Molecular Genetic Analysis of Pakistani Families With Autosomal Recessive Congenital Cataracts by Homozygosity Screening.

Purpose: To identify the genetic origins of autosomal recessive congenital cataracts (arCC) in the Pakistani population.
Methods: Based on the hypothesis that most arCC patients in consanguineous families in the Punjab areas of Pakistan should be homozygous for causative mutations, affected individuals were screened for homozygosity of nearby highly informative microsatellite markers and then screened for pathogenic mutations by DNA sequencing. A total of 83 unmapped consanguineous families were screened for mutations in 33 known candidate genes.
Results: Patients in 32 arCC families were homozygous for markers near at least 1 of the 33 known CC genes. Sequencing the included genes revealed homozygous cosegregating sequence changes in 10 families, 2 of which had the same variation. These included five missense, one nonsense, two frame shift, and one splice site mutations, eight of which were novel, in EPHA2, FOXE3, FYCO1, TDRD7, MIP, GALK1, and CRYBA4. Conclusions: The above results confirm the usefulness of homozygosity mapping for identifying genetic defects underlying autosomal recessive disorders in consanguineous families. In our ongoing study of arCC in Pakistan, including 83 arCC families that underwent homozygosity mapping, 3 mapped using genome-wide linkage analysis in unpublished data, and 30 previously reported families, mutations were detected in approximately 37.1% (43/116) of all families studied, suggesting that additional genes might be responsible in the remaining families. The most commonly mutated gene was FYCO1 (14%), followed by CRYBB3 (5.2%), GALK1 (3.5%), and EPHA2 (2.6%). This provides the first comprehensive description of the genetic architecture of arCC in the Pakistani population.

Congenital cataract (CC) is a significant cause of vision loss worldwide, causing approximately one-third of blindness in infants.[1] Approximately one-third of CCs are familial; the cataract may be isolated or be associated with other systemic abnormalities.[2] Nonsyndromic CCs may account for approximately 70% of CC cases,[3] and have an estimated frequency of 1 to 6 per 10,000 live births.[4] Congenital cataracts are clinically and genetically heterogeneous, with approximately 8.3% to 25.0% of nonsyndromic CCs being inherited, approximately 7% as autosomal recessive (ar), 76% to 89% as autosomal dominant (ad), or 2% to 10% as X-linked traits in European populations.[5-8] Currently more than 48 CC loci have been identified, and more than 35 of them have been associated with causative mutations in specific genes, as delineated in the Cat-Map database (http://cat-map.wustl.edu/, in the public domain).[9] Individuals in families having CCs as a result of the same mutation can show variable severity and morphology, probably reflecting effects of differences in their genetic backgrounds or environmental factors. Conversely, cataracts with similar morphologies can result from mutations in genes involved in disparate biological pathways, suggesting that cataract is a final endpoint for a variety of different biological insults. That causative mutations have been identified in only a subset of patients with CCs suggests that additional cataract genes have yet to be identified. Although the fraction of families with cataracts caused by uncharacterized genes appears to vary in different populations, it is difficult to estimate from existing studies, most of which examine a subset of candidate genes and do not represent an exhaustive characterization of a random set of families.

As part of an ongoing collaboration between the National Eye Institute (Bethesda, MD, USA) and the National Centre of Excellence in Molecular Biology and Allama Iqbal Medical College in Lahore, Pakistan, this study was designed to identify the genes underlying arCC in the Pakistani population. We screened 83 unlinked arCC families for homozygosity at 33 genes or loci commonly involved in CC and related disorders for possible involvement in disease. In 32 families showing homozygous regions encompassing known CC genes, the respective genes were sequenced, identifying nine disease-causing mutations in 10 families. Overall, including 3 unpublished and 30 previously reported families,[10-20] mutations or loci were detected in 43 of arCC families tested, consistent with known cataract genes or loci being responsible for cataracts in 37.1% of the entire set of families.

Subjects and Methods

Ascertainment of Families and Clinical Analysis

This study was approved by institutional review boards (IRB) of the National Centre of Excellence in Molecular Biology and the Combined Neuroscience (CNS) IRB at the National Institutes of Health. Participating subjects gave informed consent consistent with the tenets of the Declaration of Helsinki. Ophthalmological examinations were performed at the Layton Rehmatullah Benevolent Trust Hospital in Lahore, Pakistan. Detailed family and medical histories were obtained from family members. Presence and types of cataract in both affected and unaffected individuals of the families were confirmed by slit lamp biomicroscopy. A cohort of 143 CC consanguineous families was collected over a period of 10 years. Genome-wide linkage analyses using 384 highly polymorphic microsatellite markers and Sanger sequencing had identified a molecular diagnosis in 3 unpublished and 30 previously reported families.[10-21] From the remaining 110 unlinked families, 83 were selected for homozygosity screening analysis based on the availability of DNA samples and an ar inheritance pattern that in addition included consanguineous matings. In addition, control DNA samples were available from 96 unrelated, ethnically matched Pakistani individuals. Blood samples were obtained from study participants, and DNA was extracted using standard methods, as previously described.[22]

Homozygosity Mapping

Thirty-three candidate genes and loci involved in CC and related disorders or based on expression and function were selected for screening (Table 1). Each was screened for homozygosity by genotyping 1 or 2 microsatellite markers (total of 51 markers; Table 1). The screening algorithm and a summary of the results are shown in Figure 1. The microsatellite markers were selected based on reported high heterozygosity (0.75 or more) and were located within 1 to 2 megabases (Mb) of the candidate gene. If a single marker with 75% or greater heterozygosity was not available, two markers were genotyped. Information on the PCR primer reaction conditions, heterozygosity, and location was obtained from the UniSTS Human Genome Database and National Center for Biotechnology Information (NCBI) Mapview databases. The detection of homozygosity at a given locus in an affected family member was followed by genotyping a second affected family individual at the locus. A variant of the multiplexing short tandem repeat with tailed primers approach described by Oetting et al.[23] using fluorescently labeled tagged primers homologous to extensions on initial primers in a two-PCR approach was used to genotype these microsatellite markers. The PCR products were multiplex electrophoresed on an ABI 3130 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA), and fragment sizes were determined by GeneMapper version 4.0 (Applied Biosystems). Primer sequences and PCR conditions are shown in Supplementary Table S1. If the second individual also was homozygous, linkage was carried out with all available family members to confirm cosegregation of markers with disease. Polymerase chain reaction products were separated on an ABI 3130 DNA Analyzer (Applied Biosystems), and alleles were assigned with GeneMapper Software version 4.0 (Applied Biosystems).

Table 1

Summary of Gene-Specific Microsatellite Markers

Figure 1

Work flow of the present study.

Linkage Analysis

Haplotype comparisons used the Cyrillic 2.1 program (Cyrillic Software, Wallingford, Oxfordshire, UK) for inspection to identify homozygous regions common to affected individuals in each family. Two-point linkage analyses were performed with the FASTLINK version of MLINK from the LINKAGE Program Package.[24,25] Maximum logarithm of the odds (LOD) scores were calculated with ILINK from the LINKAGE Program Package. Autosomal recessive cataracts were analyzed as a fully penetrant trait with a disease allele frequency of 0.0001, and mutation frequency of 0. The marker order and physical distances between the markers were obtained from the Marshfield database and the NCBI chromosome sequence maps.

DNA Sequencing

Mutation screening of candidate gene coding regions used PCR amplification of exons and adjacent intronic regions. Primer pairs for individual exons in the critical interval were designed online with the Primer3 program (http://primer3.sourceforge.net/, in the public domain). Polymerase chain reaction primers for each exon were used for bidirectional sequencing with Big Dye Terminator Ready reaction mix per instructions of the manufacturer (Applied Biosystems). Sequencing was performed using ABI PRISM 3130 automated sequencers (Applied Biosystems) and analyzed using Mutation Surveyor (Soft Genetics, Inc., State College, PA, USA) and the Seqman program of DNASTAR Software (DNASTAR, Inc., Madison, WI, USA). Sequence changes observed were checked for cosegregation in the family and for presence or absence in at least 96 healthy control individuals as well as the 1000 Genomes (http://www.internationalgenome.org/home, in the public domain) and ExAC (http://exac.broadinstitute.org/, in the public domain) databases, although low frequencies of heterozygous changes were not considered to exclude pathogenicity.

Pathogenicity Assessment of Identified Variants

A mutation was considered novel if it was not present in the Human Mutation Database (http://www.hgmd.cf.ac.uk/ac, in the public domain) or the NCBI dbSNP database (http://www.ncbi.nlm.nih.gov/projects/SNP/index.html, in the public domain) and not in Cat-Map (http://cat-map.wustl.edu, in the public domain). A sequence variation was considered pathogenic when it cosegregated with the disease in the family; was not present in 96 randomly selected controls from the Pakistani population; altered a well-conserved amino acid, preferably in a conserved region (http://www.ebi.ac.uk/Tools/clustalw2/index.html, in the public domain); and it was judged significant in computational tests for pathogenicity. Missense variants were assessed for possible causality with the online programs Sorting Intolerant From Tolerant (SIFT; http://sift.jcvi.org/www/SIFT_enst_submit.html, in the public domain) and polymorphism phenotyping (Polyphen-2; http://genetics.bwh.harvard.edu/pph2/, in the public domain), as well as Condel (http://bg.upf.edu/fannsdb/, in the public domain), which uses input from multiple programs in its assessment.

Results

In the 83 unlinked arCC families, fluorescently labeled microsatellite markers flanking each of the 33 genes or loci were genotyped to test for homozygosity. In the first stage, a single affected individual from each family was screened, and all markers tested for the 33 loci were heterozygous in two unlinked arCC families. In the second stage, a second affected individual in each of the 81 remaining families was screened in regions that were homozygous in the first affected individual, and all markers were heterozygous in 49 families. In the remaining 32 families, two-point linkage analyses and haplotype analyses were performed with closely spaced microsatellite markers in regions that were homozygous in both the affected individuals tested in stages 1 and 2. An LOD less than −2 was obtained in 11 of the families, leaving 21 families requiring sequence analysis of candidate genes, of which mutations were identified in 10. The work flow is summarized in Figure 1 and the results of the linkage analysis are shown in Figure 2 and Table 2.

Figure 2

The 13 arCC pedigrees collected from Pakistan including 10 families that were mapped through homozygosity mapping and 3 families that were mapped by genome-wide linkage analysis (denoted by asterisks). Filled symbols denote affected individuals. Pedigrees include haplotypes for two microsatellite and gene mutations. The blackened bars correspond to affected haplotypes with alleles that cosegregate with the disease and that are homozygous in affected individuals.

Table 2

Two-Point LOD Scores of Known Cataract Gene Markers in the 13 arCC Families

Mutation Analysis of Known Cataract Genes in the Homozygous Regions

Twenty-one families had homozygous regions containing 1 or more of the 33 known genes with LOD scores higher than −2 at θ = 0 (Table 2). Sequence analysis of these genes in the corresponding families revealed nine mutations (eight of which were novel) cosegregating with and likely to be causative for CC in 10 of the respective families (Supplementary Fig. S1). In addition, mutations were identified in three families that had undergone unpublished genome-wide linkage analysis, including previously described mutations in FYCO1 and GALK1 and a novel mutation in HSF4, marked as U in Table 3. Pathogenicity of mutations was evaluated using a detailed in silico analysis (Table 3). Domain structures of the encoded genes and the location of the mutations in them are shown in Figure 3, as are the sequence conservation in nine species ranging from humans to zebrafish for missense mutations.

Table 3

Known Cataract Gene Mutations in Pakistani arCC Families

Figure 3

Domain structure and evolutionary conservation of proteins with missense mutations. Graphic overview of the proteins encoded by genes in which mutations were identified in EPHA2, FOXE3, FYCO1, TDRD7, AQP0, HSF4, GALK1, and CRYBA4. Structural or functional domains are depicted, as well as the position of the mutation. Amino acid sequence conservation around residues affected by missense mutations are shown for the five known cataract genes identified in this study. The red bar indicates the position of the mutation. The sequences of proteins or predicted translation products from nine species from humans to zebrafish have been compared and aligned.

Mutations identified included a novel homozygous substitution in EPHA2 exon 10 (c.1814C>T, p.[Thr605Ile]) in families 60061 and 60157; a novel homozygous substitution in FOXE3 (c.307G>A, p.[Glu103Lys]) in family 60039; a known homozygous substitution in FYCO1 exon 8 (c.2206C>T; p.[Gln736*]), in families 60218 and 60228; a c.1129delG frameshift mutation predicted to result in premature termination, p.(Ala377Profs*2), in family 60152; novel homozygous substitution in exon 1 (c.67T>A; p.[Tyr23Asn]) in family 60090; a c.1067T>C p.(Leu356Pro) missense mutation in family 60133; and a novel homozygous c.440G>T (p.[Gly147Val]) substitution in CRYBA4 exon 5 in family 60038.

EPH Receptor A2 (EPHA2) Variation of Uncertain Significance

This variant (rs753345828) has a reported minor allele frequency of 0.00005 in dbSNP (https://www.ncbi.nlm.nih.gov/projects/SNP/) and was not seen in 192 ethnically matched control chromosomes (96 individuals). Families 60061 and 60157 share a common haplotype of 11 consecutive SNP markers across EPHA2, suggesting that they derive the mutant allele from a common ancestor (Supplementary Table S2). The Thr605 residue is conserved among species from humans to chickens, but not in the zebrafish (Fig. 3), suggesting that it is essential for protein function. The SIFT score for this change was 0, predicting that it is deleterious to the protein. However, the PolyPhen-2 program predicts this mutation to be damaging using the HumDiv dataset but benign using the HumVar dataset for comparisons. It is also predicted to be neutral by program Condel (Table 3). Thus, although likely to be pathogenic, the significance of this sequence change is currently uncertain.

Discussion

Here, we describe the results of screening 83 unlinked arCC families for homozygosity at 33 genes or loci known to be involved in arCC. Nine disease-causing mutations were identified in 10 families, and in 11 families no mutations were identified in the linked gene (Supplementary Table S3). We also describe the results of genome-wide linkage analysis in three families for which the mutation had been identified by using a standard linkage approach, but for which the results had not yet been published. Overall, including previously and newly identified mutations, causative genes or loci were identified in 37.1% of the entire set of families studied as part of this project.

The high degree of genetic heterogeneity in arCC makes genetic screening and gene identification expensive and time-consuming. Although this can be approached efficiently by using high-throughput sequencing, this approach is generally more expensive than homozygosity mapping and requires full knowledge of the causative genes and their structure. In contrast to CC in European populations studied, 87% of the families in this project, collected in an unbiased fashion, had arCC, whereas only 13% had adCC or an ambiguous pedigree (Fig. 1). Because of these considerations and the high levels of consanguinity in our families, we chose to use homozygosity testing disease gene loci for arCC. This enables relatively rapid and economical screening of many loci and is particularly useful in analysis of consanguineous families in which regions of several centimorgans adjacent to the disease gene are expected to be identical by descent. Screening of 33 genes or loci in the present study identified putative pathogenic alterations in seven different genes in 10 (12%) of 83 families. Five missense, one nonsense, two frame shift, and one splice site mutations were detected, of which eight were novel.

It is unclear why 11 of the 21 families remaining after linkage analysis did not show a mutation in the included candidate gene. The most likely explanation is that these families were too small to yield a statistically significant LOD score (Table 2), so that the homozygosity is fortuitous and the true locus has yet to be mapped. Another possibility is that these families harbor mutations that might be missed by Sanger sequencing, either in introns or currently unidentified exons or control regions. Also, while studying offspring of consanguineous matings should decrease compound heterozygosity, it is possible that this is responsible for the dropout of some families during homozygosity mapping.

EPHA2 (OMIM 176946) belongs to the A-subclass of receptor tyrosine kinase and interacts with its cognate membrane-anchored ligands to activate cell bidirectional signaling pathway.[26] First described as a cause of ad cataracts in a Caucasian family,[27,28] homozygous recessive mutations were subsequently implicated in arCC in a Pakistani family.[10] To date, nine different mutations in EPHA2 have been reported (see Cat-Map) in 15 families, and EPHA2 has also been implicated in age-related cataract.[27,29-32] Here, we report a novel homozygous EPHA2 missense mutation in two consanguineous Pakistani families. DNA sequencing revealed the transition c.1814C>T, p.(Thr605Ile) in exon 10 located near the protein tyrosine kinase domain of the protein (Fig. 3), suggesting it might alter the tyrosine kinase activity of the EPHA2 protein.

The gene FOXE3 (OMIM 601094), on chromosome 1p33, is a member of the forkhead box gene family,[33] consisting of a single exon encoding a 319–amino acid DNA-binding transcription factor, consistent with a role in the development of the lens placode.[34] The c.307G>A, p.(Glu103Lys), mutation reported in this study is a novel homozygous missense mutation associated with posterior subcapsular cataract. This mutation occurs in a highly conserved amino acid located in the fork head domain (Fig. 3) and might change the ability of FOXE3 to bind DNA.

FYCO1 contains 18 exons and encodes for a coiled coil protein comprising 1478 amino acids (∼167 kDa).[35] Expressed widely including the eye (UniGene, https://www.ncbi.nlm.nih.gov/unigene, in the public domain), it comprises an α-helical RUN domain following by a long coiled-coil region, an FYVE zinc-finger domain, an LC3-interacting region, and a Golgi dynamics domain.[14] FYCO1 interacts directly with LC3, affecting the maturation of p40phox+ phagosomes,[36] to participate in autophagosomal trafficking.[37] The mutation seen in family 60173, c.2206C>T, p.(Gln736*), Table 3, was identified in three previously published families (60003, 60012, and 60069)[13] and one unpublished family (60237). The novel p.(Gln782Argfsx32) mutation was detected in family 60218 (Fig. 3). The c.3151–2A>C, p.(Ala1051Aspfs*27) mutation identified in family 60228 results in the inactivation of a splice acceptor site. All are predicted to cause nonsense-medicated decay of the FYCO1 mRNA and a loss of FYCO1 function in the face of the requirement for turning over large amounts of protein and organelles as part of fiber cell differentiation.

TDRD7 (OMIM 611258) belongs to a large family of Tudor domain–containing proteins, and as an RNA granule component interacts with methylated arginine residues and RNA to control the levels of mRNAs posttranscriptionally.[38] TDRD7 is highly expressed in differentiating fiber cells of the lens. To date, only two mutations in TDRD7 have been reported: a balanced chromosomal rearrangement disrupting the TDRD7 gene causing juvenile cataracts in an isolated patient, and as a cause of arCCs in a family.[38] TDRD7 has also been implicated in age-related cataract.[39] Here, we report a novel homozygous c.1129delG, p.(Ala377Profs*2) TDRD7 frameshift mutation cosegregating with cataracts in a consanguineous Pakistani family (60152; Table 3), possibly resulting in nonsense-medicated decay of the TDRD7 mRNA.

The AQP0 (major intrinsic protein [MIP], OMIM 154050) gene on chromosome 12q13 is a member of the aquaporin family, a ubiquitous family of membrane water transport proteins that confers rapid movements of water across cell membranes. This 263–amino acid intrinsic membrane protein is expressed only in terminally differentiated fiber cells, constituting more than 50% of the total membrane protein in the lens.[40,41] Members of the aquaporin family are predicted to share a unique structure with six transmembrane bilayer-spanning domains (TM1–TM6, Fig. 3).[41] Fourteen different mutations in MIP have been identified in 15 families with different types of ad cataract, as listed in Cat-Map.[9] However, this novel homozygous missense mutation is the first to be associated with ar cataracts. The c.67T>A, p.(Tyr23Asn) mutation occurs in a highly conserved amino acid located within the first transmembrane region of the protein (Fig. 3), suggesting that it might alter the water pore channel function, possibly through affecting water-permeability properties or trafficking. Consistent with the ar cataracts resulting from a loss of function, a knockout mouse model also shows bilateral cataracts.[42]

HSF4 (OMIM 116800) mutations were originally identified in ad cataract,[43] and later in ar cataract families.[44-46] A novel missense mutation, c.433G>C, p.(Ala145Pro), in the sixth exon of HSF4 was found to cosegregate with the disease phenotype in this ar congenital nuclear cataract family (Fig. 1). The Ala145 residue is conserved among different species (Fig. 3), suggesting that it is essential for protein function.

GALK1 (galactokinase, OMIM 604313) contains eight exons and is located on chromosome 17q25.1. It codes for a 392–amino acid protein containing two ATP binding sites (Fig. 3). Mutations in GALK1 cause recessive cataracts,[47] and two mutations, c.410delG, p.(Gly137Valfs*27), and c.416T>C, p.(Leu139Pro), were reported in two Pakistani families.[13] Family 60248 in this study showed ac.766C>T, p.(Arg256Trp), mutation previously reported by Asada et al.,[48] and family 61133 showed a novel c.1067T>C, p.(Leu356Pro), mutation at the junction of the second ATP binding site, which might be important for ATP binding.

The β-crystallin gene family includes three basic (CRYBB) and four acidic (CRYBA) crystallin proteins, believed to derive from a common βγ-crystallin ancestor. All have a highly conserved two-domain, four Greek key motif structure. CRYBA4 (βA4-crystallin, OMIM 123631) encodes a 196–amino acid protein. The c.440G>T, p.(Gly147Val) mutation (Table 3) is the first CRYBA4 mutation to be associated with ar cataracts, suggesting a lack of function in CRYBA4 causes the cataracts, and further that CRYBA4 might have a functional role in the lens beyond that of a structural crystallin. The mutation is in the fourth Greek key motif (Fig. 3), changes a highly conserved amino acid, and thus probably damages the protein structure. This glycine residue is a critical part of a tryptophan corner motif, occurring at the junction of a Greek key and the following β-strand of the barrel, and thought to aid in folding of the Greek key.[49] The glycine residue is two amino acids to the N-terminal of a tryptophan residue (W-2), which forms a hydrogen bond with the hydrophilic W-3 residue (Supplementary Fig. S2).

In total, members in 21 of the 83 families had at least one homozygous region harboring a mutation in a known CC gene. In 10 probands, the causative mutation was identified in the included genes. In the remaining 11 arCC families, sequencing of the known genes in the mapped loci did not reveal a mutation that cosegregated with the disease phenotype (Supplementary Table S3). Clarifying the origin of cataracts in these families remains a challenge. Next-generation sequencing (NGS) has already proven valuable in identifying novel disease genes, both through whole exome or whole genome sequencing and targeted sequencing of linkage intervals or specific genomic regions.[50,51] However, because of the expense of NGS, homozygosity mapping remains effective in terms of cost and time for localizing mutations in patients with arCC in populations with a high frequency of consanguineous matings, such as the Pakistani population.

Taken together with our previous work, mutations and loci were identified in 43 of 116 Pakistani arCC families. FYCO1 was implicated most commonly, with causative mutations identified in 13.8% (16/116) of arCC families, whereas CRYBB3 accounted for 5.2% (6/116) of arCC in the families studied. In addition, the percentage of arCC cases that can be attributed to the other genes in our study cohort is approximately 3.4% for GALK1 (4/116), approximately 2.6% for EPHA2 (3/116), approximately 1.7% each for CRYAB (2/116) and SIL1 (2/116), and approximately 0.9% each for FOXE3 (1/116), TDRD7 (1/116), MIP (1/115), HSF4 (1/116), and CRYBA4 (1/116). Figure 4 summarizes the genetic causes of arCC in the Pakistani families studied. In all, the 11 genes are responsible for approximately 32.8% of inherited cataracts in these 116 Pakistani families, with cataracts in the remaining families excluded from these loci.

Figure 4

Frequency of cataract gene mutations in the Pakistani population. Pie chart showing the frequencies of cataract gene mutations in the Pakistani population as seen in this study and our previous studies.

Overall, this work demonstrates that homozygosity mapping is an efficacious and economical initial step in localizing genetic defects of consanguineous arCC families, allowing insight into the genetic architecture of arCC in the Pakistani population. In addition, these results lay the groundwork for screening larger groups of arCC families by using a similar approach followed by NGS to identify the causative genes in all families. The current advances in conventional and genetic therapies mean that knowledge of the genetic causes of disease in these patients is becoming increasingly valuable for their medical treatment.

Accession codes. GenBank: EPHA2 mRNA, NM_004431.3; EPHA2 protein, NP_004422.2. FOXE3 mRNA, NM_012186.2, protein, NP_036318.1. FYCO1 mRNA, NM_024513.3, FYCO1 protein, NP_078789.2. TDRD7 mRNA NM_014290.2; TDRD7 protein NP_055105.2. MIP mRNA, NM_012064.3; MIP protein, NP_036196.1. GALK1 mRNA, NM_000154.1; GALK1 protein, NP_000145.1. CRYBA4 mRNA, NM_001886.2; CRYBA4 protein, NP_001877.1.

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.