Context-specific eQTLs implicate differential genomic regulatory mechanisms in obese and lean Finns. A. Ko1,2, R. C. Cantor1, E. Nikkola1, M. Alvarez1, B. Pasaniuc1,3,4, K. L. Mohlke5, M. Boehnke6, F. S. Collins7, J. Kuusisto8, M. Laakso8, P. Pajukanta1,2 1) Department of Human Genetics, David Geffen School of Medicine at UCLA, Los Angeles, CA; 2) Molecular Biology Institute at UCLA, Los Angeles, CA; 3) Department of Pathology and Laboratory Medicine, David Geffen School of Medicine at UCLA, Los Angeles ,CA; 4) Bioinformatics Interdepartmental Program, UCLA, Los Angeles, CA; 5) Department of Genetics, University of North Carolina, Chapel Hill, NC; 6) Department of Biostatistics and Center for Statistical Genetics, School of Public Health, University of Michigan, Ann Arbor, MI; 7) National Human Genome Research Institute, National Institutes of Health, Bethesda, MD; 8) Department of Medicine, University of Eastern Finland and Kuopio University Hospital, Kuopio, Finland.

   Obesity is a worldwide health problem with an alarming prevalence of 35% in the U.S. Although obesity is highly heritable (up to 70%), the increase in its prevalence during the last 2 decades points to gene and environment (GxE) interactions in its etiology. However, small GxE effects are difficult to detect in human, especially given the uncontrollable environment in human studies. We hypothesized that in obesity, the cellular environment affects activation of genomic regulatory mechanisms. To this end, we investigated whether an intermediate mRNA expression phenotype differs between obese and lean individuals, utilizing RNA-sequencing of subcutaneous adipose, a highly obesity-relevant tissue. Our design identifies GxE interactions under either an obese or a lean physiological state by investigating expression quantitative trait loci (eQTL) within subgroups ascertained initially based on their body mass index (BMI). We will also investigate how the metabolic disturbances related to obesity differentiate these two groups in future studies. We performed eQTL mapping with 650,000 genotyped SNPs (MAF>1%) and 15,000 expressed genes in ~600 individuals from the Finnish METSIM cohort subdivided into two groups of ~300 based on the BMI median. All eQTLs that were only observed in the obese group but not in the lean or overall groups were considered obese-specific (OS), and vice versa for lean-specific (LS). After correcting for multiple testing using an FDR<0.05, we discovered 6,689 cis (+/-1Mb) and 1,414 trans OS eQTLs; and 3,992 cis and 2,068 trans LS eQTLs, respectively. The frequencies of the context-specific OS and LS cis eQTL SNPs did not differ between these groups (P>0.1), suggesting that expression profiles change due to different genomic regulatory mechanisms in OS versus LS cis eQTLs. This was supported by our preliminary data demonstrating that the OS cis eQTLs were enriched for both obesity GWAS loci (P=0.003 Fishers exact (FE)), and all active enhancers (P=0.01 FE), whereas LS cis eQTLs were enriched for non-synonymous (ns) variants (P=2.5x10-16 Binomial). For instance, a common ns variant (MAF 9%) in the gene, DAK, regulates the expression of FADS2 in the lean Finns exclusively. FADS2 encodes the fatty acid desaturase 2 enzyme, a key regulator of unsaturated fatty acids. Our results suggest that the distinct cellular environment drives specific genomic regulatory mechanisms depending on obese and lean physiological conditions in adipose tissue.

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