Whole-exome sequencing identified genes known to be responsible for retinitis pigmentosa in 28 Chinese families

• Introduction
• Methods
• Results
• Discussion

Abstract

Purpose: Retinitis pigmentosa (RP) is a group of highly heterogenetic inherited retinal degeneration diseases. Molecular genetic diagnosis of RP is quite challenging because of the complicated disease-causing mutation spectrum. The aim of this study was to explore the mutation spectrum in Chinese RP patients using next-generation sequencing technology and to explore the genotype–phenotype relationship.

Method: In this study, a cost-effective strategy using whole-exome sequencing (WES) was employed to address the genetic diagnosis of 28 RP families in China. One to two patients and zero to two healthy relatives were sequenced in each family. All mutations in WES data that passed through the filtering procedure were searched in relation to 662 gene defects that can cause vision-associated phenotypes (including 89 RP genes in the RetNet Database). All patients visiting the outpatient department received comprehensive ophthalmic examinations.

Result: Twenty-five putative pathogenic mutations of 12 genes were detected by WES and were all confirmed by Sanger sequencing in 20 (20/28, 71.4%) families, including the 12 following genes: USH2A, CYP4V2, PRPF31, RHO, RP1, CNGA1, CNGB1, EYS, PRPF3, RP2, RPGR, and TOPORS. Three families were rediagnosed as having Bietti crystalline dystrophy (BCD). USH2A (4/20, 20%) and CYP4V2 (3/20, 15%) were found to be the most frequent mutated genes. Seven novel mutations were identified in this research, including mutations in USH2A1, USH2A2, PRPF31, RP2, TOPORS, CNGB1, and RPGR. Phenotype and genotype relationships in the 12 RP genes were analyzed, which revealed later disease onset and more severe visual function defects in CYP4V2.

Conclusion: Twenty-five putative pathogenic mutations of 12 genes were detected by WES, and these were all confirmed by Sanger sequencing in 20 (20/28, 71.4%) families, including seven novel mutations. USH2A and CYP4V2 were found to be the most frequent genes in this research. Phenotype and genotype relationships were revealed, and the mutation spectrum of RP in Chinese populations was expanded in this research, which may benefit future cutting-edge therapies.

Retinitis pigmentosa (RP) is a group of inherited retinal degeneration diseases that affect retinal photoreceptor cells and RPE cells. With the slow degeneration of rod cells followed by loss of cone cells, patients suffer from progressive visual field constriction and gradual or rapid vision loss until visual acuity is severely affected in their 50s to 60s; some specific types may bring about severe vision loss in early decades. The prevalence of RP worldwide was reported to be approximately 1/4,000 [1], with a prevalence of 1:1,000 to 1:4,016 in China [2-4].

RP has varied inherited patterns, including autosomal dominant (30%–40%), autosomal recessive (50%–60%), and X-linked (5%–15%) [5]. It shows great genetic heterogeneity, and to date, there have been 89 genes reported to relate to RP in the RetNet Database. The gene spectrum of RP was reported to overlap with other inherited retinal dystrophies (IRDs), including Leber congenital amaurosis (LCA), cone-rod dystrophy (CRD), macular dystrophies, and congenital stationary night blindness (CSNB) [6].

The complicated gene spectrum and inherited pattern of RP raises great challenges to doctors and researchers for genetic diagnosis. With the increasing number of gene therapy approaches in IRDs (e.g., RPE65-associated retinal dystrophies [RDs]-Luxturna [7], MERTK-associated RDs [8], and REP1-associated RDs [9]), genetic diagnosis was not only beneficial in confirming the diagnosis, predicting disease prognosis, and providing genetic consultant advice, but it was also crucial in identifying patients who could benefit from these emerging novel therapeutic techniques. With the development of next-generation sequencing (NGS), whole-exome sequencing (WES) and panel-based NGS have been widely used in molecular genetic diagnosis of IRDs [10]. Whole-genome sequencing (WGS), which is based on non-PCR technology, can provide more information about the whole genome, including introns and areas that cannot be sequenced using WES and panel-based NGS, such as large indels and copy number variants. However, WGS is much more expensive at this stage than other methods, and it is more complicated in terms of data processing, making it inapplicable for small laboratories [11]. WES, which targets the complete protein coding region in the genome, has been reported to be successful in identifying genetic defects in 60%–80% of Mendelian diseases [12]. Compared with panel-based NGS, which comprises a well-established panel including certain genes, WES can be used to detect novel mutations in IRD patients. The decreasing cost makes it more practical to apply than other NGS approaches are.

In this study, we investigated the disease-causing genes of 28 Chinese families with a clear family history of RP through WES. The results may benefit the RP gene diagnosis and the pathogenic and genotype-phenotype study of RP.

Twenty-eight Chinese families with a diagnosis of RP were recruited for this study. Of these, 9 were autosomal dominant RP (adRP) families, 17 were autosomal recessive RP (arRP) families, and 2 were X-linked RP families. WES was performed in 83 individuals from 28 RP families (including 55 RP patients and 28 of their healthy relatives), with 2 patients and 0–2 healthy relatives sequenced in each family. All individuals who were sequenced are highlighted with genotype in Figure 1. WES achieved an average of 116.75×depth and an average of 99.88% coverage rate of the exome targeted region. The mapping rate and coverage of the targeted region of each sample are shown in Appendix 2. Sanger sequencing results of each family are listed in Appendix 3.

For 28 RP families, putative pathogenic mutations of 20 (71.4%) families were identified, including the 12 following RP genes (Table 1) [16-31]: USH2A (4/20, 20%), CYP4V2 (3/20, 15%), PRPF31 (2/20, 10%), RHO (2/20, 10%), RP1 (2/20, 10%), CNGA1 (1/20, 5%), CNGB1 (1/20, 5%), EYS (1/20, 5%), PRPF3 (1/20, 5%), RP2 (1/20, 5%), RPGR (1/20, 5%), and TOPORS (1/20, 5%). Three families with CYP4V2 mutations were rediagnosed as having Bietti crystalline dystrophy (BCD). The pedigree charts of the 20 families are listed in Figure 1. All the putative genes cosegregated with the phenotype in RP families. All suspicious mutations found in each family and the reason we choosed putative mutations were illustrated in Appendix 4. Putative genes of 7 (7/9, 77.78%) autosomal dominant families, 11 (11/17, 64.71%) autosomal recessive families, and 2 (2/2, 100%) X-linked families were identified. In total, 28 mutations were identified, including 10 (35.17%) missense mutations, 9 (32.14%) frameshift mutations, 5 (17.86%) missplicing mutations, and 4 (14.27%) truncation mutations. The mutation type spectrum of each gene is listed in Appendix 5. The following seven novel mutations were identified in this research: USH2A, c.9337dupA(p.I3113fs); USH2A, c.C10498T(p.Q3500*); PRPF31, c.967_968delGA (E323Dfs*151); RP2, c.758_761delTAAT (p.L253fs*10); TOPORS, c.2323_2324delAG (p.S775*); CNGB1, c.G2006A (p.W669*); RPGR, c.T773C (p.L258P).

From the 20 families with confirmed molecular diagnoses, 33 patients visited our outpatient department. Their clinical characteristics are listed in Table 2. The mean age of all patients was 42.9 ± 14.5 years, whereas the mean age of disease onset and mean age of visual acuity decline were 11.7 ± 9.9 years and 33 ± 9.5 years. Of the 33 patients, 26 (78.8%) had an eye with BCVA lower than 0.3, whereas 20 (60.6%) had an eye with BCVA lower than 0.1. The long duration from disease onset to molecular diagnosis and poor preserved BCVA in this research indicated a late molecular diagnosis in Chinese RP patients.

Phenotype–genotype was detected in this research. Average disease duration, average visual acuity, and average disease onset age were calculated and analyzed as shown in Figure 2. Genes on the left side of the image were found to have a more severe phenotype with shorter disease duration and poor visual acuity; genes on the right side were found to have a milder phenotype. USH2A was found to have a disease onset from adolescence, but the visual function exhibited moderate defect over 35 years of disease duration. In contrast, CYP4V2 was found to have a later disease onset from the 30s, but severe visual function defects were observed in the later 17 years.

In all 12 identified RP genes, four families (20%)—RP008, RP015, RP028, and RP033—were detected to have compound heterozygous mutations in USH2A (Table 1), with six mutations. Among all mutations detected, two novel mutations were found—namely, c.C10498T (p.Q3500*) in RP015 and c.9337dupA (p.I3113Nfs*17) in RP033. These two mutations were located in the extracellular matrix protein-related regions, making the subsequent extracellular structure of more than 2,000 amino acids untranslatable, which may have led to damaging effect for Usherin protein [31]. They were identified as pathogenic mutations according to the ACMG guidelines. Patients in all four families were siblings who exhibited the arRP inheritance pattern’. Since not all mutations were novel, some of them has been reported previously. The clinical data of the patients in the four families are listed in Table 2. All patients visiting the outpatient department showed a defect in the fundus with mild to moderate peripheral bone spicule-like pigments, gray retina color, and attenuation of retinal vessels (Figure 3). Patients in family RP008 and family RP028 had hearing defects, so we revisited and rediagnosed the patients in RP008 and RP028 as having Usher syndrome type II. The other two families, RP015 and RP033, did not have obvious hearing problems; they were diagnosed as having simple RP.

CYP4V2 accounted for 15% (3/20) of mutations detected in this research. All three families with CYP4V2 mutation (RP026, RP034, RP037) showed a compound heterozygous mutation pattern, and the patients in these three families were revisited and rediagnosed as having BCD. Five mutations identified in this research had been reported previously. In the three families, all six patients (five visited our outpatient department and one provided medical materials from a local hospital) showed highly reflective crystal deposits and profound RPE atrophy in the fundus photography (Figure 4). Five patients who could complete the visual field test all showed acentric visual field islands.

RP1, RHO, and PRPF31 each accounted for two (2/20, 10%) families in this research. TOPORS, EYS, CNGA1, CNGB1, and RPGR were all identified in only one (1/20, 5%) family. All clinical data for these patients are listed in Table 2, and fundus images are shown in Appendix 3. Novel mutations in these families are elaborated on below.

One novel mutation in PRPF31 was identified as pathogenic in family RP014—namely, c.967_ 968delGA(E323Dfs*151). This novel mutation was a small deletion mutation, which led to translation frameshift and protein truncation. This may cause abnormal posttranslation after 323 amino acids, potentially leading to the abnormal function of the C-terminal domain and affecting the normal localization of protein in cells [32]. The mutation was identified as pathogenic according to the analysis of the ACMG guidelines. Two patients in RP014 showed moderate visual defect, with slight pigments in the fundus (Figure 5).

One novel mutation in RP2 was identified as pathogenic in family RP010—namely, c.758_ 761delTAAT (p.l253fs*10). This was a small deletion mutation and led to translation frameshift and protein truncation. The C-terminal domain of the RP2 (RP2 activator of ARL3 GTPase) protein has weak homology with nucleoside diphosphate kinase (NDK). The mutation causing protein truncation has been reported to relate to a more severe phenotype [33]. Moreover, Jayasundera et al. reported that two different missense mutations at amino acid 253 lead to more severe phenotypes in RP2 mutations [34]. In family RP010, the fundus of proband III:2 showed slight pigments and myopia in both eyes (oculus dexter [OD]: −5D, oculus sinister [OS]: −3.5D; Figure 5). In contrast, II:3—the mother of the proband, who was a carrier of this mutation—had high myopia of −14.5D in her left eye. In addition, II:2—the uncle of the proband, who did not come to the outpatient department of our hospital for examination—was totally blind at the age of 40 years. The local data provided showed that there was no light perception in either eye, and leopard fundus and high myopia were present in both eyes.

One novel mutation in TOPORS was identified in a large four-generation autosomal dominant family, family RP012—namely, c.2323_2324delAG, p.S775*. This small deletion mutation led to a truncated protein of 775 amino acids, resulting in partial loss of the RS domain and loss of two proline, glutamic acid, serine, and threonine (PEST) domains in the TOPORS protein. The RS domain is a region rich in arginine and serine, which may affect pre-mRNA splicing, whereas PEST domains are five residues rich in PEST elements (proline, glutamic acid, serine, and threonine), which are usually the characteristics of fast degradation protein. Loss of these crucial domains may severely affect protein function [35]. This novel mutation was identified as pathogenic according to the ACMG guidelines. Eight patients tested in RP012 carried this heterozygous mutation; they all complained about night blindness from 6 to 17 years old accompanied by constricted visual field in adult age. The proband V:I who visited our outpatient department was a 22-year-old female. She complained about night blindness from 6 years old. At presentation, she had preserved a BCVA of 1.0 in both eyes but had a constricted visual field less than 24° (Figure 5).

A truncated mutation c.G2006A (p.W669*) in CNGB1 was first reported in this research. This mutation was located in exon 10 (amino acids 661–838), which damages all key domains in CNGB1 protein, including the N-terminal glutamate rich domain (encoded by exons 1 to 16), transmembrane and pore domain (encoded by exons 21 to 26), cyclic nucleotide-binding domain (encoded by exons 29 to 31), and carboxyl terminal channel-like domain [36]. This mutation may also trigger nonsense-mediated decay and affect the normal function of protein. This novel mutation was identified as pathogenic according to the ACMG guidelines. Two patients in family RP035 who carried this heterozygous mutation were siblings. Proband II:5 was a 64-year-old male, and his younger brother II:3 was 54 years old. They complained about night blindness before 5 years old and visual loss from the age of 25 (II:5) to their 40s (II:3). Dense pigments and profound RPE atrophy were found in the fundus of proband II:5 (Figure 5).

One novel mutation of RPGR (c.T773C, p.L258P), detected in the four-generation family RP036, was considered likely pathogenic. This missense mutation was located on exon 2 and was predicted to be damaging by PolyPhen, SIFT, and Provean. It has not been reported in the ExAC, 1000 Genomes, or Annoroad Healthy person mutation databases. According to ACMG guidelines, this novel mutation was identified as pathogenic. The proband III:5 and his three female cousins experienced night blindness from early childhood and visual defects from their 40s, whereas several female members of this family complained about high myopia over −10.00D. The female family member IV:2 had a high myopia of −14.00D in the right eye and −18.00D in the left eye. She had also complained about night blindness from early childhood, and her ERG examination showed diminished rod and cone responses. The fundus appearance of two patients showed dense pigments, attenuated retinal vessels, and a pale optic nerve head; in contrast, the fundus of the carrier female showed no pigments and a leopard fundus (Figure 5).

In this research, several important findings were reported, including the following: 1) 25 putative pathogenic mutations of 12 genes were detected by WES, and they were all confirmed by Sanger sequencing in 20 (20/28, 71.4%) families, including 12 genes with USH2A and CYP4V2 as the most frequent mutated genes; 2) 7 novel mutations were identified, including USH2A, PRPF31, RP2, TOPORS, CNGB1, and RPGR; 3) the phenotype–genotype relationship in the 12 RP genes were analyzed which revealed later disease onset and more severe visual function defects in CYP4V2; and 4) late molecular diagnosis with long disease duration and poor preserved BCVA were found in Chinese RP patients.

Twelve genes were identified as putative pathogenic genes in this group of RP families, with USH2A, CYP4V2, RHO, PRPF31, and RP1 as the most frequent genes. Several studies of the mutation spectrum in Chinese RP patients were reported previously, which were found to have some differences from our research (Table 3). USH2A (4/20, 20%) was detected to most frequently harbor the mutations in this research, which is consistent with the previously reported 12%–25% proportion worldwide [1,5]. CYP4V2 (3/20, 15%) was detected to be the second most frequent mutation gene in this research; CYP4V2 encodes a member of the cytochrome P450 heme-thiolate protein superfamily, which is involved in oxidizing various substrates in the metabolic pathway. Mutations in this gene result in corneoretinal BCD [37]. This gene has been reported to account for 3% of RP patients in Caucasians [38]; it appears to be more common in East Asian countries, such as China [3] and Japan [39]. Recently, Gao et al. [40] reported a CYP4V2 frequency of 15% in a large RD cohort comprising 1,243 patients, which indicated a large group of BCD patients in China. The differences in the most frequent mutation genes between this research and previous reports may come from study scales and different inclusion criteria because some studies may exclude BCD from RP. In addition, since BCD can be easily diagnosed from a unique fundus appearance, some clinicians may use Sanger sequencing as the detecting technology.

The diagnosis rate of WES sequencing in RDs varied greatly in previous studies because of the sequencing platform selection, inheritance pattern, and proband selection [11,12]. It has been reported that WES can achieve a diagnosis rate of 41%–55% [41-44] in large RP cohorts, and a higher diagnosis rate can be achieved in larger pedigrees. Panel-based NGS can promote a diagnosis rate of 70%–80% [40,45] in RDs by carefully designing the selected genes in the panel. When compared to WES, panel-based NGS was not applicable for small research groups because of the high cost of the panel design procedure. In this research, we achieved a diagnosis rate of 71.4%. There were three factors responsible for the relatively high diagnosis rate: First, probands recruited in this research all had a clear family history and clinical diagnosis. Second, at least one patient and one healthy relative were sent for WES sequencing. Third, mutations passed through the filtering procedure were searched from among 662 gene defects that can cause vision-associated phenotypes (including 89 RP genes in the RetNet Database). With the development of NGS, the cost of WES in each patient can be relatively low, making WES a more competitive approach for molecular diagnosis in RDs.

Seven novel mutations were detected in this study. All mutations were found to be cosegregated with phenotype, and they were confirmed by Sanger sequencing. Among the seven novel mutations, six were mutations causing protein truncation, which revealed that truncated mutations were still more common in RP molecular diagnosis.

Phenotype–genotype relationships were detected in this research. For the two most frequent genes in this research, USH2A was found to have a milder phenotype compared with CYP4V2, with longer disease duration and moderate visual defect. Visual field tests in CYP4V2 patients were also found to have a unique pattern, with preserved acentric visual field islands; this indicated that different strategies should be adopted in treating BCD from other sub types of RP.

Putative genes in eight families were not identified in this research. For RP031, PRPF8 (c.C3543G, p.D1181E) has been identified as a putative gene; because the mutation and disease did not cosegregate, we excluded PRPF8 as the putative mutation in RP031. There are several possible reasons that these mutations could not be found by WES, which are as follows [12]: 1) the mutations were larger deletions or rearrangements that are not detectable by WES; 2) the mutations were in deeper intronic mutations that cannot be detected by WES; and 3) the mutations were in genes that had not been reported to be associated with RP.

In conclusion, 25 putative pathogenic mutations of 12 genes were detected by WES and were all confirmed by Sanger sequencing in 20 (20/28, 71.4%) families, including 7 novel mutations. USH2A and CYP4V2 were found to be the most frequent genes in this research. The mutation spectrum of RP in Chinese was expanded in this research, which may benefit future cutting-edge therapies.