Pharmacophore modeling of pretomanid (PA-824) derivatives for antitubercular potency against replicating and non-replicating Mycobacterium tuberculosis

Pretomanid (PA-824) is the recently (2019) approved drug for the treatment of extensively drug-resistant (XDR) TB and the multidrug-resistant (MDR) TB by US FDA. The experimental data of antitubercular activity of 543 pretomanid derivatives (total 6 datasets) against replicating (active) and non-replicating (dormant) forms of Mycobacterium tuberculosis (strain H37Rv) are available in the literature. Such vast experimental data of pretomanid derivatives against both of these endpoints, and recent approval of pretomanid molecule as a drug encouraged us to utilize this existing experimental information for the development of the 3D-pharmacophore models. The developed model (Hypo-1, MABA) showed the three physicochemical features namely, the oxygen atom of nitro group (HBA_1), fused pyran ring of imidazopyran heterocycle (HYAl_2) and the 4-fluorophenyl moiety (HYAr_3) are crucial for the antitubercular activity against replicating M. tb. Subsequently, the pharmacophore model (Hypo-1, LORA) developed against the non-replicating form of M. tb also showed the contribution of three physicochemical features namely, the 4-tri-fluoromethyl group (HYAl_2) and both the phenyl groups (HYAr_3, HYAr_4) of biaryl moiety in increasing the antitubercular activity. Both the pharmacophoric classifier models showed the classification accuracies of 82.98 and 74.42% for the training set compounds, and 63.91 and 61.60% for the test set compounds respectively, for labelling the compounds into higher and lower active classes. Both the models were also found to be retaining the higher active compounds in top 1.00% of the total number of compounds (decoys and actives), after performing the decoy set screening. Communicated by Ramaswamy H. Sarma.