Generative AI, molecular docking and molecular dynamics simulations assisted identification of novel transcriptional repressor EthR inhibitors to target Mycobacterium tuberculosis.

Abstract

Tuberculosis (TB) remains a persistent global health threat, with Mycobacterium tuberculosis (Mtb) continuing to be a leading cause of mortality worldwide. Despite efforts to control the disease, the emergence of multi-drug-resistant (MDR) and extensively drug-resistant (XDR) TB strains presents a significant challenge to conventional treatment approaches. Addressing this challenge requires the development of novel anti-TB drug molecules. This study employed de novo drug design approaches to explore new EthR ligands and ethionamide boosters targeting the crucial enzyme InhA involved in mycolic acid synthesis in Mtb. Leveraging REINVENT4, a modern open-source generative AI framework, the study utilized various optimization algorithms such as transfer learning, reinforcement learning, and curriculum learning to design small molecules with desired properties. Specifically, focus was placed on molecule optimization using the Mol2Mol option, which offers multinomial sampling with beam search. The study's findings highlight the identification of six promising compounds exhibiting enhanced activity and improved physicochemical properties through structure-based drug design and optimization efforts. These compounds offer potential candidates for further preclinical and clinical development as novel therapeutics for TB treatment, providing new avenues for combating drug-resistant TB strains and improving patient outcomes.