Spectrum of the ABCA4 Gene Mutations Implicated in Severe Retinopathies in Spanish Patients

Abstract

purpose. The purpose of this study is to describe the spectrum of mutations in the ABCA4 gene found in Spanish patients affected with several retinal dystrophies.

methods. Sixty Spanish families with different retinal dystrophies were studied. Samples were analyzed for variants in all 50 exons of the ABCA4 gene by screening with the ABCR400 microarray, and results were confirmed by direct sequencing. Haplotype analyses were also performed. For those families with only one mutation detected by the microarray, denaturing (d)HPLC was performed to complete the mutational screening of the ABCA4 gene.

results. The sequence analysis of the ABCA4 gene led to the identification of 33 (27.5%) potential disease-associated alleles among the 60 patients. These comprised 16 distinct sequence variants in 25 of the 60 subjects investigated. For autosomal recessive cone–rod dystrophy (arCRD), we found that 50% of the CRD families with the mutation had two recurrent changes (2888delG and R943Q). For retinitis pigmentosa (RP) and autosomal dominant macular dystrophy (adMD), one putative disease-associated allele was identified in 9 of the 27 and 3 of the 7 families, respectively.

conclusions. In the population studied, ABCA4 plays an important role in the pathogenesis of arCRD. However, mutations in this gene are less frequently identified in other retinal dystrophies, like RP or adMD, and therefore it is still not clear whether ABCA4 is involved as a modifying factor or the relationship is a fortuitous association.

The ATP-binding cassette transporter gene superfamily encodes membrane proteins involved in translocation of substrates across membranes. Specifically, the ABCA4 (photoreceptor-specific ATP-binding cassette transporter 4) gene encodes for a retinal protein exclusively localized at the rims of the outer segments of rod and cone photoreceptors (MIM 601691: NM_000350; GenBank: U88667; http://www.ncbi.nlm.nih.gov/Genbank; provided in the public domain by the National Center for Biotechnology Information, Bethesda, MD). ^{1 2}  

Mutations in the ABCA4 gene have been associated with autosomal recessive Stargardt disease (STGD) and have been implicated in several retinal phenotypes, such as retinitis pigmentosa (RP), cone–rod dystrophy (CRD), fundus flavimaculatus (FFM), and age-related macular dystrophy (AMD). ^{3 4 5 6} These clinical manifestations depend on the nature of the ABCA4 mutation and on the remaining protein activity. Thus, a grading system explaining these phenotypes has been proposed: Two null mutations lead to RP, two severe mutations to arCRD, two mild or moderate mutations to STGD/FFM and one milder heterozygotic mutation to AMD. ^{4 5 7 8 9} This grading system seems reasonable, since all these diseases affect primarily the central vision. However, the implication of the gene in RP has been discussed, since an RP-like fundus appearance and the presence of night blindness are not enough criteria to classify those patients as having RP. The progression from early central vision loss to complete loss of vision (both central and peripheral), in addition to night blindness and an RP-like fundus appearance in late stages, should be considered severe progressive CRD. In CRD, the cone degeneration appears early in life with central involvement of the retina, followed by a rod degeneration several years later, and can be misdiagnosed as macular dystrophy (MD) in early stages. CRD is caused by two compound mutations of the severe type that affect the ABCA4 gene. ¹⁰  

The ABCR400 microchip for ABCA4 mutations described by Jaakson et al. ¹¹ together with the denaturing (d)HPLC technique allowed us to evaluate the ABCA4 gene in arCRD and typical arRP cases in the Spanish population and to test the grading system for these diseases. Patients affected with adMD were also analyzed, as it has been proposed that the ABCA4 and STGD3 genes may interact with each other to increase the severity of the macular phenotype. ¹²  

Therefore, in this work we describe the mutation spectrum found in the ABCA4 gene in Spanish patients with different retinal dystrophies. 

Methods

Subjects and Diagnostic Criteria

A total number of 60 families were studied. The recruitment of patients and relatives was performed through six research groups involved in the Spanish Retinal Dystrophy Investigation Network (EsRetNet). This investigation adhered to the tenets of the Helsinki Declaration and was approved by all the participant institutions. Informed consent was obtained from all the adults and from the children’s parents or tutors after the nature and possible consequences of the study had been explained. 

Ophthalmic and electrophysiologic examinations were performed according to preexisting protocols, consisting of the history of the patient and his or her family, funduscopic examination after pupillary dilation, computerized testing of central and peripheral visual fields, and visual acuity testing with the best correction. Electrophysiological assessment included a full-field ERG, incorporating the protocols recommended for vision testing by the International Society for Clinical Electrophysiology of Vision and Color. ^{13 14} Diagnoses of adMD, CRD, and RP were based on the following criteria: (1) The diagnosis of adMD was determined according a combination of autosomal dominant inheritance and ocular characteristics of macular dystrophy: bilateral visual loss and a finding of generally symmetrical macular abnormalities on ophthalmoscopy. The age of onset was variable, but in most patients was during the first two decades of life. (2) The diagnosis of CRD was based on the following criteria: initial complaints of blurred central vision without a history of night blindness, poor visual acuity (typically 20/100 or worse, with progressive decline from an early age), impairment of color vision, funduscopic evidence of atrophic macular degeneration, peripheral disturbances including pigment clumping and/or pigment epithelial thinning, and greater or earlier loss of cone than rod ERG amplitude. (3) RP was diagnosed in patients who developed night blindness early in life with progressive constriction of the visual field. Signs on funduscopic examination included attenuated retinal vessels, depigmentation of the RPE, intraretinal bone spicule pigmentation, and a waxy pallor of the optic disc. The ERG responses had to be decreased in a rod–cone pattern or nonrecordable when the disease had reached its end stage. ^{15 16}  

Molecular Methods

Blood was collected by venipuncture, and genomic DNA was isolated by the salting-out method. All the exons of the ABCA4 gene were PCR amplified as described previously ¹⁷ and used in the primer extension reaction (APEX) on the ABCR400 microarray, as described elsewhere. ¹¹ The 50 exons of the ABCA4 gene, including the intron–exon junctions, were amplified by PCR to confirm the results obtained from the microarray. These fragments were electrophoresed in a 3% agarose gel and purified by using a DNA extraction kit (QIA-quick Gel Extraction Kit; Qiagen, Hilden, Germany). 

The sequencing reaction was performed with four-dye-terminator cycle sequencing ready reaction kit (dRhodamine DNA Sequencing Kit; Applied Biosystems, Foster City, CA). Sequence products were purified through thin columns (Sephadex G-501; Princetown Separations, Adelphia, NJ) and resolved in a sequencer (Prism 3100; Applied Biosystems). 

Haplotypes were constructed using four microsatellite markers flanking the ABCA4 gene (TEL-D1S435-D1S2804-D1S2868-ABCA4-D1S236-CEN). After amplification by PCR, fluorescence-labeled products were mixed and electrophoresed (Prism 3100; Applied Biosystems). 

dHPLC sample screening was performed on a DNA fragment analysis system (WAVE; Transgenomic, Omaha, NE). The PCR products were loaded (5 μL) on a C₁₈ reversed-phase column (DnaSep column; Transgenomic). Hetero- and homodimer analysis was performed with an acetonitrile gradient formed by mixing buffers A and B (WAVE Optimized; Transgenomic). The flow rate was 0.9 mL/min, and DNA was detected at 260 nm. For each DNA region, dHPLC conditions were established by a triple analysis 1° to 3°C above and below the mean melting temperature predicted by software simulation. 

Because dHPLC does not usually differentiate the wild-type from the homozygous mutant sample, all unknown samples were mixed in a 1:1 proportion with a control sample at the end of each PCR session. Before dHPLC analysis, heteroduplexes were formed by denaturing the PCR product at 95°C for 5 minutes and cooling it to room temperature. 

A splice site scoring program (http://www.fruitfly.org/seq_tools/splice.html) was used to evaluate the effect of intronic mutations. 

Results

Of the 60 families recruited, 7 had diagnosed adMD, 26 arCRD, and 27 arRP. 

The sequence analysis by the ABCR400 microarray led to the identification of 32 potential disease-associated alleles. These comprised 15 distinct sequence variants in 25 of the 60 subjects investigated. Except for two deletions, 2888delG and 5041del 15pb, the remaining variants were all single-base substitutions. Of these substitutions, 12 were missense mutations and 1 involved a splice acceptor site (Tables 1 2 and 3) . 

When the dHPLC technique was applied to 15 patients, in whom only one variant was detected by the ABC400 microarray, we detected three changes, which are shown in bold in Tables 2 and 3 . Except for the change c.6147+2T>A for which scoring programs predict a change in the splicing process, the other variants did not produce a change at the protein level. 

In the adMD families, three mutated alleles were detected in the heterozygous state in three (42%) of the families. In family ADDM-59, a complex allele [G1961E; S2255I] was detected in the index patient, but not in her affected daughter, suggesting no cosegregation of the disease within the family. Indeed, haplotype analyses were not consistent with an autosomal dominant inheritance (Table 1) . 

For family ADDM-92, a mutation in the heterozygous state, I156V, was detected. Besides, a mild allele [R943Q] was detected in family ADDM-105. 

This combined study (ABCA400 microarray and dHPLC) identified both ABCA4 disease-associated alleles underlying arCRD in 8 of the cohort of 26 patients with arCRD (31%). Only one sequence variant was identified in 5 of the 26 arCRD subjects (19%), whereas no sequence variants were detected in the remaining 15. Thus, we detected at least one mutated allele in 13 of the 26 arCRD families, representing 50% of the total CRD cases. 

Allelic segregation analyses were performed in the eight arCRD families. Except for family ARDM-134, the disease-associated haplotypes cosegregated within the families. 

The 2888delG sequence variant was the most frequently mutated allele observed among arCRD patients (5/52, 11.5%). We identified two patients harboring this homozygous 2888delG change and two other patients with this mutation as a compound heterozygote (Table 2) . 

The clinical phenotype in one individual from family ARDM-79 changed from arSTGD (age of onset, 10 years) to arCRD (age of diagnosis, 26 years), during the course of this study. This woman initially presented typical ophthalmic findings (mild pallor of the optic disc, nonspecific maculopathy with pigment spots giving a grayish aspect to the fundus, visual field with paracentral scotoma, mildly attenuated retinal vessels, and dyschromatopsia) consistent with a diagnosis of Stargardt disease (STGD). At the second examination at the age of 26 years, the patient presented extinguished ERGs (both photopic and scotopic), reduction of peripheral visual field (almost absolute scotoma), attenuated retinal vessels, and bone spiculelike pigmentation, consistent with the diagnosis of CRD (Fig. 1) . 

When the 27 families with classic RP were analyzed, no mutations were found in 18 families, and in the remaining 9 families only one sequence variant was detected. Table 3shows the allele detected in these families (the polymorphisms detected by dHPLC and described for the first time are in bold). Thus, we detected almost one sequence variant in 33% of the cases of RP. 

Discussion

To date, several studies have described mutations in the ABCA4 gene regarding a variety of retinal dystrophies. ^{5 6 18 19 20} In the current study, we analyzed the involvement of the gene ABCA4 in three types of retinal dystrophies: adMD, arCRD, and RP. 

For the first condition, adMD, three mutated alleles were detected in three families of the seven studied. For family ADDM-59, segregation analyses of the mutated complex allele [G1961E; S2255I] did not support the pathologic role of this mutation in the family. In family ADDM-92, we detected the mutation I156V, which has been associated with a STGD recessive phenotype. ²¹ The R152X mutation (located in exon 5, close to I156V) was present in a family with dominant STGD that demonstrated genetic linkage to the STGD region on 6q. The patient in this family had onset of visual symptoms beginning at the age of 24 and rapid disease progression. In fact, it has already been demonstrated that the ABCA4 and STGD3 genes can interact with each other, increasing the severity of the macular phenotype. ¹²  

In family ADDM-105, the R943Q mutation was detected. This mutation has been described as a polymorphism because it has been found in normal populations. Expression studies ²² have demonstrated that this change produces a small but detectable reduction in the nucleotidase activity and nucleotide-binding affinity of the ABCA protein. Other studies have shown this variant to be associated with G863A, leading to a severer pathogenic state in humans. Therefore, we speculate about two possibilities: First, the R943Q change could be paired with a severer mutation not found in our study; or second, R943Q could have a modulating effect on another gene implicated in adMD, not discovered yet. As discussed before, it has been reported ¹² that ABCA4 and STGD3 genes may interact with each other to increase the severity of the macular degeneration phenotype. In the case of family ADDM-92, which had the I156V mutation, the clinical phenotype seemed to be severer than that in family ADDM105, which presented the mild allele R943Q. Because the cosegregation analysis could not be performed in these families, it would be interesting to analyze the STGD3 gene to investigate how these two genes may interact. 

For CRD, both homozygous and compound heterozygous mutations in the ABCA4 gene have already been reported. In the current study, we investigated 13 CRD families that showed at least one putative pathologic ABCA4 allele, which represent 50% of the families analyzed in our study. This percentage is higher than the 33% described by Klevering et al. ²³ although, according to their estimation, mutations in the ABCA4 gene could be present in at least 67% of our cohort of CRD families. 

We detected both disease-associated alleles in eight families (Table 2) . Among them, we also identified four families carrying homozygous mutations (ARDM-79, ARDM-86, ARDM-126, and RP-267); in two of them (ARDM-86 and ARDM-126), consanguinity was proven. The 2888delG variant was the most prevalent disease-causing allele among our patients with CRD, accounting for 30% of the alleles detected. The 2888delG leads to a frameshift that produces a stop codon, and therefore the encoded protein must be severely affected. For the [G1977S;R943Q] complex allele found in homozygosis, expression analysis of the G1977S mutation has determined that it causes the inhibition of ATPase by retinal, ²⁴ whereas R943Q leads to a reduced nucleotidase activity and nucleotide-binding capacity. ²² The affected patients had an age at onset between 9 and 10 years and described night blindness at approximately 18 years of age. In the two 2888delG compound heterozygous families, one of them showed the missense mutation L11P, which affects a conserved amino acid localized in the intracytoplasmic compartment, ⁸ as a second allele, and the other family harbored the L2060R mutation, which produces an alteration in the charge of the mutated amino acid that has been associated with the CRD phenotype. ²⁰ In this work, we found that the diagnosis of the proband from family ARDM-79 had changed from arSTGD to arCRD. Regarding this, in our patients, the 2888delG variant has been associated with arCRD (either in a homozygous or heterozygous pattern), and therefore we suggest that it must be considered a severe disease-associated allele. This fact is remarkable, as patients harboring this mutation could be misdiagnosed as having arSTGD in the early stage of the disease and later be shown to have the arCRD phenotype (Figs. 2 3) . 

The family ARDM 174 presented two mutations, one the missense mutation R1640W detected by the ABCA400 microarray and with an unknown effect in the ABCR function, and the other, the c.6147+2T>A, detected by dHPLC. This last mutation affects the splicing of the intron 44, and so the donor site disappears, as predicted by the splice-site scoring program. 

The last arCRD family studied also presented two missense mutations, namely T901A and R943Q, the latter described as reducing the ATPase activity in 40% and producing minimal defects in nucleotide binding, ²² being categorized as a mild mutation. 

In the remaining families, only one pathologic allele was detected. In all the cases, they were missense mutations (Table 1) , although two of them (R943Q and S2255I) are still controversial: R943Q reduces the ATPase activity, and S2255I is supposed to have limited pathogenicity. Thus, such alleles would not be expected to cause disease if paired themselves, but could cause disease if paired with another allele of higher pathogenicity. ²⁵ Expression analysis of S2255I has not been reported, but, as proposed by Webster et al., ²⁵ we cannot exclude a limited pathologic effect of this allele in those cases presenting a severer phenotype. For instance, family ARDM-85 showed unilateral vitreous detachment, severely reduced a- and b-waves of the ERG, reduced visual fields, but normal angiofluoresceingraphy and fundus. We speculate that either the ABCA4 gene acts as a modulating factor, or other genetic factors have an effect on the phenotypic outcome of the ABCA4 mutations. Clinical manifestations of the arCRD patients are shown in Table 4 . 

In the 27 RP families, we found one mutated allele in 9 (33%). In three families we identified only the controversial change S2255I, but as we pointed out before, we could not exclude this mutation as having a role in the pathogenesis of the disease. The percentage of allele detection was 13% to 16%, depending on the inclusion of the S2255I change. This mutation detection rate is higher than the 5% described by Klevering et al. ²³ and similar to the 16% described by Wiszniewski et al. ²⁶ Therefore, a complete study is needed to determine the real role of this controversial allele in the pathogenesis of the disease. A homogeneous phenotype has been described in patients with RP with mutations in the ABCA4 gene characterized by severe loss of visual function, extensive atrophy, and early loss of ERG responses. Most of our patients presented early onset of RP, with symptoms typically starting before the age of 10 years. There were two exceptions, SRP-775 and SRP-834, who demonstrated moderate and asymmetrical RP, respectively. All the mutations identified in those patients were suggested to have functional consequences for the ABCA4 activity (Table 3) . Taking into account that eight of the patients with heterozygous RP were simplex cases and the relatively high prevalence of STGD in the population (1:10,000), it is questionable, as pointed out by Lorenz and Preising, ¹⁰ whether these mutations are in fact disease-causing in RP. The involvement of the ABCA4 gene in several retinopathies is clear, but the exactly mechanism is still unknown. 

 Supported by Grant G03/018 from Fondo de Investigación Sanitaria, Grant PGIDT04PXIC30102PN from Xunta de Galicia , and Grant LSHG-CT-2005-512036 EVI-GENORET.

 Submitted for publication March 22, 2006; revised September 6, 2006; accepted January 11, 2007.

 Disclosure: D. Valverde, None; R. Riveiro-Alvarez, None; J. Aguirre-Lamban, None; M. Baiget, None; M. Carballo, None; G. Antiñolo, None; J.M. Millán, None; B.G. Sandoval, None; C. Ayuso, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Genetic Analyses of ABCA4 Mutations in Three Families with Autosomal Dominant Macular Dystrophy

Family	Allele 1			Allele 2			Haplotype Analysis
		Nucleotide Change	Amino Acid Change	Nucleotide Change	Amino Acid Change
ADDM-59	[5582G→A; 6764G→T]	[G1961E; S22551]			Excluded
ADDM-92	466A→G	I156V			Not done
ADDM-105	2828G→A	R943Q			Not done

 No change has been detected as allele 2.

Table 2.

 

Genetic Analyses of ABCA4 Mutations in 13 arCRD Families

Family	Allele 1			Allele 2			Haplotype Analysis
		Nucleotide Change	Amino Acid Change	Nucleotide Change	Amino Acid Change
ARDM-79	2888delG	Frameshift	2888delG	Frameshift	Cosegregates
ARDM-85	6764G→T	S2255I (likely nonpathogenic)	Not detected		Not done*
ARDM-86	2888delG	Frameshift	2888delG	Frameshift	Cosegregates
ARDM-99	4297G→A	V1433I	Not detected		Not done*
ARDM-126	[2828G→A; 5929G→A]	[R943Q; G1977S]	[2828G→A; 5929G→A]	[R943Q; G1977S]	Cosegregates
ARDM-133	32T→C	L11P	2888delG	Frameshift	Cosegregates
ARDM-134	2828G→A	R943Q			Excluded
ARDM-174	4918C→T	R1640W	c.6147+2T→A	Splice	Cosegregates
ARDM-176	2888delG	Frameshift	6179T→G	L2060R	Cosegregates
RP-267	5041del 15 pb	Frameshift	5041del 15 pb	Frameshift	Cosegregates
RP-577	1140T→A	N380K	Not detected		Not done*
SRP-964	2828G→A	R943Q	Not detected		Not done*
B210	2828G→A	R943Q	2701A→G	T901A	Not done*

 The mutation detected by dHPLC is in bold.

*  In all these families, only the index case could be analyzed.

Table 3.

 

Genetic Analyses of ABCA4 Changes in Nine Families with Autosomal Recessive RP

Family	Allele 1			Allele 2
		Nucleotide Change	Amino Acid Change	Nucleotide Change	Amino Acid Change
SRP-716	6764G→T	S2255I (likely nonpathogenic)	c.858 +8T→G
SRP-766	2300T→A	V767D	c.858 +8T→G
SRP-775	466A→G	I156V	c.858 +8T→G
SRP-818	6764G→T	S2255I (likely nonpathogenic)
SRP-834	c.5547+5G→A	Splice acceptor
SRP-854	6764G→T	S2255I
B57	466A→G	I156V
B173	2828G→A	R943Q	G5466A	L1821L
B278	2701A→G	T901A

Figure 1.

 

Fundus photographs from the patient in family ARDM-79.

Figure 2.

 

Fundus photographs from the proband in family ARDM-86, homozygous for 2888delG.

Figure 3.

 

Fundus photographs from the proband in family ARDM-176, heterozygous for 2888delG.

Table 4.

 

Clinical Data from the Patients with arCRD

Family	Age at Onset (y)	Actual Age (y)	Visual Acuity	Central Vision	Night Blindness	Dyschromatopsia	ERG	Fundus
ARDM-79	10	26	Diminished	Absolute scotoma	Yes	At 10 y	Extinguished	Attenuated retinal vessels and bone spiculelike pigmentation
ARDM-85	55	63	Reduced				Severely reduced	Normal
ARDM-86	9	40	Diminished	Absolute scotoma	At 18 y	At 47 y		Mild pallor, attenuated retinal vessels, with rounded, pigmented posterior pole
ARDM-99	43	58	0,05	Central scotomata		Yes	Diminished ERG	Macular atrophy and altered RPE
ARDM-126	10	35	0,5	Absolute scotoma	At 18 y	No		RPE atrophic, with bone spiculelike pigment
ARDM-133	8	42	Diminished	Absolute scotoma	Yes	Yes		Photophobia
ARDM-134	14	48	Diminished	Central scotoma		No	Diminished ERG	RPE atrophy, reddish macula
ARDM-174	4	36	Diminished	Central scotomata	Yes	Yes	Diminished ERG	Photophobia
ARDM-176	10	32	Diminished	Central scotoma	At 25 y	No		Photophobia
RP267	6	31	0,05/Count fingers	Diminished	At 9 y	Yes		Extensive tapetoretinal degeneration, with severe macular affection
RP577	30		0,1	Tunnel vision	Yes	Yes	Diminished ERG	Mild pallor papilla, normal vessels
SRP964	7	26	Diminished	Diminished visual field	At 14 y	At 14 y		Photophobia