Development of potent and selective factor Xa inhibitors

An unexpected dynamic binding mode between coagulation factor X and Rivaroxaban reveals importance of flexibility in drug binding

Rivaroxaban (RIV) is a direct oral anticoagulant (DOAC) targeting activated coagulation factor X (FXa). An earlier study reported the F174A mutant of FXa resistant to a RIV-like inhibitor, Apixaban. In current study, the detailed molecular mechanism of the resistance has been explored by molecular dynamics simulations on the impaired interactions between RIV and FXa in the damaged S4 pocket of F174A mutant. Besides, an unexpected relative stable binding mode of S1'S1 was revealed, which required dynamic motions of Gln192 and Gln61 to allow the morpholinone moiety of RIV to shift into the S1' pocket and form strong interactions. These dynamic motions of RIV and critical residues might be important in drug design for direct inhibitors of coagulation factors.

Prediction of factor Xa inhibitors by machine learning methods

Factor Xa (FXa) inhibitors have been explored as anticoagulants for treatment and prevention of thrombotic diseases. Molecular docking, pharmacophore, quantitative structure-activity relationships, and support vector machines (SVM) have been used for computer prediction of FXa inhibitors. These methods achieve promising prediction accuracies of 69-80% for FXa inhibitors and 85-99% for non-inhibitors. Prediction performance, particularly for inhibitors, may be further improved by exploring methods applicable to more diverse range of compounds and by using more appropriate set of molecular descriptors. We tested the capability of several machine learning methods (C4.5 decision tree, k-nearest neighbor, probabilistic neural network, and support vector machine) by using a much more diverse set of 1098 compounds (360 inhibitors and 738 non-inhibitors) than those in other studies. A feature selection method was used for selecting molecular descriptors appropriate for distinguishing FXa inhibitors and non-inhibitors. The prediction accuracies of these methods are 89.1-97.5% for FXa inhibitors and 92.3-98.1% for non-inhibitors. In particular, compared to other studies, support vector machine gives a substantially improved accuracy of 94.6% for FXa non-inhibitors and maintains a comparable accuracy of 98.1% for inhibitors, based-on a more rigorous test with more diverse range of compounds. Our study suggests that machine learning methods such as SVM are useful for facilitating the prediction of FXa inhibitors.

Discovery of novel hydroxy pyrazole based factor IXa inhibitor

Synthesis and biological data of a novel selective and efficacious factor IXa inhibitor are described along with its crystal structure in factor VIIa.

Dissecting and Designing Inhibitor Selectivity Determinants at the S1 Site Using an Artificial Ala190 Protease (Ala190 uPA)

A site-directed mutant of the serine protease urokinase-type plasminogen activator (uPA), was produced to assess the contribution of the Ser190 side-chain to the affinity and selectivity of lead uPA inhibitors in the absence of other differences present in comparisons of natural proteases. Crystallography and enzymology involving WT and Ala190 uPA were used to calculate free energy binding contributions of hydrogen bonds involving the Ser190 hydroxyl group (O(gamma)(Ser190)) responsible for the remarkable selectivity of 6-halo-5-amidinoindole and 6-halo-5-amidinobenzimidazole inhibitors toward uPA and against natural Ala190 protease anti-targets. Crystal structures of uPA complexes of novel, active site-directed arylguanidine and 2-aminobenzimidazole inhibitors of WT uPA, together with associated K(i) values for WT and Ala190 uPA, also indicate a significant role of Ser190 in the binding of these classes of uPA inhibitors. Structures and associated K(i) values for a lead inhibitor (CA-11) bound to uPA and to five other proteases, as well as for other leads bound to multiple proteases, help reveal the features responsible for the potency (K(i)=11nM) and selectivity of the remarkably small inhibitor, CA-11. The 6-fluoro-5-amidinobenzimidzole, CA-11, is more than 1000-fold selective against natural Ala190 protease anti-targets, and more than 100-fold selective against other Ser190 anti-targets.

SVM-Based Feature Selection for Characterization of Focused Compound Collections

Artificial neural networks, the support vector machine (SVM), and other machine learning methods for the classification of molecules are often considered as a "black box", since the molecular features that are most relevant for a given classifier are usually not presented in a human-interpretable form. We report on an SVM-based algorithm for the selection of relevant molecular features from a trained classifier that might be important for an understanding of ligand-receptor interactions. The original SVM approach was extended to allow for feature selection. The method was applied to characterize focused libraries of enzyme inhibitors. A comparison with classical Kolmogorov-Smirnov (KS)-based feature selection was performed. In most of the applications the SVM method showed sustained classification accuracy, thereby relying on a smaller number of molecular features than KS-based classifiers. In one case both methods produced comparable results. Limiting the calculation of descriptors to only the most relevant ones for a certain biological activity can also be used to speed up high-throughput virtual screening.

Elaborate manifold of short hydrogen bond arrays mediating binding of active site-directed serine protease inhibitors

An extensive structural manifold of short hydrogen bond-mediated, active site-directed, serine protease inhibition motifs is revealed in a set of over 300 crystal structures involving a large suite of small molecule inhibitors (2-(2-phenol)-indoles and 2-(2-phenol)-benzimidazoles) determined over a wide range of pH (3.5-11.4). The active site hydrogen-bonding mode was found to vary markedly with pH, with the steric and electronic properties of the inhibitor, and with the type of protease (trypsin, thrombin or urokinase type plasminogen activator (uPA)). The pH dependence of the active site hydrogen-bonding motif is often intricate, constituting a distinct fingerprint of each complex. Isosteric replacements or minor substitutions within the inhibitor that modulate the pK(a) of the phenol hydroxyl involved in short hydrogen bonding, or that affect steric interactions distal to the active site, can significantly shift the pH-dependent structural profile characteristic of the parent scaffold, or produce active site-binding motifs unique to the bound analog. Ionization equilibria at the active site associated with inhibitor binding are probed in a series of the protease-inhibitor complexes through analysis of the pH dependence of the structure and environment of the active site-binding groups involved in short hydrogen bond arrays. Structures determined at high pH (>11), suggest that the pK(a) of His57 is dramatically elevated, to a value as high as approximately 11 in certain complexes. K(i) values involving uPA and trypsin determined as a function of pH for a set of inhibitors show pronounced parabolic pH dependence, the pH for optimal inhibition governed by the pK(a) of the inhibitor phenol involved in short hydrogen bonds. Comparison of structures of trypsin, thrombin and uPA, each bound by the same inhibitor, highlights important structural variations in the S1 and active sites accessible for engineering notable selectivity into remarkably small molecules with low nanomolar K(i) values.