A Type 2 Diabetes-Associated Functional Regulatory Variant in a Pancreatic Islet Enhancer at the ADCY5 Locus

Abstract

Molecular mechanisms remain unknown for most type 2 diabetes genome-wide association study identified loci. Variants associated with type 2 diabetes and fasting glucose levels reside in introns of ADCY5, a gene that encodes adenylate cyclase 5. Adenylate cyclase 5 catalyzes the production of cyclic AMP, which is a second messenger molecule involved in cell signaling and pancreatic β-cell insulin secretion. We demonstrated that type 2 diabetes risk alleles are associated with decreased ADCY5 expression in human islets and examined candidate variants for regulatory function. rs11708067 overlaps a predicted enhancer region in pancreatic islets. The type 2 diabetes risk rs11708067-A allele showed fewer H3K27ac ChIP-seq reads in human islets, lower transcriptional activity in reporter assays in rodent β-cells (rat 832/13 and mouse MIN6), and increased nuclear protein binding compared with the rs11708067-G allele. Homozygous deletion of the orthologous enhancer region in 832/13 cells resulted in a 64% reduction in expression level of Adcy5, but not adjacent gene Sec22a, and a 39% reduction in insulin secretion. Together, these data suggest that rs11708067-A risk allele contributes to type 2 diabetes by disrupting an islet enhancer, which results in reduced ADCY5 expression and impaired insulin secretion.

Figure 1

rs11708067 overlaps pancreatic islet open chromatin and a strong enhancer chromatin state in ADCY5 intron 3. The 50-kb region shown includes all 17 variants in strong LD (r² >0.8, 1000 Genomes phase 3 EUR) with the type 2 diabetes–associated single nucleotide polymorphism rs11708067 (in boldface type) (18 total noncoding candidate variants located within ADCY5 introns 1–3). Selected Encyclopedia of DNA Elements (ENCODE) open chromatin (DNase, FAIRE) (21,22), chromatin accessibility (ATAC-seq) (20), and RNA-seq and chromatin state tracks (28) are shown. The black vertical line above the pancreatic islet open chromatin tracks represents the 231-bp segment containing rs11708067 that was tested in luciferase reporter assays. GM12878, lymphoblastoid cells from B cells; H1 ES, embryonic stem cells; HepG2, human hepatocellular carcinoma cells; HMEC, human mammary epithelial cells; HSMM, human skeletal muscle myoblasts; HUVEC, human umbilical vein endothelial cells; islets, pancreatic islets; K562, leukemia cell line; NHEK, epidermal keratinocytes; NHLF, lung fibroblasts.

Figure 2

Evidence of allelic imbalance in H3K27ac ChIP-seq reads at rs11708067. H3K27ac ChIP-seq reads in a primary human islet sample heterozygous at rs11708067. Blue indicates reads that contain the G allele of rs11708067, red indicates reads that contain the A allele of rs11708067, and gray indicates reads in the peak that do not overlap rs11708067.

Figure 3

rs11708067 exhibits allelic differences in transcriptional activity and protein binding in 832/13 rat insulinoma cells. A: The 231-bp segments containing allele A or G of rs11708067 were cloned into a pGL4.23 luciferase reporter vector upstream of the minimal promoter in both orientations (forward and reverse). The relative luciferase activities of plasmids transfected into 832/13 cells normalized to an empty vector control are shown on the y-axis. Error bars represent SD of three to five independent clones per allele (t tests). Black triangles, empty vector; black circles, rs11708067-A allele; white circles, rs11708067-G allele. B: EMSAs with biotin-labeled probes containing either the A or G allele of rs11708067. Probes were incubated with 2.2 µg 832/13 rat insulinoma nuclear lysate. The arrows indicate differential protein binding to the A allele, which is competed away by 240-fold excess competitor DNA containing the A allele (lane 3). EMSAs were performed six separate times with consistent results. Antibodies to CEBPB and NKX2.2 (8 µg) were tested for supershifts in lanes 4 and 5, respectively, and lanes 9 and 10, respectively.

Figure 4

Deletion of the orthologous enhancer element in 832/13 cells leads to reduced Adcy5 expression and reduced insulin secretion. A: Three homozygous deletions (indicated by the genomic regions in between the Xs) of the orthologous Adcy5 enhancer region were generated using CRISPR-Cas9 technology in 832/13 rat insulinoma cells. UCSC Genome Browser chromosome 11 coordinates are shown for the rat genome rn5. PhyloP Cons indicates basewise conservation with 13 species (rat, mouse, guinea pig, human, chimp, rhesus, cow, dog, panda, opossum, chicken, turkey, and zebrafish) based on the phyloP track from the UCSC Genome Browser. Gene expression normalized to B2m control (shown on the y-axes for B and C) or insulin secretion (shown on the y-axis for D) was measured in 832/13 rat β-cells. B: Normalized Sec22a nearby control gene expression (displayed as a percent of average expression of intact control clones). C: Normalized Adcy5 expression (displayed as a percent of average expression of intact control clones). D: Insulin secretion at 3 mmol/L glucose concentration (displayed as pg insulin/µg total protein). B and C: For gene expression, error bars indicate the SD of three to five clones per sample for two independent experiments. D: Insulin secretion assays were performed in duplicate or triplicate, and error bars indicate the SD of two to five clones per sample. Unpaired t tests were performed to compare gene expression or insulin secretion of enhancer homozygous deletion clones with intact control clones. Black circles, wild-type (WT) 832/13 clones; black triangles, intact control clones; white circles, enhancer homozygous deletion clones.