Genetic analysis of transcriptional profiles for the identification of genes influencing obesity
Charlesworth, JC and Curran, JE and Johnson, MP and Goring, HHH and Dyer, TD and Comuzzie, AG and Cole, SA and Mahaney, MC and Jowett, JBM and MacCluer, JW and Collier, GR and Moses, EK and Blangero, J, Genetic analysis of transcriptional profiles for the identification of genes influencing obesity, Abstract/Session Information, American Society of Human Genetics, 56th Annual Meeting, October 9-13, 2006, New Orleans, Louisiana EJ (2006) [Conference Extract]
The identification of candidate genes for human quantitative traits is typically based upon subjective knowledge of a biological pathway that is extrapolated to a particular phenotype. In this study, we propose an objective approach to candidate gene discovery that utilizes large-scale transcriptional profiling to identify novel cis-acting genes that correlate with a given quantitative trait. Using RNA extracted from lymphocytes, we obtained genome-wide quantitative transcriptional profiles from 1,240 individuals in the San Antonio Family Heart Study. In this data set, we were able to significantly detect ~20,000 transcripts. Using quantitative trait linkage analysis, we identified over 3,000 autosomal cis-acting QTLs for which we have significant evidence for variation at the transcript’s genomic location that influences expression levels. To identify potential novel candidate genes involved in obesity, we examined correlations between expression levels of these cis-acting genes and two obesity-related indicators, the body mass index and fat mass (as measured by bioimpedance). Using high-dimensional endophenotypic search procedures, we identified 383 autosomal genes (many novel) that correlate with these obesity-related phenotypes. Several known candidate genes for obesity, including IGBP3 and CDF (which encodes adipsin) were confirmed. The novel genes identified vary in their general biological actions, from mitochondrial functions to inflammation and growth factors. Because these genes were chosen to have large cis-acting effects on transcription levels, it is likely that some of our findings reflect causal relationships with risk of obesity. Our results point to the utility of large scale family-based transcriptional data bases for identifying human quantitative trait loci.