oa The epigenetic response to disease and environmental challenges: a deep multiomics phenotyping view

Epigenetic regulation of human cellular functions allows temporal adaptation to changes in health, lifestyle and environment. Complex diseases and exposure to environmental challenges may lead to adjustments of the expression of certain enzymes, transporters, and metabolic regulator genes (1). The response to such challenges by adjustment of gene expression may be reflected through changes in the DNA methylome (2). A recent metabolomics epigenome wide association study showed that the methylation states of certain DNA cytosine-guanine (CpG) pairs are strongly associated with blood metabolite levels (3). Interestingly these metabolomics associated CpG sites are also associated with a related complex trait falling in one of two major groups, the first being disease driven i.e. diabetes and/or obesity that also associate with liver function or blood pressure, and the second being environmentally driven i.e. smoking. We performed deep molecular phenotyping covering approximately 4000 traits in blood, urine and salivary samples from a diverse population of 359 individuals from the Qatar Metabolomics Study of Diabetes (QMDiab), using array-based DNA genotyping and methylomics, NMR based lipidomics, mass-spectrometry based metabolomics, aptamer based blood-circulating proteomics, and total plasma protein N- and IgG-glycomics. We confirmed the established smoking and diabetes associations in QMDiab and identified novel multi-omics associations at these sites. In particular, we identified deep molecular phenotypes that are characteristic for the TXNIP-diabetes and the AHRR-smoking associations, including diabetes and smoking associated metabolites such as 1,5-anhydroglucitol (1,5-AG) and o-cresol sulfate respectively, diabetes and smoking associated proteins such as Sex hormone-binding globulin (SHBG) and Polymeric immunoglobulin receptor (PIGR) respectively, and diabetes and smoking associated glycans. We also replicated some of the novel associations, particularly the proteomics and glycomics associations, in two independent studies (KORA and TwinsUK). These observations show an interesting overlap between DNA methylation and pathways relevant to complex disorders and environmental challenges. We also addressed the direction of that involvement in this study. Motivated by a recent obesity Mendelian randomization study (4) that showed a causal effect of adiposity on the methylation levels of multiple CpG sites near obesity-related genes, we used Mendelian randomization to test the direction of association between the metabolite and methylation levels of selected obesity associated CpG sites in KORA. In line with that study, we also found a causal effect of metabolite levels on methylation of the given CpG sites, i.e. glycerophospholipid PC(O-36:5), glycine, and a very low density lipoprotein A (VLDL-A) on the methylation of the DHCR24, MYO5C, and CPT1A loci, respectively. The overlap of disease- and lifestyle-associated CpG sites with metabolite-associated CpG sites suggests that the organismal adjustment to disease or environmental challenges can be captured by measuring relevant intermediate molecular phenotypes (multi-omics) (5). Taken together, this study supports the hypothesis that multi-omics-associated CpG methylation can serve as a functional readout of the organism's response to the challenges induced by disease or environmental stress, serve as novel functional diagnostic and prognostic biomarkers, and indicate potential targets for therapeutic intervention.