Predictive models for nucleoside bisubstrate analogs as inhibitors of siderophore biosynthesis in Mycobacterium tuberculosis: pharmacophore mapping and chemometric QSAR study

Abstract

Inhibitors of aryl acid adenylating enzymes (AAAE), known as MbtA, involved in siderophore biosynthesis in Mycobacterium tuberculosis, are being explored as potential antitubercular agents. In this article, we report the development of a robust pharmacophore model and investigation of structure-activity relationship of several nucleoside bisubstrate analogs reported as MbtA inhibitors. The developed pharmacophore model revealed the importance of two hydrogen bond donors and one hydrogen bond acceptor features. Furthermore, it was found that an aromatic ring at the distal part of molecule away from the two aromatic rings of adenyl moiety is a critical requirement for the tight binding of inhibitor. The generated pharmacophore-based alignment was used to derive a predictive atom-based 3D-QSAR model for training set (r (2) = 0.97, SD = 0.23, F = 310.6, N = 48) and test set (Q (2) = 0.71, RMSE = 0.65, Pearson-R = 0.85, N = 15). Structure-activity relationship investigation further revealed that bulky substitutions at the C-6 position of adenyl moiety is detrimental to activity, while hydrophobic substitutions can be tolerated at C-2 position. Taken together, the PLS-generated QSAR regression cubes along with developed pharmacophore model provide a qualitative picture of the active site and can be used as a powerful tool for the rational modification of bisubstrate inhibitors of MbtA in search of better antitubercular agents. Furthermore, a three-class classification chemometric QSAR model was developed using molecular descriptors for the prediction of whole-cell activity which could be used in the predictive layer for screening of compounds before synthesis.