Repurposing of drugs for combined treatment of COVID-19 cytokine storm using machine learning

Area of Science:

• Computational drug discovery
• Virology
• Immunology

Background:

• Cytokine storm induced by SARS-CoV-2 is a primary driver of COVID-19 mortality.
• Current treatments targeting single proteins show limited efficacy due to complex cell signaling pathways.
• Five key proteins (AT1R, ADAM17, NF-κB, JAK1, STAT3) are implicated in the SARS-CoV-2 cytokine storm pathway.

Purpose of the Study:

• To identify novel therapeutics for COVID-19 by targeting multiple proteins in the cytokine storm pathway.
• To develop and apply machine learning models for predicting drug efficacy against key SARS-CoV-2 targets.
• To screen FDA-approved drugs for potential repurposing as multi-target COVID-19 therapeutics.

Main Methods:

• Developed machine learning models for five proteins (AT1R, ADAM17, NF-κB, JAK1, STAT3) using known inhibitors.
• Screened FDA-approved drugs against these models to predict activity.
• Validated machine learning predictions using PyRx software for protein-ligand docking and binding affinity calculations.

Main Results:

• Identified twenty drugs active against four proteins (score > 0.8) and eight drugs active against all five proteins (score > 0.85).
• Mitomycin C demonstrated the highest overall activity across all five proteins with an average prediction score of 0.886.
• Protein-ligand docking experiments corroborated the machine learning model findings.

Conclusions:

• Several FDA-approved drugs show potential for simultaneously targeting multiple proteins in the SARS-CoV-2 cytokine storm pathway.
• These multi-target therapies could offer synergistically effective treatments for COVID-19 by inhibiting the cytokine storm at various points.
• The study highlights the utility of machine learning and drug repurposing in identifying novel therapeutic strategies for complex diseases like COVID-19.