Ligand-based pharmacophore modeling of TNF-α to design novel inhibitors using virtual screening and molecular dynamics

Abstract

Tumor necrosis factor-α (TNF-α) is one of the promising targets for treating inflammatory (Crohn disease, psoriasis, psoriatic arthritis, rheumatoid arthritis) and various other diseases. Commercially available TNF-α inhibitors are associated with several risks and limitations. In the present study, we have identified small TNF-α inhibitors using in silico approaches, namely pharmacophore modeling, virtual screening, molecular docking, molecular dynamics simulation and free binding energy calculations. The study yielded better and potent hits that bind to TNF-α with significant affinity. The best pharmacophore model generated using LigandScout has an efficient hit rate and Area Under the operating Curve. High throughput virtual screening of SPECS database molecules against crystal structure of TNF-α protein, coupled with physicochemical filtration, PAINS test. Virtual hit compounds used for molecular docking enabled the identification of 20 compounds with better binding energies when compared with previously known TNF-α inhibitors. MD simulation analysis on 20 virtual identified hits showed that ligand binding with TNF-α protein is stable and protein-ligand conformation remains unchanged. Further, 16 compounds passed ADMET analysis suggesting these identified hit compounds are suitable for designing a future class of potent TNF-α inhibitors. Communicated by Ramaswamy H. Sarma