Computational design of new protein kinase 2 inhibitors for the treatment of inflammatory diseases using QSAR, pharmacophore-structure-based virtual screening, and molecular dynamics.

Receptor-interacting protein kinase 2 (RIPK2) plays an essential role in autoimmune response and is suggested as a target for inflammatory diseases. A pharmacophore model was built from a dataset with ponatinib (template) and 18 RIPK2 inhibitors selected from BindingDB database. The pharmacophore model validation was performed by multiple linear regression (MLR). The statistical quality of the model was evaluated by the correlation coefficient (R), squared correlation coefficient (R2), explanatory variance (adjusted R2), standard error of estimate (SEE), and variance ratio (F). The best pharmacophore model has one aromatic group (LEU24 residue interaction) and two hydrogen bonding acceptor groups (MET98 and TYR97 residues interaction), having a score of 24.739 with 14 aligned inhibitors, which were used in virtual screening via ZincPharmer server and the ZINC database (selected in function of the RMSD value). We determined theoretical values of biological activity (logRA) by MLR, pharmacokinetic and toxicology properties, and made molecular docking studies comparing binding affinity (kcal/mol) results with the most active compound of the study (ponatinib) and WEHI-345. Nine compounds from the ZINC database show satisfactory results, yielding among those selected, the compound ZINC01540228, as the most promising RIPK2 inhibitor. After binding free energy calculations, the following molecular dynamics simulations showed that the receptor protein's backbone remained stable after the introduction of ligands.