Longitudinal Study Of Whole Blood Transcriptomes In a Twin Cohort. J. Bryois1,2,3, A. Buil1,2,3, P. Ferreira1,2,3, N. Panoussis1,2,3, A. Planchon1,2,3, D. Bielser1,2,3, A. Vi˝uela4, K. Small4, T. Spector4, E. T. Dermitzakis1,2,3 1) Department of Genetic Medicine and Development, University of Geneva Medical School, Geneva, Switzerland; 2) Institute of Genetics and Genomics in Geneva (iGE3), Geneva, Switzerland; 3) Swiss Institute of Bioinformatics (SIB), Geneva, Switzerland; 4) Department of Twin Research and Genetics Epidemiology, Kings College, London, United Kingdom.
The majority of genes in human were recently found to be regulated by expression quantitative trait loci (eQTLs). Altough eQTLs studies vastly improved our understanding of the genetics of gene expression, they only provide a snapshot of the genetics of gene expression. In order to investigate the temporal aspect of the genetics of gene expression, we used RNA-seq on whole blood of females from the TwinsUK adult registry (21 DZ twin pairs, 19 MZ twin pairs, 25 Singleton) at two timepoints separated on average by 1.8 years. The twin structure of the data allowed to estimate the heritability of gene expression at the first and second timepoints (mean=0.3) and to discover that the difference in gene expression is also heritable (mean=0.2). Using ~60 unrelated invididuals, we discovered 999 genes with a cis-eQTL (5% FDR) at the first timepoint and 1018 genes with a cis-eQTL at the second timepoint. The cis-eQTLs detected at both timepoint were largely shared (1=88%) indicating that the genetic signal on gene expression is mostly stable over time. We found that 1556 genes are differentially expressed (DE) between the two timepoints and that the heritability of their difference in expression is significantly greater than for genes not DE. In addition, DE genes are on average more heritable than stable genes at the first timepoint but not at the second time point, indicating that a loss of the genetic control of gene expression is a possible cause of the differential expression. Furthermore, we found that DE genes are strongly enriched (FDR < 0.01%) in age related GO terms (ribosome, oxydative phosphorylation, parkinsons disease, hungtingtons disease, alzheimers disease and spliceosome). We observed that gene expression is on average weakly correlated (mean=0.2) between the two timepoints. However, genes with an eQTL had a significantly higher correlation between timepoints than genes without an eQTL. In addition, DE genes were significantly less correlated between timepoints than stable genes. We observed that the correlation of gene expression between timepoints is significantly correlated to the heritability of gene expression indicating that part of the correlation of gene expression between timepoints is due to genetics. Finally, using conservative approaches we discovered 39 genes (5% FDR) with a genetic effect on gene expression change between the two timepoints, providing examples of genes where the change in gene expression is genetically driven.
You may contact the first author (during and after the meeting) at