Calibration of 1,2,4-Triazole-3-Thione, an Original Zn-Binding Group of Metallo-β-Lactamase Inhibitors. Validation of a Polarizable MM/MD Potential by Quantum Chemistry

Abstract

In the context of the SIBFA polarizable molecular mechanics/dynamics (PMM/PMD) procedure, we report the calibration and a series of validation tests for the 1,2,4-triazole-3-thione (TZT) heterocycle. TZT acts as the chelating group of inhibitors of dizinc metallo-β-lactamases (MBL), an emerging class of Zn-dependent bacterial enzymes, which by cleaving the β-lactam bond of most β-lactam antibiotics are responsible for the acquired resistance of bacteria to these drugs. Such a study is indispensable prior to performing PMD simulations of complexes of TZT-based inhibitors with MBL's, on account of the anchoring role of TZT in the dizinc MBL recognition site. Calibration was done by comparisons to energy decomposition analyses (EDA) of high-level ab initio QC computations of the TZT complexes with two probes: Zn(II), representative of "soft" dications, and water, representative of dipolar molecules. We performed distance variations of the approach of each probe to each of the two TZT atoms involved in Zn ligation, the S atom and the N atom ortho to it, so that each SIBFA contribution matches its QC counterpart. Validations were obtained by performing in- and out-of-plane angular variations of Zn(II) binding in monoligated Zn(II)-TZT complexes. The most demanding part of this study was then addressed. How well does ΔE(SIBFA) and its individual contributions compare to their QC counterparts in the dizinc binding site of one MBL, L1, whose structure is known from high-resolution X-ray crystallography? Six distinct complexes were considered, namely each separate monozinc site, and the dizinc site, whether ligated or unligated by TZT. Despite the large magnitude of the interaction energies, in all six complexes ΔE(SIBFA) can match ΔE(QC) with relative errors <2% and the proper balance of individual energy contributions. The computations were extended to the dizinc site of another MBL, VIM-2, and its complexes with two other TZT analogues. ΔE(SIBFA) faithfully reproduced ΔE(QC) in terms of magnitude, ranking of the three ligands, and trends of the separate energy contributions. A preliminary extension to correlated calculations is finally presented. All these validations should enable a secure design of a diversity of TZT-containing MBL inhibitors: a structurally and energetically correct anchoring of TZT should enable all other inhibitor groups to in turn optimize their interactions with the other target MBL residues.