Open Access

Crystal structures of the A2A adenosine receptor and their use in medicinal chemistry

In Silico Pharmacology20131:22

DOI: 10.1186/2193-9616-1-22

Received: 9 December 2013

Accepted: 10 December 2013

Published: 20 December 2013


New insights into drug design are derived from the X-ray crystallographic structures of G protein-coupled receptors (GPCRs), and the adenosine receptors (ARs) are at the forefront of this effort. The 3D knowledge of receptor binding and activation promises to enable drug discovery for GPCRs in general, and specifically for the ARs. The predictability of modeling based on the X-ray structures of the A2AAR has been well demonstrated in the identification, design and modification of both known and novel AR agonists and antagonists. It is expected that structure-based design of drugs acting through ARs will provide new avenues to clinically useful agents.


G protein-coupled receptors Adenosine receptors Crystal structures Modeling Computer-aided drug discovery


A large fraction of currently marketed pharmaceuticals act via G protein-coupled receptors (GPCRs), and understanding recognition and function of GPCRs at a precise molecular level is essential for future drug discovery (Jacobson & Costanzi 2012). Action of a drug via a GPCR is either through direct binding to the receptor (as orthosteric agonist or antagonist or as allosteric modulator) or by affecting the levels of the endogenous transmitter that are available to activate it. Among the many naturally-occurring extracellular signaling molecules, nucleosides and nucleotides serve to both modulate biological processes in many organs and tissues and to maintain homeostasis. In addition to their many intracellular functions and ion channel activation, these ubiquitous signaling molecules act outside the cell through GPCRs. These receptors include eight subtypes of P2Y receptors (P2YRs), which are all activated by nucleotides but not nucleosides, and four subtypes of adenosine receptors (ARs). The increased release and production of extracellular nucleotides and adenosine, which are localized rather than systemic, are typically a direct function of stress to an organ or tissue, such as hypoxia (Chen et al. 2013). The resulting elevated level of endogenous agonist increases a tonic activation of one or more of these receptors to produce a protective response. In some cases, such as the adenosine “halo” surrounding tumor cells (Antonioli et al. 2013), the protective response is undesired clinically, and in which case an antagonist might be more applicable therapeutically. Nucleotide action through P2YRs tends to boost the innate immune system, i.e. the immediate defensive response.

There is renewed interest in developing AR or P2YR modulators for therapeutic use, due to recent advances in understanding the related biological processes and increased structural knowledge of the receptors. We and other labs are exploring structure-activity relationships (SAR) at ARs and P2YRs, in order to synthesize and characterize selective agents as pharmacological probes and as potential therapeutic agents. We emphasize four aspects of studying these receptors: 1) design and synthesis of novel and selective agonists and antagonists to expand the SAR; 2) structure-function studies of the receptor proteins; 3) exploration of the novel biological role of such receptors; and 4) conceptualization of future therapeutics. In this effort, there is a tight coupling of organic synthetic methodology, molecular modeling, structural biology, and pharmacology.

Main text and discussion

We are designing more potent and subtype-selective receptor ligands, based on an understanding of the molecular recognition in the binding site. New insights into drug design are derived from the X-ray crystallographic structures of GPCRs (Jacobson & Costanzi 2012), and ARs are at the forefront of this effort (Jaakola et al. 2008;Dore et al. 2011;Xu et al. 2011). Although few compounds that act at these two receptor families are approved for clinical use, the ARs serve as important test cases for structure-based ligand design for GPCRs in general. Since 2008, the availability of X-ray crystallographic structures of the human A2AAR in complex with xanthine and non-xanthine antagonists (Jaakola et al. 2008;Dore et al. 2011), has allowed more precise molecular modeling of the 3 dimensional structures of the A2AAR and related receptors with other bound ligands. Recently, a very high resolution structure of an antagonist-bound A2AAR (1.8 Å) was reported and shown to predict allosteric interactions at a site deeper than the orthosteric ligand binding region (Liu et al. 2013;Gutiérrez-de-Terán et al. 2013). The accuracy of many previous AR modeling and docking studies (Ivanov et al. 2009) is a fortuitous result of the close structural similarity of ARs and rhodopsin, the GPCR used for nearly a decade as the most closely related protein template for modeling Family A GPCRs.

In 2011, the first crystal structure of the A2AAR bound to an agonist, i.e. the highly substituted nucleoside UK432,097 1 (Figure 1), was determined by Raymond Stevens and coworkers at Scripps Research Inst. (Xu et al. 2011), and similar structures of the A2AAR in complex with adenosine and its mono-substituted derivative (NECA, 2) were determined by Lebon et al. at MRC Laboratory of Molecular Biology in Cambridge, UK and Heptares Therapeutics (Lebon et al. 2011). The structures of Lebon et al., which additionally show portions of the second extracellular loop (EL2) not defined in the UK432,097 complex, were determined using receptors with thermostabilizing point mutations (StaRs). All agonist-bound A2AAR structures show similar interactions involving the adenine core and the ribose ring with the receptor. Moreover, the UK432,097 complex is stabilized by multiple H-bonding, van der Waals and hydrophobic interactions on extended C2 and N6 substituents. The agonist interactions within the A2AAR evident in the X-ray structures are shown below:
Figure 1

Two of the AR agonists that were co-crystallized with the human A 2A AR (1, 2) and agonists selective for the A 2A AR (3, 4) and A 1 AR (5).

Comparison of agonist-bound and inactive conformations of the A2AAR have largely validated previous approaches to homology modeling of ARs and indicated the conformational changes needed for receptor activation, particularly within the ligand binding region. These conformational changes include a tightening of hydrophilic residues in TM3, TM5 and TM7 around the ribose moiety, which is known to form the “message” portion of the molecule (Xu et al. 2011). Many of the hydrogen bonds with ligand found in the agonist-bound A2AAR structures were predicted by earlier molecular modeling (Ivanov et al. 2009). Upon binding of nucleoside agonists to the ARs, unstable water molecules that appear in the inactive state of the A2AAR (Jaakola et al. 2008;Dore et al. 2011) are expelled from deep in the binding site. The more hydrophobic adenine moiety, which may be derivatized at the C2 and N6 positions with other mainly hydrophobic groups that extend further into the EL region, is considered the “address” portion of adenosine, with respect to receptor subtype selectivity. Some of the movements of ELs are specific to accommodate sterically bulky groups of UK432,097 and are less pronounced in other agonist-bound A2AAR structures. The implications for receptor signaling by particular agonists that produce their own distinct conformational changes of the A2AAR are not yet known.

Pharmacophore-based searches have yielded ligands having novel chemotypes, such as diverse heterocyclic antagonists of the A2AAR, a target for Parkinson’s disease (de Lera Ruiz et al. 2013). An inactive state structure of the A2AAR was applied to in silico searches of diverse chemical libraries to identify antagonists of several AR subtypes. This approach has been useful for discovery of antagonists at both the A2AAR (Katritch et al. 2010;Carlsson et al. 2010;van der Horst et al. 2011;Congreve et al. 2012;Langmead et al. 2012) and the closely related A1 and A3ARs, by homology. Four separate homology models of the human A1AR were built based on optimization with different bound (known) ligands of that subtype, and 2.2 million lead-like compounds were docked in all four models (Kolb et al. 2012). In this study, initially targeting the A1AR, 39 in silico hits were obtained and subsequently screened in binding to three AR subtypes, resulting in the following percent of successful antagonist hits identified: 21% (A1), 38% (A2A) and 36% (A3). Thus, this docking approach is useful for discovery of hits at the A1AR and at closely related subtypes. Some of the successful hits bound to two or three AR subtypes.

Beyond screenings for novel chemotypes, the structures of the A2AAR have also proved useful to design improved analogs of the existing ligands. Previous efforts to develop AR agonists for therapeutic use have faced efficacy issues and low selectivity, leading to side effects (Chen et al. 2013). Two agonists of the A3AR that were first synthesized and characterized biologically in our lab are already in advanced clinical trials for inflammatory diseases (IB-MECA, 3) and liver cancer (the 2-chloro analogue of IB-MECA, CF102). Current or planned trials of A3AR agonists are for treatment of hepatocellular cancer, rheumatoid arthritis, psoriasis, dry eye syndrome, and other inflammatory conditions. Moreover, selective A1AR agonists are sought as antiseizure agents, analgesics and for other neuroprotective functions (Luongo et al. 2013). Structural insights, enabled by molecular modeling, have guided the introduction of novel adenosine derivatives as A1 and A3AR agonists. A first step to validation of the AR agonist modeling approach was to apply it to the A2AAR, used as template, and then to extend it to the closely related A1 and A3AR by homology. Molecular modeling of agonist binding has been evaluated in light of the crystallographic structure of the agonist-bound A2AAR and found to predict interactions of 20 known nucleoside ligands (Deflorian et al. 2012).

Ligand docking and fragment searching are useful for suggesting new congeners of known AR agonists and antagonists to be synthesized in order to explore distinct subpockets in the binding site. Our structure-based synthesis of adenosine agonists guided by knowledge of the X-ray structure treated each region of the nucleoside using a different approach. For example, binding affinities following distal changes on an extended C2 chain correlated with predicted interactions in the spatially permissive EL region (Deflorian et al. 2012). In contrast, an exploration of the relatively small subpocket of the agonist-bound A2AAR structure surrounding the 5′ substituent of ribose yielded novel 5′-amide derivatives, similar to NECA (Tosh et al. 2012c). The ethyl group of NECA was replaced with small (MW < 150) but diverse molecular fragments to complement this subpocket. The majority of in silico hits that were high ranking energetically (roughly the top 1% of a set of 2000 screened structures) and tested in binding assays indeed bound with submicomolar Ki values at the A2AAR. When the inactive A2AAR structure (3EML) was used for screening these adenosine analogues with diverse 5′-amide fragments, the statistical advantage shown in the receiver operating characteristic (ROC) curve (Figure 2), clearly evident using the agonist-bound A2AAR structure (3QAK), disappeared. When an intermediate conformational model (agonist-optimized 3EML) was used for screening, modified from the inactive state through docking of an agonist and relaxing of side chains, the ROC predictability was also intermediate. Virtual hits that more potently inhibited radioligand binding at the A1AR than the A2AAR were rationalized based on specific H-bonding interactions with residues of the A1AR as predicted in the modeling.
Figure 2

Performance of virtual screening of amide fragment derivatives similar to NECA, using three different A 2A AR models, reprinted with permission from Tosh et al. (Tosh et al. 2012c ).

A comparison of the ribose binding elements of the A2AAR with the corresponding amino acid residues in other subtypes shows more significant variation in the A3AR than in the A1AR. Affinity (but not efficacy in receptor activation) of nucleosides at the A3AR seems to depend more heavily on factors outside the ribose binding pocket than is the case for other AR subtypes. This hypothesis is supported by many adenosine analogues lacking the 5′-amide group entirely, i.e. truncated at the 4′ carbon that still bind potently to the A3AR (Tosh et al. 2012b). Nevertheless, the tetrahydrofuryl core scaffold of ribose, or an isostere thereof, is still important at the A3AR as for other AR subtypes.

The latest generation of A3AR receptor agonists contains a constrained bicyclic (bicyclo[3.1.0]hexane, known as methanocarba) substitution of ribose to maintain a receptor-preferred North (N) conformation (Tosh et al. 2012a). This rigid ring system (fused cyclpentane and cyclopropane rings) in its two isomeric forms, i.e. either (N) or South (S), as applied to nucleosides was used before any structural elucidation to interrogate the ribose conformation when bound to an AR (as ribose in adenosine can readily twist to adopt a range of conformations) (Jacobson et al. 2000). The adenosine analogue having a fixed (N) conformation was clearly the more potent isomer at the A1AR, A2AAR and A3AR, with the greatest enhancement over (S) occurring at A3AR (155-fold). Indeed, the predicted (N) ribose conformation was verified when the agonist-bound A2AAR structures were determined.

We have explored rigidified (N)-methanocarba C2-arylalkynyl nucleoside analogues, such as MRS5698 4, that are highly selective A3AR agonists. The very high A3AR selectivity (typically ~10,000-fold) in this series of C2-arylalkynyl (N)-methanocarba nucleosides was consistent with a predicted outward movement of the portion of TM2 near the EL region of the A3AR, in order to preserve hydrogen-bonding interactions with conserved residues in TMs 3, 5, and 7 that lock the adenosine moiety in its deeper binding pocket. This outward movement of TM2, which is justified based on other activated GPCR structures (opsin and β2-adrenergic receptor) is not likely to occur in the A2AAR, because the EL region of this AR subtype is conformationally constrained by four disulfide bridges.

A1AR agonist MRS5474 5 also contains a constrained bicyclic substitution of ribose to maintain a receptor-preferred North (N) conformation (Tosh et al. 2012b). MRS5474 has in vivo anti-convulsant activity without typical rotorod toxicity observed for standard A1AR agonists, which are known to produce intense cardiovascular depression. The N6-dicyclopropylmethyl substitution was selected by screening (in binding assays of the synthetic analogues), in the truncated, methanocarba series, many alkyl/aryl groups known to promote A1AR agonism in the ribose series (Tosh et al. 2012b). This particular N6 group appears to be a structural sweet spot, in that it maintains selectivity and full efficacy. The very tight fit in the N6 region in receptor docking has been analyzed through homology modeling of the human A1AR, based on the agonist-bound A2AAR structure (Xu et al. 2011). Deletion of an otherwise important recognition element (H-bonding groups at the 5′ position) is compensated by a specific N6 substitution in MRS5474.

Novel fluorescent and radiolabeled probes and multivalent conjugates of strategically functionalized AR ligands have also been designed with the aid of molecular modeling (Kozma et al. 2013). Such high affinity ligands containing reporter groups are useful tools for the characterization of receptors and their oligomeric assemblies. For example, adenosine agonist binding to the heterodimers formed between the A2AAR and D2 dopamine receptor can be followed in real time using such a conjugate (Fernández-Dueñas et al. 2012). The interaction of the fluorophore of a given fluorescent chemical probe with the outer portions of the receptor, mainly distal interactions with functional groups on the receptor, may have a major influence on its affinity and selectivity. Thus, the pharmacological profile of the functionalized congener that serves as precursor of the fluorescent ligand is not always predictive of the properties of the final conjugate, and GPCR molecular modeling can provide insight into this phenomenon. One such example is a fluorescent agonist of the P2Y6R that was greatly enhanced in potency and selectivity by the addition of an AlexaFluor488 moiety at the end of a flexible chain (Jayasekara et al. 2013). According to receptor modeling, this fluorophore did not escape the outer environment, i.e. the ELs of the P2Y6R where they are predicted to anchor to specific sites. In this case, the closer template for homology modeling for the P2YRs was judged to be the CXCR4 structure, rather than the A2AAR structure.


Thus, the 3D knowledge of receptor binding and activation promises to enable drug discovery for GPCRs in general, and specifically for the ARs. The predictability of modeling based on the X-ray structures of the A2AAR has been well demonstrated in the design and modification of both known and novel AR agonists and antagonists. In silico screening by docking of chemical libraries to these new structures and to closely related receptor homology models has been validated. It is expected that structure-based design of drugs acting through ARs will provide new avenues to clinically useful agents.



This research was supported in part by the Intramural Research Program of the NIH, National Institute of Diabetes and Digestive and Kidney Diseases.

Authors’ Affiliations

Molecular Recognition Section, Laboratory of Bioorganic Chemistry, NIDDK, National Institutes of Health


  1. Antonioli L, Blandizzi C, Pacher P, Hasko G: Immunity, inflammation and cancer: a leading role for adenosine. Nat Rev Cancer 2013, 13: 842–857. doi:10.1038/nrc3613 View ArticleGoogle Scholar
  2. Carlsson J, Yoo L, Gao ZG, Irwin JJ, Shoichet BK, Jacobson KA: Structure-based discovery of A 2A adenosine receptor ligands. J Med Chem 2010, 53: 3748–3755. 10.1021/jm100240hPubMed CentralView ArticleGoogle Scholar
  3. Chen J-F, Eltzschig HK, Fredholm BB: Adenosine receptors as drug targets - what are the challenges? Nat Rev Drug Discov 2013, 12: 265–286. 10.1038/nrd3955PubMed CentralView ArticleGoogle Scholar
  4. Congreve M, Andrews SP, Dore AS, Hollenstein K, Hurrell E, Langmead CJ, Mason JS, Ng IW, Tehan B, Zhukov A, Weir M, Marshall FH: Discovery of 1,2,4-triazine derivatives as adenosine A 2A antagonists using structure based drug design. J Med Chem 2012, 55: 1898–1903. 10.1021/jm201376wPubMed CentralView ArticleGoogle Scholar
  5. Deflorian F, Kumar TS, Phan K, Gao ZG, Xu F, Wu H, Katritch V, Stevens RC, Jacobson KA: Evaluation of molecular modeling of agonist binding in light of the crystallographic structure of the agonist-bound A 2A adenosine receptor. J Med Chem 2012, 55: 538–552. 10.1021/jm201461qPubMed CentralView ArticleGoogle Scholar
  6. de Lera Ruiz M, Lim YH, Zheng J: Adenosine A 2A receptor as a drug discovery target. J Med Chem 2013. in press, Scholar
  7. Dore AS, Robertson N, Errey JC, Ng I, Hollenstein K, Tehan B, Hurrell E, Bennett K, Congreve M, Magnani F, Tate CG, Weir M, Marshall FH: Structure of the Adenosine A 2A Receptor in Complex with ZM241385 and the Xanthines XAC and Caffeine. Structure 2011, 19: 1283–1293. 10.1016/j.str.2011.06.014PubMed CentralView ArticleGoogle Scholar
  8. Fernández-Dueñas V, Gómez-Soler M, Jacobson KA, Kumar TS, Fuxe K, Borroto-Escuela DO, Ciruela F: Molecular determinants of the adenosine A 2A R-dopamine D 2 receptor-receptor allosterism: Role of the intracellular loop 3 of the dopamine D 2 receptor. J Neurochem 2012, 123: 373–384. 10.1111/j.1471-4159.2012.07956.xPubMed CentralView ArticleGoogle Scholar
  9. Gutiérrez-de-Terán H, Massink A, Rodríguez D, Liu W, Joseph JS, Katritch I, Heitman LH, Xia L, IJzerman AP, Cherezov V, Katritch V, Stevens RC: The role of a sodium ion binding site in the allosteric modulation of the A 2A adenosine G protein-coupled receptor. Structure 2013, 21: 1–11. 10.1016/j.str.2012.12.004View ArticleGoogle Scholar
  10. Ivanov AA, Barak D, Jacobson KA: Evaluation of homology modeling of GPCRs in light of the A 2A adenosine receptor crystallographic structure. J Med Chem 2009, 52: 3284–3292. 10.1021/jm801533xPubMed CentralView ArticleGoogle Scholar
  11. Jaakola VP, Griffith MT, Hanson MA, Cherezov V, Chien EYT, Lane JR, IJzerman AP, Stevens RC: The 2.6 Angstrom Crystal Structure of a Human A 2A Adenosine Receptor Bound to an Antagonist. Science 2008, 322: 1211–1217. 10.1126/science.1164772PubMed CentralView ArticleGoogle Scholar
  12. Jacobson KA, Ji X-D, Li AH, Melman N, Siddiqui MA, Shin KJ, Marquez VE, Ravi RG: Methanocarba analogues of purine nucleosides as potent and selective adenosine receptor agonists. J Med Chem 2000, 43: 2196–2203. 10.1021/jm9905965PubMed CentralView ArticleGoogle Scholar
  13. Jacobson KA, Costanzi S: New insights for drug design from the X-ray crystallographic structures of GPCRs. Mol Pharmacol 2012, 82: 361–371. 10.1124/mol.112.079335PubMed CentralView ArticleGoogle Scholar
  14. Jayasekara PS, Barrett MO, Ball CB, Brown KA, Kozma E, Costanzi S, Squarcialupi L, Balasubramanian R, Maruoka H, Jacobson KA: 4-Alkyloxyimino-Cytosine Nucleotides: Tethering Approaches to Molecular Probes for the P2Y 6 Receptor. Med Chem Commun 2013, 4: 1156–1165. 10.1039/c3md00132fView ArticleGoogle Scholar
  15. Katritch V, Jaakola VP, Lane JR, Lin J, IJzerman AP, Yeager M, Kufareva I, Stevens RC, Abagyan R: Structure-based discovery of novel chemotypes for adenosine A 2A receptor antagonists. J Med Chem 2010, 53: 1799–1809. 10.1021/jm901647pPubMed CentralView ArticleGoogle Scholar
  16. Kolb P, Phan K, Gao ZG, Marko AC, Sali A, Jacobson KA: Limits of ligand selectivity from docking to models: In silico screening for A 1 adenosine receptor antagonists. PLoS One 2012, 7: e49910. doi:10.1371/journal.pone.0049910PubMed CentralView ArticleGoogle Scholar
  17. Kozma E, Jayasekara PS, Squarcialupi L, Paoletta S, Moro S, Federico S, Spalluto G, Jacobson KA: Fluorescent ligands for adenosine receptors. Bioorg Med Chem Lett 2013, 23: 26–36. 10.1016/j.bmcl.2012.10.112PubMed CentralView ArticleGoogle Scholar
  18. Langmead CJ, Andrews SP, Congreve M, Errey JC, Hurrell E, Marshall FH, Mason JS, Richardson CM, Robertson N, Zhukov A, Weir M: Identification of novel adenosine A 2A receptor antagonists by virtual screening. J Med Chem 2012, 55: 1904–1909. 10.1021/jm201455yPubMed CentralView ArticleGoogle Scholar
  19. Lebon G, Warne T, Edwards PC, Bennett K, Langmead CJ, Leslie AG, Tate CG: Agonist-bound adenosine A 2A receptor structures reveal common features of GPCR activation. Nature 2011, 474: 521–555. 10.1038/nature10136PubMed CentralView ArticleGoogle Scholar
  20. Liu W, Chun E, Thompson AA, Chubukov P, Xu F, Katritch V, Han GW, Roth CB, Heitman LH, IJzerman AP, Cherezov V, Stevens RC: Structural basis for allosteric regulation of GPCRs by sodium ions. Science 2013, 337: 232–236.View ArticleGoogle Scholar
  21. Luongo L, Guida F, Imperatore R, Napolitano F, Gatta L, Cristino L, Giordano C, Siniscalco D, Di Marzo V, Bellini G, Petrelli R, Cappellacci L, Usiello A, de Novellis V, Rossi F, Maione S: The A 1 adenosine receptor as a new player in microglia physiology. Glia 2013, 62: 1098–1136.Google Scholar
  22. Tosh DK, Deflorian F, Phan K, Gao ZG, Wan TC, Gizewski E, Auchampach JA, Jacobson KA: Structure-guided design of A 3 adenosine receptor-selective nucleosides: combination of 2-arylethynyl and bicyclo[3.1.0]hexane substitutions. J Med Chem 2012, 55: 4847–4860. 10.1021/jm300396nPubMed CentralView ArticleGoogle Scholar
  23. Tosh DK, Paoletta S, Deflorian F, Phan K, Moss SM, Gao ZG, Jiang X, Jacobson KA: Structural sweet spot for A 1 adenosine receptor activation by truncated (N)-methanocarba nucleosides: receptor docking and potent anticonvulsant activity. J Med Chem 2012, 55: 8075–8090. 10.1021/jm300965aPubMed CentralView ArticleGoogle Scholar
  24. Tosh DK, Phan K, Gao ZG, Gakh A, Xu F, Deflorian F, Abagyan R, Stevens RC, Jacobson KA, Katritch V: Optimization of adenosine 5′-carboxamide derivatives as adenosine receptor agonists using structure-based ligand design and fragment-based searching. J Med Chem 2012, 55: 4297–4308. 10.1021/jm300095sPubMed CentralView ArticleGoogle Scholar
  25. van der Horst E, van der Pijl R, Mulder-Krieger T, Bender A, IJzerman AP: Substructure-based virtual screening for adenosine A 2A receptor ligands. ChemMedChem 2011, 6: 2302–2311. 10.1002/cmdc.201100369View ArticleGoogle Scholar
  26. Xu F, Wu H, Katritch V, Han GW, Jacobson KA, Gao ZG, Cherezov V, Stevens RC: Structure of an agonist-bound human A 2A adenosine receptor. Science 2011, 332: 322–327. 10.1126/science.1202793PubMed CentralView ArticleGoogle Scholar


© Jacobson; licensee Springer. 2013

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 (, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.