Discovery of potential protein tyrosine phosphatase non-receptor type 2 inhibitors from phytochemical database through structure- based high-throughput virtual screening and machine learning Discovery of protein tyrosine phosphatase non-receptor type 2 inhibitors

Abstract

Abstract Protein tyrosine phosphatase non-receptor type 2 (PTPN2) has recently become a viable target for cancer immunotherapy. Our study aimed to predict compounds with potential to inhibit PTPN2 for anticancer immunotherapy through a combination of structure-based and ligand-based virtual screening. Using a virtual screening method based on the protein structure of PTPN2 (PDB: 7UAD), we were able to identify 94 compounds from the PhytoHub database that have the ability to bind to PTPN2. These compounds include flavonoids, stilbenes, alkaloids, carotenoids, coumarin, ellagitannins, diterpenoids, curcuminoids, and phenolic acids. The predicted IC50 value for the PTPN2 inhibitory activity of the compounds is from 15.25 µM to 3.26 µM through our machine learning model. Among them, the leading structures of each group also had stable complexes with proteins during molecular dynamics simulations and tightly bound to proteins while exhibiting low binding free energy (MM/GBSA). The two leading groups of compounds are metabolites derived from hesperetin and trans-resveratrol (sulfate and glucuronide) that can be further investigated as potential PTPN2 inhibitors for anti-tumor immunotherapy.