Clinical and genetic characteristics of 14 patients from 13 Japanese families with RPGR-associated retinal disorder: report of eight novel variants

Variants in the retinitis pigmentosa GTPase regulator (RPGR) gene are a major cause of X-linked inherited retinal disorder (IRD). We herein describe the clinical and genetic features of 14 patients from 13 Japanese families harboring RPGR variants in a nationwide cohort. Comprehensive ophthalmological examinations were performed to classify the patients into one of the phenotype subgroups: retinitis pigmentosa (RP) and cone rod dystrophy (CORD). The mean age of onset/at examination was 13.8/38.1 years (range, 0–50/11–72), respectively. The mean visual acuity in the right/left eye was 0.43/0.43 (range, 0.1–1.7/−0.08–1.52) LogMAR unit. Eight patients had RP, and six had CORD. Whole-exome sequencing with target analyses identified 13 RPGR variants in 730 families with IRD, including 8 novel variants. An association between the phenotype subgroup and the position of variants (cutoff of amino acid 950) was revealed. To conclude, the clinical and genetic spectrum of RPGR-associated retinal disorder was first illustrated in a Japanese population, with a high proportion of novel variants. These results suggest the distinct genetic background of RPGR in the Japanese population, in which the genotype–phenotype association was affirmed. This evidence should be helpful monitoring and counseling patients and in selecting patients for future therapeutic trials.


Introduction
Inherited retinal disorder (IRD) is a major cause of blindness both in children and working populations in developed countries 1 . Retinitis pigmentosa (RP) is one of the most prevalent IRDs, and RP represents a heterogeneous group of retinal diseases characterized by progressive bilateral degeneration of rod and cone photoreceptors [1][2][3][4][5][6][7][8] . The estimated prevalence of RP in European populations is~1 in 3000-4000 individuals 2-6,9 . Different patterns of inheritance have been identified in RP and allied disorders, including autosomal dominant (AD), autosomal recessive (AR), X-linked (XL), and mitochondrial inheritance 2,3,10 .
Pathogenic variants in the RPGR gene (RP3) were first identified as a cause of XLRP in 1996 12,13 . RPGR contains 19 exons and encodes a 90-kDa protein product localized predominantly to the photoreceptor connecting cilium (CC) 12,14 . The RPGR protein contains a repeat structure highly similar to the regulator of chromosome condensation 1 (RCC1) at the N-terminus 11,12,15 . RCC1 plays a crucial role in nucleocytoplasmic transport and regulation of cell-division processing 16,17 . Later, a novel 3′ terminal exon (well-known as exon open-reading frame 15; ORF15) was identified, which includes a large 3′ terminal exon consisting of exon 15 and extending into part of intron 15 18 . Biochemical investigations revealed that RPGR-ORF15 is located in the CC, which binds to the axoneme and the basal body 19,20 . The RPGR protein plays an important role in the transportation of phototransduction components and other outer segment proteins across the CC, although the function of RPGR is not perfectly understood 3 .
XLRP caused by pathogenic RPGR variants is one of the most severe forms of RP, with early onset of disease, night blindness, myopia, severe generalized rod and cone dysfunction, and progression to legal blindness by the third or fourth decade 3,21 . Carrier females are mostly asymptomatic or mildly affected with characteristic fundus features and electrophysiological abnormalities 22 , although the severity of carriers varies.
XLCORD caused by pathogenic RPGR variants affects males with various onsets ranging from the second to the fourth decade, myopia, generalized cone rod dysfunction (occasionally with rod dysfunction), and diverse rates of progression 3 . Carrier females are mostly asymptomatic or mildly affected, with varying severity 28 .
The purpose of this study was to characterize the clinical and genetic features of patients and carriers with RPGR-RD in a large nationwide Japanese cohort by clarifying a genotype-phenotype association.

Participants
The protocol of this study adhered to the tenets of the Declaration of Helsinki, which was approved by the local ethics committee of the participating institutions from Japan (Reference: R18-029). Signed informed consent was obtained from all participants after explanation of the nature and possible consequences of this study.
Patients with a clinical diagnosis of IRD and available genetic data of whole-exome sequencing (WES) were investigated between 2008 and 2018 in the Japan Eye Genetics Consortium (JEGC; http://www.jegc.org/) study 42 . A total of 1294 subjects from 730 Japanese families registered to the JEGC database were surveyed.

Clinical investigations
Detailed demographic information was obtained, including ethnicity, sex, medical and family history, chief complaints of visual symptoms, and onset of disease. Comprehensive ophthalmological examinations were performed, including measurement of refractive errors, best corrected decimal visual acuity (BCVA) converted to the logarithm of the minimum angle of resolution (Log-MAR), fundus photography, fundus autofluorescence (FAF) imaging, spectral-domain optical coherence tomography (SD-OCT), visual field testing, and electrophysiological assessment according to the international standards of the International Society for Clinical Electrophysiology of Vision (ISCEV) 43,44 .

Phenotype subgroup
For the purpose of this study, phenotype subgroups were defined based on clinical manifestation according to the previous report; 45 RP (including rod-cone dystrophy), a progressive retinal dystrophy initially often presenting peripheral atrophy with generalized rod dysfunction greater than cone dysfunction; CORD, a progressive retinal dystrophy initially often presenting with macular atrophy with generalized cone dysfunction greater than rod dysfunction.

RPGR variant detection
Genomic DNA was extracted from all affected subjects and unaffected family members (where available for cosegregation analysis). WES with target analysis of retinal disease-associated genes (RetNet; https://sph.uth.edu/ retnet/home.htm) was performed according to previously published methods 42 . The called variants were filtered by the allele frequency in the general Japanese population (<1%) as listed in the Human Genetic Variation Database (HGVD; http://www.genome.med.kyoto-u.ac.jp/SnpDB/ about.htm). Depth and coverage for the target areas were interrogated using the Integrative Genomics Viewer (http://www.broadinstitute.org/igv/). For the purpose of this study, long-read direct sequencing was performed in seven patients with XLRP who were negative for two major XLRP-associated genes (RP2, RPGR) by WES at the National Genetic Reference Laboratory in Manchester, UK, for further screening of RPGR-ORF15 according to the previously published method 46 . Together with clinical features and pattern of inheritance, disease-causing variants were determined from the detected variants of the retinal disease-associated genes.

Participants
Fourteen affected subjects from 13 Japanese families with a clinical diagnosis of IRD and harboring RPGR variants were ascertained. All 14 affected subjects were registered as a proband (or probands) for each pedigree. Seven females from six families were also registered as carriers.
The detailed clinical information of 14 affected subjects (registered as a proband) is presented in Table 1. Pedigrees of 13 families are shown in Fig. 1. All 14 subjects and 7 carriers were originally from Japan, and no mixture of ethnicity was reported.
Onset, chief complaint, refraction, and visual acuity The mean age of onset of ten affected males with available records was 14.3 years (range, 0-50). One affected female with available records had onset of disease at the age of 9 (11-III:1).
The mean refractive error of the right/left eye of ten affected males with available records who had no refractive complication was

Retinal images, visual field, and electrophysiological findings
Fundus photographs were available in 12 affected males, and FAF images were obtained in 6 affected males. Representative fundus and FAF images of 12 affected males are presented in Fig. 2.
Central atrophy or parafoveal atrophy was identified in all 12 patients with available fundus photographs. Tigroid changes (seen in high myopic retina) were observed in six patients (6/12, 50%), and bone-spicule pigmentation (found in peripheral retinal atrophy) was detected in five patients (5/12, 41.7%). Well-marked atrophic changes were demonstrated in FAF images. A ring of high AF density was noted in all six patients with available FAF images. SD-OCT images were available in 11 affected males. Representative SD-OCT images are shown in Fig. 3. Structural disruption in the photoreceptor layers was observed in all 11 patients. Relatively preserved photoreceptor layers at the fovea were detected in five patients (5/11, 45.5%; 4-IV:1, 7-II:1, 10-III:1, 12-III:1, and 13-III:3).
Electrophysiological assessment was performed in ten affected males and two affected females. Undetectable responses in both generalized rod and cone systems were observed in five patients (5/12, 41.7%). Severely decreased generalized cone and severely decreased rod responses were demonstrated in one patient (1/12, 8.3%). Severely decreased generalized rod responses and moderately decreased generalized cone responses were identified in one patient (1/12, 8.3%). Severely decreased generalized cone and mildly decreased generalized rod responses were found in three patients (3/12, 25.0%). Moderately decreased generalized cone and generalized rod responses were noted in two patients (2/12, 16.7%).

Phenotype subgroups
Phenotype subgroup classification was performed in all 12 affected males and 2 affected females. There were six males with CORD (6/12, 50%) and six with RP (6/12, 50%). Both affected females were classified into RP.

Clinical information of carrier females
Clinical information of seven carrier females from six families was obtained. The detailed findings are described in Supplementary Table 1. Representative fundus and FAF images are presented in Supplementary Fig. 1. Representative SD-OCT images are shown in Supplementary  Fig. 2.
Abnormalities on fundus photographs, FAF, SD-OCT, visual field testing, and electrophysiological assessment were found in five out of seven carriers (5/7, 71.4%). One carrier had reduced visual acuity, and one had night blindness. Retinal atrophy was observed in two carriers, and tapetal reflexes were found in five carriers (5/7, 71.4%). Electrophysiological assessment was available in

RPGR variants
The variant data of 18 affected, 9 carriers, and 14 unaffected subjects of 13 families are summarized in Table 2.

In silico molecular genetic analysis
The detailed results of in silico molecular genetic analyses for the 13 detected RPGR variants are presented in Supplementary Table 2.
Seven frameshift variants were located in ORF15. Six variants, including three nonsense, two missense, and one frameshift variant, were in exons 5, 7, 8, 10, and 14, and all six variants except for one frameshift variant (p. Gly876ArgfsTer203) were located in the RCC1-like domain. The allele frequency for one variant (p. Glu1060ArgfsTer18) in the general population was 0.001134% in the GnomAD database. None of the 13 detected RPGR variants were found in the Japanese general population of the HGVD and iJGVD databases.
General prediction, functional prediction, and conservation were assessed for the 13 variants, and pathogenicity classification according to the ACMG guidelines was pathogenic for eight variants, including six frameshift (p.Glu746ArgfsTer23, p.Glu1031Glyf-sTer58, p.Glu1035GlyfsTer43, p.Glu1060ArgfsTer18, p.Tyr1103SerfsTer7, and p.Pro1134HisfsTer18) and two nonsense variants (p.Glu210Ter and p.Gln565Ter), likely pathogenic for three variants including two frameshift (p.Phe130SerfsTer4 and p.Gly876ArgfsTer203) and one nonsense variant (p. Gln227Ter), and uncertain significance for two missense variants (p.Gly357Asp and p.Thr278Ala).
Two missense variants (p. Gly357Asp and p. Thr278Ala) with uncertain significance were found in two probable XL families (Families 9, 10), and no other candidate variants associated with RP were detected in either of these families.
Overall, 13 disease-causing variants in the RPGR gene were ascertained in eight families with RP, and five families with CORD. Together with the clinical features of affected subjects and the model of inheritance in the

Genotype-phenotype association
The locations of variants for two phenotype subgroups were investigated. All variants in eight families with RP were located in exons 1-14 and the 5′ end of ORF15 (< amino acid 950). All variants in five families with CORD were located at the 3′ end of ORF15 (> amino acid 950). A significant genotype-phenotype association between the phenotype subgroup and the position of detected variants was revealed.

Discussion
The clinical and genetic characteristics of RPGR-RD were illustrated in a nationwide cohort of 18 affected and 14 unaffected individuals and 9 carriers from 13 Japanese families with RPGR-RD, detecting 13 variants including 8 novel variants. There were eight families with RP and five families with CORD, which was associated with the position of RPGR variants.
To the best of our knowledge, this study reports the largest cohort of RPGR-RD in the Asian population. RPGR-RD accounts for 66.7% of molecularly confirmed XLRP (12 families; RPGR-8 families, RP2-4 families) in   50,51 was not included as CORD in the JEGC cohort. Out of 13 detected RPGR variants detected in this study, five variants were located within the RCC1-like domain (5/13, 38.5%), and seven were within the ORF15 domain (7/13, 53.8%). The proportion of ORF15 variants in the present cohort was slightly lower than that of the North American population (66%, reported by Sharon et al.) 27 .
In this study, five previously reported RPGR variants were identified in five families (Families 2, 3, 5 -CORD; Families 6, 7 -RP). Three of the five previously reported variants in the present cohort were reported in European cases with RP (p. Glu746ArgfsTer23, p.Gly876Argf-sTer203, and p.Glu1031GlyfsTer58) 18,26,33 . The other two variants were reported in European cases with CORD (p. Glu1060ArgfsTer18 and p.Tyr1103SerfsTer7) 29,32 . Four of the five families (4/5, 80%) in the present cohort showed a concordant phenotype with previous reports (Families 2, 3, 6, 7). In that study, the clinical effect of these four variants was confirmed in the Japanese population. One family with CORD had a discordant phenotype with a previous report (Family 5, p. Glu1031GlyfsTer58). Given the severely affected retinal findings in two affected males in Family 5, an advanced stage of this phenotype could be described as "RP".
Eight RPGR variants were reported first in this study, including three frameshift (p.Phe130SerfsTer4, p. Glu1035GlyfsTer43, and p.Pro1134HisfsTer18), three nonsense (p.Glu210Ter, p.Gln227Ter, and p.Gln565Ter), and two missense variants (p.Thr278Ala and p. Gly357Asp). A high proportion of novel variants (8/13, 61.5%) was revealed in the Japanese cohort, which suggests a distinct genetic background for RPGR in the Japanese population compared with the European population.
In silico analysis for eight novel variants predicted pathogenic effects in four variants (p. Glu210Ter, p. Gln565Ter, p.Glu1035GlyfsTer43, and p.Pro1134Hisf-sTer18), a likely pathogenic effect in two variants (p. Gln227Ter and p.Phe130SerfsTer4), and variants of uncertain significance in two variants (p.Thr278Ala and p.Gly357Asp). These two missense variants (p. Gly357Asp and p.Thr278Ala) were located within the RCC1-like domain, where other missense diseasecausing RPGR variants are frequently found 37 . The pathogenicity of these two missense variants is uncertain, although some association with nucleocytoplasmic transport and regulation of cell-division processing may be anticipated 16,17 .
A significant association between genotype and phenotype was revealed in this study. Variants located in exons 1-14 and the 5′ end of ORF15 caused RP, and variants at the 3′ end of ORF15 caused CORD. This finding was consistent with previous reports in the European population 3,29,30 . This fact supports the prediction of the natural history of RPGR-RD in counseling patients. Notably, the mechanism underlying this genotype-phenotype association has not been clarified.
There are limitations to this study. The selection bias related to the disease severity should be inherent, since it is uncommon for genetically affected subjects with good vision to visit clinics/hospitals. This study is a crosssectional retrospective study; thus, longitudinal natural history studies in a larger cohort could provide more accurate information for disease progression of RPGR-RD. In addition, the molecular mechanisms of disease causation for some novel and previously reported variants are not yet known; therefore, further functional analysis could determine the disease causation of each variant.
In conclusion, the phenotypic and genotypic features of RPGR-RD were documented first in a large cohort of Japanese populations. A broad spectrum of phenotypic and genotypic findings was determined, revealing a considerable genotype-phenotype association. This evidence should be helpful in monitoring and counseling patients and in selecting patients for future therapeutic trials such as gene replacement therapy.