We have estimated the activation energy for hydrogen abstraction by compound I in cytochrome P450 for a diverse set of 24 small organic substrates using state-of-the-art density functional theory (B3LYP). We then show that these results can be reproduced by computationally less demanding methods, for example, by using small organic mimics of compound I with both B3LYP and the semiempirical AM1 method (mean absolute error of 3-4 kJ/mol) or by calculating the bond dissociation energy, without relaxation of the radical (B3LYP) or estimated from three-point fit to a Morse potential (AM1; errors of 4 and 5 kJ/mol, respectively). We can assign activation energies of 74, 61, 53, 47, and 30 kJ/mol to primary carbons, secondary/tertiary carbons, carbons with adjacent sp(2) or aromatic groups, ethers/thioethers, and amines, respectively, which gives a very simple and predictive model. Finally, some of the less demanding methods are applied to study the CYP3A4 metabolism of progesterone and dextromethorphan.
Journal of Medicinal Chemistry, 2006, Vol 49, Issue 22, p. 6489-6499