Rheumatoid arthritis (RA) remains one of the most prevalent autoimmune diseases worldwide. Janus kinase 3 (JAK3) is an essential enzyme for treating autoimmune diseases, including RA. Molecular modeling techniques play a crucial role in the search for new drugs by reducing time delays. In this study, the 3D-QSAR approach is employed to predict new JAK3 inhibitors. Two robust models, both field-based with R2 = 0.93, R = 0.96, and Q2 = 87, and atom-based with R2 = 0.94, R = 0.97, and Q2 = 86, yielded good results by identifying groups that may readily direct their interaction. A reliable pharmacophore model, DHRRR1, was provided in this work to enable the clear characterization of chemical features, leading to the design of 13 inhibitors with their pIC50 values. The DHRRR1 model yielded a validation result with a ROC value of 0.87. Five promising inhibitors were selected for further study based on an ADMET analysis of their pharmacokinetic properties and covalent docking (CovDock). Compared to the FDA-approved drug tofacitinib, the pharmaceutical features, binding affinity and stability of the inhibitors were analyzed through CovDock, 300 ns molecular dynamics simulations, free energy binding calculations and ADMET predictions. The results show that the inhibitors have strong binding affinity, stability and favorable pharmaceutical properties. The newly predicted molecules, as JAK3 inhibitors for the treatment of RA, are promising candidates for use as drugs.
Keywords: 3D-QSAR; JAK3; MM/GBSA; cancer; computational modeling; drug discovery; rheumatoid arthritis.