Molecular Gatekeeper Discovery: Workflow for Linking Multiple Environmental Biomarkers to Metabolomics

The exposome reflects the many exposures to various factors across the life-course that can affect health. Sensitive techniques like metabolomics can reveal the underlying molecular basis linking exposures to disease and generate hypotheses for future quantitative toxicological studies. Current applications of metabolomics are primarily to identify metabolic changes linking a single exposure and a health outcome(s); there is no general framework for multiple exposures. Here, we explore the concept of ‘molecular gatekeepers’—key metabolites that link single or multiple exposure biomarkers with correlated clusters of endogenous metabolites—to inform health-relevant biological targets. We performed untargeted metabolomics on plasma from 152 adolescent girls participating in the Growing Up Healthy Study in New York City, using liquid chromatography-high resolution mass spectrometry (LC-HRMS). We then performed network analysis to link metabolites to environmental biomarkers including five trace elements (Cd, Mn, Pb, Se, and Hg) and five perfluorinated chemicals (PFCs; n-PFOS, Sm-PFOS, n-PFOA, PFHxS, PFNA) previously measured in the same samples. We defined any metabolite associated with at least one environmental biomarker and correlated with at least one other metabolite (Spearman rho > 0.9) as a ‘molecular gatekeeper’. Associations of gatekeepers with health outcomes (e.g., body mass index, age at menarche) were tested with linear models. After removing redundant peaks, 964 (positive mode) and 1784 (negative mode) metabolite features were used for network analysis. Of 95 and 138 metabolites, respectively, associated with at least one exposure, 28 and 43 were molecular gatekeepers. Further, lysophosphatidylcholine(16:0) and taurodeoxycholate were correlated with both n-PFOA and n-PFOS, suggesting a shared dysregulation from multiple xenobiotic exposures. One annotated gatekeeper, sphingomyelin(d18:2/14:0), was significantly associated with age at menarche; yet, no direct association was detected between any exposure biomarkers and age at menarche. Thus, molecular gatekeepers may provide a general data analysis framework to discover molecular linkages between exposure biomarkers and health outcomes that may otherwise be obscured by complex interactions in direct measurements. This framework may aid in identifying vulnerable biological pathways for future exposome research.