Multi-method genome- and epigenome-wide studies of inflammatory protein levels in healthy older adults

Abstract

Background: The molecular factors which control circulating levels of inflammatory proteins are not well understood. Furthermore, association studies between molecular probes and human traits are often performed by linear modelbased methods which may fail to account for complex structure and interrelationships within molecular datasets. Methods: In this study, we perform genome-and epigenome-wide association studies (GWAS/EWAS) on the levels of 70 plasma-derived inflammatory protein biomarkers in healthy older adults (Lothian Birth Cohort 1936; n = 876; Olink® inflammation panel). We employ a Bayesian framework (BayesR+) which can account for issues pertaining to data structure and unknown confounding variables (with sensitivity analyses using ordinary least squares-(OLS) and mixed model-based approaches). Results: We identified 13 SNPs associated with 13 proteins (n = 1 SNP each) concordant across OLS and Bayesian methods. We identified 3 CpG sites spread across 3 proteins (n = 1 CpG each) that were concordant across OLS, mixedmodel and Bayesian analyses. Tagged genetic variants accounted for up to 45% of variance in protein levels (for MCP2, 36% of variance alone attributable to 1 polymorphism). Methylation data accounted for up to 46% of variation in protein levels (for CXCL10). Up to 66% of variation in protein levels (for VEGFA) was explained using genetic and epigenetic data combined. We demonstrated putative causal relationships between CD6 and IL18R1 with inflammatory bowel disease and between IL12B and Crohn's disease. Conclusions: Our data may aid understanding of the molecular regulation of the circulating inflammatory proteome as well as causal relationships between inflammatory mediators and disease.

Epigenetic scores for the circulating proteome as tools for disease prediction

“…Though testing in a holdout set and two external cohorts suggested that many of our 27 optimal proxies were robustly capturing protein expression, there were many proteins for which we did not achieve reasonable proxies. As the training sample increases, we expect convergence between proxy projections and measured proteins; however, our previous work indicates that there is a threshold for variance explained in protein expression by genome-wide DNAm 2,3 . Nevertheless, even where DNAm proxies CRP and IL6 correlate ~0.2 with measured protein levels, they provide a more stable measure of expression than proteomic measures when averaged longitudinally; they often outperform the measured proteins in relation to associations with health outcomes and lifestyle factors [13][14][15] .…”

“…To determine if pQTL mapping was possible with the proxy measures, GWAS analyses were run for the proxies corresponding to 7 proteins with GWAS significant SNPs in previous Lothian Birth Cohort 1936 analyses 2,3 (N=9 genome-wide significant SNPs; Supplementary Note 1). We replicated 7/9 sites from previous studies at P < 5 x 10 -8 , six of which were cis associations for the protein coding gene, and one of which was trans (rs46876657; Supplementary Table 13) 2,3 . Moreover, all 7 SNPs were within 75kb of a CpG that was included in the corresponding protein proxy, six of which were previously reported as methylation quantitative trait loci (mQTLs) for protein proxy CpGs [PMID 27036880] 21 .…”

“…Prior to running the elastic net models, we rank-based inverse normalised protein levels and adjusted for age, sex, cohort-specific variables and, where present, cis and trans pQTL effects identified from previous analyses 17,26,27 (Methods). Of a possible 793 proteins in KORA, 84 EpiScores had Pearson r > 0.1 and P < 0.05 when tested in an independent subset of Generation Scotland (The Stratifying Resilience and Depression Longitudinally:STRADL study, n=778) (Supplementary Table 2).…”

“…Normalised plasma protein levels were then regressed onto age, sex, four genetic components of ancestry derived from multidimensional scaling of the Illumina 610-Quadv1 genotype array (to control for population structure) and Olink® array plate. In the LBC1936 cohort, known pQTLs were adjusted for as additional covariates, through reference to previous GWAS studies in the samples 26,27 .…”

“…Plasma protein biomarkers can help to achieve this goal, as their divergent expression provides insight into the likely causes and consequences of disease in an individual, oftentimes prior to the onset of clinically relevant symptoms 1 . The recent characterization of the genome-wide genetic and epigenetic architectures of plasma proteins highlights the potential for integrated omics data to uncover causal pathways contributing to protein expression 2,3 and health outcomes [2][3][4][5][6][7] .…”