Computational studies of the cytochromes P450

Validation of standard comparative modelling techniques was carried out by comparison between several models of the mammalian P450 2C5 enzyme and experimentally determined X-ray crystal structure. Structural models of were then proposed for human cytochrome P450 2D6, a polymorphic enzyme known to be involved in the metabolism of a wide variety of pharmaceuticals. These models incorporated the only mammalian P450 crystal structure (pdb accession code 1dt6) and the impact of including this structure as a template in the modelling process was assessed using computational techniques such as active site characterisation using a number of chemical probes and principal components analysis to systematically analyse for differences between the comparative models. This analysis highlighted a single residue, Glu 216, as a residue that may be important for substrate binding, a hypothesis that was supported via docking studies of known substrates i.e. codeine and MPTP. Prediction of binding affinities was also investigated. A number of known P450 2D6 inhibitors were docked into the active site of the model and their binding affinities investigated using regression based scoring methods. The activities of 2D6 substrates were also investigated using a number of Quantitative Structure-Activity Relationship (QSAR) methods, the most successful being the grid-based Molecular Field Analysis (MFA) technique. Finally pharmacophores were used to predict the binding affinities of a number of P450 2D6 substrates. The results for these investigations compared favourably to literature precedents and served to reiterate the importance of a hydrophobic region approximately 5A in distance from a positive ionisable group, such as a basic nitrogen atom, as an indicator that a substrate is likely to be metabolised by P450 2D6.