Exome array analysis identifies new loci and low-frequency variants influencing insulin processing and secretion

Abstract

Insulin secretion has a crucial role in glucose homeostasis, and failure to secrete sufficient insulin is a hallmark of type 2 diabetes. Genome-wide association studies (GWAS) have identified loci contributing to insulin processing and secretion; however, a substantial fraction of the genetic contribution remains undefined. To examine low-frequency (minor allele frequency (MAF) 0.5-5%) and rare (MAF < 0.5%) nonsynonymous variants, we analyzed exome array data in 8,229 nondiabetic Finnish males using the Illumina HumanExome Beadchip. We identified low-frequency coding variants associated with fasting proinsulin concentrations at the SGSM2 and MADD GWAS loci and three new genes with low-frequency variants associated with fasting proinsulin or insulinogenic index: TBC1D30, KANK1 and PAM. We also show that the interpretation of single-variant and gene-based tests needs to consider the effects of noncoding SNPs both nearby and megabases away. This study demonstrates that exome array genotyping is a valuable approach to identify low-frequency variants that contribute to complex traits.

Figure 1

Manhattan plot for the fasting proinsulin analysis. Association results of the single-variant analysis (−log₁₀ P values) are plotted against genomic position (NCBI Build 37). Previously identified loci are denoted in blue and loci identified by the current study in red. Fasting proinsulin levels were log-transformed and adjusted for fasting insulin, body mass index, age, and age.

Figure 2

The MADD gene is located in a region of unusually high linkage disequilibrium on chromosome 11 from 46 to 57 Mb. Regional association results of the single-variant analysis (−log₁₀ P values) are plotted against genomic position (NCBI Build 37) for fasting proinsulin before (a) and after (b) adjustment of the lead SNPs for the common GWAS signals (rs7944584 and rs1051006) and the nonsense variant rs35233100 (MAF 3.7%) at MADD. Fasting proinsulin levels were log-transformed and adjusted for fasting insulin, body mass index, age, and age. The conditioning SNPs are indicated in blue. For clarity, only a portion of the 11 Mb region and a subset of the genes are shown.