Common genetic variants near the Brittle Cornea Syndrome locus ZNF469 influence the blinding disease risk factor central corneal thickness

Abstract

Central corneal thickness (CCT), one of the most highly heritable human traits (h(2) typically>0.9), is important for the diagnosis of glaucoma and a potential risk factor for glaucoma susceptibility. We conducted genome-wide association studies in five cohorts from Australia and the United Kingdom (total N = 5058). Three cohorts were based on individually genotyped twin collections, with the remaining two cohorts genotyped on pooled samples from singletons with extreme trait values. The pooled sample findings were validated by individual genotyping the pooled samples together with additional samples also within extreme quantiles. We describe methods for efficient combined analysis of the results from these different study designs. We have identified and replicated quantitative trait loci on chromosomes 13 and 16 for association with CCT. The locus on chromosome 13 (nearest gene FOXO1) had an overall meta-analysis p-value for all the individually genotyped samples of 4.6x10(-10). The locus on chromosome 16 was associated with CCT with p = 8.95x10(-11). The nearest gene to the associated chromosome 16 SNPs was ZNF469, a locus recently implicated in Brittle Cornea Syndrome (BCS), a very rare disorder characterized by abnormal thin corneas. Our findings suggest that in addition to rare variants in ZNF469 underlying CCT variation in BCS patients, more common variants near this gene may contribute to CCT variation in the general population.

Figure 1. Significant association of central corneal thickness on chromosome 16 from the meta-analysis of the AU and UK twin cohorts.

The top SNP rs9938149 had a p-value of 2.54×10⁻¹⁰ in the imputation set. The degree of linkage disequilibrium (LD) between rs9938149 and other SNPs is indicated by red shading. The recombination rate is displayed as a light blue line, with scale on the right hand axis. This imputed SNP is about 108 kb away from gene ZNF469 (16q24, build 36.3), and there is evidence for recombination in the intervening region.

Figure 2. Association with CCT for variants on chromosome 13 from the meta-analysis of the AU and UK twin cohorts.

The genotyped variant rs2755237 remained as the top SNP in the imputation set. It is about 20 kb from 3′ end of gene FOXO1 (13q14.1, build 36.3). The degree of linkage disequilibrium between rs2755237 and other SNPs is indicated by red shading. The recombination rate is displayed as a light blue line, with scale on the right hand axis.

Figure 3. Comparison of the allelic effects estimated from the four cohorts.

The four cohorts in comparison are the combined Australian twin cohorts (AU), the UK twin cohort (UK), the Blue Mountains population-based cohort in DNA pooling design (DNA), and the Adelaide population-based cohort in Blood pooling design (BLOOD). The mean allelic effect from each cohort is marked as red dot, with standard errors marked as bars. The overall meta-analysis p-value based on all the cohorts is shown under the x-axis label for each sub-figure. These SNPs are the most significant variants found in the meta-analysis of twin cohorts. The estimated effects from pooled samples of rs12447690 and rs9938149 around ZNF469 region, were smaller than the effects in the twin cohorts. rs2755237 near FOXO1 replicated well in pooled samples, with similar effects shown in all the cohorts. Two SNPs within gene FAM53B had an opposite direction of the effects found in the Blood pools.