Identification of potent inhibitors of potential VEGFR2: a graph neural network-based virtual screening and in vitro study.

Abstract

VEGFR2 is a transmembrane tyrosine kinase receptor expressed on vascular endothelial cells and is closely associated with tumour cell growth. A comparison of traditional Chinese medicines and natural products with existing VEGFR2 inhibitors revealed that the former exhibited superior anticancer properties while concomitantly showing a reduced incidence of adverse effects. We proposed a novel strategy for screening potential candidates targeting VEGFR2 in a Chinese medicine monomer database using a combination of AI deep learning and structure-based drug design. The graph neural network served as the final predictive model to evaluate the molecular activities within the database, resulting in the selection of six candidate compounds. Kinase inhibition assays showed that the three compounds exhibited significant inhibition of VEGFR2. Molecular docking and molecular dynamics simulations further demonstrated the stability of their binding to VEGFR2. This study identified three compounds that effectively inhibited VEGFR2, making them promising candidates in cancer treatment.