Abstract

Snakebite envenoming remains a neglected health challenge in tropical regions, where access to antivenom is limited. Phospholipase A2 (PLA2), a major venom toxin, is a promising therapeutic target. This study utilized an AI-driven drug discovery pipeline to explore 303 phytochemicals from Cyanthillium cinereum, a plant traditionally used to treat snakebite, for their potential as PLA2 inhibitors. Machine learning models (Random Forest, XGBoost, and stacked autoencoder), molecular docking, and molecular dynamics (MD) simulations were applied to identify and validate active compounds. The top candidates exhibited strong binding affinities and stable interactions with PLA2 isoforms from Crotalus durissus terrificus, Daboia russelii, and Bothrops asper. In particular, 2-Amino-3- (4-hydroxy-3-methoxyphenyl) propanoic acid demonstrated stable binding within the active side of C. d. terrificus PLA2 during MD simulations. These findings suggest that C. cinereum contains bioactive compounds with promising anti-PLA2 activity, highlighting the potential of AI-based approaches to accelerate antivenom discovery. Further pharmacological and in vivo validation is warranted to advance the development of plant-derived inhibitors for snakebite treatment.

Key words: antivenom, Cyanthillium cinereum, drug target interaction, machine learning, phospholipase A2