Identification of genetic epistasis in regulation of gene expression via variance expression quantitative trait loci. A. Brown1, A. Buil2,3,4, M. N. Davies5, A. Vi˝uela5, T. Lappalainen2,3,4, H. F. Zheng6, J. B. Richards5,6, K. S. Small5, T. D. Spector5, E. T. Dermitzakis2,3,4, R. Durbin1 1) Wellcome Trust Sanger Institute, Cambridge, United Kingdom; 2) Dep. Genetic Medicine and Development, University of Geneva, Geneva, Switzerland; 3) Institute of Genetics and Genomics in Geneva, University of Geneva, Geneva, Switzerland; 4) Swiss Institute of Bioinformatics, Switzerland; 5) Department of Twin Research and Genetic Epidemiology, King's College London, London, United Kingdom; 6) Departments of Medicine, Human Genetics, Epidemiology and Biostatistics, McGill University, Montreal, Canada.
Epistasis, in which the effect of a variant on a phenotype is modulated by one or more other variants, is frequently observed in model organisms but has proved difficult to isolate from human GWAS. One issue is the vast search space of all pairwise combinations of SNPs: variance QTL have been suggested as an alternative to an exhaustive search. Variance QTL are SNPs that effect the variance of a trait rather than the mean, and are plausible candidates to be involved in gene by environment interactions (GxE) and epistasis. We use variance QTL affecting gene expression (ve-QTL) to discover epistasis using RNA-seq from lymphoblastoid cell lines (LCLs), subcutaneous adipose tissue, skin and whole blood from ~800 female twins from the TwinsUK cohort. Using this v-eQTL approach, we reduce the search space from a quadratic scan to two linear scans. This space is further reduced by concentrating on the cis-window (1MB around transcription start site). In addition, our twin design allows exploration of variance explained by cis-trans epistasis, as well as an analysis of discordance within monozygotic (MZ) twins pairs to infer presence of GxE. The analysis of 765 LCL samples found variance in expression of 501 genes to be affected by a SNP in the cis window (a v-eQTL). Looking at discordance within MZ twin pairs we found evidence that GxE could contribute to 63% of the v-eQTL. A further scan for SNPs in epistasis with these v-eQTL discovered epistatic interactions affecting expression of 170 genes. We replicated 35 of the epistatic interactions, 30 with the same direction of effect, in a RNA-seq dataset of 465 samples from 1000 Genomes. Moreover, leveraging available sequence data from the 1000 genomes project, we were able to test for possible confounding by eQTL on rare haplotypes. We found that in 10 cases, a secondary eQTL would not explain the interaction. One example of epistasis affecting expression was the gene TRIT1, for which we found that 8.5% of variance in expression was explained by an interaction between two loci on the boundaries of two distinct enhancer regions. We have successfully exploited v-eQTL to discover replicated epistasis affecting gene expression. Our study design also suggests some of the variants were involved in GxE. We are currently exploring other tissues (adipose, skin and whole blood), focusing in these cases on GxE as environmental factors are known to play a larger role here than in LCLs.
You may contact the first author (during and after the meeting) at