A novel chemogenomics analysis of G protein-coupled receptors (GPCRs) and their ligands: a potential strategy for receptor de-orphanization
- Eelke van der Horst†1,
- Julio E Peironcely†1,
- Adriaan P IJzerman1,
- Margot W Beukers1,
- Jonathan R Lane1,
- Herman WT van Vlijmen1,
- Michael TM Emmerich2,
- Yasushi Okuno3 and
- Andreas Bender1, 4Email author
© van der Horst et al; licensee BioMed Central Ltd. 2010
Received: 4 March 2010
Accepted: 10 June 2010
Published: 10 June 2010
G protein-coupled receptors (GPCRs) represent a family of well-characterized drug targets with significant therapeutic value. Phylogenetic classifications may help to understand the characteristics of individual GPCRs and their subtypes. Previous phylogenetic classifications were all based on the sequences of receptors, adding only minor information about the ligand binding properties of the receptors. In this work, we compare a sequence-based classification of receptors to a ligand-based classification of the same group of receptors, and evaluate the potential to use sequence relatedness as a predictor for ligand interactions thus aiding the quest for ligands of orphan receptors.
We present a classification of GPCRs that is purely based on their ligands, complementing sequence-based phylogenetic classifications of these receptors. Targets were hierarchically classified into phylogenetic trees, for both sequence space and ligand (substructure) space. The overall organization of the sequence-based tree and substructure-based tree was similar; in particular, the adenosine receptors cluster together as well as most peptide receptor subtypes (e.g. opioid, somatostatin) and adrenoceptor subtypes. In ligand space, the prostanoid and cannabinoid receptors are more distant from the other targets, whereas the tachykinin receptors, the oxytocin receptor, and serotonin receptors are closer to the other targets, which is indicative for ligand promiscuity. In 93% of the receptors studied, de-orphanization of a simulated orphan receptor using the ligands of related receptors performed better than random (AUC > 0.5) and for 35% of receptors de-orphanization performance was good (AUC > 0.7).
We constructed a phylogenetic classification of GPCRs that is solely based on the ligands of these receptors. The similarities and differences with traditional sequence-based classifications were investigated: our ligand-based classification uncovers relationships among GPCRs that are not apparent from the sequence-based classification. This will shed light on potential cross-reactivity of GPCR ligands and will aid the design of new ligands with the desired activity profiles. In addition, we linked the ligand-based classification with a ligand-focused sequence-based classification described in literature and proved the potential of this method for de-orphanization of GPCRs.
G protein-coupled receptors (GPCRs) comprise a large family, more than 800 in human , of cell surface receptors that consist of seven transmembrane (TM) helices. These receptors are activated by a variety of external stimuli, including light, ions, small molecules, lipids, and proteins; moreover, the majority of therapeutic drugs act on GPCRs . Because of the limited number of target crystal structures [3–6], GPCR drug design relies largely on ligand-based approaches  such as property-based methods , pharmacophore models , and substructure methods . These methods do not require any knowledge about the target protein; however, combining them with target information often increases their potential. The resulting so-called 'chemogenomics' approaches thus involve both ligand-based and target-based aspects . They do not focus on a single group of ligands and one individual target, but rather on groups of ligands against groups of targets. The central idea is that similar targets have similar ligands [12, 13]. Therefore, relationships between targets from the sequence side can be exploited to search for novel receptor ligands on the chemical structure side.
Traditionally, the GPCR superfamily has been classified based on sequence homology of the receptors. Kolakowski grouped all seven transmembrane (7-TM) proteins into classes A to F for receptors proven to bind G-proteins and class O for the other 7-TM proteins . Class A receptors resemble rhodopsin and form the largest cluster. Later, Fredriksson et al. proposed a more elaborate classification for known and predicted human GPCRs . Surgand et al. presented a sequence-based phylogenetic classification of GPCRs viewed from a ligand perspective . By selecting residues pointing inwards into the generic binding pocket of GPCRs, the authors assembled a set of 30 residues most likely to be accessible for ligand binding. Based on these residues, phylogenetic clustering was performed. Although only a subset of residues was used, the classification was similar to classifications based on the full sequence. Applications of a grouping such as proposed by Surgand et al. constitute ligand design for related receptors, as well as de-orphanization of GPCRs . However, the study by Surgant et al. is somewhat limited by the scarcity of structural protein data where the identification of binding site residues was solely based on the structure of bovine rhodopsin. It could not yet take into account recent advances that yielded three pharmacologically relevant X-ray crystal structures, namely those of the human β2 and turkey β1 adrenoceptors, as well as of the human adenosine A2A receptor [3, 5, 6, 16]. Building further on Surgand's work, Gloriam et al. proposed an extended set of ligand-accessible residues, derived from visual inspection of the newly available X-ray GPCR crystal structures, from supporting mutagenesis data and from the evaluation of previously established residue sets . The resulting set of 44 residues was then applied to cluster class A GPCRs into a phylogenetic tree, which reflected similarities in binding site of the receptors.
Complementary to these sequence-based classifications are the ligand-based classifications of GPCRs. Approaches that use ligand similarity measures for target classification have been previously described [18, 19]. Keiser et al. related targets by pair-wise comparison of their ligands . From a set of 65 k ligands, a network was constructed connecting almost all 246 targets through sequential linkage. From this, previously unknown antagonism of methadone on the muscarinic M3 receptor and of emetine on the α2-adrenoceptor was identified.
While sequence-based similarity relies on comparison of the residues at certain positions in the sequence, there is no unambiguously defined method to measure ligand-based similarity. One way of defining ligand similarity is to consider the overlap of substructures in the molecules. Frequent substructure mining is a method for finding the most common substructures in a set of molecules [21–23]. It evaluates all possible substructures, not only discrete fragments that are present in the molecules; it is therefore an exhaustive approach, resulting in a more complete view on the structural features in the set.
In this study, we employ frequent substructure mining to determine the similarity between groups of ligands in a thorough and unbiased manner. This substructural similarity is then used for classification of GPCRs according to relatedness of substructure profiles of their ligands. The substructure-based classification of GPCRs visualizes relatedness of receptors in the form of a phylogenetic tree, which is then compared to the sequence-based phylogenetic classifications of GPCRs. The differences in tree organization are examined with methods that visualize changes in target position. Taken together, we present a (GPCR) classification from the small molecule (ligand) perspective, which facilitates analysis of target similarities and differences in ligand-binding behavior. In addition, we explore the potential of our ligand-based classification in receptor de-orphanization, i.e. the prediction of new ligands for orphan receptors.
Results and Discussion
The relationship between target clustering in the substructure tree (Figure 2) and ligand promiscuity suggests that the substructure tree may be used to identify possible side effects on receptors that are close neighbors in this tree. For instance, off-target activity of ligands can be identified. If inspection reveals a ligand to bind to receptor(s) that are phylogenetically related to the target of interest, a more detailed experimental follow-up with respect to receptor selectivity would be worthwhile.
Visual comparison of the sequence tree (Figure 1) with the substructure tree (Figure 2) reveals that the overall phylogenetic organization is similar. For instance, with the exclusion of the glycoprotein, P2Y, angiotensin, and bradykinin receptors, all other receptors represented by two subtypes occur in pairs in both the ligand tree and the sequence tree. This is also true for receptors with three subtypes present in the dataset, e.g. the three members of the α1, the α2, and the β1 adrenoceptors, as well as the bombesin receptors. Exceptions to this rule are the neuropeptide Y and vasopressin receptors. In addition, the prostanoid receptors largely group together in both trees, as do most of the aminergic receptors.
The clear distinction between the two dopamine receptor types, i.e. D1 and D5 (D1-like) versus D2, D3, and D4 (D2-like), exists both in the sequence-based classification and ligand-based classification. This is in agreement with a previous study  and also known from drugs on the market such as the benzazepines that favor D1-like over D2-like dopamine receptors. Similarly, antipsychotics such as chlorpromazine have a higher affinity for the D2-like subtypes than D1-like receptors .
The difference between ligand-based and target-based classifications may be due to convergent evolution . Functional convergence denotes how proteins that differ in sequence may fulfill the same protein function. The protein sequence of GPCR subtypes will be similar in parts that are involved in the endogenous ligand recognition but may be different in other parts, for instance those parts that play a role in recognition of other, exogenous, ligands (e.g. synthetic drugs). These may therefore have a different selectivity profile compared to the endogenous ligand.
Overall, our method proves useful for receptor de-orphanization, since for 93% of receptors studied de-ophanization performed better than random selection (AUC > 0.5) and for 35% of receptors de-orphanization performed well (AUC > 0.7).
Limitations of the work
In the present study, some targets were excluded due to insufficient availability of ligand data in the source databases. The absence of a receptor may influence the order of other receptors in the trees. Scarcity of ligand data is reflected in the substructure profiles, thereby influencing the correlations among receptors. The issue of data (in) completeness and its effect on interaction networks was recently discussed by Mestres et al.. Using three datasets of increasing complexity (more connections) that linked ligands to targets based on full chemical identity, the authors showed that an increase in the number of connections rapidly leads to shifts in connection patterns. However, our study linked targets based on overlap in substructures; as a consequence sharing of substructures rather than of ligands is sufficient for targets to be identified as related. Bender et al. and Keiser et al. already showed that overlapping ligands are not necessary to predict whether targets are close in ligand space [19, 20]. In addition, our method employs an exhaustive approach to analyze the structural features of ligands. Frequent substructure mining considers all possible substructures that occur in the ligands and is therefore unbiased, i.e. all possible substructures were evaluated, not only those intuitive to chemists, such as functional groups, ring systems (e.g. a phenyl ring), and linkers . However, in the present study less 'obvious' substructures such as ethyl or isobutyl are also considered . For a complete discussion on substructure generation and evaluation, see ref. . Our method is not limited to GPCRs alone; it is easily extended to other protein families for analysis of the differences between subfamily phylogenies, given that sufficient ligand information is available. For instance, it can be applied to the realm of enzymes to complement other chemogenomics analyses .
In this work, we presented a ligand-based phylogenetic classification that complements the well-established sequence-based classification of proteins, and applied our method to classification of GPCRs. This alternate view may contribute to our understanding of GPCR classification since it reveals relationships that are unnoticed with conventional phylogeny. Targets were analyzed based on the substructure profiles of their ligands using an unbiased approach. The overall organization of the sequence tree and the substructure tree was similar; however, substantial differences were also discovered. In the substructure tree, several clusters of subtypes were identified. For instance, it was found that the adenosine receptors group together, and that certain GPCR subfamilies that do not share sequence homology cluster because of ligand similarity. Thus, receptor similarities that signal for potential off-target effects, such as for the serotonergic receptors, are readily identified. In addition, combined with sequence-based classification, the ligand-based classification presented has proven potential (93% of receptors with AUC > 0.5 and 35% with AUC > 0.7) for de-orphanization of receptors.
Ligands for human GPCRs were collected from three publicly available data sources: the StARLITe database, as made available by ChEBI (EMBL-EBI) as part of the ChEMBL database , GLIDA , and KiDB . ChEMBL consists of a collection of more than 500,000 small molecules annotated with activity. Here, only activity values measured directly from binding studies were included. Compounds with Ki, IC50, or EC values below 10 μM were considered active. GLIDA provides biological information on GPCRs (sequences) and chemical information about ligand structures. It has links to several external databases, GPCRDB , UniProt , PubChem , and DrugBank . A reported affinity in one of these source databases classifies a compound as active, independent of the reported binding affinity. Ligands are annotated with an activity type, namely: full agonist, partial agonist, agonist, antagonist or inverse agonist. In the present study, we focused only on binding affinity and not on the activity type. This allowed us to merge the set with the rest of the data. KiDB provides information on drugs and molecular compounds that interact with GPCRs, ion channels, transporters, and enzymes. The entries in KiDB are annotated with ligand, Ki value, radiolabeled ligand, receptor name, source & tissue, species, and PubMed link to the publication(s). Our dataset consisted of ligands from all three sources, by selecting human GPCR ligands with a molecular weight between 50 and 700 Da. Only targets that had 20 or more ligands listed were used. In this study, we focused on class A (rhodopsin-like) GPCRs since the majority of targets are from class A and only a minor part from class C; combining both classes would have negatively affected homogeneity of the phylogenetic trees, thereby hampering comparison. For the same reason, we removed two singleton targets (targets that are the only member in a subfamily), the gonadotrophin-releasing hormone receptor and the ghrelin receptor. The final set consisted of 102 targets (provided in Table 1 of Additional file 3 - List of GPCRs used in this study) with 37350 unique ligands in total.
Frequent Substructure Mining
For the ligands of each receptor, the most frequently occurring substructures were determined. This was accomplished by using the frequent subgraph-mining algorithm , which finds all frequent substructures in a set of molecular graphs . For a description and a quantitative comparison of recent substructure mining algorithms, see . Briefly, starting from the smallest substructure, namely the single atoms, the algorithm finds the number of molecules in which the substructure occurs. If this occurrence is above a user-defined minimum, the minimum support value, the substructure is stored. Stored substructures are stepwise extended, and tested in a systematic manner, with the aim of testing all possible substructures that have at least one of the stored substructures as their basis. The algorithm seeks ways to test only those substructures that actually occur in the set, and that have a frequency above the set minimum. An important concept of frequent substructure mining is the a priori principle, originating from frequent item set mining . Algorithms based on the a priori principle exploit that the frequency of a substructure will be equal or lower than the frequency of the substructures it contains. Therefore, whenever the occurrence of a substructure is below the minimum support, all extensions of that substructure are discarded.
Structures were represented as labeled graphs with a special type for aromatic bonds. In this study, the minimum support value was set to 30% of the number of ligands in each activity set. At this value, the algorithm provided a large group of substructures while still being computationally feasible to work with. In addition, molecular structures were sorted in ascending order according to the number of bonds. This allowed the algorithm to prune scarce, complicated substructures that consisted of a large number of bonds, thereby reducing memory requirements. If the set of generated substructures is disproportionately large (more than 1000 times larger) compared to the majority of the other classes, the generated substructures are discarded except for those that also occur in other classes. This step was performed in order to prevent single targets from dominating the analysis. Since in practice most classes generated sets of less than 1000 substructures, a cut-off of 1 M substructures was used. Substructures with molecular weight below 50 Dalton were discarded. The frequent substructures of all classes were merged into one set, removing any duplicates. For all substructures in this set, the frequency in each subfamily was determined. To calculate the correlation between two targets, we used the substructure frequencies as features for that target. A correlation matrix was constructed by calculating the Pearson correlation coefficient for each pair of targets. Finally, a distance matrix was constructed by subtracting the values of the correlation matrix from unity and normalizing the results linearly to the interval [0;1].
To study receptor organization, receptors were clustered into a phylogenetic tree using the Neighbor-Joining (NJ) method (Neighbor from the PHYLIP package ). This method infers phylogenies from the pair-wise distances between receptors. Phylogenetic trees built from distance matrices facilitate tree comparison across domains. In addition, NJ clusters each domain equally well since it does not involve an 'evolutionary clock', a concept rooted in evolutionary biology. Two distance matrices represented the similarities of the receptors: according to the frequent substructures of their ligands and the 7-TM domain sequence alignment, both were visualized as a phylogenetic tree, with receptors as leaves of the tree. The number of branches between two leaves in the tree grows with dissimilarity of these two leaves.
The protein distances between the aligned sequences were calculated with Protdist from the PHYLIP package version 3.6. using the Jones-Taylor-Thornton matrix (default) . Both the sequence-based and ligand-based phylogenetic trees were constructed using the neighbor.exe program from the PHYLIP package. Tree construction might be influenced by the order in which targets are provided to the tree constructor. To minimize the influence on the resulting phylogenetic tree, target input order was randomized 10 times and 10 new trees were generated. From these, a consensus tree was built. MEGA4  was used for editing the layout of the trees and for visualization. Trees were rooted on the mid-points, that is, a root is placed at the mid-point of the longest distance between two taxa of the unrooted tree. Taxa were arranged for balanced shape and trees were visualized as circular trees showing only topology, i.e. branch lengths do not reflect evolutionary distance in a quantitative manner.
For the comparison of trees, several methods and visualizations are available; however, there is not a single definitive measure for tree difference. To visualize how the receptor positions change between two trees we employed a delta-delta plot.
The delta-delta plot reveals how receptor locations behave globally with respect to the median of all receptors. It was used to visualize the differences in location of each receptor in sequence space and in substructure space. This plot is an adaptation from the delta-delta plot in Garr et al.. It is a new way of tree comparison, which visualizes the differences among trees graphically, as opposed to the sole calculation of a numerical distance between two trees which is not trivial to interpret. For each receptor, the mean distance of that receptor to all other receptors was calculated. This value was plotted in a scatter plot, with each axis representing the mean distance of the respective node in one of the trees. The interpretation of this plot is as follows. Along both axes, receptors plotted far from the origin are, on average, more distant from the rest of the group, while receptors plotted close to the origin were closer to the rest of receptors. Receptors plotted near the diagonal do not change much in their mean distance to other receptors when going from one tree to the other (since they are close to the X = Y diagonal). Receptors plotted above or below the diagonal have different average distance to the other receptors between trees. For instance, consider a delta-delta plot that plots a substructure tree along the x-axis and a sequence tree along the y-axis. If a receptor is plotted above the diagonal, the mean distance of that receptor to the other receptors is larger in the sequence tree than the substructure tree; for receptors plotted below the diagonal, the opposite is true.
This experiment is repeated for every receptor (the 'orphan receptor') by temporarily removing ligands of this receptor from the dataset and predicting the position of molecules of this class in the substructure tree. A molecule from the left-out class is a hit when it is predicted to belong to one of the closest classes in sequence space. The closest classes in sequence space are found using the distance matrix from the multiple sequence alignment. Prediction of the class of a molecule is based on the Euclidean distance in substructure space. This distance is calculated as follows: for each substructure, the square of the difference between the relative frequency in a class and the molecule is calculated. The relative frequency of a substructure in a molecule is either 0 for absence, or 1 for presence of the substructure. The square root of the sum of all squared differences is the Euclidean distance between a molecule and a class. The area under the curve (AUC) of the receiver operating characteristic (ROC) plot served as a quality measure of the predictions for a class.
Instead of repeating the substructure mining for every left-out class, a lookup table of substructure occurrence was used. This table related all generated substructures with all molecules in which they occurred. Substructures that had a frequency just above the support threshold in the left-out class were not considered when analysis was performed for molecules of this class.
The authors thank all members of the Division of Medicinal Chemistry of the Leiden/Amsterdam Center for Drug Research at Leiden University for helpful discussions. In addition, the authors thank Bas Vroling from the CMBI, Radboud University, for his help with the sequence alignments.
Funding: This work was supported by the Dutch Top Institute Pharma, project number: D1-105.
- Fredriksson R, Lagerstrom MC, Lundin L-G, Schioth HB: The G-Protein-Coupled Receptors in the Human Genome Form Five Main Families. Phylogenetic Analysis, Paralogon Groups, and Fingerprints. Molecular Pharmacology 2003, 63(6):1256–1272. 10.1124/mol.63.6.1256View ArticlePubMedGoogle Scholar
- Jacoby E, Bouhelal R, Gerspacher M, Seuwen K: The 7 TM G-Protein-Coupled Receptor Target Family. Chem Med Chem 2006, 1(8):760–782.View ArticleGoogle Scholar
- Jaakola V-P, Griffith MT, Hanson MA, Cherezov V, Chien EYT, Lane JR, IJzerman AP, Stevens RC: The 2.6 Angstrom Crystal Structure of a Human A2A Adenosine Receptor Bound to an Antagonist. Science 2008, 1164772.Google Scholar
- Ballesteros J, Palczewski K: G protein-coupled receptor drug discovery: Implications from the crystal structure of rhodopsin. Curr Opin Drug Discovery Dev 2001, 4(5):561–574.Google Scholar
- Cherezov V, Rosenbaum DM, Hanson MA, Rasmussen SGF, Thian FS, Kobilka TS, Choi H-J, Kuhn P, Weis WI, Kobilka BK, et al.: High-Resolution Crystal Structure of an Engineered Human β2-Adrenergic G Protein Coupled Receptor. Science 2007, 318(5854):1258–1265. 10.1126/science.1150577View ArticlePubMedPubMed CentralGoogle Scholar
- Warne T, Serrano-Vega MJ, Baker JG, Moukhametzianov R, Edwards PC, Henderson R, Leslie AGW, Tate CG, Schertler GFX: Structure of a β1-adrenergic G-protein-coupled receptor. Nature 2008, 454(7203):486–491. 10.1038/nature07101View ArticlePubMedPubMed CentralGoogle Scholar
- Klabunde T, Hessler G: Drug Design Strategies for Targeting G-Protein-Coupled Receptors. Chem Bio Chem 2002, 3(10):928–944.View ArticlePubMedGoogle Scholar
- Balakin KV, Tkachenko SE, Lang SA, Okun I, Ivashchenko AA, Savchuk NP: Property-Based Design of GPCR-Targeted Library. J Chem Inf Comput Sci 2002, 42(6):1332–1342.View ArticlePubMedGoogle Scholar
- Chang LCW, Spanjersberg RF, von Frijtag Drabbe-Künzel JK, Mulder-Krieger T, van den Hout G, Beukers MW, Brussee J, IJzerman AP: 2,4,6-Trisubstituted Pyrimidines as a New Class of Selective Adenosine A1Receptor Antagonists. J Med Chem 2004, 47(26):6529–6540. 10.1021/jm049448rView ArticlePubMedGoogle Scholar
- Bywater R: Privileged Structures in GPCRs. In GPCRs: From Deorphanization to Lead Structure Identification. Edited by: Bourne H, Horuk R, Kuhnke J, Michel H. Springer-Verlag; 2007:75–92. full_textView ArticleGoogle Scholar
- Doddareddy MR, Westen GJPv, Horst Evd, Peironcely JE, Corthals F, IJzerman AP, Emmerich M, Jenkins JL, Bender A: Chemogenomics: Looking at biology through the lens of chemistry. Statistical Analysis and Data Mining 2009, 2(3):149–160. 10.1002/sam.10046View ArticleGoogle Scholar
- Bender A, Young DW, Jenkins JL, Serrano M, Mikhailov D, Clemons PA, Davies JW: Chemogenomic data analysis: Prediction of small-molecule targets and the advent of biological fingerprints. Comb Chem High Throughput Screening 2007, 10(8):719–731. 10.2174/138620707782507313View ArticleGoogle Scholar
- Klabunde T: Chemogenomic approaches to drug discovery: similar receptors bind similar ligands. Br J Pharmacol 2007, 152(1):5–7. 10.1038/sj.bjp.0707308View ArticlePubMedPubMed CentralGoogle Scholar
- Kolakowski LFJ: GCRDb: a G-protein-coupled receptor database. Recept Channels 1994, 2: 1–7.PubMedGoogle Scholar
- Surgand J-S, Rodrigo J, Kellenberger E, Rognan D: A chemogenomic analysis of the transmembrane binding cavity of human G-protein-coupled receptors. Proteins: Struct, Funct, Bioinf 2006, 62(2):509–538. 10.1002/prot.20768View ArticleGoogle Scholar
- Rasmussen SGF, Choi H-J, Rosenbaum DM, Kobilka TS, Thian FS, Edwards PC, Burghammer M, Ratnala VRP, Sanishvili R, Fischetti RF, et al.: Crystal structure of the human β2adrenergic G-protein-coupled receptor. Nature 2007, 450(7168):383–387. 10.1038/nature06325View ArticlePubMedGoogle Scholar
- Gloriam DE, Foord SM, Blaney FE, Garland SL: Definition of the G Protein-Coupled Receptor Transmembrane Bundle Binding Pocket and Calculation of Receptor Similarities for Drug Design. J Med Chem 2009, 52(14):4429–4442. 10.1021/jm900319eView ArticlePubMedGoogle Scholar
- Bender A, Jenkins JL, Glick M, Deng Z, Nettles JH, Davies JW: "Bayes Affinity Fingerprints" Improve Retrieval Rates in Virtual Screening and Define Orthogonal Bioactivity Space: When Are Multitarget Drugs a Feasible Concept? J Chem Inf Model 2006, 46(6):2445–2456. 10.1021/ci600197yView ArticlePubMedGoogle Scholar
- Bender A, Scheiber J, Glick M, Davies JW, Azzaoui K, Hamon J, Urban L, Whitebread S, Jenkins JL: Analysis of Pharmacology Data and the Prediction of Adverse Drug Reactions and Off-Target Effects from Chemical Structure. ChemMedChem 2007, 2(6):861–873. 10.1002/cmdc.200700026View ArticlePubMedGoogle Scholar
- Keiser MJ, Roth BL, Armbruster BN, Ernsberger P, Irwin JJ, Shoichet BK: Relating protein pharmacology by ligand chemistry. Nat Biotech 2007, 25(2):197–206. 10.1038/nbt1284View ArticleGoogle Scholar
- van der Horst E, Okuno Y, Bender A, IJzerman AP: Substructure Mining of GPCR Ligands Reveals Activity-Class Specific Functional Groups in an Unbiased Manner. J Chem Inf Model 2009, 49(2):348–360. 10.1021/ci8003896View ArticlePubMedGoogle Scholar
- Borgelt C, Berthold MR: Mining Molecular Fragments: Finding Relevant Substructures of Molecules. In Proceedings of the 2002 IEEE International Conference on Data Mining: 2002. IEEE Computer Society; 2002:51–58. full_textView ArticleGoogle Scholar
- Nijssen S, Kok JN: A quickstart in frequent structure mining can make a difference. In Proceedings of the tenth ACM SIGKDD international conference on Knowledge discovery and data mining: 2004. ACM Press, New York, USA; 2004:647–652. full_textView ArticleGoogle Scholar
- Foord SM, Bonner TI, Neubig RR, Rosser EM, Pin J-P, Davenport AP, Spedding M, Harmar AJ: International Union of Pharmacology. XLVI. G Protein-Coupled Receptor List. Pharmacol Rev 2005, 57(2):279–288. 10.1124/pr.57.2.5View ArticlePubMedGoogle Scholar
- Horn F, Bettler E, Oliveira L, Campagne F, Cohen FE, Vriend G: GPCRDB information system for G protein-coupled receptors. Nucl Acids Res 2003, 31(1):294–297. 10.1093/nar/gkg103View ArticlePubMedPubMed CentralGoogle Scholar
- Baker JG: The selectivity of β-adrenoceptor antagonists at the human β1, β2and β3adrenoceptors. Br J Pharmacol 2005, 144(3):317–322. 10.1038/sj.bjp.0706048View ArticlePubMedPubMed CentralGoogle Scholar
- Van Zwieten PA, Doods HN: Muscarinic receptors and drugs in cardiovascular medicine. Cardiovascular Drugs and Therapy 1995, 9(1):159–167. 10.1007/BF00877757View ArticlePubMedGoogle Scholar
- Voigtländer U, Jöhren K, Mohr M, Raasch A, Tränkle C, Buller S, Ellis J, Höltje H-D, Mohr K: Allosteric site on muscarinic acetylcholine receptors: identification of two amino acids in the muscarinic M2 receptor that account entirely for the M2/M5 subtype selectivities of some structurally diverse allosteric ligands in N-methylscopolamine-occupied receptors. Molecular Pharmacology 2003, 64(1):21–31. 10.1124/mol.64.1.21View ArticlePubMedGoogle Scholar
- Okuno Y, Tamon A, Yabuuchi H, Niijima S, Minowa Y, Tonomura K, Kunimoto R, Feng C: GLIDA: GPCR ligand database for chemical genomics drug discovery database and tools update. Nucl Acids Res 2008, 36(suppl_1):D907–912.PubMedPubMed CentralGoogle Scholar
- Paolini GV, Shapland RHB, van Hoorn WP, Mason JS, Hopkins AL: Global mapping of pharmacological space. Nat Biotech 2006, 24(7):805–815. 10.1038/nbt1228View ArticleGoogle Scholar
- Cuisiat S, Bourdiol N, Lacharme V, Newman-Tancredi A, Colpaert F, Vacher B: Towards a New Generation of Potential Antipsychotic Agents Combining D2 and 5-HT1A Receptor Activities. J Med Chem 2007, 50(4):865–876. 10.1021/jm061180bView ArticlePubMedGoogle Scholar
- Lawrence AJ: Optimisation of anti-psychotic therapeutics: a balancing act? Br J Pharmacol 2007, 151(2):161–162. 10.1038/sj.bjp.0707164View ArticlePubMedPubMed CentralGoogle Scholar
- Bondensgaard K, Ankersen M, Thogersen H, Hansen BS, Wulff BS, Bywater RP: Recognition of Privileged Structures by G-Protein Coupled Receptors. J Med Chem 2004, 47(4):888–899. 10.1021/jm0309452View ArticlePubMedGoogle Scholar
- Schnur DM, Hermsmeier MA, Tebben AJ: Are Target-Family-Privileged Substructures Truly Privileged? J Med Chem 2006, 49(6):2000–2009. 10.1021/jm0502900View ArticlePubMedGoogle Scholar
- Abramovitz M, Adam M, Boie Y, Carrière M-C, Denis D, Godbout C, Lamontagne S, Rochette C, Sawyer N, Tremblay NM, et al.: The utilization of recombinant prostanoid receptors to determine the affinities and selectivities of prostaglandins and related analogs. Biochim Biophys Acta, Mol Cell Biol Lipids 2000, 1483(2):285–293. 10.1016/S1388-1981(99)00164-XView ArticleGoogle Scholar
- Pettipher R, Hansel TT, Armer R: Antagonism of the prostaglandin D2 receptors DP1 and CRTH2 as an approach to treat allergic diseases. Nat Rev Drug Discov 2007, 6(4):313–325. 10.1038/nrd2266View ArticlePubMedGoogle Scholar
- Wang S, Gustafson E, Pang L, Qiao X, Behan J, Maguire M, Bayne M, Laz T: A Novel Hepatointestinal Leukotriene B4 Receptor. Cloning and Functional Characterization. J Biol Chem 2000, 275(52):40686–40694. 10.1074/jbc.M004512200View ArticlePubMedGoogle Scholar
- Yokomizo T, Izumi T, Chang K, Takuwa Y, Shimizu T: A G-protein-coupled receptor for leukotriene B4 that mediates chemotaxis. Nature 1997, 387(6633):620–624. 10.1038/42506View ArticlePubMedGoogle Scholar
- Le Crom S, Kapsimali M, Barôme P-O, Vernier P: Dopamine receptors for every species: Gene duplications and functional diversification in Craniates. Journal of Structural and Functional Genomics 2003, 3(1):161–176. 10.1023/A:1022686622752View ArticlePubMedGoogle Scholar
- Zhang J, Xiong B, Zhen X, Zhang A: Dopamine D1 receptor ligands: where are we now and where are we going. Med Res Rev 2009, 29(2):272–294. 10.1002/med.20130View ArticlePubMedGoogle Scholar
- Roth BL, Sheffler D, Potkin SG: Atypical antipsychotic drug actions: unitary or multiple mechanisms for 'atypicality'? Clinical Neuroscience Research 2003, 3(1–2):108–117. 10.1016/S1566-2772(03)00021-5View ArticleGoogle Scholar
- Coward DM: General pharmacology of clozapine. The British Journal of Psychiatry Supplement 1992, (17):5–11.Google Scholar
- Zakon HH: Convergent Evolution on the Molecular Level. Brain, Behavior and Evolution 2002, 59(5–6):250–261. 10.1159/000063562View ArticlePubMedGoogle Scholar
- Mestres J, Gregori-Puigjane E, Valverde S, Sole RV: Data completeness--the Achilles heel of drug-target networks. Nat Biotech 2008, 26(9):983–984. 10.1038/nbt0908-983View ArticleGoogle Scholar
- Bemis GW, Murcko MA: The Properties of Known Drugs. 1. Molecular Frameworks. J Med Chem 1996, 39(15):2887–2893. 10.1021/jm9602928View ArticlePubMedGoogle Scholar
- van der Horst E, IJzerman AP: Computational Approaches to Fragment and Substructure Discovery and Evaluation. In Fragment-Based Drug Discovery: A Practical Approach. Edited by: Zartler ER, Shapiro J, Chichester M. West Sussex, U.K.: John Wiley & Sons, Ltd; 2008.Google Scholar
- Bernasconi P, Min C, Galasinski S, Popa-Burke I, Bobasheva A, Coudurier L, Birkos S, Hallam R, Janzen WP: A Chemogenomic Analysis of the Human Proteome: Application to Enzyme Families. J Biomol Screen 2007, 12(7):972–982. 10.1177/1087057107306759View ArticlePubMedGoogle Scholar
- Roth BL, Lopez E, Beischel S, Westkaemper RB, Evans JM: Screening the receptorome to discover the molecular targets for plant-derived psychoactive compounds: a novel approach for CNS drug discovery. Pharmacol Ther 2004, 102(2):99–110. 10.1016/j.pharmthera.2004.03.004View ArticlePubMedGoogle Scholar
- The UniProt Consortium: The Universal Protein Resource (UniProt). Nucl Acids Res 2008, 36(suppl_1):D190–195.PubMed CentralGoogle Scholar
- Wheeler DL, Barrett T, Benson DA, Bryant SH, Canese K, Chetvernin V, Church DM, DiCuccio M, Edgar R, Federhen S, et al.: Database resources of the National Center for Biotechnology Information. Nucl Acids Res 2008, (36 Database):D13-D21.Google Scholar
- Wishart DS, Knox C, Guo AC, Shrivastava S, Hassanali M, Stothard P, Chang Z, Woolsey J: DrugBank: a comprehensive resource for in silico drug discovery and exploration. Nucl Acids Res 2006, 34(suppl_1):D668–672. 10.1093/nar/gkj067View ArticlePubMedPubMed CentralGoogle Scholar
- Wörlein M, Meinl T, Fischer I, Philippsen M: A Quantitative Comparison of the Subgraph Miners MoFa, gSpan, FFSM, and Gaston. Knowledge Discovery in Databases: PKDD 2005 2005, 392–403. full_textView ArticleGoogle Scholar
- Agrawal R, Srikant R: Fast Algorithms for Mining Association Rules in Large Databases. In Proceedings of the 20th International Conference on Very Large Data Bases: September 12 - 15 1994. Morgan Kaufmann Publishers, San Francisco, CA; 1994:487–499.Google Scholar
- Felsenstein J: PHYLIP (Phylogeny Inference Package) version 3.6. Distributed by the author. Department of Genome Sciences, University of Washington, Seattle. 2005.Google Scholar
- Tamura K, Dudley J, Nei M, Kumar S: MEGA4: Molecular Evolutionary Genetics Analysis (MEGA) Software Version 4.0. Mol Biol Evol 2007, 24(8):1596–1599. 10.1093/molbev/msm092View ArticlePubMedGoogle Scholar
- Garr CD, Peterson JR, Schultz L, Oliver AR, Underiner TL, Cramer RD, Ferguson AM, Lawless MS, Patterson DE: Solution Phase Synthesis of Chemical Libraries for Lead Discovery. J Biomol Screen 1996, 1(4):179–186. 10.1177/108705719600100404View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.