Structure-based and machine learning-assisted identification of natural phytochemicals targeting Mycobacterium leprae InhA for anti-mycobacterial therapeutics.

Mycobacterium leprae (M. leprae) is the bacterium that causes leprosy. It is a public health problem in many regions, especially in developing countries. The situation is getting worse and worse as drug resistance spreads. InhA is one of the most important proteins for M. leprae's survival. It helps make mycolic acid, an important part of the bacterial cell envelope; hence, InhA is a good target for developing new anti-leprosy drugs. In this study, we focused on identifying natural plant-derived compounds (phytochemicals) capable of inhibiting InhA function using sophisticated computer-based techniques, including molecular docking, simulations, DFT calculations, and machine learning. After structure-based virtual screening, docking scores helped us narrow down the list to Hinokiflavone, 3,29-Dibenzoyl Rarounitriol, and 4'-O-Methylochnaflavone. Molecular dynamics simulations (500 ns) showed that Hinokiflavone and 4'-O-Methylochnaflavone had stable binding and only small changes, which confirmed that the protein-ligand interactions were strong. Principal Component Analysis (PCA) and Free Energy Landscape (FEL) analyses showed that the InhA-ligand complexes were stable in shape and exhibited clear low-energy states. Frontier Molecular Orbital (FMO) analysis showed that the reactivity and electronic profiles were good, especially for Hinokiflavone, which had a small HOMO-LUMO gap. Furthermore, a machine-learning-based QSAR model was used to predict the biological activity values (pIC₅₀) of these compounds post-simulation. The best KNN model showed that the pIC₅₀ values of these compounds were greater than 7.0, exceeding the activity threshold of pIC₅₀ ≥ 6.0. This means they are active inhibitors. These results underscore the potential of these phytochemicals as InhA inhibitors for the management of M. leprae infections and offer a robust in silico prediction for subsequent experimental validation.