Genetic screening of LCA in Belgium: predominance of CEP290 and identification of potential modifier alleles in AHI1 of CEP290-related phenotypes

1Leber Congenital Amaurosis (LCA; MIM# 204000) was first described as a congenital type of retinitis pigmentosa (RP). Approximately 20% of all blind children are thought to suffer from this disease. Phenotypic features include a congenital onset, severely reduced or absent electroretinogram (ERG), nystagmus, the oculo-digital sign and a fundus aspect varying from normal to severely atrophic. Two main types of LCA have been reported, based on the presence or absence of photophobia, night blindness, hyperopia, macular/peripheral retinal abnormalities and measurable visual acuity. LCA displays variable expression, and seems to represent the extreme and severe end of a spectrum of inherited retinal disease.

2LCA is predominantly inherited in an autosomal recessive manner. So far, one locus – LCA9 – and the following 14 genes have been identified: GUCY2D, RPE65, CRX, AIPL1, RPGRIP1, CRB1, RDH12, IMPDH1, CEP290, RD3, LCA5 and SPATA7, with the involvement of TULP1 and LRAT under debate. Mutations in these genes account for ∼70% of all LCA cases. Several of them are also implicated in other retinal dystrophies: CRB1, RPE65, RDH12and SPATA7are associated with both LCA and early-onset retinal dystrophy (EORD), which often overlap.

3Several subtypes of LCA can be considered part of the ciliopathies, as four disease genes – TULP1, RPGRIP1, CEP290 and LCA5 – encode ciliary proteins. Since cilia are present throughout the whole body, mutations in ciliary genes may cause a broad phenotypic spectrum. One of the best examples is CEP290, the most frequently mutated gene in the western European LCA population. In addition to LCA, CEP290 is associated with Joubert syndrome (JS; MIM# 213300), Senior-Loken syndrome (SLS; MIM# 266900), Meckel-Grüber syndrome (MKS; MIM# 249000) and Bardet-Biedl syndrome (BBS; MIM# 209900); a range of clinically and genetically heterogeneous ciliopathies. Recent studies suggest that modifiers may play a role in the pathogenesis of ciliopathies.

4Establishing a molecular diagnosis for LCA is not only important in the context of genetic counselling and clinical prognosis, but is also essential in view of future gene therapy. Recent Phase I clinical trials for RPE65 gene replacement therapy provide hopeful prospects for the treatment of inherited retinal dystrophies. As such therapies are likely to be gene-specific, the development of robust clinical testing and efforts toward gene identification are of utmost importance.

5Current diagnostic testing for LCA generally involves chip analysis that contains known mutations in all known LCA and EORD genes (Asper Ophthalmics, Estonia). Depending on the population, causal mutations are found in approximately 55% of all cases. Only a limited number of laboratories subsequently screen an additional number of genes.

6This study includes an extensive genetic survey in order to identify the molecular cause in 91 LCA probands mainly of Belgian origin, using LCA chip analysis for 8 to 13 genes and subsequent sequencing of the following genes: CEP290 (MIM# 610142), CRB1 (MIM# 604210), RPE65 (MIM# 180069), GUCY2D (MIM# 600179), AIPL1 (MIM# 604392) and CRX (MIM# 602225). In addition, exhaustive phenotyping was performed in all patients carrying mutation(s), and the AHI1 gene was screened for modifier alleles of CEP290-related disease.

7Ninety-one consenting subjects initially diagnosed with LCA were referred for molecular testing by an ophthalmologist and/or geneticist, mainly associated with the University Hospitals of Ghent, Leuven or Brussels. Eleven probands are born from a consanguineous marriage. The inclusion criteria for LCA were bilateral visual loss before the age of 6 months accompanied by nystagmus and an undetectable or significantly reduced ERG. Twelve patients presented with additional mental retardation and/or autistic behaviour. For 18 patients with an available Magnetic Resonance Imaging (MRI), the absence of a molar tooth sign (MTS) excluded the diagnosis of JS. In addition, genotyping was performed on 11 probands with EORD (disease diagnosed beyond the first six months of life but before the age of three) and 13 with a retinal dystrophy in the context of JS (LCA-JS), SLS or cerebello-oculo-renal syndrome (CORS). These patients were not included during calculations of gene-specific contributions in isolated LCA. Genomic DNA and RNA were extracted from leukocytes using the Puregene DNA isolation kit (Gentra) and the RNeasy Mini kit (Qiagen) respectively, followed by cDNA synthesis with the iScript cDNA Synthesis kit (Bio-Rad). If available, parents and/or siblings were also genotyped. Seven of the patients were reported previously. Patient notation was performed according to their clinical diagnosis (prefixes LCA, SLS, LCA-JS, CORS and EORD), with consecutive numbering in the order of the genes involved.

8As a pre-screening method, all patients with either isolated LCA or EORD were analysed with a microarray containing 344 to 641 mutations in 8 (GUCY2D, CRX, RPE65, CRB1, RPGRIP1,AIPL1, LRAT and MERTK) to 13 (addition of TULP1, LCA5, RDH12, CEP290 and SPATA7) LCA and EORD genes (LCA chip Versions 2004–2009; Asper Ophthalmics, Estonia) (http://www.asperbio.com). Each of the mutations found by the LCA chip was subsequently confirmed through Sanger sequencing. In case of a heterozygous mutation, the coding exons and intron-exon boundaries of the involved gene were sequenced.

9Patients in whom no mutations were identified after LCA chip analysis were analysed through sequencing of all coding exons and intron-exon boundaries of CEP290, CRB1, RPE65, GUCY2D, AIPL1 and CRX, the first five genes being the most frequently mutated in LCA. At the time the LCA chip did not yet include CEP290 variants, stepwise targeted mutation analysis was performed prior to sequencing of the total coding region. We initially screened for the frequent c.2991+1655A>G mutation followed by four additional mutations: c.4723A>T (p.Lys1575X), c.5587-1G>C (splice site), c.5163del (p.Thr1722GlnfsX2) and c.3310-1_3310delinsAA (splice site). The first three mutations occurred multiple times in a previous study; the latter was found in three patients with a heterozygous c.2991+1655A>G mutation in our population. CEP290 was also screened at cDNA level in patients with only a single mutation in CEP290. To this end, cDNA screening using 16 overlapping primer sets was optimized. Four patients with CEP29 0-related LCA who presented with mental retardation, two patients with SLS, one patient with CORS and one patient with LCA-JS underwent sequencing of the AHI1 gene. For 13 patients with SLS/LCA-JS/CORS, molecular testing of CEP290 was requested. Supp. Table S1 includes all primer sequences used in this study.

10Mutation nomenclature uses numbering with the A of the initiation codon ATG as +1 (http://www.hgvs.org/mutnomen), based on the following RefSeqs: NM_201253.1 (CRB1), NM_000329.2 (RPE65), NM_000180.3 (GUCY2D), NM_014336.3 (AIPL1), NM_000554.4 (CRX), NM_025114.3 (CEP290), NM_152443.2 (RDH12), NM_020366.3 (RPGRIP1) and NM_001134831.1 (AHI1) (http://www.ncbi.nlm.nih.gov/nuccore). All mutations and variants found in CEP290 were submitted to the locus-specific mutation database CEP290base (http://medgen.ugent.be/cep290base).

11The presence of all mutations was confirmed on a second PCR product. Segregation analysis of disease alleles was performed if possible. Genomic DNA obtained from > 340 unrelated ethnically matched healthy individuals was used as a control panel. Thorough bio-informatic evaluation of novel variants was done using Alamut software (v.1.5). Variants were designated as “unclassified variant (UV)” if no consensus was seen in all prediction programs used. The Alamut output for missense changes is listed in Supp. Table S2.

12After identification of the molecular cause, clinical records were revisited, based on a clinical checklist comprising data on visual function, retinal appearance and associated (extra-) ocular features. When possible, ERG, fundus pictures, autofluorescence (AF) images and optical coherence tomography (OCT) were obtained. In case of CEP290-related LCA, neurological (MRI) and nephrological data (kidney ultrasound [US], urinary and blood parameters) were evaluated.

13As a first step, 102 probands were subjected to LCA chip analysis (91 LCA and 11 EORD). In total, 30 sequence changes assigned as mutations by Asper Ophthalmics were identified in 47 individuals. Homozygous and compound heterozygous variants in one gene were each found in 13 patients; a single heterozygous variant was identified in 17 individuals. In addition, variants within two distinct genes were found in four patients. The zygosity of p.Glu1330X (CRB1) could not be determined in LCA-36. Confirmation of each mutation through direct sequencing identified two inconsistencies. At first, LCA-58 was genotyped heterozygously for the AIPL1 mutation p.Trp278X by chip, while she was in fact homozygous. Secondly, a heterozygous p.Arg38AlafsX3 mutation in AIPL1 (LCA chip version 2006) could not be confirmed in LCA-23. Instead, a heterozygous c.111C>T (p.=) variant was identified on the same nucleotide position. This miscall has previously been described. In addition, subsequent sequencing of GUCY2D in LCA-51 revealed a heterozygous c.389del mutation that was not detected on the LCA chip. The variants c.2101C>T (p.Pro701Ser) (GUCY2D), c.3341A>G (p.Asp1114Gly) (RPGRIP1), c.286G>A p.Val96Ile (AIPL1) and C.1301C>T (p.Ala434Val) (RPE65) have already been reported as polymorphisms and were therefore discarded as mutations. Moreover, identification of the GUCY2D p.Pro701Ser variant in a homozygous state in both healthy parents from an LCA patient further supported its non-pathogenic nature. After the exclusion of these polymorphisms, variants were assigned to be mutations in 45 patients (39 LCA and 6 EORD).

14Secondly, all patients with a heterozygous mutation identified through chip analysis were subjected to screening of the relevant gene. In addition, all patients with negative chip results underwent sequencing of 6 LCA genes. In the following sections, the molecular results are discussed in detail for each of the genes.

15CEP290 was found to be the most frequently mutated gene in our cohort, accounting for 30% (27/91) of cases with isolated LCA (Table 1). Since the LCA chip did not contain CEP290 variants at the onset of this study, only a fraction of currently known mutations were detected using this technique. The c.2991+1655A>G, c.4723A>T (p.Lys1575X) and c.3310-1_3310delinsAA mutations were the most recurrent, with gene-specific allele frequencies of 49%, 11% and 6%, respectively. Similar to previous studies, most of the mutations are either nonsense, frameshift or splice site mutations. Only two missense variants were identified, of which the pathogenic effect is currently uncertain (p.Ala1566Pro and p.Leu1694Pro) (Supp. Table S2). Overall, 13 novel CEP290 mutations were identified to cause LCA. The complex allele c.3310-1_3310delinsAA has a predicted effect on splicing, which was confirmed by cDNA analysis (data not shown). The silent c.1824G>A change affects the last nucleotide of exon 18 and was also predicted to alter splicing (data not shown).

16Since CEP290 mutations may cause a phenotypic spectrum ranging from isolated LCA to more complex disorders, we analysed 13 additional probands suffering from LCA-JS, SLS or CORS. CEP290 harbored mutations in seven of them (Table 1). Six probands carried known mutations, whereas a novel p.Thr2457AlafsX27 mutation segregated in family LCA-JS-2.

17Sequencing of the entire coding region did not reveal a second mutation in LCA-27, while the pathogenic effect of one variant was uncertain in LCA-25 and LCA-26. Subsequent cDNA screening in LCA-25 and LCA-27 was normal, thereby making deep intronic splice site mutations or large exon deletions/duplications very unlikely. No RNA was available for LCA-26.

18Mutations in CRB1 were found in 15 families with LCA (16%) and 5 families with EORD (Table 1). The LCA chip allowed the identification of a homozygous or compound heterozygous CRB1 mutation in 15 probands (12 LCA and 3 EORD), and a heterozygous CRB1 mutation in 3 patients. Sequencing of the whole coding region of CRB1 in the latter revealed a known and novel mutation, respectively (LCA-29, p.Gln362X and LCA-34, c.4006-1G>T), and a novel unclassified variant (EORD-3, p.Asp491Val) on the second allele (Supp. Table S2). Sequencing of the total coding region identified compound heterozygous mutations in two additional probands. In one of them, CRB1 screening was exceptionally performed without prior LCA chip analysis, given clear clinical indications for a CRB1-related phenotype (EORD-2). Indeed, this patient was compound heterozygous for the known p.Cys948Tyr mutation, and the novel p.Cys310Tyr variant, which is predicted to disrupt a disulfide bridge (Supp. Table S2). In addition, two sisters with LCA carried two novel frameshift mutations (LCA-41). As previously described, the p.Lys801X and p.Cys948Tyr mutations were most frequent, showing gene-specific allele frequencies in the LCA cohort of 27% and 23%, respectively.

19Eight cases with LCA showed mutations in RPE65 (9%) (Table 1). LCA chip analysis identified a homozygous RPE65 mutation in LCA-43 and LCA-47 and two compound heterozygous mutations in LCA-49. In addition, a heterozygous mutation was detected in four patients through LCA chip analysis. Sequencing of RPE65 in these individuals identified three novel mutations and the known p.Phe530LeufsX40 mutation which was not yet present on the LCA chip at the time (LCA-48). Although the evidence for a pathogenic nature of the novel mutation p.Trp331dup is not conclusive, segregation in patient LCA45a, her affected aunt LCA45b and her (healthy) parents sustains a causal role. The two other novel mutations result in a frameshift (LCA-46 and LCA-50). Following sequencing of the total coding region, one additional proband was found to be homozygous for the novel RPE65 mutation p.Pro181Leu (LCA-44) (Supp. Table S2). Interestingly, segregation analysis of the mutations found in LCA-46 could only confirm segregation of p.Leu341Ser in the mother, suggesting that p.Ser121LeufsX6 occurred de novo (paternity confirmed).

20Mutations in GUCY2D were found in seven probands with LCA (8%) (Table 1). Five of them were identified with GUCY2D mutations using the LCA chip. One was homozygous for p.Phe565Ser (LCA-52), while three others carried p.Arg768Trp. One of the latter was homozygous (LCA-53); the other two were compound heterozygous for p.Phe565Ser (LCA-54) and the novel missense change p.Lys866Asn (LCA-55), respectively (Supp. Table S2). In addition, LCA-57 was heterozygous for the p.Pro575Leu variant that was previously identified in the mother of an LCA patient. However, no second mutation was found. LCA-56 was compound heterozygous for the novel missense changes p.Glu196Val and p.Pro71 1Leu (Supp. Table S2). In addition, a novel splice site mutation was identified in LCA-51 (c.2577-2A>C).

21Only two distinct AIPL1 variants were detected through LCA chip analysis in five LCA patients (5%) (Table 1). The p.Trp278X mutation occurred homozygously in four probands (LCA-58 to LCA-61). In addition, a heterozygous p.Thr1 14Ile variant was found in proband LCA-62. Direct sequencing of the AIPL1 gene identified the known variant p.Pro376Ser (missing signal on LCA chip). Segregation analysis in the parents, however, revealed a cis-allelic inheritance from the mother. No further mutations were detected following additional sequencing of AIPL1 in other patients.

22LCA chip analysis identified 2 CRX missense variants in two LCA patients (Table 1). The p.Tyr142Cys variant was previously described as a mutation. Stone and colleagues, however, considered this variant as a polymorphism based on the estimate of pathogenic probability and the identification of this variant in a patient with two disease-causing alleles in another LCA gene (LCA-63). The pathogenicity of the second variant p.Val242Met also remains unclear (LCA-64). Given their uncertain pathogenic potential, both variants were discarded as mutations for further calculations.

23Mutation screening of RDH12 was performed downstream of LCA chip results involving this gene. A heterozygous p.Ala269GlyfsX2 mutation was identified in two probands with EORD. Subsequent sequencing of RDH12 identified an additional missense change in both patients, p.Val233Asp (EORD-6) and p.Ser175Leu (EORD-7) respectively (Supp. Table S2). According to UniProt, the Ser175 residue might be a substrate binding site (http://www.uniprot.org/uniprot/Q96NR8). A known mutation located in the same codon, p.Ser175Pro, lacks the ability to catalyze the reduction of retinaldehyde to retinol in vitro. In addition, a homozygous p.Ala269GlyfsX2 mutation was identified in patient EORD-8, for which RDH12 sequencing was performed prior to LCA-chip analysis (upon request).

24Similarly, screening of RPGRIP1 was performed in the context of LCA chip analysis. In one LCA patient, a heterozygous mutation was identified. Sequencing of RPGRIP1, however, did not identify a second mutation (LCA-65).

25For the assessment of the potential involvement of a second gene in LCA, only variants with significant pathogenic potential were taken into account (see above). Two patients with CEP290-related LCA displayed a heterozygous mutation in another LCA gene: LCA-16 was heterozygous for the known p.Arg85Cys mutation in RPE65, while LCA-20 carried the common p.Lys801X mutation in CRB1.

26The AHI1 gene was sequenced as a candidate modifier gene in eight patients with CEP290-related LCA who presented with mental retardation. Four of them were diagnosed with LCA (LCA-3, LCA-20, LCA-23 and LCA-24); two patients also suffered from NPHP (SLS-2 and SLS-3) and in two other cases, the LCA phenotype was part of a JS diagnosis (CORS-1 and LCA-JS-3). A MTS was absent on brain imaging in two isolated patients with LCA (no data were available for LCA-20, LCA-24, SLS-2 and SLS-3).

27A heterozygous novel AHI1 p.Asn811Lys mutation was found in the most severely affected patient CORS-1, out of three patients with the same CEP290 genotype but different neurological involvement (SLS-2, SLS-3 and CORS-1). Moreover, AHI1 screening in the five remaining patients revealed a second heterozygous missense variant, p.His758Pro, in LCA-3. Conservation and in silico predictions for both changes suggest a possible effect on protein structure/function (Supp. Table S2). Interestingly, exonic splicing enhancer (ESE) predictions point to a change in ESEs for both variants (data not shown). Moreover, both changes are located in a conserved WD-40 repeat (http://www.uniprot.org/uniprot/Q8N157) and were absent in > 340 Belgian control individuals.

28In addition, SLS-2 was found to be heterozygous for the known p.Ser1123Phe change. Although it concerns a potentially pathogenic variant that affects a phosphorylation site and is located in a highly conserved region, this change was considered a polymorphism because of its frequency in the Dutch population and the observation that it did not segregate in a family with JS.

29Extensive ophthalmological data (best corrected visual acuity [BCVA], refraction, ERG, visual fields, color vision testing, fundus aspect both with white light and autofluorescence imaging and the presence of nystagmus, night blindness, photophobia and additional features) as well as associated manifestations in patients with CRB1, RPE65, GUCY2D, AIPL1, CRX, RPGRIP1 and RDH12 mutations are summarized in detail in Supp. Table S3. In addition, Figure 1 depicts several representative fundi from patients with an established molecular diagnosis. An MTS due to midbrain abnormalities with cerebellar vermis aplasia was demonstrated in five patients with JS-LCA/CORS and CEP290 mutations (Figure 2).

30Special attention was paid to the CEP290-related ocular phenotype, since this has been described in only a few LCA studies so far. It appeared that this phenotype displays only limited fundus alterations in the first few years of life. In a small subset of patients, no fundus abnormalities were obvious early on, while in the majority a marbleized fundus and/or salt and pepper aspect was seen during the first decade. This aspect further evolved from young adulthood into progressive outer retinal atrophy in the midperiphery with relative sparing of the central macula. Abnormalities of the central macula were absent in our patient cohort, despite the impression that CEP290-related disease is probably of a cone-rod type (based on ERG findings and the occurrence of photophobia). Of note is the presence of a hyperautofluorescent ring around the central macula on AF imaging, observed in four patients starting from the age of six (LCA-2, LCA-3, LCA-7 and LCA-25). Mild intraretinal spicular pigment migration occurred in three patients at an age between 7 and 33 years old (LCA-JS-1, LCA-6 and LCA-7). This aspect became even more pronounced at the age of 49 in patient LCA-7, where a predominant spicular pigmentation was mixed with less frequent intraretinal pigment migration with a nummular aspect. Visual acuity of this group was mostly limited to light perception. In the few patients with better preserved central vision, basic color vision was present and visual fields varied from severely concentrically constricted (LCA-8) to sparing of the central 30° at an age of 49 (LCA-7) (Supp. Table S3).

31In a cohort of 91 unrelated LCA patients, mainly originating from Belgium, a total of 61 different mutations (including 9 UVs) were found in 7 genes. Homozygous or compound heterozygous mutations were detected in 65% (59/91) of probands, whereas only one heterozygous mutation could be identified in 4% (4/91) of probands. In addition to isolated LCA, this study also identified mutations in eight probands with EORD and seven probands with syndromic LCA (SLS, LCA-JS and CORS) (Table 1).

32LCA chip analysis proved to be a powerful initial tool as mutations were found in 41% (37/91) of patients with LCA. Subsequent sequencing of 6 genes (CEP290, CRB1, RPE65, GUCY2D, AIPL1 and CRX) enabled us to identify mutations in an additional 28% of cases. Of note, the majority of mutations found in the latter probands are now included in the LCA chip, increasing its detection rate to 65% if our cohort would have been analyzed using the latest version (v8, 641 variants).

33Segregation of disease alleles was demonstrated in 41 out of 43 families available, with two exceptions. The first one is LCA-62, in which the AIPL1 variants p.Thr1 14Ile and p.Pro376Ser were located in cis on the maternal chromosome. This finding challenges a previous study proposing this genotype as causal in an LCA patient. In the second case, only one of two mutations identified in LCA-46 was found in the mother but not in father, suggesting that the other mutation arose de novo (RPE65, p.Ser121LeufsX6, nonp-aternity excluded) (Table 1). Of note, two different mutations in CEP290 were identified in CORS-1, originating from a consanguineous marriage, which illustrates that assuming homozygosity in offspring from a consanguinous mating can be a potential pitfall for the identification of the causal defect (Table 1).

34Our data demonstrate a key role for CEP290 in the Belgian LCA population, as CEP290 mutations were identified in 27 probands (30%). Although the prevalence of CEP290 mutations is not this high worldwide, this study corroborates the importance of CEP290 in the Northwestern European population. The second most frequently mutated gene in our LCA population was CRB1 (16%), followed by RPE65 (9%), GUCY2D (8%) and AIPL1 (5%), which is in agreement with previous data. The RPGRIP1 gene – in this study only investigated with the LCA chip – accounted for less than 1% of the LCA population.

35In total, 30 novel mutations/variants were identified in this study (Table 1, Supp. Table S2). Interestingly, 5 out of 16 novel mutations in CEP290 are located nearby known changes: c.384_385del (c.384_387del, c.381_382delinsT), c.2218-2A>C (c.2218-4_2222del and c.2218-15_2220del), c.3310-1_3310delinsAA (c.3310-1G>C), c.5519_5537del (c.5515_5518del) and c.5865_5867delinsGG (c.5866G>T). A similar observation was made for RDH12, in which the novel mutation p.Ser175Leu affects the same codon as the known mutation p.Ser175Pro. Overall, these regions/codons might be more prone to mutational events.

36For several genes, a limited number of recurrent mutations made up the majority of mutated alleles. This was certainly the case for CEP290, in which c.2991+1655A>G was found in 89% of all LCA patients with CEP290- related pathology. Together with p.Lys1575X and c.3310-1_3310delinsAA, a significant fraction of mutated alleles was identified in the LCA population (35/53 alleles). So far, p.Lys1575X has only been found in patients originating from northern France or Belgium. This potential founder effect is supported by our study, since all patients who carry p.Lys1575X live in Flanders (northern part of Belgium). A similar regional prevalence was seen for p.Ala1832ProfsX19, which was inherited from the Greek father of LCA-16. The same mutation occurred in an Italian patient with CORS and in two consanguineous families of Kosovar-Albanian and Kosovar origin with MKS, sharing a common haplotype. For the other genes, the following mutations presented with a gene-specific allele-frequency of at least 20% in the LCA population: p.Lys801X (CRB1, 27%), p.Cys948Tyr (CRB1, 23%), p.Phe530LeufsX40 (RPE65, 25%), p.Arg768Trp (GUCY2D, 31%), p.Phe565Ser (GUCY2D, 23%) and p.Trp278X (AIPL1, 89%). The presence of all but one of these mutations on the LCA chip significantly contributed to its high detection rate.

37In six LCA patients, only one mutation was found after sequencing of the gene apparently involved following chip testing. Notably, we might have failed to detect deep intronic and regulatory mutations or multi-exon deletions, as were recently demonstrated in CEP290. In addition, it cannot be excluded that the phenotype is caused by mutations in a different gene, as was the case for LCA-20. Furthermore, new mutations in the other known genes cannot be ruled out here. Their contribution is expected to be limited however, taking into account the high detection rate obtained using the current strategy. Finally, these patients may carry mutations in as yet unknown genes.

38In addition, the phenotypes of patients with a molecular diagnosis were extensively studied. For all genes, nystagmus and hypermetropia were recurrent features. The oculodigital sign (plus enophthalmos) was often seen in all but RPE65-related LCA patients. A relatively higher incidence of both keratoconus and cataract was observed in the CRB1-related group, which may reflect secondary effects of a more severe retinal dystrophy compared to other genes. Indeed, several retinal abnormalities such as macular atrophy and intraretinal pigment migration already became apparent in the first decade of life in patients with CRB1-related disease, being earlier than generally seen for the other genes. In addition, a yellowish discoloration of the central macula was often observed. However, this feature is not entirely gene-specific, since it also occurred in a patient carrying RDH12 mutations (EORD-7). This feature may be due to more severe outer retinal atrophy in the macula in CRB1- and RDH12-related disease, which may cause more intense scleral light reflection due to less absorption by the atrophic retinal pigment epithelium, with consequent highlighting of the macular luteal pigment.

39The RPE65-related phenotype proved to be typically associated with a fundus appearance which is essentially normal during first years and displays only later on fundus alterations which are initially mild. Visual acuity is generally somewhat better than that seen in CRB1 -related LCA. However, it seems to be the relatively slow evolution of the phenotype which makes it particularly suitable for therapeutic intervention. For two patients from our cohort (LCA-47 and LCA-49), RPE65 gene-replacement therapy resulted in better visual function. In addition, two patients with RPE65 mutations reported a period of increased visual function, possibly reflecting postnatal physiological cone maturation (LCA-44 and LCA-47). Hanein et al. classified both CRB1- and RPE65- related LCA as rod-cone dystrophies because of a predominant occurrence of night blindness. All RPE65-related phenotypes in this study correspond to this classification. In the CRB1-group, however, six patients with LCA and two with EORD also suffered from photophobia, even before the onset of night blindness in EORD-3.

40Similarly, the GUCY2D-related phenotype was found to have only limited fundus abnormalities. Although this phenotype was previously categorized as cone-rod dystrophy, one patient in our cohort had severe night blindness before photophobia became apparent (LCA-54).

41In the patients with AIPL1 mutations, an RP-like phenotype emerged by their teenage years at the latest, and a maculopathy with (partial) outer retinal atrophy was typically present in the majority of cases.

42A unique feature of RDH12-related early-onset dystrophy was the occurrence of areas with complete preservation of the chorioretina in the retinal periphery, alternating with regions of total atrophy (EORD-7).

43Notably, our study is one of the first reporting on the ocular phenotype of a larger group of LCA patients with CEP290 mutations (Supp. Table S3). In keeping with previously reported findings, it appeared to be that of a severe cone-rod type retinal dystrophy. Visual acuity was mostly limited to light perception, as recently described. Interestingly, a limited subset of patients displayed no obvious retinal abnormalities in the first years of life. In general, the fundus contained either small white dots or, more frequently, a marbleized or salt and pepper aspect in the first to second decade. In two patients aged 18 and 49, predominant spicular pigment migration was observed (LCA-6 and LCA-7), in contrast to the reported nummular pigmentation in one patient in the fourth decade. In one patient, nummular pigmentation was described in the first decade (LCA-10). Interestingly, a more severe phenotype was seen in LCA-20, who carried a heterozygous CRB1 null allele on top of two mutations in CEP290.

44Several patients with CEP290-related retinal dystrophy showed additional systemic features. Two patients with isolated LCA had several symptoms suggestive of renal dysfunction (LCA-3 and LCA-23, Supp. Table S3). LCA-3 suffered from growth retardation, polydipsia, enuresis nocturna and diurnal incontinence. Kidney US at the age of seven, however, was normal. In LCA-23, kidney US at the age of three revealed increased echogenicity and kidneys without clear cortico-medullar differentiation. Despite this observation, no clinical nephrological manifestations were present at the age of 17. Since the age of onset of end-stage renal disease caused by CEP290 mutations may exceed the age of 20, a close nephrological follow-up of these patients is required. Interestingly, both of these patients carry the recurrent c.2991+1655A>G mutation, which so far has only been reported in LCA patients without any other associated pathology. Of note, kidney US was available for only a subset of patients, and in general performed very early in life, when developing kidney disease might be difficult to detect.

45In addition, four patients suffered from recurrent otitis media (OM) (LCA-5, SLS-1, SLS-3 and CORS-1, Supp. Table S3). Although this is common in childhood, it is worth mentioning that it is also a clinical manifestation often seen in primary ciliary dyskinesia (PCD), a genetically heterogeneous disorder of motile cilia. In a few cases, PCD with OM was associated with X-linked RP, caused by mutations in RPGR. Notably, RPGR is a centrosomal protein that interacts with CEP290. Moreover, loss-of-function experiments of CEP290 in zebrafish caused developmental abnormalities of the otic cavity.

46Strikingly, 33% of patients with CEP290-related isolated LCA presented with mental retardation and/or autism, in contrast to only 8% of patients with mutations in the other genes. Subtle brain abnormalities such as broadened lateral ventricles were seen on MRI in some patients. Of note, brain imaging was not available for a subset of patients. Additional neurological manifestations included movement abnormalities (LCA-21), and dyspraxia and balance/coordination problems (LCA-18), the latter of which was also evident in two patients with syndromic CEP290-relatedLCA (SLS-2 and CORS-1) (Supp. Table S3).

47Taken together, these extra-ocular manifestations fit well into the broad clinical spectrum of CEP290 mutations, varying from isolated LCA to the lethal MKS. In addition, the CEP290 allelic spectrum is highly complex. Mutations associated with isolated LCA in this study were previously reported in other ciliopathies (Table 1), albeit always in compound heterozygosity with a different mutation in the distinct phenotypes. Moreover, identical CEP290 genotypes can display interfamilial variable expressivity and intrafamilial variation of the neurological phenotype was observed in several families with CEP290-related pathology. The complexity of CEP290-related disease is further illustrated by two cases from this study.

48The first one is SLS-1, homozygous for p.Trp7Cys. Valente and coworkers identified the same mutation in patient with CORS, also of Pakistani origin (COR22, II:1). Despite a similar ocular and renal phenotype, both patients significantly differ in their neurological phenotype. A second and even more pronounced example is the variability in both nephrological and neurological involvement in three unrelated patients with the same p.Lys1575X/p.Arg1465X genotype (SLS-2, SLS-3 and CORS-1). Patients SLS-3 and CORS-1 displayed a similar clinical course of renal disease, with renal failure at the age of 16 and 14 years, respectively. In contrast, renal insufficiency in patient SLS-2 was not substantiated until the age of 30 (Supp. Table S3). Neurological signs of these three patients ranged from a mild mental handicap (SLS-2) over severe autism in combination with moderate mental retardation (SLS-3) to severe mental retardation associated with ataxia and a MTS (CORS-1). An MRI was not available for the other two patients, however.

49The AHI1 gene was screened as a candidate modifier gene in these three patients. Strikingly, CORS-1, with the most severe nephrologic and neurologic phenotype, carries a heterozygous novel p.Asn811Lys mutation in AHI1, which was absent in the two other patients. Upon screening of AHI1 in five additional patients with CEP290-related disease and neurological involvement, a novel missense variant, p.His758Pro, was identified in LCA-3. Mutations in AHI1 encoding Jouberin are responsible for JS, with retinal involvement in 75% and renal involvement in less than 10% of all AHI1-associated patients. Interestingly, both p.Asn811Lys and p.His758Pro affect conserved residues and are located in the predicted WD40-repeat, a domain conserved across all eukaryotes, mediating functions such as vesicular trafficking. Since AHI1 and CEP290 appear to be in the same pathway through their interaction with rab8a, mutations in one of both genes may modify a phenotype caused by the other. Tory and colleagues already suggested a similar potential epistatic effect of CEP290 and AHI1 mutations on phenotypes related to NPHP1, encoding another interactor of AHI1. Strikingly, one of the AHI1 variants they described as a potential modifier for neurological involvement in patients with NPHP1 mutations, p.Arg830Trp, was recently identified as a modifier allele for retinal degeneration in patients with NPHP, independent of a primary NPHP1 mutation. Of note, four out of seven patients in the study from Tory and coworkers carrying p.Arg830Trp displayed visual impairment, with one blind individual. The p.Arg830Trp variant might affect AHI1 complex stability/formation. The variants identified here, assumed to represent a neurological modifier in patients with LCA, might disrupt interactions with other proteins, thereby influencing AHI1 function in other organ systems.

50Overall, a molecular diagnosis of CEP290 mutations might have considerable consequences towards the clinical prognosis of an individual. Given the potential involvement of ciliary modifiers and the presence of cilia throughout the whole body, the development of various additional clinical manifestations should be taken into account. As for LCA, both children and (young) adults should have a long-term close clinical neurological and nephrological follow-up, since some features have a later onset.

51In conclusion, molecular testing identified mutations in 69% of our LCA cohort, with a major involvement of CEP290. Detailed phenotyping of all patients with a molecular diagnosis revealed novel insights, mainly into the CEP290-related retinal phenotype, which is well documented for the first time in a larger patient group. The variable age-of-onset of the extra-ocular features emphasizes the importance of long-term clinical follow-up of LCA patients with CEP290 mutations. Moreover, the identification of potential modifiers of CEP290-related disease might contribute to a refined prognosis based on a molecular diagnosis. Finally, our findings fit with previous observations suggesting that the phenotype of ciliopathies is most likely determined by the synergistic effect of all variants occurring in the ciliary proteome.