A Machine Learning and Quantum Chemical Strategy to Disrupt Oncogenic Arginine Methylation

Protein arginine N-methyltransferase 1 (PRMT1) plays a key role in gene regulation by modifying proteins through asymmetric arginine dimethylation, a process often linked to cancer development. Due to its elevated activity in cancer, PRMT1 is a promising target for therapeutic intervention. This study used an integrated approach to identify and optimize selective inhibitors of PRMT1, starting from the known compound GSK3368715. Over 1,000 structurally related analogs were screened for their binding potential. Top candidates underwent quantum mechanical optimization and docking studies, which confirmed strong interactions with key active-site residues, including hydrogen bonding with Tyr57 and Glu171, and π-π stacking with Tyr53 and Phe54. Molecular dynamics simulations revealed stable binding behavior, supported by MM/GBSA binding energy calculations. One compound, 118971667, showed particularly high binding affinity and structural stability. Principal component analysis and free energy landscapes further confirmed its favorable dynamics. Additionally, a machine learning model trained on 290 known PRMT1 inhibitors predicted 118971667 to have strong inhibitory potential (pIC50 of 6.66), outperforming a reference compound. This comprehensive strategy, combining structural modeling, simulations, quantum calculations, and predictive analytics, highlights 118971667 as a promising PRMT1 inhibitor and offers valuable insights into targeting arginine methylation in cancer.