Towards more accurate pharmacophore modeling: Multicomplex-based comprehensive pharmacophore map and most-frequent-feature pharmacophore model of CDK2

Pharmacophore modeling, including ligand- and structure-based approaches, has become an important tool in drug discovery. However, the ligand-based method often strongly depends on the training set selection, and the structure-based pharmacophore model is usually created based on apo structures or a single protein-ligand complex, which might miss some important information. In this study, multicomplex-based method has been suggested to generate a comprehensive pharmacophore map of cyclin-dependent kinase 2 (CDK2) based on a collection of 124 crystal structures of human CDK2-inhibitor complex. Our multicomplex-based comprehensive pharmacophore map contains almost all the chemical features important for CDK2-inhibitor interactions. A comparison with previously reported ligand-based pharmacophores has revealed that the ligand-based models are just a subset of our comprehensive map. Furthermore, one most-frequent-feature pharmacophore model consisting of the most frequent pharmacophore features was constructed based on the statistical frequency information provided by the comprehensive map. Validations to the most-frequent-feature model show that it can not only successfully discriminate between known CDK2 inhibitors and the molecules of focused inactive dataset, but also is capable of correctly predicting the activities of a wide variety of CDK2 inhibitors in an external active dataset. Obviously, this investigation provides some new ideas about how to develop a multicomplex-based pharmacophore model that can be used in virtual screening to discover novel potential lead compounds.     

Identification of potent inhibitors of DNA methyltransferase 1 (DNMT1) through a pharmacophore-based virtual screening approach

DNA methylation is an epigenetic change that results in the addition of a methyl group at the carbon-5 position of cytosine residues. DNA methyltransferase (DNMT) inhibitors can suppress tumour growth and have significant therapeutic value. However, the established inhibitors are limited in their application due to their substantial cytotoxicity. Additionally, the standard drugs for DNMT inhibition are non-selective cytosine analogues with considerable cytotoxic side-effects. In the present study, we have designed a workflow by integrating various ligand-based and structure-based approaches to discover new agents active against DNMT1. We have derived a pharmacophore model with the help of available DNMT1 inhibitors. Utilising this model, we performed the virtual screening of Maybridge chemical library and the identified hits were then subsequently filtered based on the Naïve Bayesian classification model. The molecules that have returned from this classification model were subjected to ensemble based docking. We have selected 10 molecules for the biological assay by inspecting the interactions portrayed by these molecules. Three out of the ten tested compounds have shown DNMT1 inhibitory activity. These compounds were also found to demonstrate potential inhibition of cellular proliferation in human breast cancer MDA-MB-231 cells. In the present study, we have utilized a multi-step virtual screening protocol to identify inhibitors of DNMT1, which offers a starting point to develop more potent DNMT1 inhibitors as anti-cancer agents.     

Virtual screening against Mycobacterium tuberculosis dihydrofolate reductase: suggested workflow for compound prioritization using structure interaction fingerprints

In this study, we suggest a new workflow for the identification and prioritization of potential compounds targeted against Mycobacterium tuberculosis dihydrofolate reductase, an important folate cycle enzyme and a validated target for the development of anti-tubercular agents. First, we have performed an integrated pharmacophore and structure-based virtual screening using Maybridge small molecule database, subsequently interaction patterns from known actives to the receptor were applied for scoring and ranking the virtual screening hits using structure interaction fingerprint (SIFt)-based similarity approach. In addition, agglomerative hierarchical clustering of the structure interaction fingerprints permits the easy separation of active from inactive binding modes. Using this approach we screened 59275 Maybridge compounds and 20 compounds were prioritized as promising virtual screening hits. Though using a receptor interaction scoring approach, the results were not biased toward the chemical classes of the known actives and the proposed compounds were structurally diverse with low molecular weights and structural complexities. Our results suggest that structure-based virtual screening coupled with the SIFt should be a valuable tool for prioritization of virtual screening hits.     

A systematic methodology for large scale compound screening: A case study on the discovery of novel S1PL inhibitors

Decrease in sphingosine 1-phosphate (S1P) concentration induces migration of pathogenic T cells to the blood stream, disrupts the CNS and it is implicated in multiple sclerosis (MS), a progressive inflammatory disorder of the central nervous system (CNS), and Alzheimer’s disease (AD). A promising treatment alternative for MS and AD is inhibition of the activity of the microsomal enzyme sphingosine 1-phosphate lyase (S1PL), which degrades intracellular S1P. This report describes an integrated systematic approach comprising virtual screening, molecular docking, substructure search and molecular dynamics simulation to discover novel S1PL inhibitors. Virtual screening of the ZINC database via ligand-based and structure-based pharmacophore models yielded 10000 hits. After molecular docking, common substructures of the top ranking hits were identified. The ligand binding poses were optimized by induced fit docking. MD simulations were performed on the complex structures to determine the stability of the S1PL-ligand complex and to calculate the binding free energy. Selectivity of the selected molecules was examined by docking them to hERG and cytochrome P450 receptors. As a final outcome, 15 compounds from different chemotypes were proposed as potential S1PL inhibitors. These molecules may guide future medicinal chemistry efforts in the discovery of new compounds against the destructive action of pathogenic T cells.     

Pharmacophore modeling and virtual screening studies to design potential COMT inhibitors as new leads

Catechol-O-methyltransferase (COMT) catalyzes the methylation of catecholamines, including neurotransmitters like dopamine, epinephrine and norepinephrine, leading to their degradation. COMT has been a subject of study for its implications in numerous neurological disorders like Parkinson's disease (PD), schizophrenia, and depression. The COMT gene is associated with many allelic variants, the Val108Met polymorphism being the most clinically significant.Availability of crystal structure of both 108V and 108M forms of human soluble-COMT (S-COMT) facilitated us to use structure-based virtual screening approach to obtain new hits by screening a library of CNS permeable compounds from ZINC database. In this study, E-pharmacophore was also used to generate pharmacophore models based on a series of known COMT inhibitors. A five-point pharmacophore model consisting of one hydrogen-bond acceptor (A), two hydrogen bond donors (D), and two aromatic rings (R) was generated for both the polymorphic forms of COMT. These models were then used for filtering ZINC-CNS permeable library to obtain new hits. Physicochemical properties were also calculated for all the hits obtained from both the approaches for favorable ADME properties. These identified hits maybe of interest for further structural optimization and biological evaluation assays.     

Identification of p38α MAP kinase inhibitors by pharmacophore based virtual screening

The p38α mitogen-activated protein (MAP) kinase plays a vital role in treating many inflammatory diseases. In the present study, a combined ligand and structure based pharmacophore model was developed to identify potential DFG-in selective p38 MAP kinase inhibitors. Conformations of co-crystallised inhibitors were used in the development and validation of ligand and structure based pharmacophore modeling approached. The validated pharmacophore was utilized in database screening to identify potential hits. After Lipinski's rule of five filter and molecular docking analysis, nineteen hits were purchased and selected for in vitro analysis. The virtual hits exhibited promising activity against tumor necrosis factor-α (TNF-α) with 23–98% inhibition at 10 μM concentration. Out of these seven compounds has shown potent inhibitory activity against p38 MAP kinase with IC₅₀ values ranging from 12.97 to 223.5 nM. In addition, the toxicity study against HepG2 cells was also carried out to confirm the safety profile of identified virtual hits.     

A support vector machines approach for virtual screening of active compounds of single and multiple mechanisms from large libraries at an improved hit-rate and enrichment factor

Support vector machines (SVM) and other machine-learning (ML) methods have been explored as ligand-based virtual screening (VS) tools for facilitating lead discovery. While exhibiting good hit selection performance, in screening large compound libraries, these methods tend to produce lower hit-rate than those of the best performing VS tools, partly because their training-sets contain limited spectrum of inactive compounds. We tested whether the performance of SVM can be improved by using training-sets of diverse inactive compounds. In retrospective database screening of active compounds of single mechanism (HIV protease inhibitors, DHFR inhibitors, dopamine antagonists) and multiple mechanisms (CNS active agents) from large libraries of 2.986 million compounds, the yields, hit-rates, and enrichment factors of our SVM models are 52.4–78.0%, 4.7–73.8%, and 214–10,543, respectively, compared to those of 62–95%, 0.65–35%, and 20–1200 by structure-based VS and 55–81%, 0.2–0.7%, and 110–795 by other ligand-based VS tools in screening libraries of ≥1 million compounds. The hit-rates are comparable and the enrichment factors are substantially better than the best results of other VS tools. 24.3–87.6% of the predicted hits are outside the known hit families. SVM appears to be potentially useful for facilitating lead discovery in VS of large compound libraries.     

Generation and validation of the first predictive pharmacophore model for cyclin-dependent kinase 9 inhibitors

A three-dimensional (3D) pharmacophore modelling approach was applied to a diverse data set of known cyclin-dependent kinase 9 (CDK9) inhibitors. Diversity sampling and principal components analysis (PCA) were employed to ensure the rational selection of representative training sets. Twelve statistically robust pharmacophore models were generated using the HypoGen algorithm. The resulting models showed high homology and indicated great convergence in ascertaining pharmacophoric features essential for CDK9 inhibitory activity. One of the best models (Hypo 6) was assessed further by external predictive capability, randomization test, as well as its performance in virtual screening. The capability of the resulting models to reliably predict the inhibitory activity of external data sets and discriminate active structures from general databases would assist the identification and optimization of novel CDK9 inhibitors.     

A support vector machines approach for virtual screening of active compounds of single and multiple mechanisms from large libraries at an improved hit-rate and enrichment factor

Support vector machines (SVM) and other machine-learning (ML) methods have been explored as ligand-based virtual screening (VS) tools for facilitating lead discovery. While exhibiting good hit selection performance, in screening large compound libraries, these methods tend to produce lower hit-rate than those of the best performing VS tools, partly because their training-sets contain limited spectrum of inactive compounds. We tested whether the performance of SVM can be improved by using training-sets of diverse inactive compounds. In retrospective database screening of active compounds of single mechanism (HIV protease inhibitors, DHFR inhibitors, dopamine antagonists) and multiple mechanisms (CNS active agents) from large libraries of 2.986 million compounds, the yields, hit-rates, and enrichment factors of our SVM models are 52.4–78.0%, 4.7–73.8%, and 214–10,543, respectively, compared to those of 62–95%, 0.65–35%, and 20–1200 by structure-based VS and 55–81%, 0.2–0.7%, and 110–795 by other ligand-based VS tools in screening libraries of ≥1 million compounds. The hit-rates are comparable and the enrichment factors are substantially better than the best results of other VS tools. 24.3–87.6% of the predicted hits are outside the known hit families. SVM appears to be potentially useful for facilitating lead discovery in VS of large compound libraries.     

Chemical function-based pharmacophore development for novel, selective kappa opioid receptor agonists

In an effort to reduce or eliminate the centrally associated side effects produced by opioid analgesics there has been an interest in the preparation of peripherally acting opioid receptor agonists. These compounds would have very limited or no access to the central nervous system. As a first step towards developing peripheral kappa opioid receptor (KOP) agonists, we have developed a quantitatively predictive chemical function-based pharmacophore model of selective kappa opioid receptor agonists by using the HypoGen algorithm implemented in the Catalyst software. The input for HypoGen was a training set of 26 KOP agonists exhibiting K(i) values ranging between 0.015nM and 2300nM. The best output hypothesis consists of four features: one hydrophobic (HYD), one ring aromatic (RA), one hydrogen bond acceptor (HBA), and one positive ionizable (PI) function. The predictive power of the model could be demonstrated by internal and external validation of the generated hypothesis. The resulting Catalyst pharmacophore can be used concurrently for rapid virtual screening of chemical databases to identify novel, selective KOP agonists that may be easily restricted to target tissues by synthetic modification. It is anticipated that such an approach will lead to the generation of novel selective KOP agonists that are clinically useful for the treatment of pain through peripheral mechanisms.     

Development and screening of a polyketide virtual library for drug leads against a motilide pharmacophore

A virtual library of macrocyclic polyketide molecules was generated and screened to identify novel, conformationally constrained potential motilin receptor agonists ("motilides"). A motilide pharmacophore model was generated from the potent 6,9-enol ether erythromycin and known derivatives from the literature. The pharmacophore for each molecular conformation was a point in a distance-volume space based on presentation of the putative binding moieties. Two methods, one fragment based method and the other reaction based, were explored for constructing the polyketide virtual library. First, a virtual library was assembled from monomeric fragments using the CHORTLES language. Second, the virtual library was assembled by the in silico application of all possible polyketide synthase enzyme reactions to generate the product library. Each library was converted to low-energy 3D conformations by distance geometry and standard minimization methods. The distance-volume metric was calculated for low-energy conformations of the members of the virtual polyketide library and screened against the enol ether pharmacophore. The goal was to identify novel macrocycles that satisfy the pharmacophore. We identified three conformationally constrained, novel polyketide series that have low-energy conformations satisfying the distance-volume constraints of the motilide pharmacophore.     

In silico identification of novel inhibitors against Plasmodium falciparum dihydroorate dehydrogenase

Plasmodium falciparum causes the most fatal form of malaria and accounts for over 1 million deaths annually, yet currently used drug therapies are compromised by resistance. The malaria parasite cannot salvage pyrimidines and relies on de novo biosynthesis for survival. The enzyme dihydrooratate dehydrogenase (DHODH), a mitochondrial flavoenzyme, catalyzes the rate-limiting step of this pathway and is therefore an attractive anti-malarial chemotherapeutic target. In an effort to design new and potential anti-malarials, structure-based pharmacophore mapping, molecular docking, binding energy calculations and binding affinity predictions were employed in a virtual screening strategy to design new and potent P. falciparum dihydrooratate dehydrogenase (PfDHODH) inhibitors. A structure-based pharmacophore model was generated which consist of important interactions as observed in co-crystal of PfDHODH enzyme. The developed model was used to retrieve molecules from ChemBridge database, a freely available commercial database. A total of 87 molecules mapped on the modeled pharmacophore from the database. The retrieved hits were further screened by docking simulation, binding energy calculations and biding affinity predictions using genetic optimization for ligand docking (GOLD) and MOE. Based on these results, finally 26 chemo-types molecules were predicted as new, potential and structurally diverse PfDHODH inhibitors.     

Identification of novel PAD4 inhibitors based on a pharmacophore model derived from transition state coordinates

1.4 Protein arginine deiminases 4 (PAD4) is an attractive target for the development of novel and selective inhibitors of Rheumatoid Arthritis (RA). F-amidine is known as mechanism-based inhibitor targeting PAD4 and used as inactivators by covalently modifying the active site Cys645. To identify novel structural inhibitors of PAD4, we investigated the flexibility of protein on basis of the transition state geometry of PAD4 inhibited by F-amidine from our previous QM/MM calculation. And a pharmacophore model was generated containing four features (ADHH) using five representative structures from molecular dynamic (MD) simulation on basis of the transition state geometry of PAD4 inhibited by F-amidine. We performed virtual screening using the pharmacophore model and molecular docking methods, resulting in the discovery of two molecules with K_D (dissociation equilibrium constant) values of 112 μM and 218 μΜ against PAD4 through Surface Plasmon Resonance (SPR) experiments. These two molecules could potentially serve as PAD4 inhibitors.     

Multi-step virtual screening to develop selective DYRK1A inhibitors

Developing selective inhibitors for a particular kinase remains a major challenge in kinase-targeted drug discovery. Here we performed a multi-step virtual screening for dual-specificity tyrosine-phosphorylation-regulated kinase 1A (DYRK1A) inhibitors by focusing on the selectivity for DYRK1A over cyclin-dependent kinase 5 (CDK5). To examine the key factors contributing to the selectivity, we constructed logistic regression models to discriminate between actives and inactives for DYRK1A and CDK5, respectively, using residue-based binding free energies. The residue-based parameters were calculated by molecular mechanics-generalized Born surface area (MM-GBSA) decomposition methods for kinase–ligand complexes modeled by computer ligand docking. Based on the findings from the logistic regression models, we built a three-dimensional (3D) pharmacophore model and chose filter criteria for the multi-step virtual screening. The virtual hit compounds obtained from the screening were assessed for their inhibitory activities against DYRK1A and CDK5 by in vitro assay. Our screening identified two novel selective DYRK1A inhibitors with IC₅₀ values of several μM for DYRK1A and >100 μM for CDK5, which can be further optimized to develop more potent selective DYRK1A inhibitors.     

Selective pharmacophore design for alpha1-adrenoceptor subtypes

Alpha1-adrenoceptors are G-protein coupled receptors found in a variety of vascular tissues and responsible for vasoconstriction. Selectivity for each of the three subtypes is an important consideration in drug design in order to minimise the possibility of side effects. Using Catalyst we developed ligand-based pharmacophores from alpha(1a,b,d)-selective antagonists available in the literature using three separate training sets. Four-feature pharmacophores were developed for the alpha(1a) and alpha(1b) subtype-selective antagonists and a five-feature pharmacophore was developed for the alpha(1d) subtype-selective antagonists. The alpha(1a) pharmacophore represents both class I and II compounds with good predictivity for other compounds outside the training set as well. The alpha(1b) pharmacophore best predicts the activity of prazosin analogues as these make up the majority of alpha(1b)-selective antagonists. Unexpectedly, no positive ionisable feature was incorporated in the alpha(1b) pharmacophore. The alpha(1d) pharmacophore was based primarily on one structural class of compounds, but has good predictivity for a heterogeneous test set. Preliminary docking studies using AutoDock and optimised alpha1-adrenoceptor homology models, conducted with the antagonists prazosin (32) and 66, showed good agreement with the findings from the pharmacophores.     

Identification of non-resistant ROS-1 inhibitors using structure based pharmacophore analysis

Proto-oncogene receptor tyrosine kinase ROS-1 plays a key role in regulating a variety of cancers mainly non-small cell lung cancer (NSCLC). The marketed ROS-1 inhibitors such as Crizotinib suffer from the tribulations of growing resistance due to mutations primarily Gly2032Arg in the ROS-1 protein. To curb the problem of resistance, researchers have developed inhibitors such as Lorlatinib against the mutant protein. The present study was designed to identify inhibitors against wild type (WT) as well as mutant ROS-1 protein that will offer a broader spectrum of activity. Exploring crystal structure of ROS-1 complexed with Lorlatinib, receptor-ligand pharmacophore model was developed using Discovery Studio (DS) software. The developed pharmacophore model consisted of one hydrogen bond acceptor (HBA), one hydrogen bond donor (HBD) and two hydrophobic features (HY), subsequently utilized for virtual screening of commercially available databases and the retrieved hits were further subjected to fitness score and Lipinski’s filter. Thereafter, the retrieved hits were docked in WT and mutated (Gly2032Arg) proteins of ROS-1. Total five molecules were retrieved with good docking scores and good binding interactions within the active site of WT and mutated ROS-1. The binding energies of the ligand-receptor complexes were predicted via calculation of MM-GBSA score. To predict the stability of the ligand receptor complexes with mutant and wild type proteins, molecular dynamic simulation was performed. Thus, these identified hits show good binding affinities with WT and mutant ROS-1 proteins that may be further evaluated for their in-vitro/in-vivo activity.     

Pharmacophore modeling,virtual screening,docking and in silico ADMET analysis of protein kinase B (PKB β) inhibitors

Protein kinase B (PKB) is a key mediator of proliferation and survival pathways that are critical for cancer growth. Therefore, inhibitors of PKB are useful agents for the treatment of cancer. Herein, we describe pharmacophore-based virtual screening combined with docking study as a rational strategy for identification of novel hits or leads. Pharmacophore models of PKB β inhibitors were established using the DISCOtech and refined with GASP from compounds with IC₅₀ values ranging from 2.2 to 246 nM. The best pharmacophore model consists of one hydrogen bond acceptor (HBA), one hydrogen bond donor (HBD) site and two hydrophobic (HY) features. The pharmacophore models were validated through receiver operating characteristic (ROC) and Güner-Henry (GH) scoring methods indicated that the model-3 was statistically valuable and reliable in identifying PKB β inhibitors. Pharmacophore model as a 3D search query was searched against NCI database. Several compounds with different structures (scaffolds) were retrieved as hits. Molecules with a Q_fit value of more than 95 and three other known inhibitors were docked in the active site of PKB to further explore the binding mode of these compounds. Finally in silico pharmacokinetic and toxicities were predicted for active hit molecules. The hits reported here showed good potential to be PKB β inhibitors.     

First universal pharmacophore model for hERG1 K+ channel activators: acthER

The intra-cavitary drug blockade of hERG1 channel has been extensively studied, both experimentally and theoretically. Structurally diverse ligands inadvertently block the hERG1 K⁺ channel currents lead to drug induced Long QT Syndrome (LQTS). Accordingly, designing either hERG1 channel openers or current activators, with the potential to target other binding pockets of the channel, has been introduced as a viable approach in modern anti-arrhythmia drug development. However, reports and investigations on the molecular mechanisms underlying activators binding to the hERG1 channel remain sparse and the overall molecular design principles are largely unknown. Most of the hERG1 activators were discovered during mandatory screening for hERG1 blockade. To fill this apparent deficit, the first universal pharmacophore model for hERG1 K⁺ channel activators was developed using PHASE. 3D structures of 18 hERG1 K⁺ channel activators and their corresponding measured binding affinity values were used in the development of pharmacophore models. These compounds spanned a range of structurally different chemotypes with moderate variation in binding affinity. A five sites AAHRR (A, hydrogen-bond accepting, H, hydrophobic, R, aromatic) pharmacophore model has shown reasonable high statistical results compared to the other developed more than 1000 hypotheses. This model was used to construct steric and electrostatic contour maps. The predictive power of the model was tested with 3 external test set compounds as true unknowns. Finally, the pharmacophore model was combined with the previously developed receptor-based model of hERG1 K⁺ channel to develop and screen novel activators. The results are quite striking and it suggests a greater future role for pharmacophore modeling and virtual drug screening simulations in deciphering complex patterns of molecular mechanisms of hERG1 channel openers at the target sites. The developed model is available upon request and it may serve as basis for the synthesis of novel therapeutic hERG1 activators.     

表皮生长因子受体酪氨酸激酶抑制剂的药效团研究

根据一系列表皮生长因子受体酪氨酸激酶抑制剂的三维定量构效关系研究,得到了该类抑制剂的药效团,研究结果与Novartis的药效团模型相当类似。药效团包括一个氢键受体,一个氢键给体,一个疏水区和一个带有氯或溴原子的苯环。该药效团对于研究表皮生长因子受体酪氨酸激酶抑制剂结构与活性的关系具有重要的意义。通过三维数据库搜索可能会得到新的先导化合物。