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

Area of Science:

• Drug discovery and development
• Computational chemistry
• Parasitology

Background:

• Drug repurposing identifies new uses for existing drugs, crucial for diseases like Chagas.
• Machine learning (ML) accelerates drug discovery by screening compound libraries.
• Antiviral compounds developed during the COVID-19 pandemic offer a potential resource for other diseases.

Purpose of the Study:

• To develop a machine learning model for predicting inhibitory activity against cruzain, the target protease in Chagas disease.
• To screen a library of peptidomimetic compounds, originally designed for SARS-CoV-2 Mpro and cathepsin L, for cruzain inhibition.
• To identify novel, high-affinity cruzain inhibitors and expand the chemical space for Chagas disease therapeutics.

Main Methods:

• Development of an ExtraTrees machine learning model to predict cruzain inhibitory activity.
• Screening of a proprietary peptidomimetic library against cruzain using the ML model.
• Validation of identified hit compounds using isothermal titration calorimetry (ITC).

Main Results:

• Identification of high-affinity cruzain inhibitors from a library targeting SARS-CoV-2 Mpro and cathepsin L.
• Discovery of novel P1 moieties in inhibitors, expanding the known chemical space for cruzain.
• Selected compounds exhibited favorable enthalpic and entropic binding contributions without compromising drug-like properties (e.g., low lipophilicity).

Conclusions:

• Machine learning effectively repurposes drug libraries for neglected tropical diseases like Chagas disease.
• This study identified promising cruzain inhibitors with improved binding thermodynamics and drug-like properties.
• The findings demonstrate the value of leveraging global health compound libraries for diverse therapeutic applications.