Ligand-based 3D pharmacophore design, virtual screening and molecular docking for novel p38 MAPK inhibitors

Abstract

The p38 mitogen-activated protein kinase (MAPK) identified in human monocytes represents a potentially attractive therapeutic target for a class of cytokine suppressive anti-inflammatory compounds. However, the limitations of protein therapeutics have stimulated significant research aimed at developing orally bioavailable, small molecular inhibitors of cytokines. In this paper, virtual ligand-based 3D pharmacophore model, CoMFA modelling and molecular docking were employed to identify novel and structurally instinct inhibitors of small molecules. 3D-ligand-based pharmacophore models with four elements for p38 MAPK inhibitors were firstly developed. Following the validation of the pharmacophore models through the test data set, Fischer’s randomization test and CoMFA analysis, the final pharmacophore model generated was found to be highly predictive for identifying p38 inhibitors over drug-like non-inhibitors. Calculated enrichment factor shows 88 % of actives being recovered in the top 7 % of the database, and all actives being recovered in the first 12 % of the database screened, while a true positive rate of 0.09 with a false positive rate of only 0.067 is obtained. The well-validated pharmacophore model was then utilised to screen several databases as a 3D query to retrieve compounds. All hits obtained from database searching were further screened using ADMET filters. Twenty-seven compounds were selected based on high fit values and diverse structures. The molecular docking with the binding site determined by X-ray structures of the protein provides a structure and binding validation for active compounds, which is in agreement with that obtained by the final pharmacophore model. The binding model and mechanism of the compounds are also described.