Color fundus photography (A), fundus autofluorescence (B), and optical coherence tomography (Heidelberg Spectralis system; Heidelberg Engineering) through the fovea (C) and through peripheral pigment clumps (D) in an area marked with a white line in part A, of the affected 84-year-old male patient showing advanced retinal degeneration resembling choroideremia. Pink arrowhead indicates subretinal pigment; white arrowhead, intraretinal pigment; yellow arrowhead, retinal vessel.
Optos images (A), fundus autofluorescence (B), and optical coherence tomography (Heidelberg Spectralis system; Heidelberg Engineering) (C) showing typical retinitis pigmentosa phenotype.
A, Asymptomatic daughter of the patient carrying the RPGR c.779-5T>G variant showing a macular radial pattern or tapetal photoreceptor reflex typical of a female carrier of an RPGR pathogenic variant and not seen in choroideremia. For comparison, a female choroideremia carrier is shown in image B, where patchy X chromosomal inactivation results in areas of retinal pigment epithelium degeneration (fine granular pattern) interspersed with healthy tissue.
Amplicons of GFP transcripts generated following reverse transcription–polymerase chain reaction of transfected HEK293T cells transfected with GFP constructs containing no intron (positive control) or the wild-type or mutant RPGR intron 7 variant (A). Band density mean gray values from A were normalized to a loading control, and the relative ratios of each GFP transcript type within the total transcript population for a given sample were determined (B). Correctly spliced transcripts were identified in all transfected samples (approximately 0.5 kb) and sequence confirmed (C). Cells transfected with the reporter construct containing either the wild-type or mutant intron variant also revealed partially spliced transcripts (approximately 0.75 kb). Cells treated with mutant construct also contained unspliced GFP transcripts (approximately 0.8 kb), which were sequence confirmed (D).
Splicing assay data suggest the mutant “G” nucleotide would result in loss of recognition of the intron 7 splice acceptor site in a population of RPGR transcripts (A) with 2 potential outcomes. Option A: use of an alternative splice acceptor site within exon 8, resulting in an out-of-frame coding sequence and premature truncation of RPGR protein (B); or option B: skipping of exon 8 which is predicted to leave an in-frame truncated RPGR ORF15 protein (C).
eMethods. RNA splicing assay
eFigure 1. Sequences used in the RNA splicing assay
eFigure 2. Retinitis pigmentosa GTPase regulator (RPGR) gene structure
eFigure 3. The variants used in the GFP reporter constructs
eFigure 4. Reporter GFP constructs
eTable. Outputs provided from the Human Splicing Finder
Customize your JAMA Network experience by selecting one or more topics from the list below.
Cehajic-Kapetanovic J, McClements ME, Whitfield J, Shanks M, Clouston P, MacLaren RE. Association of a Novel Intronic Variant in RPGR With Hypomorphic Phenotype of X-Linked Retinitis Pigmentosa. JAMA Ophthalmol. 2020;138(11):1151–1158. doi:10.1001/jamaophthalmol.2020.3634
Can a mild RPGR phenotype be explained by impaired splicing caused by a novel pathogenic variant?
In this case series study, a 3-generation X-linked family with a novel RPGR intronic variant (c.779-5T>G) demonstrated a mild phenotype with unusually preserved visual acuity, where a proband was misdiagnosed as having choroideremia. The in vitro splicing assay demonstrated that the RPGR variation reduced the efficiency of intron splicing compared with wild type, leading to a population of mutant and normal transcripts.
A novel splice-site pathogenic variant in RPGR may lead to a hypomorphic X-linked phenotype with preservation of cones and visual acuity.
Pathogenic variants in retinitis pigmentosa GTPase regulator (RPGR) gene typically lead to a severe form of X-linked retinitis pigmentosa, which is associated with early severe vision loss.
To investigate an X-linked retinal degeneration family with atypical preservation of visual acuity in the presence of a novel deep intronic splice site RPGR c.779-5T>G variant.
Design, Setting, and Participants
In this case series, 3 members of an X-linked retinal degeneration family were studied by in-depth phenotyping and genetic screening at a single center. Data were collected and analyzed from November 2018 to March 2020.
Main Outcomes and Measures
Data were collected on full ophthalmic history, examination, and retinal imaging. A full retinitis pigmentosa gene panel was analyzed by next-generation sequencing. The pathogenicity of the RPGR c.779-5T>G variant was assessed by in silico splice prediction tools and by purpose-designed in vitro splicing assay.
An 84-year-old man was referred with clinical diagnosis of choroideremia and possible inclusion into a gene therapy trial. He presented with late-stage retinal degeneration and unusually preserved visual acuity (78 and 68 ETRDS letters) that clinically resembled choroideremia. His 23-year-old grandson was still in early stages of degeneration but showed a very different clinical picture, typical of retinitis pigmentosa. Next-generation sequencing identified a sole RPGR c.779-5T>G variant of undetermined pathogenicity in both cases. The daughter of the proband showed an RPGR carrier phenotype and was confirmed to carry the same variant. The molecular analysis confirmed that the RPGR c.779-5T>G variation reduced the efficiency of intron splicing compared with wild type, leading to a population of mutant and normal transcripts. The predicted consequences of the pathogenic variant are potential use of an alternative splice acceptor site or complete skipping of exon 8, resulting in truncated forms of the RPGR protein with different levels of glutamylation.
Conclusions and Relevance
These results support the importance of careful interpretation of inconsistent clinical phenotypes between family members. Using a molecular splicing assay, a new pathogenic variant in a noncoding region of RPGR was associated with a proportion of normal and hypomorphic RPGR, where cones are likely to survive longer than expected, potentially accounting for the preserved visual acuity observed in this family.
Inherited retinal degenerations are a heterogeneous group of disorders that have surpassed diabetic retinopathy as the leading cause of blindness in the working-age population.1 They display an exceptional variation in their mode of inheritance, age at onset, phenotypic features, and underlying genetic defects.2 Approximately one-third of inherited retinal degenerations are caused by X-linked disease that includes pathogenic variants in the retinitis pigmentosa GTPase regulator (RPGR) gene.3 Usually the condition manifests with severe disease in men, with early-onset and rapidly progressing sight loss that leads to legal blindness by the fourth decade of life. This distinct phenotype and the X-linked family history subsequently guide genetic testing and result in identification of causative pathogenic variants in most cases.
Herein, we present a family with X-linked pattern of inheritance and a heterogeneous phenotype where the proband had unusually preserved visual acuity for the stage of retinal degeneration.4 He showed atypical phenotype with some clinical features of choroideremia5 and was referred to our retinal genetics clinic for genetic testing and a possibility of enrollment into a choroideremia gene therapy trial. In contrast, his grandson (from an asymptomatic daughter) displayed a clinical phenotype consistent with RPGR disease. An additional challenge with this family was the presence of a sole RPGR c.779-5T>G variant of undetermined pathogenicity. Although we identified 1 report of this variant in the literature,6 there was no description of the associated phenotype or functional assessment of pathogenicity.
In this study, we hypothesized that the RPGR variant was pathogenic, but given the preservation of visual acuity, it is possible that the mutant form maintained some function giving the hypomorphic RPGR form and/or that a proportion of normal functioning RPGR was being produced, leading to a milder phenotype. We have previously reported a case of misdiagnosing choroideremia in a patient with RPGR owing to a small in-frame deletion7 and given that gene therapy clinical trials are now ongoing for both X-linked retinitis pigmentosa and choroideremia, establishing the correct diagnosis appears to be even more important to allow recruitment into the correct trial. We thus developed an in vitro functional splicing assay to confirm the pathogenicity of the splice-site variant and to elucidate the mechanism of action of this variation on the RPGR protein function.
The study design adhered to the tenets of the Declaration of Helsinki.8 This study follows the reporting guideline for case series. Institutional review boards at Oxford University Hospitals approved the studies, and patients provided written informed consent. Patients received no compensation or incentive to participate.
An X-linked retinal degeneration family was evaluated by in-depth phenotyping. All medical records were reviewed and data were collected on family history, full ophthalmic history, and clinical examination. Retinal imaging studies, including fundus autofluorescence and optical coherence tomography, were taken with the Heidelberg Spectralis system (Heidelberg Engineering). Initial next-generation sequencing revealed an RPGR c.779-5T>G variant of undetermined pathogenicity. Further genetic testing was performed involving analysis of 115 genes associated with retinitis pigmentosa (“the full RP panel”) by next-generation sequencing including sequencing of other X-linked genes (RP2 and OFD1) and autosomal dominant retinitis pigmentosa panel of exonic hotspots and the rhodopsin and RDS/peripherin genes coding regions. Because initial sequencing covered only 90% of the CHM gene, including evaluation of small insertions and deletions, additional testing was performed to cover larger events (insertions/duplications/deletions),9 fill in the exonic gaps, include the sequencing of established deep intronic CHM pathogenic variants, and cover pathogenic variants in the CHM promoter region.10 Lastly, the ORF15 region of the RPGR gene was analyzed for the presence of any additional pathogenic variants.
The pathogenicity of the RPGR c.779-5T>G variant was assessed in a population-based genome data set (the Single Nucleotide Polymorphism Database). The allelic frequency of the variant was further evaluated in gnomAD, which includes the Exome Aggregation Consortium data set. Splice prediction analysis using the Human Splice Finder tools (HSF matrices and MaxEnt analysis) was performed to identify any potential effect of the variant on RNA splicing. An RNA splicing assay was designed to investigate the influence of the mutant allele on RNA splicing in vitro. For further details of the assay, see the eMethods and eFigure 1A and B in the Supplement.
The affected 84-year-old man was referred to a specialist retinal genetics clinic with a clinical diagnosis of choroideremia. He was first aware of problems with night vision during his late teenage years. His central reading vision remained excellent until late into adulthood and began to decrease in his late 60s. The family history revealed a 23-year-old grandson (from his asymptomatic daughter) who was also affected by night vision difficulties and had no problems with central vision. There was no reported consanguinity in the family.
On clinical examination of the proband, the visual acuity was mildly affected at 78 letters on the Early Treatment Diabetic Retinopathy Score (ETDRS) chart in the left eye and 68 letters in the right eye. There were mild lens opacities in both eyes but otherwise the anterior segment examination was unremarkable. Fundus examination and imaging (Figure 1) revealed pale, barren appearance with widespread degeneration and some peripheral pigment clumping, resembling advanced choroideremia. Retinal vessels had relatively normal caliber and the optic nerve was normal color: features often associated with choroideremia rather than retinitis pigmentosa. Fundus autofluorescence imaging showed generalized hypoautofluorescence with small central scalloped islands of normal to increased autofluorescence, with evidence of central macular pigment indicating preservation of cone photoreceptors. Optical coherence tomography showed that these central retinal islands had relatively preserved underlying retinal pigment epithelium and the ellipsoid zone, consistent with high visual acuity. The foveal outer segment length, as previously measured in patients with RPGR from the inner RPE to the inner ellipsoid zone (3), was 48.5 μm in the left eye and 45.7 μm in the right eye. This was significantly higher than the average outer segment length found in the patients with RPGR: 35.5 μm compared with 61.9 μm in normal patients (3). In addition, the tomography showed outer retinal tubulations at the edges of preserved islands and peripheral retinal thinning. Moreover, an optical coherence tomography cross-section (Figure 1D) through peripheral pigment clumps in an area marked with a white line in Figure 1A showed predominantly subretinal pigment location (red arrowhead) as seen in choroideremia and other pathogenic RPE variants, although in some instances, the pigment location was suggestive of intraretinal pigment migration (white arrowhead) as described in 79% of patients with pathogenic variants in ciliary genes including the RPGR.11
Clinical assessment of the affected grandson showed excellent visual acuity at 89 ETDRS letters in the left eye and 85 letters in the right eye. Retinal examination and imaging (Figure 2) revealed a clinical phenotype typical of retinitis pigmentosa with midperipheral bone spicules. Fundus autofluorescence showed extensive midperipheral patchy hypoautofluorescence; centrally, there was a parafoveal hyperautofluorescent ring typically seen in ciliopathies including RPGR-related dystrophy. Optical coherence tomography showed good central retinal structure with preservation of the retinal pigment epithelium and the ellipsoid zone consistent with excellent visual acuity. The foveal outer segment length was normal at 60.7 μm in the left and 62.5 μm in the right eye (3).
Next-generation sequencing identified the RPGR c.779-5T>G variant of undetermined pathogenicity (eFigure 2 in the Supplement) in both cases. In addition, the grandson, but not the proband, was found to be heterozygous for C1QTNF5 c.393C>T (p.Asn131Lys) and PDE6B c.2116A>T (p.Lys706*) variants, with no related clinical features. Further genetic screening, including the CHM gene, revealed no additional pathogenic pathogenic variants. Additional family studies showed that the daughter of the proband manifested with an RPGR carrier phenotype (Figure 3), and genetic testing confirmed that she was a carrier the same genetic variant. However, genome database analysis (gnomAD) could not exclude the possibility that this was a rare polymorphism. Splice prediction analysis using the Human Splice Finder indicated that this pathogenic variant may influence the splice acceptor site, but the scores were at less than the threshold at which a change would be considered to completely disrupt a splice site (eTable in the Supplement). Given the absence of any other known pathogenic variant, this intronic pathogenic variant was thus further investigated by an in vitro splicing assay.
A shortened version of intron 7 was created in the splice assay and analyzed using the Human Splicing Finder tool. The splice donor and splice acceptor sites were predicted to act as preferential splice sites (donor 78.86; acceptor 93.46) with a strong predicted branch point in an appropriate location (branch point 86.72). A wild-type and mutant variant of the shortened intron was created and inserted into the GFP coding sequence of a reporter construct (eFigure 3 in the Supplement). If not successfully spliced, the presence of the intron would result in a premature stop codon and therefore not generate GFP expression. Successful splicing would enable complete removal of the intron and would therefore lead to GFP expression. This provided a screening step for the assay and confirmed successful splicing of both variants prior to transcript analysis (eFigure 4 in the Supplement).
Transcript analysis from cells treated with the GFP constructs containing wild-type or mutant RPGR intron 7 revealed that the single nucleotide change equivalent to the RPGR c.779-5T>G variant reduces the efficiency of the intron 7 splice acceptor site. Whereas cells treated with the wild-type GFP construct generated 72% correctly spliced GFP transcripts, cells treated with the mutant GFP construct generated only 29% correctly spliced GFP transcripts (Figure 4). Both the wild-type and mutant GFP constructs generated partially spliced transcripts. The population of these partially spliced transcripts was relatively consistent between the 1 variants at around one-third of the total GFP transcript population. Yet whereas for the wild-type intron 7 variant, the remaining transcript population consisted only of fully spliced transcripts, the mutant intron 7 variant transcript population was an even mix of fully spliced and unspliced transcripts. The wild-type intron 7 variant generated no detectable unspliced transcripts; these were only evident in cells treated with the mutant intron 7 variant.
The predicted consequence of the pathogenic variant could be use of an alternative splice acceptor site (Figure 5). If the expected acceptor site for intron 7 is skipped, a potential downstream sequence may be interpreted as an alternative splice site. The Human Splice Finder tool predicts the next most likely site lies within exon 8, which, if used as an acceptor, would lead to an out-of-frame coding sequence that would terminate within 12 amino acids and generate a truncated form of RPGR. Another possibility could be skipping of exon 8 resulting in an in-frame messenger RNA that is likely to be translated and glutamylated, resulting in a truncated form of RPGR ORF protein.
Thus, given the lack of other pathogenic variants, including any variants in the CHM gene, the presence of an RPGR female carrier phenotype, and the supporting evidence from the in vitro studies all suggest the diagnosis in this family to be X-linked retinitis pigmentosa secondary to the RPGR c.779-5T>G variant.
In this era of retinal gene therapy, considering genetic testing to individuals with inherited retinal degenerations has become even more important.12 In a rapid paradigm shift, genetic testing is no longer focused solely on counseling but it can lead to potential recruitment to clinical trials and indeed to treatment by approved retinal gene therapy for RPE65-related disease.13 The efficacy of any gene-specific therapy requires a correct genetic diagnosis, or the therapy risks replacing the wrong gene. However, genetic testing often produces ambiguous results, posing challenges in interpretation of genetic variants of unknown significance. Herein, we describe a family with X-linked inherited retinal degeneration with an unusual heterogeneous phenotype in the presence of a splice site RPGR c.779-5T>G variant of undetermined pathogenicity.
The phenotype of the proband was unusual for late-stage RPGR disease with no pigment spicules, normal vessel caliber, and normal optic nerve appearance. Rather, there was a barren appearance to the fundal periphery with some pigment clumping and a small island of central retina with a choroideremia-like autofluorescence pattern.5,14 The proband’s grandson had a phenotype that was more consistent with RPGR retinitis pigmentosa, with typical bone spicules in mid periphery. This distinction between the peripheral pigment clumping in the late-stage choroideremia and the bone spicules in retinitis pigmentosa is a very useful sign that can help differentiate between 2 conditions in cases of uncertainty. In choroideremia, the retinal pigment epithelium is thought to be the primary cell type driving the degeneration with pigment clumping and complete cell loss in the late stages.15 However, in retinitis pigmentosa, the primary loss of photoreceptors results in the migration of retinal pigment epithelium into the outer plexiform layer through areas, which lack photoreceptors giving rise to bone spicules.16 The presence of a typical RPGR female carrier phenotype, in this case the proband’s daughter, can also confirm the RPGR diagnosis.17
Another unusual feature of the proband was the preservation of the central ellipsoid zone accounting for his good visual acuity at age 84 years, despite progressive retinal degeneration, which in RPGR disease usually leads to severe vision loss by the fourth decade of life.3 It is therefore possible that the RPGR c.779-5T>G variant produces a hypomorphic form of RPGR, where cones are preserved longer than expected. The grandson was still in early stages of degeneration, but his visual acuity was excellent, suggesting normal cone function. The challenge with this variant was that there were no previously reported functional studies and the pathogenicity was undetermined. In addition, the nucleotide change was outside the RPGR ORF15 pathogenic variant hotspot region18 and splice site pathogenic variants themselves are uncommon within any part of the RPGR; the most common pathogenic variants being small deletions that lead to frameshifts followed by nonsense pathogenic variants.19 Moreover, deep intronic pathogenic variants are not well known to influence splicing, and prediction of splice sites in noncoding regions of genes is one of the most challenging aspects of gene structure recognition.20 Skipping of exon 26 in USH2A has been observed for a -6T>A pathogenic variant in intron 25,21 but to our knowledge, there has been no other reported −5 pathogenic variant linked to the RPGR disease. Indeed, the in silico analysis produced ambiguous results, confirming difficulties in using the splice prediction tools. Further molecular analysis was therefore necessary to prove the pathogenicity of the variant and demonstrate presence of some functional RPGR protein.
We therefore set up a molecular analysis whereby we designed reporter constructs to investigate the effect of the variant on RNA splicing in vitro. By comparing constructs identical but for the T/G allele at the −5 position in RPGR intron 7, we found that the single nucleotide variation does influence splicing. Cells treated with mutant variant had 2.5 times fewer fully spliced transcripts than cells treated with the reporter containing the wild-type intron variant. For both constructs, incompletely spliced transcripts were identified. Given the artificial nature of the splicing assay, this was not unexpected. Critically, the reporter construct containing the wild-type intron 7 variant generated no unspliced transcripts; they were all completely or partially spliced. In contrast, the reporter construct containing the mutant intron 7 variant did generate unspliced transcripts. These data indicate that the RPGR c.779-5T>G pathogenic variant reduces the splicing efficiency and is ignored as a splice acceptor site in at least one-third of transcripts. This suggests that alternative splice acceptor sites within exon 8 may be used or that skipping of the entire exon 8 could occur in the transcript population, but the outcome is not 100% because the change is within the variable noncanonical splice acceptor site region, and a population of normally spliced RPGR messenger RNA transcripts is thus likely to be produced.
The use of an alternative splice acceptor site within exon 8 was predicted to lead to an out-of-frame coding sequence and premature truncation of RPGR protein. However, the skipping of exon 8 was predicted to create an in-frame truncated version of the RPGR protein. The amino acids encoded by the RPGR ORF15 region undergo posttranslational modification, called glutamylation, which is essential for protein function.22 The in-frame truncated RPGR protein is not predicted to affect the ORF15 region, and therefore, any truncated protein would be fully glutamylated and thus maintain a degree of normal function. In contrast, pathogenic variants affecting the ORF15 region, especially large deletions, are associated with the level of glutamylation and are known to be detrimental for protein function, leading to degeneration.23,24 For the RPGR pathogenic variant described here, it is predicted that one-third of the transcript population will undergo alternative splicing or exon 8 skipping, generating a truncated mutant RPGR, which in case of exon 8 skipping is expected to be fully glutamylated. Thus, a population of fully functional RPGR protein is also anticipated in these patients, which would likely provide a level of protein activity that accounts for the milder RPGR phenotype with preserved cone function. However, a dominant negative effect of the mutant protein cannot be excluded. Either way, the clinical picture suggests that cones may not need as much RPGR as rods and so the effects of RPGR gene therapy would be predicted to improve cone function ahead of rod function. Early results from a first-in-human RPGR gene therapy trial suggest this, by showing reversal of photopic central visual field loss with increase in retinal sensitivity.25 We hypothesize that the difference in clinical phenotypes between the proband and the grandson is owing to the novel hypomorphic RPGR phenotype. The phenotype leads to the preservation of central cones and visual acuity, even in advanced stages, by which time the peripheral retina has markedly degenerated as seen in the proband. The grandson also has a very mild phenotype because his vision is inexplicably high for a typical RPGR patient. Long-term follow up of the grandson will allow age-matched phenotypic comparison with the proband.
The results of this study should be considered within the context of a small study sample size.
In conclusion, we report on challenges in reaching the diagnosis of X-linked retinitis pigmentosa when faced with inconsistent phenotypes within the same family in the presence of a variant of uncertain pathogenicity. The study highlights the potential value of a molecular splice assay in confirming a new pathogenic variant in a noncoding region of RPGR.
Corresponding Author: Jasmina Cehajic-Kapetanovic, PhD, Nuffield Laboratory of Ophthalmology, John Radcliffe Hospital, University of Oxford, Headley Way, Level 6 West Wing, Oxford OX3 9DU, England (firstname.lastname@example.org).
Accepted for Publication: August 3, 2020.
Published Online: September 24, 2020. doi:10.1001/jamaophthalmol.2020.3634
Author Contributions: Dr MacLaren had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Cehajic Kapetanovic, McClements, MacLaren.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Cehajic Kapetanovic, McClements, MacLaren.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: MacLaren.
Obtained funding: Cehajic Kapetanovic, MacLaren.
Administrative, technical, or material support: Shanks, Clouston, MacLaren.
Supervision: Cehajic Kapetanovic, MacLaren.
Conflict of Interest Disclosures: Dr MacLaren reported grants and personal fees from Biogen Inc during the conduct of the study. No other disclosures were reported.
Funding/Support: This work was supported by grants from National Institute for Health Research Oxford Biomedical Research Centre, Oxford, UK, and Medical Research Council UK. Dr Kapetanovic was also funded by Global Ophthalmology Awards Fellowship, Bayer, Switzerland.
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: We thank the patients for granting permission to publish this information.