Mechanistic inhibition of FtsZ-driven bacterial cytokinesis by natural products: an integrated machine learning and advanced drug discovery approach.

Tuberculosis (TB), caused by Mycobacterium tuberculosis (MTB), remains a major global health burden, particularly due to the emergence of multidrug-resistant (MDR) and extensively drug-resistant (XDR) strains. The FtsZ protein, essential for bacterial cytokinesis and lacking a human homolog, presents a selective and non-redundant drug target. In this study, we implemented a comprehensive computational pipeline to identify potential FtsZ inhibitors from the COCONUT natural product database. Initial high-throughput virtual screening and machine learning-based pIC prediction were employed to shortlist active compounds. The top candidates were further optimized using Density Functional Theory, followed by ADMET screening, redocking, and 1000-ns molecular dynamics simulations. Binding free energy estimation via MM/GBSA identified CNP0281420 (-53.40 ± 5.57 kcal/mol), CNP0277831 (-50.06 ± 4.19 kcal/mol), and CNP0310586 (-49.47 ± 3.73 kcal/mol) as top binders. These results were supported by QM/MM total energy calculations and PCA-based Free Energy Landscape (FEL) mapping, confirming their conformational stability and electronic compatibility with the FtsZ binding pocket. Overall, this integrative study highlights promising natural compounds with strong binding affinity and dynamic stability, positioning them as potential anti-TB drug candidates for future experimental validation.