(62v) Predicting Activation Enthalpies of Cytochrome-P450-Mediated Hydrogen Abstractions: Comparison of Semiempirical PM3, SAM1, and AM1 with a Density Functional Theory Method

The progression of computational chemistry in conjunction with recent advancements in computer technology and power has made accessible new territory in the study of enzyme kinetics. Here, we focus on predicting activation enthalpies of the hydrogen abstraction step of the cytochrome P450 enzyme-mediated hydroxylation reaction ? a key process in the biotransformation of xenobiotics. A rapid methodology was previously developed by Korzekwa et al. (J. Am. Chem. Soc., 112, 7042-7046, 1990) to predict the activation enthalpy of the hydrogen abstraction reaction, using the p-nitrosophenoxy (PNPO) radical as a simple surrogate for the CYP active oxygen species. The activation enthalpy is approximated with a linear regression model using reaction enthalpy and ionization potential (of the substrate radical) as predictor variables, calculated by semi-empirical (SE) method AM1. We have applied this methodology to a set of 24 substrates, using PM3 and SAM1 calculation methods in addition to AM1. Our regression models showed strong predictive capabilities, with R2 values of 0.89, 0.90, and 0.93 for AM1, PM3, and SAM1 methods, respectively. In an effort to evaluate the applicability of the SE results, our calculated activation enthalpies were compared with those calculated by a hybrid density functional theory (DFT) method, B3LYP, using a more realistic iron-oxo-porphine model (FeO), and the results revealed limitations of the PNPO radical model. Thus, predictive models developed using SE predictors provide rapid and generally internally-consistent results, but should be interpreted and used cautiously.