GPCR-IPL score: multilevel featurization of GPCR-ligand interaction patterns and prediction of ligand functions from selectivity to biased activation.

G-protein-coupled receptors (GPCRs) mediate diverse cell signaling cascades after recognizing extracellular ligands. Despite the successful history of known GPCR drugs, a lack of mechanistic insight into GPCR challenges both the deorphanization of some GPCRs and optimization of the structure-activity relationship of their ligands. Notably, replacing a small substituent on a GPCR ligand can significantly alter extracellular GPCR-ligand interaction patterns and motion of transmembrane helices in turn to occur post-binding events of the ligand. In this study, we designed 3D multilevel features to describe the extracellular interaction patterns. Subsequently, these 3D features were utilized to predict the post-binding events that result from conformational dynamics from the extracellular to intracellular areas. To understand the adaptability of GPCR ligands, we collected the conformational information of flexible residues during binding and performed molecular featurization on a broad range of GPCR-ligand complexes. As a result, we developed GPCR-ligand interaction patterns, binding pockets, and ligand features as score (GPCR-IPL score) for predicting the functional selectivity of GPCR ligands (agonism versus antagonism), using the multilevel features of (1) zoomed-out 'residue level' (for flexible transmembrane helices of GPCRs), (2) zoomed-in 'pocket level' (for sophisticated mode of action) and (3) 'atom level' (for the conformational adaptability of GPCR ligands). GPCR-IPL score demonstrated reliable performance, achieving area under the receiver operating characteristic of 0.938 and area under the precision-recall curve of 0.907 (available in gpcr-ipl-score.onrender.com). Furthermore, we used the molecular features to predict the biased activation of downstream signaling (Gi/o, Gq/11, Gs and ß-arrestin) as well as the functional selectivity. The resulting models are interpreted and applied to out-of-set validation with three scenarios including the identification of a new MRGPRX antagonist.

HCovDock: an efficient docking method for modeling covalent protein-ligand interactions.

Covalent inhibitors have received extensive attentions in the past few decades because of their long residence time, high binding efficiency and strong selectivity. Therefore, it is valuable to develop computational tools like molecular docking for modeling of covalent protein-ligand interactions or screening of potential covalent drugs. Meeting the needs, we have proposed HCovDock, an efficient docking algorithm for covalent protein-ligand interactions by integrating a ligand sampling method of incremental construction and a scoring function with covalent bond-based energy. Tested on a benchmark containing 207 diverse protein-ligand complexes, HCovDock exhibits a significantly better performance than seven other state-of-the-art covalent docking programs (AutoDock, Cov_DOX, CovDock, FITTED, GOLD, ICM-Pro and MOE). With the criterion of ligand root-mean-squared distance < 2.0 Å, HCovDock obtains a high success rate of 70.5% and 93.2% in reproducing experimentally observed structures for top 1 and top 10 predictions. In addition, HCovDock is also validated in virtual screening against 10 receptors of three proteins. HCovDock is computationally efficient and the average running time for docking a ligand is only 5 min with as fast as 1 sec for ligands with one rotatable bond and about 18 min for ligands with 23 rotational bonds. HCovDock can be freely assessed at http://huanglab.phys.hust.edu.cn/hcovdock/.

ClusterX: a novel representation learning-based deep clustering framework for accurate visual inspection in virtual screening.

Molecular clustering analysis has been developed to facilitate visual inspection in the process of structure-based virtual screening. However, traditional methods based on molecular fingerprints or molecular descriptors limit the accuracy of selecting active hit compounds, which may be attributed to the lack of representations of receptor structural and protein-ligand interaction during the clustering. Here, a novel deep clustering framework named ClusterX is proposed to learn molecular representations of protein-ligand complexes and cluster the ligands. In ClusterX, the graph was used to represent the protein-ligand complex, and the joint optimisation can be used efficiently for learning the cluster-friendly features. Experiments on the KLIFs database show that the model can distinguish well between the binding modes of different kinase inhibitors. To validate the effectiveness of the model, the clustering results on the virtual screening dataset further demonstrated that ClusterX achieved better or more competitive performance against traditional methods, such as SIFt and extended connectivity fingerprints. This framework may provide a unique tool for clustering analysis and prove to assist computational medicinal chemists in visual decision-making.

Crystal structure of the Mycobacterium tuberculosis VirS regulator reveals its interaction with the lead compound SMARt751.

Ethionamide (ETO) is a prodrug that is primarily used as a second-line agent in the treatment of tuberculosis. Among the bacterial ETO activators, the monooxygenase MymA has been recently identified, and its expression is regulated by the mycobacterial regulator VirS. The discovery of VirS ligands that can enhance mymA expression and thereby increase the antimycobacterial efficacy of ETO, has led to the development of a novel therapeutic strategy against tuberculosis. This strategy involves the selection of preclinical candidates, including SMARt751. We report the first crystal structure of the AraC-like regulator VirS, in complex with SMARt751, refined at 1.69 Å resolution. Crystals were obtained via an in situ proteolysis method in the requisite presence of SMARt751. The elucidated structure corresponds to the ligand-binding domain of VirS, adopting an α/ß fold with structural similarities to H-NOX domains. Within the VirS structure, SMARt751 is situated in a completely enclosed hydrophobic cavity, where it forms hydrogen bonds with Asn11 and Asn149 as well as van der Waals contacts with various hydrophobic amino acids. Comprehensive structural comparisons within the AraC family of transcriptional regulators are conducted and analyzed to figure out the effects of the SMARt751 binding on the regulatory activity of VirS.

Identification of Ligands for Ion Channels: TRPM2.

Transient receptor potential melastatin 2 (TRPM2) is a calcium-permeable, nonselective cation channel with a widespread distribution throughout the body. It is involved in many pathological and physiological processes, making it a potential therapeutic target for various diseases, including Alzheimer's disease, Parkinson's disease, and cancers. New analytical techniques are beneficial for gaining a deeper understanding of its involvement in disease pathogenesis and for advancing the drug discovery for TRPM2-related diseases. In this work, we present the application of collision-induced affinity selection mass spectrometry (CIAS-MS) for the direct identification of ligands binding to TRPM2. CIAS-MS circumvents the need for high mass detection typically associated with mass spectrometry of large membrane proteins. Instead, it focuses on the detection of small molecules dissociated from the ligand-protein-detergent complexes. This affinity selection approach consolidates all affinity selection steps within the mass spectrometer, resulting in a streamlined process. We showed the direct identification of a known TRPM2 ligand dissociated from the protein-ligand complex. We demonstrated that CIAS-MS can identify binding ligands from complex mixtures of compounds and screened a compound library against TRPM2. We investigated the impact of voltage increments and ligand concentrations on the dissociation behavior of the binding ligand, revealing a dose-dependent relationship.

Pharmacology of Compounds Targeting Cation-Chloride Cotransporter Physiology.

Transporters of the solute carrier family 12 (SLC12) carry inorganic cations such as Na+ and/or K+ alongside Cl across the plasma membrane of cells. These tightly coupled, electroneutral, transporters are expressed in almost all tissues/organs in the body where they fulfil many critical functions. The family includes two key transporters participating in salt reabsorption in the kidney: the Na-K-2Cl cotransporter-2 (NKCC2), expressed in the loop of Henle, and the Na-Cl cotransporter (NCC), expressed in the distal convoluted tubule. NCC and NKCC2 are the targets of thiazides and "loop" diuretics, respectively, drugs that are widely used in clinical medicine to treat hypertension and edema. Bumetanide, in addition to its effect as a loop diuretic, has recently received increasing attention as a possible therapeutic agent for neurodevelopmental disorders. This chapter also describes how over the past two decades, the pharmacology of Na+ independent transporters has expanded significantly to provide novel tools for research. This work has indeed led to the identification of compounds that are 100-fold to 1000-fold more potent than furosemide, the first described inhibitor of K-Cl cotransport, and identified compounds that possibly directly stimulate the function of the K-Cl cotransporter. Finally, the recent cryo-electron microscopy revolution has begun providing answers as to where and how pharmacological agents bind to and affect the function of the transporters.

End-to-end protein-ligand complex structure generation with diffusion-based generative models.

BACKGROUND: Three-dimensional structures of protein-ligand complexes provide valuable insights into their interactions and are crucial for molecular biological studies and drug design. However, their high-dimensional and multimodal nature hinders end-to-end modeling, and earlier approaches depend inherently on existing protein structures. To overcome these limitations and expand the range of complexes that can be accurately modeled, it is necessary to develop efficient end-to-end methods. RESULTS: We introduce an equivariant diffusion-based generative model that learns the joint distribution of ligand and protein conformations conditioned on the molecular graph of a ligand and the sequence representation of a protein extracted from a pre-trained protein language model. Benchmark results show that this protein structure-free model is capable of generating diverse structures of protein-ligand complexes, including those with correct binding poses. Further analyses indicate that the proposed end-to-end approach is particularly effective when the ligand-bound protein structure is not available. CONCLUSION: The present results demonstrate the effectiveness and generative capability of our end-to-end complex structure modeling framework with diffusion-based generative models. We suppose that this framework will lead to better modeling of protein-ligand complexes, and we expect further improvements and wide applications.

Assessment of protein-ligand complexes in CASP15.

CASP15 introduced a new category, ligand prediction, where participants were provided with a protein or nucleic acid sequence, SMILES line notation, and stoichiometry for ligands and tasked with generating computational models for the three-dimensional structure of the corresponding protein-ligand complex. These models were subsequently compared with experimental structures determined by x-ray crystallography or cryoEM. To assess these predictions, two novel scores were developed. The Binding-Site Superposed, Symmetry-Corrected Pose Root Mean Square Deviation (BiSyRMSD) evaluated the absolute deviations of the models from the experimental structures. At the same time, the Local Distance Difference Test for Protein-Ligand Interactions (lDDT-PLI) assessed the ability of models to reproduce the protein-ligand interactions in the experimental structures. The ligands evaluated in this challenge range from single-atom ions to large flexible organic molecules. More than 1800 submissions were evaluated for their ability to predict 23 different protein-ligand complexes. Overall, the best models could faithfully reproduce the geometries of more than half of the prediction targets. The ligands' size and flexibility were the primary factors influencing the predictions' quality. Small ions and organic molecules with limited flexibility were predicted with high fidelity, while reproducing the binding poses of larger, flexible ligands proved more challenging.

Machine learning classification can reduce false positives in structure-based virtual screening.

With the recent explosion in the size of libraries available for screening, virtual screening is positioned to assume a more prominent role in early drug discovery's search for active chemical matter. In typical virtual screens, however, only about 12% of the top-scoring compounds actually show activity when tested in biochemical assays. We argue that most scoring functions used for this task have been developed with insufficient thoughtfulness into the datasets on which they are trained and tested, leading to overly simplistic models and/or overtraining. These problems are compounded in the literature because studies reporting new scoring methods have not validated their models prospectively within the same study. Here, we report a strategy for building a training dataset (D-COID) that aims to generate highly compelling decoy complexes that are individually matched to available active complexes. Using this dataset, we train a general-purpose classifier for virtual screening (vScreenML) that is built on the XGBoost framework. In retrospective benchmarks, our classifier shows outstanding performance relative to other scoring functions. In a prospective context, nearly all candidate inhibitors from a screen against acetylcholinesterase show detectable activity; beyond this, 10 of 23 compounds have IC50 better than 50 µM. Without any medicinal chemistry optimization, the most potent hit has IC50 280 nM, corresponding to Ki of 173 nM. These results support using the D-COID strategy for training classifiers in other computational biology tasks, and for vScreenML in virtual screening campaigns against other protein targets. Both D-COID and vScreenML are freely distributed to facilitate such efforts.

Acetylcholinesterase inhibition studies of alkaloid components from Crinum asiaticum var. sinicum: in vitro assessments by molecular docking and molecular dynamics simulations.

Alkaloids are among the most important and best-known secondary metabolites as sources of new drugs from medicinal plants and marine organisms. A phytochemical investigation of the whole plant of Crinum asiaticum var. sinicum resulted in the isolation of seven alkaloids (1-7), including one new dimeric compound, bis-(-)-8-demethylmaritidine (1). Their structures were elucidated using NMR and HR-ESI-MS. The absolute configuration of new compound 1 was established by circular dichroism spectroscopy. All isolated compounds were evaluated for their inhibitory effects on acetylcholinesterase (AChE) activity in vitro. Among them, compound 1 exhibited the most potent AChE inhibition. Moreover, molecular docking and molecular dynamics simulations were carried out for the most active compound to investigate their binding interactions and dynamics behavior of the AChE protein-ligand complex. Therefore, compound 1 may be a potential candidate for effectively treating Alzheimer's disease.

Structures of the TetR-like transcription regulator RcdA alone and in complexes with ligands.

RcdA is a helix-turn-helix (HTH) transcriptional regulator belonging to the TetR family. The protein regulates the transcription of curlin subunit gene D, the master regulator of biofilm formation. Moreover, it was predicted that it might be involved in the regulation of up to 27 different genes. However, an effector of RcdA and the environmental conditions which trigger RcdA action remain unknown. Herein, we report the first crystal structures of RcdA in complexes with ligands, trimethylamine N-oxide (TMAO) and tris(hydroxymethyl)aminomethane (Tris), which might serve as RcdA effectors. Based on these structures, the ligand-binding pocket of RcdA was characterized in detail. The conservation of the amino acid residues forming the ligand-binding cavity was analyzed and the comprehensive search for RcdA structural homologs was performed. This analysis indicated that RcdA is structurally similar to multidrug-binding TetR family members, however, its ligand-binding cavity differs significantly from the pockets of its structural homologs. The interaction of RcdA with TMAO and Tris indicates that the protein might be involved in alkaline stress response.

Binding affinity prediction for protein-ligand complex using deep attention mechanism based on intermolecular interactions.

BACKGROUND: Accurate prediction of protein-ligand binding affinity is important for lowering the overall cost of drug discovery in structure-based drug design. For accurate predictions, many classical scoring functions and machine learning-based methods have been developed. However, these techniques tend to have limitations, mainly resulting from a lack of sufficient energy terms to describe the complex interactions between proteins and ligands. Recent deep-learning techniques can potentially solve this problem. However, the search for more efficient and appropriate deep-learning architectures and methods to represent protein-ligand complex is ongoing. RESULTS: In this study, we proposed a deep-neural network model to improve the prediction accuracy of protein-ligand complex binding affinity. The proposed model has two important features, descriptor embeddings with information on the local structures of a protein-ligand complex and an attention mechanism to highlight important descriptors for binding affinity prediction. The proposed model performed better than existing binding affinity prediction models on most benchmark datasets. CONCLUSIONS: We confirmed that an attention mechanism can capture the binding sites in a protein-ligand complex to improve prediction performance. Our code is available at https://github.com/Blue1993/BAPA .

Amide temperature coefficients in characterizing the allosteric effects of ligand binding on local stability in proteins.

Proteins can stabilize upon binding a ligand. Due to allosteric effects, the changes in stability can occur at regions far from the protein:ligand interface. Efficient methods to measure the changes in local stability upon ligand binding will be useful to understand allostery and may be helpful in protein engineering. In this work, we suggest the measurement of backbone amide temperature coefficients to probe the effect of ligand binding on the local stability of ß-sheet rich proteins. The method was applied for two protein:ligand complexes with different binding affinities. The protein includes a beta-sheet network connected by hydrogen bonds. The measured temperature coefficient data captured the stabilizing effect of ligand binding, which propagated across the beta-sheet network of the protein. Intriguingly, the impact on the local and global stability of the protein was proportional to the strength of protein:ligand interaction.

MDMS: Software Facilitating Performing Molecular Dynamics Simulations.

Molecular Dynamics Made Simple (MDMS) is software that facilitates performing molecular dynamics (MD) simulations of solvated protein/protein-ligand complexes with Amber, one of the most popular MD codes. It guides users through the whole process of running MD starting with choosing a protein structure, preparing the model, parametrization of the system, establishing parameters for controlling MD, and finally running simulations. By accommodating every step required for running MD, this software ensures that the simulations performed by a user will provide as realistic insight as it is possible. Its sequential structure and a text-based interface ensure ease of use, while the flexibility required for complex cases is still preserved. MDMS also provides a very time-efficient and streamlined method to start MD simulations, which makes it a feasible tool for both novices and experienced computational chemists. © 2019 Wiley Periodicals, Inc.

A Phenotarget Approach for Identifying an Alkaloid Interacting with the Tuberculosis Protein Rv1466.

In recent years, there has been a revival of interest in phenotypic-based drug discovery (PDD) due to target-based drug discovery (TDD) falling below expectations. Both PDD and TDD have their unique advantages and should be used as complementary methods in drug discovery. The PhenoTarget approach combines the strengths of the PDD and TDD approaches. Phenotypic screening is conducted initially to detect cellular active components and the hits are then screened against a panel of putative targets. This PhenoTarget protocol can be equally applied to pure compound libraries as well as natural product fractions. Here we described the use of the PhenoTarget approach to identify an anti-tuberculosis lead compound. Fractions from Polycarpa aurata were identified with activity against Mycobacterium tuberculosis H37Rv. Native magnetic resonance mass spectrometry (MRMS) against a panel of 37 proteins from Mycobacterium proteomes showed that a fraction from a 95% ethanol re-extraction specifically formed a protein-ligand complex with Rv1466, a putative uncharacterized Mycobacterium tuberculosis protein. The natural product responsible was isolated and characterized to be polycarpine. The molecular weight of the ligand bound to Rv1466, 233 Da, was half the molecular weight of polycarpine less one proton, indicating that polycarpine formed a covalent bond with Rv1466.

Structural basis for selective inhibition of human PKG Iα by the balanol-like compound N46.

Activation of protein kinase G (PKG) Iα in nociceptive neurons induces long-term hyperexcitability that causes chronic pain. Recently, a derivative of the fungal metabolite balanol, N46, has been reported to inhibit PKG Iα with high potency and selectivity and attenuate thermal hyperalgesia and osteoarthritic pain. Here we determined co-crystal structures of the PKG Iα C-domain and cAMP-dependent protein kinase (PKA) Cα, each bound with N46, at 1.98 Å and 2.65 Å, respectively. N46 binds the active site with its external phenyl ring, specifically interacting with the glycine-rich loop and the αC helix. Phe-371 at the PKG Iα glycine-rich loop is oriented parallel to the phenyl ring of N46, forming a strong π-stacking interaction, whereas the analogous Phe-54 in PKA Cα rotates 30° and forms a weaker interaction. Structural comparison revealed that steric hindrance between the preceding Ser-53 and the propoxy group of the phenyl ring may explain the weaker interaction with PKA Cα. The analogous Gly-370 in PKG Iα, however, causes little steric hindrance with Phe-371. Moreover, Ile-406 on the αC helix forms a hydrophobic interaction with N46 whereas its counterpart in PKA, Thr-88, does not. Substituting these residues in PKG Iα with those in PKA Cα increases the IC50 values for N46, whereas replacing these residues in PKA Cα with those in PKG Iα reduces the IC50, consistent with our structural findings. In conclusion, our results explain the structural basis for N46-mediated selective inhibition of human PKG Iα and provide a starting point for structure-guided design of selective PKG Iα inhibitors.

Both protein dynamics and ligand concentration can shift the binding mechanism between conformational selection and induced fit.

This study aimed to shed light on the long debate over whether conformational selection (CS) or induced fit (IF) is the governing mechanism for protein-ligand binding. The main difference between the two scenarios is whether the conformational transition of the protein from the unbound form to the bound form occurs before or after encountering the ligand. Here we introduce the IF fraction (i.e., the fraction of binding events achieved via IF), to quantify the binding mechanism. Using simulations of a model protein-ligand system, we demonstrate that both the rate of the conformational transition and the concentration of ligand molecules can affect the IF fraction. CS dominates at slow conformational transition and low ligand concentration. An increase in either quantity results in a higher IF fraction. Despite the many-body nature of the system and the involvement of multiple, disparate types of dynamics (i.e., ligand diffusion, protein conformational transition, and binding reaction), the overall binding kinetics over wide ranges of parameters can be fit to a single exponential, with the apparent rate constant exhibiting a linear dependence on ligand concentration. The present study may guide future kinetics experiments and dynamics simulations in determining the IF fraction.

Mutation design of a thermophilic Rubisco based on three-dimensional structure enhances its activity at ambient temperature.

Ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) plays a central role in carbon dioxide fixation on our planet. Rubisco from a hyperthermophilic archaeon Thermococcus kodakarensis (Tk-Rubisco) shows approximately twenty times the activity of spinach Rubisco at high temperature, but only one-eighth the activity at ambient temperature. We have tried to improve the activity of Tk-Rubisco at ambient temperature, and have successfully constructed several mutants which showed higher activities than the wild-type enzyme both in vitro and in vivo. Here, we designed new Tk-Rubisco mutants based on its three-dimensional structure and a sequence comparison of thermophilic and mesophilic plant Rubiscos. Four mutations were introduced to generate new mutants based on this strategy, and one of the four mutants, T289D, showed significantly improved activity compared to that of the wild-type enzyme. The crystal structure of the Tk-Rubisco T289D mutant suggested that the increase in activity was due to mechanisms distinct from those involved in the improvement in activity of Tk-Rubisco SP8, a mutant protein previously reported to show the highest activity at ambient temperature. Combining the mutations of T289D and SP8 successfully generated a mutant protein (SP8-T289D) with the highest activity to date both in vitro and in vivo. The improvement was particularly pronounced for the in vivo activity of SP8-T289D when introduced into the mesophilic, photosynthetic bacterium Rhodopseudomonas palustris, which resulted in a strain with nearly two-fold higher specific growth rates compared to that of a strain harboring the wild-type enzyme at ambient temperature. Proteins 2016; 84:1339-1346. © 2016 Wiley Periodicals, Inc.

Native Mass Spectrometry in Fragment-Based Drug Discovery.

The advent of native mass spectrometry (MS) in 1990 led to the development of new mass spectrometry instrumentation and methodologies for the analysis of noncovalent protein-ligand complexes. Native MS has matured to become a fast, simple, highly sensitive and automatable technique with well-established utility for fragment-based drug discovery (FBDD). Native MS has the capability to directly detect weak ligand binding to proteins, to determine stoichiometry, relative or absolute binding affinities and specificities. Native MS can be used to delineate ligand-binding sites, to elucidate mechanisms of cooperativity and to study the thermodynamics of binding. This review highlights key attributes of native MS for FBDD campaigns.

NMR structure calculation for all small molecule ligands and non-standard residues from the PDB Chemical Component Dictionary.

An algorithm, CYLIB, is presented for converting molecular topology descriptions from the PDB Chemical Component Dictionary into CYANA residue library entries. The CYANA structure calculation algorithm uses torsion angle molecular dynamics for the efficient computation of three-dimensional structures from NMR-derived restraints. For this, the molecules have to be represented in torsion angle space with rotations around covalent single bonds as the only degrees of freedom. The molecule must be given a tree structure of torsion angles connecting rigid units composed of one or several atoms with fixed relative positions. Setting up CYANA residue library entries therefore involves, besides straightforward format conversion, the non-trivial step of defining a suitable tree structure of torsion angles, and to re-order the atoms in a way that is compatible with this tree structure. This can be done manually for small numbers of ligands but the process is time-consuming and error-prone. An automated method is necessary in order to handle the large number of different potential ligand molecules to be studied in drug design projects. Here, we present an algorithm for this purpose, and show that CYANA structure calculations can be performed with almost all small molecule ligands and non-standard amino acid residues in the PDB Chemical Component Dictionary.