Decoding Microbial Reductive Dechlorination of 209 Polychlorinated Biphenyl Congeners through Experiment-Aided Quantum Chemistry and Machine Learning: Environmental Science & Technology

Polychlorinated biphenyls (PCBs) persist globally as legacy pollutants with a complex structural diversity that complicates the understanding of their microbial conversion processes and remediation. In this study, high-throughput enzymatic assays, quantum chemical calculations, and machine learning were integrated to elucidate the reductive dechlorination pathways and reactivity of all 209 PCB congeners. By coupling Hirshfeld charge analysis with empirically derived steric effects, 98.3% accuracy was achieved in predicting dechlorination pathways across diverse Dehalococcoides isolates and enrichment cultures containing distinct organohalide-respiring bacteria. Furthermore, XGBoost models incorporating electronic, steric, and physicochemical descriptors were developed to quantify the dechlorination reactivity of PCBs, revealing that the steric effect-corrected Hirshfeld charge and PCB solubility primarily control microbial reductive dechlorination potential. The model successfully captured the observed trends in the dechlorination reactivity of PCBs across multiple dechlorinating cultures and predicted that 11 of the 12 dioxin-like PCB congeners were susceptible to microbial reductive dechlorination, highlighting intrinsic microbial detoxification potential under anaerobic conditions. This integrative framework unveils the first full picture of microbial dechlorination pathways and reactivity for the entire PCB family, providing mechanistic insight into how molecular properties dictate halogen removal. The findings advance the predictive understanding of organohalide respiration and offer a roadmap for designing microbiome-based bioremediation strategies for persistent halogenated pollutants like PCBs.