Computer-based methods for analyzing inhibitors of dengue virus NS2B/NS3 protease and of Wee1 kinase

Abstract

Homology models of the Dengue virus (DV) NS2B/NS3 protease complexed with a tetra-peptidic inhibitor were constructed. Molecular dynamics (MD) simulations of these complexes were carried out to rationalize the ligand interaction. The validated model was then used for molecular docking studies of small-molecule inhibitors. The results derived from MD simulations of the complex between DV NS2B/NS3 protease and the tetra-peptidic inhibitor as well as with small-molecule inhibitors revealed that residues at the C-terminus of NS2B (Asp81-Ser85), at the S1 pocket (Leu128-Thr134 of NS3), His51, Asp75, Ser135, Gly151, Asn152, Gly153 and Tyr161 of NS3 are important for inhibitor interaction. Results also demonstrated that NS2B is important for stabilizing the binding pocket of NS3 as well as for stabilzing the binding of the tetra-peptidic inhibitor. A stepwise virtual screening (VS) for DV NS2B/NS3 protease inhibitors was carried out by combining pharmacophore and molecular docking-based screening with subsequent binding free energy calculation. Hit compounds were selected from commercial compound libraries and proposed for biological testing using the DV NS2B/NS3 protease. In the second part, a 3D-QSAR model using the CoMFA approach was constructed for pyrrolocarbazole derivatives reported as Wee1 kinase inhibitors. The derived model was found to be robust and predictive, indicated by good statistical values (r2 = 0.870, q2LOO = 0.764 and r2pred. = 0.790). The analysis of the graphical CoMFA contour plot provided insight into the relevant interactions of the inhibitors and the essential features of potent Wee1 kinase inhibitors. Subsequently, a structure-based approach was developed to predict the binding activities of Wee1 kinase inhibitors. Linear interaction energy models for pyrrolocarbazole and pyridopyrimidine derivative were established. The obtained models yielded a good correlation between the experimental binding affinities and the calculated binding free energies. A carried out enrichment study showed that most of the true active compounds could be obtained by screening only the first 10% of compound databases containing actives and decoys. Based on the validated linear interaction energy models and VS carried out on different compound collections, several hits were selected for biological testing against Wee1 kinase.