3D-QSAR and free energy landscape analysis for predicting novel JAK3 inhibitors in rheumatoid arthritis therapy

JAK3 is a main tyrosine kinase within the JAK kinase family, whose inhibition plays a crucial role in the treatment of rheumatoid arthritis (RA). Multiple selective JAK3 inhibitors with different scaffolds exist. Decernotinib, a pyrrolopyridine-based RA candidate, showed early potential but was discontinued during Phase II. Currently, no selective JAK3 inhibitor is available for rheumatoid arthritis. In this research, an integrated ligand- and structure-based approach is employed to develop novel JAK3 inhibitors. Two 3D-QSAR models were developed based on CoMFA and CoMSIA methods, using a library of pyrimidine-based JAK3 inhibitor structures. The best models have been chosen based on statistical values of r² greater than 0.85 for CoMFA and CoMSIA. The models underwent internal and external validation. The electrostatic, steric, hydrogen bond acceptor, hydrogen bond donor, and hydrophobic contour maps using CoMFA and CoMSIA methods were generated, and 103 novel compounds were predicted based on all of these features. To explore the binding orientations and the associated binding free energies of the predicted compounds within JAK3’s active site, molecular docking followed by MM-GBSA analyses was conducted. Molecular dynamics simulations were additionally performed to assess the results for the top five predicted compounds. Among the predicted compounds, compound 56 stood out as the most promising candidate, displaying superior binding stability. Additionally, the free energy landscape for the JAK3-56 complex revealed stable conformations. Generally, 3D-QSAR pharmacophore modeling and MD observations will inform lead optimization and the creation of new, effective, selective JAK3 inhibitors.