Identification of novel PAD2 inhibitors using pharmacophore-based virtual screening, molecular docking, and MD simulation studies

Abstract

In the realm of epigenetic regulation, Protein arginine deiminase 2 (PAD2) stands out as a therapeutic target due to its significant role in neurological disorders, rheumatoid arthritis (RA), multiple sclerosis (MS), and various cancers. To date, no in silico studies have focused on PAD2 for lead compound identification. Therefore, we conducted structure-based pharmacophore modeling, virtual screening, molecular docking, molecular dynamics (MD) simulations, and essential dynamics studies using PCA and free energy landscape analyses to identify repurposed drugs and selective inhibitors against PAD2. The best pharmacophore model, 'Pharm_01,' had a selectivity score of 10.485 and an excellent ROC curve quality of 0.972. Pharm1 consisted of three hydrogen bond donors (HBD) and two hydrophobic (Hy) features (DDDHH). A virtual screening of about 9.2 million compounds yielded 2575 hits using a fit value threshold of 2.5 and drug-likeness criteria. Molecular docking identified the top ten molecules, which were verified using MD simulations. Stability was verified using MM-PBSA studies, whereas conformational differences were investigated using PCA and free energy landscape analyses. Two hits (Leads 1 and 2) from the DrugBank dataset showed promise for repurposing as PAD2 inhibitors, while one hit compound (Lead 8) from the ZINC database emerged as a novel PAD2 inhibitor. These findings indicate that the discovered compounds may be potent PAD2 inhibitors, necessitating additional preclinical and clinical research to produce viable treatments for cancer and neurological disorders.

Keywords: Drug repurposing; MD simulation; PAD2; PCA analyses; Structure-based Pharmacophore Model; Virtual screening.

Fig. 1

Structural and sequence similarities between PAD2 and PAD4, a Superimposed protein structures of PAD2 (green) and PAD4 (yellow) for the identification of the active site (red) of PAD2 based on the catalytic site information of PAD4, b 3D interactions of “AFM-30a” highlighted in red color with PAD2 (green) and PAD4 (yellow)in superimposed form, c Sequence similarity between PAD2 and PAD4. In the active site, important catalytic site residues encircled with red color were conserved in both isoforms.

Fig. 2

Structure-based pharmacophore model generation and validation. (a) 2D interaction analysis of human PAD2 with AFM-30a, (b) Pharmacophoric features in the presence of receptor-ligand interactions, (c) Best pharmacophore model (DDDHH), (d) ROC curve displaying the quality of the model.

Fig. 3

Visualization of the PAD2 protein structure, showcasing the predicted binding site with grid spheres and key binding site residues. (a) Complete PAD2 protein structure with the grid sphere representing the predicted binding site. (b) Close-up view of the protein binding site with the grid sphere highlighting the docking region. (c) Detailed view of the binding site residues with side chains within the binding pocket.

Fig. 4

2D receptor-ligand interactions of the final hit compounds in the active site of PAD2.

Fig. 5

RMSD analysis of the filtered compounds throughout the 100ns MD run.

Fig. 6

Protein-protein H-bond analysis of the filtered compounds throughout the 100ns MD run.

Fig. 7

PCA plot illustrating the most significant principal components of motion for the Cα atoms of PAD2 in complex with the control compound (AFM-30a) and selected lead compounds.

Fig. 8

Gibbs free energy landscape. (a) Control, (b) Lead 1, (c) Lead 2, (d) Lead 5, (e) Lead 8, and (f) Lead 9.

Fig. 9

Binding mode analyses of lead 2 compared to other leads and control compound (AFM-30a). The color-coded interactions underscore the significant contributions of hydrogen bonds (green color), pi-alkyl (pink color), pi-pi stacking bond (dark pink color), pi-sulfur (orange color), and pi-cation (golden color) in the inhibitory activity of these lead compounds against PAD2.

Fig. 10

Graphical summary of the study on PAD2 inhibitor discovery. This figure highlights the stepwise approach, including pharmacophore-based virtual screening, molecular docking, molecular dynamics simulations, and binding energy calculations, leading to the identification of PAD2 inhibitors.