A Novel GUCY2D Mutation, V933A, Causes Central Areolar Choroidal Dystrophy

Abstract

Purpose.: To identify the genetic cause of central areolar choroidal dystrophy (CACD) in a large Northern Irish family.

Methods.: We previously reported linkage of the locus for CACD in this family to an interval of approximately 5 cM on chromosome 17p13 flanked by polymorphic markers D17S1810 and CHLC GATA7B03. We undertook sequence capture, massively parallel sequencing and computational alignment, base-calling and annotation to identify a causative mutation. Conventional sequencing was used to confirm the results.

Results.: Deep sequencing identified a single-base substitution in guanylate cyclase 2D, membrane (retina-specific) gene (GUCY2D). The novel mutation segregated with the disease phenotype and resulted in substitution of valine by alanine at position 933, within the catalytic domain of the protein. It altered a motif that is strongly conserved in a large number of distantly related proteins across several species and was predicted to have a damaging effect on protein activity.

Conclusions.: Mutations in GUCY2D have previously been associated with dominant cone–rod dystrophies (CORD6) and recessive forms of Leber's congenital amaurosis. This is the first report of a GUCY2D mutation causing CACD and adds to our understanding of genotype–phenotype correlation in this heterogeneous group of choroidoretinal dystrophies.

Introduction

Central areolar choroidal dystrophy (CACD) is a rare inherited disorder of the retina characterized by progressive central visual loss. ^1–4 Electrophysiological abnormalities in affected people can be detected before either the onset of symptoms or the appearance of visible clinical features. Paracentral scotomas and reduced visual acuity may be apparent by the third or fourth decade, when parafoveal pigmentary changes may be visible on examination. Over time, a depigmented macular lesion develops, with widespread photoreceptor loss. Central vision is severely impaired by the seventh decade. The retinal pigment epithelium, choriocapillaris and overlying retina become atrophic in the affected lesions. 

Members of a family with three living generations presented with mild loss of central vision in their 30s and 40s. Color vision was normal in the early stages, there was no photophobia, and on examination, no temporal pallor of the optic disc. Fluorescein angiography demonstrated small areas of parafoveal retinal pigment epithelium atrophy in the second generation and also RPE pigment clumping. However, there was no evidence of a bull's eye maculopathy on angiography. Subsequently a large macular area of well circumscribed atrophy of the RPE, choriocapillaris and neural retina developed and enlarged over time (Fig. 1). 

Figure 1. 

 

Color fundus images of proband in second generation (A) and two of his sons in the third generation (B, C). Images show a well-circumscribed area of atrophy of the retinal pigment epithelium, choriocapillaris, and neural retina. This was larger in the older members of the family (A, B). Fluorescein angiogram of a patient from the third generation (D) demonstrating discrete areas of parafoveal retinal pigment epithelium atrophy and pigment clumping. Late angiography frame of proband in the second generation (E) highlights the central area of choroidal atrophy.

We previously mapped the disease locus in this family with dominantly inherited CACD to a 5-cM interval on chromosome 17p13 between microsatellite markers D17S1810 and GATA7B03 (now named rs4331391). ^5,6 We have now performed massively parallel sequencing on the entire linkage region and report that CACD in this family is caused by a novel mutation (V933A) within the guanylate cyclase 2D, membrane (retina-specific) gene (GUCY2D). This gene is expressed predominately within the photoreceptors, where it is responsible for synthesis of cGMP, which is vital for recovery of the dark state after light activation of the visual signal transduction cascade. ⁷  

Methods

Patients and Clinical Studies

The family affected by autosomal dominant CACD has been reported previously. Diagnosis was based on previous clinical ophthalmic examination, including stereo fundus photography, fluorescein angiography, and electrophysiological testing; the findings in the affected cases agreed with Krill's classification of CACD. ² One affected individual independently received the same diagnosis after investigation in a US retinal center. Biological samples were collected with informed consent, in accordance with the tenets of the Declaration of Helsinki and with local ethics committee approval. Anonymized DNA samples from this historical project are part of the Centre for Public Health, QUB, DNA bank, which was approved by the Office for Research Ethics Committees, Northern Ireland. 

Molecular Methods

The entire linkage region on chromosome 17 between 5,030 and 8,199 kb of National Centre for Biotechnology Information (NCBI) build 37 (from USP6 to SLC25A35) was targeted by a custom sequence capture array (Nimblegen, Madison, WI). Five DNA samples were enriched 252- to 432-fold by sequence capture: from one affected individual, one pooled sample of four affected individuals, and three unrelated control samples containing DNA from two, two, and three individuals, respectively. Following library preparation, captured samples were sequenced on a HiSeq 2000 (Illumina, San Diego, CA) with 50 base pair (bp) single-end reads (GATC Biotech, Konstanz, Germany). 

Sequence reads were aligned to the National Center for Biotechnology Information (NCBI) v37 reference sequence with the Burrows-Wheeler Aligner (BWA) 0.5.9 aln algorithm. ⁸ Sorting, indexing, and removal of duplicate reads were performed with SAMtools 0.1.14. ⁹ Functions of Genome Analysis Toolkit (GATK) were used to recalibrate, realign, and to call polymorphisms (Unified Genotyper). Functional effects were annotated using snpEff 2.0.5 with the GRCh37.64 reference set. ¹⁰ Annotated single nucleotide polymorphism (SNP) data were searched for instances where reads from both the affected individual and affected pool showed a functionally significant variant that was absent in the unaffected controls. Exon 15 of GUCY2D was also sequenced conventionally using dye terminal chemistry (ABI, Warrington, UK) and detection on an ABI 3130 xl genetic analyzer. DNA was amplified by PCR (forward primer 5′-GGTGACAAGAGGCAATCGCTTCG-3′; reverse primer 5′-TAAAGAGGGAGATGGGCTGGAGC-3′) and sequenced with the forward primer using standard protocols. The predicted effect of variation on function of GUCY2D was assessed using Sorting Intolerant from Tolerant (SIFT) ¹¹ (through SNPnexus ¹² ) and PolyPhen-2. ¹³  

Results

All 146 annotated genes located within the CACD linkage region ¹⁴ on chromosome 17 were searched for potentially functional mutations that may cause the phenotype. Average sequencing read depth ranged from 37 to 58 in the five samples, with 83% to 91% of bases covered by at least 15 reads in five samples. Four non-synonymous coding variants were present only in DNA from CACD cases. A novel variant was found in GUCY2D, which encodes retinal guanylate cyclase, at chromosome 17 position 7,918,674 bp, where a T to C transition resulted in a missense mutation from valine to alanine at amino acid 933 (V933A; Fig. 2). The mutant C allele was found in 28% (17/60) and 44% (20/45) of reads from the two samples of CACD DNA at this base, and the wild-type T found in all 175 reads from the pooled control samples (comprising seven individuals). Transmission of V933A on the affected CACD haplotype was confirmed by conventional sequencing of exon 15 of GUCY2D in a further four family members. This mutation was present in heterozygous state in DNA sequenced from the two affected family members in which recombination defined the CACD linkage region (individuals II6 and III3 from Ref 6). It was not found in an unaffected sibling (III4) who carried the same unaffected haplotype as her affected sister III3, confirming the presence of the novel mutation on the affected haplotype. It was not found in any of 1092 individuals in the March 2012 Variant Call Format (VCF) release of the 1000 genomes project. ¹⁵  

Figure 2. 

 

Multiple alignment of massively parallel sequence reads from exon 15 of GUCY2D in DNA pools of unaffected controls (top 3 panels), a single affected CACD case (fourth panel), and a pool of four affected CACD cases of the same family (fifth panel). All nucleotides that vary from the reference sequence, shown below, are indicated. Read depth and heterozygosity are designated at the top of each subpanel.

The other three variants that were identified were known polymorphisms present in dbSNP: rs114991387 and rs2304449 in the ubiquitin specific peptidase 6 gene (USP6), both of which are found at a frequency of approximately 0.01 in Europeans; and the less common rs142059605 in the cytochrome b5 domain containing 1 gene (CYB5D1), which is present at a frequency of 0.006 in Europeans. No variants in non-coding RNA genes segregated with the CACD cases. 

Analysis of the V933A mutation by SIFT and by Polyphen-2 (both based on conservation of amino acids among related proteins) suggested that the amino acid change within the catalytic domain of the enzyme would be functionally damaging (Table 1). Several relatively common coding polymorphisms in GUCY2D were found to be tolerated (Table 2). Conversely, most of the rare missense mutations that are thought to be pathogenic, contributing to the Leber's congenital amaurosis 1 (LCA1) or cone–rod dystrophies (CORD6) phenotypes, were not tolerated (Fig. 3; Table 1). Lack of tolerance was assumed for pathogenic frameshifting and nonsense mutations. 

Figure 3. 

 

Exonic and domain structure of the GUCY2D gene showing the location of damaging missense and in-frame indel mutations that cause LCA (in red), COD or CORD (in blue), and CACD (in green). Frameshifting mutations are not shown. I949T causes autosomal recessive CORD, ³³ whereas all other COD, CORD, and CACD mutations cause dominantly inherited disease.

Table 1. 

 

Exonic Mutations in GUCY2D That Cause CORD6, COD, LCA1, or CACD

Chromosome 17 Position	Wild-Type Allele	Mutant Allele	Mutation*	Phenotype	Inheritance Pattern	SIFT Prediction†
7,906,368	G	A	M1I	LCA	AR	Damaging
7,906,368	G	C	M1I	LCA	AR	Damaging
7,906,417–7,906,464	–	48 bp INS	G18–L33 dup	LCA	AR	?
7,906,486	C	T	L41F	LCA	AR	Tolerated
7,906,679	G	A	C105Y	LCA	AR	Damaging
7,906,752	C	A	N129K	LCA	AR	Tolerated
7,907,422	T	C	L325P	LCA	AR	Damaging
7,907,422	T	G	L325R	LCA	AR	Tolerated
7,907,385	C	T	R313C	LCA	AR	Tolerated
7,909,706	A	G	Y351C	LCA	AR	Damaging
7,911,300	C	T	R540C	LCA	AR	Damaging
7,912,849	T	C	F565S	LCA	AR	Damaging
7,917,236	C	T	R768W	LCA	AR	Damaging
7,917,318	G	A	R795Q	LCA	AR	Damaging
7,918,078	C	T	P858S	LCA	AR	Damaging
7,918,676	G	C	A934P	LCA	AR	Damaging
7,918,713	C	T	A946V	LCA	AR	Damaging
7,918,738	T	C	L954P	LCA	AR	Damaging
7,918,803	G	T	R976L	LCA	AR	Damaging
7,919,099	C	T	R995W	LCA	AR	Damaging
7,919,141	A	C	M1009L	LCA	AR	Damaging
7,919,257	A	C	H1019P	LCA	AR	Damaging
7,919,256–7,919,265	CACGTGAACT	TGTC	Del 1019–1022; INS VT	LCA	AR	Damaging
7,918,017 and 7,918,018	GC	CA	E837D+R838S	CORD	AD	Damaging
7,918,018	C	T	R838C	CORD	AD	Damaging
7,918,018	C	G	R838G	COD	AD	Damaging
7,918,019	G	C	R838P	CORD/COD	AD	Damaging
7,918,019	G	A	R838H	CORD/COD	AD	Damaging
7,918,722	T	C	I949T	CORD	AR	Damaging
7,918,674	T	C	V933A	CACD	AD	Damaging

 AR, autosomal recessive; AD, autosomal dominant; INS, insertion; dup, duplication; Del, deletion.

*   Small frameshifting deletions, nonsense, and splicing mutations that are predicted to result in the absence of protein product in LCA patients are excluded.

†   SIFT predicts approximately 70% of damaging mutations in human variation databases correctly and may have a 20% false-positive error rate. ¹¹

Table 2. 

 

Exonic SNPs in GUCY2D and Frequencies in European, Asian, and African Populations in the 1000 Genomes Project

SNP	Chromosome 17 Location	Alleles	AA Position	Wild-Type AA	Mutant AA	SIFT Prediction	EUR	ASN	AFR
rs9905402	7,906,426	T/C	21	W	R	Tolerated	0.02	0.02	0.34
rs61749665	7,906,519	G/T	52	A	S	Tolerated	0.33	0.67	0.12
NA	7,910,844	A/G	522	K	E	Tolerated	0.01	0.01	0.00
rs28743021	7,912,879	C/T	575	P	L	Tolerated	0.00	0.00	0.07
rs34598902	7,915,912	C/T	701	P	S	Tolerated	0.02	0.23	0.03
rs8069344	7,917,279	T/A	782	L	H	Tolerated	0.11	0.00	0.45
rs2816	7,923,564	T/C	3′ UTR	–	–	–	0.52	0.06	0.03
rs138596240	7,923,609	A/G	3′ UTR	–	–	–	0.02	0.00	0.02

 EUR, European; ASN, Asian; AFR, African; –, no insertion.

Discussion

Our investigation of the genetic cause of central areolar choroidal dystrophy in a large Irish family revealed a functionally damaging mutation, V933A in exon 15 of GUCY2D, within the known linkage region. Because of its role and expression, we previously considered this gene to be a strong candidate gene for carrying the CACD mutation. In 1997, the putative exons of GUCY2D were sequenced in the same CACD family, but the V933A mutation was not identified (using early sequencing technology), causing this gene to be excluded from further study at that time. We therefore recently revisited the project using modern massively parallel sequencing and bioinformatics methods, which offered a high degree of likelihood of finding the causal mutation and had the advantage of being independent of prior knowledge of gene function or the functional significance of polymorphisms. This method also offered the possibility of identifying structural polymorphisms that might otherwise be missed by conventional sequencing. 

GUCY2D is a membrane-bound homodimer of two monomers arranged in opposite orientation to form a cyclase with two active sites within a single cleft. ¹⁶ Exons 2–4 of GUCY2D encode the extracellular part of the protein, and exon 5 encodes the hydrophobic domain, which spans the membrane. ¹⁷ The intracellular part is usually categorized into kinase-like (exons 6–11), dimerization (exons 12–14), and catalytic (exons 15–19) domains, although mutations affecting the dimer interfaces could also affect functional activity. ¹⁸ The enzyme converts guanosine 5′-triphosphate (GTP) to cGMP, the second messenger responsible for regulating the phototransduction cascade in the retina by controlling flow through sodium and voltage-gated calcium channels.Photoactivation leads to breakdown of cGMP to 5′-GMP by phosphodiesterase, sequentially causing closure of sodium channels, hyperpolarization, and closure of calcium channels. When Ca²⁺ levels fall, GUCY2D replenishes cGMP levels, permitting reopening of the ion channels and recovery of the dark state. This is facilitated by guanylate cyclase–activating protein, which releases its bound Ca²⁺ and is then able to activate guanylate cyclase. ¹⁹  

Rare mutations affecting GUCY2D, ^20–36 its chief activating protein, GUCA1A, ³⁷ and a small number of additional retinally expressed genes give rise to phenotypes of cone or cone–rod dystrophies or Leber's congenital amaurosis. LCA is inherited in an autosomal recessive manner and presents as severe retinal dysfunction shortly after birth, caused by total deficiency or greatly reduced activity of guanylate cyclase protein. Many of the severe LCA mutations are small deletions or duplications, which can be predicted to cause frameshifts or splicing anomalies leading to nonsense-mediated decay of RNA from the mutant alleles. The known non-synonymous coding variants are restricted mainly to the extracellular, kinase-like, and catalytic domains. Extracellular domain mutations have normal enzyme activity, with the exception of M3I, which delays initiation of translation and diminishes activity approximately 8-fold. ³⁸ Heterozygote carriers of these mutations can show cone ERG abnormalities. ³⁹ The catalytic domain mutations that have been studied in detail tend to either impair basal activity or to lessen the increased activity upon stimulation by GCAP1 or GCAP2. ⁴⁰  

The importance of the variants listed in Table 1, which are predicted by SIFT to be tolerated, such as L41F, N129K, and R313C, is somewhat uncertain. Many of the apparent LCA-causing mutations have been identified in consanguineous families, and it is therefore possible that some of these represent rare polymorphisms in populations of African descent. 

With a single exception, which we believe to be erroneous, ²² all reported cases of dominantly inherited CORD or cone dystrophy (COD) due to GUCY2D carry a mutation in exon 13 affecting codon 838, which is likely to affect the dimerization of guanylate cyclase. Codon 838 is a hot spot for mutations, which arise de novo in most families, and which therefore do not share common haplotype backgrounds. Codon 838 variants have higher affinity for GCAP1, increased residual activity, and possible failure to inactivate at high Ca²⁺ levels. ⁴¹  

We report a novel mutation affecting the catalytic domain of guanylate cyclase in a family with CACD linked to chromosome 17p. Functional studies will be required to investigate the effect of the V933A mutation on catalytic activity. The CACD phenotype is distinctive, though it shares features in common with COD and CORD. In patients with CACD, cones are affected at an earlier stage than rods, though the phenotype is less severe and of later onset than in most COD and CORD families with GUCY2D mutations. In some aspects, the V933A mutation may be more comparable to a severe expressed LCA mutation, which is penetrant in heterozygotes, who exhibit less severe disease at a later age. This is supported by the report of a mutation at the adjacent amino acid (A934P) in one consanguineous family with LCA. ²¹ For non-synonymous mutations, the enzymatic activity of dimers composed of one mutant and one wild-type monomer may hold the key to whether mutations cause dominant or recessive forms of retinal degeneration. Variants in GUCY2D range from common SNPs of no pathologic significance to non-synonymous mutations, which can affect the activity and regulation of guanylate cyclase by a variety of mechanisms, and null mutations that result in the absence of protein. The range of diseases caused by GUCY2D mutations (LCA, COD, CORD, and CACD) reflects the combination of mutational variants and dosage. Further molecular and clinical studies are necessary to understand the role of mutations fully and to better correlate genotype and phenotype in these closely related pathologies. These are vital steps towards ultimately informing personalized treatment regimens to prevent loss of sight. 

Acknowledgments

We thank the family who participated in the study for their longstanding interest and the clinicians who recruited and cared for the family. Bioinformatics work was conducted on the QUB High Performance Computing Dell cluster. 

Footnotes

 Supported by grants from the Public Health Agency (Northern Ireland) (DTB) and Guide Dogs for the Blind (WM).

Footnotes

 Disclosure: A.E. Hughes, None; W. Meng, None; A.J. Lotery, None; D.T. Bradley, None