Different efficacy of adenosine and NECA derivatives at the human A3 adenosine receptor: Insight into the receptor activation switch

Biochemical Pharmacology 87 (2014) 321–331

Contents lists available at ScienceDirect

Biochemical Pharmacology journal homepage: www.elsevier.com/locate/biochempharm

Different efficacy of adenosine and NECA derivatives at the human A3 adenosine receptor: Insight into the receptor activation switch Diego Dal Ben a, Michela Buccioni a, Catia Lambertucci a, Sonja Kachler b, Nico Falgner b, Gabriella Marucci a, Ajiroghene Thomas a, Gloria Cristalli a, Rosaria Volpini a, Karl-Norbert Klotz b,* a b

School of Pharmacy, Medicinal Chemistry Unit, University of Camerino, Via S. Agostino 1, I-62032 Camerino, Italy Universita¨t Wu¨rzburg, Institut fu¨r Pharmakologie und Toxikologie, Versbacher Str. 9, D-97078, Wu¨rzburg, Germany

A R T I C L E I N F O

A B S T R A C T

Article history: Received 17 July 2013 Accepted 15 October 2013 Available online 23 October 2013

A3 Adenosine receptors are promising drug targets for a number of diseases and intense efforts are dedicated to develop selective agonists and antagonists of these receptors. A series of adenosine derivatives with 2-(ar)-alkynyl chains, with high affinity and different degrees of selectivity for human A3 adenosine receptors was tested for the ability to inhibit forskolin-stimulated adenylyl cyclase. All these derivatives are partial agonists at A3 adenosine receptors; their efficacy is not significantly modified by the introduction of small alkyl substituents in the N6-position. In contrast, the adenosine-50 N-ethyluronamide (NECA) analogs of 2-(ar)-alkynyladenosine derivatives are full A3 agonists. Molecular modeling analyses were performed considering both the conformational behavior of the ligands and the impact of 2- and 50 -substituents on ligand–target interaction. The results suggest an explanation for the different agonistic behavior of adenosine and NECA derivatives, respectively. A sub-pocket of the binding site was analyzed as a crucial interaction domain for receptor activation. ß 2013 Elsevier Inc. All rights reserved.

Keywords: Adenosine receptor NECA A3 Agonist Efficacy Molecular modeling

1. Introduction Adenosine (Ado) is an ubiquitous metabolite that regulates the function of virtually every cell type via one or several of four subtypes of G protein-coupled receptors (GPCRs) known as A1, A2A, A2B, and A3 [1]. In human, the A3 adenosine receptor (A3AR) is highly expressed in particular in immune cells, lung, and liver and at lower densities in heart, aorta, and brain and it is involved in a variety of key physiological processes such as

Chemical compounds studied in this article: Adenosine (PubChem CID: 60961); NECA (PubChem CID: 448222); HENECA (PubChem CID: 164437); PHPNECA (PubChem CID: 44339675). Abbreviations: NECA, adenosine-50 -N-ethyluronamide; Ado, adenosine; GPCR, G protein-coupled receptor; AR, adenosine receptor; CCPA, 2-chloro-N6-cyclopentyladenosine; HEMADO, 2-hexyn-1-yl-N6-methyladenosine; CPA, N6-cyclopentyladenosine; HENECA, 2-hexyn-1-yl-adenosine-50 -N-ethyluronamide; HEADO, 2hexyn-1-yl-adenosine; PENECA, 2-phenylethynyladenosine-50 -N-ethyluronamide; PEADO, 2-phenylethynyladenosine; PEMADO, N6-methyl-2-phenylethynyladenosine; PHPNECA, 2-(3-hydroxy-3-phenyl)propyn-1-yl-adenosine-50 -N-ethyluronamide; PHPADO, 2-(3-hydroxy-3-phenyl)propyn-1-yl-adenosine; PHPMADO, 2-(3hydroxy-3-phenyl)propyn-1-yl-N6-methyladenosine; MECA, adenosine-50 -Nmethyluronamide; TM, transmembrane; EL, extracellular; ns, nanosecond; ps, picosecond. * Corresponding author. E-mail address: [email protected] (K.-N. Klotz). 0006-2952/$ – see front matter ß 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.bcp.2013.10.011

release of inflammatory mediators and inhibition of tumor necrosis factor-a production [2–6]. These data make A3AR an attractive therapeutic target and the design and synthesis of potent and selective agonists for this adenosine receptor (AR) subtype could be helpful to provide tools for further characterization and evaluation of the physio-pathological role of the protein and for the development of new drugs with antiinflammatory, anticancer, and cardioprotective potential [7–14]. Over the decades a large number of agonists with distinct patterns of selectivity for the different AR subtypes have been developed [1,3,15–19], with the Ado scaffold generally believed as mandatory for their development. However, a series of novel compounds structurally unrelated to Ado was discovered that potently stimulates AR subtypes [20,21]. Considering Ado derivatives, previous studies reported substitutions and modifications to the Ado core resulting in partial [22,23] or complete loss of agonist efficacy [24,25], while among others the 2-, N6-, and 50 -positions were found to be key sites for tolerable modifications to the Ado scaffold without altering the agonistic properties. In particular, substitution of the 2-position may dramatically affect the pharmacological characteristics of Ado and selective agonists for all subtypes except A2BAR have been developed employing a structurally diverse array of substituents in this position. As an example, 2-chloro-N6-cyclopentyladenosine (CCPA) was introduced as an A1AR selective agonist [26,27] while CGS

D. Dal Ben et al. / Biochemical Pharmacology 87 (2014) 321–331

322

Table 1 Affinity (Ki nM) values of A3AR agonists (values are from ref [40]).

. cpd 1 1a 1b 1c 1d 2 2a 2b 2c 2d 3 3a 3b 3c 3d

0

2-Hexyn-1-yl-adenosine-5 -N-ethyluronamide (HENECA) 2-Hexyn-1-yl-adenosine (HEADO) 2-Hexyn-1-yl-N6-methyladenosine (HEMADO) 2-Hexyn-1-yl-N6-ethyladenosine 2-Hexyn-1-yl-N6-isopropyladenosine 2-Phenylethynyladenosine-50 -N-ethyluronamide (PENECA) 2-Phenylethynyladenosine (PEADO) N6-Methyl-2-phenylethynyladenosine (PEMADO) N6-Ethyl-2-phenylethynyladenosine N6-Isopropyl-2-phenylethynyladenosine 2-(3-Hydroxy-3-phenyl)propyn-1-yl-adenosine-50 -N-ethyluronamide (PHPNECA) 2-(3-Hydroxy-3-phenyl)propyn-1-yl-adenosine (PHPADO) 2-(3-Hydroxy-3-phenyl)propyn-1-yl-N6-methyladenosine (PHPMADO) 2-(3-Hydroxy-3-phenyl)propyn-1-yl-N6-ethyladenosine 2-(3-Hydroxy-3-phenyl)propyn-1-yl-N6-isopropyladenosine

21680 with a larger 2-(p-(2-carboxyethyl)phenylethylamino)moiety is a potent A2AAR agonist [28,29]. Further highly potent A2AAR agonists were also identified in a series of 2-(N0 alkylidenehydrazino)Ados [30,31]. It turned out that activation of the A3AR is more affected by structural changes of Ado than activation of other AR subtypes. In particular modifications in the 2- and N6-position as well as the ribose will influence the efficacy of Ado derivatives at this receptor [32–37]. It was found out that the introduction of an alkynyl group in the 2-position is a successful strategy to develop Ado derivatives with high affinity for A1 and A2A AR. However, it proved to be a particularly successful approach for the development of high affinity agonists with remarkable selectivity for A3AR [37–43]. [3H]2-hexyn-1-yl-N6-methyladenosine ([3H]HEMADO) constitutes such an example and was recently introduced as an A3AR selective radioligand [44]. Normally, Ado derivatives that were characterized as agonists at one receptor subtype would stimulate other subtypes as well if affinity allows for binding. A surprising observation was made that the above cited A1AR agonist CCPA behaves as antagonist at the human A3AR subtype in contrast to N6-cyclopentyladenosine (CPA) that lacks the 2-chloro substitution [45]. A further previous report [35] described 2-substituted Ado derivatives with varying efficacy at the different ARs. Considering Ado derivatives presenting alkynyl substituents, it was reported that the introduction of 8-substituents in 2-alkynylAdos leads to the development of compounds presenting as partial agonists at A2AAR [46], while the transfer of the same alkynyl substituents from the 2- to 8-position of Ado changes the pharmacological profile of the compounds from A3AR agonist to antagonist [25,47]. These data prompted us to take a careful look at the functional behavior of selected 2-substituted Ado derivatives which we

R1

R2

A3AR Ki (nM)

nC4H9 nC4H9 nC4H9 nC4H9 nC4H9 Ph Ph Ph Ph Ph (R,S)-CH(OH)Ph (R,S)-CH(OH)Ph (R,S)-CH(OH)Ph (R,S)-CH(OH)Ph (R,S)-CH(OH)Ph

H H CH3 C2H5 CH(CH3)2 H H CH3 C2H5 CH(CH3)2 H H CH3 C2H5 CH(CH3)2

2.4 4.7 1.1 2.3 9.7 6.2 16 3.4 4.9 17 0.42 3.3 0.76 0.97 2.3

reported in previous studies as potent agonists of A3AR [37,40]. These compounds (Table 1) represented a series of Ado derivatives bearing in the 2-position (ar)-alkynyl chains and in the N6-position small alkyl groups. All derivatives are endowed with low nanomolar affinity and different degrees of selectivity for the human A3AR [40,43]. Further modification of these compounds by the replacement of the hydroxyl group in the 50 -position of the sugar moiety with an N-ethylcarboxamido function (obtaining the NECA derivatives) showed to enhance A3AR affinity and selectivity. In the present study, these molecules were tested for their ability to inhibit forskolin-stimulated adenylyl cyclase activity and hence to analyze their efficacy profiles at the human A3AR. Molecular modeling analyses were then carried out to get a rationalization of the activities identified for the presented series of compounds. The crystal structure of the A2AAR in complex with ZM241385 (PDB ID: 3EML [48]) was employed as template to develop a homology model for the A3AR that was used for docking studies of the analyzed ligands. 2. Materials and methods 2.1. Biology All AR agonists (Table 1) were synthesized as described earlier [40,49–52]. All other compounds including guanine nucleotides were from Sigma-RBI, Taufkirchen, Germany. [a-32P]ATP was from PerkinElmer, Rodgau, Germany. Media and fetal calf serum for cell culture were from PanSystems, Aidenbach, Germany, penicillin (100 U/ml), streptomycin (100 mg/ml), L-glutamine and G-418 were purchased from Gibco-Life Technologies, Eggenstein, Germany. All other materials were from sources as described earlier [28,37].

D. Dal Ben et al. / Biochemical Pharmacology 87 (2014) 321–331

2.1.1. Cells and cell culture CHO cells were stably transfected with the human A3AR as described recently [44]. All cell culture procedures followed protocols previously described [28]. In brief, cells were grown adherently and maintained in Dulbecco’s Modified Eagles Medium at 37 8C in 5% CO2/95% air. The medium was supplemented with nutrient mixture F12 (DMEM/F12, 1:1) without nucleosides, containing 10% fetal calf serum, penicillin (100 U/ml), streptomycin (100 mg/ml), L-glutamine (2 mM) and Geneticin (G-418, 0.2 mg/ml). 2.1.2. Membrane preparation and adenylyl cyclase activity Crude membrane fractions for measurement of adenylyl cyclase activity were prepared as described before [28]. The preparation used only one high speed centrifugation of the homogenate and the resulting crude membrane pellet was resuspended in 50 mM Tris/HCl, pH 7.4 and was immediately used for the measurement of adenylyl cyclase activity. The membranes (about 30 mg of protein) were incubated with [a-32P]ATP (about 200,000 cpm) at 37 8C in the presence of 10 mM forskolin to stimulate adenylyl cyclase and 0.5 mM Ro 201724 as a PDE inhibitor. After 20 min the reaction was stopped by addition of ice cold ZnAc/Na2CO3. The accumulated [32P]cAMP was separated from [a-32P]ATP by chromatography on Alumina columns and quantified. Inhibition of forskolin-stimulated (10 mM) adenylyl cyclase activity was given as percent of the cyclase activity in the absence of an agonist. Each ligand concentration was tested in duplicates and each experiment was repeated 3–6 times. For statistical analysis Student’s t-test was used. Statistically significant differences were assumed for p < 0.5. 2.2. Molecular modeling All molecular modeling studies were performed on a 2 CPU (PIV 2.0–3.0 GHZ) Linux PC. Conformational analysis, homology modeling, energy minimization, and docking studies were carried out using Molecular Operating Environment (MOE, version 2010.10) suite [53]. Manual docking and Monte Carlo studies of the Ado binding mode were done using MOE and Schrodinger Macromodel (ver. 8.0) [54] with Schrodinger Maestro interface. Docking analyses for the compounds were then performed with MOE. All ligand structures were optimized using RHF/AM1 semiempirical calculations and the software package MOPAC implemented in MOE was utilized for these calculations [55]. 2.2.1. Conformational analysis A conformational analysis of selected ligands was performed by using the Conformational Search tool of MOE. LowMode method was employed with default settings, that are 10,000 iterations for conformer generation, 500 energy minimization steps performed on each generated conformer, and RMSD limit as 0.25 A˚ for duplicate removal process. For each molecule, MOE generated an output database that was further analyzed to measure the adenine-sugar dihedral angles (obtained by using the MOE Conformation Geometry tool). 2.2.2. Refinement of A2AAR structural model Several crystal structures of the human A2AAR have been reported to date, the great majority obtained as complexes of the receptor with an antagonist (hence in an inactive state) and three of them were obtained as complexes with agonists (i.e. Ado, NECA, UK-432097). As the aim of this work was to simulate the interaction of analyzed compounds with the binding site in an inactive state and the early steps of reciprocal ligand–target

323

adaptation and receptor activation, a crystal structure of A2AAR in complex with an agonist would have represented a conformational state of the binding site as already modified by the presence of an agonist. As consequence, the human A3AR model was developed by using as a template the crystal structure of the A2AAR in complex with an antagonist, in particular the high affinity antagonist ZM241385 (PDB ID: 3EML [48]; available at the RCSB Protein Data Bank; 2.6 A˚ resolution). The crystal structure was complemented with the Pro149-His155 and the Lys209-Ala221 segments (whose structure was not solved by X-ray in the first case or replaced by the T4L segment in the second case) and with the hydrogen atoms and then subjected to AMBER99 [56] energy minimization performed until the RMS gradient of the potential energy was less than 0.05 kJ mol1 A˚1. 2.2.3. Preliminary docking analysis with Ado A preliminary docking analysis was performed by manually docking the Ado structure into the A2AAR crystal structure binding site. The localization and orientation of the cocrystallized A2AAR antagonist ZM241385 and the recently reported crystal structure of the A2AAR–Ado complex [57] helped to establish the binding pocket for the preliminary Ado docking analysis. The obtained A2AAR–Ado complex was then subjected to energy minimization refinement with the same protocol as above. 2.2.4. Homology modeling of the human A3AR Homology models of the human A3AR were built using the A2AAR–Ado complex as a template. The alignment of the AR primary sequences was built within MOE. Template-target alignment was used for the homology modeling protocol, while the boundaries identified from the X-ray crystal structure of the human A2AAR were applied for the corresponding sequence of A3AR. The missing loop domains of the A3AR were built by the loop search method implemented in MOE. Once the heavy atoms were modeled, all hydrogen atoms were added, and the protein coordinates were then minimized with MOE using the AMBER99 force field. The energy minimization was performed with the same protocol as above. The homology modeling algorithm generated 30 A3AR models and the best one in terms of potential energy and rotamers/dihedrals quality was chosen for the further refinement stages. Reliability and quality of this model was checked by using the Protein Geometry Monitor application within MOE, which provides a variety of stereochemical measurements for inspection of the structural quality in a given protein, like backbone bond lengths, angles and dihedrals, Ramachandran w–c dihedral plots, and side chain rotamer and nonbonded contact quality. 2.2.5. Ado binding mode refinement The A3AR model in complex with Ado was subjected to Monte Carlo analysis to explore the favorable binding conformations. This analysis was conducted by the Monte Carlo Conformational Search protocol implemented in Schrodinger Macromodel. The input structure consisted of the ligand and a shell of receptor amino acids within the specified distance (6 A˚) from the ligand. A second external shell of all the residues within a distance of 8 A˚ from the first shell was kept fixed. During the Monte Carlo conformational searching, the input structure was modified by random changes in user-specified torsion angles (for all input structure residues), and molecular position (for the ligand). Hence, the ligand was left free to be continuously re-oriented within the binding site and the conformation of both ligand and internal shell residues could be explored and reciprocally relaxed. The method consisted of 10,000 Conformational Search steps with MMFF94s force field [58–64]. The best A3AR–Ado complex was saved. The final complex served

324

D. Dal Ben et al. / Biochemical Pharmacology 87 (2014) 321–331

as input in MOE and was subjected to energy minimization with the same protocol as above. 2.2.6. Insertion of A3AR homology models into the membrane bilayer model A membrane bilayer model (consisting of phosphatidylcholine residues) was built with the aid of the ‘‘membrane builder’’ tool within VMD software suite [65]. The membrane model was then loaded in MOE and a water box was added with the ‘‘water soak builder’’ tool; the obtained lipid–solvent system (box dimensions: 106  106  106 A˚) consisted of 270 lipid molecules solvated by about 28,000 water molecules. The homology model of the human A3AR was then placed in the hydrated lipid bilayer. Membrane and solvent residues giving bad contacts with the receptor were minimized (and in some cases removed) to avoid carrying out molecular dynamics simulations starting from highly unstable systems. A ‘‘bad contact’’ was defined as the presence of a distance of less than 2.0 A˚ between two non-hydrogen atoms of the system. The obtained receptor–membrane complex was then subjected to a first energy minimization process, keeping fixed the receptor coordinates and leaving free to move a shell of membrane and solvent residues within a distance of 10 A˚ from receptor atoms. In this phase, OPLS-AA force field [66] (distance dependent dielectric model) was employed. The whole complex was checked to verify the eventual presence of some large vacuum regions in the lipidsolvent in the proximity of the receptor. The system was then subjected with the same settings to a molecular dynamics simulation of 100 ps at 310 K, after the system had been heated to the final temperature (30 ps). The aim of this first simulation was to relax membrane and solvent residues working as interface between the receptor model and the rest of the bilayer-solvent environment. A final energy minimization stage was performed until the RMS gradient of the potential energy was less than 0.05 kJ mol1 A˚1. In this minimization stage, the coordinates of both A3AR–Ado complex and the ‘‘10 A˚ residues shell’’ were left free. After removal of Ado from the ligand binding site, the final receptor structure was used for docking studies of the analyzed compounds. 2.2.7. Molecular docking and post-docking analysis All compound structures were docked into the A3AR binding site by using the MOE Dock tool. This method is divided into a number of stages: Conformational Analysis of ligands. The algorithm generated conformations from a single 3D conformation by conducting a systematic search. In this way, all combinations of angles were created for each ligand. Placement. A collection of poses was generated from the pool of ligand conformations using the Alpha Triangle placement method. Poses were generated by superposition of ligand atom triplets and triplet points in the receptor binding site. The receptor site points are alpha sphere centers which represent locations of tight packing. At each iteration a random conformation was selected, a random triplet of ligand atoms and a random triplet of alpha sphere centers were used to determine the pose. Scoring. Poses generated by the placement methodology were scored using two available methods implemented in MOE, the London dG scoring function that estimates the free energy of binding of the ligand from a given pose, and Affinity dG scoring that estimates the enthalpic contribution to the free energy of binding. Top 30 poses for each ligand were output in a MOE database. Poses were then subjected to MMFF94 force field [58–64] energy minimization until the RMS gradient of the potential energy was less than 0.05 kJ mol1 A˚1. Receptor coordinates were kept fixed. Minimized poses were then rescored using the London dG and Affinity dG scoring tools and the dock-pKi predictor, which allows estimating the pKi for each ligand using the scoring.svl script retrievable at the SVL exchange service

(Chemical Computing Group, Inc. SVL exchange: http://svl.chemcomp.com) The algorithm is based upon an empirical scoring function consisting of a directional hydrogen-bonding term, a directional hydrophobic interaction term, and an entropic term (ligand rotatable bonds immobilized in binding). For each compound, top docking pose according to at least two out of the three above cited scoring functions was then selected for further analyses. 2.2.8. Molecular dynamics analysis of A3AR–ligand complexes Each A3AR–ligand complex selected from docking analysis was subjected to OPLS-AA force field energy minimization until the RMS gradient of the potential energy was less than 0.05 kJ mol1 A˚1. AMBER99 partial charges of receptor and MOPAC output partial charges of ligands were conserved. Both receptor and ligand coordinates were left free, the same for the shell of membrane and solvent residues within a distance of 10 A˚ from receptor atoms. Each minimized complex was then subjected to molecular dynamics simulation that was carried out for 10 ns at 310 K, after the system had been heated to the final temperature (1 ns). A final energy minimization stage was performed with the same procedure as above. 3. Results 3.1. Biology We have previously presented a number of 2-substituted Ado derivatives with high A3AR affinity (Table 1) and various degrees of selectivity for this receptor. The ability of these compounds to inhibit forskolin-stimulated cAMP production via human A3AR was studied as a measure of agonistic efficacy. Fig. 1 shows the inhibition of forskolin-stimulated adenylyl cyclase activity of the analyzed derivatives in comparison to the prototypical AR agonist NECA. The inhibitory effect of the full agonist NECA (100 mM) is indicated by a horizontal line in each figure for comparison. Considering the 2-hexynyl derivatives 1, 1a–d, Fig. 1A shows that the ribose-modified HENECA (1) inhibits adenylyl cyclase similar to NECA and behaves, therefore, as a full A3AR agonist. HEADO (1a) with the unmodified ribose shows only partial agonistic activity. Compounds 1b, 1c, and 1d differ from HEADO (1a) by a methyl-, ethyl-, and isopropyl-substituent, respectively, in the N6-position. These N6-substituents do only marginally change the partial agonistic activity observed for HEADO (1a) with the unsubstituted N6-position. The maximal inhibition of adenylyl cyclase by all adenosine derivatives (10 mM) is significantly different from the inhibition mediated by the corresponding NECA derivative HENECA (p < 0.05, Student’s t-test). Fig. 1B shows the A3AR efficacy of 2-phenylethynyl derivatives with the same pattern of 50 - and N6-substitutions as shown in Fig. 1A for HEADO analogs. Again, the ribose-modified PENECA (2) shows full agonistic activity as it inhibits forskolin-stimulated adenylyl cyclase activity to a similar extent as the reference compound NECA. In contrast, compounds 2a–d with the unmodified ribose are all partial agonists showing a lower efficacy than the corresponding HEADO derivatives 1a–d. The methyl-, ethyl-, or isopropyl-substituents in the N6-position of compounds 1b–d do not affect the efficacy to a significant degree. Of all compounds tested the 2-phenylethynyl adenosine derivatives show the lowest efficacy although PENECA presents as a full agonist. As a third 2-substituted adenosine derivative we studied PHPADO (3a) and analogs. The ribose-modified PHPNECA (3) was previously introduced as a nonselective high-affinity agonist [37]. Fig. 1C confirms that PHPNECA (3) is a full A3AR agonist. As in the case of HEADO (1a) and PEADO (2a), a reduced efficacy is shown for the 2-(3-hydroxy-3-phenyl)propyn-1-yl-derivative PHPADO (3a)

D. Dal Ben et al. / Biochemical Pharmacology 87 (2014) 321–331

325

Fig. 1. Efficacy of 2-substituted Ado and NECA derivatives. Shown is the A3AR-mediated inhibition of forskolin-stimulated adenylyl cyclase activity (100%) with the Ado derivatives bearing the 2-hexyn-1-yl (A), 2-phenylethynyl (B), and 2-(3-hydroxy-3-phenyl)propyn-1-yl (C) substituents. The horizontal line represents maximal inhibition by the full agonist NECA for comparison. All adenosine derivatives caused a significantly reduced inhibition of adenylyl cyclase compared to the respective NECA derivatives (p < 0.05, Student’s t-test).

with an intact ribose and no effect on the efficacy by an additional methyl-substituents in the N6-position was observed. Larger substituents like ethyl or isopropyl seem to further reduce the efficacy to some degree. Taken together these results, we can conclude that in general the presence of a hydroxymethyl or an N-ethylcarboxamido function at the 50 -position of the sugar ring seems to be of critical importance to determine partial or full agonistic activity of 2alkynyl-substituted Ado and NECA derivatives, respectively, while the presence of substituents in the N6-position seems to play only a minor role in the determination of the efficacy of these compounds. Table 2 Results of conformational analysis of compounds 1, 1a, 2, 2a, 3, 3a. Number of generated conformations (n. conf.), position of first syn conformation and its relative (%) position, and percentage of syn conformations are indicated. cpd

n. conf.a

first syn-conf.b

rel. pos.c

% syn-conf.d

1 1a 2 2a 3 3a

682 674 273 227 442 383

214 131 79 48 136 74

31.4 19.4 28.9 21.1 27.8 19.3

9.7 34.7 12.8 28.6 14.5 32.6

a Number of generated conformations in the conformational analysis output database. b Position of the first syn-conformation in the conformational analysis output database ranked according to Potential Energy parameter. c Relative position of the first syn-conformation obtained according to the equation rel. pos. = 100  (first syn-conf.)/(n. conf.). d Amount of the syn-conformations out of the total number of generated conformations.

3.2. Molecular modeling Molecular modeling methods were employed to analyze the conformational behavior of the studied compounds and to possibly identify structural factors responsible for their different efficacy at A3AR. In a first stage of this work, a conformational analysis study was performed with a special focus on the compounds tendency to assume anti- or syn-conformations. The anti/syn conformation topic is of interest due to the fact that, as described in the introduction, the presence of 8-substituents in 2-alkynylAdo derivatives led to the development of compounds presenting as partial agonists at A2AAR and this pharmacological behavior was supposed to be related to a possibly slightly higher stability of the syn-conformation compared to the anti-conformation [46]. A second reason of interest is that the shift from A3AR agonist to antagonist profile given by the transfer of the same alkynyl substituents from the 2- to 8-position of Ado was interpreted on the basis of the ability of 8-alkynyl substituted Ado derivatives to assume only the syn-conformations due to the steric hindrance of the substituent in the 8-position [25,47]. A third set of data is available from previous studies reporting that for Ado and NECA both conformations are observed in solution, with the synconformation appearing as the prevalent one according to NMR studies and the only one observed for NECA in X-ray studies [67]. In contrast, the recently reported crystal structures of A2AAR in complex with Ado and NECA show that the two ligands are inserted in the binding site in the anti-conformation [57]. For the conformational analysis (MOE, version 2010.10 [53]) we considered only the N6-unsubstituted (prototypical) NECA and Ado compounds (1, 1a, 2, 2a, 3, 3a) given that the presence of

D. Dal Ben et al. / Biochemical Pharmacology 87 (2014) 321–331

326

substituents in the N6-position seems to play a minor role for the agonistic activity of Ado derivatives. Results of conformational analysis are summarized in Table 2, which reports for each analyzed compound the number of generated conformations (n. conf.), the position (first syn-conf.) of the first syn-conformation in the output database (in which the conformations were sorted according to a Potential Energy parameter, with the lowest energy conformations as first and the highest energy conformation as last), its relative position (rel. pos., in %), and the number (% syn-conf.) of the syn-conformations out of the total number of generated conformations. The relative position (rel. pos.) of the first syn-conformation was obtained according to the following equation:

rel: pos: ¼ 100 

first syn-conf : n: conf :

The obtained data (Table 2) indicate that the Ado derivatives present a significant number (about one third) of syn-conformations while NECA derivatives (1, 2, 3) exist at a higher relative percentage in the anti-conformations (90.3%, 87.2%, and 85.5%, respectively). Even considering the relative position (rel. pos.) of the first syn-conformation, it can be observed from Table 2 that the first syn-conformation of Ado derivatives appears at about the 20% of the output database, while in the case of NECA derivatives the relative position is at about 30% of the conformations library. As example of comparison, Fig. 2 shows anti- and synconformations generated for compounds 2a and 2. In the case of the Ado derivative (Fig. 2B), the anti-conformation can be stabilized by an internal H-bond interaction between the N3 atom and the 20 -OH group, while the syn-conformation can be stabilized by a hydrogen-bond between the same nitrogen atom and the OH

Fig. 2. (A) Numbering of Ado atoms. (B and C) Superimposition of anti- and synconformations of compounds 2a (B) and 2 (C); in both cases the two conformations were superimposed to match the adenine scaffold. Anti- and syn-conformations are colored dark and light, respectively.

group in the 50 -position. In the case of the NECA analog (Fig. 2C), the anti-conformation can be stabilized similarly to the Ado derivative, while the syn-conformation can be made stable by a hydrogen-bond between the scaffold nitrogen N3 atom and the polar amide hydrogen atom of 50 -N-ethylcarboxamido function, although the 2-phenylethynyl group may be an obstacle for the Nethylcarboxamido function to stably interact with the purine scaffold. This sterical hindrance may provide an explanation for the higher prevalence of the anti-conformation in NECA derivatives. These results still allow the presence of syn- and anticonformations for both Ado and NECA derivatives, even if the latter analogs show a reduced tendency to assume the synconformation. In a second part of the molecular modeling study, molecular dynamics methods were employed to analyze the behavior of the compounds within the A3AR binding site and to interpret the efficacy of the NECA and Ado derivatives, respectively. In particular, this study was aimed at simulating firstly the interaction of analyzed compounds with the binding site in an inactive state and secondly the early steps of reciprocal ligand– target adaptation and receptor activation. A special focus was devoted to the identification of potentially different structural features of the receptor binding site at the end of the simulation and to the comparison of full and partial agonist dynamics. It must be underlined that this study did not want to predict the conformational changes of the entire receptor protein and hence the effects of the ligand–target interaction on the intracellular domains of the receptor. On this base, the human A3AR model was developed by using the crystal structure of the A2AAR [48] in complex with the high affinity antagonist ZM241385 as a template and then fully embedded into a membrane bilayer model. In this sense, a crystal structure of A2AAR in complex with an agonist would have represented a conformational state of the binding site as already modified by the presence of an agonist. Like in the case of the step of conformational analysis, in this stage we considered only compounds without the N6-substitution (1, 1a, 2, 2a, 3, 3a). The protocol followed to carry out this modeling work was a combination of procedures utilized in two previous studies [68,69]. In particular, the modeling approach started from a preliminary manual docking analysis of Ado within the A2AAR crystal structure binding site followed by energy minimization. The A2AAR–Ado complex was then used as a template to build a homology model of human A3AR that was then refined with energy minimization and subjected to Monte Carlo analysis to explore the favorable Ado binding conformations. The input structure consisted of the ligand and a shell of receptor amino acids within 6 A˚ distance from the ligand. A second external shell of all the residues within a distance of 8 A˚ from the first shell was kept fixed. During the Monte Carlo conformational searching, the input structure was modified by random changes in torsion angles (for all input structure residues), and molecular position (for the ligand). Hence, the ligand was left free to be continuously re-oriented and re-positioned within the binding site and the conformation of both ligand and internal shell residues could be explored and reciprocally relaxed. This stage was crucial to provide A3AR binding site conformations able to accommodate the nucleoside before docking and dynamics analyses. The best Ado-receptor complex was energetically minimized and then inserted into a membrane bilayer (consisting of phosphatidylcholine residues) previously built with the aid of the ‘‘membrane builder’’ tool within the VMD software suite [65] and a water soak box was added within MOE [53]. Lipids and water molecules clashing with the protein were removed. The obtained receptor–membrane complex was subjected to a first energy minimization process (keeping fixed the receptor coordinates and leaving free to move a shell of membrane and solvent residues within a distance of 10 A˚ from receptor atoms) and then checked

D. Dal Ben et al. / Biochemical Pharmacology 87 (2014) 321–331

327

group and the N-7 atom gave H-bonding with the conserved asparagine residue Asn2506.55. In particular, the N6-amine was located in a sub-pocket built by Val169 (EL2), Ile2536.58, Val259 (EL3), and Leu2647.39 residues. The role of these amino acids in providing a hydrophobic environment suitable for the introduction in the N6-position of small alkyl groups was already described [68]. The binding data in Table 1 confirm the beneficial effect of introduction of small alkyl groups in this position on the A3AR affinity for all Ado derivatives. Each A3AR–ligand complex was subjected to molecular dynamics. In this phase and for each receptor–ligand–membrane system, both receptor and ligand atoms were left free to move, just as for the shell of membrane and solvent residues within a distance of 10 A˚ from the receptor. The molecular dynamics simulation was carried out for 10 ns at 310 K, after the system had been heated to the final temperature (1 ns). After molecular dynamics simulation, each complex was then subjected to energy minimization and evaluation (post-dynamics conformations). The analysis of the obtained ligand–target complexes was focused in particular on the orientation of the ligand within the A3AR binding site, the conformation of the ligand itself, and the

Fig. 3. (A) Post-docking conformation of PEADO (2a) in the A3AR model. Key receptor residues in proximity of the N6- and 50 -position of ligand are indicated. (B) Close-up view of the ‘‘50 -sub-pocket’’ between TM3-TM5-TM6 domains with indication of involved residues.

for the potential presence of vacuum regions in the lipid-solvent phase in the proximity of the receptor. The system was then subjected to molecular dynamics simulation (100 ps after 30 ps of heating to 310 K) with the same settings. The aim of this first simulation was to relax membrane and solvent residues working as interface between the receptor model and the rest of the bilayersolvent environment. After an energy minimization stage, the receptor–membrane model was used for docking studies of selected compounds within MOE. Top score docking conformation for each compound was selected and each A3AR–ligand complex was subjected to energy minimization (post-docking conformations). The minimized docking conformations of the analyzed compounds presented highly similar features to the binding mode of MECA (50 -N-methylcarboxamidoadenosine) derivatives bearing a methyl group in the N6-position and an arylalkynyl substituent in the 2-position that we previously reported [42,68]. Both Ado and NECA derivatives were located in the binding site with an anticonformation, with the adenine scaffold positioned between transmembrane domains 3 (TM3), TM6, and TM7, the 8- and 9positions pointing toward the core of the receptor, and the 2-substituents pointing outward (Fig. 3A). The ribose moiety was located deeply between the transmembrane helix bundle, in a region between TM3, TM6, and TM7 in close proximity to the conserved tryptophan (Trp2436.48, 6.48 being the helix position according to the numbering system suggested by Ballesteros and Weinstein [70]) side chain. The ribose ring presented the hydroxyl groups in the 20 - and 30 -position for H-bond interaction with the Ser2717.42 and His2727.43 side chains. The 50 -group was located in a sub-pocket (from now on called 50 -sub-pocket) between TM3 (Leu913.33, Thr943.36, and His953.37), TM5 (Ser1815.42 and Ile1865.47), and TM6 (Trp2436.48) residues (Fig. 3B). The N6-amino

Fig. 4. Post-docking and post-dynamics conformations of A3AR–Ado derivatives 1a, 2a, and 3a (A, B, and C, respectively). Post-docking receptor–ligand system is lightcolored, while post-dynamics system is dark-colored. Ligand surfaces are colored accordingly. Key binding site residues are indicated.

328

D. Dal Ben et al. / Biochemical Pharmacology 87 (2014) 321–331

Fig. 6. Conformation of HEADO (1a) derivative within A3AR model before molecular dynamics simulation. The glycosyl torsion angle x (defined by O-C10 -N9-C4) is indicated.

Fig. 5. Post-docking and post-dynamics conformations of A3AR–NECA derivatives 1, 2, and 3 (A, B, and C, respectively). Post-docking receptor–ligand system is lightcolored, while post-dynamics system is dark-colored. Ligand surfaces are colored accordingly. Key binding site residues are indicated.

local modifications of the receptor binding site following the interaction with ligand, with a special focus on the 50 -sub-pocket. Considering the first parameter, the comparison of the postdocking and the post-dynamics conformations showed that the ligands approximately maintained the initial position and orientation although a slight displacement toward the extracellular environment was observed. Nevertheless, some minor differences were observed between Ado and NECA compounds. In particular, in the case of Ado derivatives, the 2-substituent moved outward while the ribose moiety moved toward the sub-pocket between TM3 and TM7 (Fig. 4). Hence, the displacement of Ado derivatives can be described as a rotation of the molecules toward the vertical receptor axis. In the case of NECA derivatives, the whole compound slightly moved outward (Fig. 5) while still maintaining the original orientation. As reported above, recently reported crystal structures of the A2AAR in complex with Ado and NECA show that these two ligands are inserted in the binding site with almost identical conformations, positions, and orientations [57]. The different reorientations between Ado and NECA derivatives observed in this study were interpreted considering two structural factors. On the

one hand, the rigidity of the substituted alkynyl group in the 2position did not allow this moiety to be easily accommodated within the receptor pocket and the displacement during molecular dynamics took the 2-substituent far from the proximal receptor residues in TM2, TM7, and EL2 domains. This structural factor had a similar effect on both Ado and NECA derivatives, even if the effect could be modulated by the different groups linked to the alkynyl moiety (n-butyl group for HEADO derivatives and HENECA, phenyl ring for PEADO derivatives and PENECA, hydroxyl(phenyl)methyl group for PHPADO derivatives and PHPNECA) and hence by the different rigidity of the entire 2-substituent. On the other hand, the presence of a hydroxyl or an N-ethylcarboxamido group in the 50 position (Ado and NECA derivatives, respectively) caused a different anchoring interaction within the 50 -sub-pocket residues. In particular, the 50 -N-ethylcarboxamido group of NECA derivatives was inserted in this sub-pocket before and after molecular dynamics simulation, while during the simulation the 50 -hydroxyl function of Ado derivatives exited from this sub-pocket because of the smaller size of the hydroxyl group compared to the 50 -Nethylcarboxamido substituent and because of its reduced interaction with the sub-pocket residues in contrast to the corresponding NECA group. Considering the conformation of the ligand within the target binding site, the combination of the exiting of the 50 -hydroxyl group of Ado derivatives from the 50 -sub-pocket together with the displacement of the same molecules with a rotation toward the vertical axis made the ribose group be located in a broader space between TM2, TM3, TM6, and TM7. As a consequence, the ribose group of Ado derivatives was more free to rotate about the N9-C10 bond and this rotation was observed in some cases during molecular dynamics. In particular, while at the beginning of the simulation all Ado derivatives HEADO (1a), PEADO (2a), and PHPADO (3a) presented a glycosyl torsion angle x (defined by O-C10 -N9-C4, see HEADO in Fig. 6 as example) clearly indicating an anti-conformation (x = 172.18, x = 142.58, x = 173.78, respectively), at the end of the dynamics simulation the same parameter showed values still within the anti-conformation range but indicating a slight shift (average 358) toward an anti/syn intermediate state (x = 167.18, x = 179.18, x = 128.58, respectively). In the case of NECA derivatives HENECA (1), PENECA (2), and PHPNECA (3), before the dynamics study the torsion angle x indicated an anticonformation (x = 146.38, x = 144.68, x = 137.18, respectively) that was conserved at the end of the simulation (x = 154.28,

D. Dal Ben et al. / Biochemical Pharmacology 87 (2014) 321–331

Fig. 7. A. Conformation of PEADO (2a) shown in the A3AR model before (light) and after (dark) molecular dynamics simulation. It can be seen the different orientation of the ribose moiety in the two conformations. B. Conformation of PENECA (2) shown in the A3AR model before (light) and after (dark) molecular dynamics simulation. The ribose moiety maintained the initial orientation.

x = 145.78, x = 143.88, respectively). Fig. 7 illustrates the comparison of the post-docking and post-dynamics conformations of PEADO (2a) and PENECA (2), highlighting the different orientation of the sugar moiety of Ado derivative after dynamics simulation. Our analysis considered also local modifications of the receptor binding site following the dynamics study. Reference data were chosen from recent studies [71,72] that compared several crystal structures of GPCRs with the aim to describe a set of conformational changes of receptor structures related to receptor activation. Further information could be obtained from the comparison of antagonist- and agonist-bound crystal structures of the b1adrenergic receptor [73,74] and the A2AAR [48,75]. At the binding site level, the inter-helical interactions between TM3 and TM5 as well as the reciprocal orientation of TM3, TM5, and TM6 are reported to play a key role during the receptor activation process. In particular, an interaction between residues 3.37 and 5.46 (according to the numbering system suggested by Ballesteros and Weinstein [70]; these residues in the A3AR are His95 and Trp185 and correspond to Thr126 and Ser215 in the b1-adrenergic receptor and Gln89 and Cys185 in the A2AAR, respectively) links TM3 and TM5 in the antagonist-bound crystal structures while it is lost in the agonist-bound receptors. The agonist-bound structure of the A2AAR also reveals interaction of Gln893.37 with Asn1815.42 (hence with a residue located one helix turn above residue 5.46 in the antagonist-bound structure). In the case of A3AR, His953.37 and Trp1855.46 showed a hydrophobic/p interaction before the dynamics study. After molecular dynamics experiments an

329

Fig. 8. Comparison of conformational modifications between antagonist- and agonist-bound crystal structures of A2AAR (A) and post-docking and post-dynamics A3AR–PENECA (2) complex (B). (A) Crystal structures of A2AAR in complex with antagonist ZM241385 (PDB ID: 3EML [48]) and agonist UK-432097 (PDB ID: 3QAK [75]) are colored light and dark, respectively. The different conformations and interactions of residues Gln893.37, Cys1855.46, Asn1815.42, and Trp2466.48 are highlighted. (B) Post-docking and post-dynamics conformations of A3AR–PENECA complex are colored light and dark, respectively. The different conformations and interactions of residues His953.37, Trp1855.46, Ser1815.42, and Trp2436.48 are highlighted.

increased distance between these two residues was observed for both Ado and NECA derivatives. The His953.37 side chain is located in proximity to Ser1815.42 and possibly building a H-bond with its hydroxyl group, as in the case of Gln893.37 which is interacting with Asn1815.42 in the agonist-bound A2AAR (Fig. 8). The conserved tryptophan6.48 residue (Trp243 in the case of the A3AR) kept the initial conformation during dynamics simulations, but the position of its side chain was slightly displaced toward the TM6 backbone without changing the rotamer conformation (Figs. 7 and 8). These results seem to be in accordance with the observed positions of residues in analogous positions of the above cited antagonist- and agonist-bound A2AAR crystal structures. 4. Discussion Full and partial agonists of AR based on the Ado scaffold have been developed by numerous laboratories. The introduction of an alkynyl group in the 2-position of Ado led to the development of several derivatives that did not show full agonistic activity at all AR subtypes. In particular, the partial agonism of the adenosine derivative HEADO (1a) at A2AAR was previously reported [46]. The present study focused on the analysis of efficacy of a series of Ado

330

D. Dal Ben et al. / Biochemical Pharmacology 87 (2014) 321–331

and corresponding NECA derivatives bearing in the 2-position (ar)alkynyl chains and endowed with high affinity and different degrees of selectivity for the A3AR. Interestingly, all of the 2alkynylAdo derivatives showed only partial agonistic efficacy at A3AR and the introduction of small alkyl substituents in the N6position of the same derivatives only modulated the affinity for this receptor but did not significantly modify the efficacy of the compounds. On the contrary, the modification of the 50 -group of 2alkynylAdos to obtain the corresponding NECA derivatives produced full agonists. Docking analysis of these compounds into an A3AR homology model showed that the binding modes present similar position, conformation, and interaction with the receptor binding site. Subsequent molecular dynamics studies highlighted the key role of the rigid 2-alkynyl group in causing a partial reorientation of the compounds with respect to the original binding mode. This effect was particularly evident for the Ado derivatives that showed a rotation toward the vertical axis of the receptor. In addition, the presence of a 50 -N-ethylcarboxamido group allowed the NECA derivatives to conserve the original interaction with the receptor residues within the 50 -sub-pocket, while the unmodified ribose moiety of Ado derivatives partially lost the analogous interaction with the receptor and got a higher rotational and conformational freedom compared to the 2alkynylNECAs. Furthermore, dynamic changes of residues located in TM3–TM5–TM6 and constituting part of the binding site of the A3AR were in agreement to what was observed in the comparison of antagonist- and agonist-bound A2AAR crystal structures. This demonstrates an effect of Ado and NECA derivatives on residues reported to be involved in the receptor activation mechanism. Taken together all these factors, we can hypothesize that the ability of 2-alkynylNECAs to fully activate the human A3AR could be due to the cooperative roles of the 2-alkynyl groups in providing a particular orientation of ligands and the 50 -substituents acting as a constraint for productive ligand–target interaction. Conversely, following ligand reorientation due to the 2-alkynyl group, the unmodified ribose of Ados results in a less constrained position and is subjected to a slight rotation about the glycosyl torsion angle x, leading to a ligand–target interaction able to only partially activate the receptor. Although we cannot state that Ado derivatives have a clear tendency to assume a syn-conformation within the A3AR binding site, we can hypothesize that the different ‘‘occupancy’’ of 50 -sub-pocket after ligand reorientation and hence the different interaction with TM3, TM5, and TM6 residues is crucial for the distinct efficacy of Ado versus NECA derivatives, resulting in partial or full agonistic activity, respectively. Acknowledgments This work was supported by a grant from the Italian Ministry for University and Research (PRIN2010) and by Fondo di Ricerca di Ateneo (University of Camerino). References [1] Fredholm BB, IJzerman AP, Jacobson KA, Linden J, Muller CE. International Union of Basic and Clinical Pharmacology. LXXXI. Nomenclature and classification of adenosine receptors – an update. Pharmacol Rev 2011;63: 1–34. [2] Fishman P, Bar-Yehuda S. Pharmacology and therapeutic applications of A3 receptor subtype. Curr Top Med Chem 2003;3:463–9. [3] Moro S, Gao ZG, Jacobson KA, Spalluto G. Progress in the pursuit of therapeutic adenosine receptor antagonists. Med Res Rev 2006;26:131–59. [4] Headrick JP, Peart J. A3 adenosine receptor-mediated protection of the ischemic heart. Vascul Pharmacol 2005;42:271–9. [5] Merighi S, Mirandola P, Varani K, Gessi S, Leung E, Baraldi PG, et al. A glance at adenosine receptors: novel target for antitumor therapy. Pharmacol Ther 2003;100:31–48. [6] Gessi S, Merighi S, Varani K, Leung E, Mac Lennan S, Borea PA. The A3 adenosine receptor: an enigmatic player in cell biology. Pharmacol Ther 2008;117:123–40.

[7] Cristalli G, Volpini R. Adenosine receptors: chemistry and pharmacology. Curr Top Med Chem 2003;3:355–469. [8] Jacobson KA, Gao ZG. Adenosine receptors as therapeutic targets. Nat Rev Drug Discov 2006;5:247–64. [9] Borea PA, Gessi S, Bar-Yehuda S, Fishman P. A3 adenosine receptor: pharmacology and role in disease. Handb Exp Pharmacol 2009;193:297–327. [10] Bar-Yehuda S, Silverman MH, Kerns WD, Ochaion A, Cohen S, Fishman P. The anti-inflammatory effect of A3 adenosine receptor agonists: a novel targeted therapy for rheumatoid arthritis. Expert Opin Investig Drugs 2007;16:1601–13. [11] Bar-Yehuda S, Luger D, Ochaion A, Cohen S, Patokaa R, Zozulya G, et al. Inhibition of experimental auto-immune uveitis by the A3 adenosine receptor agonist CF101. Int J Mol Med 2011;28:727–31. [12] Cohen S, Stemmer SM, Zozulya G, Ochaion A, Patoka R, Barer F, et al. CF102 an A3 adenosine receptor agonist mediates anti-tumor and anti-inflammatory effects in the liver. J Cell Physiol 2011;226:2438–47. [13] Morschl E, Molina JG, Volmer JB, Mohsenin A, Pero RS, Hong JS, et al. A3 adenosine receptor signaling influences pulmonary inflammation and fibrosis. Am J Respir Cell Mol Biol 2008;39:697–705. [14] Varani K, Padovan M, Vincenzi F, Targa M, Trotta F, Govoni M, et al. A2A and A3 adenosine receptor expression in rheumatoid arthritis: upregulation, inverse correlation with disease activity score and suppression of inflammatory cytokine and metalloproteinase release. Arthritis Res Ther 2011;13:R197. [15] Klotz K-N. Adenosine receptors and their ligands. Naunyn-Schmiedeberg’s Arch Pharmacol 2000;362:382–91. [16] Cristalli G, Muller CE, Volpini R. Recent developments in adenosine A2A receptor ligands. Handb Exp Pharmacol 2009;193:59–98. [17] Jacobson KA, Klutz AM, Tosh DK, Ivanov AA, Preti D, Baraldi PG. Medicinal chemistry of the A3 adenosine receptor: agonists, antagonists, and receptor engineering. Handb Exp Pharmacol 2009;193:123–59. [18] Kalla RV, Zablocki J, Tabrizi MA, Baraldi PG. Recent developments in A2B adenosine receptor ligands. Handb Exp Pharmacol 2009;193:99–122. [19] Kiesman WF, Elzein E, Zablocki J. A1 adenosine receptor antagonists, agonists, and allosteric enhancers. Handb Exp Pharmacol 2009;193:25–58. [20] Beukers MW, Chang LC, von Frijtag Drabbe Kunzel JK, Mulder-Krieger T, Spanjersberg RF, Brussee J, et al. New, non-adenosine, high-potency agonists for the human adenosine A2B receptor with an improved selectivity profile compared to the reference agonist N-ethylcarboxamidoadenosine. J Med Chem 2004;47:3707–9. [21] Chang LC, von Frijtag Drabbe Kunzel JK, Mulder-Krieger T, Spanjersberg RF, Roerink SF, van den Hout G, et al. A series of ligands displaying a remarkable agonistic-antagonistic profile at the adenosine A1 receptor. J Med Chem 2005;48:2045–53. [22] van der Wenden EM, Carnielli M, Roelen HC, Lorenzen A, von Frijtag Drabbe Kunzel JK, IJzerman AP. 50 -Substituted adenosine analogs as new high-affinity partial agonists for the adenosine A1 receptor. J Med Chem 1998;41:102–8. [23] van Tilburg EW, van der Klein PA, von Frijtag Drabbe Kunzel J, de Groote M, Stannek C, Lorenzen A, et al. 50 -O-alkyl ethers of N,2-substituted adenosine derivatives: partial agonists for the adenosine A1 and A3 receptors. J Med Chem 2001;44:2966–75. [24] Lohse MJ, Klotz K-N, Diekmann E, Friedrich K, Schwabe U. 20 ,30 -Dideoxy-N6cyclohexyladenosine: an adenosine derivative with antagonist properties at adenosine receptors. Eur J Pharmacol 1988;156:157–60. [25] Volpini R, Costanzi S, Lambertucci C, Vittori S, Klotz K-N, Lorenzen A, et al. Introduction of alkynyl chains on C-8 of adenosine led to very selective antagonists of the A3 adenosine receptor. Bioorg Med Chem Lett 2001; 11:1931–4. [26] Lohse MJ, Klotz K-N, Schwabe U, Cristalli G, Vittori S, Grifantini M. 2-ChloroN6-cyclopentyladenosine: a highly selective agonist at A1 adenosine receptors. Naunyn-Schmiedeberg’s Arch Pharmacol 1988;337:687–9. [27] Klotz K-N, Lohse MJ, Schwabe U, Cristalli G, Vittori S, Grifantini M. 2-ChloroN6-[3H]cyclopentyladenosine ([3H]CCPA) – a high affinity agonist radioligand for A1 adenosine receptors. Naunyn-Schmiedeberg’s Arch Pharmacol 1989;340:679–83. [28] Klotz K-N, Hessling J, Hegler J, Owman C, Kull B, Fredholm BB, et al. Comparative pharmacology of human adenosine receptor subtypes – characterization of stably transfected receptors in CHO cells. Naunyn-Schmiedeberg’s Arch Pharmacol 1998;357:1–9. [29] Hutchison AJ, Webb RL, Oei HH, Ghai GR, Zimmerman MB, Williams M. CGS 21680C, an A2 selective adenosine receptor agonist with preferential hypotensive activity. J Pharmacol Exp Ther 1989;251:47–55. [30] Niiya K, Olsson RA, Thompson RD, Silvia SK, Ueeda M. 2-(N0 -alkylidenehydrazino)adenosines: potent and selective coronary vasodilators. J Med Chem 1992;35:4557–61. [31] Niiya K, Thompson RD, Silvia SK, Olsson RA. 2-(N0 -aralkylidenehydrazino)adenosines: potent and selective coronary vasodilators. J Med Chem 1992; 35:4562–6. [32] van Tilburg EW, von Frijtag Drabbe Kunzel J, de Groote M, IJzerman AP. 2,50 Disubstituted adenosine derivatives: evaluation of selectivity and efficacy for the adenosine A1, A2A, and A3 receptor. J Med Chem 2002;45:420–9. [33] Gao ZG, Blaustein JB, Gross AS, Melman N, Jacobson KA. N6-Substituted adenosine derivatives: selectivity, efficacy, and species differences at A3 adenosine receptors. Biochem Pharmacol 2003;65:1675–84. [34] Gao ZG, Jeong LS, Moon HR, Kim HO, Choi WJ, Shin DH, et al. Structural determinants of efficacy at A3 adenosine receptors: modification of the ribose moiety. Biochem Pharmacol 2004;67:893–901.

D. Dal Ben et al. / Biochemical Pharmacology 87 (2014) 321–331 [35] Gao ZG, Mamedova LK, Chen P, Jacobson KA. 2-Substituted adenosine derivatives: affinity and efficacy at four subtypes of human adenosine receptors. Biochem Pharmacol 2004;68:1985–93. [36] Elzein E, Palle V, Wu Y, Maa T, Zeng D, Zablocki J. 2-Pyrazolyl-N6-substituted adenosine derivatives as high affinity and selective adenosine A3 receptor agonists. J Med Chem 2004;47:4766–73. [37] Klotz K-N, Camaioni E, Volpini R, Kachler S, Vittori S, Cristalli G. 2-Substituted N-ethylcarboxamidoadenosine derivatives as high-affinity agonists at human A3 adenosine receptors. Naunyn-Schmiedeberg’s Arch Pharmacol 1999;360: 103–8. [38] Cristalli G, Camaioni E, Costanzi S, Vittori S, Volpini R, Klotz K-N. Characterization of potent ligands at human recombinant adenosine receptors. Drug Dev Res 1998;45:176–81. [39] Volpini R, Costanzi S, Lambertucci C, Vittori S, Cristalli G. Purine nucleosides bearing 1-alkynyl chains as adenosine receptor agonists. Curr Pharm Des 2002;8:2285–98. [40] Volpini R, Costanzi S, Lambertucci C, Taffi S, Vittori S, Klotz K-N, et al. N6-alkyl2-alkynyl derivatives of adenosine as potent and selective agonists at the human adenosine A3 receptor and a starting point for searching A2B ligands. J Med Chem 2002;45:3271–9. [41] Volpini R, Dal Ben D, Lambertucci C, Taffi S, Vittori S, Klotz K-N, et al. N6methoxy-2-alkynyladenosine derivatives as highly potent and selective ligands at the human A3 adenosine receptor. J Med Chem 2007;50:1222–30. [42] Volpini R, Buccioni M, Dal Ben D, Lambertucci C, Lammi C, Marucci G, et al. Synthesis and biological evaluation of 2-alkynyl-N6-methyl-50 -N-methylcarboxamidoadenosine derivatives as potent and highly selective agonists for the human adenosine A3 receptor. J Med Chem 2009;52:7897–900. [43] Dal Ben D, Buccioni M, Lambertucci C, Marucci G, Volpini R, Cristalli G. The importance of alkynyl chain presence for the activity of adenine nucleosides/ nucleotides on purinergic receptors. Curr Med Chem 2011;18:1444–63. [44] Klotz K-N, Falgner N, Kachler S, Lambertucci C, Vittori S, Volpini R, et al. [3H]HEMADO – a novel tritiated agonist selective for the human adenosine A3 receptor. Eur J Pharmacol 2007;556:14–8. [45] Gao ZG, Jacobson KA. 2-Chloro-N6-cyclopentyladenosine, adenosine A1 receptor agonist, antagonizes the adenosine A3 receptor. Eur J Pharmacol 2002;443:39–42. [46] van Tilburg EW, Gremmen M, von Frijtag Drabbe Kunzel J, de Groote M, IJzerman AP. 2,8-Disubstituted adenosine derivatives as partial agonists for the adenosine A2A receptor. Bioorg Med Chem 2003;11:2183–92. [47] Costanzi S, Lambertucci C, Vittori S, Volpini R, Cristalli G. 2- and 8-alkynyladenosines: conformational studies and docking to human adenosine A3 receptor can explain their different biological behavior. J Mol Graph Model 2003;21:253–62. [48] Jaakola VP, Griffith MT, Hanson MA, Cherezov V, Chien EY, Lane JR, et al. The 2.6 angstrom crystal structure of a human A2A adenosine receptor bound to an antagonist. Science 2008;322:1211–7. [49] Cristalli G, Eleuteri A, Vittori S, Volpini R, Lohse MJ, Klotz K-N. 2-Alkynyl derivatives of adenosine and adenosine-50 -N-ethyluronamide as selective agonists at A2 adenosine receptors. J Med Chem 1992;35:2363–8. [50] Cristalli G, Volpini R, Vittori S, Camaioni E, Monopoli A, Conti A, et al. 2-Alkynyl derivatives of adenosine-50 -N-ethyluronamide: selective A2 adenosine receptor agonists with potent inhibitory activity on platelet aggregation. J Med Chem 1994;37:1720–6. [51] Cristalli G, Camaioni E, Vittori S, Volpini R, Borea PA, Conti A, et al. 2-Aralkynyl and 2-heteroalkynyl derivatives of adenosine-50 -N-ethyluronamide as selective A2a adenosine receptor agonists. J Med Chem 1995;38:1462–72. [52] Camaioni E, Di Francesco E, Vittori S, Volpini R, Cristalli G. Adenosine receptor agonists: synthesis and biological evaluation of the diastereoisomers of 2-(3hydroxy-3-phenyl-1-propyn-1-yl)NECA. Bioorg Med Chem 1997;5:2267–75. [53] Molecular Operating Environment, C.C.G., Inc., 1255 University St., Suite 1600, Montreal, Quebec, Canada H3B 3X3.

331

[54] Mohamadi F, Richards NGJ, Guida WC, Liskamp R, Lipton M, Caufield C, et al. MacroModel – an integrated software system for modeling organic and bioorganic molecules using molecular mechanics. J Comput Chem 1990;11: 440–67. [55] Stewart JJ. MOPAC: a semiempirical molecular orbital program. J Comput Aided Mol Des 1990;4:1–105. [56] Cornell WD, Cieplak P, Bayly CI, Gould IR, Merz KM, Ferguson DM, et al. A second generation force field for the simulation of proteins, nucleic acids, and organic molecules. J Am Chem Soc 1995;117:5179–97. [57] Lebon G, Warne T, Edwards PC, Bennett K, Langmead CJ, Leslie AG, et al. Agonist-bound adenosine A2A receptor structures reveal common features of GPCR activation. Nature 2011;474:521–5. [58] Halgren TA, Merck Molecular Force, Field I. Basis, form, scope, parameterization, and performance of MMFF94. J Comput Chem 1996;17:490–519. [59] Halgren TA. Merck molecular force field. II. MMFF94 van der Waals and electrostatic parameters for intermolecular interactions. J Comput Chem 1996;17:520–52. [60] Halgren TA. Merck molecular force field. III. Molecular geometries and vibrational frequencies for MMFF94. J Comput Chem 1996;17:553–86. [61] Halgren TA. Merck molecular force field. IV. Conformational energies and geometries for MMFF94. J Comput Chem 1996;17:587–615. [62] Halgren TA. MMFF VI. MMFF94s option for energy minimization studies. J Comput Chem 1999;20:720–9. [63] Halgren TA. MMFF VII. Characterization of MMFF94, MMFF94s, and other widely available force fields for conformational energies and for intermolecular-interaction energies and geometries. J Comput Chem 1999;20:730–48. [64] Halgren TA, Nachbar R. Merck molecular force field. V. Extension of MMFF94 using experimental data, additional computational data, and empirical rules. J Comput Chem 1996;17:616–41. [65] Humphrey W, Dalke A, Schulten K. VMD – visual molecular dynamics. J Mol Graphics 1996;14:33–8. [66] Jorgensen WL, Maxwell DS, Tirado-Rives J. Development and testing of the OPLS all-atom force field on conformational energetics and properties of organic liquids. J Am Chem Soc 1996;117:11225–36. [67] Moos WH, Hamilton HW, Ortwine DF, Taylor MD, McPhail AT. Adenosine receptor agonists. X-ray crystal structure of Neca 1-(6-amino-9H-purin-9-Yl)1-deoxy-N-ethyl-b-D-ribofuranuronamide. Nucleosides Nucleotides 1989;8: 449–61. [68] Dal Ben D, Buccioni M, Lambertucci C, Marucci G, Thomas A, Volpini R, et al. Molecular modeling study on potent and selective adenosine A3 receptor agonists. Bioorg Med Chem 2010;18:7923–30. [69] Dal Ben D, Antonini I, Buccioni M, Lambertucci C, Marucci G, Vittori S, et al. Molecular modeling studies on the human neuropeptide S receptor and its antagonists. ChemMedChem 2010;5:371–83. [70] Ballesteros JA, Weinstein H. Integrated methods for the construction of threedimensional models and computational probing of structure-function relations in G protein-coupled receptors. Methods Neurosci 1995;25:366–428. [71] Deupi X, Standfuss J, Schertler G. Conserved activation pathways in G-proteincoupled receptors. Biochem Soc Trans 2012;40:383–8. [72] Trzaskowski B, Latek D, Yuan S, Ghoshdastider U, Debinski A, Filipek S. Action of molecular switches in GPCRs – theoretical and experimental studies. Curr Med Chem 2012;19:1090–109. [73] Warne T, Serrano-Vega MJ, Baker JG, Moukhametzianov R, Edwards PC, Henderson R, et al. Structure of a b1-adrenergic G-protein-coupled receptor. Nature 2008;454:486–91. [74] Warne T, Moukhametzianov R, Baker JG, Nehme R, Edwards PC, Leslie AG, et al. The structural basis for agonist and partial agonist action on a b1-adrenergic receptor. Nature 2011;469:241–4. [75] Xu F, Wu H, Katritch V, Han GW, Jacobson KA, Gao ZG, et al. Structure of an agonist-bound human A2A adenosine receptor. Science 2011;332:322–7.