Identification\nof Potential Tryptase Inhibitors from\nFDA-Approved Drugs Using Machine Learning, Molecular Docking, and\nExperimental Validation

This study explores the innovative use of machine learning\n(ML)\nto identify novel tryptase inhibitors from a library of FDA-approved\ndrugs, with subsequent confirmation via molecular docking and experimental\nvalidation. Tryptase, a significant mediator in inflammatory and allergic\nresponses, presents a therapeutic target for various inflammatory\ndiseases. However, the development of effective tryptase inhibitors\nhas been challenging due to the enzyme’s complex activation\nand regulation mechanisms. Utilizing a machine learning model, we\nscreened an extensive FDA-approved drug library to identify potential\ntryptase inhibitors. The predicted compounds were then subjected to\nmolecular docking to assess their binding affinity and conformation\nwithin the tryptase active site. Experimental validation was performed\nusing RBL-2H3 cells, a rat basophilic leukemia cell line, where the\nefficacy of these compounds was evaluated based on their ability to\ninhibit tryptase activity and suppress β-hexosaminidase activity\nand histamine release. Our results demonstrated that several FDA-approved\ndrugs, including landiolol, laninamivir, and cidofovir, significantly\ninhibited tryptase activity. Their efficacy was comparable to that\nof the FDA-approved mast cell stabilizer nedocromil and the investigational\nagent APC-366. These findings not only underscore the potential of\nML in accelerating drug repurposing but also highlight the feasibility\nof this approach in identifying effective tryptase inhibitors. This\nresearch contributes to the field of drug discovery, offering a novel\npathway to expedite the development of therapeutics for tryptase-related\npathologies.