Identification of a non‐coding SNP associated with risk for non‐syndromic orofacial clefting with allele‐specific effects on IRF6 expression in vitro

Abstract

Introduction Genome wide association studies (GWAS) have identified loci associated with risk for non‐syndromic cleft lip with or without cleft palate (NSCL/P). At every locus there are multiple single nucleotide polymorphisms (SNPs) associated with disease and it is a challenge to distinguish functional SNPs from those merely in linkage disequilibrium with functional SNPs. Objective At many of the associated loci, the presumed risk‐relevant gene is expressed in oral epithelium. We hypothesize that at such loci functional SNPs have allele‐specific effects on the activity of oral epithelium enhancers. Our objective is to identify the candidate functional SNPs for NSCL/P at each of the loci identified by GWAS. Materials and Methods To test SNPs for their effects on enhancer activity we carry out massively parallel reporter assays (MPRAs) on 890 SNPs from 8 loci in the GMSM‐K human fetal oral epithelium cell line. To test the SNP‐target gene association we use the Activity‐by‐Contact method, which incorporates open chromatin, H3K27Ac, and HiC data. To confirm MPRA results on top‐scoring SNPs we perform luciferase reporter assays. We use CRISPR‐mediated homology directed repair to engineer GMSM‐K cells to be homozygous for risk or non‐risk alleles for promising SNPs and assess the expression level of the target gene in engineered cells of each genotype. Assays that pass these assays are tested in allele‐specific reporter assays in zebrafish and mouse embryos. Results Using the described methods we tested 608 NSCL/P‐associated SNPs in the IRF6 locus and 11 such SNPs in the FOXE1 locus. Two SNPs in the first group and one in the second had consistent allele‐specific effects in the MPRA and in luciferase assays. Importantly, for one SNP in the IRF6 locus, we have engineered cells to be homozygous for the non‐risk‐associated or risk‐associated allele and found expression of IRF6 is lower in the latter. In vivo enhancer reporter assays in mouse and zebrafish embryos are ongoing. Conclusion Out of more than 600 NSCL/P‐risk‐associated SNPs in the IRF6 locus, we have evidence that a particular one directly contributes to pathogenesis for this disorder. Significance This study illustrates a method to screen among multiple non‐coding SNPs identified by GWAS to identify candidates for those that are functional. Moreover, identifying the functional SNP near IRF6 is a step towards illuminating how a common variant contributes risk for this disorder.