Stage-resolved non-target reactomics and toxicity screening across a full-scale WWTP

Understanding how complex organic contaminant mixtures change in composition, structure, and transformation behavior across wastewater treatment plants (WWTPs) is essential for improving pollutant control. However, stage-resolved knowledge of these processes remains limited. Here, we integrated non-targeted screening, paired mass distance (PMD) reactomics, CANOPUS-based structural annotation, and MS2Tox-based toxicity prediction to characterize compositional changes, dominant transformation behaviors, and toxicity implications throughout a full-scale anaerobic-anoxic-oxic (A/A/O) treatment and disinfection process through wastewater sampling in a WWTP. Across the treatments, chemical features progressively shifted toward lower molecular weight and higher hydrophilicity. Structural classification revealed the common pollutants (organic acids and derivatives, benzenoids, lipids and lipid-like molecules) and specific pollutants present in each unit. Reactomics revealed that methylation and dehydrogenation/oxidation were the most frequent transformation types, and each unit exhibited several distinct reaction types (e.g., stage-specific alkylation/dealkylation reactions during disinfection). Further correlating structure and reaction types reveal that pollutant transformations in wastewater are characterized by small mass shifts and generally retained the structural category of the reactant. Subsequently, toxicity prediction results suggested that most compounds exhibited no (65%) to low (31.6%) toxicity in WWTP. Biological treatment was associated with more predicted detoxification events, whereas chlorination disinfection showed more predicted toxicity-increase events that were frequently associated with alkylation-/methylation-related PMD signals. This study provides a stage-resolved interpretation of contaminant transformation across WWTP treatment units by linking feature attenuation, structural redistribution, PMD-derived reaction signatures, and MS2Tox-predicted toxicity shifts.