Dynamic structure-based pharmacophore model development: a new and effective addition in the histone deacetylase 8 (HDAC8) inhibitor discovery

Abstract

Histone deacetylase 8 (HDAC8) is an enzyme involved in deacetylating the amino groups of terminal lysine residues, thereby repressing the transcription of various genes including tumor suppressor gene. The over expression of HDAC8 was observed in many cancers and thus inhibition of this enzyme has emerged as an efficient cancer therapeutic strategy. In an effort to facilitate the future discovery of HDAC8 inhibitors, we developed two pharmacophore models containing six and five pharmacophoric features, respectively, using the representative structures from two molecular dynamic (MD) simulations performed in Gromacs 4.0.5 package. Various analyses of trajectories obtained from MD simulations have displayed the changes upon inhibitor binding. Thus utilization of the dynamically-responded protein structures in pharmacophore development has the added advantage of considering the conformational flexibility of protein. The MD trajectories were clustered based on single-linkage method and representative structures were taken to be used in the pharmacophore model development. Active site complimenting structure-based pharmacophore models were developed using Discovery Studio 2.5 program and validated using a dataset of known HDAC8 inhibitors. Virtual screening of chemical database coupled with drug-like filter has identified drug-like hit compounds that match the pharmacophore models. Molecular docking of these hits reduced the false positives and identified two potential compounds to be used in future HDAC8 inhibitor design.

Keywords: Lipinski’s rule; molecular docking; molecular dynamics simulation; pharmacophore; virtual screening.

Figure 1

The 3D structure of human histone deacetylase 8 (HDAC8) bound with an inhibitor (PDB code 2V5X). A zoomed view of inhibitor binding region shows the tunnel-like active site in mesh form. The hydroxamic acid part of inhibitor binds close to the metal (Zn²⁺) ion whereas the aliphatic chain and hydrophobic cap groups occupy the tunnel and surface of the active site.

Figure 2

The 2D chemical structures of HDAC8 inhibitors used in this study are displayed with two clinically proven HDAC inhibitors for structural comparison.

Figure 3

Comparison of (a) root mean square deviation (RMSD) of backbone atoms; (b) root mean square fluctuation (RMSF); and (c) potential energy values of 5 ns molecular dynamic (MD) simulations for three systems.

Figure 4

Distance between the hydroxamic acid moieties of inhibitors and two catalytically important histidine residues (a) H142 and (b) H143 in the mechanism of HDAC8 enzyme.

Figure 5

The histogram obtained from the clustering analyses of last 4 of 5 ns conformational regions of HDAC8-inhibitor complex MD simulation trajectories. Dark and light gray color cylinders represent the clusters obtained from HDAC8-C1 and HDAC8-C2 complexes, respectively.

Figure 6

Overlay of the representative structures obtained from 5 ns MD simulations of HDAC8-C1 (white) and HDAC8-C2 (green) complexes. Important amino acid residues of HDAC enzyme and the inhibitors are shown in stick and ball-stick forms, respectively.

Figure 7

Structure-based pharmacophore model generated from the HDAC8-C1 complex. Secondary structure of protein is shown in cartoon and amino acid residues are shown in stick form. The C1 inhibitor is shown in ball-stick representation. The identified pharmacophoric features are shown in green, cyan and magenta for HA, HY and HD features, respectively.

Figure 8

Structure-based pharmacophore model generated from the HDAC8-C2 complex. Secondary structure of protein is shown in cartoon and amino acid residues are shown in stick form. The C2 inhibitor is shown in ball-stick representation. The identified pharmacophoric features are shown in green, cyan and magenta for HA, HY and HD features, respectively.

Figure 9

Generated pharmacophore models (a) Pharm-A and (b) Pharm-B are shown with their inter-feature distance constraints. Green, cyan, and magenta colors represent HA, HY and HD features, respectively.

Figure 10

Binding modes and molecular interactions of (a) inhibitor C1 (b) inhibitor C2 (c) hit 1 and (d) hit 2 at the active sites of two different inhibitor-induced conformations. White and green cartoons represent C1- and C2-induced conformations of HDAC8 enzyme. Amino acid residues are shown in stick form whereas the ligands are shown in ball-stick form. Hydrogen atoms are not shown for clear view.

Figure 11

The 2D chemical structures of identified hits.