Biology meets bioinformatics: Validating the predicted role of Alzheimer's GWAS SNPs on micro-RNA function

Abstract

The risk to develop non-Mendelian Alzheimer's disease (AD) is mainly genetically determined. While recent genome-wide association studies (GWAS) uncovered several single nucleotide polymorphisms (SNPs) associated with AD risk, the functional basis for most of these signals remains to be determined. Here, we explore the hypothesis that GWAS SNPs associated with AD may interfere with micro-RNA (miRNA) function. miRNAs are a class of RNA involved in post-transcriptional regulation of protein expression assumed to play important roles in nervous system development and function. We developed a bioinformatic algorithm to systematically assess and rank the potential role of AD-associated SNPs (or their proxies) in affecting miRNA-mRNA binding in 3'UTR regions (see companion abstract by Schilling et al.). For this study, we selected the top 14 “microSNPs” for in vitro follow up. The wild-type 3'UTR of each target gene was subcloned downstream of a renilla luciferase reporter, the AD-associated allele was introduced via site-directed mutagenesis, followed by co-transfection of both constructs with the respective miRNA into 293HEK cells. Luciferase activity was measured 24 hours after transfection. Pilot data using a different set of AD-unrelated markers suggests that approximately two thirds of the SNPs identified by our in silico tool also show significant effects in vitro. From a list of more than 15,000 AD-associated SNPs emerging from GWAS meta-analyses, we identified 736 with strong predicted effects on miRNA-mRNA binding using our novel in silico tool. Interestingly, this included one SNP from the current AlzGene “Top Results” list [see www.alzgene.org], i.e. rs610932 in MS4A6A (membrane-spanning 4-domains, subfamily A, member 6A) whose minor G-allele is predicted to lead to an increased protein expression. This and 13 additional SNPs are currently being tested via luciferase reporter assays to assess which of the in silico effects can be validated in vitro. We have developed a systematic whole-genome pipeline employing both in silico and in vitro experiments to assess the potential role of AD-associated SNPs on miRNA-mediated protein expression. Our approach can be applied to other complex diseases and facilitates the functional interpretation of findings from GWAS or whole-genome sequencing studies.