Expanding Computational Resources for the Discovery of Molecular Biomarkers of Exposure

Abstract

Recurrent exposure to genotoxic chemicals and other harmful environmental agents is directly related to the development of adverse phenotypes, carcinogenesis, and developmental disorders through chemical-mediated toxicity. Systems toxicology aims to chemically profile and computationally model the toxicity pathways to gain mechanistic insights into the outcomes of environmental exposure and develop modern safety guidelines. Analytical methods are being developed to perform broad, unbiased characterization of molecular biomarkers of exposure to improve the understanding of the molecular consequences and physiological responses following the introduction of reactive electrophilic chemicals. There are few computational workflows available to support this rapid development of new analytical technology. This thesis introduces novel computational resources for the parameterized analysis of discovery-driven, systems toxicology results generated by liquid chromatography-coupled mass spectrometry. Chapter 1 of this thesis presents a comprehensive review of existing analytical technology for the molecular profiling of exposure biomarkers. This review focuses on discovering toxicologically relevant compounds using constant ion monitoring, fragmentation filtering, in-source collision-induced dissociation, mass defect filtering, or isotope pattern filtering. The mechanism of discovery of each analytical method is examined and the strengths and limitations of each approach are discussed. Chapter 2 of this thesis presents a novel computational workflow for discovering molecular biomarkers of exposure using fragmentation filtering. In this study, an original module named DFBuilder for the metabolomics data processing software MZmine is presented. The application of this software tool for the discovery of covalently modified DNA nucleosides is evaluated. In this work, a novel colibactin-derived, E. coli associated DNA adduct product discovery is highlighted. Chapter 3 of this thesis extends the application of the DFBuilder module to discover urinary metabolites produced from detoxifying reactive electrophilic chemicals in tobacco cigarette smoke. This work presents the first reported application of a high-resolution mass spectrometry method for profiling mercapturic acid conjugates in positive ion mode. The combination of this novel analytical method and computational workflow discovers numerous prospective mercapturic acid signatures never reported in human urine. Statistical evaluation of these results demonstrates that many of these products are associated with cigarette usage. Chapter 4 of this thesis reports a novel mass spectral library of conjugated mercapturic acids. Using multiple fragmentation strategies, this library represents thousands of mass spectra collected in positive and negative ion polarity. This work serves as a foundation of resources necessary for verifying discovery results produced from the analytical methods presented in this thesis. Metadata insights from this library that will help future efforts to characterize mercapturic acid conjugates are discussed. This thesis concludes with a summary and future perspective evaluating the remaining computational challenges in exposomics and analytical chemistry-based approaches for systems toxicology. The areas that most need support are highlighted, and the capacity of emerging computational solutions to improve experimental outcomes is discussed.