Background Retinitis pigmentosa (RP) shows substantial genetic heterogeneity. It has been estimated that in approximately 60%–80% of RP cases, the genetic diagnosis can be found using whole exome sequencing (WES). In this study, the purpose was to identify causative variants in individuals with genetically unexplained retinal disease, which included one consanguineous family with two affected siblings and one case with RP.
Methods To identify the genetic defect, WES was performed in both probands, and clinical analysis was performed. To obtain insight into the function of KIAA1549 in photoreceptors, mRNA expression, knockdown and protein localisation studies were performed.
Results Through analysis of WES data, based on population allele frequencies, and in silico prediction tools, we identified a homozygous missense variant and a homozygous frameshift variant in KIAA1549 that segregate in two unrelated families. Kiaa1549 was found to localise at the connecting cilium of the photoreceptor cells and the synapses of the mouse retina. Both variants affect the long transcript of KIAA1549, which encodes a 1950 amino acid protein and shows prominent brain expression. The shorter transcript encodes a 734 amino acid protein with a high retinal expression and is affected by the identified missense variant. Strikingly, knockdown of the long transcript also leads to decreased expression of the short transcript likely explaining the non-syndromic retinal phenotype caused by the two variants targeting different transcripts.
Conclusion In conclusion, our results underscore the causality of segregating variants in KIAA1549 for autosomal recessive RP. Moreover, our data indicate that KIAA1549 plays a role in photoreceptor function.
- clinical genetics
- molecular genetics
- vision research
Statistics from Altmetric.com
Retinitis pigmentosa (RP; MIM: #268000) encompasses a clinical and genetic heterogeneous group of progressive inherited retinal dystrophies (IRD). RP is characterised by the primary degeneration of rod photoreceptor cells, followed by the loss of cone photoreceptor cells and retinal pigment epithelium (RPE). With a prevalence of approximately 1 in 4000 persons, it is considered the most common form of IRD.1 RP typically displays night blindness in early adulthood or adolescence, followed by the progressive loss of the peripheral visual field. The visual acuity can be relatively preserved until the advanced disease stages, but RP leads to severe visual impairment or blindness in a large number of patients.2
Besides clinical heterogeneity, RP is also characterised by its broad range of genetic heterogeneity. A large number of genes have been implicated in the pathogenesis of RP, and pathogenic variants can be inherited in a recessive, dominant or X-linked manner (RetNet; see Web Resources). Recently, it has been estimated that in only 60%–80% of RP cases the genetic explanation can be found using whole exome sequencing (WES), which is currently the most widely applied method for disease gene identification.3 4 A better understanding of the underlying disease mechanisms, the role of variants in the pathogenesis of disease in currently known RP genes, and genotype–phenotype correlations are required to provide further insights towards developing therapeutic approaches.
Recently, KIAA1549 (GenBank: NM_001164665; MIM: *613344) has been proposed as a candidate RP gene; however, supporting evidence is limited. In an autosomal recessive RP (arRP) family, a homozygous frameshift variant in KIAA1549 was described to be the only variant remaining after applying filtering criteria on WES data.4
In this study, we report on homozygous variants in KIAA1549 in two families with arRP. In addition, protein localisation studies have been performed to provide insight in the involvement of KIAA1549 in photoreceptor function, supporting its role as an RP gene.
Subjects and clinical examinations
Two families with individuals with genetically unexplained RP were included in this study: one consanguineous Iranian family with two affected siblings and one case from the Netherlands (figure 1A).
This study adhered to the tenets of the Declaration of Helsinki. All subjects provided informed consent prior to inclusion in the study.
Clinical data were collected from the medical records of two patients from family A (A-II:1 and A-II:2) and one patient from family B (B-II:1), including information regarding best-corrected Snellen visual acuity and results of slit-lamp biomicroscopy and ophthalmoscopy. In patient A-II:2 and B-II:1, fundus photography, spectral-domain optical coherence tomography (SD-OCT; Spectralis, Heidelberg Engineering) and Goldmann kinetic perimetry were performed, and full-field electroretinography (ERG) was recorded according to the International Society for Clinical Electrophysiology of Vision guidelines and assessed by applying local standard values.5 In addition, fundus autofluorescence (Spectralis, Heidelberg Engineering) images were available for patient B-II:1.
WES and variant interpretation
Genomic DNA was isolated from peripheral blood using standard isolation methods, and WES was performed in both probands. For proband A-II:2, exome enrichment was performed using the Agilent SureSelect Human All Exome V6 kit. Read mapping along the hg19 reference genome (GrCH37/hg19) and variant calling were performed using BWA V.0.78 and the haplotype caller module of GATK (Broad Institute). CNV detection was performed using CoNIFER V.0.2.2. Exome enrichment for proband B-II:1 was carried out with the Agilent SureSelect XT Human All Exon V5 enrichment kit. Mapping of sequencing reads along the hg19 reference genome and variant calling were performed using Lifescope V.1.3 (Life Technologies). CNV detection was performed using ExomeDepth V.1.1.1.
For both datasets, the obtained variants were filtered based on population allele frequencies ≤0.5% in gnomAD, ExAC, dbSNP and an inhouse exome database (containing 15 576 alleles). Only nonsense, indels, splice site (−14/+14 nucleotides), missense and synonymous variants were assessed. Missense variants were only assessed when predicted to be possibly pathogenic by at least one in silico predictor: a Grantham score ≥80, PhyloP ≥2.7 or CADD-Phred score ≥15. Synonymous variants were only assessed when predicted to have an effect on splicing by one of the splice prediction tools that are embedded in the AlamutVisual software (V.2.10). Candidate genes in which remaining variants were found were compared with currently known IRD-associated genes listed on RetNet (accessed on 1 June 2018). Validation of found variants and segregation analysis were performed by Sanger sequencing. Primer sequences and PCR conditions are available on request.
KIAA1549 expression in human tissues
KIAA1549 expression was determined in human adult tissues using commercially available cDNA panels. Total RNA derived from heart, lung, brain, kidney and bone marrow (Bio-Chain) and total RNA derived from skeletal muscle, liver, duodenum, stomach, spleen, thymus and testis (Stratagene) were used. Total RNA from retina was obtained from a healthy anonymous donor. Subsequently, cDNA was prepared using the iScript cDNA Synthesis kit (Bio-RaD) and purified with NucleoSpin Gel and PCR Clean-up Columns (Machery-Nagel). Quantitative PCR was performed using GoTaq qPCR Master Mix (Promega) according to manufacturer’s protocol. Transcript-specific intron-spanning primers have been designed and validated for the long (NM_001164665) and short transcript (XM_935390) of KIAA1549 and for the reference gene GUSB (MIM: #611499). Primer locations and sequences can be found in online supplementary table S1. Amplifications were performed with the Applied Biosystem Fast 7900 System (Applied Biosystems). All PCR reactions were executed in duplicate, and relative gene expression levels compared with the reference gene GUSB were determined with the delta-delta Ct method.
Supplementary file 2
Immunofluoresence of Kiaa1549 in mouse retinal sections
An eye obtained from a healthy 2-month-old mouse was dissected and cryoprotected for 30 min with 10% sucrose in phosphate buffered saline (PBS) before embedding Tissue-Tek OCT (Sakura). Subsequently, sections were frozen in isopentane cooled by liquid nitrogen. For immunofluorescence, unfixed cryosections (7 µm) were permeabilised in 0.01% Tween20 in PBS for 20 min. After washing with PBS, blocking was performed for 1 hour using a blocking solution containing 0.1% ovalbumin and 0.5% fish gelatin in PBS. Primary antibodies against Kiaa1549 (1:500; cat.# HPA019560, Sigma-Aldrich) and Centrin (1:500; cat.# 04–1624, Millipore) were diluted in blocking solution and incubated on the sections overnight at 4°C. Subsequently, sections were rinsed with PBS and incubated with secondary antibodies goat antirabbit Alexa 568 and goat antimouse Alexa 488 (1:500; Molecular Probes) and DAPI (1:8000; Molecular Probes) in blocking solution for 45 min. Finally, sections were postfixed with 4% paraformaldehyde (PFA) for 10 min before mounting with Prolong Gold (Molecular Probes). Sections were analysed using a Zeiss Axio Imager Z2 fluorescence microscope equipped with an Apotome using several magnifications.
Knockdown of KIAA1549 in vitro using siRNAs
Silencer Select siRNAs targeting KIAA1549 (s33562 and s33563) and non-targeting Negative Control No. 1 were obtained from Thermo Fisher Scientific (online supplementary table S2). For transfection, hTERT-RPE1 or HEK293 cells (ATCC) were transfected with a single siRNA in duplicate (15 nM final concentration), using Lipofectamine RNAiMax transfection reagent (Thermo Fisher Scientific) according to manufacturer’s protocol. After 24 hours of transfection, cells were serum starved (0.2% FCS) for 48 hours to induce ciliogenesis. To assess the effect of the siRNAs on KIAA1549 expression, RNA was isolated using the NucleoSpin RNA kit (Macherey-Nagel), and expression was quantified by qPCR. To evaluate the effect of knockdown of the long transcript on expression of the short transcript, HEK293 cells were used. HEK293 cells express both the long and short transcript abundantly, unlike hTERT-RPE1 cells, which only express the long transcript.
For immunofluorescence, transfected hTERT-RPE1 cells were fixed with 2% PFA for 20 min and permeabilised using 1% Triton X-100 in PBS for 5 min. Subsequently, cells were blocked with 2%bovine serum albumin (BSA) in PBS for 45 min. Primary antibodies against the primary cilium (anti-ARL13B; 1:500; cat.# 17711-1AP; ProteinTech) and the ciliary transition zone (anti-RPGRIP1L; 1:500; cat.# SNC039)6 diluted in blocking solution were incubated for 1 hour. After incubation with secondary antibodies in blocking solution for 45 min, samples were mounted by VECTASHIELD containing DAPI (Vector Laboratories). Cells were imaged using a Zeiss Axio Imager Z2 fluorescence microscope and a 63× magnification. Percentage of ciliated cells and cilium length were calculated using Fiji Is Just ImageJ (Fiji).7 Each experiment was performed three independent times.
Identification of KIAA1549 variants
To identify the genetic defect underlying the arRP in two affected siblings of an Iranian consanguineous family (family A; figure 1A), exome sequencing was performed in individual A-II:2. After analysis of the WES data, the frameshift variant c.52del (Hg19:g.138,665,964del; p.(Arg18Alafs*64)) was detected in the candidate RP gene KIAA1549 (Family A). This variant is located in the second largest homozygous region of 13.2 Mb. Presence of the homozygous variant was confirmed, and segregation analysis was performed using Sanger sequencing. The variant is absent from population frequency databases gnomAD, ExAC, dbSNP and the inhouse database. Moreover, the variant is absent from the Iranome database, which contains WES data of 800 healthy individuals from eight major ethnic groups in Iran. The variant causes a frameshift in exon 1 and is predicted to result in degradation of KIAA1549 mRNA due to nonsense-mediated decay. Previously, a heterozygous variant in CRB1 was reported for family A in both affected siblings.8 Analysis of the WES data of patient A-II:2 did not yield additional variants or copy number variants in this gene. Moreover, the c.2816A>G (p.(Asn894Ser)) variant was predicted to be benign by in silico predictions suggesting that CRB1 variants are unlikely to cause disease in this family. No CNVs or other compound heterozygous or homozygous variants were detected in currently known IRD-associated genes. Also, no heterozygous candidate variants were found in causative genes related to the patient’s phenotype. WES was performed in patient B-II:1 affected with RP (family B, figure 1A) and revealed a homozygous missense variant in KIAA1549; c.4686C>A (Hg19:g.138,554,373G>T; p.(His1562Gln)). After analysis, this was the only homozygous variant remaining in an IRD-associated gene and no compound heterozygous variants were observed. The homozygous variant was validated in the proband, and segregation analysis was performed by Sanger sequencing in two unaffected siblings. The moderately conserved mutated histidine residue is located in a highly conserved region (figure 1B), has a CADD-Phred score of 24.2, a PhyloP score of 0.53 and a Grantham score of 24. Additionally, the prediction tools MutationTaster,9 PolyPhen-210 and SIFT11 predicted the variant to be disease causing (p value: 0.835), possibly damaging (HumDiv: 0.889; HumVar: 0.651) and tolerated (0.17), respectively. Moreover, putative changes are predicted by Human Splicing Finder. The binding sites of the splicing factors SRp40 and SF2/ASF are no longer present, and a potential creation of an exonic splicing silencer site is predicted.12 Heterozygous variants in currently known IRD-associated genes were found in ABCA4 (c.2588G>C(;)5603A>T; p.[Gly863Ala, Gly863del](;)(Asn1868Ile) and CDHR1 (c.512C>G; p.(Thr171Ser)), but no second pathogenic alleles could be detected for these genes. No CNVs were detected in regions overlapping with known IRD-associated genes.
Clinical data were collected from the medical records of two patients from family A (A-II:1, A-II:2) and one patient from family B (B-II:1). An overview of the clinical characteristics of the three affected individuals with damaging KIAA1549 variants at the most recent examination is provided in table 1, and clinical images of patient A-II:2 and B-II:1 are shown in figure 2. All affected individuals were diagnosed with RP. They all initially experienced night blindness, followed by a gradual decline of their visual fields and visual acuity. The age of onset varied from the first decade (patient A-II:1) to the fifth decade (patient B-II:1), and all patients were myopic. Cortical cataract was observed in patient A-II:1 (age 38 years), whereas patient A-II:2 underwent a cataract extraction at the age of 38 years (right eye) and 52 years (left eye). Ophthalmoscopy revealed characteristic RP features in all three patients, including attenuated retinal vessels, waxy pallor of the optic disc and bone spicule pigmentation (figure 2A,D). In addition, nummular deep pigmentations were visible in the midperiphery. SD-OCT imaging in patient A-II:2 showed profound atrophy of the outer retinal layers with preservation of the photoreceptors in the fovea. This patient was treated for Coats-like exudative vasculopathy related to her RP in the past. Fundus autofluorescense imaging in patient B-II:1 showed the characteristic hyperautofluorescent ring that represents the transition zone between intact and degenerated photoreceptor outer segments, corresponding with a preserved ellipsoid zone within the ring on SD-OCT and loss of the ellipsoid zone external to the ring (figure 2E,F). In addition, the SD-OCT image of patient B-II:1 showed evident cystoid macular oedema that was refractory to topical treatment with non-steroidal anti-Inflammatory drugs, steroids, as well as both topical and oral carbonic anhydrase inhibitor treatments (figure 2F). Electrophysiology examination demonstrated a generalised retinal dystrophy with non-recordable rod and cone-driven responses in patient A-II:2 at the age of 32 years and severely reduced rod and cone-driven responses in patient B-II:1 at 54 years of age. Finally, perimetric analysis revealed a severely constricted visual field up to 5° in patient A-II:2 and a complete and partial ring scotoma in the right and left eye of patient B-II:1, respectively. No visual field testing was performed in patient A-II:1, yet he reported severe visual field constriction. All patients were in good general health, and no non-ocular conditions were reported.
Expression of KIAA1549 transcripts in human tissues
To gain knowledge on the specific role of KIAA1549, its relative expression was determined in a set of human adult tissues. Two major KIAA1549 isoforms have been identified (Uniprot: Q9HCM3), a long primary isoform (NM_001164665) of 1950 amino acids (aa) and a short isoform (XM_935390) of 734 aa (figure 3A), which is produced from an alternative transcript transcribed from an alternative promoter sequence located in intron 8. The nomenclature of all genetic or protein elements is based on the long isoform. The expression of both KIAA1549 transcripts and of the reference gene GUSB was evaluated in cDNA of human tissues by qPCR (figure 3B). The long transcript showed a low to moderate expression in retina and other tissues, such as heart and kidney, and is predominantly expressed in brain as has been previously described.13 On the contrary, the less characterised short transcript showed an abundant expression in the retina (~200 times; normalised for GUSB), compared with only minimal expression in brain and other tissues. This suggests that the short KIAA1549 isoform may harbour a retina-specific function.
Localisation of Kiaa1549 in mouse retina sections
To confirm the presence of KIAA1549 in the retina, as well as to define its specific localisation in this tissue, immunofluorescence was performed in retina sections obtained from a healthy 2-month-old mouse. Costaining was performed with anti-Centrin, a well-defined marker for the connecting cilium within the photoreceptor cell.14 Results showed that Kiaa1549 colocalised with Centrin, and thus is located at the connecting cilium of the photoreceptor cells (figure 4). Moreover, positive staining of Kiaa1549 was also observed at the outer plexiform layer of the mouse retina. This layer contains neural synapses between the photoreceptors and the bipolar and horizontal cells in the retina.
Assessing the function of KIAA1549
To investigate whether KIAA1549 plays a role in ciliogenesis, an in vitro study was performed in which KIAA1549 was knocked down in hTERT-RPE1 cells with two different siRNAs targeting the long transcript. The efficiency of KIAA1549 knockdown was validated by qPCR analysis, and induced ciliogenesis was studied using immunocytochemistry. The percentage of ciliated cells and average cilium length showed no significant difference when comparing cells treated with KIAA1549-targeting siRNAs or the non-targeting siRNA (online supplementary figure S1), suggesting that KIAA1549 does not have a function in the formation of cilia; however, KIAA1549 may be involved in other processes that are performed at the primary cilium.
Supplementary file 1
In this study, we report on two families in which arRP is associated with homozygous variants in KIAA1549. The retinal phenotype in both families is typical for RP, and patients’ complaints started with night blindness with subsequent constriction of the visual field. Fundus examination revealed the hallmark RP features. However, patients in family A are more severely affected compared with the patient in family B, which is displayed in a lower age at onset, severely constricted visual fields and more severely reduced ERG responses.
In family A, a homozygous frameshift variant was found in exon 1 (c.52del; p.(Arg18Alafs*64)) by WES. Although putative alternative start codons are present in exon 2, exon 1 encodes the signal peptide of the protein (aa 1–60). Therefore, a shorter protein is potentially mislocalised, impairing protein function. In family B, a homozygous missense variant was found in exon 14 (c.4686C>A; p.(His1562Gln)), which affects a highly conserved region of the protein. The damaging nature of these variants is supported by a probability of loss of function intolerance score of 1.00 (Scale 0–1) in gnomAD (accessed on 1 June 2018) and that no homozygous variants have been reported in the entire KIAA1549 gene. Combined, this suggests that the identified KIAA1549 variants in both families can be associated to the RP phenotype of the patients. Regardless, the presence of pathogenic variants present in non-coding regions uncovered by WES cannot be ruled out.
KIAA1549 encodes a transmembrane protein and is described to be predominantly expressed in the brain and is involved in oncogenesis when fused to BRAF (MIM: *164757).4 15 BRAF-KIAA1549 inframe fusion genes are caused by a 2 Mb tandem duplication at 7q34 and are found to induce BRAF kinase activity and consequently, activation of the MAPK pathway, which is involved in the development of cancer. For this reason, these fusion genes are the major cause (66%) for pilocytic astrocytomas, the most frequently occurring central nervous system tumour in children and young adolescents.
Besides this role in oncogenesis, knowledge about the function of KIAA1549 is limited. Recently, a homozygous truncating variant in KIAA1549 was found in an arRP family with two affected siblings in a study performed by Abu-Safieh et al.4 Involvement of KIAA1549 in photoreceptor function was suggested; however, no functional data were provided.4 Nevertheless, KIAA1549 is reported to be among the top 4% of genes being enriched for binding sites for the photoreceptor specific transcription factor CRX.16 Kiaa1549 expression was evaluated in a Nrl−/− knockout mouse that is characterised by degenerated rod photoreceptors. In this mouse, Kiaa1549 expression was found to be reduced ~88% (wildtype (WT): 106 reads, knockout (KO): 13 reads) when compared with the wild-type mouse, based on number of sequencing reads.16
In this study, expression levels of the major short and long transcripts of KIAA1549 have been evaluated in a set of human tissues, which demonstrated that both isoforms are present in the retina, of which the expression of the transcript encoding the short isoform is significantly higher in retina compared with other tissues. We hypothesise that both isoforms are required for the correct function of the protein in the retina, as the homozygous frameshift variant affecting the long isoform has detrimental consequences as observed in family A and the family previously described in the study of Abu-Safieh et al. By performing an in vitro experiment in which HEK293 cells were transfected with KIAA1549-targeting siRNAs that specifically recognise the long transcript of KIAA1549 (online supplementary table S1), also a significant decrease in expression of the short transcript was observed (online supplementary Figure S3), which suggests a functional dependency between the two transcripts. Hence, observed variants in the long transcript likely cause a decrease in the abundant retinal expression of the short transcript and thereby could lead to retinal degeneration. The fact that the identified variants have different consequences on the two KIAA1549 transcripts could explain the phenotypic differences observed among the affected individuals. The phenotype of the family described by Abu-Safieh et al (a non-recordable ERG at age 35 years) (personal communication Professor F S Alkuraya and N Patel, PhD) is more comparable with family A (Non-recordable ERG at age 32 years in patient A-II:2) than family B (severely reduced photopic and moderately reduced scotopic ERG at age 54 years), which may be in line with the genotype having a damaging variant in the long transcript. Identification of additional families with KIAA1549-associated RP are required to provide deeper insight into a possible phenotype–genotype correlation.17
Supplementary file 3
In addition, we showed localisation of KIAA1549 at the connecting cilium of mouse photoreceptor cells, providing the first information on KIAA1549 function in photoreceptors. Moreover, KIAA1549 localisation was also noted at the outer plexiform layer of the mouse retina. Proteins localised at the ribbon synapses of the outer plexiform layer are often structural or synaptic vesicle proteins or are involved synaptic vesicle trafficking.18 19 The KIAA1549 antibody will recognise both isoforms and thus does not provide additional knowledge on alternative localisation of the isoforms. Hypothetically, the long and short isoforms may harbour a unique function at either one of the identified locations. Additional research is required to unravel the functional differences between the short isoform and the ubiquitously expressed long isoform of KIAA1549.
Besides localisation in the photoreceptor, there is additional evidence for ciliary function is at the molecular level. A recent study based on proximity-dependent biotinylation revealed an interaction between KIAA1549 and TMEM17 (MIM: *614950).20 TMEM17 is a part of the Meckel syndrome (MKS) protein complex located in the ciliary transition zone, in which it facilitates cilium formation. Also, pathogenic variants in genes encoding proteins in this complex are known to cause (severe) ciliopathies.21 TMEM17 pathogenic variants have been reported to cause oral-facial-digital syndrome type 6 (MIM: #277170).21 The MKS complex contains both cytoplasmic and transmembrane proteins and functions as a barrier preventing rapid diffusion of transmembrane proteins between cilia and plasma membranes.22 The interaction between KIAA1549 and TMEM17 was only observed in cells in non-ciliated conditions, which suggests that the interaction is involved in a cilium-related process.13
We have studied the role of KIAA1549 in ciliogenesis, by knocking down the expression of the gene in hTERT-RPE1 cells using siRNAs. siRNA-transfected cells did not show a difference in percentage of ciliated cells or ciliumlength, suggesting that KIAA1549 does not have a direct role in the cilium formation explaining the non-syndromic phenotype observed in the patients of family A and B, as well as the family of Abu-Safieh et al, which is restricted to the retina. Pathogenic variants that do affect genes essential for ciliogenesis, such as TMEM17, would give rise to a phenotype likely affecting multiple organs as in ciliopathies. Transmembrane proteins present at the transition zone are often involved in the sensing and transducing of extracellular signals. Like TMEM17, KIAA1549 is a transmembrane protein; therefore, it is plausible that KIAA1549 may be involved in these processes at the primary cilium of the photoreceptors specifically.22
In conclusion, by employing WES, we have identified that homozygous frameshift or missense variants in KIAA1549 are associated with RP in two families. We demonstrated retina-specific expression of the short isoform of KIAA1549 and provide evidence that damaging variants targeting the long transcript may cause RP by reducing the expression of the short transcript. Moreover, we showed that KIAA1549 resided in the connecting cilium of the mouse retina, thereby providing supporting evidence that KIAA1549 might act as an essential photoreceptor protein.
AlamutVisual version 2.10, http://www.interactive-biosoftware.com/alamut- visual/
BWA version 0.78 , https://bio-bwa.sourceforge.net/
CoNIFER version 0.2.2, https://conifer.sourceforge.net
ExomeDepth version 1.1.10, https://cran.r-project.org/web/packages
HaplotypeCaller GATK, https://www.broadinstitute.org/gatk/
Human Splicing Finder V3, https://umd.be/HSF3/
MutationTaster, https://www.mutationtaster.org OMIM, https://www.omim.org/
We would like to thank Theo A Peters, Sanne Broekman, Nisha Patel, Fozwan S Alkuraya, Thanh- Minh T Nguyen and Maartje van de Vorst for expert technical assistance.
Contributors SEdB performed sequencing analyses, RNA expression and protein localisation studies. SKV, CBH and LIvdB collected clinical cases and performed clinical examinations of patients. SEdB, EdV, HK, FPMC and SR contributed significantly to design of the study. SEdB, SKV and SR wrote the manuscript. All authors reviewed and approved the manuscript.
Funding The study was financially supported by DCN Radboudumc grant (to FPMC and HK), as well as the Rotterdamse Stichting Blindenbelangen, the Stichting Blindenhulp, the Stichting tot Verbetering van het Lot der Blinden and the Stichting Blinden-Penning (to FPMC and SR)
Competing interests None declared.
Patient consent Not required.
Ethics approval This study was approved by the Institutional Review Boards of the participating centres.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.