A hypomorphic variant of choroideremia is associated with a novel intronic mutation that leads to exon skipping

ABSTRACT Introduction Molecular confirmation of pathogenic sequence variants in the CHM gene is required prior to enrolment in retinal gene therapy clinical trials for choroideremia. Individuals with mild choroideremia have been reported. The molecular basis of genotype–phenotype associations is of clinical relevance since it may impact on selection for retinal gene therapy. Methods and Materials Genetic testing and RNA analysis were undertaken in a patient with mild choroideremia to confirm the pathogenicity of a novel intronic variant in CHM and to explore the mechanism underlying the mild clinical phenotype. Results A 42-year-old male presented with visual field loss. Fundoscopy and autofluorescence imaging demonstrated mild choroideremia for his age. Genetic analysis revealed a variant at a splice acceptor site in the CHM gene (c.1350-3C > G). RNA analysis demonstrated two out-of-frame transcripts, suggesting pathogenicity, without any detectable wildtype transcripts. One of the two out-of-frame transcripts is present in very low levels in healthy controls. Discussion Mild choroideremia may result from +3 or −3 splice site variants in CHM. It is presumed that the resulting mRNA transcripts may be partly functional, thereby preventing the development of the null phenotype. Choroideremia patients with such variants may present challenges for gene therapy since there may be residual transcript activity which could result in long-lasting visual function which is atypical for this disease.


Introduction
Retinal gene therapy clinical trials for choroideremia have generated interest in CHM gene variants of unknown significance, especially when they are associated with unexpected phenotypes.Choroideremia is a rare X-linked inherited retinal disease, affecting approximately 1 in 50,000 males.It is caused by mutations in the CHM gene that encodes for Rab escort protein 1 (REP 1) protein.The primary affected cell type appears to be the retinal pigment epithelium (RPE), although some independent photoreceptor and choriocapillaris degeneration has been demonstrated (1).
Patients typically present within the first two decades of life with nyctalopia and progressive peripheral vision loss due to retinal degeneration in the mid-peripheral retina with centripetal and centrifugal progression, typically leaving a central residual island of retinal tissue at the posterior pole (2).Surviving RPE adopts either a smooth or mottled appearance on fundus autofluorescence (FAF) (3).The smooth FAF region represents an earlier stage of the disease, characterised by a preserved ellipsoid zone.The mottled zone indicates more advanced stages due to a mostly disrupted ellipsoid zone (4).The RPE changes correlate with the health of the overlying ellipsoid zone in choroideremia visualised with optical coherence tomography (OCT) imaging (4).Individuals with mild choroideremia are characterised by delayed progression of the chorioretinal degeneration (4,5).Visual acuity (VA) is preserved until the degeneration encroaches on the fovea, which often occurs after 40 years of age (6)(7)(8)(9).
REP1 deficiency causes non-syndromic retinal degeneration in humans.REP1 protein is expressed in all human cells and is involved in the prenylation and transport of Rab proteins, such as RAB27A or RAB6A, which are the key regulators of membrane trafficking (10).Residual levels of full-length CHM transcript have been correlated with this mild phenotype (5).Recent research suggests that a +3 deletion in intron 7 leads to mild choroideremia mRNA expression and milder disease (5).This is significant as the first genotype-phenotype correlation in choroideremia and also suggests that the quantity of CHM mRNA therapy may only need to be 1% of the level of normal CHM expression to slow choroideremia progression (5).
The genetic mechanism of choroideremia helps inform the relationship between genotype and phenotype (5).The molecular basis of genotype-phenotype associations in choroideremia is of clinical relevance since they may impact on patient selection for retinal gene therapy.Herein, we report the workup of a novel splice site variant in the CHM gene in a patient with a mild choroideremia phenotype, with RNA and protein analysis which suggests a potential mechanism behind the mild phenotype.

Materials and methods
A patient was referred to the Oxford Eye Hospital, a tertiary referral centre for inherited retinal diseases.Data was collected retrospectively from routine clinical care.He had undergone full clinical examination, including VA, slit lamp, and fundoscopic examination.To assess the integrity of the central retinal sensitivity, mesopic Macular Integrity Assessment (MAIA) microperimetry (Centervue SpA, Padova, Italy) was performed using the 10-2 grid, 4-2 testing strategy and Goldmann size III, with no formal dark adaptation (11).Retinal imaging included ultra-wide field imaging (Optos, Dunfermline, Scotland), short-wavelength FAF, and OCT imaging (Heidelberg Spectralis, Heidelberg Engineering, Heidelberg Germany).Electroretinography was not undertaken due to the advanced disease stage.
Blood samples were collected for genetic testing as part of the NHS genomic medicine service (12).Molecular genetic testing was undertaken using Next-Generation Sequencing (NGS) with all variants confirmed by Sanger sequencing (13).Enrichment was performed with the Twist Human Core Exome Multiplex Hybridization Kit (Twist Bioscience, San Francisco, USA), with target regions defined as coding exons ±10 bp.NGS was performed using the Illumina NovaSeq6000 by the High-Throughput Genomics Group at the Wellcome Trust Centre for Human Genetics, Oxford.The analytical sensitivity of the NGS for single nucleotide substitutions was 99.94% (95% CI: 99.89-99.97%)for regions covered to 20×, for hemizygous single nucleotide substitutions was 98.4% (95% CI: 95.95-99.56%)for regions covered to ≥10×, and for small insertions/deletions (indels) was 97.16% (95% CI: 95.49-98.34%)for regions covered to 20× (12).
Molecular analyses were performed on peripheral blood mononuclear cells (PBMC) isolated from additional whole blood samples.To determine the presence of wild-type and aberrant transcripts, cDNA was synthesized by reverse transcription from total RNA extracted from PBMCs (14).The region comprising exons 9 to 12 of the CHM transcript was amplified by PCR using a forward primer that binds within exon 8 (5'-CTTTATATGGCCAAGGAGAAC-3') and a reverse primer that binds to exon 13 (5'-TAGAAGATGTGCAAGTCAAATG -3') (14).Amplified cDNA products were separated by electrophoresis on an agarose gel and visualized under ultraviolet light.The PCR products were subsequently analysed by Sanger sequencing to identify the amplified region.
The expression and function of REP1 protein were analysed by Western blot and prenylation assay, respectively, in PBMCs.A prenylation assay to measure the activity of REP1 protein was performed as described by Patricio M.I. and colleagues (15).In summary, the incorporation of biotin-containing isoprenoids into a substrate, in this case, RAB6A, by REP1 was assessed in a prenylation reaction.The reaction products were separated on a 10% sodium dodecyl sulfate polyacrylamide gels (CriterionTM Precast Gels, Bio-Rad Laboratories Ltd., Hemel Hempstead, UK) and blotted onto polyvinylidene difluoride (PVDF) membranes (Trans-Blot Turbo Midi PVDF, Bio-Rad) using the Trans-Blot Turbo Transfer Starter System (Bio-Rad).The membranes were blocked with blocking buffer and incubated for 1 hour with the primary antibodies anti-human REP1 (MABN52, Merck Millipore, dilution 1:2,500) and anti-Biotin (IRDYe 800CW Streptavidin, #926-32230, Li-Cor, dilution 1:5,000).Mouse monoclonal anti-β-Actin antibody (AM4302, Thermo Fisher Scientific, dilution 1:50,000) was used as loading control.After several washes, the membrane was incubated with IRDye 800CW donkey antimouse (#926-32212, Li-Cor) to detect REP1 and βActin.Odyssey XF Dual-Mode Imaging System was used to image the membranes and visualize the protein bands.

Results
A 42-year-old male was referred to our clinic for assessment of choroideremia after experiencing night vision problems.He was diagnosed with choroideremia at 15 years of age, but he reported that he did not have any symptoms at the time of diagnosis.His mother was a confirmed carrier of choroideremia.Moreover, his great-grandfather had been diagnosed with retinitis pigmentosa, although this could have been choroideremia incorrectly diagnosed (16).Segregation analysis was not available from the laboratory, whilst further family pedigree information was not sufficient to produce a diagrammatic representation of the patient's heritage.
On examination, he had a VA of 6/6 in the right eye and 6/ 7.5 in the left eye.The anterior segment of both eyes appeared unremarkable with no cataracts.Posterior segment assessments showed a characteristic central island of preserved retina surrounded by extensive RPE and peripheral choroidal atrophy in both eyes.
Fundus autofluorescence imaging showed asymmetric islands of residual RPE with sharply demarcated borders.The centre of the island had a smooth texture, with a mottled zone more peripherally (Figure 1b).This fundus autofluorescence appearance was typical of choroideremia.The area of the residual island was 7.65 mm 2 in the right eye and 13.90 mm 2 in the left (Figure 1b).
OCT imaging of both maculae showed a constricted ellipsoid zone with extensive RPE degeneration and choroidal atrophy, consistent with a choroideremia phenotype.Choroidal hypertransmission revealed the extent of RPE loss.Central retinal mean sensitivity, as indicated by mesopic microperimetry, was 2.5 dB with 0% fixation losses and 5.5 dB with 0% fixation losses for the right and left eyes, respectively.
Figure 2 is an adapted figure from an article by Aylward et al. (17) which illustrates residual FAF area (mm 2 , shown on a log scale) and age in a cohort of 56 eyes.Whilst this 42-year-old patient is predicted to have 5 mm 2 baseline FAF area in both eyes, the right eye shows 7.65 mm 2 baseline FAF area, typically seen in a 35-year-old, and the left eye shows 13.90 mm 2 baseline FAF area, more likely seen in a 29-year-old.This patient, as a 42year-old, is an outlier in terms of his preserved baseline FAF area, suggesting he has a mild phenotype for his age.Figure 2 indicates that his FAF area in the left eye is outside the limits of the 95% confidence intervals of the regression, while his right eye is just on the upper 95% confidence interval.
Genetic testing identified a novel hemizygous variant (NM_000390.3:c.1350-3C>G): a transversion of cytosine by guanine three bases into the splice acceptor site at location ChrX:g.85,155,717(GRCh37(hg19)), at the boundary of intron 10 and exon 11.This variant has not been reported before and was of uncertain clinical significance (12).No other variant was found that could be associated with the phenotype.
In silico prediction tools, such as "Mutation Taster," predict pathogenicity and strongly suggest that this intronic variant leads to a cryptic splice acceptor site at the end of intron 10, leading to aberrant splicing and an out of frame transcript (12,14).To further explore the pathogenicity of this novel variant, additional investigation was undertaken.Extra blood samples were taken from the patient for RNA testing to detect whether the intronic variant identified resulted in aberrant splicing, and for protein analysis, to determine the expression and activity of REP1.
RNA splicing analysis identified two out-of-frame transcripts, present in approximately equal proportions, without any wildtype transcript detected in the patient's sample (Figure 3).Sequencing analysis confirmed that the c.1350-3C > G variant resulted in a unique aberrant transcript with out-of-frame skipping of exon 11, r.1350_1413del p.(Gln451PhefsTer4), detected in the gel as the ca.350 bp band.The other aberrant transcript detected resulted from skipping of exons 10 and 11, r.1245_1413del p.(Cys416PhefsTer4); shown in the gel as the ca.245 bp band.This transcript was also out of frame but has been detected at very low levels in normal controls (14) but has not been explored further.Although the methodology used for this test is not quantitative, the level of r.1245_1413del p. (Cys416PhefsTer4) appeared significantly elevated in the patient compared to controls.
To understand if the patient expresses functional REP1 protein, we analysed the expression of REP1 and the presence of biotinylated RAB6A after a prenylation reaction, as described above.This showed that REP1 protein is not expressed in patient's cells with the mutation.However, biotinylated RAB6A protein is detected after the prenylation reaction in these samples, indicating that there is some residual prenylation activity even in the absence of REP1 protein (Figure 4).This residual activity could explain the mild phenotype observed in this patient.
Figure 5 summarizes the production of the two out-of-frame transcripts of interest to the following reclassification conversation.Table 1 summarises the results from the genetic workup and categorises each finding alongside the American College of Medical Genetics and Genomics (ACMG) guidelines to provide evidence for reclassification.This result provides substantial additional evidence in support of pathogenicity of the c.1350-3C>G variant.This result is consistent with a genetic diagnosis of choroideremia, and the patient is hemizygous for a CHM gene splice-site variant.Table 1 sets out how the variant satisfies the ACMG criteria for being collectively classified as a "likely pathogenic" variant, with one "strong," one "moderate" and two "supporting" pieces of evidence of pathogenicity (18).The mutation appears to be novel as it has not been reported before and is not present in population genome aggregation databases, supporting the "PM2" criteria for the likely pathogenicity of the variant (Table 1).17) showing a correlation between residual FAF area (mm 2 , shown on a log scale) and age in a cohort of 56 eyes (28 patients) for the right (hollow circles) and left (solid triangles) eyes.The green line illustrates that the patient, as a 42-year-old, is an outlier in terms of baseline FAF area for his age, within the orange 95% confidence interval lines.Four different samples from healthy donors (no CHM gene mutation) were included in this assay.While REP1 expression was not detected in two of the controls, the four samples show a high concentration of biotinylated RAB6A compared with the patient, which indicates a high prenylation activity in controls.Three technical replicates were conducted with patient's samples.REP1 protein expression is not detected in the patient carrying the c.1350-3C>G mutation.However, a 24 kDa band corresponding with biotinylated RAB6A is detected in the patient's samples, which suggests that some residual prenylation activity is present.The 42 kDa band corresponding to the βActin, used as the loading control, is detected in all the samples.

Discussion
We report a patient with a novel CHM "−3" splice site variant with a mild choroideremia phenotype.This is reinforced by RNA and protein analyses suggesting pathogenicity, with the absence of wildtype transcripts in RNA analysis and no REP1 protein detected on western blot.Mild choroideremia has been suggested to result from +3 or −3 splice site variants in the CHM gene arising from residual mRNA expression which could prevent the development of the null phenotype (19).The transcript (r.1245_1413del p.(Cys416PhefsTer4)), which was reported to be present at low levels in controls, may have escaped nonsensemediated decay, and thus contributed to prenylation of RAB6A and the subsequent mild phenotype.Choroideremia patients with such variants may not be suitable for gene therapy, since this prenylation activity could lead to toxic effects due to the overexpression of functional CHM gene.On the other hand, low levels of expression of the r.1245_1413del transcript in controls could suggest that there is a low risk of an interaction with the wildtype CHM transcript.
Whilst the reduction of prenylation of Rab GTPases is associated with REP1 deficiency from mutations such as in this case, multiple other factors contribute to Rab prenylation rates (20).REP2 protein (encoded by CHML gene), homologue of REP1, also participates in prenylation of Rabs, and compensates for the loss of REP1 in most human tissues except the eye (21).This could explain why biotinylated RAB6A is detected in this patient's peripheral blood even in the absence of REP1.The level of prenylation of Rabs, specifically in the eye, in these patients remains to be elucidated.
Other reported CHM gene variants which have led to a mild phenotype include a putative frameshift variant in a CHM c.1335dup, in which mRNA analysis identified a splicing alteration that restored the reading frame of the mutant transcript, which in turn may support residual activity by producing an aberrant protein (22).A similar mechanism of mild phenotypes in a cohort of 30 choroideremia patients was associated with a splice site variant outside the canonical donor sequence (c.940+3delA); these patients expressed residual levels of full-length CHM transcript (5).However, not all splice site variants cause a mild phenotype.Neighbouring canonical donor splice site variant (c.940+2T>A) appeared to be severe with the absence of CHM transcript (5).Residual incomplete or abnormal transcripts may enable a mild phenotype since CHM transcript expression has been correlated with the choroideremia phenotype (23).In particular, "+3/-3" mutations that are outside the invariant dinucleotide sequence may enable a mild phenotype of choroideremia (5).By contrast, three splice acceptor site mutations had severe phenotypes in other diseases: a c.1389-1G>A mutation in intron 7 in the LMNA gene (NM_170707.2) (24) and the mutations c.1350-2A>G and c.1350-2A>C are associated with severe symptomatic phenotypes (25).In comparison to the Figure 2 phenotypes which are potentially mild due to non-canonical splice site mutations, there are also canonical splice site mutations which are likely to be associated with more severe phenotypes.
The eye is especially suitable for gene therapy trials due to the accessibility of injections and surgical interventions (26), and the subretinal space benefits from immune privilege (27).Subretinal gene therapy injection is a safe and effective method to deliver the CHM gene to the photoreceptors and the RPE (28,29).However, if contraindicated due to the unknown effects of aberrant transcripts, alternative therapeutic techniques should be considered, such as CRISPR (Clustered Regularly Interspaced Short Palindromic Repeats).CRISPR-Cas9 base editing systems allow the theoretical modification of single bases within genomic DNA, which can be used to modify both splice donor and acceptor sites; adenine base editors (ABE) have the capability of converting a target A/T base pair to G/C in genomic DNA (30).Cas9 has been used to extend survival time in mice with Spinal Muscular Atrophy (SMA) by CRISPR/Cas9-mediated disruption of intronic splicing silencers to enhance exon 7 inclusion in iPSC-derived motor neurons and mice (31).In Duchenne Muscular Dystrophy (DMD), CRISPR genome-editing systems have been successfully demonstrated in human cells (32).These possibilities make reporting of novel mutations, such as in this patient, crucial for the future design of gene editing clinical trials (33).Alternatively, an AAV construct, with a mirtron that blocks the expression of the residual transcripts and includes a codon optimised CHM gene, may offer another therapeutic opportunity for patients (34).
Table 1.Evidence in this case to satisfy the ACMG criteria for classifying pathogenic variants.
The American College of Medical Genetics and Genomics (ACMG) Criteria for Classifying Pathogenic Variants (18) Evidence in this case to satisfy requirements Strong evidence of pathogenicity PS3 Well-established in vitro or in vivo functional studies supportive of a damaging effect on the gene or gene product RNA splicing analysis demonstrated splicing abnormalities in a blood sample.Figure 4 confirms the absence of REP1 protein.

Moderate evidence of pathogenicity PM2
Absent from controls in Exome Sequencing, 100 Genomes, or ExAC This variant has not been reported before in unaffected human populations.The variant is not listed in the databases of human variation curated by the Genome Aggregation Database (gnomAD) (more than 170,000 alleles screened at this locus).GnomAD V.2.1.1(exomes),accessed December 2022 (12).Supporting evidence of pathogenicity

PP3
Multiple lines of computational evidence support a deleterious effect on the gene or gene products In silico prediction tools strongly suggest this variant leads to a dysfunctional splice acceptor site.

PP4
Patient's phenotype or family history is highly specific for a disease with a single genetic etiology Figure 1 details the clinical presentation characteristic of a patient with choroideremia.The patient's great-grandfather had been diagnosed with retinitis pigmentosa, although this could have been choroideremia incorrectly diagnosed (16).

Conclusion
We report a novel "likely pathogenic" splice site mutation (c.1350-3C>G) in the CHM gene.This variant has been associated with mild choroideremia phenotype due to residual RNA transcripts that could enable prenylation activity sufficient to maintain some RPE function.As a result, we argue that this individual may not be suitable for AAV gene replacement therapy due to unknown interactions between the therapeutic full-length transgene and residual transcripts.

Figure 1 .
Figure 1.(a) Ultra-widefield optomap images of both eyes showing extensive retinal degeneration with choroidal atrophy consistent with choroideremia.(b) Shortwavelength autofluorescence imaging showing typical patterns of sharply demarcated areas of remaining hyperfluorescent tissue against atrophic retina giving an absent autofluorescent signal, with the following FAF areas on 55 degree images: right 7.65 mm 2 ((OD) RE FAF) and left 13.90 mm 2 ((OS) LE FAF).(c) Spectral-domain OCT scans showing ellipsoid zone shortening, intraretinal cysts and tubulations.Choroidal hypertransmission reveals the extent of RPE loss.(d) Right and left eye microperimetry sensitivity maps showing retinal sensitivities within the residual island of tissue.

Figure 2 .
Figure 2. Adapted figure from an article by Aylward et al.(17) showing a correlation between residual FAF area (mm 2 , shown on a log scale) and age in a cohort of 56 eyes (28 patients) for the right (hollow circles) and left (solid triangles) eyes.The green line illustrates that the patient, as a 42-year-old, is an outlier in terms of baseline FAF area for his age, within the orange 95% confidence interval lines.

Figure 3 .
Figure 3. Agarose gel electrophoresis of PCR products.The wild-type transcript is only detected in the two control samples.Lower molecular weight products are detected in the patient's sample.These bands correspond to a transcript with skipping of exon 11 (ca.350 base pairs (bp) band) which results from the mutation c.1350-3C>G, and a transcript with skipping of exons 10 and 11 (ca.245 bp band) that has also been observed at low levels in healthy subjects(14).Indeed, this band is also visible in the control samples, although with less intensity than that observed in the patient.Please see supplementary FigureS1for raw Sanger sequencing data.NTC: no template control.

Figure 4 .
Figure 4. Analysis of REP1 protein expression and function in the CHM patient.Four different samples from healthy donors (no CHM gene mutation) were included in this assay.While REP1 expression was not detected in two of the controls, the four samples show a high concentration of biotinylated RAB6A compared with the patient, which indicates a high prenylation activity in controls.Three technical replicates were conducted with patient's samples.REP1 protein expression is not detected in the patient carrying the c.1350-3C>G mutation.However, a 24 kDa band corresponding with biotinylated RAB6A is detected in the patient's samples, which suggests that some residual prenylation activity is present.The 42 kDa band corresponding to the βActin, used as the loading control, is detected in all the samples.

Figure 5 .
Figure 5.The location of the genetic variant and the impact on pre-mRNA splicing, detailing the two out-of-frame transcripts and lack of REP1 protein.Figure created by BioRender.com;published under MacLaren lab BioRender license.