Machine Learning-Guided Repositioning of a SARS-CoV-2-Targeting Molecular Series as Cruzain Inhibitors

Drug repurposing and repositioning are concepts that involve identifying alternative therapeutic uses for existing drug candidates or molecular series. During the COVID-19 pandemic, hundreds of antivirals were developed, many of which remain unexplored for other diseases. Concurrently, machine learning (ML) has become a valuable tool in early drug discovery for screening the most promising compounds for a target. In this work, an ExtraTrees ML model is developed to predict inhibitory activity against cruzain, the main cysteine protease of Trypanosoma cruzi, the causative agent of Chagas disease. The model is used to screen a proprietary library of peptidomimetic compounds originally designed to target SARS-CoV-2 M^pro and human cathepsin L. High-affinity cruzain inhibitors are identified, some containing P1 moieties not previously reported in cruzain inhibitors, expanding the known chemical space for this target. Selected hits are validated using isothermal titration calorimetry and some compounds display more favorable enthalpic and entropic contributions to binding than similar peptidomimetic nitrile-based inhibitors. Notably, this is achieved without highly lipophilic R-groups, preserving drug-like properties. This work also highlights how compound libraries derived from global health efforts can be effectively repurposed for neglected tropical diseases with ML models.