Predictive Modeling and Mechanistic Validation of Synergistic Pimodivir Combinations for Anti-influenza therapy via PB2cap Affinity Enhancement

The rapid discovery of synergistic antiviral combinations remains a critical challenge in combating virus diseases. Here, we present a machine learning (ML) framework integrating drug characterization and cellular activity data to predict efficacious single and combinatorial therapies against influenza A. Our model identified pimodivir—a PB2 cap-binding inhibitor—paired with epinephrine bitartrate or L-adrenaline as top candidates, demonstrating broad-spectrum synergy against pandemic H1N1 and recent circulating bovine H5N1 viruses. Experimental validation revealed these combinations enhance pimodivir’s binding affinity to the PB2cap domain. Minigenome assays confirmed dose-dependent synergistic suppression of viral polymerase activity, despite neither adjuvant showing standalone efficacy. Multi-model synergy scoring methods (ZIP, Loewe, HSA, Bliss) and cytopathic effect profiling further validated the therapeutic potential of the selected drug combinations. Our findings offer a roadmap for rational combination therapy design against influenza and potential other RNA viruses.