Mutation Analysis of 3 Genes in Patients With Leber Congenital Amaurosis FREE

LEBER CONGENITAL amaurosis (LCA) is a term used to refer to a group of disorders characterized by severe visual impairment in the first year of life, nystagmus, and a markedly abnormal electroretinogram. The fundus appearance can initially be normal, but macular colobomas, optic disc edema, bone-spicule–like pigmentation, and vascular attenuation are observed.^{1 - 3} Most of the genes that cause the clinical phenotype known as LCA are transmitted in an autosomal recessive fashion. A small but significant fraction of patients with LCA will have additional neurologic problems, including developmental delay. An equally small percentage of children who are initially diagnosed as having LCA will retain surprisingly good visual function for decades. This wide range of potential clinical outcomes can make it difficult to provide specific prognostic information in individual cases. There is no medical or surgical intervention that can alter the course of LCA.

In 1996, a significant advance in the understanding of LCA occurred when Perrault and coworkers⁴ identified the first gene whose mutation was capable of causing the LCA phenotype, retinal guanylate cyclase (GUCY2D). This was followed in successive years by the identification of 2 additional LCA-causing genes, the retinal pigment epithelium–specific gene known as RPE65⁵ and the photoreceptor-specific homeobox gene known as CRX.⁶ Recently, the TULP1 gene encoding the Tubby-like protein 1 has also been shown to produce an LCA phenotype but only in a single Dominican family.⁷ The identification of these disease genes raised the possibility of identifying specific genotype-phenotype correlations that could be used to provide more accurate diagnostic and prognostic information to patients. Improved understanding of a disease gene also has the potential to provide sufficient new insight into the pathophysiological features of a disease so that effective therapy can be developed.

This study was undertaken to determine the frequency of mutations in the GUCY2D, RPE65, and CRX genes in 176 individuals affected with LCA and to look for genotype-phenotype correlations that might be clinically useful. During this study, we were able to deduce a full molecular explanation for a challenging set of clinical findings that we observed in one of the families. This family is described in detail to demonstrate the potential utility of molecular diagnosis in the management of rare, clinically heterogeneous disorders such as LCA.

Informed consent was obtained from all study patients or their legal guardians. One hundred patients were from the United States, 39 from India, 28 from Canada, 3 from Israel, 2 from Switzerland, 1 from Italy, 1 from Jordan, 1 from Pakistan, and 1 from Honduras. DNA was extracted from peripheral blood using a previously described protocol.⁸ The proband of each family was screened for mutations in the coding sequence of the GUCY2D, RPE65, and CRX genes by single-strand conformation polymorphism analysis. Ninety-six control subjects from Iowa were screened in an identical fashion for each gene. An additional 47 control patients were from India. These Indian control samples were screened for variations in the 5 exons, in which sequence variations were found in affected Indian patients. The primer sequences used for the single-strand conformation polymorphism screening have been previously described.^{4 - 5 ,9} The polymerase chain reaction amplification products were denatured for 3 minutes at 94°C and then electrophoresed on 6% polyacrylamide–5% glycerol gels at 25 W for about 3 hours. The gels were then stained with silver nitrate.¹⁰ Polymerase chain reaction products from samples with aberrant migration patterns were then sequenced bidirectionally with fluorescent dideoxynucleotides on an automated sequencer (model 377; Applied Biosystems, Foster City, Calif). The clinical records of patients who were found to harbor a sequence change were reviewed, and this phenotypic information was correlated with the associated genotype wherever possible. Refractions were calculated as spherical equivalents in diopters (D). Two patients harboring CRX gene mutations (Table 1) have been described previously^{6 ,12} and are included in this report because they were part of the original cohort.

A family of Jordanian ancestry consisting of 2 parents and 3 sons was initially seen for diagnosis and genetic counseling. The parents, who were first cousins, each had a normal visual acuity. They also had essentially normal fundus examination results, although careful biomicroscopy revealed subtle yellow dots at the level of the retinal pigment epithelium in their maculae. Of the 3 sons, the eldest was aged 16 years and had a history of congenital esotropia. His visual acuity was 20/25 OU. He had several subretinal gliotic nodules in the macula of each eye in addition to some yellow deposits at the level of the retinal pigment epithelium (Figure 1, A). The middle brother was aged 13 years and had a visual acuity of 20/20 OU. He had subtle yellow deposits in each macula that were more prominent than those of his parents but much less prominent than those of his older brother (Figure 1, B). The youngest brother was aged 9 years. His visual acuity was counting fingers OU. Historically, he had searching, wandering eye movements since the first year of life and a nonrecordable electroretinogram. His fundus examination revealed absence of the foveal light reflex and a diffusely abnormal retinal pigment epithelium (Figure 1, C).

Electro-oculography was performed on the parents and the elder brothers. The Arden ratios (normal, >1.8) were found to be abnormal in all 4 individuals: father, 1.4 OD, 1.3 OS; mother, 1.5 OD, 1.2 OS; eldest brother, 1.0 OD, 1.0 OS; and middle brother, 1.3 OD, 1.6 OS. The clinical and electrophysiologic findings suggested that the parents and the elder brothers might be affected with a mild form of Best disease and that the LCA phenotype in the youngest brother might be due to 2 copies of a Best disease allele. However, linkage analysis revealed the middle brother and the youngest brother to have the same haplotypes at the Best locus, which ruled out this explanation.

The discovery of the Best disease gene (VMD2)¹³ and the discoveries of 3 genes responsible for some cases of LCA allowed the findings in this family to be studied at the gene level. This family was screened for mutations in the VMD2, RPE65, CRX, and GUCY2D genes. The parents were found to be heterozygous for a bestrophin mutation (Gly222Val) and an RPE65 mutation (Glu462stop). The eldest brother was found to be homozygous for the VMD2 mutation (Gly222Val), while the middle brother was found to be heterozygous for this same mutation. The youngest brother was found to be homozygous for an RPE65 mutation (Glu462stop) and the Gly222Val VMD2 mutation (Figure 1, D).

One hundred ten instances of 35 different coding sequence variations were observed in this study. Of these, 19 were judged to be probable disease-causing mutations because they: (1) would be expected to alter the amino acid sequence of the proteins encoded by the CRX, GUCY2D, or RPE65 gene in a nonconservative fashion; (2) were present in 1 or more patients with LCA; and (3) were not present in control individuals. These are the same criteria used in similar studies.^{14 - 15} Four additional sequence variations were identified as possible disease-causing mutations for the same reasons except that the amino acid alterations were conservative. Of the 176 probands, 28 (15.9%) were found to harbor at least 1 of these probable or possible disease-causing mutations. The relative contribution of each gene to the total number of mutations was as follows: CRX, 2.8%; GUCY2D, 6.3%; and RPE65, 6.8%. Nonconservative changes were found in 5 of 5 patients with CRX, 8 of 11 patients with GUCY2D, and 10 of 12 patients with RPE65 gene mutations. Two mutations (presumably in different alleles) were found in 0 of 5 patients with CRX mutations, 2 of 11 patients with GUCY2D mutations, and 7 of 12 patients with RPE65 mutations. Thus, evidence for 2 abnormal alleles was found in 9 (32%) of the probands who harbored probable or possible disease-causing mutations (Table 1). Four of the sequence variations we observed have been previously reported (Table 1).^{2 ,5 - 6 ,16}

Three of the probands with probable or possible disease-causing sequence variations had an affected sibling available for study. In these families, the mutant alleles segregated as expected for an autosomal recessive disease (Figure 2).

Some clinical details of patients with probable disease-causing mutations are shown in Table 2. The visual acuities of these patients ranged from 20/50 to no light perception (Table 2). No patients who harbored mutations in 1 of these 3 genes were found to have mental retardation, neurologic abnormalities, developmental delay, or macular colobomas.

The identification of any disease-causing gene has several potential benefits for affected patients, including (1) more accurate diagnosis; (2) more accurate prognosis; (3) insight into the pathophysiological mechanism, which may allow the development of efficacious treatment; (4) the development of novel therapies; and (5) the creation of transgenic animal models, which can be used to evaluate existing or novel therapies relatively quickly. The potential value of molecular diagnosis for patients affected with the clinical phenotype of LCA is well illustrated by the Jordanian family who harbored mutations in the RPE65 and VMD2 genes (Figure 1). It would have been virtually impossible to counsel this family based on clinical information alone.

A major challenge for a large mutation screening study such as this one is to determine which of the observed sequence variations are actually responsible for disease. Inference of pathogenicity of an individual sequence change is usually based on a combination of the following: (1) the relative frequency of an allele in the disease population with respect to a control population; (2) cosegregation with disease within families; and (3) specific features of the mutation itself, including the degree to which it would be expected to alter the function of the gene's protein product. Although in principle it is possible to apply many of these arguments regardless of the inheritance pattern of the disease, in actual practice, autosomal recessive diseases are more challenging than the other 2 common mendelian inheritance patterns. The reasons for this are numerous and include the following: (1) 2 mutant alleles are present in affected patients with recessive diseases; (2) recessive diseases are frequently caused by loss of function alleles,¹⁷ which include mutations that occur outside the coding sequence; (3) families affected with dominant diseases tend to have a higher proportion of affected individuals, allowing greater testing for segregation of the disease; and (4) heterozygous carriers of recessive disease in the normal population make it more difficult to statistically argue for an allele's pathogenicity based on its absence from the normal population. Most of these difficulties can be overcome if a sufficiently large population affected with the disease can be studied.

Similarly, there are several obstacles to discerning the phenotype associated with autosomal recessive alleles. In dominant diseases and X-linked recessive diseases, only a single mutant allele is present in an affected individual and the phenotype associated with this allele can thus be reasonably ascribed to its presence. In contrast, 2 different alleles are likely to be present in a patient with an autosomal recessive disease from an outbred population. This makes it difficult to discern the contribution that each mutant allele makes to the overall phenotype.

Moreover, since no mutation detection strategy is 100% sensitive, one is not infrequently faced with interpreting apparent heterozygosity in affected patients. For example, CRX is known to cause autosomal dominant and recessive LCA, cone rod dystrophy, and retinitis pigmentosa.^{6 ,18 - 19} Mutations in the GUCY2D gene have been found in patients with autosomal dominant cone rod dystrophy,²⁰ and mutations in the RPE65 gene have been associated with autosomal recessive retinitis pigmentosa. Thus, the presence of a single heterozygous sequence change in the CRX or GUCY2D gene of an isolated patient with a retinal degeneration has at least 4 possible explanations: (1) it may represent a dominant disease-causing allele, (2) it may represent the only detectable member of a pair of recessive disease-causing alleles, (3) it may represent one member of a pair of digenic disease-causing alleles, and (4) it may represent a non–disease-causing polymorphism that has no role at all in the patient's retinal disease.

DIGENIC INHERITANCE, although possible, is unlikely to be involved in a large fraction of LCA cases for the following reasons. Vertical transmission and variable expressivity are rarely reported in LCA pedigrees.⁵ In addition, in the present study, we did not observe a single case of double heterozygosity as might be expected if any 2 of these 3 genes commonly causes disease via a digenic mechanism. The Jordanian family described in this article represents an extraordinary chance occurrence of 2 inherited retinal diseases. This differs significantly from true digenic disease in that the disease phenotype in each of the 3 siblings in the Jordanian family is fully explained by the presence of mutations in only a single gene (VMD2 in the elder siblings and RPE65 in the youngest).

In this study, we screened 176 unrelated individuals with the clinical diagnosis of LCA to determine the frequency and type of sequence variations present in 3 genes previously associated with this phenotype. The most striking result is one that has been noted in other studies: most patients affected with LCA do not exhibit coding sequence variations in these 3 genes. Thus, although it is certain that some disease-causing alleles have been missed by our assays, it is also quite likely that additional genes capable of causing the phenotype of LCA must exist. In this study, the median visual acuity of patients with mutations in these 3 genes was 20/2000, with a range from 20/50 to no light perception.

We were able to correlate visual acuity and refraction with the genotype in 17 patients and compare our results with those of previous reports. Leber congenital amaurosis has been associated with hyperopia in several clinical studies^{21 - 23} and in a genotype-phenotype study.²⁴ In our patients (Table 2), the mean refraction was hyperopic for all 3 genes (CRX, +0.2 D; GUCY2D, +2.8 D; and RPE65, +1.3 D), but the range of refractions was broad (CRX, −0.5 to +5.5 D; GUCY2D, +1.5 to +4.0 D; and RPE65, −1.6 to +3.8 D). Macular colobomas, neurologic abnormalities, and developmental delay have been associated with the diagnosis of LCA. Interestingly, none of the patients who exhibited sequence variations in these 3 genes exhibited any of these clinical features. This suggests that these abnormalities are more commonly associated with other genetic causes of LCA.

In a clinical series²³ of 75 patients in whom LCA was initially diagnosed, it was discovered that misdiagnosis occurred in 30 of these patients. Common revised diagnoses included congenital stationary night blindness, achromatopsia, Joubert syndrome, Zellweger syndrome, and infantile Refsum disease. A 40% clinical misdiagnosis rate illustrates the potential value of molecular diagnosis of this disease.

Thirty-nine of the patients in this study were from India, where the relatively high rate of consanguinity^{25 - 28} makes it more likely that individuals affected with recessive disease will be homozygous for their mutant allele. The relative paucity of mutations in these 3 genes in the Indian patients suggests that 1 or more additional genes is responsible for a relatively larger fraction of the Indian population with LCA than in western countries. This underscores the need to conduct mutation screens in a global fashion because certain genes may be more easily discovered and characterized in some populations than in others. Moreover, genes that may cause LCA commonly in one country may cause it less commonly but still significantly in another. Study of populations with a higher rate of consanguineous marriages and, hence, a higher frequency of homozygous affected individuals will also allow better recognition of genotype-phenotype relations for specific alleles.

In summary, a sample of 176 families gathered from 9 different countries is sufficient only for a preliminary assignment of pathogenicity of certain alleles and for the most general genotype-phenotype correlations. Nevertheless, molecular diagnosis may be of significant value in counseling the families of some individuals affected with LCA.

Accepted for publication November 19, 1999.

This study was supported in part by grants EY10539 and EY05627 from the National Institutes of Health, Bethesda, Md; the Foundation Fighting Blindness, Hunt Valley, Md; the Grousbeck Family Foundation, San Francisco, Calif; the Carver Charitable Trust, Muscatine, Iowa; the Daniel Matzkin Research Fund; and an unrestricted grant from Research to Prevent Blindness Inc, New York, NY.

We thank William Scott, MD, Douglas Fredrick, MD, Ehud Zamir, MD, Saul Merin, MD, Francis Munier, MD, Arlene Drack, MD, Terry Schwartz, MD, Jean Bennett, MD, Benedetto Falsini, MD, and Michael Vaphiades, MD, for sharing their patients with us for this study; and Luan Streb, Kim Vandenburgh, Christine Taylor, Heidi Haines, Louisa Affatigato, Robin Hockey, Gretel Beck, and David B. Hanna for their excellent technical assistance.

Reprints: Edwin M. Stone, MD, PhD, Department of Ophthalmology and Visual Sciences, University of Iowa Hospitals and Clinics, 200 Hawkins Dr, Iowa City, IA 52242 (e-mail: edwin-stone@uiowa.edu).