Reaction-guided metabolomics accelerates high-throughput characterization of the xenobiotic metabolites for human exposome

Human exposome features a highly expansive chemical space and substantial individual variability. Although screenings of xenobiotic compounds have revealed exposure landscapes of specific compounds, significant bottlenecks remain in profiling their biotransformed products for comprehensive exposome-wide analysis, including limitations to known metabolites, challenges in new metabolite annotation and low throughput. In this study, we develop an untargeted metabolomics-based compound metabolite discovery network (CMDN) to facilitate high-throughput annotation of xenobiotic metabolites. CMDN integrates a triple-layered architecture comprising a differential expression metabolic space, a rule-based pseudo-MS1 candidature space and an MS2 spectrum similarity network. The utilities and advantages are demonstrated using pesticides as a representative example, given their widespread human exposure and well-documented toxicity. Coupled with enzymatic biotransformation assays involving 1,021 pesticides, CMDN non-redundantly annotated 2,886 bio-transformed derivatives. Following multi-channel validation including standard verification, retention time prediction, murine studies, and enzymatic kinetics, identified metabolites were screened in a human cohort, revealing a novel, diverse and extensive exposure spectrum. Collectively, this study establishes, for the first time, a scalable workflow for the annotation and screening of previously undercharacterized xenobiotic metabolites with unprecedented throughput, representing a significant advancement towards the characterization, interpretation, and prioritization of the human exposome.