Identifying genetic variation contributing to keratoconus

Keratoconus (OMIM 148300) is a complex disease characterised by progressive stromal thinning and conical protrusion of the cornea. These abnormalities usually develop in the second decade of life and can lead to severe visual impairment as a result of high myopia and irregular astigmatism. Due to the early onset of disease, keratoconus affects individuals during their most productive years and quality of life estimates demonstrate a significant decline over time that is disproportional to visual acuity. The global incidence of keratoconus is approximately 1 in 50,000 and the prevalence in Caucasians is reported between 55 and 265 per 100,000 individuals. While keratoconus is a complex disease, it has a strong genetic basis. Familial examples display both autosomal dominant and recessive inheritance patterns and first-degree relatives have at least a 15-fold increased risk of developing keratoconus, compared to the general population. Current treatment options for keratoconus are suboptimal, require invasive surgical procedures and have limited applicability and effectiveness in early-onset and advanced disease. Currently, keratoconus is the second leading indication for corneal transplants in Australia and there is a great need to better understand keratoconus aetiology and pathophysiology to aid early diagnosis and the development of novel treatments to improve patient outcomes. Therefore, the overarching aim of this project was to identify genetic variants involved in keratoconus susceptibility through the assessment of individuals and families in our keratoconus DNA repository. This aim was achieved through three distinct studies: a candidate gene screen; novel variant discovery in families with multiple cases of severe and early-onset keratoconus; and association analysis in a large case-control cohort. The gene-screen aimed to determine if rare, protein-coding variants predicted to be pathogenic were enriched in in keratoconus-candidate genes in our cohort of keratoconus cases of European descent compared to ethnically matched controls. By combining whole exome-sequencing and targeted gene sequencing, a total of 22 keratoconus candidate genes were assessed in 385 unrelated cases, making this the most comprehensive study of its kind to date. Two control cohorts were used for comparison, including up to 396 population controls (unassessed for eye disease) and 230 individuals without keratoconus. The candidate genes were selected from the literature and included genes near associated loci, genes harbouring putative disease-causing variants in family studies, and genes proposed to play a role in keratoconus based on a known function in the cornea or involvement in a postulated disease process. Specifically, this study examined the frequency of potentially pathogenic variants in MPDZ, RXRA, RAB3GAP1, FOXO1, BANP, ZNF469, HGF, COL5A1, IMMP2L, FNDC3B, NFIB, ILRN, SLC4A11, CAST, COL4A3, COL4A4, TF, SOD1, VSX1, RAD51, IL1A and IL1B. For all genes, no difference was observed in the frequency of rare, protein-coding potentially pathogenic variants between the cases and controls, suggesting that rare protein-coding variation in these genes do not play a major role in keratoconus susceptibility. The second study utilised whole genome sequencing (WGS) to investigate the genetic basis of disease in two families with severe, early-onset keratoconus. The aim of this study was to identify putatively disease-causing variants that segregate with disease in these families. One family was Jordanian, KCNSW01, with eight cases of keratoconus and three unaffected family members across three generations. While the parents in the first generation are both apparently unaffected, the inheritance pattern in the second-, and third-, generation was indicative of autosomal dominant inheritance of keratoconus. The second family, KSA197, was a family of Italian heritage with two affected brothers born to unaffected, second-cousin parents. Both brothers are affected with keratoconus and one has an unaffected child, consistent with autosomal recessive inheritance. WGS was obtained for 11 individuals from KCNSW01 and five family members from KSA197. A subset of ~250,000 single nucleotide polymorphisms (SNPs) across the autosomes were extracted from the WGS and used to conduct linkage analysis in KCNSW01 and homozygosity mapping in KCNSW01. Two linkage regions with equal maximum logarithm of the odds (LOD) scores of 2.1 were identified in KCNSW01: 17q12 and 20p13-12.2. The disease-associated haplotypes at 17q12 and 20p13-12.2 were inherited from the matriarch and patriarch (both unaffected), respectively. All affected individuals carry both haplotypes, suggesting digenic inheritance of keratoconus in this family. For KSA197, a single homozygous region shared by the two affected brothers was identified at 16p12.1. For both families, variants that segregated with the disease-associated haplotype(s) were extracted and further investigated. No rare protein-coding variants fulfilled the criteria for putatively disease-causing in either family. Non-coding variation that segregated with disease were therefore prioritised based on the minor allele frequencies in the gnomAD database, predictions of deleteriousness and pathogenicity and whether or not they were located in known regulatory regions. One putatively disease-causing variant was identified in KSA197 and a total of 44 were identified KCNSW01, including a compelling variant located in an untranslated region of the spermine oxidase gene (SMOX). This novel candidate gene encodes the SMOX protein which plays a role in apoptosis and the cellular response to ultraviolet radiation and oxidative stress. These pathways are biologically-plausible in the keratoconus disease process and therefore the specific variant and the gene should be further investigated. The overall aim of the third study was to identify putatively functional variants with a role in keratoconus susceptibility. This study was separated into specific aims. Aim 1 was designed to identify novel keratoconus-associated variants in a large case-control study. As keratoconus is a disease in which the cornea progressively thins, it was hypothesised that variants that contribute to central corneal thickness (CCT) in the general population would also contribute to keratoconus risk. Therefore, 72 SNPs known to contribute to CCT were assessed for association in a cohort of 536 keratoconus cases and 2,574 controls of European descent. Five SNPs were significantly associated with keratoconus following correction for multiple testing, including a novel association at rs2268578 in an intronic region of the lumican gene (LUM). The remaining four SNPs ‚Äö- rs1536482 and rs3132303 located between RXRA and COL5A1, rs2755238 in the second intron of FOXO1 and rs66720556 between MPDZ and NFIB ‚Äö- had previously shown either a significant or suggestive association with keratoconus. Aim 2 focused on identifying the functional variants that underlie the associations at keratoconus-associated loci. Six regions were fine-mapped: RXRA-COL5A1, FOXO1, FNDC3B, MPDZ-NFIB, RAB3GAP1 and the novel locus LUM. This analysis was conducted in a cohort of 487 keratoconus cases and 626 unaffected controls with genome-wide genotyping data. To appropriately capture the association peaks during fine-mapping, variants across the surrounding gene were included for intronic loci, whereas the region encompassing the flanking genes were included for intergenic loci. Strong association peaks were observed at all loci except the previously reported RAB3GAP1 locus, thus this locus was not further analysed. To further assess variation carried on the risk-associated haplotypes at remaining loci, keratoconus patients carrying the risk allele for the top SNP determined in Aim 2 were selected for re-sequencing in Aim 3. A total of 178 cases and 62 controls were re-sequenced across the five loci. Variants at each locus were filtered to identify those that were carried on the risk-associated haplotype; in high LD with the top SNP as measured by D' to ensure the capture of both common and rare variants; and were more common in the cases compared to the controls and all populations available in Genome Aggregation Database (gnomAD). These variants were further prioritised based on deleteriousness/pathogenicity predictions and whether or not the variant was likely to disrupt a regulatory region. This analysis identified putatively functional variants at all five loci, and proposed rs79728429 as a functional variant at the FOXO1 (rs2721051) locus. From this work, it was further hypothesised that that rs79728429 alters the expression of a novel uncharacterised gene, AL133318.1, and that this altered expression in the cornea confers an increased susceptibility to keratoconus at this locus. This dissertation comprehensively investigated genetic variation in keratoconus susceptibility in a cohort of Australian keratoconus patients of European descent using three distinct studies and methodologies. In the largest study of its kind to date, this project demonstrated that rare coding variation in 22 keratoconus-candidate genes were unlikely to contribute broadly to keratoconus. The family-based study demonstrated strong evidence of digenic inheritance of keratoconus in one family with the discovery of two linkage regions of equal significance (17q12 and 20p13-12.2) and identified a homozygous region on 16p12.1 in a family with recessive disease. Putatively-disease causing variants within these regions were identified and prioritised for further investigation, including an appealing variant located in the 5' UTR of SMOX. Finally, this project identified a novel keratoconus-associated locus overlapping LUM with rs3759221 as the top SNP. Putatively functional variants were prioritised at five key keratoconus-associated loci, including a compelling novel putatively functional variant (rs79728429) at t...