Identifying Novel ATX Inhibitors via Combinatory Virtual Screening Using Crystallography-Derived Pharmacophore Modelling, Docking Study, and QSAR Analysis

Abstract

Autotaxin (ATX) is considered as an interesting drug target for the therapy of several diseases. The goal of the research was to detect new ATX inhibitors which have novel scaffolds by using virtual screening. First, based on two diverse receptor-ligand complexes, 14 pharmacophore models were developed, and the 14 models were verified through a big test database. Those pharmacophore models were utilized to accomplish virtual screening. Next, for the purpose of predicting the probable binding poses of compounds and then carrying out further virtual screening, docking-based virtual screening was performed. Moreover, an excellent 3D QSAR model was established, and 3D QSAR-based virtual screening was applied for predicting the activity values of compounds which got through the above two-round screenings. A correlation coefficient r², which equals 0.988, was supplied by the 3D QSAR model for the training set, and the correlation coefficient r² equaling 0.808 for the test set means that the developed 3D QSAR model is an excellent model. After the filtering was done by the combinatory virtual screening, which is based on the pharmacophore modelling, docking study, and 3D QSAR modelling, we chose nine potent inhibitors with novel scaffolds finally. Furthermore, two potent compounds have been particularly discussed.

Keywords: 3D QSAR model; autotaxin inhibitor; docking calculation; pharmacophore model; virtual screening.

Figure 1

Schematic representation of the autotaxin (ATX) binding pocket and the different inhibitor binding modes reported to date. (This figure is cited from the reference 22) Type I, II, III, and IV inhibitors are represented in green, purple, cyan, and orange, respectively, on the schema with matching colors for carbon atoms on the exemplified structures. The gray dotted lines represent secondary ligands modeled next to the inhibitor as shown with gray carbons on the illustrations. Zinc ions are depicted in magenta. (A) Binding mode of type I inhibitors LPA (lysophosphatidic acid) 22:6 (a), HA-155 (b), and “compound 2”(c). (B) Binding mode of type II inhibitors PAT-494 (d), PAT-078 (e), and CRT0273750 (f). (C) Binding mode of type III inhibitors PAT-343 (g) and tauroursodeoxycholic acid (h). (D) Binding mode of type IV inhibitors GLPG1690 (i) and compound 9 (j).

Figure 2

Chemical structures of representative ATX inhibitors. PDB codes of co-crystallized structures with ATX are indicated in brackets.

Figure 3

(A) All chemical features identified based on the 5M7M complex. (B) All chemical features identified based on the 5MHP complex. Feature colors: blue, hydrophobic feature; green, hydrogen acceptor feature; orange, aromatic ring feature.

Figure 4

Binding modes of 7HR (yellow) (A) and 7NB (yellow) (B) in the active site of ATX. Ligands complexed with their receptors are also shown for comparison, 7HR and 7NB indicated in green stick form.

Figure 5

(A) The scaffold of 31 autotaxin inhibitors. (B) The alignment result of 31 Mcl-1 inhibitors based on the poses acquired by the docking study calculation.

Figure 6

Plots of predicted autotaxin inhibitory activities versus experimental of training set and test set.

Figure 7

(A) 3D QSAR model coefficients on electrostatic potential grids. Blue represents positive coefficients; red represents negative coefficients. (B) 3D QSAR model coefficients on van der Waals grids. Green represents positive coefficients; yellow represents negative coefficients.

Figure 8

The workflow chart of the study. A combinatory virtual screening (VS) protocol based on the pharmacophore model, molecular docking study, and the 3D QSAR model was utilized to discover novel inhibitors targeting autotaxin.

Figure 9

The 3D chemical structures of the final nine selected compounds.

Figure 10

(A) Mapping of 5M7M 01 with compound cpd4. (B) The possible binding pose of cpd1 in the autotaxin active site. Compound 7HR complexed with autotaxin is also shown for comparison (in gray stick form).

Figure 11

(A) Mapping of 5MHP 03 with compound cpd7. (B) The possible binding pose of cpd7 in the autotaxin active site. Compound 7NB complexed with autotaxin is also shown for comparison (in gray stick form).

Figure 12

The mappings of cpd7 with isosurface-EP (–, red; +, blue) (A), isosurface-VMD (–, yellow; +, green) (B) grids. cpd7 is presented in green stick form. Compound 7NB complexed with autotaxin is also shown for comparison (in gray stick form).

Figure 13

The mappings of cpd4 with isosurface-EP (–, red; +, blue) (A), isosurface-VMD (–, yellow; +, green) (B) grids. cpd4 is presented in purple stick form. Compound 7HR complexed with autotaxin is also shown for comparison (in gray stick form).