Application of the dynamic quantitative structure-activity relationship method for modeling antibacterial activity of quinolone derivatives.

Abstract

The dynamic approach to quantitative structure-activity relationship (QSAR) was recently introduced to mimic the multiplicity of 3D-molecular shapes taken from the chemical at the different stages of the processes conditioning the endpoint under investigation. In difference with the conventional QSAR methods, where the structure of each compound is described by a single conformation (usually the one with the lowest calculated energy), the dynamic QSPR is aiming to account for the effects of the different solvent environments at the various reaction steps under which different conformations should be active. The core of the new methodology is the 3DGEN algorithm for an exhaustive 3D molecular design and the related system for an interactive conformation screening, based on the: chemical expertise, stereoelectronic parameter ranges and parameter distributions, depending on hypothesis on interaction mechanism The new methodology is incorporated in the OASIS (optimized approach based on structural indices set) computer system for QSAR/QSPR (quantitative structure activity/property relationship). In the present work it was applied to model in vitro (inhibition of Escherichia coli DNA gyrase) and in vivo (MICs against gram-negative as well as gram positive bacteria) antimicrobial activity (AMA) of quinolone derivatives. It was found that AMA is conditioned by molecular geometry as described by pair of topological indices and electron-acceptor properties, as assessed by the energies of LUMO (Lowest Unoccupied Molecular Orbital) orbitals, charges, bond orders and polarizability of the specific molecular sites. Interaction hypothesis is created, according to which polar-polar intermolecular interactions and bond breaking (cycle "opening", analogous to that of beta-lactam moiety in cephalosporins) condition biological activity. The derived QSAR models are significant according to the conventional statistical criteria as well as to the structure-activity causality requirements stated in literature. The best QSARs are obtained for in vitro AMA (r2 = 0.93 and s2 = 0.003), whereas for in-vivo activity correlations found are with lower statistics (0.54 < r2 < 0.74 and 0.005 < s2 < 0.03). The results are statistically better than those obtained by Computer automated Structure Evaluation (CASE) method.