Integrating bioinformatics and physiology to describe genetic effects in complex polygenic diseases

Type 2 diabetes mellitus (T2DM) results from interaction between genetic and environmental factors. The worldwide prevalence of T2DM is increasing rapidly due to reduction in physical activity, increase in dietary intake, and the aging of the population. This thesis has focused on dissecting the genetic contribution in T2DM using largescale genomic approaches with a particular emphasis on analysis of gene transcripts in different tissues, predominantly muscle. In paper I, we identified TXNIP as a gene whose expression is powerfully suppressed by insulin yet stimulated by glucose. In healthy individuals, its expression was inversely correlated to total body measures of glucose uptake. Forced expression of TXNIP in cultured adipocytes significantly reduced glucose uptake, while silencing with RNA interference in adipocytes and in skeletal muscle enhanced glucose uptake, confirming that the gene product is also a regulator of glucose uptake. TXNIP expression is consistently elevated in the muscle of pre-diabetics and diabetics, although in a panel of 4,450 Scandinavian individuals, we found no evidence for association between common genetic variation in the TXNIP gene and T2DM. TXNIP regulates both insulindependent and insulin-independent pathways of glucose uptake in human skeletal muscle. Combined with recent studies that have implicated TXNIP in pancreatic β-cell glucose toxicity, our data suggest that TXNIP might play a key role in defective glucose homeostasis preceding overt T2DM. In paper II, we investigated molecular mechanisms associated with insulin sensitivity in skeletal muscle by relating global skeletal muscle gene expression to physiological measures of the insulin sensitivity. We identified 70 genes positively and 110 genes inversely correlated with insulin sensitivity in human skeletal muscle. Most notably, genes involved in a mammalian target-of-rapamycin signaling pathway were positively whereas genes encoding extracellular matrix structural constituent such as extracellular matrix-receptor, cell communication, and focal adhesion pathways were inversely correlated with insulin sensitivity. More specifically, expression of CPT1B was positively and that of LEO1 inversely correlated with insulin sensitivity, a finding which was replicated in an independent study of 9 non-diabetic men. These data suggest that a high capacity of fat oxidation in mitochondria is reflected by a high expression of CPT1B which is a marker of insulin sensitivity. In paper III, we investigated molecular mechanisms associated with maximal oxygen uptake (VO2max) and type 1 fibers in human skeletal muscle. We identified 66 genes positively and 83 genes inversely correlated with VO2max and 171 genes positively and 217 genes inversely correlated with percentage of type 1 fibers in human skeletal muscle. Genes involved in oxidative phosphorylation (OXPHOS) showed high expression in individuals with high VO2max, whereas the opposite was not the case in individuals with low VO2max. Instead, genes such as AHNAK and BCL6 were associated with low VO2max. Also, expression of the OXPHOS genes, NDUFB5 and ATP5C1, increased with exercise training and decreased with aging. In contrast, expression of AHNAK in skeletal muscle decreased with exercise training and increased with aging. These findings indicate that VO2max closely reflects expression of OXPHOS genes, particularly that of NDUFB5 and ATP5C1 in skeletal muscle and high expression of these genes suggest good muscle fitness. In contrast, a high expression of AHNAK was associated with a low VO2max and poor muscle fitness. In paper IV, we combined results from the Diabetes Genetics Initiative (DGI) and the Wellcome Trust Case Control Consortium (WTCCC) genome-wide association (GWA) studies with genome-wide expression profiling in pancreas, adipose tissue, liver, and skeletal muscle in patients with or without T2DM or animal models thereof to identify novel T2DM susceptibility loci. We identified 453 single nucleotide polymorphisms (SNPs) associated with T2DM with P < 0.01 in at least one of the GWA studies and 150 genes that were located in vicinity of these SNPs. Out of these 150 genes, we identified 41 genes differentially expressed using publicly available gene expression profiling data. Most notably, we were able to identify four genes namely IGF2BP2, CDKAL1, TSPAN8, and NOTCH2 for which SNPs located in vicinity of these genes have shown association with T2DM in different populations. In addition, we identified a SNP (rs27582) in the CAST gene which was associated with future risk of T2DM (odds ratio (OR) = 1.10, 95% CI: 1.00-1.20, P < 0.05) in a prospective study of 16,061 Swedish individuals followed for more than 25 years; this association was stronger in lean individuals (OR = 1.19, 95% CI: 1.03-1.36, P = 0.024). Moreover in the Botnia Prospective Study (BPS) involving 2,770 individuals followed for more than 7 years, carriers of the A-allele were more insulin resistant than carriers of the G-allele as indicated by higher fasting insulin concentrations (regression coefficient (β) = 0.048, P = 0.017) and higher HOMA-IR index (β = 0.044, P = 0.025) as well as lower insulin sensitivity index during OGTT (β = -0.039, P = 0.039) at follow-up. In conclusion, using gene expression in different tissues from patients with T2DM and animal models is a powerful tool for prioritizing SNPs from GWA studies for replication studies. We thereby identified association of a variant (rs27582) in the CAST gene with T2DM and insulin resistance