CYP isoform specificity toward drug metabolism: analysis using common feature hypothesis.

Three dimensional pharmacophoric maps were generated for each isoforms of CYP2C9, CYP2D6 and CYP3A4 separately using independent training sets consist of highly potent substrates (seven substrates for each isoform). HipHop module of CATALYST software was used in the generation of pharmacophore models. The best pharmacophore model was chosen out of the several models on the basis of (i) highest ranking score, (ii) better fit value among training set, (iii) capability to screen substrates from data set and (iv) efficiency to identify the isoform specificity. The individual pharmacophore models (CYP2C9-hypo1, CYP2D6-hypo1 and CYP3A4-hypo1) are characterized by the pharmacophoric features XZDH, RPZH and XYZHH for the CYP2C9, CYP2D6 and CYP3A4 respectively. Each of the chosen models was validated by using data sets of CYP substrates. This comparative study of CYP substrates demonstrates the importance of acidic character along with HBD and HBAl features for CYP2C9, basic character with ring aromatic features for CYP2D6 and hydrophobic features for CYP3A4. Acidity, basicity and hydrophobicity features arising from the functional groups of the substrates are also responsible for demonstrating CYP isoform specificity. Hence, these chemical features are incorporated in the decision tree along with pharmacophore maps. Finally, a decision tree based on chemical features and pharmacophore features was generated to identify the isoform specificity of novel query molecule toward the three isoforms.