Exploring natural inhibitors of Mycobacterium tuberculosis glutamine synthetase using molecular modeling, scaffold hopping, and machine learning-based bioactivity prediction

Tuberculosis (TB), caused by Mycobacterium tuberculosis (Mtb), remains a significant global health burden, mainly due to the emergence of drug-resistant strains. Glutamine synthetase is an essential enzyme in nitrogen metabolism and cell wall biosynthesis, hence a key target for therapeutic intervention against Mtb infection. This study applied computational drug discovery to identify potential Glutamine synthetase inhibitors from plant-derived natural organic compounds. A total of 1,250 compounds had been virtually screened, yielding 1,125 with binding energies that ranged between -10.2 and to -2.5 kcal/mol. Re-docking and molecular dynamics simulations confirmed their stable binding, with Amentoflavone showing the most robust interaction profile. PCA and FEL analyses revealed that Amentoflavone and Lithospermic acid exhibited the most stable conformational state. Scaffold hopping was then applied to generate two novel analogs retaining favorable features of the lead scaffold. To quantitatively assess bioactivity, molecular descriptors were used to train machine learning regression models. Gaussian Process and Extra Trees Regressors achieved the highest accuracy (R² ∼1.0) and lowest RMSE among various algorithms. These models were then used to predict pIC₅₀ values for the lead and designed compounds, with Lithospermic acid scoring the highest (7.49), followed by Tanshinone IIA (7.22), Derivative_1 (7.18), and Amentoflavone (6.92), all of which outperformed the control compound (6.79), indicating superior predicted bioactivity. These findings support the predictive capacity of the computational model and highlight high-priority candidates for future experimental validation against drug-resistant TB.