Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The G protein-coupled receptors (GPCRs) are a principal target of currently used drugs. Traditionally, the development of such drugs has been concentrated on optimizing the interaction of molecules with the binding site for endogenous agonists and competitive antagonists (orthosteric site). In addition to the orthosteric site, drugs can interact with binding sites on the receptor molecule that are distinct from the orthosteric site (i.e., allosteric sites) (Bruns and Fergus, 1990). Several lines of evidence suggest that allosteric modulators could offer advantages over orthosteric ligands (Christopoulos, 2002; Christopoulos and Kenakin, 2002), such as greater selectivity.

In the field of ion channel-coupled GABA_A receptors, benzodiazepines as positive allosteric modulators have been used successfully in therapeutics. By contrast, directly acting agonists for GABA_A receptors have not found therapeutic application because of their potential side effects (Macdonald and Olsen, 1994). The positive allosteric modulator for nicotinic receptors, galanthamine, has also been successfully used in the treatment of Alzheimer's disease (Olin and Schneider, 2002). In the field of GPCRs, it is encouraging that, in terms of therapeutic application, positive allosteric modulators for the calcium-sensing GPCR are now in clinical trials for the treatment of primary hyperparathyroidism (Conigrave et al., 2000).

Allosteric modulators of adenosine receptors (ARs) are of potential clinical use (Linden, 1997). Allosteric modulation of the A₁AR has been extensively investigated (Bruns and Fergus, 1990; Linden, 1997; Kourounakis et al., 2001). Allosteric modulation of the A_2AAR has also been reported (Gao and IJzerman, 2000). Allosteric modulators for the A₃AR, including DU124183, VUF5455, and amiloride derivatives, have recently been identified and characterized (Fig. 1) (Gao et al., 2001, 2002b, 2003). However, the identity of amino acid residues in the A₃AR that are involved in the allosteric modulation has not previously been studied.

View larger version (10K): [in this window] [in a new window]   Fig. 1.   Chemical structures of allosteric modulators.

To identify amino acid residues that may participate in the response to allosteric modulators, we examined the roles of selected residues from transmembrane helical domains (TMs) 1, 2, 3, 5, 6 and 7 and the second extracellular loop (EL) of the human A₃AR by site-directed mutagenesis. The residues were selected based on proximity to the putative sodium binding site, the conserved DRY motif in TM3, or the putative ligand binding site (Jacobson et al., 2001).

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Materials. The QuikChange site-directed mutagenesis kit was purchased from Stratagene (La Jolla, CA). A monoclonal antibody (12CA5) against a hemagglutinin epitope and adenosine deaminase were obtained from Roche Molecular Biochemicals (Indianapolis, IN) and goat anti-mouse IgG antibody conjugated with horseradish peroxidase was from Sigma (St. Louis, MO). Oligonucleotides used were synthesized by Bioserve Biotechnologies (Laurel, MD). The vector pcDNA3 was purchased from Invitrogen (Carlsbad, CA). Human A₃AR cDNA was provided by M. Atkinson, A. Townsend-Nicholson, and P. R. Schofield (Garvan Medical Institute, Sydney, Australia) and was subcloned in pcDNA3 as pcDNA3/hA₃R. N⁶-(4-amino-3-[¹²⁵I]iodobenzyl)adenosine-5'-N-methyluronamide ([¹²⁵I]I-AB-MECA; 2000 Ci/mmol) and [³H]8-ethyl-4-methyl-2-phenyl-(8R)-4,5,7,8-tetrahydro-1H-imidazo[2.1-i]purin-5-one ([³H]PSB-11) were from Amersham Biosciences (Buckinghamshire, UK); 2-chloro-N⁶-(3-iodobenzyl)adenosine-5'-N-methyluronamide (Cl-IB-MECA) and 5-(N,N-hexamethylene)amiloride (HMA) were from Sigma (St. Louis, MO). DU124183 and VUF5455 were from Leiden/Amsterdam Center for Drug Research (Leiden, The Netherlands). All the enzymes used in this study were obtained from New England Biolabs (Beverly, MA).

Site-Directed Mutagenesis. The protocols used were as described in the QuikChange site-directed mutagenesis kit (La Jolla, CA). Mutations were confirmed by DNA sequencing.

Numbering Scheme of GPCRs. The standardized numbering system of van Rhee and Jacobson (1996) was used to identify residues in the transmembrane domains of various receptors. Each residue is identified by two numbers: the first corresponds to the TM in which it is located, and the second indicates its position relative to the most conserved residue in that helix, arbitrarily assigned to 50. For example, His3.37 is the histidine in TM3 located 13 residues before the most conserved arginine, R3.50; Asn7.45 corresponds to Asn274.

Transient Expression of Wild-Type and Mutant Receptors in COS-7 Cells. COS-7 (African green monkey kidney) cells were grown in 100-mm cell culture dishes containing Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, 100 µg/ml streptomycin, and 2 µmol/ml glutamine. Cells were washed with phosphate-buffered saline containing calcium and then transfected with plasmid DNA (10 µg/dish) using the DEAE-dextran method (Cullen, 1987) for 1 h. The cells were then treated with 100 µM chloroquine for 3 h in culture medium and cultured for an additional 48 h at 37°C and 5% CO₂.

Membrane Preparation. After 48 h of transfection, COS-7 cells were harvested and homogenized using a Polytron homogenizer. The homogenates were centrifugated at 20,000g for 20 min, and the resulting pellet was re-suspended in 50 mM Tris-HCl buffer, pH 8.0, and stored at 80°C in aliquots. The protein concentration was determined by using the method of Bradford (1976).

[¹²⁵I]I-AB-MECA Binding Assay. For competitive binding assay, each tube contained 100 µl of membrane suspension, 50 µl of [¹²⁵I]I-AB-MECA (final concentration, 0.5 nM), and 50 µl of increasing concentrations of the test ligands in Tris-HCl buffer (50 mM, pH 8.0) containing 10 mM MgCl₂ and 1 mM EDTA. Nonspecific binding was determined using 10 µM Cl-IB-MECA. The mixtures were incubated at 25°C for 60 min. For dissociation kinetics, the protocols used were as we described previously (Gao et al., 2001). Briefly, [¹²⁵I]I-AB-MECA (0.5 nM) was preincubated with WT and mutant receptor membranes (8-20 µg protein) for 1 h at 25°C. Dissociation was started by adding 10 µM Cl-IB-MECA in the absence or presence of an allosteric modulator. Binding reactions were terminated by filtration through Whatman GF/B filters under reduced pressure using a MT-24 cell harvester (Brandell, Gaithersburg, MD). Filters were washed three times with 9 ml of ice-cold buffer. Radioactivity was determined in a -counter (5500B; Beckman Coulter, Fullerton, CA).

Binding of the Selective Antagonist, [³H]PSB-11, to A₃ARs. Membranes (60-100 µg protein) were incubated with 8 nM [³H]PSB-11 (Müller et al., 2002) at 25°C in a total assay volume of 400 µl for 60 min. Nonspecific binding was measured in the presence of 10 µM Cl-IB-MECA. Binding reactions were terminated by filtration through Whatman GF/B filters under reduced pressure using a MT-24 cell harvester (Brandel, Gaithersburgh, MD).

Statistical Analysis. Binding parameters were estimated using Prism software (GraphPAD, San Diego, CA). IC₅₀ values obtained from competition curves were converted to K_i values using the Cheng and Prusoff (1973) equation. Data were expressed as mean ± S.E.

Computational Methods. All calculations were performed on an SGI Octane workstation (SGI, Mountain View, CA). All ligand structures were constructed using the "Sketch Molecule" of SYBYL 6.8 (Tripos Inc., St. Louis, MO). A conformational search of allosteric modulators was performed by grid search, rotating in 60°, 180°, and 60° for all flexible bonds and 0 or 180° for the amide bond with Merck Molecular force field (MMFF94) and charge. After clustering the low-energy conformers from the result of the conformational search, the representative ones from all groups were reoptimized by semiempirical molecular orbital calculations using the PM3 method in the MOPAC 6.0 package. Each of the lowest energy conformation from PM3 optimization was used for DISCO (DIStance COmparisons) computation (Martin et al., 1993), followed by default option.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Agonist and Antagonist Binding to WT and Mutant A₃ARs. Before studying the possible involvement of residues in the allosteric modulation, we first examined the agonist and antagonist binding properties of WT and mutant A₃ARs using an agonist radioligand, [¹²⁵I]I-AB-MECA, and an antagonist radioligand, [³H]PSB-11. Several of the A₃AR mutations examined here (Table 1) were also studied previously (Gao et al., 2002a), whereas others were constructed based on predictions of molecular modeling and by homology to other GPCRs.

Effects of Mutations on the Rate of Dissociation of the Agonist Radioligand, [¹²⁵I]I-AB-MECA, from Human A₃ARs Expressed in COS-7 Cells in the Absence or Presence of Various Allosteric Modulators. The dissociation rates of [¹²⁵I]I-AB-MECA from WT and mutant human A₃ARs in the absence or presence of DU124183, HMA, and VUF5455 were examined. As shown in Fig. 2, the dissociation rate of [¹²⁵I]I-AB-MECA from the WT receptor was decreased by DU124183 (10 µM), HMA (100 µM), and VUF5455 (10 µM) to a similar extent, roughly by half. Surprisingly, F182A^5.43 and N274A^7.45 mutations completely abolished the allosteric effects of all three modulators. However, in the N30A^1.50 and D58N^2.50 mutant receptors, only HMA but not DU124183 or VUF5455 decreased the dissociation rate. In contrast to the D58N^2.50 mutation in TM2, the D107N^3.49 mutation in TM3 eliminated the allosteric effects of DU124183 but not HMA or VUF5455. The dissociation rates of [¹²⁵I]I-AB-MECA from T94A^3.36, H95A^3.37, K152A^EL2, W243A^6.48, L244A^6.49, and S247A^6.52 mutant receptors were reduced by the allosteric modulators by approximately half, similar to the WT receptor. The dissociation rates (k₁) of [¹²⁵I]I-AB-MECA in the absence and presence of various allosteric modulators are summarized in Table 2.

View larger version (25K): [in this window] [in a new window]   Fig. 2.   The dissociation of the agonist radioligand [125I]I-AB-MECA from WT and mutant ARs in the absence and presence of various allosteric modulators. (, control; , DU124183; , HMA; , VUF5455). The dissociation rates of [125I]I-AB-MECA from N30A1.50 and H95A3.37 were determined at 4°C, whereas WT and other mutant receptors were measured at 25°C. The concentration of [125I]I-AB-MECA was 0.5 nM. The mean k1 values calculated from three independent experiments are listed in Table 2.

Effects of Mutations on the Rate of Dissociation of the Antagonist Radioligand [³H]PSB-11 from Human A₃ARs Expressed on COS-7 Cells in the Absence and Presence of Various Allosteric Modulators. [³H]PSB-11 is a newly developed antagonist radioligand for A₃ receptors (Müller et al., 2002). Sodium ions (100 mM) exerted only slight effect on the equilibrium binding of [³H]PSB-11 to A₃ receptors. The K_d values of [³H]PSB-11 in the absence of and presence of sodium ions were 4.7 ± 0.8 and 3.8 ± 1.1 nM, respectively.

1

View larger version (22K): [in this window] [in a new window]   Fig. 3.   The dissociation of the antagonist radioligand [3H]PSB-11 from WT (A) and D58N (B) mutant ARs in the presence or absence of sodium ions and HMA. The final concentration of [3H]PSB-11 used was 8 nM and the incubation temperature was 25°C. The mean k1 values calculated from three independent experiments are listed in Table 3.

View this table: [in this window] [in a new window]   TABLE 3 Dissociation rates of the antagonist radioligand [3H]PSB-11 from WT and mutant receptors in the absence and presence of allosteric modulators Results were expressed as means ± S.E. from three independent experiments. The dissociation rates (k1) were determined at 25°C. The final concentration of [3H]PSB-11 used in this experiment was 8 nM.

Effects of Sodium Ions on the Equilibrium Binding of the Agonist Radioligand [¹²⁵I]I-AB-MECA to WT and Mutant A₃ARs. As described above, the effects of sodium ions on the dissociation rate of the antagonist [³H]PSB-11 from some mutant receptors could not be determined because of the loss of high-affinity antagonist binding. Alternatively, we examined the effect of sodium ions on the equilibrium binding of the agonist radioligand, [¹²⁵I]I-AB-MECA (0.5 nM), to WT and mutant receptors. As shown in Fig. 4, 100 mM NaCl induced an approximately 80% inhibition of the binding of [¹²⁵I]I-AB-MECA to WT receptors. The D58N^2.50, D107N^3.49, and F182A^5.43 mutant receptors were completely insensitive to 100 mM sodium ions, whereas N30A^1.50 and N274A^7.45 mutations induced a modest but significant increase of agonist binding in the presence of 100 mM NaCl. In the case of the T94A^3.36 and H95A^3.37 mutant receptors, 100 mM NaCl induced an approximately 50% inhibition of the agonist binding (Fig. 4). The effect of sodium ions in the S247A^6.52 mutant receptor was essentially the same as that in WT (Fig. 4). Similarly, the K152A^EL2, W243A^6.48, and L244A^6.49 mutations also did not significantly modify the modulatory effect of sodium ions. The percentage inhibition of [¹²⁵I]I-AB-MECA binding to these three mutant receptors by 100 mM NaCl was 73, 79, and 82%, respectively.

View larger version (50K): [in this window] [in a new window]   Fig. 4.   Effect of sodium ions on the equilibrium binding of the agonist radioligand, [125I]I-AB-MECA to WT and mutant ARs. [125I]I-AB-MECA (0.5 nM) was incubated with membranes (8-20 µg of protein) from COS-7 cells expressing WT or mutant receptors in total assay volume of 200 µl of Tris-HCl buffer, pH 8.0, containing 10 mM MgCl2 and 1 mM EDTA at 25°C for 60 min. Data were from three independent experiments performed in triplicate.

1

Competition by Allosteric Modulators of the Binding of [¹²⁵I]I-AB-MECA to WT and Mutant Receptors. All allosteric modulators tested in this study competed for the binding of [¹²⁵I]I-AB-MECA to human A₃ receptors (Gao et al., 2001, 2002b, 2003). Thus, we examined the effects of mutations on this binding. The IC₅₀ values are listed in Table 1. The N30A^1.50 and H95A^3.37 mutations induced a 3- to 30-fold decrease in affinity of the allosteric modulators. The D58N^2.50, T94A^3.36, D107N ^3.49, L244A^6.49, S247A^6.52, and N274A^7.45 mutations induced only a slight reduction in the affinity of these allosteric modulators for ligand-unoccupied receptors. The K152A mutations did not influence the affinity of HMA but induced a 3- and 8-fold decrease of affinity of DU124183 and VUF5455, respectively. Interestingly, the W243A^6.48 mutation induced a 4-fold increase of the affinity of HMA (Fig. 5), but resulted in a 3- and 10-fold decrease of the affinity for DU124183 and VUF5455, respectively. Hence, the W243A^6.48 mutation selectively enhanced the affinity of the amiloride analog, HMA. To further confirm the unique binding properties of amiloride analogs at this mutant receptor, we further tested the effect of amiloride on WT and W243A^6.48 mutant receptors. The IC₅₀ values of amiloride were 74 ± 12 and 36 ± 8 µM for WT and W243A^6.48 mutant receptors, respectively, consistent with the gain in affinity observed with the amiloride analog, HMA.

View larger version (16K): [in this window] [in a new window]   Fig. 5.   Competition of allosteric modulators for the binding of [125I]I-AB-MECA to WT and H95A and W243A mutant ARs. A, HMA; B, DU124183; C, VUF5455. The final concentration of [125I]I-AB-MECA used was 0.5 nM. Data are from a representative experiment in duplicate. The Ki values listed in Table 1 are from three independent experiments in duplicate.

Molecular Modeling. The present mutagenesis results supported the definition of putative orthosteric binding sites for agonist and antagonist that were consistent with the previously reported A₃AR model (Gao et al., 2002a,c). Previous docking results suggested that L91^3.33, T94^3.36, H95^3.37, K152^EL2, Q167^EL2, F168^EL2, F182^5.43, F239^6.44, W243^6.48, L246^6.51, S247^6.52, N250^6.55, I268^7.39, S271^7.42, H272^7.43, and N274^7.45 lined the orthosteric site, being located within 5 Å of the putative binding sites for nonselective AR ligands NECA and CGS15943. An extension of the binding site delineated by the side chains of V141^4.56, M177^5.38, and V178^5.39 served as a binding region for both the N⁶-substituent of Cl-IB-MECA and the benzyl group of MRS1220. The current mutational results indicated that M177^5.38, V178^5.39, N250^6.55, S271^7.42, and H272^7.43 may be critically involved in binding to A₃AR agonist and antagonist, consistent with the previous molecular model (Gao et al., 2002a,c). It should be noted that the loss of ligand binding may also be caused by the incorrect folding of the receptor, because the expression of these receptors on the cell surface was not determined. Mutation of L244^6.49 or S247^6.52 was without the effect on agonist/antagonist affinity, and these residues did not participate in direct interaction with the ligand according to the model, notwithstanding their proximity to the orthosteric site.

View larger version (48K): [in this window] [in a new window]   Fig. 6.   Rhodopsin-based molecular model of the A3AR, showing the previously defined docking mode of the A3-selective agonist Cl-IB-MECA (colored according to atom type) and a putative allosteric binding site for VUF5455 (orange) on TM 7. Shown are selected residues near the allosteric modulator and residues F182 and N274, which were found to be important for the allosteric action of VUF5455. The secondary structure of each TM is displayed by ribbon using different colors, which indicate TMs 1, 2, 3, 4, 5, 6, and 7 with red, orange, yellow, green, cyan, blue, and purple color, respectively. The coordinates of this A3AR model are available as from the Protein Data Bank ftp site under the name pdblo74.ent.Z. See http://www.rcsb.org/pdb/pdb_news2002.html#models_removal2 for more information.

View larger version (49K): [in this window] [in a new window]   Fig. 7.   Major interhelical H-bonding networks proposed for the hA3AR based on the rhodopsin structure, in the D582.50 region (A) and involving the DRY motif (B). The H-bonds are represented in yellow.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

We examined the effects on allosteric modulation and ligand binding of the mutation of amino acid residues of the human A₃AR that are hypothesized to be near one of three loci: the putative sodium binding site, the putative ligand binding site, and the DRY motif in TM3. Because it has been established that these three heterocyclic A₃AR allosteric modulators (DU124183, VUF5455, and HMA) also compete for radioligand binding at the orthosteric binding site, their affinities at WT and mutant receptors were examined. As for other heterocyclic antagonists (Gao et al., 2002a), the most substantial loss of affinity was observed for the H95A mutation (10- to 30-fold). Also, VUF5455 was selectively reduced in affinity at the N30A, D107N, K152A, and W243A mutant receptors. Conversely, DU124183 was selectively reduced in affinity at the F182A mutant receptor. Thus, the structurally diverse allosteric modulators have specific residues associated with the binding of each.

It was shown that certain residues contributed differently to ligand binding and to allosteric modulation either by heterocyclic derivatives or by sodium ions, suggesting a possible separation of the structural requirements for all of these processes. Most notably, the Ala substitution of Phe182^5.43 or Asn274^7.45 abolished the ability of all of the allosteric modulators, including HMA, DU124183, and VUF5455, to influence the agonist dissociation rate. Thus, although not critical for agonist binding, Phe182^5.43 and Asn274^7.45 were required in allosteric modulation.

Also, specific regions of the receptor were involved in ligand binding but seemed not to influence allosteric effects. For example, W243A^6.48 mutation caused a 4-fold increase of the affinity HMA for the unoccupied receptors, but it did not influence the effect of HMA on the dissociation of the agonist [¹²⁵I]I-AB-MECA. Concerning the mutations that evidently did not affect the allosterism, it must be noted that the effects of the allosteric ligands on the mutant receptors were studied at single concentrations of the compounds. Therefore, the present study can not rule out the possibility that a null mutation may in some instances be a counterbalancing effect on both parameters.

It was demonstrated that DU124183 (10 µM) inhibits [¹²⁵I]I-AB-MECA binding at WT by ~ 80% but only 10% at H95A mutant (Fig. 5B), yet the dissociation rates are halved for both WT and H95A mutant (Fig. 2). One explanation is that the mutation of H95 to Ala decreased the competitive binding potency but not the allosteric binding potency of DU124183. Another possible reason is that the temperature dependence of the affinity of DU124183 in WT (measured at 25°C) and H95A mutant receptors (measured at 4°C) may be different.

As we discussed in relation to the role of the highly conserved W243^6.48 in activation of the A₃AR (Gao et al., 2002a,c), the indole side chain displayed a characteristic movement in the model exclusively upon docking of agonist, whereas the unaltered geometry of the inactive conformation in the A₃AR could accommodate the binding of antagonist. Consistent with N274^7.45 being constrained by the indole side chain of W243^6.48 in the inactive conformation and W243^6.48 participating in the hydrophobic interaction with the antagonist MRS1220 but not with the agonist Cl-IB-MECA, the N274A^7.45 mutation caused a dramatic affinity decrease for the antagonist MRS1220 but had little effect on binding of the agonist Cl-IB-MECA. Thus, mutation of those residues predicted in modeling to take part in the stabilizing the inactive conformation may also affect the binding of ligand indirectly.

Molecular modeling and experimental results suggested that multiple binding modes of the allosteric modulators may be possible, considering their small size and low affinity compared with orthosteric ligands. Multiple binding/modulatory sites have already been proposed to be present in some other GPCRs. There seem to be two distinct allosteric sites on the M₁ receptor (Birdsall et al., 2001), and more than one allosteric site exists on the ₁-adrenergic receptor (Leppik and Birdsall, 2000). Some agonists (e.g., demox) also suggested the possibility of interaction with both orthosteric and allosteric sites of the M₂ muscarinic receptor (Angeli et al., 2002).

The possible location of the allosteric sites has been previously probed in several other GPCRs, although to a limited extent. In A₁ARs, a T277A^7.42 mutation both diminished agonist affinity and inhibited the effect of the A₁ receptor enhancer PD81,723 (Kourounakis et al., 2001). Distinct allosteric sites on the D₄ dopamine receptors have also been demonstrated by the study of sodium and zinc ions and methylisobutylamiloride (Schetz and Sibley, 2001). A single amino acid in TM7 has been found to be critical to the muscarinic receptor subtype selectivities of caracurine V derivatives and alkane-bisammonium ligands (Buller et al., 2002). In the M₁ muscarinic receptor, two residues, W101 and W400, are important for gallamine binding to the receptor, both in the unoccupied and in the antagonist (N-methylscopolamine)-bound state (Matsui et al., 1995). The latter residue may have a structural role, because it was important for the binding of ACh and number of antagonists. Studies of residues responsible for the m2 receptor selectivity of gallamine have used both chimeric and point-mutated receptors. Residues have been found in EL2 and EL3, especially the "EDGE" sequence in EL2 of M₂ muscarinic receptors (in M₁, the corresponding sequence is LAGQ) (Leppik et al., 1994; Gnagey et al., 1999). It has been suggested that the gallamine binding site in the M₁ muscarinic receptor is located in the region of the receptor structure close to W400 and the LAGQ sequence, and the cleft represents the access route to the binding site of ACh and other polar agonists and antagonists (Birdsall et al., 2001).

Through the study of muscarinic allosteric modulators, it was suggested that the allosteric site comprises the extracellular contact points located above the orthosteric site (Christopoulos et al., 1998), which is itself postulated to be located in the upper third of the inner transmembrane pore (Wess, 1993). An allosteric binding site for various charged molecules may be near the orthosteric site of the muscarinic receptor but at a more extracellular level (Christopoulos and Kenakin, 2002). Thus, a relatively vacant region at the top of TM6 and TM7 of the A₃ adenosine receptor was searched for the possibility location of an allosteric binding site. An energetically stabilized mode of docking of VUF5455 to the A₃AR/Cl-IB-MECA complex at this region of TM7 was identified (Fig. 6).

Sodium ion binding to GPCRs is considered another form of allosteric modulation (negative) of the binding of agonists. The Asp^2.50 residue in TM2, which is conserved among GPCRs, is not generally important for ligand binding in the human A₃AR; however, it is critical in sodium modulation at this and many other GPCRs. NaCl (100 mM) caused an ~80% inhibition of agonist binding in the WT A₃AR but did not show any effect on D58N^2.50, D107N^3.49, and F182A^5.43 mutant receptors. NaCl decreased the dissociation rate of the antagonist radioligand [³H]PSB-11 at the WT, but not the D58N^2.50 mutant receptor. N30^1.50 is not part of either the putative ligand binding site or the putative sodium binding site at D58^2.50, although it is in a H-bond network with the latter. Mutation of this residue caused a great decrease in affinity of both the agonist Cl-IB-MECA and the antagonist MRS1220, could play a role in stabilizing both ligand binding and sodium ion binding. Also, D58^2.50 and N274^7.45, which did not form part of the ligand binding site, seemed to be important in the H-bonding network involving D58^2.50. Furthermore, we found that Asp^2.50 is not the only residue associated with the modulatory effects of sodium ions. Asp^3.49, Phe^5.43, and Asn^7.45 are also critically involved in the allosteric effects of sodium binding at the human A₃AR. It should be noted that we have not modeled the sodium binding site, so we are unable to distinguish direct or indirect effects of mutations on sodium modulation.

The corresponding Asp^2.50 residue has been intensively studied in other GPCRs (Beukers et al., 1999). It has been demonstrated that this Asp residue is critically involved in sodium modulation in ₂-adrenergic receptors (Horstman et al., 1990), D₂ and D₄ dopamine receptors (Neve et al., 2001; Schetz and Sibley, 2001), the A₁AR (Barbhaiya et al., 1996), and neurotensin receptors (Martin et al., 1999). Here, we further demonstrated that in the A₃AR, both this residue and Asn30^1.50 are critically involved in the allosteric modulation by sodium ions but not by the amiloride analog HMA. Furthermore, Asn30^1.50 and Asp58^2.50 were also demonstrated to be involved in the modulation by DU124183 and VUF5455. Thus, the present results demonstrated that Asn^1.50, Asp^2.50, and Asn^7.45 are critically involved in the sodium modulation in the human A₃AR. Similar results were indicated for sodium effects on the D₂ dopamine receptors (Neve et al., 2001) suggesting that Asn^1.50, Asp^2.50, Ser^3.39, Asn^7.45, and Ser^7.46 might act in concert in sodium binding.

Homology modeling of the dopamine D₂ receptor has suggested the presence of a pyramidal sodium-binding pocket defined by residues Asp80^2.50, Ser121^3.39, Asn124^3.42, and Ser420^7.46 at each vertex of the base and Asn423^7.49 at the apex (Teeter et al., 1994). N52A^1.50, N52L^1.50, and S121L^3.39 mutations profoundly altered the properties in the mutant receptors, which exhibited no detectable binding of radioligands, and S121A^3.19, S121N^3.39, S420A^3.46, and S420N^3.46 showed little or no dependence on sodium. With respect to the highly conserved (D/E)R(Y/W) motif in GPCRs, E134^3.49 in rhodopsin formed a salt-bridge with the guanidinium group of the adjacent R135^3.50 (Palczewski et al., 2000). The calculated A₃AR was stabilized by a number of interhelical H-bonds in a similar fashion: D107^3.49 H-bonded with T46^IL1, T122^IL2, R108^3.50, and R111^IL2 (Fig. 7B). Thus, mutation of residues that are important for the intramolecular TM network may affect the stability of the inactive conformation and consequently pharmacological properties of the A₃AR. In the sodium experiment, the Ala mutant receptors of D58^2.50, D107^3.49, and F182^5.43 eliminated the modulation by sodium ions. It was suggested D58^2.50 and D107^3.49 might act as counter ions in direct sodium binding; however, F182^5.43, which is separated from the putative sodium binding pocket, might play an indirect role in stabilizing an activated conformational state or in cooperativity, interacting with the lipid membrane or in aromatic-aromatic interaction.

In the inactive conformation of the A₃AR, three amino acids, D58^2.50, D107^3.49, F182^5.43 were far from each other. Three distances between C atoms of D58^2.50 to D107^3.49, D58^2.50 to F182^5.43, and D107^3.49 to F182^5.43 were 22.8, 21.3, and 24.4 Å, respectively. Thus, binding of sodium could affect the overall conformation of the A₃AR. Depending on their class of allosteric modulators, DU124183 and VUF5455 binding affected the sodium binding site D58^2.50, whereas HMA affected only F182^5.43 and N274^7.45. Only DU124183 affected another anionic residue, D107^3.49, involved in sodium binding. The existence of multiple binding conformation was thus possible, depending on the allosteric modulator as suggested in ligand-selective receptor conformation.

In summary, the present study demonstrated that a number of residues in TMs are critically involved in ligand binding and/or allosteric modulation. It was demonstrated that not only Asp58^2.50, but also Asn30^1.50, Asp107^3.49, Phe182^5.43, and Asn274^7.45 are critically involved in sodium binding. Phe182^5.43 and Asn274^7.45 are not critically involved in agonist binding, but they are required for the allosteric modulation by the known A₃AR allosteric modulators.