Phenocopy of a heterozygous carrier of X-linked retinitis pigmentosa due to mosaicism for a RHO variant

We describe both phenotype and pathogenesis in two male siblings with typical retinitis pigmentosa (RP) and the potentially X-linked RP (XLRP) carrier phenotype in their mother. Two affected sons, two unaffected daughters, and their mother underwent detailed ophthalmological assessments including Goldmann perimetry, color vision testing, multimodal imaging and ISCEV-standard electroretinography. Genetic testing consisted of targeted next-generation sequencing (NGS) of known XLRP genes and whole exome sequencing (WES) of known inherited retinal disease genes (RetNet-WES). Variant validation and segregation analysis were performed by Sanger sequencing. The mutational load of the RHO variant in the mother was assessed in DNA from leucocytes, buccal cells and hair follicles using targeted NGS. Both affected sons showed signs of classical RP, while the mother displayed patches of hyperautofluorescence on blue light autofluorescence imaging and regional, intraretinal, spicular pigmentation, reminiscent of a carrier phenotype of XLRP. XLRP testing was negative. RetNet-WES testing revealed RHO variant c.404G > C p.(Arg135Pro) in a mosaic state (21% of the reads) in the mother and in a heterozygous state in both sons. Targeted NGQSS of the RHO variant in different maternal tissues showed a mutation load between 25.06% and 41.72%. We report for the first time that somatic mosaicism of RHO variant c.404G > C p.(Arg135Pro) mimics the phenotype of a female carrier of XLRP, in combination with heterozygosity for the variant in the two affected sons.

Genetic and genomic study. Peripheral blood samples were collected in EDTA tubes and DNA was extracted using the QiAmp, Gentra Puregene Cell kit (Qiagen, Antwerp, Belgium) or the ReliaPrep Large-Volume HT gDNA Isolation System (Promega, Leiden, the Netherlands) according to the manufacturer's protocol. DNA from buccal mucosa and hair follicles was extracted using NaOH and Tris-HCl. The coding and exonintron regions of the RPGR (ENST00000378505), RP2 (ENST00000218340) genes, as well as the deep intronic OFD1 IVS9 + 706A > G mutation were tested in the proband (III.1) using targeted next-generation sequencing (NGS). PCR-enrichment of all coding exons and flanking intron sequences was followed by sequencing by synthesis (MiSeq, Illumina). A variant of class 3, 4 or 5 was confirmed by Sanger sequencing. Individuals II.1 and III.1 underwent whole exome sequencing on a HiSeq 3000 (Illumina) after exome enrichment using the SureSelect Human All Exon V6 kit (Agilent). Read alignment and variant calling were performed using the CLC Genomics Workbench (v7. 5.4). Variants were annotated and filtered using in-house software. Variants were scored heterozygous or homozygous in case of a variant allele frequency of 20-69% and minimum 70% respectively. A selection of 265 RetNet genes was assessed (gene panel version v5). Variants were assessed on the basis of predictions done in Alamut HT/Alamut Batch. (Likely) pathogenic variants were confirmed using Sanger sequencing. Nucleotide numbering was done following HGVS-guidelines (http://www.hgvs.org) with nucleotide ' A' of the ATG as 'c.1' . Classification of variants was based on the ACMG guidelines with adaptations [9][10][11][12]  www.nature.com/scientificreports/ from blood using bidirectional Sanger sequencing of RHO exon 2 and data analysis using SeqPilot (JSI Medical Systems GmbH, Germany). To assess the mutation load in multiple tissues (blood, buccal mucosa and hair follicles), targeted deep sequencing was performed using a flexible targeted next-generation protocol, consisting of singleplex-PCR, followed bij NexteraXT library preparation and sequencing on a MiSeq instrument as previously described 13 . Variants were filtered with a minimum allele frequency threshold of 1%. The sequencing depth varied from 147 to 17,523 reads (after removal of duplicate reads and overlapping paired-end reads).

Results
Phenotypic characteristics.  14 . On blue-light autofluorescence imaging (BAF) a mottled aspect of a mostly hypoautofluorescent retinal periphery, with a small hyperautofluorescent ring surrounding the central macula was seen (Fig. 2). Near-infrared autofluorescence imaging showed a fairly homogeneous hyperautofluorescent central macular area with diffuse hypo-autofluorescence around it. Both eyes were affected equally on fundoscopy and autofluorescence imaging (Fig. 2). Goldmann visual fields showed a normal central sensitivity and a loss of sensitivity in both the pericentral and midperipheral regions, however with normal peripheral limits. Color vision testing was normal. ISCEV-standard full-field flash ERG showed absent scotopic rod-specific responses, significantly decreased combined rod-cone responses to intense flashes and significantly decreased and delayed photopic cone-specific responses (Fig. 3). A clinical diagnosis of RP was made based on all of the findings above.
Individual III.2. The younger son was first seen at the age of 19. He complained of nyctalopia and photophobia since he was a toddler. His general medical history revealed an attention deficit disorder (ADD). His BCVA was 20/20 in both eyes. Slit-lamp examination was unremarkable. Fundoscopy showed spicular intraretinal pigment www.nature.com/scientificreports/ migration and RPE alterations in the retinal periphery. Autofluorescence imaging with blue light revealed a mottled hypo-autofluorescent retinal periphery. Around the macula, a small, hyperautofluorescent ring with a hypo-autofluorescent zone around it was observed. Goldmann visual fields showed decreased pericentral sensitivity with normal peripheral limits. Color vision testing was normal. ISCEV-standard full-field flash ERG revealed considerably reduced and delayed scotopic rod-specific and photopic cone-specific all responses were decreased and delayed (data not shown). A diagnosis of RP was made based on the above findings. Imaging and functional testing showed individual III.2 was less severely affected than his elder brother (III.1). Both eyes were affected equally.

Sisters III.3 and III.4.
Both had a completely normal extensive ophthalmological examination. The RHO c.404G > C p.(Arg135Pro) variant was not found in both individuals.
Individual II.1. Individual II.1, the mother of the proband was first examined thoroughly at the age of 45. She had undergone radial keratotomy to correct moderate myopia in both eyes in her twenties. BCVA was 20/20 in both eyes. Slit-lamp biomicroscopy showed scars of the radial keratotomies in both eyes, but was otherwise unremarkable. Only when questioned specifically about symptoms of RP, she reported difficulty navigating at night, as well as mild photophobia. Fundoscopy showed bilateral, but only very limited sectors of intraretinal pigment migration and RPE alterations, especially nasally to the optic disc. Blue light autofluorescence imaging revealed patches of hyperautofluorescence randomly distributed and interspersed by normal areas throughout the retina in both eyes. The left eye was more affected than the right eye. (Fig. 2). This patchy distribution is a typical finding seen in female carriers of X-linked RP, due to the female X-inactivation, also known as lyonization 15,16 . Goldmann visual fields were normal except for a very mild decrease in sensitivity in the midperipheral fields. www.nature.com/scientificreports/ ISCEV-standard full field-flash ERG revealed a reduction of the amplitudes of scotopic, rod-specific and mixed rod-cone responses as well as photopic cone-specific responses to 70% of normal values, with a mild delay of the peak times for all (Fig. 3). Based on the findings in II.1 and her sons, and on phenotypes in this family, as well as the pedigree, an X-linked RP was suspected.
Genomic profiling. Given the suspicion of XLRP, genetic testing for XLRP was requested for individual III.1. Panel testing for RPGR (including ORF15), RP2, and a deep intronic OFD1 mutation did not reveal any (likely) pathogenic variants leading to further genetic testing. RetNet-WES analysis in both III.1 and his mother II.1 revealed a known variant in the RHO gene, transversion c.404G > C, leading to a missense change p. (Arg135Pro). Interestingly, the RHO variant was shown to occur in a mosaic state (variant allele frequency (VAF) of ~ 21%) in the mother (II.1) (Fig. 4). This was confirmed by Sanger sequencing. To investigate the mutation load in other tissues than blood in an accurate way, targeted NGS, characterized by a high coverage, was performed on DNA from maternal leucocytes, buccal cells and hair follicles (II.1) and revealed a VAF ranging from 25.06 to 41.72% in the different tissues, with the highest VAF found in hair follicles and the lowest in the patient's blood (Fig. 5).
The son was shown to be heterozygous for the RHO variant (Fig. 4). Targeted testing showed presence of the variant in both affected brothers (III.1 and III.2) and absence in the unaffected sisters (III.3 and III.4). This variant is a known RHO missense variant, already previously reported in ADRP 2,15 . Classification of the variant following ACMG and ACGS criteria revealed this is a likely pathogenic variant (class 4) (Table 1) 9-12 .

Discussion
We report a family including a female with a suspected X-linked RP carrier-like phenotype, who proved to be mosaic for the dominant variant c.404G > C p.(Arg135Pro) in the RHO gene, and her two sons with a classic RP phenotype, heterozygous for the same RHO variant.
RP can be inherited in an autosomal dominant, autosomal recessive, or X-linked fashion, depending on the gene involved. Analysis of the pedigree can be performed as a first step to discriminate between the different modes of inheritance. In ADRP, vertical transmission can be seen. In such families, the patients often display a similar disease course and comparable clinical manifestations. However, specific cases of ADRP with variable expressivity and non-penetrance have been described. For example, in PRPF31-associated ADRP, different patients from the same family with the same mutation can have a variable disease severity or can be even totally asymptomatic 17 . In XLRP, affected male individuals usually display a more severe and early-onset phenotype, while most female carriers show either RP characteristics later in life or stay asymptomatic 15 . In the latter case, thorough functional testing with electroretinography, BAF and fundus imaging, will reveal retinal abnormalities.
In the family reported here, BAF imaging in the mother revealed patches of hyper-autofluorescence randomly distributed and interspersed by normal areas throughout the retina in both eyes. This is a typical finding in female carriers of XLRP, due to X-inactivation 18 . Interocular asymmetry in the female proband was also a clinical aspect  www.nature.com/scientificreports/ suggestive of XLRP. Indeed, 9% of 125 females heterozygous for an RPGR mutation show interocular asymmetry 6 . Given this clinical presentation and the RP phenotype seen in her sons, genetic testing of the XLRP genes was conducted. However, no mutation was found. Subsequent RetNet-WES analysis in III.1 and his mother II.1 identified a variant in the RHO gene c.404G > C p. (Arg135Pro). Interestingly, mosaicism of this variant was shown in the mother II.1, estimated to be 21% on DNA from leukocytes, and heterozygosity was shown in her affected son III.1. Co-segregation of the variant with the disease was shown. Genetic mosaicism is characterized by two or more sets of cell lines with different genotypes in the same individual, resulting from a postzygotic mutation. A variety of factors, including the biological function of the gene, the intrinsic effect of the mutation, the moment when the mutation occurred, and its tissue distribution will determine the type of mosaicism and its phenotypic consequences. This was demonstrated by Cao et al. who investigated the allelic fraction and clinical effects of 120 clinically relevant mosaic single nucleotide variants. They showed that an alternate allele fraction (AAF) of 13-24% for mosaic variants in MTOR and PIK3CA could lead to Smith-Kingsmore syndrome, Cowden syndrome 5, and/or megalocephaly-capillary malformationpolymicrogyria syndrome, while a comparable AAF of 16-30% for mosaic variants in CACNA1A was found in asymptomatic parents of children affected with epileptic encephalopathies 19,20 .
Mosaicism can lead to a wide spectrum of phenotypes, ranging from pigment variations to neurofibromatosis type I and osteogenesis imperfecta type II 21 . Additionally, a few cases of mosaicism in eye-related phenotypes have been reported. In particular, it has been shown that approximately 15% of sporadic retinoblastoma cases are caused by postzygotic RB1 mutations, that intrafamilial phenotypic variability in aniridia cases can be explained by PAX6 mosaicism 22,23 . In 2016, the contribution of RHO mutations to ADRP in the Israeli and Palestinian populations was studied. Segregation analysis of a specific RP-causing RHO mutation [c.548_638dup p.(Ile214Alafs*147)] revealed that the mutation originated from a mosaic individual who did not show any clinical signs of RP. The degree of mosaicism in this individual was 13% 24 . In contrast, the mosaic individual (II.1) studied here, showed a variable mutation load ranging from 21 to 42% in the different tissues tested (Fig. 5). Indeed, it has been reported that some pathogenic mosaic variants differ in their abundance depending on the tissue, which may be explained partly by differential negative and positive selective pressures in different tissues 21,25,26 . Moreover, a more accurate quantification of mosaicism, even at a lower degree, is possible using deep NGS 27 . Individual II.1 had phenotypical signs of sectorial RP and a clearly less severe phenotype than that observed in her affected sons. It therefore seems that when a certain threshold of mutation load in the retina is met, clinical disease becomes manifest. However, it is not possible to know the exact mutation load in the retina and how this would compare with the mutation load in blood cells or other types of cells. While the mutation reported by Beryozkin et al. has a predicted loss-of-function effect, with a mutation load expected to be proportional to the amount of gene activity, the mutation identified here is known to have a dominant-negative effect, likely resulting in lost gene activity that is presumably higher than the mutation load 11,15,23 .
When counselling mosaicism, the recurrence risk can theoretically be between 0 and 50%, depending on the mutation load in the gametes. In case of somatic mosaicism, the mutation cannot be transmitted to the next generation. In case of germline mosaicism however, a clinically unaffected person can transmit the mutation to an offspring. A combination of both somatic and gonadal mosaicism is also possible. Pre-implantation genetic testing should therefore be discussed as a reproductive option.
In conclusion, we report two male siblings with typical signs of RP and their mother with clinical characteristics highly suggestive for XLRP. While XLRP testing was negative, WES revealed mosaicism of RHO variant c.404G > C p.(Arg135Pro) in the mother and heterozygosity of this variant in her affected sons. This study emphasizes that mosaicism for a dominant mutation should be considered in families suggestive of XLRP but testing negative.