Hydrophilic Side Chains in the Third and Seventh Transmembrane Helical Domains of Human A_2A Adenosine Receptors Are Required for Ligand Recognition

SUMMARY

Hydrophilic residues of the G protein-coupled human A_2A adenosine receptor that are potentially involved in the binding of the ribose moiety of adenosine were targeted for mutagenesis. Residues in a T₈₈QSS₉₁ sequence in the third transmembrane helical domain (TM3) were individually replaced with alanine and other amino acids. Two additional serine residues in TM7 that were previously shown to be involved in ligand binding were mutated to other uncharged, hydrophilic amino acids. The binding affinity of agonists at T88 mutant receptors was greatly diminished, although the receptors were well expressed and bound antagonists similar to the wild-type receptor. Thus, mutations that are specific for diminishing the affinity of ribose-containing ligands (i.e., adenosine agonists) have been identified in both TM3 and TM7. The T88A and T88S mutant receptor fully stimulated adenylyl cyclase, with the dose-response curves to CGS 21680 highly shifted to the right. A Q89A mutant gained affinity for all agonist and antagonist Iigands examined in binding and functional assays. Q89 likely plays an indirect role in ligand binding. S9OA, S91A, and S277C mutant receptors displayed only moderate changes in ligand affinity. A 5281 N mutant gained affinity for all adenosine denyatives (agonists), but antagonist affinity was generally diminished, with the exception of a novel tetrahydnobenzothiophenone derivative.

Adenosine acts as a neuromodulator in the central and peripheral nervous systems and as a homeostatic regulator in a variety of other systems, including the cardiovascular, renal, and immune systems (1). Four pharmacologically distinct adenosine receptor subtypes, A₁, A_2A, A_2B, and A₃, have been cloned (2, 3). Activation of adenosine A_2A receptors, in general, increases the energy supply in various organs. The regulation of blood pressure by centrally (4) and peripherally (5) mediated mechanisms involves A_2A receptors. Activation of A_2A receptors results in vasodilatation, and this effect has been examined as a potential antihypertensive therapy using selective A_2A agonists such as CGS 21680 {2-[4-[(2-carboxyethyl)phenyl]ethyl-amino]-5′-N-ethylcarboxamidoadenosine) (6). A_2A receptors, present in platelets, where they inhibit aggregation, and in the liver, have also been investigated for therapeutic applications. In the brain, A_2A receptors occur primarily in the striatum, where they are colocalized with D₂ dopamine receptors (7). Adenosine acts in a manner opposite to dopamine and thus elicits locomotor depression (8). Diseases in which the dopaminergic system is hyperactive [e.g., schizophrenia (9) and Huntington’s disease (8)] may be mitigated by A_2A receptor activation. Parkinson’s disease, in which the dopaminergic system is hyporesponsive, may be amenable to treatment with selective A_2A receptor antagonists (10, 11).

We characterized A adenosine receptors through the design and use of novel ligand probes, including radioligands and affinity labels (1, 12), and biotinylated probes and fluorescent labels (13, 14). In addition, site-directed mutagenesis (15) and molecular modeling (16) were used for A_2A receptor characterization. A rhodopsin-based model (15) of the human A_2A-receptor has been proposed. This model is highly consistent with mutagenesis results regarding orientation of individual amino acid residues within the central ligand-binding cavity, thus implicating residues in TM5, TM6, and TM7 of the A_2A receptor in ligand recognition (15). In particular, two histidine residues (H250 and H278) and S277 are essential for ligand binding. The adenine moiety likely interacts with a pocket of aromatic amino acids in TM6, and the ribose moiety likely interacts with a hydrophilic region in TM7. Mutagenesis studies of A₁ adenosine receptors have identified some of the corresponding residues as necessary for ligand binding. The corresponding histidine residues (17), a hexapeptide region of TM5 (18), and 1274 and 5277 in TM7 of bovine A₁ receptors (19, 20) are involved in ligand binding. Thus, the position of the receptor-bound adenosine is likely to be similar in these subtypes.

In this study, we introduced the novel finding that residues of TM3 are essential for the binding of ligands to the human A_2A receptor and further explore the role of the previously modified serine residues in TM7 (15). A long-range goal of this investigation is the design of more selective pharmacological agents based on structural differences in receptors.

Experimental Procedures

Materials

Human A_2A adenosme receptor cDNA (pSVLA_2A) was provided by Dr. Marlene A. Jacobson (Merck Research Labs, West Point, PA). Taq polymerase for the PCR was purchased from PerkinElmer Cetus (Norwalk, CT). All enzymes used in this study were ohtamed from New England Biolabs (Beverly, MA). The agonists CGS 21680, NECA, R-PIA, CADO, and DPMA and the antagonists XAC and CGS 15943 {9-chloro-2-(fiiryl)[1,2,4}triazolo[1,5-c]quinazolin-5-amine) were from RBI (Natick, MA). [³H]CGS 21680 (41.2 Ci/mmol) and [3HIXAC (118 Ci/mmol) were obtained from DuPont-New England Nuclear (Boston, MA), and [³H]adenine (15 Ci/mmol) was purchased from American Research Chemicals Inc. (St. Louis, MO). IB-MECA was prepared as described previously (21). BTH₄ was obtained from Maybridge Chemicals (Trevillet, UK). Chemical structures of the agonist ligands used in this study may be found in Kim et al. (15). FBS and o-phenylenediamine dihydrochloride were purchased from Sigma Chemical Co. (St. Louis, MO). The Sequenase Kit, ATP, and cAMP were from United States Biochemical (Cleveland, OH). All oligonucleotides used were synthesized by Bioserve Biotechnologies (Laurel, MD). A monoclonal antibody (12CA5) against an HA epitope was purchased from Boehringer-Mannheim Biochemicals (Indianapolis, IN), and goat anti-mouse IgG (γ-chain specific) antibody conjugated with horseradish peroxidase was purchased from Sigma. DEAE-dextran was obtained from Pharmacia LKB (Piscataway, NJ). Rolipram was a gift from Schering AG (Berlin, Germany). SCH 58261 {5-amino-7(phenylethyl)-2-(2-furyl)-pyrazolo[4,3-e]1,2,4-triazolo[1,5-c]pyrimidine) was a gift from Dr. E. Ongini (Schering, Milan, Italy) and Prof. P. Baraldi (Urnversity of Ferrara, Italy). ZM 241385 {4-(2-[7-amino-2-(2-furyl)[1,2,4}triazolo[2,3a][1,3,5]triazinyl-amino]ethyl)-phenol) was a gift from Dr. S. Poucher (Zeneca, Macclesfield, UK).

Plasmid construction and site-directed mutagenesis

The coding region of pSVLA_2A was subcloned into the pcD cDNA expression vector (22), yielding pcDA_2A. All mutations were introduced into pcDA_2A by using standard PCR mutagenesis techniques (23). The accuracy of all PCR-derived sequences was confirmed by dideoxy sequencing of the mutant plasmids (24).

Epitope tagging

A 9-amino acid sequence derived from the influenza virus HA protein (TAC CCC TAC GAC GTC CCC GAC TAC GCC; peptide sequence: YPYDVPDYA) was inserted after the second methionine residue at the extracellular amino terminus of the A_2A adenosine receptor gene (15). Oligonucleotides containing the HA-tag sequence were designed and used to generate PCR fragments, which were then used to replace the homologous wild-type pcDA_2A sequences.

Transient expression of mutant receptors in COS-7 cells

COS-7 cells (2 × 10⁶) were seeded into 100-mm culture dishes contaming 10 ml of Dulbecco’s modified Eagle’s medium supplemented with 10% FBS. Cells were transfected with plasmid DNA (4 μg of DNA/dish) according to the DEAE-dextran method (25) ~24 hr later and grown for an additional 72 hr at 37°.

Membrane preparation and radioligand binding assay

Cells were scraped into ice-cold lysis buffer (4 ml of 50 mM Tris, pH 6.8, at room temperature, containing 10 m i MgCl₂). Harvested cells were homogenized using a Polytron homogenizer (Brinkmann, Westbury, NY) and then spun at 27,000 × g for 15 mm. Cell membranes (pellet) were resuspended in the same buffer.

For saturation and competition binding experiments, each tube contained 100 μl of membrane suspension (containing 2 units/ml adenosine deaminase (Boehringer-Mannheim Biochemicals), 50 μl of radioligand, and either 50 μl of buffer/competitor (50 mM Tris, pH 6.8, 10 mM MgCl₂) or 50 μl of 80 μm CADO in buffer (to determine nonspecific binding). The mixtures were incubated at 25° for 120 mm, filtered, and washed three times with ~5 ml of ice-cold buffer/wash using a Brandel cell harvester (Gaithersburg, MD). Pharmacological parameters were analyzed using the KaleidaGraph program (version 3.01; Abelbeck/Synergy Software, Reading, PA). Statistical analysis was performed using the alternate t test (InStat version 2.04; GraphPad, San Diego, CA).

cAMP determination

cAMP levels were determined by measurement of the conversion of [³H]ATP to [³H]cAMP. One day after transfection, cells were transferred from 100-mm dishes into six-well dishes (~3 × 10⁵ cells/well) and incubated with culture media contaming 2 μCi/ml [³H]adenine. After 24 hr, the cultures were washed and incubated with 1 ml/well Hanks’ balanced salt solution containing 0.1 mM rolipram for 15 mm at 37°. The cells were incubated with different concentrations ofthe agonist CGS 21680 (in culture media) for 30 mm at 37°. The reaction was terminated by aspiration of the media and the addition of 1 ml of ice-cold 5% trichloroacetic acid containing 1 mM ATP and 1 mM cAMP. After a 30-mm incubation at 4°, cell lysates were eluted through sequential chromatography on Dowex and alumina columns (26). cAMP formation is expressed as percentage of maximal stimulation of conversion of [³H]ATP to [³H]cAMP (27). At agonist concentrations of > 100 μm, a stimulation was observed in nontransfected COS-7 cells (15), and these values were subtracted from values obtained in the transfected mutant receptor cells.

ELISA

For indirect cellular ELISA measurements, cells were transferred to 96-well dishes (4–5 × 10⁴ cells/well) 1 day after transfection. At ~48 hr after splitting, cells were fixed in 4% formaldehyde in phosphate-buffered saline for 30 mm at room temperature. After washing with phosphate-buffered saline three times and blocking with Dulbecco’s modified Eagle’s medium (containing 10% FBS), cells were incubated with HA-specific monoclonal antibody (12CA5; 20 μg/ml) for 3 hr at 37°. Plates were washed and incubated with a 1:2000 dilution of a peroxidase-conjugated goat anti-mouse IgG antibody (Sigma) for 1 hr at 37°. Hydrogen peroxide and o-phenylenediamine (each 2.5 mm in 0.1 m phosphate/citrate buffer, pH 5.0) served as substrate and chromogen, respectively. The enzymatic reaction was stopped after 30 mm at room temperature with 1 m H₂SO₄ solution containing 0.05 m Na₂SO₃, and the color development was measured bichromatically in the BioKinetics reader (EL 312; Bio Tek Instruments, Winooski, VT) at 490 nm and at 630 nm (base-line).

Results

Sequence alignments for selected transmembrane regions of adenosine receptors and other GPCRs are shown in Fig. 1. The residues of the human A_2A receptor selected as targets for site-directed mutagenesis are shown in bold; they include hydrophilic residues potentially involved in the binding of the ribose moiety, which was suggested in our previous studies (15, 16) to occur at TM7 (from mutagenesis results) and at TM3 (predicted by molecular modeling). Mutated residues include 591 and 5281, which are highly conserved among GPCRs. Another mutated residue, T88, is conserved among all known adenosine receptor sequences. Other mutated residues include those that are conserved between A₁ and A₂ receptors (Q89, 590) or between A₂ and A₃ receptors (5277). Each of these amino acid residues was individually replaced with alanine and/or other amino acids (see below). In addition, each mutant contained an epitope-tag sequence included at the amino terminus for immunological detection (see below). The pharmacological properties were compared with those of the wild-type receptor that was similarly modified.

Fig. 1.

Location of mutations carried out in this study, illustrated through an alignment of the TM3 and TM7 of selected receptor subtypes. Bold, residues mutated in A_2A receptors (present study) and in a chimeric bovine A₁/rat A₃ construct (18). Accession numbers are h A2a (human) P29274, rA2a (rat) P30543, cA2a (dog) P11617, gpA2a (guinea pig) U04201 , hA2b (human) P29275, hAl (human) P30542, hA3 (human) P33765, rA3 (rat) P28647, m3 (rat) P08483, β₂ (hamster) P04274, and NK1 (human) P25103. In TM3, the canine and guinea pig A_2A sequences are identical to the rat A_2A sequence.

Ligand binding properties of mutant human A adenosine receptors

Radioligand saturation studies and competition binding experiments using a fixed concentration of either the agonist [³H]CGS 21680 at a concentration of 15 nm (6) or the antagonist [³H]XAC at a concentration of 2 nm (28) were carried out on the wild-type and mutant receptors. The agonists selected for competition (Table 1) included adenosine derivatives modified at the 2 position (CADO), the N6 position (DPMA, ^R-PIA, and ADAC), the 5′ position (NECA), the 5′ and 2 positions (CGS 21680), and the 5′ and N6 positions (IB-MECA). A diverse set of adenosine antagonists (Fig. 2), including XAC, a potent and nonselective nonxanthine (CGS 15943), several recently reported (29, 30) potent and highly selective A_2A antagonists (SCH 58261 and ZM 241385), and a novel structure (BTH₄) that does not contain any nitrogen atoms (31), were studied in competition for [³H]CGS 21680 binding.

TABLE 1. Binding characteristics of wild-type and mutant human A_2A-adenosine receptors using the antagonist radioligand [³H]XAC.

Data are presented as mean ± standard deviation of two or three independent experiments, each performed in duplicate. Each sample contained 7–11 μg of membrane protein/tube. Agonist and antagonist binding affinities [K_i values, structures in Fig. 2 and Jacobson et al., (35)] were determined in ³H]XAC (1.0 nM) competition binding studies using membrane homogenates prepared from transiently transfected COS-7 cells, as described in Experimental Procedures. K_i values were calculated from IC₅₀ values by using the KaleidaGraph program. All constructs contain an HA-tag sequence at the amino terminus (15).

	Construct
	Compound	Wild-type	T88A	T88S	188R
Bmax (pmol/mg)	[3H]XAC	14.9 ± 1.2	6.54 ± 1.61b	6.26 ± 070b	5.54 ± 0.47b
Kd (nM)	[3H]XAC	8.83 ± 1.46	10.8 ± 18d	5.76 ± l.82d	4.21 ± 1.17c
Ki (nM)	Agonists
	CADO	118 ± 22	16,400 ± 3,200c	776 ± 313c	9,140 ± 2,590c
	DPMA	65.5 ± 0.3	5,710 ± 2,520c	655 ± 134c	4,090 ± 760c
	NECA	19.1 ± 1.7	1,590 ± 280c	778 ± 224c	2,660 ± 1,390c
	CGS 21680	26.7 ± 6.5	22,100 ± 2,900b	422 ± 180c	4,620 ± 620b
	Antagonists
	CGS 15943	1.20 ± 0.30	7.98 ± 1.71c	6.28 ± 1.68c	5.00 ± 052b
	ZM 241385	0.758 ± 0.135	3.47 ± 2.41d	1.16 ± 0.20d	2.00 ± 0.71d

^ap <0.001.

^bp <0.01.

^cp <0.05.

^dNot significant.

FIg. 2.

Structures of agonist (A) and antagonist (B) ligands used in this study.

Among TM3 mutations, amino acid substitution of T88 distinguished between agonist and antagonist ligands (Table 1). The specific binding of [³H]CGS 21680 was greatly diminished (i.e., <2% of the specific binding of 15 nm [³H]CGS 21680 observed with the wild-type receptor) in the T88A, T88S, and T88R mutant receptors, whereas binding of [³H]XAC was similar to that ofthe wild-type receptor. The K_d determined for [³H]XAC at the wild-type receptor (8.83 ± 1.46 nM) was not significantly different from the K_d at the T88 mutant receptors (4.2–10.8 nM). The receptor densities, B_max, in membranes of COS-7 cells transfected with the T88 mutant receptors (6–7 pmollmg protein) were approximately half those of the COS-7 cells expressing wild-type receptors (14.9 ± 1.2 μmol/mg of protein). K_i values for competition of binding of [³H]XAC by agonists in T88A and T88R mutant receptors indicated a 60–830-fold lower affinity than in wildtype receptors. Even the T88S mutant receptor, which differs only in the absence of a methyl group, bound agonists with 6.6-fold (for CADO) to 41-fold (for NECA) lower affinity. Affinities of the antagonists XAC and ZM 241385 were unchanged in the T88S mutant receptor, whereas CGS 15943 was 5.2-fold weaker than in wild-type receptors. In a comparison of T88S and T88A mutants, the largest contrast in ligand affinity occurred for the agonist CGS 21680. The removal of the serine hydroxyl group of the T88S mutant, already impaired in its ability to bind agonists, diminished by 52-fold the affinity of this adenosine derivative. Thus, the requirements for agonist binding at position 88 are highly specific, whereas the binding of antagonists was less dramatically affected by mutation at this site.

The T88 mutant receptors were found to be properly expressed on the cell surface using an ELISA (15). To estimate approximate levels of receptor protein present in the plasma membrane, a standard curve was constructed from different batches oftransfected COS-7 cells expressing different levels of HA-tagged A_2A wild-type receptors (see Experimental Procedures and Ref. 15). This ELISA procedure does not interfere with the intactness of the plasma membrane barrier; thus, the assay is specific for receptor molecules having the proper orientation (extracellular amino terminus). Expression levels for the various mutants (HA-tagged wild-type receptor = 100%) determined according to this method were as follows (six experiments): T88A, 77.3 ± 12.5%; T88S, 89.8 ± 43.0%; and T88R, 134 ± 38.3%. The combination of ELISA and radioligand binding results indicates that the T88 residue is important, either directly or indirectly, for the high affinity binding of agonist ligands.

In contrast, replacement of other amino acids, located carboxyl-terminally to T88 in TM3, did not have as detrimental an effect on binding of the agonist radioligand [³H]CGS 21680 (Table 2). The S9OA mutant receptor displayed a slight trend toward higher affinity of both agonists and antagonists relative to the wild-type receptor (K_d of [³H]CGS 21680 was 2-fold lower). In competition binding studies at the S91A mutant receptor, only slight changes were noted in K_i values relative to the wild-type receptor. CA.DO, CGS 21680, and ADAC (N6-substituted) were approximately half as potent in displacing radioligand binding as the wild-type receptor.

TABLE 2. Binding characteristics of wild type and mutant human Aı-adenosine receptors using the agonist radioligand [³H]CGS 21680.

Data are presented as mean ± standard deviation of two or three independent experiments, each performed in duplicate. Each sample contained 7-11 μg of membrane protein/tube. Agonist and antagonist binding affinities [K_i values, structures in Fig. 2 and Jacobson et al. (35)] were determined in [³H]CGS 21 680 (15 nM) competition binding studies using membrane homogenates prepared from transiently transfected COS-7 cells, as described in Experimental Procedures. K_i values were calculated from IC₅₀ values by using the KaleidaGraph program. All constructs contain an HA-tag sequence at the amino terminus (15).

	Construct
	Compound	Wild-type	Q89A	S9OA	S91A	S277C	S281N
Bmax (pmol/mg)	[3H]CGS 21680	11.2 ± 0.3	13.2 ± 1.5d	12.1 ± 1.2d	14.3 ± 2.1d	1.20 ± 0.01a	6.26 ± 0.86c
Kd (nM)	[3H]CGS 21680	31.0 ± 1.0	6.70 ± 0.86a	16.1 ± 35c	59.7 ± 77c	37.2 ± 1.2b	12.3 ± 1.88
Ki (nM)	Agonists
	CADO	100 ± 14	19.1 ± 6.2b	19.4 ± 4.0c	224 ± 7b	384 ± 22a	7.43 ± l.80b
	DPMA	77.3 ± 4.8	16.6 ± 3.0a	11.1 ± 1.7b	88.4 ± 28.6d	355 ± 113d	5.92 ± 2.00b
	NECA	27.6 ± 0.9	1.90 ± 0.22a	5.90 ± 0.40a	35.5 ± 1.6b	129 ± 12b	4.50 ± 0.50a
	R-PIA	299 ± 83	28.6 ± 10.0c	76.1 ± 20.0c	314 ± 13d	1590 ± 570d	17.6 ± 5.4c
	IB-MECA	435 ± 100	3.21 ± 0.07c	78.5 ± 13.6c	614 ± 212d	1755 ± 400c	53.6 ± 15.1c
	ADAC	1330 ± 50	41.2 ± 2.6b	675 ± 83c	2530 ± 560d	3580 ± 168b	156 ± 88a
	Antagonists
	SCH 58261	1.85 ± 0.15	0.246 ± 0.084a	1.50 ± 0.50d	2.30 ± 1.12d	4.70 ± 0.50c	7.85 ± 0.55b
	CGS 15943	1.95 ± 0.45	0.455 ± 0.165b	0.925 ± 0.475d	1.95 ± 0.56d	1.53 ± 11d	5.10 ± 0.90c
	XAC	8.71 ± 1.60	2.01 ± 0.72c	16.6 ± 16b	18.1 ± 2.3c	10.3 ± 4.6d	82.3 ± 2.1a
	BTH4	859 ± 14	133 ± 15a	508 ± 101c	1880 ± 30a	279 ± 17a	245 ± 23a
	BTH4	859 ± 14	133 ± 15a	508 ± 101c	1880 ± 30a	279 ± 17a	245 ± 23a
	ZM 241385	0.525 ± 0.035	0.197 ± 0.041a	0.360 ± 0.080d	1.09 ± 0.11c	1.35 ± 0.35d	1.93 ± 0.23b

^ap <0.001.

^bp <0.01.

^cp <0.05.

^dNot significant.

The mutation of Q89 gave an unanticipated enhancement of affinity for both agonist and antagonist ligands (Tables 2 and 3). From saturation binding, it was determined that the Q89A mutant receptor had a 4.6-fold greater affinity for [³H]CGS 21680 (6.70 ± 0.86 nm) than the wild-type receptor (31.0 ± 1.0 nm). The affinities ofthe competing ligands at the Q89A mutant receptor (Table 2) were dramatically higher than in the wild-type receptor. IB-MECA had the greatest ratio of affinities (136-fold versus the wild-type receptor; Fig. 3A), whereas the potencies of ADAC (Fig. 3B) and NECA were enhanced 32- and 15-fold, respectively. Other N⁶-modified analogues, such as R-PIA and DPMA, and the C2-modified analogue CADO were 5–12-fold more potent in binding at the Q89A mutant receptor. The antagonists generally displayed a 4–5-fold enhancement of affinity at the Q89A mutant receptor versus the wild-type receptor.

TABLE 3. Radioligand binding characteristics of Q89 mutant human A_2A-denosine receptors using an agonist radioligand.

Data are presented as mean ± standard deviation of two or three independent experiments, each performed in duplicate. Each sample contained 7-11 μg of membrane protein/tube. Saturation of binding of [³H]CGS 21680 using membrane homogenates prepared from transiently transfected COS-7 cells, as described in Experimental Procedures. All constructs contain the HA-epitope tag sequence at the amino terminus (15).

Constructa	Kd	B max
	nM	pmol/mg
Wild-type	31.0 ± 1.0	11.2 ± 0.3
Q89A	6.70 ± 0.86b	13.2 ± 1.5e
Q89N	98.7 ± 23.6d	8.71 ± 0.69c
Q89S	93.3 ± 11.5d	14.6 ± 34e
Q89L	35.6 ± 0.7c	8.68 ± 0.32c
Q89H	46.8 ± 0.4c	11.4 ± 0.2e
Q89R	50.8 ± 10.8e	4.68 ± 0.00b

^aSpecific binding versus [³H]XAc (5.5 nM) showed levels comparable to HA-tagged wild-type receptors for Q89A, Q89N, Q89S, Q89L, and Q89H mutant receptors. For the Q89R mutant, <8% of the specific binding found for wild-type receptors was observed for Q89R.

^***p <0.001;

^**p<0.01;

^*p <0.05: n.s. not significant.

^bp <0.001.

^cp <0.01.

^dp <0.05.

^eNot significant.

Fig. 3.

Displacement of binding of the agonist radioligand [³H]CGS 21680 from HA-tagged A_2A wildtype (WT) and Q89A mutant receptors expressed in COS-7 cells. Competitors used were IB-MECA (A) and ADAC (B). Competition binding studies were carried out using membrane homogenates prepared from transfected COS-7 cells, as described in Experimental Procedures. Data are from a representative experiment performed in duplicate.

Other amino acids were substituted at position 89 to probe the contribution to ligand recognition of factors such as size, polarity, aromaticity, or charge (Table 3). Substitution of Q89 by residues, charged or uncharged, larger than alanine (Q89S, Q89N, Q89L, Q89H, and Q89R) did not preclude the high affinity binding of[³H]CGS 21680. Histidine was chosen because it occurs at this position in A₃ receptors; however, no selective enhancement of the affinity of the A₃ receptor-selective agonist IB-MECA was observed. The mutant receptors Q89S and Q89N had a lower affinity (3-fold) for the radioligand than wild-type receptors. The mutant receptor Q89A was unique in having increased affinity for [3HICGS 21680. Thus, for binding of the agonist radioligand, there existed a great deal of tolerance for substitution at position 89.

Competition binding was studied for three ligands (NECA, IB-MECA, and XAC) at this set of mutant receptors at position 89 (Table 3). The affinities showed an approximate dependence on the size of the amino acid side chain. Solvent-accessible surface, a theoretical value reported for each amino acid (36) that was used as a relative steric indicator for ordering the amino acids (Fig. 4), fell within the range of 224 Ų for A to 355 Ų for R, and affinity varied over several orders of magnitude. For each competitor, the K_i value tended to increase as a function of the size of residue 89. The effects on affinity relative to wild-type receptors ranged from a large enhancement (for the two agonists NECA and IB-MECA at alanine, serine, asparagine, and leucine mutants) to substantial reduction (for the antagonist XAC at H and R mutant receptors).

Fig. 4.

Plot of affinity (K_i in nA_2A) in competition binding experiments of three adenosine receptor ligands at Q89 mutant receptors as a function of the amino acid residue. The amino acids were arranged in order of increasing solvent-accessible surface area, using theoretical values for each isolated amino acid as calculated by Hubbard et al. (36). K_i values for the agonist (NECA and lB-MECA) and antagonist (XAC) ligands (structures in Fig. 2) were determined in [³H]CGS 21680 (15 nM) competition binding studies using membrane homogenates prepared from transiently transfected COS-7 cells, as described in Experimental Procedures. K_i values were calculated from IC₅₀ values by using the KaleidaGraph program. All constructs contain the HA epitope tag sequence at the amino terminus (15). ***, p < 0.001; **, p < 0.01; *, p < 0.05; n.s., not significant.

The TM7 mutant receptors S277C and S281N, both small polar substitutions capable of hydrogen bonding, were also constructed (Table 2). In a previous study (15), these serine residues were mutated to alanine and found to be required for ligand recognition (5277 for agonists and 5281 for all ligands). Changes in the affinities of agonists at the S277C mutant receptor were relatively modest (i.e., ≤5-fold lower than at the wild-type receptor). Some antagonists (XAC and CGS 15943) were approximately equipotent at S277C mutant receptors and at wild-type receptors. The nonxanthine A_2A-receptor antagonists SCH 58261 and ZM 241385 were moderately diminished in affinity, whereas the novel tetrahydrobenzothiophenone derivative BTH₄ displayed 3-fold enhanced affinity at S277C mutant receptors. At S281N mutant receptors, the affinities of agonists, including those substituted at N⁶, C2, or C5′ positions, were 6–17-fold higher than at wild-type receptors. The agonist order (decreasing) for enhancement of affinity produced by the S281N mutation was R-PIA (17.0-fold) > CADO (13.5-fold) = DPMA (13.1-fold) > ADAC (8.5-fold) = IB-MECA (8.1-fold) > NECA (6.1-fold) > CGS 21680 (2.5-fold). Antagonist affinity at the S281N mutant receptor was generally diminished (3–10-fold) compared with wild-type receptors, except for BTH₄ (3.5-fold enhancement).

Functional assay

To determine whether T88A mutant receptors that lacked high affinity radioligand binding were still functional at high agonist concentrations, their ability to mediate increases in intracellular cAMP levels in transfected COS-7 cells was studied. Rolipram was used as an inhibitor of phosphodiesterases. The T88A and T88S mutant receptors showed a dose-dependent stimulation of cAMP production after treatment with CGS 21680, with EC₅₀ values of 14.4 ± 0.423 and 0.323 ± 0.048 μm, respectively. Thus, functional potency of CGS 21680 was far less at the T88A and T88S mutant receptors than at wild-type receptors (EC₅₀ = 0.915 ± 0.213 nM) by factors of 15,700 and 353, respectively. A maximal response of these mutant receptors was reached at ~10⁻⁴ M. The T88R mutant receptor however, showed minimal stimulation of adenylyl cyclase even at 1 mm CGS 21680 (Fig. 5).

Fig. 5.

Stimulation of adenylyl cyclase in COS-7 cells transiently expressing HA-tagged A_2A wild-type (WT) or mutant A_2A-adenosine receptors in the presence of 2 units/mI adenosine deaminase and 0.1 m rolipram. The following receptors were studied: wild-type, T88A, T88R, and T88S mutant receptors. Transfected COS-7 cells were incubated for 30 mm at 37° (for details, see Experimental Procedures) with increasing concentrations of CGS 21680. Data are presented as percentage of maximal increase in cAMP above basal levels in the absence of CGS 21680 for a representative experiment. For each curve, the maximal stimulation represents a 4–5-fold stimulation over basal levels. At agonist concentrations of ≥100 μm, the stimulation observed in nontransfected COS-7 cells (15) was subtracted. EC₅₀ values (average of three independent experiments, each carried out in duplicate) were wild-type receptor, 0.915 ± 0.213 nm; T88A, 14.4 ± 0.423 μm; and T88S, 0.323 ± 0.048 μm.

The 2- and 5′-disubstituted adenosine agonist CGS 21680 acting at the Q89A mutant receptor elicited a dose-dependent stimulation of cAMP production (Fig. 6, top), with an EC₅₀ value of 0.223 ± 0.054 nm [i.e., 4-fold more potent (p < 0.05) than at wild-type receptors (EC₅₀ = 0.915 ± 0.213 nm)]. A similar gain of potency was observed for the N⁶-substituted compound DPMA (EC₅₀ = 10.8 ± 1.4 nm at the wild-type and 2.73 ± 0.59 nm at the Q89A mutant receptor; a 4-fold increase, p < 0.05; Fig. 6, bottom). This effect was even more pronounced for the 5′-substituted compound NECA (EC₅₀ = 44.5 ± 6.2 nm at the wild-type and 2.91 ± 0.54 n t at the Q89A mutant receptor), whereas the Q89A mutation induced a 15.3-fold increase in potency (p < 0.01, Fig. 6, middle). Thus, there was a parallel between the enhanced binding affinities of agonists at the Q89A mutant receptor (Table 2) and their functional potencies. No increase in basal adenylyl cyclase activity was observed for this mutant receptor.

Fig. 6.

Stimulation of adenylyl cyclase in COS-7 cells transiently expressing HA-tagged A_2A wild-type (WT) or mutant A_2A-adenosine receptors in the presence of 2 units/mI adenosine deaminase and 0.1 m rolipram. The following receptors were studied: wild-type and Q89A mutant receptors with CGS 21680 (top), NECA (middle), or DPMA (bottom). Transfected COS-7 cells were incubated for 30 mm at 37° (for details, see Experimental Procedures) with increasing concentrations of agonist. Data are presented as percentage of maximal increase in cAMP above basal levels in the absence of agonist for a representative experiment. For each curve, the maximal stimulation represents a 4-5-fold stimulation over basal levels. EC₅₀ values were calculated averaged over three independent experiments, each carried out in duplicate.

Discussion

Ligand binding and stimulation of adenylyl cyclase in TM3 mutant human A adenosine receptors

The current study demonstrates clearly that hydrophilic residues of TM3 of the human A_2A receptor are involved in ligand binding. Several TM3 mutants prepared in this study are highly unnatural in their ligand binding properties, having either enhanced (Q89A) or greatly diminished (T88A and T88R) affinity for various ligands. Alanine scanning mutagenesis showed that T88 is essential for high affinity agonist binding, whereas only moderate changes in ligand binding affinity occur on replacement of 590 and S91. T88A and T88R mutations had similar detrimental effects on agonist binding, whereas high affinity antagonist binding was still observed in these mutant receptors. Affinity in the T88S receptor mutant was decreased as well but to a lesser extent. Thus, a section of TM3 seems to be involved in ligand (especially agonist) recognition. This is consistent with a molecular model based on a rhodopsin template (15), which predicted that T88 and Q89 are in proximity to the ribose moiety of adenosine.

At high agonist concentrations, the T88A and T88S mutant receptors were active functionally in the stimulation of adenylyl cyclase (Fig. 5), with the dose-response curves right-shifted by 15,700- and 350-fold, respectively (p < 0.01 ). This change in potency reflects the relative agonist affinity shifts in the binding assays. The T88R mutant receptor seemed to be functionally impaired (Fig. 5), even at agonist concentrations that clearly displaced bound [³H]XAC (Table 1).

For single amino acid replacements of Q89, the ligand affinity varied from dramatic increases (e.g., IB-MECA at Q89A) to decreases for the larger amino acid substitutions. The trend of inverse dependence of ligand affinity on steric bulk ofthe side chain at position 89 (Fig. 4) was shown for a number of structurally divergent amino acids substituted at this position through mutagenesis. Because electronic factors of the amino acids seemed to be not as important as steric factors and because three selected competing ligands were similarly affected, it is hypothesized that residue Q89 affects the size of the crevice that constitutes the binding site. Diminishment ofthe size ofa sterically limiting side chain (i.e., in the Q89A mutant receptor) would increase the accessibility of the binding site, thus lowering the K_i values. This mechanism either might involve the side chain acting as an energy barrier to a ligand occupying the binding site or may be more indirect (e.g., by changing interhelical distances). Several other residues of the human A_2A receptor (i.e., Y271) have been postulated to affect ligand binding through interhelical contacts (15), although alanine substitution decreased agonist affinity in that case.

The finding of enhanced affinity for all ligands, both agonists and antagonists, on single amino acid replacement, as in the Q89A mutant receptor, is a rather rare finding. An increase in agonist affinity, but not antagonist affinity, is often observed in constitutively active mutants (32, 33). The Q89A mutation, however, did not affect basal adenylyl cyclase levels. Even though affinity and potency of agonists were both increased in the Q89A mutant receptor versus the wild-type receptor, this mutant receptor was not constitutively active.

In the S90A mutant A_2A receptor, both agonist and antagonist affinities were only slightly affected. Mutation ofS9l to alanine had virtually no effect on ligand affinity.

Structurally related differences in ligand binding in TM7 mutant human A_2A adenosine receptors

In our previous study (15), we reported that residue S277 is importa.nt for agonist recognition. We now present data regarding a new mutant, S277C, in which ligand binding affinities are only slightly affected (agonist affinity was unaffected for CGS 21680 and decreased by 3–5-fold for other agonists). This mutant to a certain extent rescues ligand binding compared with the earlier S277A mutant, in which agonist affinity decreased 43–1070-fold relative to the wild-type receptor.

This study has revealed differences in affinity shifts between agonists and antagonists [e.g., S281N (agonists become more potent and antagonists less potent) and mutations of T88 (agonists selectively become much less potent)]. Therefore, a partially different set of amino acid residues in the receptor is involved in agonist versus antagonist binding. The results of chemical modification and conformational analysis (34, 35) as well as mutagenesis studies of adenosine receptors (15, 17) have suggested that the ribose moiety is coordinated to the histidine residue of TM7, common to all adenosine receptors. Furthermore, changes in affinity of the ribose-modified agonist NECA at A₁ receptors are associated with mutation of an adjacent threonine residue (19, 20). Mutations that are specific for diminishing the affinity of ribose-containing ligands (i.e., adenosine agonists) have been identified in both TM3 and TM7, consistent with the previously reported molecular models of human A_2A receptors that predicted that the ribose moiety of adenosine may bridge these domains (15, 16).

It was proposed (15) that 5281 probably is not directly involved in ligand binding, yet in the current study the S281N mutant receptor distinguishes between agonists and antagonists. All agonists examined displayed an increased affinity (2.5-fold for CGS 21680 to 17-fold for R-PIA) toward the S281N mutant receptor, whereas affinity for most antagonists was moderately decreased (3–9-fold). The A₁-selective antagonist XAC showed the largest decrease in affinity. The A₂-selective antagonists SCH 58261 and ZM 241385 exhibited an intermediate decrease, and the nonselective antagonist CGS 15943 had the smallest decrease. Interestingly, the novel tetrahydrobenzothiophenone derivative BTH₄ showed a moderate increase in affinity for the S281N mutant receptor. Unlike other antagonists, BTH₄ displayed enhanced affinity at both TM7 mutant receptors examined in this study (Table 2). Thus, the binding mode of this antagonist seems to be unique, as is its chemical structure. It is among the least purine-like of known adenosine antagonists, which are almost exclusively nitrogen-containing heterocycles.

Although there are several models for mapping the xanthine binding site onto the partially overlapping adenosine binding site (34, 37), no such exercises have been performed for the nonxanthine antagonists SCH 58261, ZM 241385, and CGS 15943. The current data suggest that the mode of nonxanthine antagonist binding to the receptor may differ significantly from the modes proposed for xanthine-based antagonists.

Structural homology of human A_2A adenosine receptors to other GPCRs

In the biogenic amine GPCRs, there is an essential, conserved aspartate residue in TM3 that coordinates the positively charged secondary nitrogen group of the endogenous ligand. This has been demonstrated for adrenoceptors (38, 39) and receptors for dopamine (40, 41), histamine (42), serotonin (43, 44), and acetylcholine (45, 46). Adenosine, being uncharged at physiological pH, would not benefit from a strong electrostatic interaction with the receptor at this position. In all adenosine receptors, the residue homologous to the above-mentioned aspartate corresponds to a valine (V84 in the human A_2A adenosine receptor). The current study has identified important residues for ligand recognition at positions in the human A_2A adenosine receptor deeper in the membrane than V84.

The essential T88 of A_2A receptors would be expected to be facing approximately the same direction as that of the essential aspartate residue involved in recognition of the secondary nitrogen of biogenic amines at their receptors, being four residues (i.e., one helical turn) closer to the cytoplasmic side. Residue C118 of the human D₂ receptor, equivalent to T88, reacts with a thiol reagent, indicating that this residue is solvent exposed and thus faces the central binding cavity (41).

The S505R mutant of the human thyrotropin receptor (position equivalent to T88) was constitutively active (47), whereas none of the T88 mutants in the current study were constitutively active. Residue T88 of the A_2A receptor aligns with Vi 16 of the NK1 receptor (Fig. 1). Mutation of this residue to leucine reversed the selectivity of antagonists (48). There is no direct evidence that Q89 is pointing directly into the binding cavity. By analogy with rhodopsin, residue E122, the equivalent to Q89, is facing the binding cavity. This residue was found to coordinate and neutralize the Schiff base of retinal (formed with K296), based on a blue shift in the absorption spectrum of the E122Q mutant (49). Residue A120 of the human D₂ receptor (41), the position equivalent to 590, was found not to be exposed to the ligand binding cavity because the A12OC mutant receptor did not react with thiol reagents.

Almost one full helical turn down from T88 is 591. This residue is conserved as a serine in 98 of 156 GPCR sequences probed (e.g., m3 muscarinic receptor, hamster β₂-adrenoceptor: Fig. 1) and therefore is not likely to confer ligand binding specificity. Mutations at this site in the hamster β₂-adrenoceptor led to decreased receptor expression and improper post-transcriptional processing (S12OA; Ref. 50), although the 591A mutant A_2A adenosine receptor had a Bmax value for [³H]CGS 21680 binding similar to wild-type receptors. This site was shown to be facing the binding cavity in the human D₂ receptor (S121C; Ref. 41).

Also in TM7, homology to amino acids known to be involved in ligand binding was found. In the rat ml muscarinic receptor, the mutant receptor C407S, corresponding to residue 5277 ofthe A_2A receptor, had decreased agonist affinity (51). Residue S281 corresponds to the essential S319 in the hamster β₂-adrenergic receptor (50). In the human luteinizing hormone receptor, the naturally occurring mutation at the same site, S616Y, results in hypogonadism and decreased agonist affinity (52).

Conclusions

Results of the current study strongly imply that residues of TM3 are involved in ligand recognition and extend the findings of Kim et al. (15) concerning TM7. Thus, it seems that hydrophilic residues in both TM3 and TM7 are important for recognition of the ribose moiety. The binding affinity and functional activity of agonists at T88 mutant receptors were greatly diminished. A Q89A mutant gained affinity for all agonist and antagonist ligands examined. Q89 likely plays an indirect role in ligand binding. Furthermore, the fact that all of the residues examined in this study are conserved among most adenosine receptors (species and subtypes) suggests that the proposed mode of binding of agonists is common or very similar in these receptors. Divergent effects of mutagenesis within TM3 and TM7 on the binding of adenosine derivatives, xanthines, and nonxanthine antagonists suggest nonidentical mechanisms of binding.

ABBREVIATIONS

TM

(helical) transmembrane domain

ADAC

N⁶-[4-[[[[4-[[[(2-aminoethyl)amino]carbonyl]methyl]anilino]carbonyl]methyl]-phenyl]adenosine

BTH₄

ethyl 3-benzylthio-4,5,6,7-tetrahydrobenzo[c]thiophen-4-one-1-carboxylate

CADO

2-chloroadenosine

DPMA

N⁶-[2-(3,5-dimethoxyphenyl)-2-(2-methylphenyl)ethyl]adenosine

ELISA

enzyme-linked immunosorbent assay

FBS

fetal bovine serum

GPCR

G protein-coupled receptor

HA

hemagglutinin

IB-MECA

N⁶-(3-iodobenzyl)adenosine-5′-N-methyluronamide

NECA

5′-N-ethylcarboxamidoadenosine

PCR

polymerase chain reaction

R-PIA

(R)-N⁶-phenylisopropyladenosine

XAC

(8-[4-[[[[(2-aminoethyl)-amino]carbonyl]methyl]oxy]phenyl]-1,3-dipropylxanthine)