Unraveling viral drug targets: a deep learning-based approach for the identification of potential binding sites.

The coronavirus disease 2019 (COVID-19) pandemic has spurred a wide range of approaches to control and combat the disease. However, selecting an effective antiviral drug target remains a time-consuming challenge. Computational methods offer a promising solution by efficiently reducing the number of candidates. In this study, we propose a structure- and deep learning-based approach that identifies vulnerable regions in viral proteins corresponding to drug binding sites. Our approach takes into account the protein dynamics, accessibility and mutability of the binding site and the putative mechanism of action of the drug. We applied this technique to validate drug targeting toward severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) spike glycoprotein S. Our findings reveal a conformation- and oligomer-specific glycan-free binding site proximal to the receptor binding domain. This site comprises topologically important amino acid residues. Molecular dynamics simulations of Spike in complex with candidate drug molecules bound to the potential binding sites indicate an equilibrium shifted toward the inactive conformation compared with drug-free simulations. Small molecules targeting this binding site have the potential to prevent the closed-to-open conformational transition of Spike, thereby allosterically inhibiting its interaction with human angiotensin-converting enzyme 2 receptor. Using a pseudotyped virus-based assay with a SARS-CoV-2 neutralizing antibody, we identified a set of hit compounds that exhibited inhibition at micromolar concentrations.

Intermolecular Interactions in G Protein-Coupled Receptor Allosteric Sites at the Membrane Interface from Molecular Dynamics Simulations and Quantum Chemical Calculations.

Allosteric modulators are called promising candidates in G protein-coupled receptor (GPCR) drug development by displaying subtype selectivity and more specific receptor modulation. Among the allosteric sites known to date, cavities at the receptor-lipid interface represent an uncharacteristic binding location that raises many questions about the ligand interactions and stability, the binding site structure, and how all of these are affected by lipid molecules. In this work, we analyze interactions in the allosteric sites of the PAR2, C5aR1, and GCGR receptors in three lipid compositions using molecular dynamics simulations. In addition, we performed quantum chemical calculations involving the symmetry-adapted perturbation theory (SAPT) and the natural population analysis to quantify the strength of intermolecular interactions. We show that besides classical hydrogen bonds, weak polar interactions such as O-HC, O-Br, and long-range electrostatics with the backbone amides contribute to the stability of allosteric modulators at the receptor-lipid interface. The allosteric cavities are detectable in various membrane compositions. The availability of polar atoms for interactions in such cavities can be assessed by water molecules from simulations. Although ligand-lipid interactions are weak, lipid tails play a role in ligand binding pose stability and the size of allosteric cavities. We discuss physicochemical aspects of ligand binding at the receptor-lipid interface and suggest a compound library enriched by weak donor groups for ligand search in such sites.

A Facile Approach to Bis(isoxazoles), Promising Ligands of the AMPA Receptor.

A convenient synthetic approach to novel functionalized bis(isoxazoles), the promising bivalent ligands of the AMPA receptor, was elaborated. It was based on the heterocyclization reactions of readily available electrophilic alkenes with the tetranitromethane-triethylamine complex. The structural diversity of the synthesized compounds was demonstrated. In the electrophysiological experiments using the patch clamp technique on Purkinje neurons, the compound 1,4-phenylenedi(methylene)bis(5-aminoisoxazole-3-carboxylate) was shown to be highly potent positive modulator of the AMPA receptor, potentiating kainate-induced currents up to 70% at 10-11 M.

Imidazolidine-2,4,5- and pirimidine-2,4,6-triones - New primary pharmacophore for soluble epoxide hydrolase inhibitors with enhanced water solubility.

A series of inhibitors of the soluble epoxide hydrolase (sEH) containing imidazolidine-2,4,5-trione or pirimidine-2,4,6-trione has been synthesized. Inhibition potency of the described compounds ranges from 8.4 µM to 0.4 nM. The tested compounds possess higher water solubility than their preceding ureas. Molecular docking indicates new bond between the triones and the active site of sEH that in part explain the observed potency of the new pharmacophores. While less potent than the corresponding ureas, the modifications of urea group reported herein yield compounds with higher water solubility, thus permitting easier formulation.

Bioisosteric substitution of adamantane with bicyclic lipophilic groups improves water solubility of human soluble epoxide hydrolase inhibitors.

A series of inhibitors of the soluble epoxide hydrolase (sEH) containing lipophilic groups of natural origin (camphanyl, norcamphanyl, furan-2-yl) were developed. Inhibitory potency ranging from 0.4 nM to 2.16 µM were obtained. While having the same level of inhibitory activity bicyclic ureas are up to 10-fold more soluble than the corresponding ureas containing adamantyl or 4-trifluoromethoxyphenyl substituents. This makes them easier to formulate, more bioavailable and thus more promising as therapeutic sEH inhibitors. Endo/exo-form of compound 2b derived from l-camphor is 14-fold more potent than the corresponding analogue derived from d-camphor (IC50 = 3.7 nM vs. 50.6 nM) indicating enantiomeric preference.

Baseline Model for Predicting Protein-Ligand Unbinding Kinetics through Machine Learning.

Derivation of structure-kinetics relationships can help rational design and development of new small-molecule drug candidates with desired residence times. Efforts are now being directed toward the development of efficient computational methods. Currently, there is a lack of solid, high-throughput binding kinetics prediction approaches on bigger datasets. We present a prediction method for binding kinetics based on the machine learning analysis of protein-ligand structural features, which can serve as a baseline for more sophisticated methods utilizing molecular dynamics (MD). We showed that the random forest algorithm is capable of learning the protein binding site secondary structure and backbone/side-chain features to predict the binding kinetics of protein-ligand complexes but still with inferior performance to that of MD-based descriptor analysis. MD simulations had been applied to a limited number of targets and a series of ligands in terms of kinetics analysis, and we believe that the developed approach may guide new studies. The method was trained on a newly curated database of 501 protein-ligand unbinding rate constants, which can also be used for testing and training the binding kinetics prediction models.

Comparative Study of Multitask Toxicity Modeling on a Broad Chemical Space.

Acute toxicity is one of the most challenging properties to predict purely with computational methods due to its direct relationship to biological interactions. Moreover, toxicity can be represented by different end points: it can be measured for different species using different types of administration, etc., and it is questionable if the knowledge transfer between end points is possible. We performed a comparative study of prediction multitask toxicity for a broad chemical space using different descriptors and modeling algorithms and applied multitask learning for a large toxicity data set extracted from the Registry of Toxic Effects of Chemical Substances (RTECS). We demonstrated that multitask modeling provides significant improvement over single-output models and other machine learning methods. Our research reveals that multitask learning can be very useful to improve the quality of acute toxicity modeling and raises a discussion about the usage of multitask approaches for regulation purposes. Our MultiTox models are freely available in OCHEM platform ( ochem.eu/multitox ) under CC-BY-NC license.

Adamantyl thioureas as soluble epoxide hydrolase inhibitors.

A series of inhibitors of the soluble epoxide hydrolase (sEH) containing one or two thiourea groups has been developed. Inhibition potency of the described compounds ranges from 50 µM to 7.2 nM. 1,7-(Heptamethylene)bis[(adamant-1-yl)thiourea] (6f) was found to be the most potent sEH inhibitor, among the thioureas tested. The inhibitory activity of the thioureas against the human sEH is closer to the value of activity against rat sEH rather than murine sEH. While being less active, thioureas are up to 7-fold more soluble than ureas, which makes them more bioavailable and thus promising as sEH inhibitors.

Characterization of the mechanism of bile salt hydrolase substrate specificity by experimental and computational analyses.

Bile salt hydrolases (BSHs) are currently being investigated as target enzymes for metabolic regulators in humans and as growth promoters in farm animals. Understanding structural features underlying substrate specificity is necessary for inhibitor design. Here, we used a multidisciplinary workflow including mass spectrometry, mutagenesis, molecular dynamic simulations, machine learning, and crystallography to demonstrate substrate specificity in Lactobacillus salivarius BSH, the most abundant enzyme in human and farm animal intestines. We show the preference of substrates with a taurine head and a dehydroxylated sterol ring for hydrolysis. A regression model that correlates the relative rates of hydrolysis of various substrates in various enzyme mutants with the residue-substrate interaction energies guided the identification of structural determinants of substrate binding and specificity. In addition, we found T208 from another BSH protomer regulating the hydrolysis. The designed workflow can be used for fast and comprehensive characterization of enzymes with a broad range of substrates.

Ureas derived from camphor and fenchone reveal enantiomeric preference of human soluble epoxide hydrolase.

The soluble epoxide hydrolase (sEH) is a potential target to treat cardiovascular, renal and neuronal diseases. A series of sEH inhibitors containing naturally occurring lipophilic groups (originating from camphor and fenchone) were developed. Inhibitory potency ranging from 0.7 nM to 6.47 µM was obtained. It was discovered that ureas derived from L-camphor were more active against sEH (2.3-fold average) than the corresponding analogues derived from D-camphor indicating enantiomeric preference of sEH. Ureas derived from fenchone possess lower activity against sEH (ca. 80-fold on average) than their camphor-derived analogs due to the specific structure of the lipophilic fragment and show less enantiomeric preference (1.75-fold on average). Moreover, fenchone-derived ureas show no consistency in enantiomeric preference. Endo/exo-form of compound L-3a derived from L-camphor is 4-fold more potent than the corresponding analogue prepared from D-camphor (IC50 = 0.7 nM vs. 2.8 nM) making it the most promising sEH inhibitor among the tested series.

Biphenyl scaffold for the design of NMDA-receptor negative modulators: molecular modeling, synthesis, and biological activity.

NMDA (N-methyl-d-aspartate) receptor antagonists are promising tools for the treatment of a wide variety of central nervous system impairments including major depressive disorder. We present here the activity optimization process of a biphenyl-based NMDA negative allosteric modulator (NAM) guided by free energy calculations, which led to a 100 times activity improvement (IC50 = 50 nM) compared to a hit compound identified in virtual screening. Preliminary calculation results suggest a low affinity for the human ether-a-go-go-related gene ion channel (hERG), a high affinity for which was earlier one of the main obstacles for the development of first-generation NMDA-receptor negative allosteric modulators. The docking study and the molecular dynamics calculations suggest a completely different binding mode (ifenprodil-like) compared to another biaryl-based NMDA NAM EVT-101.

Probe Confined Dynamic Mapping for G Protein-Coupled Receptor Allosteric Site Prediction.

Targeting G protein-coupled receptors (GPCRs) through allosteric sites offers advantages over orthosteric sites in identifying drugs with increased selectivity and potentially reduced side effects. In this study, we developed a probe confined dynamic mapping protocol that allows the prediction of allosteric sites at both the GPCR extracellular and intracellular sides, as well as at the receptor-lipid interface. The applied harmonic wall potential enhanced sampling of probe molecules in a selected area of a GPCR while preventing membrane distortion in molecular dynamics simulations. The specific probes derived from GPCR allosteric ligand structures performed better in allosteric site mapping compared to commonly used cosolvents. The M2 muscarinic, ß2 adrenergic, and P2Y1 purinergic receptors were selected for the protocol's retrospective validation. The protocol was next validated prospectively to locate the binding site of [5-fluoro-4-(hydroxymethyl)-2-methoxyphenyl]-(4-fluoro-1H-indol-1-yl)methanone at the D2 dopamine receptor, and subsequent mutagenesis confirmed the prediction. The protocol provides fast and efficient prediction of key amino acid residues surrounding allosteric sites in membrane proteins and facilitates the structure-based design of allosteric modulators.

graphDelta: MPNN Scoring Function for the Affinity Prediction of Protein-Ligand Complexes.

In this work, we present graph-convolutional neural networks for the prediction of binding constants of protein-ligand complexes. We derived the model using multi task learning, where the target variables are the dissociation constant (K d), inhibition constant (K i), and half maximal inhibitory concentration (IC50). Being rigorously trained on the PDBbind dataset, the model achieves the Pearson correlation coefficient of 0.87 and the RMSE value of 1.05 in pK units, outperforming recently developed 3D convolutional neural network model K deep.

Novel Positive Allosteric Modulators of AMPA Receptors Based on 3,7-Diazabicyclo[3.3.1]nonane Scaffold.

A series of new positive allosteric modulators (PAMs) of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors based on 3,7-diazabicyclo[3.3.1]nonane scaffold have been designed, synthesized, and analyzed. In electrophysiological patch clamp studies, several compounds have demonstrated a sub-nanomolar potency. Compound 4 in in vivo tests showed anti-amnestic properties in the scopolamine-induced model of amnesia in the step-through passive avoidance or maximal electroshock experiments in rats at 0.01 mg/kg showing a significant "dose-response" advantage over memantine. Based on the analysis of the flexible docking results of PAMs, the cyclothiazide-like mechanism of binding mode was suggested as the major site for the interaction with AMPA receptors.

Chemical space exploration guided by deep neural networks.

A parametric t-SNE approach based on deep feed-forward neural networks was applied to the chemical space visualization problem. It is able to retain more information than certain dimensionality reduction techniques used for this purpose (principal component analysis (PCA), multidimensional scaling (MDS)). The applicability of this method to some chemical space navigation tasks (activity cliffs and activity landscapes identification) is discussed. We created a simple web tool to illustrate our work (http://space.syntelly.com).

Computational characterization of the glutamate receptor antagonist perampanel and its close analogs: density functional exploration of conformational space and molecular docking study.

Perampanel approved by FDA in 2012 is a first-in-class antiepileptic drug which inhibits α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor currents. It is markedly more active than many of its close analogs, and the reasons for this activity difference are not quite clear. Recent crystallographic studies allowed the authors to identify the location of its binding site. Unfortunately, the resolution is low, and the detailed description of perampanel binding mode is still in part speculative. Here we provide a detailed DFT-level conformational analysis of perampanel in a vacuum and in the solvents, mimicking the protein environment, followed by quantum theory of atoms in molecules (QTAIM), non-covalent interactions (NCI), and natural bond orbital (NBO) analyses. The findings indicate the electrostatic nature of the intramolecular interactions which contribute to energy differences of the conformations in a vacuum whereas the increase of dielectric constant leads to the energy equalization of conformations. Based on these results, the docking study was performed to investigate possible binding modes of perampanel and its close analogs in AMPA receptors. The influence of the pyridine nitrogen and cyano group position was explained based on the results of conformational analysis and molecular docking. These findings may contribute to the design of novel antiepileptic drugs and the development of novel approaches to treat neurodegenerative diseases and major depressive disorder.

Bivalent AMPA receptor positive allosteric modulators of the bis(pyrimidine) series.

The first example of a novel class of AMPA receptor positive allosteric modulators of the bis(pyrimidine) series having a hydroquinone linker has been obtained and showed a potency to increase kainate-induced currents at subnanomolar concentrations.

Cytokinin activity of N6-benzyladenine derivatives assayed by interaction with the receptors in planta, in vitro, and in silico.

Biological effects of hormones in both plants and animals are based on high-affinity interaction with cognate receptors resulting in their activation. The signal of cytokinins, classical plant hormones, is perceived in Arabidopsis by three homologous membrane receptors: AHK2, AHK3, and CRE1/AHK4. To study the cytokinin-receptor interaction, we used 25 derivatives of potent cytokinin N6-benzyladenine (BA) with substituents in the purine heterocycle and/or in the side chain. The study was focused primarily on individual cytokinin receptors from Arabidopsis. The main in planta assay system was based on Arabidopsis double mutants retaining only one isoform of cytokinin receptors and harboring cytokinin-sensitive reporter gene. Classical cytokinin biotest with Amaranthus seedlings was used as an additional biotest. In parallel, the binding of ligands to individual cytokinin receptors was assessed in the in vitro test system. Quantitative comparison of results of different assays confirmed the partial similarity of ligand-binding properties of receptor isoforms. Substituents at positions 8 and 9 of adenine moiety, elongated linker up to 4 methylene units, and replacement of N6 by sulfur or oxygen have resulted in the suppression of cytokinin activity of the derivative toward all receptors. Introduction of a halogen into position 2 of adenine moiety, on the contrary, often increased the ligand activity, especially toward AHK3. Features both common and distinctive of cytokinin receptors in Arabidopsis and Amaranthus were revealed, highlighting species specificity of the cytokinin perception apparatus. Correlations between the extent to which a compound binds to a receptor in vitro and its ability to activate the same receptor in planta were evaluated for each AHK protein. Interaction patterns between individual receptors and ligands were rationalized by structure analysis and molecular docking in sensory modules of AHK receptors. The best correlation between docking scores and specific binding was observed for AHK3. In addition, receptor-specific ligands have been discovered with unique properties to predominantly activate or block distinct cytokinin receptors. These ligands are promising for practical application and as molecular tools in the study of the cytokinin perception by plant cells.