Design and quantitative structure-activity relationship of 3-amidinobenzyl-1H-indole-2-carboxamides as potent, nonchiral, and selective inhibitors of blood coagulation factor Xa

Dose-Response Activity-Based DNA-Encoded Library Screening

Dose-response, or "conforming" behavior, increases confidence in a screening hit's authenticity. Here, we demonstrate dose-response solid-phase DNA-encoded library (DEL) screening. Compound dose in microfluidic droplets is modulated via the UV intensity of photocleavage from DEL beads. A 55,296-member DEL was screened at different UV intensities against model enzyme drug targets factor Xa (FXa) and autotaxin (ATX). Both screens yielded photochemical dose-dependent hit rates (FXa hit rates of 0.08/0.05% at 100/30% UV exposure; ATX hit rates of 0.24/0.08% at 100/20% UV exposure). FXa hits contained structures reflective of FXa inhibitors and four hits inhibited FXa (IC50 = 4.2 ± 0.1, 7.4 ± 0.3, 9.0 ± 0.3, and 19 ± 2 μM.) The top ATX hits (two dihydrobenzamidazolones and a tetrahydroisoquinoline) were validated as inhibitors (IC50 = 7 ± 2, 13 ± 2, and 1 ± 0.3 μM). Photochemical dose-response DEL screening data prioritized hits for synthesis, the rate-limiting step in DEL lead identification.

A comparison between 2D and 3D descriptors in QSAR modeling based on bio-active conformations

QSAR models are widely and successfully used in many research areas. The success of such models highly depends on molecular descriptors typically classified as 1D, 2D, 3D, or 4D. While 3D information is likely important, e. g., for modeling ligand-protein binding, previous comparisons between the performances of 2D and 3D descriptors were inconclusive. Yet in such comparisons the modeled ligands were not necessarily represented by their bioactive conformations. With this in mind, we mined the PDB for sets of protein-ligand complexes sharing the same protein for which uniform activity data were reported. The results, totaling 461 structures spread across six series were compiled into a carefully curated, first of its kind dataset in which each ligand is represented by its bioactive conformation. Next, each set was characterized by 2D, 3D and 2D + 3D descriptors and modeled using three machine learning algorithms, namely, k-Nearest Neighbors, Random Forest and Lasso Regression. Models' performances were evaluated on external test sets derived from the parent datasets either randomly or in a rational manner. We found that many more significant models were obtained when combining 2D and 3D descriptors. We attribute these improvements to the ability of 2D and 3D descriptors to code for different, yet complementary molecular properties.

Probing Factor Xa Protein-Ligand Interactions: Accurate Free Energy Calculations and Experimental Validations of Two Series of High-Affinity Ligands

The accurate prediction of protein-ligand binding affinity belongs to one of the central goals in computer-based drug design. Molecular dynamics (MD)-based free energy calculations have become increasingly popular in this respect due to their accuracy and solid theoretical basis. Here, we present a combined study which encompasses experimental and computational studies on two series of factor Xa ligands, which enclose a broad chemical space including large modifications of the central scaffold. Using this integrated approach, we identified several new ligands with different heterocyclic scaffolds different from the previously identified indole-2-carboxamides that show superior or similar affinity. Furthermore, the so far underexplored terminal alkyne moiety proved to be a suitable non-classical bioisosteric replacement for the higher halogen-π aryl interactions. With this challenging example, we demonstrated the ability of the MD-based non-equilibrium free energy calculation approach for guiding crucial modifications in the lead optimization process, such as scaffold replacement and single-site modifications at molecular interaction hot spots.

Interpretation of Structure-Activity Relationships in Real-World Drug Design Data Sets Using Explainable Artificial Intelligence

In silico models based on Deep Neural Networks (DNNs) are promising for predicting activities and properties of new molecules. Unfortunately, their inherent black-box character hinders our understanding, as to which structural features are important for activity. However, this information is crucial for capturing the underlying structure-activity relationships (SARs) to guide further optimization. To address this interpretation gap, "Explainable Artificial Intelligence" (XAI) methods recently became popular. Herein, we apply and compare multiple XAI methods to projects of lead optimization data sets with well-established SARs and available X-ray crystal structures. As we can show, easily understandable and comprehensive interpretations are obtained by combining DNN models with some powerful interpretation methods. In particular, SHAP-based methods are promising for this task. A novel visualization scheme using atom-based heatmaps provides useful insights into the underlying SAR. It is important to note that all interpretations are only meaningful in the context of the underlying models and associated data.

Structural Characterization of Haemophilus influenzae Enolase and Its Interaction with Human Plasminogen by In Silico and In Vitro Assays

Haemophilus influenzae is the causal agent of invasive pediatric diseases, such as meningitis, epiglottitis, pneumonia, septic arthritis, pericarditis, cellulitis, and bacteremia (serotype b). Non-typeable H. influenzae (NTHi) strains are associated with localized infections, such as otitis media, conjunctivitis, sinusitis, bronchitis, and pneumonia, and can cause invasive diseases, such as as meningitis and sepsis in immunocompromised hosts. Enolase is a multifunctional protein and can act as a receptor for plasminogen, promoting its activation to plasmin, which leads to the degradation of components of the extracellular matrix, favoring host tissue invasion. In this study, using molecular docking, three important residues involved in plasminogen interaction through the plasminogen-binding motif (₂₅₁EFYNKENGMYE₂₆₂) were identified in non-typeable H. influenzae enolase (NTHiENO). Interaction with the human plasminogen kringle domains is conformationally stable due to the formation of four hydrogen bonds corresponding to enoTYR₂₅₃-plgGLU₁ (K2), enoTYR₂₅₃-plgGLY₃₁₀ (K3), and enoLYS₂₅₅-plgARG₄₇₁/enoGLU₂₅₁-plgLYS₄₆₈ (K5). On the other hand, in vitro assays, such as ELISA and far-western blot, showed that NTHiENO is a plasminogen-binding protein. The inhibition of this interaction using polyclonal anti-NTHiENO antibodies was significant. With these results, we can propose that NTHiENO–plasminogen interaction could be one of the mechanisms used by H. influenzae to adhere to and invade host cells.

Application of Deep Neural Network Models in Drug Discovery Programs

In silico driven optimization of compound properties related to pharmacokinetics, pharmacodynamics, and safety is a key requirement in modern drug discovery. Nowadays, large and harmonized datasets allow to implement deep neural networks (DNNs) as a framework for leveraging predictive models. Nevertheless, various available model architectures differ in their global applicability and performance in lead optimization projects, such as stability over time and interpretability of the results. Here, we describe and compare the value of established DNN-based methods for the prediction of key ADME property trends and biological activity in an industrial drug discovery environment, represented by microsomal lability, CYP3A4 inhibition and factor Xa inhibition. Three architectures are exemplified, our earlier described multilayer perceptron approach (MLP), graph convolutional network-based models (GCN) and a vector representation approach, Mol2Vec. From a statistical perspective, MLP and GCN were found to perform superior over Mol2Vec, when applied to external validation sets. Interestingly, GCN-based predictions are most stable over a longer period in a time series validation study. Apart from those statistical observations, DNN prove of value to guide local SAR. To illustrate this important aspect in pharmaceutical research projects, we discuss challenging applications in medicinal chemistry towards a more realistic picture of artificial intelligence in drug discovery.

Sensitivity of Binding Free Energy Calculations to Initial Protein Crystal Structure

Binding free energy calculations using alchemical free energy (AFE) methods are widely considered to be the most rigorous tool in the computational drug discovery arsenal. Despite this, the calculations suffer from accuracy, precision, and reproducibility issues. In this publication, we perform a high-throughput study of more than a thousand AFE calculations, utilizing over 220 μs of total sampling time, on three different protein systems to investigate the impact of the initial crystal structure on the resulting binding free energy values. We also consider the influence of equilibration time and discover that the initial crystal structure can have a significant effect on free energy values obtained at short timescales that can manifest itself as a free energy difference of more than 1 kcal/mol. At longer timescales, these differences are largely overtaken by important rare events, such as torsional ligand motions, typically resulting in a much higher uncertainty in the obtained values. This work emphasizes the importance of rare event sampling and long-timescale dynamics in free energy calculations even for routinely performed alchemical perturbations. We conclude that an optimal protocol should not only concentrate computational resources on achieving convergence in the alchemical coupling parameter (λ) space but also on longer simulations and multiple repeats.

Automated De Novo Design in Medicinal Chemistry: Which Types of Chemistry Does a Generative Neural Network Learn?

Artificial intelligence offers promising solutions for property prediction, compound design, and retrosynthetic planning, which are expected to significantly accelerate the search for pharmacologically relevant molecules. Here, we investigate aspects of artificial intelligence based de novo design pertaining to its integration into real-life workflows. First, different chemical spaces were used as training sets for reinforcement learning (RL) in combination with different reward functions. With the trained neuronal networks different biologically active molecules could be regenerated. Excluding molecules with substructures such as five-membered rings from training spaces nevertheless produced results containing these moieties. Furthermore, different scoring functions in RL were investigated and produced different design ensembles. In summary, some of these design proposals are close in chemical space to the query, thus supporting lead optimization, while 3D-shape or QSAR (quantitative structure-activity relationship) models produced significantly different proposals by sampling a broader region of the chemical space, thus supporting lead generation. Therefore, RL provides a good framework to tailored design approaches for different discovery phases.

Can System Truncation Speed up Ligand-Binding Calculations with Periodic Free-Energy Simulations?

We have investigated whether alchemical free-energy perturbation calculations of relative binding energies can be sped up by simulating a truncated protein. Previous studies with spherical nonperiodic systems showed that the number of simulated atoms could be reduced by a factor of 26 without affecting the calculated binding free energies by more than 0.5 kJ/mol on average ( Genheden, S.; Ryde, U. J. Chem. Theory Comput. 2012 , 8 , 1449 ), leading to a 63-fold decrease in the time consumption. However, such simulations are rather slow, owing to the need of a large cutoff radius for the nonbonded interactions. Periodic simulations with the electrostatics treated by Ewald summation are much faster. Therefore, we have investigated if a similar speed-up can be obtained also for periodic simulations. Unfortunately, our results show that it is harder to truncate periodic systems and that the truncation errors are larger for these systems. In particular, residues need to be removed from the calculations, which means that atoms have to be restrained to avoid that they move in an unrealistic manner. The results strongly depend on the strength on this restraint. For the binding of seven ligands to dihydrofolate reductase and ten inhibitors of blood-clotting factor Xa, the best results are obtained with a small restraining force constant. However, the truncation errors were still significant (e.g., 1.5-2.9 kJ/mol at a truncation radius of 10 Å). Moreover, the gain in computer time was only modest. On the other hand, if the snapshots are truncated after the MD simulations, the truncation errors are small (below 0.9 kJ/mol even for a truncation radius of 10 Å). This indicates that postprocessing with a more accurate energy function (e.g., with quantum chemistry) on truncated snapshots may be a viable approach.

Novel Anthranilamide-Based FXa Inhibitors: Drug Design, Synthesis and Biological Evaluation

Factor Xa (FXa) plays a significant role in the blood coagulation cascade and it has become a promising target for anticoagulation drugs. Three oral direct FXa inhibitors have been approved by the FDA for treating thrombotic diseases. By structure-activity relationship (SAR) analysis upon these FXa inhibitors, a series of novel anthranilamide-based FXa inhibitors were designed and synthesized. According to our study, compounds 1a, 1g and 1s displayed evident FXa inhibitory activity and excellent selectivity over thrombin in in vitro inhibition activities studies. Compounds 1g and 1s also exhibited pronounced anticoagulant activities in in vitro anticoagulant activity studies.

Synthesis of spiro isoindolinone-indolines and 1,2-disubstituted indoles from 2-iodobenzamide derivatives

Copper catalyzed C-terminal and N-terminal attack of 2-alkynylanilines to 2-iodobenzamides afforded isoindolinones and 1,2-disubstituted indoles, respectively.

Application of Molecular Modeling to Development of New Factor Xa Inhibitors

In consequence of the key role of factor Xa in the clotting cascade and absence of its activity in the processes that do not affect coagulation, this protein is an attractive target for development of new blood coagulation inhibitors. Factor Xa is more effective and convenient target for creation of anticoagulants than thrombin, inhibition of which may cause some side effects. This study is aimed at finding new inhibitors of factor Xa by molecular computer modeling including docking SOL and postdocking optimization DISCORE programs. After validation of molecular modeling methods on well-known factor Xa inhibitors the virtual screening of NCI Diversity and Voronezh State University databases of ready-made low molecular weight species has been carried out. Seventeen compounds selected on the basis of modeling results have been tested experimentally in vitro. It has been found that 12 of them showed activity against factor Xa (IC50 = 1.8-40 μM). Based on analysis of the results, the new original compound was synthesized and experimentally verified. It shows activity against factor Xa with IC50 value of 0.7 μM.

Computer-aided drug discovery

Computational approaches are an integral part of interdisciplinary drug discovery research. Understanding the science behind computational tools, their opportunities, and limitations is essential to make a true impact on drug discovery at different levels. If applied in a scientifically meaningful way, computational methods improve the ability to identify and evaluate potential drug molecules, but there remain weaknesses in the methods that preclude naïve applications. Herein, current trends in computer-aided drug discovery are reviewed, and selected computational areas are discussed. Approaches are highlighted that aid in the identification and optimization of new drug candidates. Emphasis is put on the presentation and discussion of computational concepts and methods, rather than case studies or application examples. As such, this contribution aims to provide an overview of the current methodological spectrum of computational drug discovery for a broad audience.

Strong nonadditivity as a key structure-activity relationship feature: distinguishing structural changes from assay artifacts

Nonadditivity in protein–ligand affinity data represents highly instructive structure–activity relationship (SAR) features that indicate structural changes and have the potential to guide rational drug design. At the same time, nonadditivity is a challenge for both basic SAR analysis as well as many ligand-based data analysis techniques such as Free-Wilson Analysis and Matched Molecular Pair analysis, since linear substituent contribution models inherently assume additivity and thus do not work in such cases. While structural causes for nonadditivity have been analyzed anecdotally, no systematic approaches to interpret and use nonadditivity prospectively have been developed yet. In this contribution, we lay the statistical framework for systematic analysis of nonadditivity in a SAR series. First, we develop a general metric to quantify nonadditivity. Then, we demonstrate the non-negligible impact of experimental uncertainty that creates apparent nonadditivity, and we introduce techniques to handle experimental uncertainty. Finally, we analyze public SAR data sets for strong nonadditivity and use recourse to the original publications and available X-ray structures to find structural explanations for the nonadditivity observed. We find that all cases of strong nonadditivity (ΔΔpK_i and ΔΔpIC₅₀ > 2.0 log units) with sufficient structural information to generate reasonable hypothesis involve changes in binding mode. With the appropriate statistical basis, nonadditivity analysis offers a variety of new attempts for various areas in computer-aided drug design, including the validation of scoring functions and free energy perturbation approaches, binding pocket classification, and novel features in SAR analysis tools.

A large-scale test of free-energy simulation estimates of protein-ligand binding affinities

We have performed a large-scale test of alchemical perturbation calculations with the Bennett acceptance-ratio (BAR) approach to estimate relative affinities for the binding of 107 ligands to 10 different proteins. Employing 20-Å truncated spherical systems and only one intermediate state in the perturbations, we obtain an error of less than 4 kJ/mol for 54% of the studied relative affinities and a precision of 0.5 kJ/mol on average. However, only four of the proteins gave acceptable errors, correlations, and rankings. The results could be improved by using nine intermediate states in the simulations or including the entire protein in the simulations using periodic boundary conditions. However, 27 of the calculated affinities still gave errors of more than 4 kJ/mol, and for three of the proteins the results were not satisfactory. This shows that the performance of BAR calculations depends on the target protein and that several transformations gave poor results owing to limitations in the molecular-mechanics force field or the restricted sampling possible within a reasonable simulation time. Still, the BAR results are better than docking calculations for most of the proteins.

Thermodynamics of Water in an Enzyme Active Site: Grid-Based Hydration Analysis of Coagulation Factor Xa

Water molecules in the active site of an enzyme occupy a complex, heterogeneous environment, and the thermodynamic properties of active-site water are functions of position. As a consequence, it is thought that an enzyme inhibitor can gain affinity by extending into a region occupied by unfavorable water or lose affinity by displacing water from a region where it was relatively stable. Recent advances in the characterization of binding-site water, based on the analysis of molecular simulations with explicit water molecules, have focused largely on simplified representations of water as occupying well-defined hydration sites. Our grid-based treatment of hydration, GIST, offers a more complete picture of the complex distributions of water properties, but it has not yet been applied to proteins. This first application of GIST to protein-ligand modeling, for the case of Coagulation Factor Xa, shows that ligand scoring functions based on GIST perform at least as well as scoring functions based on a hydration-site approach (HSA), when applied to exactly the same simulation data. Interestingly, the displacement of energetically unfavorable water emerges as the dominant factor in the fitted scoring functions, for both GIST and HSA methods, while water entropy plays a secondary role, at least in the present context.

Site-Identification by Ligand Competitive Saturation (SILCS) assisted pharmacophore modeling

Database screening using receptor-based pharmacophores is a computer-aided drug design technique that uses the structure of the target molecule (i.e. protein) to identify novel ligands that may bind to the target. Typically receptor-based pharmacophore modeling methods only consider a single or limited number of receptor conformations and map out the favorable binding patterns in vacuum or with a limited representation of the aqueous solvent environment, such that they may suffer from neglect of protein flexibility and desolvation effects. Site-Identification by Ligand Competitive Saturation (SILCS) is an approach that takes into account these, as well as other, properties to determine 3-dimensional maps of the functional group-binding patterns on a target receptor (i.e. FragMaps). In this study, a method to use the FragMaps to automatically generate receptor-based pharmacophore models is presented. It converts the FragMaps into SILCS pharmacophore features including aromatic, aliphatic, hydrogen-bond donor and acceptor chemical functionalities. The method generates multiple pharmacophore hypotheses that are then quantitatively ranked using SILCS grid free energies. The pharmacophore model generation protocol is validated using three different protein targets, including using the resulting models in virtual screening. Improved performance and efficiency of the SILCS derived pharmacophore models as compared to published docking studies, as well as a recently developed receptor-based pharmacophore modeling method is shown, indicating the potential utility of the approach in rational drug design.

5-Chlorothiophene-2-carboxylic acid [(S)-2-[2-methyl-3-(2-oxopyrrolidin-1-yl)benzenesulfonylamino]-3-(4-methylpiperazin-1-yl)-3-oxopropyl]amide (SAR107375), a selective and potent orally active dual thrombin and factor Xa inhibitor

Compound 15 (SAR107375), a novel potent dual thrombin and factor Xa inhibitor resulted from a rational optimization process. Starting from compound 14, with low factor Xa and modest anti-thrombin inhibitory activities (IC50's of 3.5 and 0.39 μM, respectively), both activities were considerably improved, notably through the incorporation of a neutral chlorothiophene P1 fragment and tuning of P2 and P3-P4 fragments. Final optimization of metabolic stability with microsomes led to the identification of 15, which displays strong activity in vitro vs factor Xa and thrombin (with Ki's of 1 and 8 nM, respectively). In addition 15 presents good selectivity versus related serine proteases (roughly 300-fold), including trypsin (1000-fold), and is very active (0.39 μM) in the thrombin generation time (TGT) coagulation assay in human platelet rich plasma (PRP). Potent in vivo activity in a rat model of venous thrombosis following iv and, more importantly, po administration was also observed (ED50 of 0.07 and 2.8 mg/kg, respectively). Bleeding liability was reduced in the rat wire coil model, more relevant to arterial thrombosis, with 15 (blood loss increase of 2-fold relative to the ED80 value) compared to rivaroxaban 2 and dabigatran etexilate 1a.

Inclusion of multiple fragment types in the site identification by ligand competitive saturation (SILCS) approach

The site identification by ligand competitive saturation (SILCS) method identifies the location and approximate affinities of small molecular fragments on a target macromolecular surface by performing molecular dynamics (MD) simulations of the target in an aqueous solution of small molecules representative of different chemical functional groups. In this study, we introduce a set of small molecules to map potential interactions made by neutral hydrogen bond donors and acceptors and charged donor and acceptor fragments in addition to nonpolar fragments. The affinity pattern is obtained in the form of discretized probability or, equivalently, free energy maps, called FragMaps, which can be visualized with the target surface. We performed SILCS simulations for four proteins for which structural and thermodynamic data is available for multiple diverse ligands. Good overlap is shown between high affinity regions identified by the FragMaps and the crystallographic positions of ligand functional groups with similar chemical functionality, thus demonstrating the validity of the qualitative information obtained from the simulations. To test the ability of FragMaps in providing quantitative predictions, we calculate the previously introduced ligand grid free energy (LGFE) metric and observe its correspondence with experimentally measured binding affinity. LGFE is computed for different conformational ensembles and improvement in prediction is shown with increasing ligand conformational sampling. Ensemble generation includes a Monte Carlo sampling approach that uses the GFE FragMaps directly as the energy function. The results show that some but not all experimental trends are predicted and warrant improvements in the scoring methodology. In addition, the potential utility of atom-based free energy contributions to the LGFE scores and the use of multiple ligands in SILCS to identify displaceable water molecules during ligand design are discussed.

Structural role of exon-coded fragment of polypeptide chains in selected enzymes

This paper discusses the structural role of fragments encoded by individual exons in proteins. Selected enzymes (hydrolases, transferases, ligases) reveal the presence of at least one exon fragment whose contribution to the protein's hydrophobic core is in line with theoretical expectations. This phenomenon is confirmed by quantitative analysis of the hydrophobicity density distribution in protein molecules. Results are compared with a 3D Gaussian function, treated as an "idealized" distribution of hydrophobicity density, with the highest values observed near the center of the molecule and near-zero values on its surface. At least one accordant exon fragment has been identified in each of the proteins subjected to analysis. On the basis of these results the authors propose that accordant exons are responsible for tertiary structural stabilization of proteins by ensuring the generation of a stable hydrophobic core.

Lead optimization mapper: automating free energy calculations for lead optimization

Alchemical free energy calculations hold increasing promise as an aid to drug discovery efforts. However, applications of these techniques in discovery projects have been relatively few, partly because of the difficulty of planning and setting up calculations. Here, we introduce lead optimization mapper, LOMAP, an automated algorithm to plan efficient relative free energy calculations between potential ligands within a substantial library of perhaps hundreds of compounds. In this approach, ligands are first grouped by structural similarity primarily based on the size of a (loosely defined) maximal common substructure, and then calculations are planned within and between sets of structurally related compounds. An emphasis is placed on ensuring that relative free energies can be obtained between any pair of compounds without combining the results of too many different relative free energy calculations (to avoid accumulation of error) and by providing some redundancy to allow for the possibility of error and consistency checking and provide some insight into when results can be expected to be unreliable. The algorithm is discussed in detail and a Python implementation, based on both Schrödinger's and OpenEye's APIs, has been made available freely under the BSD license.

Application of Drug-Perturbed Essential Dynamics/Molecular Dynamics (ED/MD) to Virtual Screening and Rational Drug Design

We present here the first application of a new algorithm, essential dynamics/molecular dynamics (ED/MD), to the field of small molecule docking. The method uses a previously existing molecular dynamics (MD) ensemble of a protein or protein-drug complex to generate, with a very small computational cost, perturbed ensembles which represent ligand-induced binding site flexibility in a more accurate way than the original trajectory. The use of these perturbed ensembles in a standard docking program leads to superior performance than the same docking procedure using the crystal structure or ensembles obtained from conventional MD simulations as templates. The simplicity and accuracy of the method opens up the possibility of introducing protein flexibility in high-throughput docking experiments.

Potent direct inhibitors of factor Xa based on the tetrahydroisoquinoline scaffold

Direct inhibition of coagulation factor Xa (FXa) carries significant promise for developing effective and safe anticoagulants. Although a large number of FXa inhibitors have been studied, each can be classified as either possessing a highly flexible or a rigid core scaffold. We reasoned that an intermediate level of flexibility will provide high selectivity for FXa considering that its active site is less constrained in comparison to thrombin and more constrained as compared to trypsin. We studied several core scaffolds including 1,2,3,4-tetrahydroisoquinoline-3-carboxylic acid for direct FXa inhibition. Using a genetic algorithm-based docking and scoring approach, a promising candidate 23 was identified, synthesized, and found to inhibit FXa with a K(i) of 28 μM. Optimization of derivative 23 resulted in the design of a potent dicarboxamide 47, which displayed a K(i) of 135 nM. Dicarboxamide 47 displayed at least 1852-fold selectivity for FXa inhibition over other coagulation enzymes and doubled PT and aPTT of human plasma at 17.1 μM and 20.2 μM, respectively, which are comparable to those of clinically relevant agents. Dicarboxamide 47 is expected to serve as an excellent lead for further anticoagulant discovery.

The role of structural information in the discovery of direct thrombin and factor Xa inhibitors

The quest for novel medications to treat thromboembolic disorders such as venous thrombosis, pulmonary embolism and stroke received a boost when the 3D structures of two major players in the blood coagulation cascade were determined in 1989 and 1993. Structure-guided design of inhibitors of thrombin (factor IIa, fIIa) and factor Xa (fXa) eventually led to the discovery of potent, selective, efficacious, orally active and safe compounds that proved successful in clinical studies. In 2008, the direct thrombin inhibitor dabigatran etexilate developed by Boehringer Ingelheim became the first novel antithrombotic molecular entity to enter the market in 50 years. Additional compounds targeting factor Xa were subsequently granted marketing authorization or are in late-stage clinical studies. In this review, I use selected case studies to describe the discovery of novel fIIa and fXa inhibitors, with a particular emphasis on the pre-eminent role that structural information played in this process.

Protein pharmacophore selection using hydration-site analysis

Virtual screening using pharmacophore models is an efficient method to identify potential lead compounds for target proteins. Pharmacophore models based on protein structures are advantageous because a priori knowledge of active ligands is not required and the models are not biased by the chemical space of previously identified actives. However, in order to capture most potential interactions between all potentially binding ligands and the protein, the size of the pharmacophore model, i.e. number of pharmacophore elements, is typically quite large and therefore reduces the efficiency of pharmacophore based screening. We have developed a new method to select important pharmacophore elements using hydration-site information. The basic premise is that ligand functional groups that replace water molecules in the apo protein contribute strongly to the overall binding affinity of the ligand, due to the additional free energy gained from releasing the water molecule into the bulk solvent. We computed the free energy of water released from the binding site for each hydration site using thermodynamic analysis of molecular dynamics (MD) simulations. Pharmacophores which are colocalized with hydration sites with estimated favorable contributions to the free energy of binding are selected to generate a reduced pharmacophore model. We constructed reduced pharmacophore models for three protein systems and demonstrated good enrichment quality combined with high efficiency. The reduction in pharmacophore model size reduces the required screening time by a factor of 200-500 compared to using all protein pharmacophore elements. We also describe a training process using a small set of known actives to reliably select the optimal set of criteria for pharmacophore selection for each protein system.

Comparison of end-point continuum-solvation methods for the calculation of protein-ligand binding free energies

We have compared the predictions of ligand-binding affinities from several methods based on end-point molecular dynamics simulations and continuum solvation, that is, methods related to MM/PBSA (molecular mechanics combined with Poisson-Boltzmann and surface area solvation). Two continuum-solvation models were considered, viz., the Poisson-Boltzmann (PB) and generalised Born (GB) approaches. The nonelectrostatic energies were also obtained in two different ways, viz., either from the sum of the bonded, van der Waals, nonpolar solvation energies, and entropy terms (as in MM/PBSA), or from the scaled protein-ligand van der Waals interaction energy (as in the linear interaction energy approach, LIE). Three different approaches to calculate electrostatic energies were tested, viz., the sum of electrostatic interaction energies and polar solvation energies, obtained either from a single simulation of the complex or from three independent simulations of the complex, the free protein, and the free ligand, or the linear-response approximation (LRA). Moreover, we investigated the effect of scaling the electrostatic interactions by an effective internal dielectric constant of the protein (ϵ(int) ). All these methods were tested on the binding of seven biotin analogues to avidin and nine 3-amidinobenzyl-1H-indole-2-carboxamide inhibitors to factor Xa. For avidin, the best results were obtained with a combination of the LIE nonelectrostatic energies with the MM+GB electrostatic energies from a single simulation, using ϵ(int) = 4. For fXa, standard MM/GBSA, based on one simulation and using ϵ(int) = 4-10 gave the best result. The optimum internal dielectric constant seems to be slightly higher with PB than with GB solvation.

Binding affinities in the SAMPL3 trypsin and host-guest blind tests estimated with the MM/PBSA and LIE methods

We have estimated affinities for the binding of 34 ligands to trypsin and nine guest molecules to three different hosts in the SAMPL3 blind challenge, using the MM/PBSA, MM/GBSA, LIE, continuum LIE, and Glide score methods. For the trypsin challenge, none of the methods were able to accurately predict the experimental results. For the MM/GB(PB)SA and LIE methods, the rankings were essentially random and the mean absolute deviations were much worse than a null hypothesis giving the same affinity to all ligand. Glide scoring gave a Kendall's τ index better than random, but the ranking is still only mediocre, τ = 0.2. However, the range of affinities is small and most of the pairs of ligands have an experimental affinity difference that is not statistically significant. Removing those pairs improves the ranking metric to 0.4-1.0 for all methods except CLIE. Half of the trypsin ligands were non-binders according to the binding assay. The LIE methods could not separate the inactive ligands from the active ones better than a random guess, whereas MM/GBSA and MM/PBSA were slightly better than random (area under the receiver-operating-characteristic curve, AUC = 0.65-0.68), and Glide scoring was even better (AUC = 0.79). For the first host, MM/GBSA and MM/PBSA reproduce the experimental ranking fairly good, with τ = 0.6 and 0.5, respectively, whereas the Glide scoring was considerably worse, with a τ = 0.4, highlighting that the success of the methods is system-dependent.

Profiling the structural determinants for the selectivity of representative factor-Xa and thrombin inhibitors using combined ligand-based and structure-based approaches

The current study deciphers the combined ligand- and structure-based computational insights to profile structural determinants for the selectivity of representative diverse classes of FXa-selective and thrombin-selective as well as dual FXa-thrombin high affinity inhibitors. The thrombin-exclusive insertion 60-loop (D-pocket) was observed to be one of the most notable recognition sites for the known thrombin-selective inhibitors. Based on the topological comparison of four common active-site pockets (S1-S4) of FXa and thrombin, the greater structural disparity was observed in the S4-pocket, which was more symmetrical (U-shaped) in FXa as compared to thrombin mainly due to the presence of L99 and I174 residues in latter in place of Y99 and F174 respectively in former protease. The S2 pocket forming partial roof at the entry of 12 Å deep S1-pocket, with two extended β-sheets running antiparallel to each other by undergoing U-turn (∼180̊), has two conserved glycine residues forming H-bonds with the bound ligand for governing ligand binding affinity. The docking, scoring, and binding pose comparison of the representative high-affinity and selective inhibitors into the active sites of FXa and thrombin revealed critical residues (S214, Y99, W60D) mediating selectivity through direct- and long-range electrostatic interactions. Interestingly, most of the thrombin-selective inhibitors attained S-shaped conformation in thrombin, while FXa-selective inhibitors attained L-shaped conformations in FXa. The role of residue at 99th position of FXa and thrombin toward governing protease selectivity was further substantiated using molecular dynamics simulations on the wild-type and mutated Y99L FXa bound to thrombin-selective inhibitor 2. Furthermore, predictive CoMFA (FXa q² = 0.814; thrombin q² = 0.667) and CoMSIA (FXa q² = 0.807; thrombin q² = 0.624) models were developed and validated (FXa r²(test) = 0.823; thrombin r(2)(test) = 0.816) to feature molecular determinants of ligand binding affinity using the docking-based conformational alignments (DBCA) of 141 (88(train)+53(test)) and 39 (27(train)+11(test)) nonamidine class of potent FXa (0.004 ≤ K(i) (nM) ≤ 4700) and thrombin (0.001 ≤ K(i) (nM) ≤ 940) inhibitors, respectively. Interestingly, the ligand-based insights well corroborated with the structure-based insights in terms of the role of steric, electrostatic, and hydrophobic parameters for governing the selectivity for the two proteases. The new computational insights presented in this study are expected to be valuable for understanding and designing potent and selective antithrombotic agents.

Substituent diversity-directed synthesis of indole derivatives

This paper reports a versatile, good-yielding, solution-phase method that is a substituent diversity-directed synthesis of 1H-indoles (6-13, 17-20) and 1-hydroxyindoles (14, 15) starting from commercially available 1,5-difluoro-2,4-dinitrobenzene. The synthetic products possessed the maximum six diversity points.

Molecular dynamics simulations of Factor Xa: insight into conformational transition of its binding subsites

Protein flexibility and conformational diversity is well known to be a key characteristic of the function of many proteins. Human blood coagulation proteins have multiple substrates, and various protein-protein interactions are required for the smooth functioning of the coagulation cascade to maintain blood hemostasis. To address how a protein may cope with multiple interactions with its structurally diverse substrates and the accompanied structural changes that may drive these changes, we studied human Factor X. We employed 20 ns of molecular dynamics (MD) and steered molecular dynamics (SMD) simulations on two different conformational forms of Factor X, open and closed, and observed an interchangeable conformational transition from one to another. This work also demonstrates the roles of various aromatic residues involved in aromatic-aromatic interactions, which make this dynamic transition possible.

An improved method to predict the entropy term with the MM/PBSA approach

A method is suggested to calculate improved entropies within the MM/PBSA approach (molecular mechanics combined with Poisson-Boltzmann and surface area calculations) to estimate protein-ligand binding affinities. In the conventional approach, the protein is truncated outside ~8 A from the ligand. This system is freely minimised using a distance-dependent dielectric constant (to simulate the removed protein and solvent). However, this can lead to extensive changes in the molecular geometry, giving rise to a large standard deviation in this term. In our new approach, we introduce a buffer region approximately 4 A outside the truncated protein (including solvent molecules) and keep it fixed during the minimisation. Thereby, we reduce the standard deviation by a factor of 2-4, ensuring that the entropy term no longer limits the precision of the MM/PBSA predictions. The new method is tested for the binding of seven biotin analogues to avidin, eight amidinobenzyl-indole-carboxamide inhibitors to factor Xa, and two substrates to cytochrome P450 3A4 and 2C9. It is shown that it gives more stable results and often improved predictions of the relative binding affinities.

Role of the active-site solvent in the thermodynamics of factor Xa ligand binding

Understanding the underlying physics of the binding of small-molecule ligands to protein active sites is a key objective of computational chemistry and biology. It is widely believed that displacement of water molecules from the active site by the ligand is a principal (if not the dominant) source of binding free energy. Although continuum theories of hydration are routinely used to describe the contributions of the solvent to the binding affinity of the complex, it is still an unsettled question as to whether or not these continuum solvation theories describe the underlying molecular physics with sufficient accuracy to reliably rank the binding affinities of a set of ligands for a given protein. Here we develop a novel, computationally efficient descriptor of the contribution of the solvent to the binding free energy of a small molecule and its associated receptor that captures the effects of the ligand displacing the solvent from the protein active site with atomic detail. This descriptor quantitatively predicts (R(2) = 0.81) the binding free energy differences between congeneric ligand pairs for the test system factor Xa, elucidates physical properties of the active-site solvent that appear to be missing in most continuum theories of hydration, and identifies several features of the hydration of the factor Xa active site relevant to the structure-activity relationship of its inhibitors.