Ligand and decoy sets for docking to G protein-coupled receptors

Discovering allatostatin type-C receptor specific agonists

Currently, there is no pesticide available for the selective control of the pine processionary moth (Thaumetopoea pityocampa-specific), and conventional methods typically rely on mechanical techniques such as pheromone traps or broad-spectrum larvicidal chemicals. As climate change increases the range and dispersion capacity of crop and forest pests, outbreaks of the pine processionary occur with greater frequency and significantly impact forestry and public health. Our study is carried out to provide a T. pityocampa-specific pesticide targeting the Allatostatin Type-C Receptor (AlstR-C). We use a combination of computational biology methods, a cell-based screening assay, and in vivo toxicity and side effect assays to identify, for the first time, a series of AlstR-C ligands suitable for use as T. pityocampa-specific insecticides. We further demonstrate that the novel AlstR-C targeted agonists are specific to lepidopteran larvae, with no harmful effects on coleopteran larvae or adults. Overall, our study represents an important initial advance toward an insect GPCR-targeted next-generation pesticide design. Our approach may apply to other invertebrate GPCRs involved in vital metabolic pathways.

Virtual Screening of a Chemically Diverse "Superscaffold" Library Enables Ligand Discovery for a Key GPCR Target

The advent of ultra-large libraries of drug-like compounds has significantly broadened the possibilities in structure-based virtual screening, accelerating the discovery and optimization of high-quality lead chemotypes for diverse clinical targets. Compared to traditional high-throughput screening, which is constrained to libraries of approximately one million compounds, the ultra-large virtual screening approach offers substantial advantages in both cost and time efficiency. By expanding the chemical space with compounds synthesized from easily accessible and reproducible reactions and utilizing a large, diverse set of building blocks, we can enhance both the diversity and quality of the discovered lead chemotypes. In this study, we explore new chemical spaces using reactions of sulfur(VI) fluorides to create a combinatorial library consisting of several hundred million compounds. We screened this virtual library for cannabinoid type II receptor (CB2) antagonists using the high-resolution structure in conjunction with a rationally designed antagonist, AM10257. The top-predicted compounds were then synthesized and tested in vitro for CB2 binding and functional antagonism, achieving an experimentally validated hit rate of 55%. Our findings demonstrate the effectiveness of reliable reactions, such as sulfur fluoride exchange, in diversifying ultra-large chemical spaces and facilitate the discovery of new lead compounds for important biological targets.

Deep reinforcement learning enables better bias control in benchmark for virtual screening

Virtual screening (VS) has been incorporated into the paradigm of modern drug discovery. This field is now undergoing a new wave of revolution driven by artificial intelligence and more specifically, machine learning (ML). In terms of those out-of-the-box datasets for model training or benchmarking, their data volume and applicability domain are limited. They are suffering from the biases constantly reported in the ML application. To address these issues, we present a novel benchmark named MUBDsyn. The utilization of synthetic decoys (i.e., presumed inactives) is the main feature of MUBDsyn, where deep reinforcement learning was leveraged for bias control during decoy generation. Then, we carried out extensive validations on this new benchmark. First, we confirmed that MUBDsyn was superior to the classical benchmarks in control of domain bias, artificial enrichment bias and analogue bias. Moreover, we found that the assessment of ML models based on MUBDsyn was less biased as revealed by the analysis of asymmetric validation embedding bias. In addition, MUBDsyn showed better setting of benchmarking challenge for deep learning models compared with NRLiSt-BDB. Overall, we have proven that MUBDsyn is the close-to-ideal benchmark for VS. The computational tool is publicly available for the easy extension of MUBDsyn.

Identification and quantification of kukoamine A and kukoamine B as novel μ-opioid receptor agonists in potato and other solanaceous plants

Activation of the µ-opioid receptor (µOR) by food components could lead to reward effects or to the modulation of motor functions in the gastrointestinal tract. In an unbiased search for novel µOR agonists in food, a three-step virtual-screening process selected 22 promising candidates with potential to interact with the µOR. Radioligand binding studies showed that ten of these substances indeed bind to the receptor. Functional assays identified kukoamine A as a full agonist (EC50 = 5.6 µM) and kukoamine B as a partial agonist (EC50 = 8.7 µM) to µOR. After extraction, both kukoamines were analyzed by LC-MS/MS in potato, tomato, pepper, and eggplant. Depending on the potato variety, up to 16 µg of kukoamine A and 157 µg of kukoamine B per gram dry weight could be determined in the whole tuber, mainly concentrated in the potato peel. Cooking did not influence the kukoamine contents.

Machine learning- and structure-based discovery of a novel chemotype as FXR agonists for potential treatment of nonalcoholic fatty liver disease

Farnesoid X receptor (FXR) is a promising target for drug discovery against nonalcoholic fatty liver disease (NAFLD). However, no FXR agonist has been approved for NAFLD so far. The R & D of FXR agonists are somewhat hindered by the lack of effective and safe chemotypes. To this end, we developed a multi-stage computational workflow to screen the Specs and ChemDiv chemical library for FXR agonists, which consisted of machine learning (ML)-based classifiers, shape-based and electrostatic-based models, a FRED-based molecular docking protocol, an ADMET prediction protocol and substructure search. As a result, we identified a novel chemotype that has never been reported before, with compound XJ02862 (ChemDiv ID: Y020-6413) as the representative. By designing an asymmetric synthesis strategy, we were able to prepare four isomers of compound XJ02862. Interestingly, one of the isomers, 2-((S)-1-((2S,4R)-2-methyl-4-(phenylamino)-3,4-dihydroquinolin-1(2H)-yl)-1-oxopropan-2-yl)hexahydro-1H-isoindole-1,3(2H)-dione (XJ02862-S2), showed potent FXR agonistic activity in HEK293T cells. The molecular docking, molecular dynamics simulations and site-directed mutagenesis suggested the hydrogen bond between compound XJ02862-S2 and HIS294 of FXR is essential for ligand binding. We further demonstrated that compound XJ02862-S2 had no agonistic effect on TGR5. Further biological experiments have shown that compound XJ02862-S2 could ameliorate hypercholesterolemia, hepatic steatosis, hyperglycemia, insulin resistance (IR) in high-fat-diet induced obese (DIO) mice. In term of molecular mechanism, compound XJ02862-S2 regulates the expression of FXR downstream genes involved in lipogenesis, cholesterol transport and bile acid biosynthesis and transport. Taken together, we have discovered a novel chemotype as potent FXR agonists for NAFLD by computational modeling, chemical synthesis and biological evaluation.

How good are AlphaFold models for docking-based virtual screening?

A crucial component in structure-based drug discovery is the availability of high-quality three-dimensional structures of the protein target. Whenever experimental structures were not available, homology modeling has been, so far, the method of choice. Recently, AlphaFold (AF), an artificial-intelligence-based protein structure prediction method, has shown impressive results in terms of model accuracy. This outstanding success prompted us to evaluate how accurate AF models are from the perspective of docking-based drug discovery. We compared the high-throughput docking (HTD) performance of AF models with their corresponding experimental PDB structures using a benchmark set of 22 targets. The AF models showed consistently worse performance using four docking programs and two consensus techniques. Although AlphaFold shows a remarkable ability to predict protein architecture, this might not be enough to guarantee that AF models can be reliably used for HTD, and post-modeling refinement strategies might be key to increase the chances of success.

Binding and Activation of Serotonergic G-Protein Coupled Receptors by the Multimodal Antidepressant Vortioxetine

G-protein coupled receptors (GPCRs) are important pharmacological targets. Despite substantial progress, important questions still remain concerning the details of activation: how can a ligand act as an agonist in one receptor but as an antagonist in a homologous receptor, and how can agonists activate a receptor despite lacking polar functional groups able to interact with helix 5 as is the case for the related adrenergic receptors? Studying vortioxetine (VXT), an important multimodal antidepressant drug, may elucidate both questions. Herein, we present a thorough in silico analysis of VXT binding to 5-HT_1A, 5-HT_1B, and 5-HT₇ receptors and compare it with available experimental data. We are able to rationalize the differential mode of action of VXT at different receptors, but also, in the case of the 5-HT_1A receptor, we observe the initial steps of activation that inform about an activation mechanism that does not involve polar interaction with helix 5. The results extend our current understanding of agonist and antagonist action at aminergic GPCRs.

Synthon-based ligand discovery in virtual libraries of over 11 billion compounds

Structure-based virtual ligand screening is emerging as a key paradigm for early drug discovery owing to the availability of high-resolution target structures^1-4 and ultra-large libraries of virtual compounds^5,6. However, to keep pace with the rapid growth of virtual libraries, such as readily available for synthesis (REAL) combinatorial libraries⁷, new approaches to compound screening are needed^8,9. Here we introduce a modular synthon-based approach-V-SYNTHES-to perform hierarchical structure-based screening of a REAL Space library of more than 11 billion compounds. V-SYNTHES first identifies the best scaffold-synthon combinations as seeds suitable for further growth, and then iteratively elaborates these seeds to select complete molecules with the best docking scores. This hierarchical combinatorial approach enables the rapid detection of the best-scoring compounds in the gigascale chemical space while performing docking of only a small fraction (<0.1%) of the library compounds. Chemical synthesis and experimental testing of novel cannabinoid antagonists predicted by V-SYNTHES demonstrated a 33% hit rate, including 14 submicromolar ligands, substantially improving over a standard virtual screening of the Enamine REAL diversity subset, which required approximately 100 times more computational resources. Synthesis of selected analogues of the best hits further improved potencies and affinities (best inhibitory constant (K_i) = 0.9 nM) and CB₂/CB₁ selectivity (50-200-fold). V-SYNTHES was also tested on a kinase target, ROCK1, further supporting its use for lead discovery. The approach is easily scalable for the rapid growth of combinatorial libraries and potentially adaptable to any docking algorithm.

Ligand-Based Virtual Screening Based on the Graph Edit Distance

Chemical compounds can be represented as attributed graphs. An attributed graph is a mathematical model of an object composed of two types of representations: nodes and edges. Nodes are individual components, and edges are relations between these components. In this case, pharmacophore-type node descriptions are represented by nodes and chemical bounds by edges. If we want to obtain the bioactivity dissimilarity between two chemical compounds, a distance between attributed graphs can be used. The Graph Edit Distance allows computing this distance, and it is defined as the cost of transforming one graph into another. Nevertheless, to define this dissimilarity, the transformation cost must be properly tuned. The aim of this paper is to analyse the structural-based screening methods to verify the quality of the Harper transformation costs proposal and to present an algorithm to learn these transformation costs such that the bioactivity dissimilarity is properly defined in a ligand-based virtual screening application. The goodness of the dissimilarity is represented by the classification accuracy. Six publicly available datasets-CAPST, DUD-E, GLL&GDD, NRLiSt-BDB, MUV and ULS-UDS-have been used to validate our methodology and show that with our learned costs, we obtain the highest ratios in identifying the bioactivity similarity in a structurally diverse group of molecules.

Encoding mu-opioid receptor biased agonism with interaction fingerprints

Opioids are potent painkillers, however, their therapeutic use requires close medical monitoring to diminish the risk of severe adverse effects. The G-protein biased agonists of the μ-opioid receptor (MOR) have shown safer therapeutic profiles than non-biased ligands. In this work, we performed extensive all-atom molecular dynamics simulations of two markedly biased ligands and a balanced reference molecule. From those simulations, we identified a protein-ligand interaction fingerprint that characterizes biased ligands. Then, we built and virtually screened a database containing 68,740 ligands with proven or potential GPCR agonistic activity. Exemplary molecules that fulfill the interacting pattern for biased agonism are showcased, illustrating the usefulness of this work for the search of biased MOR ligands and how this contributes to the understanding of MOR biased signaling.

Combination of pose and rank consensus in docking-based virtual screening: the best of both worlds

The use of high-throughput docking (HTD) in the drug discovery pipeline is today widely established. In spite of methodological improvements in docking accuracy (pose prediction), scoring power, ranking power, and screening power in HTD remain challenging. In fact, pose prediction is of critical importance in view of the pose-dependent scoring process, since incorrect poses will necessarily decrease the ranking power of scoring functions. The combination of results from different docking programs (consensus scoring) has been shown to improve the performance of HTD. Moreover, it has been also shown that a pose consensus approach might also result in database enrichment. We present a new methodology named Pose/Ranking Consensus (PRC) that combines both pose and ranking consensus approaches, to overcome the limitations of each stand-alone strategy. This approach has been developed using four docking programs (ICM, rDock, Auto Dock 4, and PLANTS; the first one is commercial, the other three are free). We undertook a thorough analysis for the best way of combining pose and rank strategies, and applied the PRC to a wide range of 34 targets sampling different protein families and binding site properties. Our approach exhibits an improved systematic performance in terms of enrichment factor and hit rate with respect to either pose consensus or consensus ranking alone strategies at a lower computational cost, while always ensuring the recovery of a suitable number of ligands. An analysis using four free docking programs (replacing ICM by Auto Dock Vina) displayed comparable results.

The new methodology named Pose/Ranking Consensus (PRC) combines both pose and ranking consensus strategies. It displays an enhanced performance in terms of enrichment factor and hit rate, ensuring the recovery of a suitable number of ligands.

Efficiency of Homology Modeling Assisted Molecular Docking in G-protein Coupled Receptors

BACKGROUND

Molecular docking is in regular practice to assess ligand affinity on a target protein crystal structure. In the absence of protein crystal structure, the homology modeling or comparative modeling is the best alternative to elucidate the relationship details between a ligand and protein at the molecular level. The development of accurate homology modeling (HM) and its integration with molecular docking (MD) is essential for successful, rational drug discovery.

OBJECTIVE

The G-protein coupled receptors (GPCRs) are attractive therapeutic targets due to their immense role in human pharmacology. The GPCRs are membrane-bound proteins with the complex constitution, and the understanding of their activation and inactivation mechanisms is quite challenging. Over the past decade, there has been a rapid expansion in the number of solved G-protein-coupled receptor (GPCR) crystal structures; however, the majority of the GPCR structures remain unsolved. In this context, HM guided MD has been widely used for structure-based drug design (SBDD) of GPCRs.

METHODS

The focus of this review is on the recent (i) developments on HM supported GPCR drug discovery in the absence of GPCR crystal structures and (ii) application of HM in understanding the ligand interactions at the binding site, virtual screening, determining receptor subtype selectivity and receptor behaviour in comparison with GPCR crystal structures.

RESULTS

The HM in GPCRs has been extremely challenging due to the scarcity in template structures. In such a scenario, it is difficult to get accurate HM that can facilitate understanding of the ligand-receptor interactions. This problem has been alleviated to some extent by developing refined HM based on incorporating active /inactive ligand information and inducing protein flexibility. In some cases, HM proteins were found to outscore crystal structures.

CONCLUSION

The developments in HM have been highly operative to gain insights about the ligand interaction at the binding site and receptor functioning at the molecular level. Thus, HM guided molecular docking may be useful for rational drug discovery for the GPCRs mediated diseases.

Discovery of Novel CCR5 Ligands as Anticolorectal Cancer Agents by Sequential Virtual Screening

C-C chemokine receptor type 5 (CCR5) is a member of the G protein-coupled receptor. CCR5 and its interaction with chemokine ligands have been crucial for understanding and tackling human immunodeficiency virus (HIV)-1 entry into target cells. In recent years, the change in CCR5 expression has been related to the progression of different cancer types. Patients treated with the CCR5 ligand, maraviroc (MVC), showed a deceleration in tumor development especially for metastatic colorectal cancer. Based on the crystal structure of CCR5, we herein describe a multistage virtual screening protocol including pharmacophore screening, molecular docking, and protein-ligand interaction fingerprint (PLIF) postdocking filtration for discovery of novel CCR5 ligands. The applied virtual screening protocol led to the identification of four hits with binding modes showing access to the major and minor pockets of the MVC binding site. Compounds 2-4 showed a decrease in cellular proliferation upon testing on the metastatic colorectal cancer cell line, SW620, displaying 12, 16, and 4 times higher potency compared to MVC, respectively. Compound 3 induced apoptosis by arresting cells in the G0/G1 phase of the cell cycle similar to MVC. Further in vitro assays showed compound 3 drastically decreasing the CCR5 expression and cellular migration 48 h post treatment, indicating its ability to inhibit metastatic activity in SW620 cells. The discovered hits represent potential leads for the development of novel classes of anticolorectal cancer agents targeting CCR5.

Property-Unmatched Decoys in Docking Benchmarks

Enrichment of ligands versus property-matched decoys is widely used to test and optimize docking library screens. However, the unconstrained optimization of enrichment alone can mislead, leading to false confidence in prospective performance. This can arise by over-optimizing for enrichment against property-matched decoys, without considering the full spectrum of molecules to be found in a true large library screen. Adding decoys representing charge extrema helps mitigate over-optimizing for electrostatic interactions. Adding decoys that represent the overall characteristics of the library to be docked allows one to sample molecules not represented by ligands and property-matched decoys but that one will encounter in a prospective screen. An optimized version of the DUD-E set (DUDE-Z), as well as Extrema and sets representing broad features of the library (Goldilocks), is developed here. We also explore the variability that one can encounter in enrichment calculations and how that can temper one's confidence in small enrichment differences. The new tools and new decoy sets are freely available at http://tldr.docking.org and http://dudez.docking.org.

VSPrep: A KNIME Workflow for the Preparation of Molecular Databases for Virtual Screening

Drug discovery is a challenging and expensive field. Hence, novel in silico tools have been developed in early discovery stage to identify and prioritize novel molecules with suitable physicochemical properties. In many in silico drug design projects, molecular databases are screened by virtual screening tools to search for potential bioactive molecules. The preparation of the molecules is therefore a key step in the success of well-established techniques such as docking, similarity or pharmacophore searching. We review here the lists of several toolkits used in different steps during the cleaning of molecular databases, integrated within a KNIME workflow. During the first step of the automatic workflow, salts are removed, and mixtures are split to get one compound per entry. Then compounds with unwanted features are filtered. Duplicated entries are then deleted while considering stereochemistry. As a compromise between exhaustiveness and computational time, most distributed tautomers at physiological pH are computed. Additionally, various flags are applied to molecules by using either classical molecular descriptors, similarity search to known libraries or substructure search rules. Moreover, stereoisomers are enumerated depending on the unassigned chiral centers. Then, three-dimensional coordinates, and optionally conformers, are generated. This workflow has been already applied to several drug design projects and can be used for molecular database preparation upon request.

Ligand- and Structure-Based Analysis of Deep Learning-Generated Potential α2a Adrenoceptor Agonists

The α2a adrenoceptor is a medically relevant subtype of the G protein-coupled receptor family. Unfortunately, high-throughput techniques aimed at producing novel drug leads for this receptor have been largely unsuccessful because of the complex pharmacology of adrenergic receptors. As such, cutting-edge in silico ligand- and structure-based assessment and de novo deep learning methods are well positioned to provide new insights into protein-ligand interactions and potential active compounds. In this work, we (i) collect a dataset of α2a adrenoceptor agonists and provide it as a resource for the drug design community; (ii) use the dataset as a basis to generate candidate-active structures via deep learning; and (iii) apply computational ligand- and structure-based analysis techniques to gain new insights into α2a adrenoceptor agonists and assess the quality of the computer-generated compounds. We further describe how such assessment techniques can be applied to putative chemical probes with a case study involving proposed medetomidine-based probes.

Structure-Based Virtual Screening of Ultra-Large Library Yields Potent Antagonists for a Lipid GPCR

Cysteinyl leukotriene G protein-coupled receptors, CysLT1R and CysLT2R, regulate bronchoconstrictive and pro-inflammatory effects and play a key role in allergic disorders, cardiovascular diseases, and cancer. CysLT1R antagonists have been widely used to treat asthma disorders, while CysLT2R is a potential target against uveal melanoma. However, very few selective antagonist chemotypes for CysLT receptors are available, and the design of such ligands has proved to be challenging. To overcome this obstacle, we took advantage of recently solved crystal structures of CysLT receptors and an ultra-large Enamine REAL library, representing a chemical space of 680 M readily available compounds. Virtual ligand screening employed 4D docking models comprising crystal structures of CysLT1R and CysLT2R and their corresponding ligand-optimized models. Functional assessment of the candidate hits yielded discovery of five novel antagonist chemotypes with sub-micromolar potencies and the best Ki = 220 nM at CysLT1R. One of the hits showed inverse agonism at the L129Q constitutively active mutant of CysLT2R, with potential utility against uveal melanoma.

Benchmarking Data Sets from PubChem BioAssay Data: Current Scenario and Room for Improvement

Developing realistic data sets for evaluating virtual screening methods is a task that has been tackled by the cheminformatics community for many years. Numerous artificially constructed data collections were developed, such as DUD, DUD-E, or DEKOIS. However, they all suffer from multiple drawbacks, one of which is the absence of experimental results confirming the impotence of presumably inactive molecules, leading to possible false negatives in the ligand sets. In light of this problem, the PubChem BioAssay database, an open-access repository providing the bioactivity information of compounds that were already tested on a biological target, is now a recommended source for data set construction. Nevertheless, there exist several issues with the use of such data that need to be properly addressed. In this article, an overview of benchmarking data collections built upon experimental PubChem BioAssay input is provided, along with a thorough discussion of noteworthy issues that one must consider during the design of new ligand sets from this database. The points raised in this review are expected to guide future developments in this regard, in hopes of offering better evaluation tools for novel in silico screening procedures.

Learning the Edit Costs of Graph Edit Distance Applied to Ligand-Based Virtual Screening

BACKGROUND

Graph edit distance is a methodology used to solve error-tolerant graph matching. This methodology estimates a distance between two graphs by determining the minimum number of modifications required to transform one graph into the other. These modifications, known as edit operations, have an edit cost associated that has to be determined depending on the problem.

OBJECTIVE

This study focuses on the use of optimization techniques in order to learn the edit costs used when comparing graphs by means of the graph edit distance.

METHODS

Graphs represent reduced structural representations of molecules using pharmacophore-type node descriptions to encode the relevant molecular properties. This reduction technique is known as extended reduced graphs. The screening and statistical tools available on the ligand-based virtual screening benchmarking platform and the RDKit were used.

RESULTS

In the experiments, the graph edit distance using learned costs performed better or equally good than using predefined costs. This is exemplified with six publicly available datasets: DUD-E, MUV, GLL&GDD, CAPST, NRLiSt BDB, and ULS-UDS.

CONCLUSION

This study shows that the graph edit distance along with learned edit costs is useful to identify bioactivity similarities in a structurally diverse group of molecules. Furthermore, the target-specific edit costs might provide useful structure-activity information for future drug-design efforts.

High-Throughput Docking Using Quantum Mechanical Scoring

Today high-throughput docking is one of the most commonly used computational tools in drug lead discovery. While there has been an impressive methodological improvement in docking accuracy, docking scoring still remains an open challenge. Most docking programs are rooted in classical molecular mechanics. However, to better characterize protein-ligand interactions, the use of a more accurate quantum mechanical (QM) description would be necessary. In this work, we introduce a QM-based docking scoring function for high-throughput docking and evaluate it on 10 protein systems belonging to diverse protein families, and with different binding site characteristics. Outstanding results were obtained, with our QM scoring function displaying much higher enrichment (screening power) than a traditional docking method. It is acknowledged that developments in quantum mechanics theory, algorithms and computer hardware throughout the upcoming years will allow semi-empirical (or low-cost) quantum mechanical methods to slowly replace force-field calculations. It is thus urgently needed to develop and validate novel quantum mechanical-based scoring functions for high-throughput docking toward more accurate methods for the identification and optimization of modulators of pharmaceutically relevant targets.

Structure-based discovery of potent and selective melatonin receptor agonists

Melatonin receptors MT1 and MT2 are involved in synchronizing circadian rhythms and are important targets for treating sleep and mood disorders, type-2 diabetes and cancer. Here, we performed large scale structure-based virtual screening for new ligand chemotypes using recently solved high-resolution 3D crystal structures of agonist-bound MT receptors. Experimental testing of 62 screening candidates yielded the discovery of 10 new agonist chemotypes with sub-micromolar potency at MT receptors, with compound 21 reaching EC50 of 0.36 nM. Six of these molecules displayed selectivity for MT2 over MT1. Moreover, two most potent agonists, including 21 and a close derivative of melatonin, 28, had dramatically reduced arrestin recruitment at MT2, while compound 37 was devoid of Gi signaling at MT1, implying biased signaling. This study validates the suitability of the agonist-bound orthosteric pocket in the MT receptor structures for the structure-based discovery of selective agonists.

Pose Filter-Based Ensemble Learning Enables Discovery of Orally Active, Nonsteroidal Farnesoid X Receptor Agonists

Farnesoid X receptor (FXR) agonists can reverse dysregulated bile acid metabolism, and thus, they are potential therapeutics to prevent and treat nonalcoholic fatty liver disease. The low success rate of FXR agonists' R&D and the side effects of clinical candidates such as obeticholic acid make it urgent to discover new chemotypes. Unfortunately, structure-based virtual screening (SBVS) that can speed up drug discovery has rarely been reported with success for FXR, which was likely hindered by the failure in addressing protein flexibility. To address this issue, we devised human FXR (hFXR)-specific ensemble learning models based on pose filters from 24 agonist-bound hFXR crystal structures and coupled them to traditional SBVS approaches of the FRED docking plus Chemgauss4 scoring function. It turned out that the hFXR-specific pose filter ensemble (PFE) was able to improve ligand enrichment significantly, which rendered 3RUT-based SBVS with its PFE the ideal approach for FXR agonist discovery. By screening of the Specs chemical library and in vitro FXR transactivation bioassay, we identified a new class of FXR agonists with compound XJ034 as the representative, which would have been missed if the PFE was not coupled. Following that, we performed in-depth biological studies which demonstrated that XJ034 resulted in a downtrend of intracellular triglyceride in vitro, significantly decreased the serum/liver TG in high fat diet-induced C57BL/6J obese mice, and more importantly, showed metabolic stabilities in both plasma and liver microsomes. To provide insight into further structure-based lead optimization, we solved the crystal structure of hFXR complexed with compound XJ034, uncovering a unique hydrogen bond between compound XJ034 and residue Y375. The current work highlights the power of our pose filter-based ensemble learning approach in terms of scaffold hopping and provides a promising lead compound for further development.

The role of water and protein flexibility in the structure-based virtual screening of allosteric GPCR modulators: an mGlu5 receptor case study

Stabilizing unique receptor conformations, allosteric modulators of G-protein coupled receptors (GPCRs) might open novel treatment options due to their new pharmacological action, their enhanced specificity and selectivity in both binding and signaling. Ligand binding occurs at intrahelical allosteric sites and involves significant induced fit effects that include conformational changes in the local protein environment and water networks. Based on the analysis of available crystal structures of metabotropic glutamate receptor 5 (mGlu₅) we investigated these effects in the binding of mGlu₅ receptor negative allosteric modulators. A large set of retrospective virtual screens revealed that the use of multiple protein structures and the inclusion of selected water molecules improves virtual screening performance compared to conventional docking strategies. The role of water molecules and protein flexibility in ligand binding can be taken into account efficiently by the proposed docking protocol that provided reasonable enrichment of true positives. This protocol is expected to be useful also for identifying intrahelical allosteric modulators for other GPCR targets.

Key Topics in Molecular Docking for Drug Design

Molecular docking has been widely employed as a fast and inexpensive technique in the past decades, both in academic and industrial settings. Although this discipline has now had enough time to consolidate, many aspects remain challenging and there is still not a straightforward and accurate route to readily pinpoint true ligands among a set of molecules, nor to identify with precision the correct ligand conformation within the binding pocket of a given target molecule. Nevertheless, new approaches continue to be developed and the volume of published works grows at a rapid pace. In this review, we present an overview of the method and attempt to summarise recent developments regarding four main aspects of molecular docking approaches: (i) the available benchmarking sets, highlighting their advantages and caveats, (ii) the advances in consensus methods, (iii) recent algorithms and applications using fragment-based approaches, and (iv) the use of machine learning algorithms in molecular docking. These recent developments incrementally contribute to an increase in accuracy and are expected, given time, and together with advances in computing power and hardware capability, to eventually accomplish the full potential of this area.

Influence of the Structural Accuracy of Homology Models on Their Applicability to Docking-Based Virtual Screening: The β2 Adrenergic Receptor as a Case Study

How accurate do structures of the β₂ adrenergic receptor (β₂AR) need to be to effectively serve as platforms for docking-based virtual screening campaigns? To answer this research question, here, we targeted through controlled virtual screening experiments 23 homology models of the β₂AR endowed with different levels of structural accuracy. Subsequently, we studied the correlation between virtual screening performance and structural accuracy of the targeted models. Moreover, we studied the correlation between virtual screening performance and template/target receptor sequence identity. Our study demonstrates that docking-based virtual screening campaigns targeting homology models of the β₂AR, in the majority of the cases, yielded results that exceeded random expectations in terms of area under the receiver operating characteristic curve (ROC AUC). Moreover, with the most effective scoring method, over one-third and one-quarter of the models yielded results that exceeded random expectation also in terms of enrichment factors (EF1, EF5, and EF10) and BEDROC (α = 160.9), respectively. Not surprisingly, we found a detectable linear correlation between virtual screening performance and structural accuracy of the ligand-binding cavity. We also found a detectable linear correlation between virtual screening performance and structural accuracy of the second extracellular loop (EL2). Finally, our data indicate that, although there is no detectable linear correlation between virtual screening performance and template/β₂AR sequence identity, models built on the basis of templates that show high sequence identity with the β₂AR, especially within the ligand-biding cavity, performed consistently well. Conversely, models with lower sequence identity displayed performance levels that ranged from very good to random, with no apparent correlation with the sequence identity itself.

Ligand-Based Virtual Screening Using Graph Edit Distance as Molecular Similarity Measure

Extended reduced graphs provide summary representations of chemical structures using pharmacophore-type node descriptions to encode the relevant molecular properties. Commonly used similarity measures using reduced graphs convert these graphs into 2D vectors like fingerprints, before chemical comparisons are made. This study investigates the effectiveness of a graph-only driven molecular comparison by using extended reduced graphs along with graph edit distance methods for molecular similarity calculation as a tool for ligand-based virtual screening applications, which estimate the bioactivity of a chemical on the basis of the bioactivity of similar compounds. The results proved to be very stable and the graph editing distance method performed better than other methods previously used on reduced graphs. This is exemplified with six publicly available data sets: DUD-E, MUV, GLL&GDD, CAPST, NRLiSt BDB, and ULS-UDS. The screening and statistical tools available on the ligand-based virtual screening benchmarking platform and the RDKit were also used. In the experiments, our method performed better than other molecular similarity methods which use array representations in most cases. Overall, it is shown that extended reduced graphs along with graph edit distance is a combination of methods that has numerous applications and can identify bioactivity similarities in a structurally diverse group of molecules.

Exponential consensus ranking improves the outcome in docking and receptor ensemble docking

Consensus-scoring methods are commonly used with molecular docking in virtual screening campaigns to filter potential ligands for a protein target. Traditional consensus methods combine results from different docking programs by averaging the score or rank of each molecule obtained from individual programs. Unfortunately, these methods fail if one of the docking programs has poor performance, which is likely to occur due to training-set dependencies and scoring-function parameterization. In this work, we introduce a novel consensus method that overcomes these limitations. We combine the results from individual docking programs using a sum of exponential distributions as a function of the molecule rank for each program. We test the method over several benchmark systems using individual and ensembles of target structures from diverse protein families with challenging decoy/ligand datasets. The results demonstrate that the novel method outperforms the best traditional consensus strategies over a wide range of systems. Moreover, because the novel method is based on the rank rather than the score, it is independent of the score units, scales and offsets, which can hinder the combination of results from different structures or programs. Our method is simple and robust, providing a theoretical basis not only for molecular docking but also for any consensus strategy in general.

Maximal Unbiased Benchmarking Data Sets for Human Chemokine Receptors and Comparative Analysis

Chemokine receptors (CRs) have long been druggable targets for the treatment of inflammatory diseases and HIV-1 infection. As a powerful technique, virtual screening (VS) has been widely applied to identifying small molecule leads for modern drug targets including CRs. For rational selection of a wide variety of VS approaches, ligand enrichment assessment based on a benchmarking data set has become an indispensable practice. However, the lack of versatile benchmarking sets for the whole CRs family that are able to unbiasedly evaluate every single approach including both structure- and ligand-based VS somewhat hinders modern drug discovery efforts. To address this issue, we constructed Maximal Unbiased Benchmarking Data sets for human Chemokine Receptors (MUBD-hCRs) using our recently developed tools of MUBD-DecoyMaker. The MUBD-hCRs encompasses 13 subtypes out of 20 chemokine receptors, composed of 404 ligands and 15756 decoys so far and is readily expandable in the future. It had been thoroughly validated that MUBD-hCRs ligands are chemically diverse while its decoys are maximal unbiased in terms of "artificial enrichment", "analogue bias". In addition, we studied the performance of MUBD-hCRs, in particular CXCR4 and CCR5 data sets, in ligand enrichment assessments of both structure- and ligand-based VS approaches in comparison with other benchmarking data sets available in the public domain and demonstrated that MUBD-hCRs is very capable of designating the optimal VS approach. MUBD-hCRs is a unique and maximal unbiased benchmarking set that covers major CRs subtypes so far.

Mechanism of the G-protein mimetic nanobody binding to a muscarinic G-protein-coupled receptor

Protein-protein binding is key in cellular signaling processes. Molecular dynamics (MD) simulations of protein-protein binding, however, are challenging due to limited timescales. In particular, binding of the medically important G-protein-coupled receptors (GPCRs) with intracellular signaling proteins has not been simulated with MD to date. Here, we report a successful simulation of the binding of a G-protein mimetic nanobody to the M₂ muscarinic GPCR using the robust Gaussian accelerated MD (GaMD) method. Through long-timescale GaMD simulations over 4,500 ns, the nanobody was observed to bind the receptor intracellular G-protein-coupling site, with a minimum rmsd of 2.48 Å in the nanobody core domain compared with the X-ray structure. Binding of the nanobody allosterically closed the orthosteric ligand-binding pocket, being consistent with the recent experimental finding. In the absence of nanobody binding, the receptor orthosteric pocket sampled open and fully open conformations. The GaMD simulations revealed two low-energy intermediate states during nanobody binding to the M₂ receptor. The flexible receptor intracellular loops contribute remarkable electrostatic, polar, and hydrophobic residue interactions in recognition and binding of the nanobody. These simulations provided important insights into the mechanism of GPCR-nanobody binding and demonstrated the applicability of GaMD in modeling dynamic protein-protein interactions.

Decoys Selection in Benchmarking Datasets: Overview and Perspectives

Virtual Screening (VS) is designed to prospectively help identifying potential hits, i.e., compounds capable of interacting with a given target and potentially modulate its activity, out of large compound collections. Among the variety of methodologies, it is crucial to select the protocol that is the most adapted to the query/target system under study and that yields the most reliable output. To this aim, the performance of VS methods is commonly evaluated and compared by computing their ability to retrieve active compounds in benchmarking datasets. The benchmarking datasets contain a subset of known active compounds together with a subset of decoys, i.e., assumed non-active molecules. The composition of both the active and the decoy compounds subsets is critical to limit the biases in the evaluation of the VS methods. In this review, we focus on the selection of decoy compounds that has considerably changed over the years, from randomly selected compounds to highly customized or experimentally validated negative compounds. We first outline the evolution of decoys selection in benchmarking databases as well as current benchmarking databases that tend to minimize the introduction of biases, and secondly, we propose recommendations for the selection and the design of benchmarking datasets.

Improving virtual screening of G protein-coupled receptors via ligand-directed modeling

G protein-coupled receptors (GPCRs) play crucial roles in cell physiology and pathophysiology. There is increasing interest in using structural information for virtual screening (VS) of libraries and for structure-based drug design to identify novel agonist or antagonist leads. However, the sparse availability of experimentally determined GPCR/ligand complex structures with diverse ligands impedes the application of structure-based drug design (SBDD) programs directed to identifying new molecules with a select pharmacology. In this study, we apply ligand-directed modeling (LDM) to available GPCR X-ray structures to improve VS performance and selectivity towards molecules of specific pharmacological profile. The described method refines a GPCR binding pocket conformation using a single known ligand for that GPCR. The LDM method is a computationally efficient, iterative workflow consisting of protein sampling and ligand docking. We developed an extensive benchmark comparing LDM-refined binding pockets to GPCR X-ray crystal structures across seven different GPCRs bound to a range of ligands of different chemotypes and pharmacological profiles. LDM-refined models showed improvement in VS performance over origin X-ray crystal structures in 21 out of 24 cases. In all cases, the LDM-refined models had superior performance in enriching for the chemotype of the refinement ligand. This likely contributes to the LDM success in all cases of inhibitor-bound to agonist-bound binding pocket refinement, a key task for GPCR SBDD programs. Indeed, agonist ligands are required for a plethora of GPCRs for therapeutic intervention, however GPCR X-ray structures are mostly restricted to their inactive inhibitor-bound state.

G protein-coupled receptors (GPCRs) are a major target for drug discovery. These receptors are highly dynamic membrane proteins, and have had limited tractability using with biophysical screens that are widely adopted for globular protein targets. Thus, structure-based virtual screening (SBVS) holds great promise as a complement to physical screening for rational design of novel drugs. Indeed, the increasing number of atomic-detail GPCR X-ray crystal structures has coincided with an increase in prospective SBVS studies that have identified novel compounds. However, experimentally solved GPCR structures do not meet the full demand for SBVS, as the GPCR structural landscape is incomplete, lacking both in coverage of available GPCRs, and diversity in both receptor conformations and the chemistry of co-crystalised ligands. Here we present a novel computational GPCR binding pocket refinement method that can generate predictive GPCR/ligand complexes with improved SBVS performance. This ligand-directed modeling workflow uses parallel processing and efficient algorithms to search the GPCR/ligand conformational space faster and more efficiently than the widely used protein refinement method molecular dynamics. In this study, the resulting models are evaluated both structurally, and in retrospective SBVS. We demonstrate improved performance of refined models over their starting structures in the majority of our test cases.

Hybrid approach to structure modeling of the histamine H3 receptor: Multi-level assessment as a tool for model verification

The crucial role of G-protein coupled receptors and the significant achievements associated with a better understanding of the spatial structure of known receptors in this family encouraged us to undertake a study on the histamine H3 receptor, whose crystal structure is still unresolved. The latest literature data and availability of different software enabled us to build homology models of higher accuracy than previously published ones. The new models are expected to be closer to crystal structures; and therefore, they are much more helpful in the design of potential ligands. In this article, we describe the generation of homology models with the use of diverse tools and a hybrid assessment. Our study incorporates a hybrid assessment connecting knowledge-based scoring algorithms with a two-step ligand-based docking procedure. Knowledge-based scoring employs probability theory for global energy minimum determination based on information about native amino acid conformation from a dataset of experimentally determined protein structures. For a two-step docking procedure two programs were applied: GOLD was used in the first step and Glide in the second. Hybrid approaches offer advantages by combining various theoretical methods in one modeling algorithm. The biggest advantage of hybrid methods is their intrinsic ability to self-update and self-refine when additional structural data are acquired. Moreover, the diversity of computational methods and structural data used in hybrid approaches for structure prediction limit inaccuracies resulting from theoretical approximations or fuzziness of experimental data. The results of docking to the new H3 receptor model allowed us to analyze ligand-receptor interactions for reference compounds.

VSPrep: A General KNIME Workflow for the Preparation of Molecules for Virtual Screening

Over the past decades, virtual screening has proved itself to be a valuable asset to identify new bioactive compounds. The vast majority of commonly used techniques can be described in three steps: pre-processing the dataset i. e. small (ligands) and eventually larger (receptors) molecules, execute the method and finally analyse the results. Hence, the preparation of ligands is a critical step for success of commonly used virtual screening approaches such as protein-ligand docking, similarity or pharmacophore search. We present here a new workflow, VSPrep, for the pre-processing of small molecules; it is based on freely accessible tools for academics and is integrated within the KNIME platform. It can be used to perform several chemoinformatics tasks such as molecular database cleaning, tautomer and stereoisomer enumeration, focused library design and conformer generation. Additionally, graphical reports of the results are provided to the user as a convenient analysis tool.

Discovery of Novel and Selective Adenosine A2A Receptor Antagonists for Treating Parkinson's Disease through Comparative Structure-Based Virtual Screening

Among non-dopaminergic strategies for combating Parkinson's disease (PD), antagonism of the A_2A adenosine receptor (AR) has emerged to show great potential. In this study, on the basis of two crystal structures of the A_2A AR with the best capability to distinguish known antagonists from decoys, docking-based virtual screening (VS) was conducted to identify novel A_2A AR antagonists. A total of 63 structurally diverse compounds identified by VS were submitted to experimental testing, and 11 of them exhibited substantial activity against the A_2A AR (K_i < 10 μM), including two compounds with K_i below 1 μM (compound 43, 0.42 μM; compound 51, 0.27 μM) and good A_2A/A₁ selectivity (fold < 0.1). Compounds 43 and 51 demonstrated antagonistic activity according to the results of cAMP measurements (cAMP IC₅₀ = 1.67 and 1.80 μM, respectively) and showed good efficacy in the haloperidol-induced catalepsy (HIC) rat model for PD at doses of up to 30 mg/kg. Further lead optimization based on a substructure searching strategy led to the discovery of compound 84 as an excellent A_2A AR antagonist (A_2A K_i = 54 nM, A_2A/A₁ fold < 0.1, cAMP IC₅₀ = 0.3 μM) that exhibited significant improvement in anti-PD efficacy in the HIC rat model.

Structural features embedded in G protein-coupled receptor co-crystal structures are key to their success in virtual screening

Structure based drug discovery on GPCRs harness atomic detail X-ray binding pockets and large libraries of potential drug lead candidates in virtual screening (VS) to identify novel lead candidates. Relatively small conformational differences between such binding pockets can be critical to the success of VS. Retrospective VS on GPCR/ligand co-crystal structures revealed stark differences in the ability of different structures to identify known ligands, despite being co-crystallized with the same ligand. When using the OpenEye toolkit and the ICM modeling package, we identify criteria associated with the predictive power of binding pockets in VS that consists of a combination of ligand/receptor interaction pattern and predicted ligand/receptor interaction strength. These findings can guide the selection and refinement of GPCR binding pockets for use in SBDD programs and may also provide a potential framework for evaluating the ability of computational GPCR binding pocket refinement tools in improving the predictive power of binding pockets.

Identification of the Beer Component Hordenine as Food-Derived Dopamine D2 Receptor Agonist by Virtual Screening a 3D Compound Database

The dopamine D2 receptor (D2R) is involved in food reward and compulsive food intake. The present study developed a virtual screening (VS) method to identify food components, which may modulate D2R signalling. In contrast to their common applications in drug discovery, VS methods are rarely applied for the discovery of bioactive food compounds. Here, databases were created that exclusively contain substances occurring in food and natural sources (about 13,000 different compounds in total) as the basis for combined pharmacophore searching, hit-list clustering and molecular docking into D2R homology models. From 17 compounds finally tested in radioligand assays to determine their binding affinities, seven were classified as hits (hit rate = 41%). Functional properties of the five most active compounds were further examined in β-arrestin recruitment and cAMP inhibition experiments. D2R-promoted G-protein activation was observed for hordenine, a constituent of barley and beer, with approximately identical ligand efficacy as dopamine (76%) and a Ki value of 13 μM. Moreover, hordenine antagonised D2-mediated β-arrestin recruitment indicating functional selectivity. Application of our databases provides new perspectives for the discovery of bioactive food constituents using VS methods. Based on its presence in beer, we suggest that hordenine significantly contributes to mood-elevating effects of beer.

Three-Dimensional Biologically Relevant Spectrum (BRS-3D): Shape Similarity Profile Based on PDB Ligands as Molecular Descriptors

The crystallized ligands in the Protein Data Bank (PDB) can be treated as the inverse shapes of the active sites of corresponding proteins. Therefore, the shape similarity between a molecule and PDB ligands indicated the possibility of the molecule to bind with the targets. In this paper, we proposed a shape similarity profile that can be used as a molecular descriptor for ligand-based virtual screening. First, through three-dimensional (3D) structural clustering, 300 diverse ligands were extracted from the druggable protein-ligand database, sc-PDB. Then, each of the molecules under scrutiny was flexibly superimposed onto the 300 ligands. Superimpositions were scored by shape overlap and property similarity, producing a 300 dimensional similarity array termed the "Three-Dimensional Biologically Relevant Spectrum (BRS-3D)". Finally, quantitative or discriminant models were developed with the 300 dimensional descriptor using machine learning methods (support vector machine). The effectiveness of this approach was evaluated using 42 benchmark data sets from the G protein-coupled receptor (GPCR) ligand library and the GPCR decoy database (GLL/GDD). We compared the performance of BRS-3D with other 2D and 3D state-of-the-art molecular descriptors. The results showed that models built with BRS-3D performed best for most GLL/GDD data sets. We also applied BRS-3D in histone deacetylase 1 inhibitors screening and GPCR subtype selectivity prediction. The advantages and disadvantages of this approach are discussed.

Getting the most out of PubChem for virtual screening

INTRODUCTION

With the emergence of the 'big data' era, the biomedical research community has great interest in exploiting publicly available chemical information for drug discovery. PubChem is an example of public databases that provide a large amount of chemical information free of charge.

AREAS COVERED

This article provides an overview of how PubChem's data, tools, and services can be used for virtual screening and reviews recent publications that discuss important aspects of exploiting PubChem for drug discovery.

EXPERT OPINION

PubChem offers comprehensive chemical information useful for drug discovery. It also provides multiple programmatic access routes, which are essential to build automated virtual screening pipelines that exploit PubChem data. In addition, PubChemRDF allows users to download PubChem data and load them into a local computing facility, facilitating data integration between PubChem and other resources. PubChem resources have been used in many studies for developing bioactivity and toxicity prediction models, discovering polypharmacologic (multi-target) ligands, and identifying new macromolecule targets of compounds (for drug-repurposing or off-target side effect prediction). These studies demonstrate the usefulness of PubChem as a key resource for computer-aided drug discovery and related area.

GPCR-Bench: A Benchmarking Set and Practitioners' Guide for G Protein-Coupled Receptor Docking

Virtual screening is routinely used to discover new ligands and in particular new ligand chemotypes for G protein-coupled receptors (GPCRs). To prepare for a virtual screen, we often tailor a docking protocol that will enable us to select the best candidates for further screening. To aid this, we created GPCR-Bench, a publically available docking benchmarking set in the spirit of the DUD and DUD-E reference data sets for validation studies, containing 25 nonredundant high-resolution GPCR costructures with an accompanying set of diverse ligands and computational decoy molecules for each target. Benchmarking sets are often used to compare docking protocols; however, it is important to evaluate docking methods not by "retrospective" hit rates but by the actual likelihood that they will produce novel prospective hits. Therefore, docking protocols must not only rank active molecules highly but also produce good poses that a chemist will select for purchase and screening. Currently, no simple objective machine-scriptable function exists that can do this; instead, docking hit lists must be subjectively examined in a consistent way to compare between docking methods. We present here a case study highlighting considerations we feel are of importance when evaluating a method, intended to be useful as a practitioners' guide.

Pharmacophore-Map-Pick: A Method to Generate Pharmacophore Models for All Human GPCRs

GPCR-based drug discovery is hindered by a lack of effective screening methods for most GPCRs that have neither ligands nor high-quality structures. With the aim to identify lead molecules for these GPCRs, we developed a new method called Pharmacophore-Map-Pick to generate pharmacophore models for all human GPCRs. The model of ADRB2 generated using this method not only predicts the binding mode of ADRB2-ligands correctly but also performs well in virtual screening. Findings also demonstrate that this method is powerful for generating high-quality pharmacophore models. The average enrichment for the pharmacophore models of the 15 targets in different GPCR families reached 15-fold at 0.5 % false-positive rate. Therefore, the pharmacophore models can be applied in virtual screening directly with no requirement for any ligand information or shape constraints. A total of 2386 pharmacophore models for 819 different GPCRs (99 % coverage (819/825)) were generated and are available at http://bsb.kiz.ac.cn/GPCRPMD.

Dynamics and structural determinants of ligand recognition of the 5-HT6 receptor

In order to identify molecular models of the human 5-HT6 receptor suitable for virtual screening, homology modeling and membrane-embedded molecular dynamics simulations were performed. Structural requirements for robust enrichment were assessed by an unbiased chemometric analysis of enrichments from retrospective virtual screening studies. The two main structural features affecting enrichment are the outward movement of the second extracellular loop and the formation of a hydrophobic cavity deep in the binding site. These features appear transiently in the trajectories and furthermore the stretches of uniformly high enrichment may only last 4-10 ps. The formation of the inner hydrophobic cavity was also linked to the active-like to inactive-like transition of the receptor, especially the so-called connector region. The best structural models provided significant and robust enrichment over three independent ligand sets.

In Silico Investigation of the Neurotensin Receptor 1 Binding Site: Overlapping Binding Modes for Small Molecule Antagonists and the Endogenous Peptide Agonist

The neurotensin receptor 1 (NTSR1) belongs to the family of 7TM, G protein-coupled receptors, and is activated by the 13-amino-acid peptide neurotensin (NTS) that has been shown to play important roles in neurological disorders and the promotion of cancer cells. Recently, a high-resolution x-ray crystal structure of NTSR1 in complex with NTS8-13 has been determined, providing novel insights into peptide ligand recognition by 7TM receptors. SR48692, a potent and selective small molecule antagonist has previously been used extensively as a tool compound to study NTSR1 receptor signaling properties. To investigate the binding mode of SR48692 and other small molecule compounds to NTSR1, we applied an Automated Ligand-guided Backbone Ensemble Receptor Optimization protocol (ALiBERO), taking receptor flexibility and ligand knowledge into account. Structurally overlapping binding poses for SR48692 and NTS8-13 were observed, despite their distinct chemical nature and inverse pharmacological profiles. The optimized models showed significantly improved ligand recognition in a large-scale virtual screening assessment compared to the crystal structure. Our models provide new insights into small molecule ligand binding to NTSR1 and could facilitate the structure-based design of non-peptide ligands for the evaluation of the pharmacological potential of NTSR1 in neurological disorders and cancer.