Machine learning-guided discovery of spirocyclic inhibitors targeting Mycobacterium tuberculosis FtsZ

Tuberculosis continues to pose serious threats to global public health and calls for the design of new drugs that can target key bacterial proteins. This research used an integrated approach involving computational techniques to discover new drugs that can inhibit the FtsZ protein of Mycobacterium tuberculosis using the Life Chemicals compound library as a starting point. Sequence analysis revealed very low sequence identity (~ 12-13%) between FtsZ and human cytoskeletal proteins, indicating substantial evolutionary divergence. Five lead compounds were discovered through structure-based screening, exhibiting good binding affinities in the range of - 7.18 to - 6.69 kcal/mol. Binding free energy studies showed that F3411-4559 was the most active compound (binding free energy = - 69.49 kcal/mol) compared with the reference (binding free energy = - 65.78 kcal/mol). A pharmacokinetics study suggested good drug-likeness in terms of excellent intestinal permeability and suitable physicochemical parameters. The quantum chemistry study demonstrated improved electronic reactivity for F3411-4594. The molecular dynamics studies proved that all the protein-ligand complexes were stable, with good structures and conformations. The predictive models also indicated high inhibition activity (pIC50 = 7.75-8.51). In conclusion, spirocyclic derivatives are promising candidates for the development of novel anti-tuberculosis drugs.