RESUMO
Zn-metalloproteins are a major class of targets for drug design. They constitute a demanding testing ground for polarizable molecular mechanics/dynamics aimed at extending the realm of quantum chemistry (QC) to very long-duration molecular dynamics (MD). The reliability of such procedures needs to be demonstrated upon comparing the relative stabilities of competing candidate complexes of inhibitors with the recognition site stabilized in the course of MD. This could be necessary when no information is available regarding the experimental structure of the inhibitor-protein complex. Thus, this study bears on the phosphomannose isomerase (PMI) enzyme, considered as a potential therapeutic target for the treatment of several bacterial and parasitic diseases. We consider its complexes with 5-phospho-d-arabinonohydroxamate and three analog ligands differing by the number and location of their hydroxyl groups. We evaluate the energy accuracy expectable from a polarizable molecular mechanics procedure, SIBFA. This is done by comparisons with ab initio quantum-chemistry (QC) calculations in the following cases: (a) the complexes of the four ligands in three distinct structures extracted from the entire PMI-ligand energy-minimized structures, and totaling up to 264 atoms; (b) the solvation energies of several energy-minimized complexes of each ligand with a shell of 64 water molecules; (c) the conformational energy differences of each ligand in different conformations characterized in the course of energy-minimizations; and (d) the continuum solvation energies of the ligands in different conformations. The agreements with the QC results appear convincing. On these bases, we discuss the prospects of applying the procedure to ligand-macromolecule recognition problems. © 2016 Wiley Periodicals, Inc.