REVENG: 3D-DRIVEN REVERSE ENGINEERING OF METABOLIC BIOLOGICAL PATHWAYS

Elucidating the specific enzymes responsible for metabolite formation is a cornerstone of drug discovery; it is essential for mapping metabolic pathways, assessing drug-drug interactions, driving lead optimization to enhance metabolic properties, and mitigating toxicity risk. While liquid chromatography-mass spectrometry (LC-MS/MS) excels at structural Metabolite Identification (MetID), experimental reaction phenotyping remains a major resource-intensive bottleneck. To bridge this gap, we present RevEng, an innovative computational workflow integrated within MetaSite 7¹. Taking the parent drug and a target metabolite (for example, an experimentally detected LC-MS/MS metabolite) as inputs, RevEng structurally reverse-engineers the metabolic space to reconstruct all potential metabolic pathways, including complex multi-step biotransformations, and assigns a probability score to the enzymes involved in each step. This inverse pathway prediction relies entirely on structure-based 3D docking and intrinsic chemical reactivity, using dedicated enzyme-specific predictive models for 23 Phase I (CYP and non-CYP) and 20 Phase II major human drug-metabolizing enzymes^2,3. These models are interaction- and reaction-based and are therefore fully independent of training-set-derived substrate knowledge. Enzyme active sites are characterized using GRID Molecular Interaction Fields (MIFs)⁴, while the docking procedure is specifically designed to mimic the biological and chemical determinants of each enzymatic reaction by evaluating substrate–residue interactions within the catalytic site. The resulting interaction energy, combined with an assessment of the molecule’s intrinsic reactivity toward a specific biotransformation, enables prediction of the most likely site of metabolism, related metabolites structures and the likelihood of metabolism by the enzyme under consideration. The workflow was rigorously validated against proprietary in-house experimental datasets, and the predicted inverse pathways were benchmarked against established literature data for diverse chemical entities. The 3D-docking engine successfully unraveled complex metabolic fates, accurately backtracking both single-step biotransformations and sequential, multi-step cascades (such as Phase I functionalization followed by Phase II conjugation). Beyond pathway elucidation, the prediction of the enzymes responsible for the formation of observed metabolites can directly drive experimental design, enabling the detection of metabolites using targeted enzymatic assays rather than complex biological matrices, thus leading to a better characterization of the metabolic profile of a drug. Furthermore, these predictions provide invaluable guidance for targeted metabolite synthesis and for anticipating potential toxicity liabilities. In conclusion, RevEng, in combination with the other advanced tools integrated within MetaSite 7, successfully bridges analytical chemistry and mechanism-based 3D docking, offering a chemically accurate and structurally-driven solution for automated reaction phenotyping and intelligent experimental design in early drug discovery.

1. Isoherranen, N., & Zanger, U. M. (2025). Cytochrome P450 enzymes in drug metabolism and precision medicine: Current challenges and future directions. ACS Pharmacology & Translational Science, 8(2), 114-131.
2. Cruciani, G., Carosati, E., De Boeck, B., Ethirajulu, K., Mackie, C., Howe, T., & Vianello, R. (2005). MetaSite: Understanding Metabolism in Human Cytochromes from the Perspective of the Chemist. Journal of Medicinal Chemistry, 48(22), 6970-6979.
3. Di, L., & Obach, R. S. (2024). The evolving landscape of non-CYP enzymes in drug metabolism: Current status and clinical relevance. Drug Metabolism Reviews, 56(2), 145-168.
4. Artese, A., Cross, S., Costa, G., Distinto, S., Parrotta, L., Alcaro, S., Ortuso, F., & Cruciani, G. (2013). Molecular interaction fields in drug discovery: recent advances and future perspectives. WIREs Computational Molecular Science, 3(6), 594-613.