Novel FXa Inhibitor Identification through Integration of Ligand- and Structure-Based Approaches

Abstract

Factor Xa (FXa), a vitamin K-dependent serine protease plays a pivotal role in the coagulation cascade, one of the most interesting targets for the development of new anticoagulants. In the present work, we performed a virtual screening campaign based on ligand-based shape and electrostatic similarity search and protein-ligand docking to discover novel FXa-targeted scaffolds for further development of inhibitors. From an initial set of 260,000 compounds from the NCI Open database, 30 potential FXa inhibitors were identified and selected for in vitro biological evaluation. Compound 5 (NSC635393, 4-(3-methyl-4H-1,4-benzothiazin-2-yl)-2,4-dioxo-N-phenylbutanamide) displayed an IC₅₀ value of 2.02 nM against human FXa. The identified compound may serve as starting point for the development of novel FXa inhibitors.

Keywords: blood coagulation cascade; enzyme inhibitors; factor Xa; protein-ligand docking; shape-based screening; virtual screening.

Figure 1

Currently approved and investigational FXa inhibitors.

Figure 2

(A) Schematic representation of FXa protein structural domains showing the location of the binding site and ions within the heavy chain (green cartoon) and the light chain (light blue cartoon); (B) The X-ray crystal structure of Apixaban (yellow carbons) in complex with human FXa (PDBid 2P16). Essential amino acids and binding pockets are indicated; hydrogen bonds between Apixaban and FXa residues and structural waters are shown as green dotted lines.

Figure 3

Schematic overview of the procedure used for selecting the FXa-ligand complexes.

Figure 4

Schematic overview of the virtual screening workflow and the procedure used for selecting the virtual screening hits whose bioactivity was experimentally tested. The number of compounds that passed each step and the programs used are shown. From an initial set of near 260,000 compounds, 300 compounds were identified as putative FXa inhibitors by the virtual screening workflow. Thirty of these compounds were selected for in vitro testing.

Figure 5

Schematic overview of ROCS shape-based queries and electrostatic grids based on the center ligand corresponding to each cluster.

Figure 6

ROCS shape-based query validation statistics for Cluster 2. (A) ROC plot and (B) validation statistics using the active ligands of cluster 2 or the FXa actives from the DUD-E database. (C) Superposition of Rivaroxaban over the cluster 2 shape-based query, and (D) Comparison of crystallographic position of Rivaroxaban (PDB id 2W26) with the position obtained for compound 5 with ROCS and FRED runs.

Figure 7

Comparison between the predicted binding mode and in vitro FXa inhibition for (A) compound 5 (NSC635393) and (B) Rivaroxaban.

Figure 8

(A) In vitro aPTT and PT of compound 5 (NSC635393) and Rivaroxaban using human plasma; (B) Tautomerization of compound 5 (NSC635393) to its enol form, with predicted physicochemical and plasma protein binding properties.