Abstract
EnzyDock is a multistate, multiscale CHARMM-based docking program which enables mechanistic docking, i.e., modeling enzyme reactions by docking multiple reaction states, like substrates, intermediates, transition states, and products to the enzyme, in addition to standard protein-ligand docking. To achieve docking of multiple reaction states with similar poses (i.e., consensus docking), EnzyDock employs consensus pose restraints of the docked ligand states relative to a docking template. In the current work, we present an implementation of a Maximum Common Substructure (MCS)-guided docking strategy using EnzyDock, enabling the automatic detection of similarity among query ligands. Specifically, the MCS multistate approach is employed to efficiently dock ligands along enzyme reaction coordinates, including reactants, intermediates, and products, which allows efficient and robust mechanistic docking. To demonstrate the effectiveness of the MCS strategy in modeling enzymes, it is first applied to two highly complex enzyme reaction cascades catalyzed by the diterpene synthase CotB2 and the Diels-Alderase LepI. In addition, the MCS strategy is applied to dock enzyme inhibitors using cocrystallized inhibitors or substrates to guide the docking in the enzymes dihydrofolate reductase and the SARS-CoV-2 enzyme M(pro). The latter case exemplifies the use of MCS with EnzyDock's covalent docking capabilities and QM/MM scoring option. We show that different protocols of the implemented MCS algorithm are needed to obtain mechanistic consistency (i.e., similar poses) in mechanistic docking or to accurately dock chemically diverse ligands in inhibitor docking. Although the current implementation is specific for EnzyDock, the findings should be general and transferable to additional docking programs.