An automated system for the analysis of G protein-coupled receptor transmembrane binding pockets: alignment, receptor-based pharmacophores, and their application

GPCRLigNet: rapid screening for GPCR active ligands using machine learning

Molecules with bioactivity towards G protein-coupled receptors represent a subset of the vast space of small drug-like molecules. Here, we compare machine learning models, including dilated graph convolutional networks, that conduct binary classification to quickly identify molecules with activity towards G protein-coupled receptors. The models are trained and validated using a large set of over 600,000 active, inactive, and decoy compounds. The best performing machine learning model, dubbed GPCRLigNet, was a surprisingly simple feedforward dense neural network mapping from Morgan fingerprints to activity. Incorporation of GPCRLigNet into a high-throughput virtual screening workflow is demonstrated with molecular docking towards a particular G protein-coupled receptor, the pituitary adenylate cyclase-activating polypeptide receptor type 1. Through rigorous comparison of docking scores for molecules selected with and without using GPCRLigNet, we demonstrate an enrichment of potentially potent molecules using GPCRLigNet. This work provides a proof of principle that GPCRLigNet can effectively hone the chemical search space towards ligands with G protein-coupled receptor activity.

Truly Target-Focused Pharmacophore Modeling: A Novel Tool for Mapping Intermolecular Surfaces

Pharmacophore models are an accurate and minimal tridimensional abstraction of intermolecular interactions between chemical structures, usually derived from a group of molecules or from a ligand-target complex. Only a limited amount of solutions exists to model comprehensive pharmacophores using the information of a particular target structure without knowledge of any binding ligand. In this work, an automated and customable tool for truly target-focused (T²F) pharmacophore modeling is introduced. Key molecular interaction fields of a macromolecular structure are calculated using the AutoGRID energy functions. The most relevant points are selected by a newly developed filtering cascade and clustered to pharmacophore features with a density-based algorithm. Using five different protein classes, the ability of this method to identify essential pharmacophore features was compared to structure-based pharmacophores derived from ligand-target interactions. This method represents an extremely valuable instrument for drug design in a situation of scarce ligand information available, but also in the case of underexplored therapeutic targets, as well as to investigate protein allosteric pockets and protein-protein interactions.

Linear and Kernel Model Construction Methods for Predicting Drug-Target Interactions in a Chemogenomic Framework

Identification of drug-target interactions is a crucial process in drug discovery. In this chapter, we present protocols for recent advancements in machine learning methods for predicting drug-target interactions from heterogeneous biological data in a chemogenomic framework, in which prediction is based on the chemical structure data of drug candidate compounds and translated genomic sequence data of target candidate proteins. Most existing methods are based on either linear modeling or kernel modeling. To illustrate linear modeling, we introduce sparsity-induced binary classifiers and sparse canonical correlation analysis. To illustrate kernel modeling, we introduce pairwise kernel-based support vector machines and kernel-based distance learning. Workflows for using these techniques are presented. We also discuss the characteristics of each method and suggest some directions for future research.

A Structural Framework for GPCR Chemogenomics: What's In a Residue Number?

The recent surge of crystal structures of G protein-coupled receptors (GPCRs), as well as comprehensive collections of sequence, structural, ligand bioactivity, and mutation data, has enabled the development of integrated chemogenomics workflows for this important target family. This chapter will focus on cross-family and cross-class studies of GPCRs that have pinpointed the need for, and the implementation of, a generic numbering scheme for referring to specific structural elements of GPCRs. Sequence- and structure-based numbering schemes for different receptor classes will be introduced and the remaining caveats will be discussed. The use of these numbering schemes has facilitated many chemogenomics studies such as consensus binding site definition, binding site comparison, ligand repurposing (e.g. for orphan receptors), sequence-based pharmacophore generation for homology modeling or virtual screening, and class-wide chemogenomics studies of GPCRs.

Sparse Modeling to Analyze Drug-Target Interaction Networks

Most drugs produce their phenotypic effects by interacting with target proteins, and understanding the molecular features that underpin drug-target interactions is crucial when designing a novel drug. In this chapter, we introduce the protocols that have driven recent advances in sparse modeling methods for analyzing drug-target interaction networks within a chemogenomic framework. In this approach, the chemical structures of candidate drug compounds are correlated with the genomic sequences of the candidate target proteins. We demonstrate the use of sparse canonical correspondence analysis and sparsity-induced binary classifiers to extract the underlying molecular features that are most strongly involved in drug-target interactions. We focus on drug chemical substructures and protein domains. Workflows for applying these methods are presented, and an application is described in detail. We consider the characteristics of each method and suggest possible directions for future research.

Reprint of Pharmacological and molecular characterization of the positive allosteric modulators of metabotropic glutamate receptor 2

The metabotropic glutamate receptor 2 (mGlu₂) plays an important role in the presynaptic control of glutamate release and several mGlu₂ positive allosteric modulators (PAMs) have been under assessment for their potential as antipsychotics. The binding mode of mGlu₂ PAMs is better characterized in functional terms while few data are available on the relationship between allosteric and orthosteric binding sites. Pharmacological studies characterizing binding and effects of two different chemical series of mGlu₂ PAMs are therefore carried out here using the radiolabeled mGlu₂ agonist ³[H]-LY354740 and mGlu₂ PAM ³[H]-2,2,2-TEMPS. A multidimensional approach to the PAM mechanism of action shows that mGlu₂ PAMs increase the affinity of ³[H]-LY354740 for the orthosteric site of mGlu2 as well as the number of ³[H]-LY354740 binding sites. ³[H]-2,2,2-TEMPS binding is also enhanced by the presence of LY354740. New residues in the allosteric rat mGlu2 binding pocket are identified to be crucial for the PAMs ligand binding, among these Tyr^3.40 and Asn^5.46. Also of remark, in the described experimental conditions S731A (Ser^5.42) residue is important only for the mGlu₂ PAM LY487379 and not for the compound PAM-1: an example of the structural differences among these mGlu₂ PAMs. This study provides a summary of the information generated in the past decade on mGlu₂ PAMs adding a detailed molecular investigation of PAM binding mode. Differences among mGlu₂ PAM compounds are discussed as well as the mGlu2 regions interacting with mGlu₂ PAM and NAM agents and residues driving mGlu2 PAM selectivity. This article is part of the Special Issue entitled 'Metabotropic Glutamate Receptors, 5 years on'.

Model-Based Discovery of Synthetic Agonists for the Zn2+-Sensing G-Protein-Coupled Receptor 39 (GPR39) Reveals Novel Biological Functions

The G-protein-coupled receptor 39 (GPR39) is a G-protein-coupled receptor activated by Zn²⁺. We used a homology model-based approach to identify small-molecule pharmacological tool compounds for the receptor. The method focused on a putative binding site in GPR39 for synthetic ligands and knowledge of ligand binding to other receptors with similar binding pockets to select iterative series of minilibraries. These libraries were cherry-picked from all commercially available synthetic compounds. A total of only 520 compounds were tested in vitro, making this method broadly applicable for tool compound development. The compounds of the initial library were inactive when tested alone, but lead compounds were identified using Zn²⁺ as an allosteric enhancer. Highly selective, highly potent Zn²⁺-independent GPR39 agonists were found in subsequent minilibraries. These agonists identified GPR39 as a novel regulator of gastric somatostatin secretion.

A Real-World Perspective on Molecular Design

We present a series of small molecule drug discovery case studies where computational methods were prospectively employed to impact Roche research projects, with the aim of highlighting those methods that provide real added value. Our brief accounts encompass a broad range of methods and techniques applied to a variety of enzymes and receptors. Most of these are based on judicious application of knowledge about molecular conformations and interactions: filling of lipophilic pockets to gain affinity or selectivity, addition of polar substituents, scaffold hopping, transfer of SAR, conformation analysis, and molecular overlays. A case study of sequence-driven focused screening is presented to illustrate how appropriate preprocessing of information enables effective exploitation of prior knowledge. We conclude that qualitative statements enabling chemists to focus on promising regions of chemical space are often more impactful than quantitative prediction.

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.

3-iodothyronamine differentially modulates α-2A-adrenergic receptor-mediated signaling

Most in vivo effects of 3-iodothyronamine (3-T1AM) have been thus far thought to be mediated by binding at the trace amine-associated receptor 1 (TAAR1). Inconsistently, the 3-T1AM-induced hypothermic effect still persists in Taar1 knockout mice, which suggests additional receptor targets. In support of this general assumption, it has previously been reported that 3-T1AM also binds to the α-2A-adrenergic receptor (ADRA2A), which modulates insulin secretion. However, the mechanism of this effect remains unclear. We tested two different scenarios that may explain the effect: the sole action of 3-T1AM at ADRA2A and a combined action of 3-T1AM at ADRA2A and TAAR1, which is also expressed in pancreatic islets. We first investigated a potential general signaling modification using the label-free EPIC technology and then specified changes in signaling by cAMP inhibition and MAPKs (ERK1/2) determination. We found that 3-T1AM induced Gi/o activation at ADRA2A and reduced the norepinephrine (NorEpi)-induced MAPK activation. Interestingly, in ADRA2A/TAAR1 hetero-oligomers, application of NorEpi resulted in uncoupling of the Gi/o signaling pathway, but it did not affect MAPK activation. However, 3-T1AM application in mice over a period of 6 days at a daily dose of 5 mg/kg had no significant effects on glucose homeostasis. In summary, we report an agonistic effect of 3-T1AM on the ADRA2A-mediated Gi/o pathway but an antagonistic effect on MAPK induced by NorEpi. Moreover, in ADRA2A/TAAR1 hetero-oligomers, the capacity of NorEpi to stimulate Gi/o signaling is reduced by co-stimulation with 3-T1AM. The present study therefore points to a complex spectrum of signaling modification mediated by 3-T1AM at different G protein-coupled receptors.

Chemical genomics approach for GPCR-ligand interaction prediction and extraction of ligand binding determinants

Chemical genomics research has revealed that G-protein coupled receptors (GPCRs) interact with a variety of ligands and that a large number of ligands are known to bind GPCRs even with low transmembrane (TM) sequence similarity. It is crucial to extract informative binding region propensities from large quantities of bioactivity data. To address this issue, we propose a machine learning approach that enables identification of both chemical substructures and amino acid properties that are associated with ligand binding, which can be applied to virtual ligand screening on a GPCR-wide scale. We also address the question of how to select plausible negative noninteraction pairs based on a statistical approach in order to develop reliable prediction models for GPCR-ligand interactions. The key interaction sites estimated by our approach can be of great use not only for screening of active compounds but also for modification of active compounds with the aim of improving activity or selectivity.

Chemogenomic approaches to infer drug-target interaction networks

The identification of drug-target interactions from heterogeneous biological data is critical in the drug development. In this chapter, we review recently developed in silico chemogenomic approaches to infer unknown drug-target interactions from chemical information of drugs and genomic information of target proteins. We review several kernel-based statistical methods from two different viewpoints: binary classification and dimension reduction. In the results, we demonstrate the usefulness of the methods on the prediction of drug-target interactions from chemical structure data and genomic sequence data. We also discuss the characteristics of each method, and show some perspectives toward future research direction.

Signatures of the ATP-binding pocket as a basis for structural classification of the serine/threonine protein kinases of gram-positive bacteria

Eukaryotic-like serine/threonine protein kinases (ESTPKs) are widely spread throughout the bacterial genomes. These enzymes can be potential targets of new antibacterial drugs useful for the treatment of socially important diseases such as tuberculosis. In this study, ESTPKs of pathogenic, probiotic, and antibiotic-producing Gram-positive bacteria were classified according to the physicochemical properties of amino acid residues in the ATP-binding site of the enzyme. Nine residues were identified that line the surface of the adenine-binding pocket, and ESTPKs were classified based on these signatures. Twenty groups were discovered, five of them containing >10 representatives. The two most abundant groups contained >150 protein kinases that belong to the various branches of the phylogenetic tree, whereas certain groups are genus- or even species-specific. Homology modeling of the typical representatives of each group revealed that the classification is reliable, and the differences between the protein kinase ATP-binding pockets predicted based on their signatures are apparent in their structure. The classification is expected to be useful for the selection of targets for new anti-infective drugs.

Prediction of Compound-protein Interactions with Machine Learning Methods

In silico prediction of compound-protein interactions from heterogeneous biological data is critical in the process of drug development. In this chapter the authors review several supervised machine learning methods to predict unknown compound-protein interactions from chemical structure and genomic sequence information simultaneously. The authors review several kernel-based algorithms from two different viewpoints: binary classification and dimension reduction. In the results, they demonstrate the usefulness of the methods on the prediction of drug-target interactions and ligand-protein interactions from chemical structure data and genomic sequence data.

Differential modulation of Beta-adrenergic receptor signaling by trace amine-associated receptor 1 agonists

Trace amine-associated receptors (TAAR) are rhodopsin-like G-protein-coupled receptors (GPCR). TAAR are involved in modulation of neuronal, cardiac and vascular functions and they are potentially linked with neurological disorders like schizophrenia and Parkinson's disease. Subtype TAAR1, the best characterized TAAR so far, is promiscuous for a wide set of ligands and is activated by trace amines tyramine (TYR), phenylethylamine (PEA), octopamine (OA), but also by thyronamines, dopamine, and psycho-active drugs. Unfortunately, effects of trace amines on signaling of the two homologous β-adrenergic receptors 1 (ADRB1) and 2 (ADRB2) have not been clarified yet in detail. We, therefore, tested TAAR1 agonists TYR, PEA and OA regarding their effects on ADRB1/2 signaling by co-stimulation studies. Surprisingly, trace amines TYR and PEA are partial allosteric antagonists at ADRB1/2, whereas OA is a partial orthosteric ADRB2-antagonist and ADRB1-agonist. To specify molecular reasons for TAAR1 ligand promiscuity and for observed differences in signaling effects on particular aminergic receptors we compared TAAR, tyramine (TAR) octopamine (OAR), ADRB1/2 and dopamine receptors at the structural level. We found especially for TAAR1 that the remarkable ligand promiscuity is likely based on high amino acid similarity in the ligand-binding region compared with further aminergic receptors. On the other hand few TAAR specific properties in the ligand-binding site might determine differences in ligand-induced effects compared to ADRB1/2. Taken together, this study points to molecular details of TAAR1-ligand promiscuity and identified specific trace amines as allosteric or orthosteric ligands of particular β-adrenergic receptor subtypes.

Ligand discovery from a dopamine D3 receptor homology model and crystal structure

G protein-coupled receptors (GPCRs) are intensely studied as drug targets and for their role in signaling. With the determination of the first crystal structures, interest in structure-based ligand discovery increased. Unfortunately, for most GPCRs no experimental structures are available. The determination of the D(3) receptor structure and the challenge to the community to predict it enabled a fully prospective comparison of ligand discovery from a modeled structure versus that of the subsequently released crystal structure. Over 3.3 million molecules were docked against a homology model, and 26 of the highest ranking were tested for binding. Six had affinities ranging from 0.2 to 3.1 μM. Subsequently, the crystal structure was released and the docking screen repeated. Of the 25 compounds selected, five had affinities ranging from 0.3 to 3.0 μM. One of the new ligands from the homology model screen was optimized for affinity to 81 nM. The feasibility of docking screens against modeled GPCRs more generally is considered.

Snooker: a structure-based pharmacophore generation tool applied to class A GPCRs

G-protein coupled receptors (GPCRs) are important drug targets for various diseases and of major interest to pharmaceutical companies. The function of individual members of this protein family can be modulated by the binding of small molecules at the extracellular side of the structurally conserved transmembrane (TM) domain. Here, we present Snooker, a structure-based approach to generate pharmacophore hypotheses for compounds binding to this extracellular side of the TM domain. Snooker does not require knowledge of ligands, is therefore suitable for apo-proteins, and can be applied to all receptors of the GPCR protein family. The method comprises the construction of a homology model of the TM domains and prioritization of residues on the probability of being ligand binding. Subsequently, protein properties are converted to ligand space, and pharmacophore features are generated at positions where protein ligand interactions are likely. Using this semiautomated knowledge-driven bioinformatics approach we have created pharmacophore hypotheses for 15 different GPCRs from several different subfamilies. For the beta-2-adrenergic receptor we show that ligand poses predicted by Snooker pharmacophore hypotheses reproduce literature supported binding modes for ∼75% of compounds fulfilling pharmacophore constraints. All 15 pharmacophore hypotheses represent interactions with essential residues for ligand binding as observed in mutagenesis experiments and compound selections based on these hypotheses are shown to be target specific. For 8 out of 15 targets enrichment factors above 10-fold are observed in the top 0.5% ranked compounds in a virtual screen. Additionally, prospectively predicted ligand binding poses in the human dopamine D3 receptor based on Snooker pharmacophores were ranked among the best models in the community wide GPCR dock 2010.

Docking-based virtual screening of potential human P2Y12 receptor antagonists

Platelet plays essential roles in hemostasis and its dysregulation can lead to arterial thrombosis. P2Y12 is an important platelet membrane adenosine diphosphate receptor, and its antagonists have been widely developed as anti-coagulation agents. The current P2Y12 inhibitors available in clinical practice have not fully achieved satisfactory anti-thrombotic effects, leaving room for further improvement. To identify new chemical compounds as potential anti-coagulation inhibitors, we constructed a three-dimensional structure model of human P2Y12 by homology modeling based on the recently reported G-protein coupled receptor Meleagris gallopavo β1 adrenergic receptor. Virtual screening of the modeled P2Y12 against three subsets of small molecules from the ZINC database, namely lead-like, fragment-like, and drug-like, identified a number of compounds that might have high binding affinity to P2Y12. Detailed analyses of the top three compounds from each subset with the highest scores indicated that all of these compounds beard a hydrophobic bulk supplemented with a few polar atoms which bound at the ligand binding site via largely hydrophobic interactions with the receptor. This study not only provides a structure model of P2Y12 for rational design of anti-platelet inhibitors, but also identifies some potential chemicals for further development.

Allosteric modulation of metabotropic glutamate receptors: structural insights and therapeutic potential

Allosteric modulation of G protein-coupled receptors (GPCRs) represents a novel approach to the development of probes and therapeutics that is expected to enable subtype-specific regulation of central nervous system target receptors. The metabotropic glutamate receptors (mGlus) are class C GPCRs that play important neuromodulatory roles throughout the brain, as such they are attractive targets for therapeutic intervention for a number of psychiatric and neurological disorders including anxiety, depression, Fragile X Syndrome, Parkinson's disease and schizophrenia. Over the last fifteen years, selective allosteric modulators have been identified for many members of the mGlu family. The vast majority of these allosteric modulators are thought to bind within the transmembrane-spanning domains of the receptors to enhance or inhibit functional responses. A combination of mutagenesis-based studies and pharmacological approaches are beginning to provide a better understanding of mGlu allosteric sites. Collectively, when mapped onto a homology model of the different mGlu subtypes based on the β(2)-adrenergic receptor, the previous mutagenesis studies suggest commonalities in the location of allosteric sites across different members of the mGlu family. In addition, there is evidence for multiple allosteric binding pockets within the transmembrane region that can interact to modulate one another. In the absence of a class C GPCR crystal structure, this approach has shown promise with respect to the interpretation of mutagenesis data and understanding structure-activity relationships of allosteric modulator pharmacophores.

The family of G protein-coupled receptors: an example of membrane proteins

The G protein coupled receptors belong to the largest group of membrane proteins that regulates many essential physiological properties and represents an important class of drug targets. In this chapter, we show how the synergy between a laboratory experiment and computational modeling leads to structural delineation of the ligand binding pocket and how the knowledge of ligand-protein interactions is used for rational local and global drug design in which the structural knowledge of a particular receptor and its ligands is used for drug design of this particular GPCR and others.

Definition of the G protein-coupled receptor transmembrane bundle binding pocket and calculation of receptor similarities for drug design

Recent advances in structural biology for G-protein-coupled receptors (GPCRs) have provided new opportunities to improve the definition of the transmembrane binding pocket. Here a reference set of 44 residue positions accessible for ligand binding was defined through detailed analysis of all currently available crystal structures. This was used to characterize pharmacological relationships of Family A/Rhodopsin family GPCRs, minimizing evolutionary influence from parts of the receptor that do not generally affect ligand binding. The resultant dendogram tended to group receptors according to endogenous ligand types, although it revealed subdivision of certain classes, notably peptide and lipid receptors. The transmembrane binding site reference set, particularly when coupled with a means of identifying the subset of ligand binding residues, provides a general paradigm for understanding the pharmacology/selectivity profile of ligands at Family A GPCRs. This has wide applicability to GPCR drug design problems across many disease areas.

Transmembrane helical domain of the cannabinoid CB1 receptor

Brain cannabinoid (CB(1)) receptors are G-protein coupled receptors and belong to the rhodopsin-like subfamily. A homology model of the inactive state of the CB(1) receptor was constructed using the x-ray structure of beta(2)-adrenergic receptor (beta(2)AR) as the template. We used 105 ns duration molecular-dynamics simulations of the CB(1) receptor embedded in a 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) bilayer to gain some insight into the structure and function of the CB(1) receptor. As judged from the root mean-square deviations combined with the detailed structural analyses, the helical bundle of the CB(1) receptor appears to be fully converged in 50 ns of the simulation. The results reveal that the helical bundle structure of the CB(1) receptor maintains a topology quite similar to the x-ray structures of G-protein coupled receptors overall. It is also revealed that the CB(1) receptor is stabilized by the formation of extensive, water-mediated H-bond networks, aromatic stacking interactions, and receptor-lipid interactions within the helical core region. It is likely that these interactions, which are often specific to functional motifs, including the S(N)LAxAD, D(E)RY, CWxP, and NPxxY motifs, are the molecular constraints imposed on the inactive state of the CB(1) receptor. It appears that disruption of these specific interactions is necessary to release the molecular constraints to achieve a conformational change of the receptor suitable for G-protein activation.

Protein-ligand interaction prediction: an improved chemogenomics approach

MOTIVATION

Predicting interactions between small molecules and proteins is a crucial step to decipher many biological processes, and plays a critical role in drug discovery. When no detailed 3D structure of the protein target is available, ligand-based virtual screening allows the construction of predictive models by learning to discriminate known ligands from non-ligands. However, the accuracy of ligand-based models quickly degrades when the number of known ligands decreases, and in particular the approach is not applicable for orphan receptors with no known ligand.

RESULTS

We propose a systematic method to predict ligand-protein interactions, even for targets with no known 3D structure and few or no known ligands. Following the recent chemogenomics trend, we adopt a cross-target view and attempt to screen the chemical space against whole families of proteins simultaneously. The lack of known ligand for a given target can then be compensated by the availability of known ligands for similar targets. We test this strategy on three important classes of drug targets, namely enzymes, G-protein-coupled receptors (GPCR) and ion channels, and report dramatic improvements in prediction accuracy over classical ligand-based virtual screening, in particular for targets with few or no known ligands.

AVAILABILITY

All data and algorithms are available as Supplementary Material.

Virtual screening of GPCRs: an in silico chemogenomics approach

BACKGROUND

The G-protein coupled receptor (GPCR) superfamily is currently the largest class of therapeutic targets. In silico prediction of interactions between GPCRs and small molecules in the transmembrane ligand-binding site is therefore a crucial step in the drug discovery process, which remains a daunting task due to the difficulty to characterize the 3D structure of most GPCRs, and to the limited amount of known ligands for some members of the superfamily. Chemogenomics, which attempts to characterize interactions between all members of a target class and all small molecules simultaneously, has recently been proposed as an interesting alternative to traditional docking or ligand-based virtual screening strategies.

RESULTS

We show that interaction prediction in the chemogenomics framework outperforms state-of-the-art individual ligand-based methods in accuracy both for receptor with known ligands and without known ligands. This is done with no knowledge of the receptor 3D structure. In particular we are able to predict ligands of orphan GPCRs with an estimated accuracy of 78.1%.

CONCLUSION

We propose new methods for in silico chemogenomics and validate them on the virtual screening of GPCRs. The methods represent an extension of a recently proposed machine learning strategy, based on support vector machines (SVM), which provides a flexible framework to incorporate various information sources on the biological space of targets and on the chemical space of small molecules. We investigate the use of 2D and 3D descriptors for small molecules, and test a variety of descriptors for GPCRs. We show that incorporating information about the known hierarchical classification of the target family and about key residues in their inferred binding pockets significantly improves the prediction accuracy of our model.

Prediction of drug-target interaction networks from the integration of chemical and genomic spaces

Motivation: The identification of interactions between drugs and target proteins is a key area in genomic drug discovery. Therefore, there is a strong incentive to develop new methods capable of detecting these potential drug–target interactions efficiently.

Results: In this article, we characterize four classes of drug–target interaction networks in humans involving enzymes, ion channels, G-protein-coupled receptors (GPCRs) and nuclear receptors, and reveal significant correlations between drug structure similarity, target sequence similarity and the drug–target interaction network topology. We then develop new statistical methods to predict unknown drug–target interaction networks from chemical structure and genomic sequence information simultaneously on a large scale. The originality of the proposed method lies in the formalization of the drug–target interaction inference as a supervised learning problem for a bipartite graph, the lack of need for 3D structure information of the target proteins, and in the integration of chemical and genomic spaces into a unified space that we call ‘pharmacological space’. In the results, we demonstrate the usefulness of our proposed method for the prediction of the four classes of drug–target interaction networks. Our comprehensively predicted drug–target interaction networks enable us to suggest many potential drug–target interactions and to increase research productivity toward genomic drug discovery.

Availability: Softwares are available upon request.

Contact:Yoshihiro.Yamanishi@ensmp.fr

Supplementary information: Datasets and all prediction results are available at http://web.kuicr.kyoto-u.ac.jp/supp/yoshi/drugtarget/.

Processing and classification of chemical data inspired by insect olfaction

The chemical sense of insects has evolved to encode and classify odorants. Thus, the neural circuits in their olfactory system are likely to implement an efficient method for coding, processing, and classifying chemical information. Here, we describe a computational method to process molecular representations and classify molecules. The three-step approach mimics neurocomputational principles observed in olfactory systems. In the first step, the original stimulus space is sampled by "virtual receptors," which are chemotopically arranged by a self-organizing map. In the second step, the signals from the virtual receptors are decorrelated via correlation-based lateral inhibition. Finally, in the third step, olfactory scent perception is modeled by a machine learning classifier. We found that signal decorrelation during the second stage significantly increases the accuracy of odorant classification. Moreover, our results suggest that the proposed signal transform is capable of dimensionality reduction and is more robust against overdetermined representations than principal component scores. Our olfaction-inspired method was successfully applied to predicting bioactivities of pharmaceutically active compounds with high accuracy. It represents a way to efficiently connect chemical structure with biological activity space.

Trace amine-associated receptor 1-Family archetype or iconoclast?

Interest has recently been rekindled in receptors that are activated by low molecular weight, noncatecholic, biogenic amines that are typically found as trace constituents of various vertebrate and invertebrate tissues and fluids. The timing of this resurgent focus on receptors activated by the "trace amines" (TA) beta-phenylethylamine (PEA), tyramine (TYR), octopamine (OCT), synephrine (SYN), and tryptamine (TRYP) is the direct result of 2 publications that appeared in 2001 describing the cloning of a novel G protein-coupled receptor (GPCR) referred to by their discoverers Borowsky et al. as TA1 and Bunzow et al. as TA receptor 1 (TAR1). When heterologously expressed in Xenopus laevis oocytes and various eukaryotic cell lines, recombinant rodent and human TAR dose-dependently couple to the stimulation of adenosine 3',5'-monophosphate (cAMP) production. Structure-activity profiling based on this functional response has revealed that in addition to the TA, other biologically active compounds containing a 2-carbon aliphatic side chain linking an amino group to at least 1 benzene ring are potent and efficacious TA receptor agonists with amphetamine (AMPH), methamphetamine, 3-iodothyronamine, thyronamine, and dopamine (DA) among the most notable. Almost 100 years after the search for TAR began, numerous TA1/TAR1-related sequences, now called TA-associated receptors (TAAR), have been identified in the genome of every species of vertebrate examined to date. Consequently, even though heterologously expressed TAAR1 fits the pharmacological criteria established for a bona fide TAR, a major challenge for those working in the field is to discern the in vivo pharmacology and physiology of each purported member of this extended family of GPCR. Only then will it be possible to establish whether TAAR1 is the family archetype or an iconoclast.

Discovery of the first nonpeptidic, small-molecule, highly selective somatostatin receptor subtype 5 antagonists: a chemogenomics approach

We disclose the first selective, nonpeptidic, small-molecule somatostatin receptor subtype 5 (SST5R) antagonists that were identified by a chemogenomics approach based on the analysis of the homology of amino acids defining the putative consensus drug binding site of SST5R. With this strategy, opioid, histamine, dopamine, and serotonine receptors were identified as the closest neighbors of SST5R. The H1 antagonist astemizole was chosen as a seed structure and subsequently transformed into a SST5 receptor antagonist with nanomolar binding affinity devoid of the original target activity.

The binding site for neohesperidin dihydrochalcone at the human sweet taste receptor

Background

Differences in sweet taste perception among species depend on structural variations of the sweet taste receptor. The commercially used isovanillyl sweetener neohesperidin dihydrochalcone activates the human but not the rat sweet receptor TAS1R2+TAS1R3. Analysis of interspecies combinations and chimeras of rat and human TAS1R2+TAS1R3 suggested that the heptahelical domain of human TAS1R3 is crucial for the activation of the sweet receptor by neohesperidin dihydrochalcone.

Results

By mutational analysis combined with functional studies and molecular modeling we identified a set of different amino acid residues within the heptahelical domain of human TAS1R3 that forms the neohesperidin dihydrochalcone binding pocket. Sixteen amino acid residues in the transmembrane domains 2 to 7 and one in the extracellular loop 2 of hTAS1R3 influenced the receptor's response to neohesperidin dihydrochalcone. Some of these seventeen residues are also part of the binding sites for the sweetener cyclamate or the sweet taste inhibitor lactisole. In line with this observation, lactisole inhibited activation of the sweet receptor by neohesperidin dihydrochalcone and cyclamate competitively, whereas receptor activation by aspartame, a sweetener known to bind to the N-terminal domain of TAS1R2, was allosterically inhibited. Seven of the amino acid positions crucial for activation of hTAS1R2+hTAS1R3 by neohesperidin dihydrochalcone are thought to play a role in the binding of allosteric modulators of other class C GPCRs, further supporting our model of the neohesperidin dihydrochalcone pharmacophore.

Conclusion

From our data we conclude that we identified the neohesperidin dihydrochalcone binding site at the human sweet taste receptor, which overlaps with those for the sweetener cyclamate and the sweet taste inhibitor lactisole. This readily delivers a molecular explanation of our finding that lactisole is a competitive inhibitor of the receptor activation by neohesperidin dihydrochalcone and cyclamate. Some of the amino acid positions crucial for activation of hTAS1R2+hTAS1R3 by neohesperidin dihydrochalcone are involved in the binding of allosteric modulators in other class C GPCRs, suggesting a general role of these amino acid positions in allosterism and pointing to a common architecture of the heptahelical domains of class C GPCRs.

Chemogenomic approaches to rational drug design

Paradigms in drug design and discovery are changing at a significant pace. Concomitant to the sequencing of over 180 several genomes, the high-throughput miniaturization of chemical synthesis and biological evaluation of a multiple compounds on gene/protein expression and function opens the way to global drug-discovery approaches, no more focused on a single target but on an entire family of related proteins or on a full metabolic pathway. Chemogenomics is this emerging research field aimed at systematically studying the biological effect of a wide array of small molecular-weight ligands on a wide array of macromolecular targets. Since the quantity of existing data (compounds, targets and assays) and of produced information (gene/protein expression levels and binding constants) are too large for manual manipulation, information technologies play a crucial role in planning, analysing and predicting chemogenomic data. The present review will focus on predictive in silico chemogenomic approaches to foster rational drug design and derive information from the simultaneous biological evaluation of multiple compounds on multiple targets. State-of-the-art methods for navigating in either ligand or target space will be presented and concrete drug design applications will be mentioned.

Structural determinants of odorant recognition by the human olfactory receptors OR1A1 and OR1A2

An interaction of odorants with olfactory receptors is thought to be the initial step in odorant detection. However, ligands have been reported for only 6 out of 380 human olfactory receptors, with their structural determinants of odorant recognition just beginning to emerge. Guided by the notion that amino acid positions that interact with specific odorants would be conserved in orthologs, but variable in paralogs, and based on the prediction of a set of 22 of such amino acid positions, we have combined site-directed mutagenesis, rhodopsin-based homology modelling, and functional expression in HeLa/Olf cells of receptors OR1A1 and OR1A2. We found that (i) their odorant profiles are centred around citronellic terpenoid structures, (ii) two evolutionary conserved amino acid residues in transmembrane domain 3 are necessary for the responsiveness of OR1A1 and the mouse ortholog Olfr43 to (S)-(-)-citronellol, (iii) changes at these two positions are sufficient to account for the differential (S)-(-)-citronellol responsiveness of the paralogs OR1A1 and OR1A2, and (iv) the interaction sites for (S)-(-)-citronellal and (S)-(-)-citronellol differ in both human receptors. Our results show that the orientation of odorants within a homology modelling-derived binding pocket of olfactory receptor orthologs is defined by evolutionary conserved amino acid positions.

Comparison of the binding pockets of two chemically unrelated allosteric antagonists of the mGlu5 receptor and identification of crucial residues involved in the inverse agonism of MPEP

Fenobam [N-(3-chlorophenyl)-N'-(4,5-dihydro-1-methyl-4-oxo-1H-imidazole-2-yl)urea], a clinically validated non-benzodiazepine anxiolytic, has been shown to be a potent and non-competitive metabotropic glutamate (mGlu)-5 receptor antagonist. In the present study, we have used the site-directed mutagenesis coupled with three-dimensional receptor-based pharmacophore modelling to elucidate the interacting mode of fenobam within the seven-transmembrane domain (7TMD) of mGlu5 receptor and its comparison with that of 2-methyl-6-(phenylethynyl)pyridine (MPEP), the prototype antagonist. The common residues involved in the recognition of MPEP and fenobam include Pro654(3.36), Tyr658(3.40), Thr780(6.44), Trp784(6.48), Phe787(6.51), Tyr791(6.55) and Ala809(7.47). The differentiating residues between both modulators' interacting modes are Arg647(3.29), Ser657(3.39) and Leu743(5.47). Our data suggest that these chemically unrelated mGlu5 antagonists act similarly, probing a functionally unique region of the 7TMD. Using [3H]inositol phosphates accumulation assay, we have also identified the critical residues involved in the inverse agonist effect of MPEP. The mutation W784(6.48)A completely blocked the inverse agonist activity of MPEP; two mutations F787(6.51)A and Y791(6.55)A, caused a drastic decrease in the MPEP inverse agonism. Furthermore, these three mutations led to an increased efficacy of quisqualate without having any effect on its potency. The fact that the residues Trp784(6.48) and Phe787(6.51) are essential equally in antagonism and inverse agonism effects emphasizes again the key role of these residues and the involvement of a common transmembrane network in receptor inactivation by MPEP.

Integrating molecular design resources within modern drug discovery research: the Roche experience

Various computational disciplines, such as cheminformatics, ADME modeling, virtual screening, chemogenomics search strategies and classic structure-based design, should be seen as one multifaceted discipline contributing to the early drug discovery process. Although significant resources enabling these activities have been established, their true integration into daily research should not be taken for granted. This article reviews value-adding activities from target assessment to lead optimization, and highlights the technical and process-related aspects that can be considered essential for performance and alignment within the research organization.