Structure-activity Relationship Studies on Inhibitors of Dihydroorotate Dehydrogenase Based on Pharmacophore Model

The pharmacophore model of dihydroorotate dehydrogenase inhibitors was established guide the investigation on virtual screening for new dihydroorotate dehydrogenase inhibitors. Based on the training set composed of 31 DHODH inhibitors, pharmacophore models were generated by HypoGen program of the Catalyst software. The IC50 of the inhibitors varied from 7 to 63000 nmol/L. The pharmacophore models were a set of 3D pharmacophore features, which were constructed by the generation of conformational models and alignments of conformations based on the training set. The best model was validated to be highly predictive by two methods, namely, test set prediction and CatScramble method. This model consisted of two hydrogen-bond acceptors, and two hydrophobic regions. These features in the three-dimensional arrangement in the pharmacophore model could be characterized as hydrogen bond interaction, hydrophobic interaction and pi-pi interaction between ligand and acceptor. These features play an important role in determining the activities of bioactive molecules. The model's correlation coefficient between the estimated and true activities for compounds constituting the training set were 0.8405, and for compounds from microbial metabolites were 0.8788. Using the pharmacophore model of DHODH inhibitors, 59 compounds with good estimated activities were found from microbial metabolites. New DHODH inhibitors may be found from these candidate compounds.