A simple computational tool for accurate, quantitative prediction of one–electron reduction potentials of hypoxia–activated tirapazamine analogues

Abstract

The reduction potentials of bioreductively-activated drugs represent an important design parameter to be accommodated in the course of creating lead compounds and improving the efficacy of older generation drugs. Reduction potentials are traditionally reported as single–electron reduction potentials, E(1), measured against reference electrodes under strictly defined experimental conditions. More recently, computational chemists have described redox properties in terms of a molecule’s highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO), in electron volts (eV). The relative accessibility of HOMO/LUMO data through calculation using today’s computer infrastructure and simplified algorithms make the calculated value (LUMO) attractive in comparison to the accepted but rigorous experimental determination of E(1). This paper describes the correlations of eV (LUMO) to E(1) for three series of bioreductively–activated benzotriazine di-N-oxides (BTDOs), ring-substituted BTDOs, ring-added BTDOs and a selection of aromatic nitro compounds. The current computational approach is a closed–shell calculation with a single optimization. Gas phase geometry optimization was followed by a single–point DFT (Density Functional Theory) energy calculation in the gas phase or in the presence of polar solvent. The resulting DFT–derived LUMO energies (eV) calculated for BTDO analogues in gas phase and in presence of polar solvent (water) exhibited very strong linear correlations with high computational efficiency (r2 = 0.9925) and a very high predictive ability (MAD = 7 mV and RMSD = 9 mV) when compared to reported experimentally determined single–electron reduction potentials.