Integrated analysis of pancreatic islet eQTLs and regulatory state maps identifies putative causal mechanisms at T2D associated loci. M. van de Bunt1,2, J. E. Manning Fox3, K. J. Gaulton2, A. Barrett1, X. Q. Dai3, M. Ferdaoussi3, P. E. MacDonald3, M. I. McCarthy1,2,4, A. L. Gloyn1,4 1) Oxford Centre for Diabetes, Endocrinology & Metabolism, University of Oxford, Oxford, United Kingdom; 2) Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford, United Kingdom; 3) Alberta Diabetes Institute, Li Ka Shing Centre, University of Alberta, Edmonton, Canada; 4) Oxford NIHR Biomedical Research Centre, Churchill Hospital, Oxford, United Kingdom.

   To date around 80 loci have been found associated with type 2 diabetes (T2D). There is compelling evidence that many of these signals act through islet regulatory mechanisms. To define underlying causal mechanisms we therefore need to connect risk-associated regulatory variants to their effector transcripts. We aimed to further mechanistic insight by integrating maps of islet regulatory elements and expression quantitative trait loci (eQTLs). We profiled transcript levels in 142 human islet samples through RNA sequencing on the Illumina HiSeq2000 platform, and obtained genotypes using the Illumina HumanOmni2.5-Exome array and imputation into the 1000 Genomes Phase 1v3 cosmopolitan panel. Reads were aligned to the genome (GRCh37) with TopHat2 and raw read counts were used for exon quantification. To account for potential noise introduced by, for example, differences in donor characteristics and islet purity, exon counts were normalized using PEER. We used data from the 120 individuals of European ancestry in this study to derive islet exon eQTLs. For 74 known common T2D association loci, genetic data were integrated with islet eQTLs and islet ChromHMM regulatory state maps. To be considered, at a given locus the most significant eQTL variant (of all variants with P<0.01) had to be in strong LD (r2>0.8) with the lead T2D variant, and either the GWAS or eQTL lead variant had to overlap an islet-active regulatory state annotation. In total, our analysis uncovered a single potential effector transcript at 14 loci. We have started to functionally validate these genes, beginning with the most significant coinciding islet cis-eQTL at ZMIZ1 (P=8.1x10-6). Immunohistochemistry showed that within the pancreas ZMIZ1 localizes to the islet. We then determined gene expression levels in non-diabetic and T2D islets (n=6 each), which showed significantly higher ZMIZ1 levels in T2D islets (P=0.01) in line with the observed higher ZMIZ1 expression level for the T2D risk allele (=0.55). Overexpression of ZMIZ1 in human islets (n=6) resulted in a significant reduction in insulin exocytosis (P<0.0001). Conversely, knockdown of ZMIZ1 in T2D islets (n=3) increased insulin exocytosis (P<0.0001), and could rescue the blunted exocytotic response normally seen in T2D. Our results demonstrate the power of integrating multiple islet genomic annotations to deliver effector transcripts and molecular mechanisms underlying T2D association signals.

You may contact the first author (during and after the meeting) at