Genotype–Phenotype Analysis of RPGR Variations: Reporting of 62 Chinese Families and a Literature Review

Purpose RPGR is the most common cause of X-linked retinitis pigmentosa (RP), of which female carriers are also frequently affected. The aim of the current study was to explore the RPGR variation spectrum and associated phenotype based on the data from our lab and previous studies. Methods Variants in RPGR were selected from exome sequencing data of 7,092 probands with different eye conditions. The probands and their available family members underwent comprehensive ocular examinations. Similar data were collected from previous reports through searches in PubMed, Web of Science, and Google Scholar. Systematic analyses of genotypes, phenotypes and their correlations were performed. Results A total of 46 likely pathogenic variants, including nine missense and one in-frame variants in RCC1-like domain and 36 truncation variants, in RPGR were detected in 62 unrelated families in our in-house cohort. In addition, a total of 585 variants, including 491 (83.9%) truncation variants, were identified from the literature. Systematic analysis of variants from our in-house dataset, literature, and gnomAD suggested that most of the pathogenic variants of RPGR were truncation variants while pathogenic missense and in-frame variants were enriched in the RCC1-like domain. Phenotypic variations were present between males and female carriers, including more severe refractive error but better best corrected visual acuity (BCVA) in female carriers than those in males. The male patients showed a significant reduction of BCVA with increase of age and males with exon1-14 variants presented a better BCVA than those with ORF15 variants. For female carriers, the BCVA also showed significant reduction with increase of age, but BCVA in females with exon1-14 variants was not significant difference compared with those with ORF15 variants. Conclusion Most pathogenic variants of RPGR are truncations. Missense and in-frame variants located outside of the RCC1-like domain might be benign and the pathogenicity criteria for these variants should be considered with greater caution. The BCVA and refractive error are different between males and female carriers. Increase of age and location of variants in ORF15 contribute to the reduction of BCVA in males. These results are valuable for understanding genotypes and phenotypes of RPGR.

Purpose: RPGR is the most common cause of X-linked retinitis pigmentosa (RP), of which female carriers are also frequently affected. The aim of the current study was to explore the RPGR variation spectrum and associated phenotype based on the data from our lab and previous studies.
Methods: Variants in RPGR were selected from exome sequencing data of 7,092 probands with different eye conditions. The probands and their available family members underwent comprehensive ocular examinations. Similar data were collected from previous reports through searches in PubMed, Web of Science, and Google Scholar. Systematic analyses of genotypes, phenotypes and their correlations were performed.
Results: A total of 46 likely pathogenic variants, including nine missense and one inframe variants in RCC1-like domain and 36 truncation variants, in RPGR were detected in 62 unrelated families in our in-house cohort. In addition, a total of 585 variants, including 491 (83.9%) truncation variants, were identified from the literature. Systematic analysis of variants from our in-house dataset, literature, and gnomAD suggested that most of the pathogenic variants of RPGR were truncation variants while pathogenic missense and in-frame variants were enriched in the RCC1-like domain. Phenotypic variations were present between males and female carriers, including more severe refractive error but better best corrected visual acuity (BCVA) in female carriers than those in males. The male patients showed a significant reduction of BCVA with increase of age and males with exon1-14 variants presented a better BCVA than those with ORF15 variants. For female carriers, the BCVA also showed significant reduction with increase of age, but BCVA in females with exon1-14 variants was not significant difference compared with those with ORF15 variants.
Conclusion: Most pathogenic variants of RPGR are truncations. Missense and in-frame variants located outside of the RCC1-like domain might be benign and the pathogenicity

INTRODUCTION
Retinitis pigmentosa (RP) is a common type of inherited retinal degenerations (IRD) characterized by impaired dark adaptation and night blindness, progressive visual field defects and pigmentary retinopathy, affecting approximately one in 3,500-4,000 people worldwide (Berger et al., 2010;Traboulsi, 2010;Sundaram et al., 2012;Zhang, 2016). RP can be inherited as an autosomal dominant, autosomal recessive, or X-linked trait, with these categories accounting for approximately 30-40%, 50-60%, and 5-15% of RP patients, respectively (Bunker et al., 1984;Grondahl, 1987;Hartong et al., 2006). X-linked RP is one of the most severe forms of human retinal degeneration (Bird, 1975). Affected males usually suffer nyctalopia and severe and rapid progressive loss of peripheral vision with an early onset, followed by progressive central visual loss during the second to fourth decades of life, while female carriers may present a wide range of phenotypes, ranging from asymptomatic to severe phenotype (Bird, 1975;Fishman et al., 1986;Banin et al., 2007). Additionally, the phenotype of X-linked RP generally shows great phenotypic heterogeneity, including interfamily heterogeneity, in terms of the age of onset, clinical severity, rate of progression, and prevailing damage to rods and cones (Fahim et al., 2011). Variants in retinitis pigmentosa GTPase regulator (RPGR, OMIM 312610) account for 70-80% (Sharon et al., 2003;Pelletier et al., 2007;Shu et al., 2007) of X-linked RP cases. This protein localizes to the connecting cilium in photoreceptors and is thought to play a role in protein transport (Roepman et al., 2000;Hong et al., 2003).
In 2007, a study provided an overview of RPGR genotypes and the associated phenotypic variation (Shu et al., 2007). However, the widespread application of next-generation sequencing (NGS) in recent years has increased the number of variants identified in RPGR and expanded the known phenotypic spectrum of patients. Further comprehensive analysis of RPGR genotype-phenotype relationships would be expected. In addition, most of the patients previously reported to show variants in RPGR were recruited from America or Europe.
In this study, we performed a summary of the genotypes and corresponding phenotypes in RPGR from our database and the literature. The pathogenicity of the variants in RPGR and genotype-phenotype correlations were further assessed and summarized.

Samples
In an ongoing study of genetic eye diseases, we recruited 7092 probands with different eye conditions from the pediatric and Genetic Eye Clinic of the Zhongshan Ophthalmic Center, and we collected the available clinical data of the probands and their available family members with RPGR variations. This study was performed in accordance with the Declaration of Helsinki, and written informed consent was obtained from participating individuals or their guardians. Our study was approved by the Institutional Review Board of Zhong Shan Ophthalmic Center. All patients included in the study underwent exome sequencing [whole-exome sequencing (WES) and targeted exome sequencing (TES)]. The rare variants were defined as variants with a minor allelic frequency of less than 0.01 in general population from gnomAD database and patients with likely pathogenic variants were subsequently discriminated from the rare variants of RPGR in this study. Genotype-phenotype correlation was investigated by statistical analyses on different groups of patients with likely pathogenic variants of RPGR according to the variants in certain regions. In addition, patients with rare variants in RPGR were summarized based on our data (Tables 1, 2).

Exome Sequencing
Exome sequencing, including WES and TES, was conducted in the patients included in our study. Whole-exome sequencing was performed on 5,307 probands using a commercial service as described in our previous study . Genomic DNA from the probands was sheared and fragments of an approximate 150 bp were selected. Exome was captured by an Agilent SureSelect Human All Exon Enrichment Kit (Agilent, Santa Clara, CA, United States). Library quality was assessed using an Agilent 2100 Bioanalyzer and were then sequenced on the Illumina HiSeq platform (Illumina, San Diego, CA, United States) with an average depth of at least 125-fold. After filtering out low quality reads, and remaining clean data was verified by aligning the sequencing with the UCSC hg19 1 CORD, cone-rod dystrophy; DM, disease-causing mutations; HM, high myopia; MD, macular degeneration; NA, not available; NSC, no splicing change; RP, retinitis pigmentosa; SA, splicing acceptor; SD, splicing donor; *, termination codon; /, not applicable. BDGP, Berkeley Drosophila Genome Project; HGMD, the Human Gene Mutation Database; HSF, Human Splicing Finder; ()¸: Previously reported by our lab. None of the variants were recorded in gnomAD except c.3027_3028delGG. Evidence that variant is likely pathogenic: 1 = segregate with inherited eye diseases in one or more families (males with variants were affected); 2 = variants identified in one or more families with eye disease accompany with RP, CORD, COD, MD, or HM; 3 = at least three of four predicting tools are pathogenic; 4 = MAF ≤ 4.7 × 10 −5 or absence in gnomAD database; 5 = other known IRD pathogenic variants were not identified; 6 = variants does not find in controls; 7 = variants are de novel.
Frontiers in Genetics | www.frontiersin.org Evidence that variant is benign or likely benign: 1 = does not segregate with inherited eye diseases in one or more families (males with variants were unaffected); 2 = variant identified in one or more families with eye disease other than RP, CORD, COD, MD, or HM; 3 = at least two of four predicting tools are benign; 4 = MAF ≥ 4.7 × 10 −5 in gnomAD database; 5 = identified other known IRD pathogenic variants; 6 = verified in controls.
All variants are not recorded in HGMD except the variant c.1630A>G.
Frontiers in Genetics | www.frontiersin.org reference using the Burrows-Wheeler Aligner (BWA 2 ). Variants were detected by SAMTOOLS 3 and were annotated and predicted by SnpEff 4 , ANNOVAR 5 , and dbNSFP 6 , respectively. Targeted-exome sequencing was conducted on 1,785 probands by our lab as described in our previous study . Approximately 200 bp fragments were obtained from genomic DNA using a Bioruptor Plus (Diagenode, Liege, Belgium). A paired-end library was prepared using a KAPA HTP Library Preparation kit (Roche, Basel, Switzerland). Targeted exome was captured using a designed NimbleGen SeqCap EZ Prime Choice kit (Roche, Basel, Switzerland). Library quality was assessed using an Agilent 2100 Bioanalyzer and were then sequenced on an Illumina Nextseq550 Analyzer using the Illumina NextSeq550 Mild output v2 kit (150 PE) (Illumina, San Diego, CA, United States) with an average depth of 250-fold. Variant calling and annotation were analyzed using the StrandNGS software (Karnataka, India) according to the manufacturer's instructions. The UCSC Genome Browser on Human hg19 Assembly was used as an alignment reference. The dbNSFP was used for predictions of missense variants. The list of 126 target genes, including RPGR, in TES has been described in our previous study . Variants in RPGR identified through WES and TES were selected and filtered via multistep bioinformatics analyses as previously reported (Xu et al., 2014;Li et al., 2015;Sun et al., 2015;Zhou L. et al., 2018). Additionally, we used CADD 7 and REVEL 8 to further predict the severity of the missense variants in RPGR. Data from the Genome Aggregation Database (gnomAD 9 ) and Human Genome Mutation Database (HGMD 10 ) were included as references for evaluating the pathogenicity of the variants in RPGR. Selected remaining variants were verified by Sanger sequencing. The pedigrees and sequence diagrams of potential likely pathogenic variants are shown in Supplementary  Figures 1, 2.

Phenotype Analysis in Our Lab
Probands and available family members with variants in RPGR were recruited for further comprehensive ocular examinations. All of the examinations were performed by the same experienced team of ophthalmologists. A detailed family and ophthalmic history were obtained. The comprehensive ocular examinations included best corrected visual acuity (BCVA), refractive error (RE), and spectral domain-optical coherence tomography (SD-OCT).
Refractive error was measured using an autorefractometer (Topcon KR-8000, Paramus, NJ, United States) after mydriasis with tropicamide (Mydrin-P, Santen Pharmaceutical, Japan). An optical biometer (IOL master V5.0, Carl Zeiss Meditec AG, Germany) was used to detect the ocular biometric axial length. Full-field electroretinogram (ERG) responses were recorded in patients in accordance with the standards of the International Society for Clinical Electrophysiology of Vision for evaluating retinal disorders, using an Espion ERG system (Diagnosys LLC, United States). Optical coherence tomography of the macular and optic disks was performed via SD-OCT (Optovue, Inc., United States).
Clinical data were collected to perform further comparisons between genders, ages, locations and variation types. Spherical equivalent refraction (SER) was calculated by adding spherical refraction to half the cylindrical refraction.

Statistical Analysis
Analyses were performed using R software and SPSS version 23. Logistic regression was used to screen out the factors influencing BCVA in males and females. Median (IQR, interquartile range) were used for continuous parameters. Mann-Whitney U test was used to compare the BCVA and refractive error among groups, namely (1) patients with variants in exon1-14; (2) patients with variants in ORF15; (3) patients with variants in RCC1like domain; (4) patients with missense and in-frame variants; (5) patients with truncation variants. The corrected significant P-value for this study should be less than 0.017 (α = 0.05/3) according to the Bonferroni method.

Identification of RPGR Variants in 7,092 Probands With Different Eye Conditions in Our Lab
A total of 121 variants, including 15 polymorphisms, eight 3 UTR variants, one synonymous variant and 97 rare variants, were detected in 7,092 probands. Of the 97 rare variants, 46 FIGURE 2 | Comparison of phenotypes according to different factors. (A) Comparison of logMAR BCVA between males and females. The BCVA of female carriers was better than that of male patients. (B) Comparison of refractive error (RE) between males and females. Spherical equivalent refraction represents the severity of RE. The RE of female carriers was more serious than that of males. (C) Scatterplots of logMAR BCVA and age, the two fitted lines correspond to male (black) and female (red) patients. A significant reduction of BCVA with increase of age in both males and females. (D) Scatterplots of RE and age, the two fitted lines correspond to male (black) and female (red) patients. The trends of the two lines are basically smooth. BCVA, best corrected visual acuity. ***, P value less than 0.001. likely pathogenic variants (11 novels) and 51 likely benign variants were identified. Among the 46 likely pathogenic variants, nine missense variants, one in-frame variant and 17 truncation variants were located in exon1-14, and the remaining 19 truncation variants were located in ORF15 ( Table 1). The 46 likely pathogenic variants were identified in 62 families, of which truncation variants were identified in 52 (83.9%, 52/62), while missense and in-frame variants were identified in nine (14.5%, 9/62) and one (1.6%, 1/62) family, respectively. Of the other 51 likely benign variants, 21 missense variants and one in-frame variant, were identified in exon1-14, while 16 missense and 13 in-frame variants were detected in ORF15 ( Table 2).

Review of RPGR Genotypes From Our Lab and Previous Literature
A total of 585 variants have been reported in previous literature, including 491 truncations, 84 missenses, and 10 in-frame variants. Of the 94 missense and in-frame variants, 81 were located in the RCC1-like domain, while the remaining 13 were located outside the domain (Supplementary Table 1). A total of 585 previously reported variants, combining 46 likely pathogenic variants with our laboratory data, a total of 606 variants were analyzed (25 variants were repetitive).

Pathogenicity Evaluation of Missense and In-Frame Variants Located Outside of the RCC1
A total of 57 missense and in-frame variants were located outside of the RCC1 region, including 45 variants from our in-house cohort and 13 from literature were identified (one variant was repetitive) ( Table 2 and Supplementary Table 1).
The following lines of evidence suggested that these variants in RPGR might not be disease causing. (1) Missense and inframe variants were significantly enriched outside of the RCC1 region according to the gnomAD database, and the frequency was obviously high (Figure 1). (2) Most of these variants were FIGURE 3 | (A) The logMAR BCVA of male patients with variants in exon1-14 and ORF15 showed that patients with variants in exon1-14 have a better visual acuity. (B) Patients with variants in RCC1-like domain were no significant difference compared to those in ORF15. (C) Comparison of logMAR BCVA between M + I and T, there was no significant difference in variation type. (D) ROC curves suggested that our model shows high sensitivity and specificity in distinguishing different degrees of BCVA. The datasets used for AUC analysis were from available males' data and were randomly divided into two independent datasets (training and test datasets) by the R-software. BCVA, best corrected visual acuity; E1-14, exon1-exon14; RCC1, RCC1-like domain; M + I, missense and in-frame; T, truncation. ns, no statistical significance; **, P value less than 0.01. identified in one or more probands with different eye conditions other than RP or closely relative early onset high myopia (HM), cone-rod dystrophy (CORD), cone-dystrophy (COD), or macular degeneration (MD) ( Table 2). (3) All but two missense variants (c.37G > A and c.1519A > G) located outside of the RCC1 were predicted to be benign by at least two of four prediction tools (90% cutoff score: 0.29 in REVEL and 21.5 in CADD) ( Table 2). (4) A few patients showed variants in other known IRD genes, and some variants were verified in unaffected controls. (5) Segregation analysis contributed further evidence that missense and in-frame variants in non-RCC1 regions are not disease causing, and the corresponding pedigrees are shown in Supplementary Figure 3. (6) A previous study reported frequent in-frame deletions of 3-36 bp in healthy controls, suggesting that in-frame variants are benign (Karra et al., 2006). In addition, Zhang Q. et al. (2019) developed an in vitro assay illustrating that some variations located outside of the RCC1 region might be non-disease-causing polymorphisms.

BCVA in Patients With RPGR Variations
The clinical data of the probands and available families with pathogenic variants from our database and previous studies are summarized in Supplementary Tables 2, 3. The statistical results table were shown in Supplementary Table 4. BCVA showed a significant reduction with increase of age in both males and females (r = 0.479 and r = 0.216, respectively) ( Figure 2C). Better BCVA in female carriers (0.10 [0.00, 0.30] logMAR) than in male patients (0.40 [0.17, 0.60] logMAR) (P = 7.41E-25) (Figure 2A). Logistic regression was used to screen out the factors influencing BCVA in males and females, and the receiver operating characteristic (ROC) curves suggested that our model showed high sensitivity and specificity in distinguishing the different degrees of BCVA ( Figure 3D). For males, the variation type was not associated with BCVA (P = 0.183) ( Figure 3C). The BCVA of male patients with variants in exon1-14 (0.36 [0.17, 0.48] logMAR) was significantly better than that of male patients with variants in ORF15 (0.40 [0.20,0.70] logMAR) (P = 0.005) after age adjustment, however, the comparison between RCC1 and ORF15 was no significant difference (P = 0.048) (Figures 3A,B). BCVA was not associated with location or variation type in female carriers (all P > 0.05, respectively) (Supplementary Figures 5A-C).

Refractive Error in Patients With RPGR Variations
Spherical equivalent refraction was used to assess the severity of the RE. The percentage of female carriers with high myopia was significantly greater than that of males (109/165 and 51/179, respectively). Females with variants in RPGR showed a more serious of SER than males (−8.00 [−12.00, −4.19] in female carriers and −3.72 [−6.99, −1.28] in male patients, P = 5.46E-10) ( Figure 2B). Logistic regression showed that RE was unrelated to age, location or variation type in both male patients and female carriers (all P > 0.05) (Figure 2D and Supplementary  Figures 4, 5D-F).
In addition, the fundus changes vary widely among patients with RPGR variants, including gray-white fundal spots, tessellated fundus, retinal degeneration to macular degeneration in males and female carriers.

DISCUSSION
In this study, 97 rare RPGR variations were detected in our in-house exome sequence data. A total of 585 variants were identified from previous studies. All in-house data and previous literature data were combined for further genotypephenotype analysis.
Enrichment and the frequency analyses showed that the benign variants were enriched in non-RCC1 regions. Multistep bioinformatics analyses provided evidence that the corresponding prediction scores were lower than those of variants in the RCC1 region. In addition, segregation and phenotypic consistency analyses further confirmed the benign nature of the variants. A few families also showed variants in other known IRD genes, and some variants were verified in unaffected controls. In previous studies, three families with compound heterozygous variants in RPGR, one allele was an in-frame variant in ORF15, and the other allele was a truncation variant (Pelletier et al., 2007;Neidhardt et al., 2008). Moreover, in-frame variants in ORF15 (spanning 3-36 bp) in healthy individuals were reported in a previous study, suggesting that at least some in-frame variants in ORF15 of RPGR might not be causative (Karra et al., 2006). An in vitro assay developed in a previous study illustrated that some variations located outside of the RCC1 regions might be non-disease-causing polymorphisms (Zhang Q. et al., 2019). Taken together, these findings suggest that at least some missense changes and in-frame variants in the non-RCC1 region might not be pathogenic. Interestingly, several truncation variants at C-terminal region of RPGR had a high frequency in the gnomAD database, but all of them were low-confidence. If the high frequency of these truncations were validated, the pathogenicity of truncations around and downstream of these variants should be considered with greater caution.
More than 85% of the patients with pathogenic RPGR variants had RP. The remainder were diagnosed with a variety of X-linked retinal diseases, including IRD, CORD, COD, high myopia, and MD, among others. The BCVA of the probands with RPGR was age depended, and the BCVA of female carriers was better than that of male patients. In addition to age, the location of the variants in RPGR might play important roles in male patients with BCVA but not in female patients. Male patients with variants in exon1-14 retained better BCVA.
Based on our analysis, there were no significant differences in the SER with regard to the variation type, location or age in either males or females. These results suggest that progression of myopia is relatively slow in patients with variants in RPGR. Because some probands exhibited high myopia in the early stage, the specific screening of RPGR was initially not carried out in many of these patients. This emphasizes the importance of performing a comprehensive examination of patients with early-onset high myopia and of considering the possibility that RPGR variants may exist in these patients. RE was only associated with gender and was more serious in females than in males.
In summary, the results of this study suggested that missense and in-frame variants located outside the RCC1-like domain are likely benign. The pathogenicity criteria for RPGR should be considered with greater caution. Increase of age and location of variants in ORF15 contribute to the reduction of BCVA in males. These results are valuable for understanding genotypes and phenotypes of RPGR.

DATA AVAILABILITY STATEMENT
The data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Institutional Review Board of Zhong Shan Ophthalmic Center. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

AUTHOR CONTRIBUTIONS
XX, SL, and QZ recruited patients. JY, LZ, WS, XX, and SL collected the clinical data. XX and QZ performed whole exome analysis. QZ, JY, and LZ performed the bioinformatic analysis and designed the study. JY, LZ, JO, WS, and QZ discussed the results and wrote the manuscript. All authors reviewed and approved the manuscript.

FUNDING
This study was supported by grants from the National Natural Science Foundation of China (81371058 and 81970837) and the Fundamental Research Funds of the State Key Laboratory of Ophthalmology.

ACKNOWLEDGMENTS
The authors thank all patients and family members for their participation.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fgene.