|Home | About | Journals | Submit | Contact Us | Français|
It has been well documented that mutations in the same retinal disease gene can result in different clinical phenotypes due to difference in the mutant allele and/or genetic background. To evaluate this, a set of consanguineous patient families with Leber congenital amaurosis (LCA) that do not carry mutations in known LCA disease genes was characterized through homozygosity mapping followed by targeted exon/whole-exome sequencing to identify genetic variations. Among these families, a total of five putative disease-causing mutations, including four novel alleles, were found for six families. These five mutations are located in four genes, ALMS1, IQCB1, CNGA3, and MYO7A. Therefore, in our LCA collection from Saudi Arabia, three of the 37 unassigned families carry mutations in retinal disease genes ALMS1, CNGA3, and MYO7A, which have not been previously associated with LCA, and 3 of the 37 carry novel mutations in IQCB1, which has been recently associated with LCA. Together with other reports, our results emphasize that the molecular heterogeneity underlying LCA, and likely other retinal diseases, may be highly complex. Thus, to obtain accurate diagnosis and gain a complete picture of the disease, it is essential to sequence a larger set of retinal disease genes and combine the clinical phenotype with molecular diagnosis.
Leber congenital amaurosis (LCA; MIM# 204000) is an early-onset retinal dystrophy that is often diagnosed at birth or within the first year of life. The clinical features of LCA include blindness or severe vision impairment, congenital nystagmus, amaurotic pupils, and reduced or absence of signal in electroretinogram (ERG) [Franceschetti and Dieterle, 1954]. It is estimated to affect one in every 30,000~80,000 individuals and represents 5% of all retinal dystrophies [Dharmaraj et al., 2000; Kaplan et al., 1990; Leber, 1869; Stone, 2007].
LCA is a genetically heterogeneous and predominantly autosomal recessive disease. Up to now, mutations in 16 genes have been associated with LCA: aryl hydrocarbon receptor interacting protein-like 1 (AIPL1), centrosomal protein 290 kDa (CEP290), crumbs homolog 1 (Drosophila) (CRB1), cone-rod homeobox (CRX), guanylate cyclase 2D (GUCY2D), inosine 5′-monophosphate dehydrogenase 1 (IMPDH1), IQ motif containing B1 (IQCB1, also known as NPHP5), Leber congenital amaurosis 5 (LCA5), lecithin retinol acyltransferase (LRAT), orthodenticle homeobox 2 (OTX2), retinal degeneration 3 (RD3), retinol dehydrogenase 12 (RDH12), retinal pigment epithelium-specific protein 65 kDa (RPE65), retinitis pigmentosa GTPase regulator interacting protein 1 (RPGRIP1), spermatogenesis associated 7 (SPATA7), and tubby-like protein 1 (TULP1). [Bowne et al., 2006; den Hollander et al., 2001, 2007a,b, 2006; Dryja et al., 2001; Estrada-Cuzcano et al., 2011; Freund et al., 1998; Friedman et al., 2006; Henderson et al., 2009; Janecke et al., 2004; Marlhens et al., 1997; Mataftsi et al., 2007; Perrault et al., 2004, 1996; Senechal et al., 2006; Sohocki et al., 2000; Stone et al., 2011; Wang et al., 2009]. These genes are involved in various physiological pathways that are important for retinal function, including phototransduction (GUCY2D), maintenance of photoreceptor cell function (AIPL1, TULP1, RD3), visual cycle (LRAT, RPE65, RDH12), centrosomal or ciliary function (CEP290, IQCB1, LCA5, RPGRIP1), retinal development (CRB1, CRX, OTX2), and guanine nucleotide biosynthesis (IMPDH1) [Azadi et al., 2010; den Hollander et al., 2007a, 2006; Dizhoor, 2000; Escobar-Henriques and Daignan-Fornier, 2001; Furukawa et al., 1997; Mehalow et al., 2003; Murga-Zamalloa et al., 2009; Nishida et al., 2003; O’Byrne et al., 2005; Otto et al., 2005; Ramamurthy et al., 2004; Redmond et al., 1998; Thompson et al., 2005; Xi et al., 2005].
Recently, it was reported that certain mutant alleles in syndromic ocular disease genes may cause nonsyndromic LCA. For example, mutations in gene IQCB1 were previously reported to cause Senior-Løken syndrome (SLSN), which is characterized by retinal defects and nephronophthisis [Otto et al., 2005]. However, several nonsense and frameshift mutations in IQCB1 have been found in LCA patients without nephronophthisis or overt renal disease, suggesting that mutations in IQCB1 may cause LCA without having other syndromic phenotypes [Estrada-Cuzcano et al., 2011; Stone et al., 2011]. Considering that many nonsyndromic and syndromic ocular diseases, such as achromatopsia, Alstrom syndrome (ALMS), Batten disease, and SLSN, are associated with “LCA-like ocular phenotypes” [den Hollander et al., 2008], it is conceivable that some percentage of patients diagnosed with LCA are actually caused by either a specific allele or combination of disease gene alleles or mis-assignment due to the absence of syndromic features at the time of diagnosis.
Previously, we collected 37 consanguineous families with recessive LCA from Saudi Arabia. PCR and Sanger sequencing were performed for these families to screen for mutations in all known LCA genes [Li et al., 2009]. Among them, mutations have been identified in nine families. To identify the underlying mutations in the remaining consanguineous LCA disease families, homozygosity mapping using high-density SNP genotyping arrays followed by targeted or whole-exome sequencing was performed. Interestingly, mutations have been identified in four genes, ALMS1, IQCB1 (also known as NPHP5), CNGA3, and MYO7A, in six consanguineous LCA families, accounting for 16% of our collection (six of the 37) (Table 1). Recent studies have linked mutations in IQCB1 with LCA without renal failure [Estrada-Cuzcano et al., 2011; Stone et al., 2011]. Whereas mutations in ALMS1, CNGA3, and MYO7A are known to be associated with syndromic or nonsyndromic eye diseases, they have not been previously linked to LCA. Therefore, our results support an emerging theme that a significant fraction of patients diagnosed with LCA may be accounted for by mutations in syndromic and other retinal disease genes. These results highlight the importance of combining molecular diagnosis with clinical findings for diseases with high genetic heterogeneity in order to obtain accurate diagnosis and devise a proper treatment strategy.
We obtained blood samples and pedigrees after receiving informed consent from all individuals. Approval was obtained from the Institutional Review Boards of the participating centers. LCA families KKESH24, KKESH28, KKESH34, KKESH72, and KKESH88 were obtained by Dr. Lewis through the King Khaled Eye Specialist Hospital (KKESH) in Riyadh, Saudi Arabia. Blood samples were collected from all available family members, and DNA was extracted with the Qiagen blood genomic DNA extraction kit (QIAGEN Inc., Valencia, CA) following the protocol provided by the manufacturer. The pedigrees of these families are shown in Figure 1. The number of affected and unaffected members of each family is listed in Table 2 and can be seen in Figure 1. All of the LCA patients were diagnosed with typical LCA clinical features (Table 2). Patients have experienced vision defects since birth or as early as 2 years old. No significant syndromic phenotypes were observed, except that a patient in KKESH88 showed midfacial hypoplasia and psychomotor delay.
Homozygosity mapping using high-density SNP arrays was performed on these six consanguineous LCA families (Fig. 1). SNP genotypes were obtained for multiple affected members and unaffected siblings from each pedigree. Homozygous blocks greater than 1 Mb in size were identified from each individual. Based on recessive inheritance models, candidate disease loci are defined by shared homozygous blocks among all affected individuals but those are heterozygous or absent in unaffected siblings. The homozygous regions for six families range from 2000 bp to more than 57 Mb in size (Table 3).
Whole-exome capture sequencing provides an alternative way to sequence large numbers of coding regions at a lower price than traditional PCR and Sanger sequencing. One affected member from each family was chosen for NimbleGen whole-exome capture followed by Illumina HiSeq paired-end sequencing. For each patient, a total of about 4.4 to 100 million reads were uniquely mapped to the targeted gene-coding regions with an average coverage of 9× to 467×. SNPs were identified based on filtering criteria (posterior probability cutoff = 0.9, minimum coverage = 3). A total of 52,188–370,000 SNPs were identified for further analysis (Supp. Tables S1–S6).
Direct PCR and sequencing were used to validate the mutations found in human ALMS1 (RefSeq: NM_015120.4; MIM# 606844), IQCB1 (RefSeq: NM_001023570.2; MIM# 609237), CNGA3 (RefSeq: NM_001298.2; MIM# 600053), and MYO7A (Ref-Seq: NM_000260.3; MIM# 276903). Primers were designed using primer design tool Primer3 (http://biotools.umassmed.edu/bioapps/primer3_www.cgi) [Rozen and Skaletsky, 2000]. Each amplicon was 300–500 base pairs in length and was sequenced directly with an ABI3730 machine in both forward and reverse directions. Each read was aligned to the reference sequence and base changes were identified with the Sequencher program.
For mutations, nucleotide numbering reflects cDNA numbering with +1 corresponding to the A of the ATG translation initiation codon in the reference sequence, according to journal guidelines (www.hgvs.org/mutnomen). The initiation codon is codon 1.
Candidate disease loci in consanguineous families were determined based on the property of identity-by-descent (IBD). To identify underlying mutations in the LCA family collection from Saudi Arabia, homozygosity mapping and targeted exon/whole-exome sequencing were performed. The pedigrees of the six families, KKESH2, KKESH24, KKESH28, KKESH34, KKESH72, and KKESH88, reported here are shown in Figure 1. The clinical phenotypes of affected members of these families are listed in Table 2.
As shown in Figure 1, one of the biggest families in our collection is KKESH72. It is a four generation family containing six affected and 29 unaffected members. Four of the affected and one of the unaffected members were genotyped using SNP arrays. Although multiple homozygous blocks are present in each individual, only one homozygous region located on chromosome 2 from 71.5 to 74.3 Mb is shared among all four affected members and is either heterozygous or absent in the unaffected individual (Table 3). Consistent with previous data, no known LCA gene is located within this region, indicating that this region must contain a novel LCA disease gene. A custom-designed padlock probe set covering all exons in this region was used for capture sequencing as described in the methods section. A total of 4.45 million reads were generated, with an average sequencing coverage of 467×. For individual KKESH72#13, a total of 15 homozygous rare variants were detected that cause amino acid (aa) changes and reside within the homozygous block. Among them, only one nonsense mutation was identified. This mutation is located in the sixteenth exon of ALMS1 (c.10945G>T, p. E3649X) (Fig. 2A and B and Supp. Table S1). Interestingly, this nonsense mutation has been previously reported to be associated with ALMS [Marshall et al., 2007]. This mutation was confirmed by direct Sanger sequencing (Fig. 2B) and segregates with the disease in this family where affected individuals are homozygous for this mutation and other unaffected individuals are either heterozygous or do not carry the mutation. ALMS1 encodes a protein of 4,169 aa that localizes to centrosomes and ciliary basal bodies, and is important for normal cilium function [Li et al., 2007]. This nonsense mutation in exon 16 will either give rise to a protein with a truncated C-terminus missing 520 aa that is likely to have partial function or cause complete loss of function due to nonsense-mediated mRNA decay [Li et al., 2007]. To further check the potential effect of this mutation in splicing, ESEfinder 3.0 [Cartegni et al., 2003; Smith et al., 2006] was used to identify potential exonic splicing enhancers (ESE) around this variant. Interestingly, three potential ESEs are identified that might be affected by the mutation (Supp. Table S7). In the case that the 16th exon is skipped due to the mutation, it will lead to shift of the open reading frame and truncation of the protein. Multiple mutations in human ALMS1 have been reported to cause the ALMS, which is characterized by cone–rod retinal dystrophy, cardiomyopathy, type 2 diabetes mellitus, obesity, and hearing loss [Bond et al., 2005; Collin et al., 2002; Hearn et al., 2002; Marshall et al., 2007]. Patients in the KKESH72 family are reported to have nystagmus, cataracts, fundus defects, and other retinal defects (Table 2). In contrast to the phenotype in ALMS, no other syndromic phenotypes such as hearing loss or obesity were observed.
It has been recently reported that some of the mutant alleles in IQCB1 cause nonsyndromic LCA [Estrada-Cuzcano et al., 2011; Stone et al., 2011]. KKESH24, KKESH88, and KKESH28 are three consanguineous LCA families with one affected member in each family and four, six, and four unaffected members, respectively (Fig. 1). Based on homozygosity mapping, multiple homozygous blocks were identified for each family (Table 3). Whole-exome sequencing was conducted for one affected member from each family: KKESH24#5, KKESH88#5, and KKESH28#5. For the three KKESH-affected individuals, a total of 107, 117, and 81 million reads were generated, resulting average sequence coverage of 40×, 21×, and 30×, respectively. A total of 6, 4, and 27 homozygous rare variants that cause aa changes reside in each family’s homozygous block (Supp. Tables S2–S4). Among these candidate SNPs, only one obvious loss-of-function mutation is found for each family, all of which locate in IQCB1. One mutation is a novel splice acceptor site change (c.1130-1G>C) in KKESH24#5 and KKESH88#5, and the other is a novel nonsense mutation (c.1479C>A, p.Y493X) in KKESH28#5 (Fig. 2C and D). Both mutations are confirmed by direct Sanger sequencing (Fig. 2D) and are likely to be pathogenic. First, both mutations segregate with the disease in each family, with affected members being homozygous for the mutations and other unaffected individuals as either heterozygous or not carrying the mutation. Second, these mutations are rare as they are not found in 200 matching controls, the dbSNP130 database, or the 1,000 Genome database. Third, IQCB1 encodes an IQ-domain protein, which is important for cilium function and colocalizes with RPGR at the connecting cilia of photoreceptors [Otto et al., 2005]. Human mutations in IQCB1 are associated with LCA [Estrada-Cuzcano et al., 2011; Stone et al., 2011] and SLSN [Otto et al., 2005], which is characterized by nephronophthisis and retinal degeneration. The patients in these three KKESH families show typical LCA phenotypes such as nystagmus, nonrecordable ERGs, and other visual defects. However, no kidney defects were observed at the time of diagnosis.
KKESH2 is a four generation family with one affected and six unaffected members (Fig. 1). One unaffected and one affected member were genotyped by SNP arrays. Ten homozygous blocks ranging from 5.3 to 35.5 Mb were identified in the affected member and are not present in the unaffected member (Table 3). Whole-exome sequencing was conducted for one affected individual, KKESH2#4. A total of 100 million reads were generated with 36× coverage. SNP calling reveals 11 homozygous rare variants that lead to aa changes and reside within the homozygous region (Supp. Table S5). Based on gene annotation, we focus on a missense change (c.1579C>A, p.L527M) in the CNGA3 gene. This variant was confirmed by direct Sanger sequencing (Fig. 3A and B) and is likely to be pathogenic. First, it is located within the homozygous region chr2: 94.7–100.8 Mb and segregates with the disease in this family, given the affected member is homozygous for this mutation and other unaffected individuals are either heterozygous or do not carry the mutation. Second, this variant is rare as it is absent in all 200 normal matching controls. In addition, it is not recorded in the dbSNP130 database and the 1,000 Genome database. Third, CNGA3 encodes a member of the family of cyclic nucleotide-gated channel alpha 3 proteins, which are important for normal vision and olfactory signaling transduction. The p.L527M mutation resides in the cGMP-binding domain of CNGA3, which is important for the cyclic nucleotide gating mechanism (Fig. 3C). This mutated leucine is conserved across vertebrate species, from human to stickleback, further suggesting functional importance of this residue (Fig. 3D). Human mutations in this gene are reported in hereditary cone photoreceptor disorder, which is characterized by cone photoreceptor dysfunction, and achromatopsia, which leads to early vision loss, nystagmus, photophobia, color blindness, but has normal scotopic responses [Johnson et al., 2004; Wissinger et al., 2001]. Affected members of the KKESH2 family show early-onset nystagmus (at 5 months old), sluggish pupils, no visual response, and nonrecordable ERG at 10 months of age (Table 2). Taken together, these data suggest that the p.L527M mutation in CNGA3 is associated with the LCA phenotype in the KKESH2 family.
KKESH34 is a large consanguineous family containing four affected and 11 unaffected individuals (Fig. 1). All the affected and three unaffected members were genotyped by SNP arrays. The only homozygous region that was shared by affected members and not present in unaffected members resides on chr11: 5463161491827532, with a size of 37.2 Mb (Table 3). Since the size of the candidate region is large, whole-exome sequencing was performed for the patient KKESH34#6. A total of 7.7 million reads were generated, with coverage of 27×. A total of six homozygous rare variants that lead to aa changes were identified within the candidate region (Supp. Table S6). Based on conservation score (data not shown) and gene annotations, we focus on a novel missense variant (c.578C>T, p.T193I) in the MYO7A gene (Fig. 4A and B). This variant is rare, as it is absent from 200 matching controls, the dbSNP database, and 1,000 Genome database. It segregates with the disease in this family, given that all the affected members are homozygous for this mutation and other unaffected individuals are either heterozygous or do not carry the mutation. MYO7A encodes an unconventional myosin, myosin VIIA, which is involved in the transportation of opsin in photoreceptor cilia [Liu et al., 1999]. Also, it was recently demonstrated that myosin VIIA protein may affect the localization and function of the visual retinoid cycle enzyme, RPE65, which has been associated with LCA [Lopes et al., 2011]. The mutation identified (p.T193I) is a novel allele residing in the motor domain, which is responsible for binding filamentous actin and hydrolyzing ATP (Fig. 4C). Human mutations in MYO7A are associated with Usher syndrome, characterized by deafness and progressive vision loss. All patients of KKESH34 have had poor vision since birth. They all also have nystagmus, neuroepithelial atrophy, and nonrecordable ERGs (Table 2). However, patients in the KKESH34 family do not exhibit hearing loss. Therefore, it is likely that this mutation in MYO7A will only partially affect protein function, resulting in retinal specific clinical phenotype.
We performed homozygosity mapping and whole-exome sequencing to identify mutations affecting a collection of consanguineous LCA families. Our experiments show that homozygosity mapping coupled with whole-exome sequencing is a very powerful tool for identifying mutations, even for families with a small number of affected members. For example, in the KKESH28 family, there was only one affected member and four unaffected members (Fig. 1 and Table 2). Homozygosity mapping alone discovered 12 homozygous blocks whose widths range from 2087 bp to 51.7 Mb, making the approach of PCR and Sanger sequencing impractical (Table 3). By using whole-exome sequencing, we were able to discover 27 rare SNPs that lead to aa changes and reside within the homozygous regions (Supp. Table S4). Among these candidate SNPs, a nonsense mutation (c.1479C>A, p.Y493X) was identified, suggesting this approach can be effective in identifying mutations in families with a small number of affected members.
Interestingly, we have identified families carrying mutations in ALMS1, IQCB1, CNGA3, and MYO7A genes that have been previously associated with other syndromic or nonsyndromic retinal diseases whose phenotypes overlap with that of LCA. The mutations include a previously reported nonsense mutation (c.10945G>T, p. E3649X) in ALMS1 in family KKESH72, a novel splicing change (c.1130-1G>C) in IQCB1 in two families, KKESH24 and KKESH88, a novel nonsense mutation (c.1479C>A, p.Y493X) of IQCB1 in family KKESH28, a novel missense mutation (c.1579C>A, p.L527M) in CNGA3 in family KKESH2, and a novel missense mutation (c.578C>T, p.T193I) in MYO7A in family KKESH34 (Table 1). Based on the phenotype of the patients, it is likely that in each case these missense mutations result in partial loss-of-function of the gene. These alleles are not only useful for molecular diagnoses, but further study of the functional consequence of these mutations will likely provide additional insight into the molecular mechanisms by which these proteins act.
Our findings support the emerging theme that a significant fraction of patients diagnosed with LCA may carry mutations in other syndromic or nonsyndromic retinal disease genes. This is likely due to several reasons. First, it is possible that patients diagnosed as having LCA are too young at the time of diagnosis to present with other syndromic phenotypes. For example, patients who carry mutations in IQCB1 may not show kidney defects until they are in their teens. Second, it is possible that different alleles in the same gene can result in different clinical presentations. For example, patients in family KKESH34 carry a missense mutation (c.578C>T, p.T193I) in MYO7A and show typical LCA phenotypes without hearing problems. It is possible that this mutation results in a partial loss-of-function allele that is only critical for function in the eye. Indeed, similar phenomena have been observed for many retinal disease genes. For example, partial loss-of-function mutations in CEP290 lead to LCA [Coppieters et al., 2010; den Hollander et al., 2006], while null alleles are also associated with recessive Meckel syndrome [Baala et al., 2007; Frank et al., 2008], Bardet-Biedl syndrome [Leitch et al., 2008], and Joubert syndrome [Sayer et al., 2006; Valente et al., 2006]. This phenomenon is also observed in other LCA causal genes, such as CRX [Freund et al., 1997; Swain et al., 1997] and RDH12 [Fingert et al., 2008; Janecke et al., 2004]. Third, it is possible that the same mutation can lead to a different clinical presentation, presumably due to modifications from the genetic background and/or the environment. In our report, the nonsense mutation (c.10945G>T, p. E3649X) in the ALMS1 gene identified in family KKESH72 has been previously reported to cause ALMS [Marshall et al., 2007]. However, none of the patients in this family show other syndromic phenotypes, such as hearing loss or obesity, which are typical for ALMS. It is possible that the different genetic background in the LCA patients from family KKESH72 lead to the different clinical presentations. Indeed, it has been reported recently that a common allele in RPGIP1L can modify the retinal degeneration phenotype in ciliopathies [Katsanis et al., 2001; Khanna et al., 2009]. In addition, in this report, the same mutation in IQCB1 has been identified in both KKESH24#5 and KKESH88#3 patients. It is interesting to note that while the eye phenotype in both patients is quite similar, KKESH88#3 also shows additional neural defects such as midfacial hypoplasia and psychomotor delay. With the whole-exome sequencing data, it is possible to check if KKESH88#3 carries potential genetic modifiers that can potentially explain the clinic phenotype. Indeed, in KKESH88#3, rare variants have been found in gene ACAT1 and FGFR2, which is associated with psychomotor delay and midfacial hypoplasia, respectively (Supp. Table S8) [Korman, 2006; Tartaglia et al., 1997].
In summary, it is evident that in addition to the 16 known LCA genes, mutations in several other syndromic or nonsyndromic eye disease genes may also lead to the LCA phenotype. Due to the high genetic heterogeneity of LCA, it is likely to be informative to sequence a larger set of retinal genes along with the known LCA genes. Combining accurate molecular diagnoses with the clinical phenotypes of LCA patients will be an essential step to proper treatment of this disease in the future.
Contract grant sponsor: Retina Research Foundation and the National Eye Institute (R01EY018571 to R.C.).
We are indebted to John Cavender, M.D., the Research Director of the King Khalid Eye Specialist Hospital at the time of this study, to the Research Council of KKESH for financial support, and to the staff of the KKESH Research Department for their diligent commitment to this program. In addition, we thank the family reported here for their willing cooperation in this study. We would like to thank Dr. Molly Bray for the SNP genotyping. Dr. Lewis is a Senior Scientific Investigator of Research to Prevent Blindness, New York. HW was supported by postdoctoral fellowship F32EY19430.
Additional Supporting Information may be found in the online version of this article.