Journal of Research in Pharmacy

Öz

Hepatitis C Virus (HCV) is a blood-borne RNA virus that causes inflammation of the liver that can lead to liver cirrhosis and hepatocellular carcinoma [1]. Non-structural protein 5B (NS5B) is an essential component of HCV for viral transcription and genome replication [2]. As there is no close mammalian analog for this enzyme, it has been the focus of many drug discovery projects [3]. In the present work, a combination of different computer-aided drug design approaches such as ensemble docking, binding free energy calculations, and quantitative structure-activity relationship (QSAR) model generation was applied to identify novel inhibitors of NS5B. In the first step, all available protein structures in Protein Data Bank in a complex with thumb site 2 inhibitors were collected. Then, an automated KNIME [4] workflow was generated to select a few representative structures of the conformational changes in the binding pocket upon ligand binding. In total eight NS5B-inhibitor complexes were selected for further in silico work. Next, a virtual combinatorial library was obtained using the privileged substructures of known NS5B inhibitors. Different congeneric series of compounds including phenylalanine derivatives, thiophene-2-carboxylic acid derivatives, and anthranilic acid derivatives were used for the database formation. Upon ligand preparation, over 182.000 molecules were built. Consequently, known thumb site 2 inhibitors were docked with GLIDE-SP [5-7] and rescored with Prime MM-GBSA [8,9] protocol implemented in Schrödinger software to estimate the docking and binding free energy scores that will be used as a threshold for filtering the newly produced combinatorial library. In addition, categorical and numerical QSAR models were generated based on the known thumb site 2 inhibitors and used in the post-filtering step. Compounds that were predicted as actives will be visually analyzed and selected further for synthesis and biological evaluation.