The Phenotypic Variability of Retinal Dystrophies Associated With Mutations in CRX, With Report of a Novel Macular Dystrophy Phenotype

Abstract

Purpose.: To present a detailed phenotypic and molecular study of a series of 18 patients from 11 families with retinal dystrophies consequent on mutations in the cone–rod homeobox (CRX) gene and to report a novel phenotype.

Methods.: Families were ascertained from a tertiary clinic in the United Kingdom and enrolled into retinal dystrophy studies investigating the phenotype and molecular basis of inherited retinal disease. Eleven patients were ascertained from the study cohorts and a further seven from investigation of affected relatives. Detailed phenotyping included electrodiagnostic testing and retinal imaging. Bidirectional Sanger sequencing of all exons and intron-exon boundaries of CRX was performed on all 18 reported patients and segregation confirmed in available relatives.

Results.: Based on clinical characteristics and electrophysiology, four patients had Leber congenital amaurosis (LCA), two had rod–cone dystrophy (RCD), five had cone–rod dystrophy (CORD), one had cone dystrophy (COD), and six had macular dystrophy with different phenotypes observed within 5 of 11 families. The macular dystrophy patients presented between 35 to 50 years of age and had visual acuities at last review ranging from 0.2 to 1.5 logMAR (20/32 to 20/630 Snellen). All 18 patients were heterozygous for a mutation in CRX with seven novel mutations identified. There was no evident association between age of onset and position or type of CRX mutation. De novo mutations were confirmed in three patients.

Conclusions.: Mutations in CRX demonstrate significant phenotypic heterogeneity both between and within pedigrees. A novel, adult-onset, macular dystrophy phenotype is characterized, further extending our knowledge of the etiology of dominant macular dystrophies.

Introduction

The cone–rod homeobox gene CRX (MIM #602225) encodes a transcription factor vital for both the development and survival of photoreceptors.^1–3 It is expressed predominantly in photoreceptors and the pineal gland and has a high homology to the OTX family of homeobox genes.^1,3 It acts by binding to promoter enhancer regions of specific retinal genes, thus influencing retinal development. It is co-expressed in the retina with other transcription factors including NRL (neural retina leucine zipper) and NR2E3.^3,4 

A locus for autosomal dominant cone–rod dystrophy (CORD2, MIM #120970) was identified in 1994 and mapped to 19q13.⁵ The gene responsible for CORD at this locus was identified as CRX.¹ Subsequently, it became evident that mutations in CRX may be associated with a range of different retinal phenotypes including CORD, Leber congenital amaurosis (LCA7, MIM #613829), retinitis pigmentosa (RP, MIM #268000), and cone dystrophy (COD).^6–8 

CRX mutations are rare accounting for approximately 1.7% of LCA cases with the majority of mutations arising de novo.⁹ To date, 42 probable disease causing mutations and one whole exon deletion have been reported, all in the heterozygous state except for three case reports of homozygous disease in LCA and severe RP.^1,6–33 

The aim of this study is to report the phenotypic heterogeneity of retinal dystrophies arising from mutations in CRX and describe a previously unreported association with adult-onset ‘bulls-eye' macular dystrophy. 

Methods

The study protocol adhered to the tenets of the Declaration of Helsinki and received approval from the local ethics committee. Written, informed consent was obtained from all participants prior to their inclusion in this study with parental written consent provided on behalf of the children involved in this study. 

Families were identified from the following three study cohorts of patients referred to a tertiary clinic in the United Kingdom: 

1.  
   Six families from a large ongoing survey of families with LCA or childhood onset retinal dystrophy, with genetic screening by candidate gene Sanger sequencing, arrayed primer extension (APEX) microarray (Asper Biotech Ltd., Tartu, Estonia), next generation sequencing of 105 retinal genes (Manchester Centre for Genomic Medicine, Manchester, UK), or exome-analysis (AROS Applied Biotechnology, Aarhus, Denmark);
2.  
   Two families from a cohort of 28 unrelated probands with adult-onset macular dystrophy or CORD and no known molecular diagnosis who were screened with exome analysis; and
3.  
   Three families from a further panel of 65 probands with adult-onset macular dystrophy, or CORD who were Sanger sequenced for CRX.

Each patient with a CRX mutation underwent a full clinical examination including visual acuity and dilated fundus examination. Age permitting, retinal fundus imaging was obtained by conventional 35° fundus color photographs (Topcon Great Britain Ltd., Berkshire, UK), 30° or 55° fundus autofluorescence (FAF) imaging, and spectral-domain optical coherence tomography (OCT) scans (Spectralis, Heidelberg Engineering Ltd., Heidelberg, Germany). In older children and adult patients, a full-field electroretinogram (ERG) and pattern electroretinogram (PERG) was performed using gold foil electrodes to incorporate the International Society for Clinical Electrophysiology of Vision standards; recording in infants and young children was performed with surface electrodes as previously described.^34–37 

For the purposes of this study, clinical subgroups of disease have been defined based on clinical presentation and electrophysiology: LCA presenting within the first 6 months of life and with a nondetectable ERG; CORD, a progressive retinal dystrophy with cone dysfunction greater than rod on ERG; RP, also termed rod–cone dystrophy (RCD), a progressive retinal dystrophy with rod dysfunction greater than cone on ERG; COD, a progressive retinal dystrophy with isolated cone dysfunction on ERG; and macular dystrophy presenting with macular dysfunction and normal full-field ERG. 

Molecular Biology

Genomic DNA was isolated from peripheral blood lymphocytes using the Puregene kit (Gentra Puregene Blood Extraction Kit, QIAGEN, Manchester, UK). 

APEX screening (Asper Biotech Ltd.) was performed using a genotyping microarray containing more than 300 disease-causing variants and common polymorphisms for eight retinal dystrophy genes (AIPL1, CRB1, CRX, GUCY2D, RPE65, RPGRIP1, LRAT, and MERTK) as previously described.³⁸ 

Next generation sequencing of the coding regions of 105 retinal genes (1874 exons) was performed at the Manchester Centre for Genomic Medicine (Manchester, UK) with enrichment using a SureSelect Target Enrichment Kit (Agilent Technologies Inc., Santa Clara, CA, USA) then run on a SOLiD 4 sequencer (Life Technologies, Grant Island, NY, USA), more information available on request. 

Whole-exome sequencing was performed at AROS Applied Biotechnology using a solution-phase Agilent SureSelect 38-Mb exome capture (SureSelect Human All Exon Kit; Agilent Technologies Inc., Santa Clara, USA) and the Illumina HiSeq2000 sequencer (Illumina, San Diego, CA, USA). Reads were aligned to the hg19 human reference sequence using Novoalign (version 2.05; Novocraft, Selangor, Malaysia). The ANNOVAR tool (in the public domain, Openbioinformatics.org) was used to annotate single nucleotide polymorphisms (SNPs) and small insertions/deletions. 

Bidirectional Sanger sequencing of all exons and intron–exon boundaries of CRX was performed on all 18 reported patients and segregation confirmed in available relatives. DNA was amplified using specifically designed primers by PCR and the resulting fragments were sequenced using standard protocols (Supplementary Table S1). 

Mutation nomenclature was assigned in accordance with GenBank Accession number NM_000554.4 with nucleotide position 1 corresponding to the A of the ATG initiation codon. Variants were identified as novel if not previously reported in the literature and if absent from the dbSNP (available in the public domain at http://www.ncbi.nlm.nih.gov/projects/SNP/), NHLBI GO Exome Sequencing Project (EVS, Seattle, Washington; available in the public domain at http://evs.gs.washington.edu/EVS/), and 1000 genomes project (available in the public domain at http://www.1000genomes.org/), all accessed July 9, 2014.³⁹ The likely pathogenicity of novel missense variants was assessed using the predictive algorithms of ‘Sorting Intolerant from Tolerant' (SIFT; available in the public domain at http://sift.jcvi.org) and Polymorphism Phenotyping v2 (PolyPhen-2; available in the public domain at http://genetics.bwh.harvard.edu/pph2).^40,41 Where relevant, potential splice site disruption was assessed using Splice Site Prediction by Neural Network (available in the public domain at http://www.fruitfly.org/seq_tools/splice.html).⁴² Variants likely to affect function were assessed for segregation in available family members. 

Phenotype–Genotype Correlation

Age of presentation was used as a surrogate and approximate metric of severity in order to test associations between mutation position and phenotype severity. First, a quantitative analysis was performed of the mutations by plotting the mutation position against age of presentation. Using the same data, the hypothesis that mutations affecting residues earlier in the gene are generally more severe than others was tested by computing the nonparametric Spearman correlation coefficient. Second, a qualitative comparison was performed by dividing the mutations into two mutually exclusive groups: missense variants affecting the homeodomain, and those that were frameshifting/premature termination codons (PTC) in the carboxyl end of the gene. A comparison of the median age of the two groups was made, and tested for significance using the Mann-Whitney U test. Statistical analysis was performed using IBM SPSS Statistics version 22 (Portsmouth, UK). 

A review of all reported mutations in the CRX gene was undertaken from published literature giving a context for the mutations discovered in this cohort. Conservation of CRX homeodomain residues between species and between paralogues within humans was analyzed using Clustal Omega (accessed in the public domain at http://www.ebi.ac.uk/Tools/msa/clustalo/) with protein sequences identified from Ensembl (accessed in the public domain http://www.ensembl.org).^43,44 The location of mutations arising within the homeodomain were plotted against the consensus sequences. 

Results

The clinical data are summarized in Table 1 with pedigrees shown in Figure 1. From the screened cohorts, 11 affected patients were ascertained: (1) six patients with two from candidate gene Sanger sequencing, two from APEX screening, one from next generation sequencing, and one from exome analysis, (2) two patients from exome analysis, and (3) three patients from targeted Sanger sequencing. A further seven affected relatives from five families were recruited giving a total of 18 affected patients from 11 families. There were six simplex cases of which de novo disease could be confirmed in three (19090, 19512, 20046). A dominant pedigree was evident in four families. 

Figure 1

 

Pedigrees of families.

Table 1

 

Key Phenotypic Features of Patients

Family (Sex)	Age of Onset	Age Last Review	Diagnosis	Fundus	Age at Last Electrophysiology, Key Findings	Latest VA, logMAR, (Snellen Equivalent)	Latest Refractive Error
19090 (f)	Birth	2 y	LCA	Blonde fundus	11 mo: probably undetectable	R NPL L NPL	Unknown
19512 (m)	3 mo	2.5 y	LCA	Blonde fundus, central macular atrophy, thin peripheral retina	9 mo: undetectable	R NPL L NPL	R +3.00 DS L +3.00 DS
20046(f)	Birth	7 mo	LCA	Blonde fundus	7 mo: markedly attenuated	R PL L PL	R +3.50 DS L +3.50 DS
5126.1 (m)	12 y	50 y	CORD	Macular atrophy, peripheral retinal atrophy, bone spicules, attenuated arterioles, pale optic discs	47 y: very severe generalized retinal dysfunction	R HM L CF	R +3.25/-2.00 × 165 L +3.75/-3.50 × 5
5126.2 (m)	12 y	27 y	CORD	Macular atrophy, subtle peripheral RPE mottling	25 y: undetectable PERG, generalized retinal dysfunction, worsening of cone function	R 1.0 L 1.1(R 20/200 L 20/250)	R +1.25/-2.75 × 29L +1.25/-3.00 × 174
5126.3 (f)	14 y	25 y	CORD	Macular atrophy, peripheral retinal RPE pigment change	15 y: undetectable PERG, subnormal rod ERGs, moderately severe reduction cone responses	R 1.0 L 0.8(R20/200, L 20/125)	R +6.5/-3.50 × 175L +6.00/-3.50 × 20
17489.1(m)	3.5 y	16 y	RCD	Small yellow spots R macula, pale optic discs, attenuated arterioles, mid peripheral hypopigmentary granular change	11 y: Undetectable PERG and rod ERG, and a markedly subnormal cone specific ERG	R 1.0 L 1.2(R 20/200, L 20/320)	R +0.50 DSL +0.50 DS
17489.2(f)	53 y	53 y	Macular  dystrophy	Mild disc pallor only	53 y: Bilateral macular dysfunction, normal ERGs	R 0.3 L 0.5(R 20/40, L 20/63)	Unknown
18280.1(f)	49 y	56 y	Macular  dystrophy	Ring of RPE atrophy in maculae	52 y: PERG not definitely detectable, normal ERGs	R 0.2 L 0.3(R 20/32 L 20/40)	R +0.25/0.25 × 180L +0.25/0.25 × 45
18280.2 (f)	50 y	56 y	Macular  dystrophy	Ring of RPE atrophy in maculae	51 y: Undetectable PERG, normal ERGs	R 0.6 L 0.3(R 20/80 L 20/40)	R +1.00/-0.75 × 90L +0.50 DS
18280.3(m)	32 y	42 y	CORD	Ring RPE atrophy in maculae with peripheral RPE pigmentary change	35 y: Undetectable PERG, subnormal rod ERG, markedly subnormal cone ERG	R 1.0 L 1.0(R 20/200 L 20/200)	R −2.50/-0.50 × 180L −3.50/-0.50 × 180
19161.1 (f)	50 y	56 y	Macular  dystrophy	Mild RPE mottling maculae	52 y: Markedly reduced PERG, normal ERGs	R 0.0 L 0.0(R 20/20 L 20/20)	Hyperopic
19161.2 (f)	45 y	52 y	Cone  dystrophy	Ring of RPE atrophy in maculae	48 y: Severely reduced PERG, significantly reduced and delayed cone ERGs	R 0.5 L 0.2(R 20/63 L 20/32)	R +1.75 DSL +1.75 DS
19990.1(f)	6 y	27 y	RCD	Macular atrophy, peripheral extensive pigmentary retinopathy	26 y: Severe generalized loss of photoreceptor function	R PL L PL	Unknown
19990.2(f)	Birth	2 y	LCA	Ring of RPE atrophy in maculae, mottled peripheral RPE change	20 mo: undetectable PERG and ERG	R HM L HM	R +6.00/-2.00 × 180L +2.00/-1.25 × 180
712(f)	11 y	73 y	CORD	Pale discs, blonde posterior pole, attenuated arterioles, peripheral pigmentary clumps	60 y: PERG undetectable on L, residual activity on R, severe generalized cone dysfunction with rod involvement	R 0.8 L 1.0(R 20/125 L 20/200)	R −11.00/-1.00 × 10L −9.50/-2.00 × 175
4663(f)	42 y	67 y	Macular  dystrophy	Macular atrophy	67 y: PERG extinguished, normal ERGs	R 1.3 L 1.5(R 20/400 L 20/630)	Hyperopic
16711(m)	35 y	63 y	Macular  dystrophy	Macular atrophy	54 y, PERG undetectable, normal ERGs	R 1.0 L 1.0(R 20/200 L 20/200)	Unknown

 VEP, visual evoked potential; R, right; L, left; NPL, no perception of light; PL, perception of light; CF, counting fingers; HM, hand movements.

Twelve patients had generalized photoreceptor dysfunction with four LCA, two RCD, five CORD, and one COD. 

An unexpected group of adult onset macular dystrophy in six patients was identified with initial identification of this phenotype from exome analysis of two cases. Two were asymptomatic with their disease identified incidentally. One (patient 17489.2) was identified after all family members of patient 17489.1 were examined in the clinic and the other (patient 19161.1) was identified after routine visual field testing at the optometrist showed centrocecal scotomata. This patient still had acuity of 0.0 logMAR (20/20 Snellen) after 6 years of follow up. Deterioration of acuity with time was documented in three cases. Patient 4663, the most severely affected of the macular dystrophy group, deteriorated from 0.2 logMAR (20/32 Snellen) each eye to right 1.3 logMAR (20/400 Snellen) and left 1.5 logMAR (20/630 Snellen) during 16 years follow up. Patient 16711 had incidental peripapillary changes similar to angioid streaks without any other identifiable features in the fundus or systemically (Fig. 2); this was thought to be an incidental finding unrelated to his macular dystrophy. 

Figure 2

 

Fundus imaging of patients 5126.2, 17489.1, 17489.2, 19161.1, 18280.1, 18280.3, 16711, 4663: (a) right fundus photograph, (b) right 30° or 55° fundus autofluorescence (FAF) imaging, (c) right optical coherence tomography (OCT). Patient 4663, (a.1) fundus photograph from 1998, (a.2) fundus photograph from 2014.

A common fundus feature in all phenotypes was that of macular atrophy, present in 14 of 18 cases. It was not present in two LCA patients, nor in both members of family 17489, although these latter cases did have inner segment ellipsoid (ISe) band disruption at the maculae on OCT (Fig. 2). Three of four LCA cases had noticeably blonde fundi at presentation. 

Fundus autofluorescence imaging and OCT scans were available in 12 of 18 patients; the LCA patients were all too young for imaging and FAF and Spectralis OCT was unavailable in the COD patient. Fundus autofluorescence imaging demonstrated a reduced ring of autofluorescence parafoveally in the CORD patients, an extensive loss of autofluorescence in the RCD patients and a ring of increased autofluorescence at the macula with loss of autofluorescence within the ring in all macular dystrophy patients. 

On OCT, four of the CORD patients had loss of the ISe band with retinal thinning at the macula on OCT, with patient 9 showing disruption of the ISe band but no macular thinning. The two patients with RCD had thin retina with loss of the ISe band at the maculae on OCT. The macular dystrophy patients had disruption of the ISe band at the maculae on OCT with patient 17489.2 the least affected. 

Electrophysiology was performed in all cases (Fig. 3, Table 1). Patients with LCA had an undetectable ERG, whereas those with later onset generalized photoreceptor dystrophy had subnormal and delayed full field ERGs. The PERG was universally reduced in the macular dystrophy patients, with normal full-field ERGs. 

Figure 3

 

Full-field ERG and PERG features from one eye of patients 19161.1, 18280.3, 17489.1, and healthy control (age 24). Patient 19161.1, macular dystrophy, demonstrates subnormal PERG with normal ERGs for age (53 years at ERG).^54,55 Patient 18280.3, CORD, demonstrates subnormal rod and bright flash ERGs (DA 0.01; DA 11.0), delayed and subnormal cone flicker ERGs (LA 3.0 30 Hz), and markedly subnormal single flash ERG (LA 3.0 2 Hz) with undetectable PERG. Patient 17489.1, RCD, demonstrates undetectable rod ERG (DA 0.01); severe reduction in the bright flash ERGs (DA 11.0), and markedly delayed and subnormal cone flicker and single flash ERGs (LA 3.0 30Hz; LA 3.0 2 Hz). Pattern ERG was unrecordable due to nystagmus.

Four families demonstrated intrafamilial phenotypic heterogeneity (Fig.1). Family 17489 segregates macular dystrophy and RCD; family 18280 macular dystrophy and CORD; family 19161 macular dystrophy and cone dystrophy; and family 19990 RCD and LCA. Family 17489 is particularly unusual as the son has early-onset retinal dystrophy with rod–cone dysfunction, the mother a mild, asymptomatic macular dystrophy, and the daughter optic atrophy with normal ERGs that presented age 3. Both she and her father screened negative for mutations in CRX. The identified heterozygous mutation in this family has previously been reported.¹⁰ Phenotypic homogeneity was present in only one family (5126), all affected members having CORD. 

Molecular analysis was performed on all patients and available family members (Table 2). Seven novel mutations were identified, five frameshifting and one PTC mutation in exon 4, and one missense mutation in exon 3. The novel missense mutation is predicted to be pathogenic based on Sift and Polyphen2 scores (Table 2). Segregation analysis confirmed de novo mutations in three of the LCA cases. 

Table 2

 

Mutations in CRX Found in This Patient Series, With Predicted Damage Based on SIFT and Polyphen-2 Scores and the First Report of Each Mutation in the Literature

Family	Diagnosis	Variation: Nucleotide, Protein	Predicted Effect	Segregation	First Reported
19090	LCA	c.570delC, p.Tyr191Metfs*3	Frameshifting	Both parents negative	This paper (for  mutation)
19512	LCA	c.571delT, p.Tyr191Metfs*3	Frameshifting	Both parents negative	Rivolta 200126
20046	LCA	c.570delC, p.Tyr191Metfs*3	Frameshifting	Both parents negative	This paper
5126	CORD	c.568_590del, p.Pro190Glyfs*38	Frameshifting	Present in all 3 affected patients, other family DNA unavailable	This paper
17489	RCD, macular  dystrophy	c.121C>T, p.Arg41Trp	Pathogenic (Sift score 0,  damaging; Polyphen2  score 1.0, probably  damaging)	Present in affected patient and affected mother, absent in father and sister	Swain 199710
18280	Macular  dystrophy, CORD	c.774T>A, p.Tyr258*	Truncating mutation	Present in all 3 affected patients, further family DNA unavailable	This paper
19161	Macular  dystrophy, Cone  dystrophy	c.605delG, p.Cys202Sfs*17	Frameshifting	Present in both affected patients, further family DNA unavailable	This paper
19990	RCD, LCA	c.624T>G, p.Tyr208*	Truncating	Present in affected patient and daughter, further family DNA unavailable	Stone 20079
712	CORD	c.821delG, p.Gly274Alafs*97	Frameshifting	No other DNA available	This paper
4663	Macular dystrophy	c.582delC, p.Tyr195Thrfs*23	Frameshifting	No other DNA available	This paper
16711	Macular dystrophy	c.272G>A, p.Arg91Lys	Pathogenic (Sift score 0,  damaging, Polyphen2  score 0.992, probably  damaging)	No other DNA available	This paper

Nine further missense variants in 11 patients were identified (Supplementary Table S2^10,45,53). Based on predictive algorithms, previous reports, and the presence of the variant in control population databases (EVS), eight of these most likely represent benign changes with the ninth predicted to be damaging but not segregating with known disease within the family. Four of these nine variants are novel and include two synonymous changes, c.355A>C (p.Arg119Arg) and c.561C>T (p.Thr187Thr), and two nonsynonymous changes, c.127C>T (p.Arg43Cys) and c.526C>T (p.Arg176Trp). The novel synonymous changes are predicted to be tolerated on Sift analysis, arise more than 100 base pairs from any intron–exon boundary and are not predicted to affect splicing based on in silico analysis. Variant c.365G>A, found in two patients with macular dystrophy and cone dystrophy, onset childhood and early 30's, respectively, has been previously reported as an apparently benign variant.⁴⁵ Variants c.472G>A and c.101-12A>G were found in two patients both with predominantly macular dystrophy and mild full-field ERG abnormalities. Both variants have been previously reported in healthy controls and have minor allele frequencies on 1000 genomes of 0.023 and 0.013, respectively.¹⁰ The c.526C>T variant was identified in a RCD patient hemizygous for a novel RPGR splice site mutation (identified in the same exome sequencing experiment) and segregation analysis found the same heterozygous CRX change in the mother, who had normal acuity, fundus examination and electrophysiology at the age of 35. The c.127C>T variant predicted to be damaging was identified in a macular dystrophy patient on exome-analysis, and subsequently also identified in her older brother and mother, both of whom are asymptomatic but unavailable for further examination. Age of onset in this patient was 22 years old with visual acuity at last review age 27 of 1.0 logMAR each eye (20/200 Snellen). 

Coding CRX variants previously reported in the 1000 genomes project (n = 2577 exomes) comprise 53 single base substitutions (32 nonsynonymous), 17 frameshift indels (1 in exon 2 and the rest spread throughout exon 4) and one splice variant. In contrast, in the larger NHLBI EVS database (n = 6503 exomes), we found 38 single base substitutions (23 nonsynonymous) but no frameshift indel and no splice altering variant. To understand the difference in the number of frameshift calls, we analyzed a set of 2700 whole exomes of nonretinal dystrophy patients (UCL consortium), using a recently developed joint calling strategy (haplotype caller/gVCF pipeline implemented in GATK release 3.1, in the public domain at https://www.broadinstitute.org/gatk/) to minimize spurious calls. In that dataset, we identified 21 single base substitutions (13 nonsynonymous), one splice altering variant, and no frameshift indels. This comparison suggests that the 1000 Genomes frameshift coding indels may be spurious. 

A quantitative analysis was performed of the mutations by plotting the mutation position against age of presentation. A more severely affected subset of patients could be expected to have a lower median age of presentation than the remainder of the cohort. Such an analysis would be expected to demonstrate critical gene regions after which severity might change, thus exposing clinically significant functional domains.⁴⁶ On plotting the position of mutation against age of presentation (Supplementary Fig. S1) there was no evident position after which severity differed, nor was there a correlation between position and severity (Spearman correlation coefficient r_s = 0.093, P = 0.787). A comparison of median age of presentation in those with homeodomain missense mutations versus frameshifting/PTC mutations showed no statistical difference (Mann-Whitney test, U = 7.00, P = 0.634). 

Forty-three mutations of possible pathogenicity have been previously reported (Supplementary Table S3). A further four variants were excluded from further analysis for the following reasons: the first was from a single report of a mutation in exon 2, c.24dupG (p.Pro9Alafs*61) in a patient with LCA, but the mutation did not segregate with disease and the patient in question also had severe bilateral sensorineural hearing loss, a feature not otherwise reported with CRX mutations²⁹; two mutations, c.720_742dup23 (p.Gln248Profs*19) and c.753delC (p.Ser252Profs*119), are part of a screen on a microarray are also excluded as they are unpublished and no further information is provided as to patient phenotype³⁸; the fourth, c.351dupC (p.Lys118Glnfs*56) is also unpublished.¹⁵ 

Analysis of the 43 remaining likely pathogenic mutations indicates a pattern of missense mutations in exon 3 and frameshifting or stop mutations in exon 4 (Fig. 4). There are two exceptions to this, c.344G>A (p.Arg115Gln) located early in exon 4 and reported in a single patient with limited phenotype and segregation data; and c.887T>G (p.Phe296Cys), located in the extreme C terminal and reported in a single family with CORD that is predicted to be damaging.^31,45 There is a single report of a whole exon deletion of CRX, identified in a sibling pair with LCA who had biallelic disease with a compound heterozygous missense mutation on the other allele.³³ The CRX homeodomain is highly conserved throughout species (Supplementary Fig. S2) with all reported homeodomain mutations arising within highly conserved residues (Supplementary Fig. S2). Analysis of the homeodomain between human paralogues reveals that four reported mutations arise in residues that are not conserved or only partially conserved, p.Glu42Lys reported to cause LCA, p.Ala56Thr reported to cause LCA, p.Asp65His reported to cause RCD in a homozygous state and p.Lys88Asn reported to cause LCA. All mutations found within the homeodomain that are associated with LCA, therefore alter amino acid residues that are not conserved between paralogue human proteins. 

Figure 4

 

Schematic diagram of CRX structure with all published likely pathogenic mutations and seven novel mutations from this paper. Mutations in bold with # for patients in this paper.

 

MD, macular dystrophy; hom, homozygous; het, heterozygous

Discussion

This series of 18 patients with retinal dystrophy consequent on CRX mutations represents the largest report to date and identifies a new associated macular dystrophy phenotype. A group of patients with an autosomal dominant macular dystrophy presenting in their third to fifth decades is described. There was documented progression of the macular disease. Two cases were asymptomatic at presentation. 

At the severe end of the CRX disease spectrum are the LCA cases presenting in infancy with severe loss of vision. Rod–cone dystrophy presented within the first decade of life, CORD in the second to third decades and macular dystrophy in adulthood, with a range of 35 to 53 years. The macular dystrophy was characterized by a ‘bulls-eye' appearance similar to that observed in other macular dystrophies including the autosomal dominant form associated with PROM1.⁴⁷ Variable expressivity is also a feature of PROM1 with some patients manifesting generalized retinal dysfunction. 

Macular atrophy was present in 14 of 18 cases. Also commonly found was a disrupted ISe band on OCT, identified in all investigated patients. Electrophysiology further characterized the phenotypes demonstrating a normal ERG in all macular dystrophy cases confirming that the dysfunction was confined to the macula. Two cases with macula dystrophy have normal ERGs, age 60 and 67 years, indicating no evidence for peripheral retinal dysfunction developing with time. In families 17489 and 18280 who cosegregate macular dystrophy with generalized retinal dysfunction, it is noted that the older affected family members have the mildest phenotype. 

There are four exons in CRX, the first noncoding, producing a 299 amino acid protein. The protein has strong homology to OTX1 and OTX2 in three regions; the homeodomain at residues 39-99, which binds to DNA, the 13 residue WSP motif at residues 158-170, which is of unknown function, and the OTX tail at residues 284-295, a specific carboxyl terminus motif, of unknown function.¹ Broadly, the mutations that arise in the homeodomain are missense mutations; those in the remainder of the last exon, exon 4 are frameshifting with no mutation reported within the OTX tail (Fig. 4). In all cases, it is predicted that abnormal protein is produced, as avoidance of nonsense mediated decay would be predicted for the premature terminations according to the classical rules for this phenomenon.⁴⁸ The significance of the two distinct classes of mutation in the different protein regions is not clear. As noted, no clear genotype–phenotype association could be deduced from the complete data set or from the cases newly reported here. 

All but one reported homeodomain mutation arise within the alpha helices that are important in binding to the major groove of DNA via several hydrogen bonds, the exception being a homozygous mutation, c.193G>C; p.Asp65His.^13,16 One homeodomain mutation associated with LCA, c.264G>T; p.Lys74Asn has been demonstrated to disrupt DNA binding in a molecular model as well as interfere with the normal function of co-expressed transcription factors such as NRL in vitro, by a postulated dominant-negative effect.¹⁶ The homeodomain mutations arise in residues fully conserved in CRX between species but not fully conserved in human paralogues of the CRX protein (Supplementary Figs. S2, S3). Interestingly, all LCA associated mutations were located in nonconserved residues in these paralogues suggesting importance of these residues in CRX-specific function. 

Heterozygous knockout mice (+/−) have no retinal phenotype at 6 months of age, whereas homozygous knockout mice (−/−) have severe loss of rod and cone function, but neither types model the dominant human disease.⁴⁹ A spontaneous mouse mutant, Crx^Rip/+ was recently identified in which a dominant frameshifting mutation on one allele produced a truncated protein and a phenotype similar to congenital, blinding, dominant CRX retinal dystrophy in humans.⁵⁰ There is a lack of normal photoreceptor differentiation in the Crx^Rip/+ mouse, with arrested development at an early stage and inactive and immature photoreceptors identified histologically. There were two main mechanisms by which this mutant caused disease: firstly by a lack of normal DNA binding by the CRX mutant compared with wild type leading to loss of normal transactivation of photoreceptor genes and secondly by a dominant negative effect of this mutant allele, whereby the normal dimerization of the CRX protein is affected. The mutant CRX blocks the binding of OTX2 to the Nrl promoter preventing transactivation of Nrl and in turn the transactivation of essential rod and cone gene expression. Ectopic Nrl expression in the Crx^Rip/+ mouse partially rescued the poorly differentiated rod photoreceptor precursors. Recently reported knock-in mouse models have severely reduced retinal function and demonstrated the association of CRX expression level on disease severity as well as the ability of the mutant allele to interfere with wild-type function.⁵¹ One model with a truncating mutation had more severe disease than another, with a missense mutation mirroring the reported human phenotypes for those specific mutations. These mouse models may provide excellent opportunities for further functional analysis of CRX-related disease and the investigation of potential treatments. The evidence to date, including this paper and the mouse models, suggests that haploinsufficiency of CRX is not, in itself, disease-causing. Instead, an allele has to be both nonfunctional and expressed (hence, the lack of PTCs early in the gene), such that the abnormal protein partly abrogates the normal one expressed from the other allele. 

Seven novel mutations are reported herein; one missense in exon 3 with one PTC and five frameshifting mutations in exon 4. Four families demonstrated the large degree of clinical variability that can occur between those sharing the same CRX mutant allele. This heterogeneity may be due to (1) the influence of polymorphisms in the CRX promoter region, (2) polymorphisms in co-expressed transcription factors such as NRL, (3) the impact of environmental factors, (4) stochastic factors, that is small perturbations of CRX function causing larger and later effects on the degree of degeneration, (5) variable levels of expression of the mutant allele, which in a mouse model has been correlated with variable severity, or (6) variable levels in expression of the wild-type allele, which in PRPF31-related disease has been correlated with variable penetrance.^51,52 Study of the promoter region and of RNA transcripts may help to elucidate this further. 

A further novel mutation, c.127C>T, in a patient with macular dystrophy did not segregate with known eye disease in the family, with the asymptomatic mother and brother heterozygous for the variant. Given the examples in this series of asymptomatic presentation, it cannot be excluded that these other family members are affected. Unfortunately, examination and retinal imaging was not possible in the family members to clarify this issue. The patient presented at age 22, younger than the other macular dystrophy patients in this series, and also has a more severe reduction in visual acuity than the other macular dystrophy patients. 

Only 3 of 11 reported families had a family history of possible retinal dystrophy in antecedent generations prior to this study. Parental sequencing in three families confirmed de novo mutations. DNA was unavailable from antecedents in the remaining five families, although the observed pedigrees make it possible that these mutations may also have arisen de novo. Alternatively, given the relatively mild presentation of the macular dystrophy phenotype, disease in antecedents might have been unreported. These eight families highlight the difficulties in genetic counseling in patients with apparent simplex retinal disease. The possibility of de novo mutation, or mild unreported disease in antecedents, suggests a greater risk in subsequent generations than if autosomal recessive inheritance was assumed. It would also affect the risk to other siblings. Clinicians should therefore have a low threshold for screening CRX in patients presenting with any of the phenotypes consistent with those presented here. 

Acknowledgments

The authors thank the cooperation and help provided by the family members in this study. They also thank the colleagues who referred patients to them and to those who contributed to the assembly of the retinal dystrophy panels particularly Panos Sergouniotis, Naushin Waseem, Ravinder Chana, Beverley Scott, Sophie Devery, Genevieve Wright, Alan Bird, Michel Michaelides, Robert Henderson, Arun Dev Borman, Kaoru Fujinami and Alice Davidson. Also, Professor Veronica van Heyningen for her support and encouragement and the UCL exome consortium for access to control data. 

Presented at the annual meeting of the International Society for Genetic Eye Diseases & Retinoblastoma, Ghent, Belgium, August 2013. 

Supported by grants from The National Institute for Health Research and Biomedical Research Centre at Moorfields Eye Hospital and the UCL Institute of Ophthalmology (London, UK), The Foundation Fighting Blindness (Owings Mills, MD, USA), Fight For Sight (London, UK), Moorfields Eye Hospital Special Trustees (London, UK), Macular Society (Andover, UK), RP Fighting Blindness (Buckingham, UK), and Rosetrees Trust (Edgware, UK). 

Disclosure: S. Hull, None; G. Arno, None; V. Plagnol, None; S. Chamney, None; I. Russell-Eggitt, None; D. Thompson, None; S.C. Ramsden, None; G.C.M. Black, None; A. Robson, None; G.E. Holder, None; A.T. Moore, None; A.R. Webster, None