Identification of novel type 1 diabetes candidate genes by integrating genome-wide association data, protein-protein interactions, and human pancreatic islet gene expression

Abstract

Genome-wide association studies (GWAS) have heralded a new era in susceptibility locus discovery in complex diseases. For type 1 diabetes, >40 susceptibility loci have been discovered. However, GWAS do not inevitably lead to identification of the gene or genes in a given locus associated with disease, and they do not typically inform the broader context in which the disease genes operate. Here, we integrated type 1 diabetes GWAS data with protein-protein interactions to construct biological networks of relevance for disease. A total of 17 networks were identified. To prioritize and substantiate these networks, we performed expressional profiling in human pancreatic islets exposed to proinflammatory cytokines. Three networks were significantly enriched for cytokine-regulated genes and, thus, likely to play an important role for type 1 diabetes in pancreatic islets. Eight of the regulated genes (CD83, IFNGR1, IL17RD, TRAF3IP2, IL27RA, PLCG2, MYO1B, and CXCR7) in these networks also harbored single nucleotide polymorphisms nominally associated with type 1 diabetes. Finally, the expression and cytokine regulation of these new candidate genes were confirmed in insulin-secreting INS-1 β-cells. Our results provide novel insight to the mechanisms behind type 1 diabetes pathogenesis and, thus, may provide the basis for the design of novel treatment strategies.

FIG. 1.

Overview of study design. Flowchart demonstrating the major steps in the study process. ND, gene expression not detected.

FIG. 2.

Networks enriched for both cytokine-regulated and type 1 diabetes genes. Protein interaction networks for networks 3 and 4. These networks contained excess of type 1 diabetes input genes and were significantly enriched in differentially expressed genes in our model comparing unstimulated human pancreatic islets with the same islet preparations stimulated with proinflammatory cytokines. Genes significantly upregulated are shown as red nodes; genes demonstrating significant downregulation are shown as blue nodes. A summary of gene function for differentially regulated genes in the networks can be found in Supplementary Table 4. Type 1 diabetes input proteins have red labels, and their chromosomal region is also shown. Edges between proteins that physically interact are colored in orange. The width of the edges depends on the confidence score to each protein association in STRING. Networks were displayed and colored using Inkscape.

FIG. 3.

Network enriched for cytokine-regulated but not type 1 diabetes genes. Protein interaction network for network 16. This network did not contain excess of type 1 diabetes input genes as compared with randomly generated networks of similar size and topological characteristics, but it was significantly enriched for cytokine-regulated genes in human islets. Furthermore, this network exclusively contains physical interactions. Genes significantly upregulated are shown as red nodes; genes demonstrating significant downregulation are shown as blue nodes. A summary of gene function for differentially regulated genes in the networks can be found in Supplementary Table 4. Type 1 diabetes input proteins have red labels, and their chromosomal region is also shown. The width of the edges depends on the confidence score to each protein association in STRING. Inkscape was used to display and color the network.

FIG. 4.

Gene expression profiling in INS-1 cells of novel candidate genes. Gene expression was evaluated in INS-1 cells after stimulation with IL-1β and IL-1β + IFN-γ and correlated to basal expression level. Results were normalized against PPIA as endogenous control and shown as average fold changes. Significance levels (paired t test) for comparisons against control: *P < 0.05, **P < 0.01, ***P < 0.001. ND, gene expression not detected.