Genetic and Clinical Findings in an Ethnically Diverse Cohort with Retinitis Pigmentosa Associated with Pathogenic Variants in CERKL

Autosomal recessive retinitis pigmentosa is caused by mutations in over 40 genes, one of which is the ceramide kinase-like gene (CERKL). We present a case series of six patients from six unrelated families diagnosed with inherited retinal dystrophies (IRD) and with two variants in CERKL recruited from a multi-ethnic British population. A retrospective review of clinical data in these patients was performed and included colour fundus photography, fundus autofluorescence (AF) imaging, spectral domain–optical coherence tomography (SD–OCT), visual fields and electroretinogram (ERG) assessment where available. Three female and three male patients were included. Age at onset ranged from 7 years old to 45 years, with three presenting in their 20s and two presenting in their 40s. All but one had central visual loss as one of their main presenting symptoms. Four patients had features of retinitis pigmentosa with significant variation in severity and extent of disease, and two patients had no pigment deposition with only macular involvement clinically. Seven variants in CERKL were identified, of which three are novel. The inherited retinopathies associated with the CERKL gene vary in age at presentation and in degree of severity, but generally are characterised by a central visual impairment early on.


Introduction
Inherited retinal dystrophies (IRD) are a heterogeneous group of disorders associated with the dysfunction or death of photoreceptors, resulting in varying severity of visual loss. IRD has an incidence of 1 in 2000-3000, affecting an estimated two million people worldwide [1]. Retinitis pigmentosa (RP; Mendelian Inheritance in Man (MIM) #268000) is the most common IRD, affecting approximately 1 in 4000 people [2], characterised by symptoms of nyctalopia, peripheral visual field loss with a clinical appearance of intraretinal bone spicule pigmentation, pale discs and attenuated vasculature. RP is associated with significant genotypic and phenotypic heterogeneity, with 41 genes described in cM region on chromosome 2q31-32 in a consanguineous Spanish family by Bayes and colleagues [3]. Subsequently, Tuson and colleagues identified a novel gene from the region that was expressed in the retina, named ceramide kinase-like (CERKL) [4]. The coding exons of CERKL were sequenced in the original RP26 family, and all patients were found to be homozygous for a nonsense mutation c.769C > T, p.Arg257* (now annotated as c.847C > T, p.Arg283*, using the numbering based on transcript NM_001030311, which consists of 14 exons) [4]. A further unrelated Spanish family was also identified with the same homozygous change [4].
The CERKL gene is composed of 14 exons but alternative splicing produces multiple transcripts ( Figure 1) [8,9]. Most of the previous studies examining variants in CERKL used isoform NM_201548, but this only contains 13 exons and is missing exon 5. Table 1 lists all the previously published mutations and the novel changes described in the patients in the present study, using NM_001030311 (which consists of 14 coding exons) as the reference. To date, 39 different mutations in CERKL have been identified (see Figure 1 and Table 1 for summary). The majority of these CERKL-associated IRD studies are case reports and many include the results from next-generation sequencing (NGS) testing with minimal phenotype data. There are, however, four studies reporting on at least four families, but these pertain to one ethnicity only in each study (Yemenite Jewish, Spanish, Tunisian and Finnish) [10][11][12][13]. In these reported populations, CERKL is a significant gene contributing to autosomal recessive RP; presumably due to a founder mutation effect.
The development of next-generation sequencing (NGS) techniques over the past 20 years led to diagnostic laboratories having the capability of screening known RP genes simultaneously [14]. Although this increased the detection rate, enabling a comprehensive genotype/phenotype analysis of cohorts of patients with same gene variants, it also identifies variants in other genes whose relevance may be difficult to determine. We report phenotypic and genotypic characterisation of six patients from six unrelated families with different ethnicities, and report three novel and four previously described variants in CERKL. The CERKL gene on 2q31-32 consists of 14 exons and spans approximately 12 kb of genomic DNA. Exons are shown as boxes and introns as lines; all are to scale. The position of all the CERKL variants reported to date are marked. The mutations identified in our cohort are marked, those novel to this study are shown below in red and those reported previously are highlighted in bold above the structure (see Table 1 for details). The development of next-generation sequencing (NGS) techniques over the past 20 years led to diagnostic laboratories having the capability of screening known RP genes simultaneously [14]. Although this increased the detection rate, enabling a comprehensive genotype/phenotype analysis of cohorts of patients with same gene variants, it also identifies variants in other genes whose relevance may be difficult to determine. We report phenotypic and genotypic characterisation of six patients from six unrelated families with different ethnicities, and report three novel and four previously described variants in CERKL.
The CERKL gene on 2q31-32 consists of 14 exons and spans approximately 12 kb of genomic DNA. Exons are shown as boxes and introns as lines; all are to scale. The position of all the CERKL variants reported to date are marked. The mutations identified in our cohort are marked, those novel to this study are shown below in red and those reported previously are highlighted in bold above the structure (see Table 1 for details).  [17,18] c.238 + 1G > A Intron 1 Splicing defect [11,19] Whole exon deletion Exons 1 and 2 Loss of function [20] Whole exon deletion Exon 2 Loss of function [17] c.316C > T Exon 2 CGT > TGT p.(Arg106Cys) [10,16]

Materials and Methods
The clinical notes of patients seen at the Oxford Eye Hospital, identified to have CERKL variants by the Oxford Medical Genetics Laboratory, were reviewed to obtain full ophthalmic details, history of symptom onset and family pedigrees. The study adhered to the tenets of the Declaration of Helsinki and was approved by the Central Oxford Research Ethics Committee and the Research and Development Department of the Oxford Radcliffe Hospitals NHS Trust (RetGene 08/H0302/96). Informed consent was obtained from all participants.

Clinical Studies
Clinical data collected included best corrected visual acuity, slit-lamp biomicroscopy, digital colour fundus photography, short-wavelength (488) fundus autofluorescence (Spectralis; Heidelberg Engineering, Heidelberg, Germany), spectral domain-optical coherence tomography (SD-OCT; Spectralis, Heidelberg Engineering, Heidelberg, Germany), Goldmann visual field analysis and electrodiagnostic testing, where possible. The Heidelberg software was used for retinal segmentation of the ganglion cell layer (GC) and the inner plexiform layer (IPL), which were then used to measure the thickness of the GC-IPL layers in the areas temporal and nasal to the area of macular atrophy.

CERKL Mutation Screening
CERKL variants were identified by targeted sequencing performed by the Oxford Medical Genetics Laboratory using their next generation sequencing (NGS) phenotype panels. Enrichment for the CERKL gene was achieved as part of a customised HaloPlex enrichment system kit (Agilent Technologies, Santa Clara, USA) designed to capture the coding exons and at least 10 bp of the flanking introns of 111 retinal genes in the retinal dystrophy panel [14]. HaloPlex reactions were prepared as per the manufacturer's instructions. Libraries were pooled into batches of 14 and sequenced on an Illumina MiSeq instrument (Illumina) using a MiSeq v3 kit (Illumina, San Diego, USA), as per the manufacturer's instructions. Reads were aligned using Burrow-Wheeler Aligner BWA [15] and variants were called using Platypus [49]. All variants identified by NGS were confirmed by Sanger sequencing and investigated according to the Association of Clinical Genetic Science guidelines [50]. This included filtering of candidate variants by minor allele frequency against the ExAC dataset [51].
Chromosome 2 position was based on build GRCh37/hg19 and nucleotide and protein numbering is based on CERKL transcript NM_001030311, comprising 14 exons. The Genome Aggregation Database (gnomAD) (available online: http://gnomad.broadinstitute.org, accessed on 1 July 2020) was used to determine the carrier frequency of the minor allele in the general population. In silico analysis using 3 different prediction methods to determine the deleteriousness of the variants, Polyphen2 [52] Sorting Intolerant from Tolerance (SIFT) [53] and Mutation Taster [54] was carried out on the variants identified. Variants in other genes in these patients identified by this methodology were also reported, and family segregation studies were performed where possible.

Clinical Analysis
The clinical data of six patients from six unrelated families ( Figure 2) with a diagnosis of a retinal dystrophy and with at least two variants in CERKL were reviewed ( Figure 3, Tables 2 and 3). The study comprised three females and three males, with age at presentation ranging from childhood to mid-40s. Symptom type and onset and clinical phenotypes are summarised in Table 2. Autofluorescence (AF) and optical coherence tomography (OCT) imaging are summarised in Table 3 and seen in Figure 3. to mid-40s. Symptom type and onset and clinical phenotypes are summarised in Table 2. Autofluorescence (AF) and optical coherence tomography (OCT) imaging are summarised in Table 3 and seen in Figure 3.    Table 2; Table 3 for further details. The thickness of the ganglion cell-inner plexiform layer (GC-IPL) layers in the areas temporal and nasal to the area of macular atrophy were also measured (see Table S1.  Tables 2 and 3 for further details. The thickness of the ganglion cell-inner plexiform layer (GC-IPL) layers in the areas temporal and nasal to the area of macular atrophy were also measured (see Table S1).   He was fit and well other than a diagnosis of Marfan Syndrome. There was no family history of eye disease. Examination revealed visual acuities of 6/24 in the right and hand movements only in the left. Fundoscopy and imaging demonstrated bilateral attenuated vessels, intraretinal bone spicule pigmentation and patches of peripheral atrophy with central macular atrophy. AF imaging revealed significant central atrophy with foveal sparing. Patchy AF signal was noted in the mid-to far-periphery. OCT imaging showed significant central atrophy bilaterally (Figure 3, Tables 2 and 3). Visual field testing and electrophysiology testing were not available.  Tables 2 and 3). Goldmann visual fields showed central reduction in sensitivity extending to 20 degrees on the right and left bilaterally (see Figure S1). Electrophysiology showed extinguished PERG, reduction in a-and b-wave amplitudes on the rod ERG and reduction in b-wave amplitude for the standard bright white flash. Cone responses showed reduction in amplitudes and the 30 Hz flicker was delayed and reduced.

Patient C
A 56-year-old white Caucasian male, compound heterozygous for c.847C > T, p.(Arg283*) and c.566_569delinsGTG, p.(Lys189Serfs*5) CERKL mutations, first noticed symptoms of nyctalopia and visual loss aged 25 years. Regarding family history, his maternal great uncle was said to be visually impaired but the aetiology of this is unknown. Examination revealed visual acuities of no perception of light in both eyes. Fundoscopy and imaging revealed pale discs, grossly attenuated vasculature and central macular atrophy. Widespread intraretinal bone spicule pigmentation was noted primarily in the mid-periphery, and patchy nummular atrophic and hyper pigmented patches were observed in the mid-periphery to periphery mainly in the temporal retina. AF imaging demonstrated loss-of-signal consistent with central atrophy with a strip of minimal central sparing. There were also peripheral areas of loss-of-signal consistent with the chorioretinal atrophic patches. Fundus fluorescein angiography clearly demonstrated the central and inferior retinal patches of total retinal loss (see Figure S2). OCT imaging revealed loss of outer retina bilaterally (Figure 3, Tables 2 and 3). Visual field testing was not performed due to poor visual acuity and electrophysiology was not available.  Tables 2 and 3). Goldmann visual fields showed bilateral constriction to less than 10 degrees (see Figure S1). Electrophysiology testing showed no measurable components for PERG, cone and 30 Hz ERG, rod and standard flash and maximal ERG.

Patient E
A 58-year-old Indian male, homozygous for a c.1045_1046delAT, p.(Met349Valfs*20) mutation in CERKL, noted symptoms of nyctalopia and a slow gradual deterioration in vision in his teens, with a sharp deterioration in the last five years. His brother is also affected by the same condition (see Figure 2). Examination revealed visual acuities of perception of light bilaterally. Fundoscopy and imaging demonstrated significantly attenuated vessels, pale discs, and dense intraretinal bone spicule pigmentation scattered throughout the fundus, with central atrophy. AF imaging showed complete loss-of-signal in patches throughout the fundi, indicating severe coalescing atrophy. OCT imaging revealed loss of outer retina bilaterally (Figure 3, Tables 2 and 3). Visual field testing was not possible due to poor visual acuity. Electrophysiology testing was not available.  Tables 2 and 3). Neither visual fields nor electrophysiology were available.

Genetic Analysis
The six patients described here all had two disease-causing variants in CERKL (Figure 1). Three were homozygotes and three were compound heterozygotes ( Table 2). Seven variants in CERKL were identified in total, three of which were novel ( Figure 1, and Table 1). The most common mutation in our cohort was the nonsense variant c.847C > T, p.(Arg283*) in exon 6, which was identified in three of our patients (Patient A is homozygous and Patients B and C are both heterozygous). This is also the most common CERKL variant reported to date (Table 1). Another previously described nonsense mutation (c.1090C > T p.(Arg364*)) was found in Patient B [22,46]. We also found two missense mutations, one of which was novel ( Figure 1, Table 1). In silico analysis was performed using three different prediction methods to determine the deleteriousness of the novel variant. Two of them, SIFT [53], and Mutation Taster [54], predicted the c.1393C > T variant to be damaging and disease-causing, respectively. However, Polyphen2 [52] predicted it to be benign. The remaining variants were a previously reported deletion, a novel indel and a novel inversion. Patient C was a compound heterozygote with c.847C > T p.(Arg283*) and the novel c.566_569delinsGTG, resulting in a frameshift and premature termination of the protein (p. (Lys189Serfs*5)). Patient D and Patient E were homozygous for a novel inversion and a novel indel respectively, the inversion (c. 1617-16_1630inv30) resulting in a 30bp inversion in exon 14, causing the splice acceptor site to be lost. The novel indel in Patient E causes a frameshift in exon 8, resulting in a premature termination of the protein.
The mutations published to date, including the novel variants reported in this study, are illustrated schematically in Figure 1, demonstrating no significant clustering and they are located throughout the entire gene.
As the patients were screened using either a panel of 55 or 111 IRD genes, variants in other genes were also described in three of them (Table S2). Patient D was heterozygous for a PDE6A variant (c. 769C > T p.(Arg257*)) of known pathogenicity and patient E was heterozygous for an IMPG2 variant (c. 789C > G p.(Ser263Arg)) of uncertain pathogenicity. Patient F was heterozygous for USH2A c.3812-3_3837dup p.(Met1280*) of presumed pathogenicity, as well as for NRL c.11c > T p(Pro4Leu) of uncertain pathogenicity; these are discussed in more detail below (Table S2).

Discussion
This study of six patients from different families with pathogenic variants in CERKL is one of the largest multi-ethnic studies describing CERKL-associated IRD to date. Of the seven different pathogenic variants, three are novel. The phenotype, although variable and with different ages of onset, shows significant similarity in symptom presentation, with five of the six reporting a deterioration in central vision. Three patients were affected in their twenties and two in their forties, similar to the literature. Patient E presented at the age of seven and demonstrated the most severe phenotype. However, in the Finnish cohort described by Avela et al. the early-onset patients did not always have severe phenotypes [10]. Patients A, B, C and F showed clear central atrophy with varying degrees of a residual foveal sparing. These features were seen but less clearly demarcated in patient D, where the infrared imaging showed the central atrophic changes more distinctly and could just be an earlier stage of the same phenotype. Indeed, the phenotype of rod cone dystrophy with initially preserved central vision despite earlier macular involvement was described clearly by Khan and Abu-Safieh, who reported an almost identical phenotype [44]. Also, the peripheral nummular chorioretinal atrophic areas seen in patient D were described by Fernandez et al. in their seven patients with the c.769C > T p.(Arg257*) variant [13]. However, for patient E the changes were so advanced that, although there was a macular atrophic area, the same phenotype was not distinguishable.
For patient F, the ring of AF surrounding the atrophic change was different in its extent compared to the rest of this cohort. However, a similar phenotype was described in CERKL IRD before in a patient compound heterozygous for c.375C > G, and c.193G > T in Avela et al.'s Finnish CERKL cohort (patient 4 in Avela et al.) [10].
As described in other reports, peripheral involvement was variable, ranging from mild pigment clumping to severe widespread atrophy, as seen in patient E. Two of our cohort (B and F) had no visible peripheral changes at all.
Patients D, E and F also possess variants in other known IRD genes (Table S2), but none of these variants were considered to be disease-causing, either because they were seen in a heterozygous state in genes causing autosomal recessive disease or because the phenotype previously described for the gene was not consistent. For example, Patient D was heterozygous for a PDE6A variant (c.769C > T p.(Arg257*)) of known pathogenicity and patient E was heterozygous for an IMPG2 variant (c.789C > G p.(Ser263Arg)) of uncertain pathogenicity, but in both cases a second variant was not found (Table S2). Additionally, segregation studies for patient E identified the same CERKL variant (p.(Met349Valfs*20)) in the heterozygous state in his unaffected father. His unaffected mother is deceased and so was unavailable for analysis. Patient F was heterozygous for variants in USH2A and NRL (Table S2). The USH2A, c.3812-3_3837dup p.(Met1280*) is of presumed pathogenicity, however, the phenotype is not typical of an USH2A-related retinopathy and patient F is not deaf (although not all patients with USH2A variants have hearing impairment). The NRL c.11C > T p(Pro4Leu) variant is of uncertain pathogenicity and NRL-associated RP recessive or dominant is rare, and usually associated with a severe early-onset phenotype, which is not present in patient F, who was 40 years old at the age of diagnosis [55]. In addition, although only the segregation of the CERKL variants were investigated, her affected brother possesses the same two CERKL variants, whereas her unaffected brother has neither.
Herein, we describe a cohort of six ethnically diverse patients, including white Caucasian, Indian, Pakistani, and Kashmiri origins. Previous reports in Finnish [10], Yemeni Jewish [11] and Spanish populations [13] reported the CERKL variants (c.375C > G, p.(Cys125Trp); c.238 + 1G > A and c.847C > T, p.(Arg283*) respectively) as a significant cause of arRP due to founder mutations. Three of our cohort, which did not include any patients from these populations, did have the c.847C > T variant found in the Spanish population. However, in other populations, CERKL variants are a rare cause of arRP, so there are limited phenotypic data in the literature [10,43]. The CERKL phenotype reported was variable in severity, degree of field loss and presence or lack of abnormal pigmentation. However there appeared to be a typical phenotype of early macular involvement with progression over time with concurrent varying degrees of progressive photoreceptor degeneration. The macular atrophy, seen particularly clearly on autofluorescence imaging, may be associated with a preserved small residual island of tissue [36]. As yet, there is no clear explanation for its persistence. It is also observed in some other IRDs, such as ABCA4 retinopathies; Bax et al. made the observation that these patients were in their fifth decade and that foveal sparing is observed in patients with a low rate of progression [56]. The Finnish cohort described by Avela et al. demonstrates a wide range of severity, as also seen in our cohort. Recently, Yu et al. demonstrated progressive degeneration of rod and cone outer segments in a Cerkl-knockout zebrafish model, with rod degeneration preceding cone degeneration [57]. There was no clearly observed genotype phenotype correlation, either in our cohort or in the literature, in patients with pathogenic variants in CERKL. The variants described to date (Table 1) are distributed across the entire gene (Figure 1), and the exact function of CERKL remains unclear. Although it shows 50% similarity to the ceramide kinase (CERK) protein, which phosphorylates ceramide to ceramide 1-phosphate and encodes a potential diacylglycerol kinase (DAG) domain, there is no evidence of any kinase activity in CERKL [6]. Indeed, studies examining the transcriptional complexity of CERKL in the retina demonstrate that there are isoforms that do not contain the DAG domain [8,9]. There is some evidence for CERKL having a role in protecting retinal cells from injury caused by oxidative stress [9,58]. A recent study describing the generation of a mouse model using CRISPR-Cas9 editing to delete the Cerkl locus showed that total ablation of the locus was embryonically lethal [7]. The authors therefore generated a model where the knockout allele (Cerkl KO ) was in trans with a knockdown allele (Cerkl KD ). These animals showed defects in the photoreceptor outer segments but did not initially show any alteration in the electrophysiological response, although older mice did show some changes [7]. This model may prove to be useful in gaining some understanding of the function of CERKL/Cerkl in the mammalian retina.
To date, there are no therapeutic strategies to correct or treat the retinal degeneration caused by mutations in CERKL. However, gene therapy for CERKL using adeno-associated viral (AAV) vectors is a possibility due to a combination of the recessive nature of the disease and small coding sequence at 1599 base pairs, which is easily encodable by adeno-associated viral (AAV) vectors similar to Luxturna, which is approved for Leber's Congenital Amaurosis (LCA) type II. One problem, however, is that CERKL may also have a vital role in the inner retina [4], but in this series we did not identify any significant inner retinal changes. Identifying the correct mRNA sequence will be essential in developing future gene therapy treatments, but CERKL has many known splice isoforms [8]. Hence, our observation of novel mutations in exons 2, 3, 6 and 8 confirms the likely critical role of these exons in the CERKL isoform required in photoreceptors.

Conclusions
In conclusion, CERKL mutations are an uncommon cause of arRP, but they are a significant cause of disease in populations with founder mutations [10,11,13]. The importance of seeking further information from family segregation studies is highlighted by the families described in this study. Identifying the correct genetic variant will be of significant importance when potential therapies become available. It is highly likely that, in some cases, revisiting the genotype, particularly in cases where more than one IRD gene variant is identified, will be required. This study adds to the phenotype spectrum of pathogenic variants in CERKL and highlights specific difficulties in diagnosis when more than one gene variant is present.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4425/11/12/1497/s1, Figure S1: Goldmann visual fields showing two timepoints in patients B and D; Figure S2: Fundus fluorescein angiography; Table S1: The thickness of the ganglion cell layer-inner plexiform layer in the nasal and temporal retina; Table S2: Variants identified in IRD genes other than CERKL.