevation of fatty acids and retinopathy. To protect hepatic function and lipid metabolism, the patient should avoid certain medications and noxious substances (e.g., alcohol). Furthermore, and in contrast to Usher syndrome, therapeutic options could exist for patients with mild PBD who may benefit from a phytanic acid-restricted diet and extracorporeal lipid apheresis Ruether et al. 2010;Kohlsch€ utter et al. 2012). In the ideal case, an effective diet could also counteract progression or even manifestation of retinal degeneration. Nine patients (in whom biallelic Usher or, in one case, Heimler syndrome gene mutations were found) had been diagnosed with non-syndromic deafness. If AAV-based gene addition or read-through approaches should become available as regular therapies, such early diagnosed patients could benefit immensely from their early genetic diagnosis.
While exome sequencing has become a reasonable "one-test-solution" for genetically heterogeneous conditions with a significant proportion of patients lacking mutations in the known disease genes, we propose NGS (large) panel analysis for Usher(-like) syndromewith the advantage of technically feasible CNV discovery, a very high diagnostic yield, and uncovering conditions mimicking Usher syndrome.

Supporting Information
Additional Supporting Information may be found online in the supporting information tab for this article: Figure S1. Confirmation of a heterozygous intragenic deletion of CDH23 exons 19-27 in patient P103 by array-CGH (244k Agilent Technologies microarray). Table S1. Deafness genes analyzed by targeted NGS.