A Gα_s Carboxyl-Terminal Peptide Prevents G_s Activation by the A_2A Adenosine Receptor

Experimental Procedures

Materials.

[³H]CGS21680 (39.5 Ci/mmol) was purchased from NEN Life Science Products (Boston, MA), and [α-³²P]ATP (30 Ci/mmol) and [³H]cAMP (25 Ci/mmol) were from Amersham Pharmacia Biotech (Little Chalfont, Buckinghamshire, UK). [³H]SCH58261 (77 Ci/mmol) and SCH58261 were generous gifts of Dr. E. Ongini (Schering-Plough Research Institute, Milan, Italy). Myokinase, creatine kinase, leupeptin, GTP, and GTPγS were obtained from Roche Molecular Biochemicals (Mannheim, Germany). CGS21680 was from Research Biochemicals International (Natick, MA). Bacitracin, benzamidine, and phenylmethylsulfonyl fluoride (PMSF) were products of Fluka Chemie AG (Buchs, Switzerland). Adenosine deaminase, papaverine, and 5′-N-ethylcarboxamidoadenosine (NECA) were from Sigma Chemical Co. (St. Louis, MO). All other reagents were from standard commercial sources and of the highest grade available.

Membrane Preparation.

Male Sprague-Dawley rats (150–200 g) were sacrificed by cervical dislocation, and striatal tissue was isolated from the brain by dissection. For radioligand binding assays, membranes were prepared essentially as described previously (Mazzoni et al., 1993). Membrane protein concentration was determined by the method of Lowry et al. (1951) using BSA as a standard. For the adenylyl cyclase assay, striatal tissue was suspended in 10 volumes of ice-cold buffer containing 10 mM HEPES/NaOH (pH 7.5), 10 mM MgCl₂, 1 mM EGTA, 1 mM dithiothreitol (DTT) (buffer A) and homogenized with 12 strokes of a tight Teflon-glass homogenizer at 4°C. The homogenate was diluted 6-fold with buffer A and centrifuged at 46,000g for 20 min at 4°C. The pellet was resuspended in 10 volumes of ice-cold buffer A containing 1 mM benzamidine, 200 μg/ml bacitracin, and 2 U/ml adenosine deaminase, and incubated for 30 min at 30°C. The membrane suspension was centrifuged at 46,000g for 20 min at 4°C. The pellet was resuspended in 10 volumes of ice-cold buffer A containing protease inhibitors (as above) and centrifuged at 46,000g for 20 min at 4°C. This washing step was repeated. The final pellet was resuspended in 50 mM HEPES/NaOH, pH 7.5 (buffer B) and used immediately for the adenylyl cyclase assay. Protein concentration was determined using the Coomassie Blue binding method (Bradford, 1976) using BSA as a standard.

Peptide Synthesis.

Peptides were synthesized by the continuous-flow solid phase method using Fmoc chemistry on an automatic synthesizer (Milligen 9050; Millipore, Bedford, MA). Crude peptides were purified by reversed HPLC on a preparative Vydac C₁₈ column (2.2 × 25 cm) (Beckman System Gold, San Ramon, CA) using a 15 to 30% gradient of acetonitrile in 0.1% trifluoroacetic acid/distilled water (v/v). After lyophilization, purity was checked by analytical HPLC and electrospray mass spectrometry using a mass spectrometer (VG Quattro; Micromass, Altricham, UK) equipped with a standard Electrospray ion source. Molecular weight calculations were performed by deconvolution using MassLynx software, version 2.00 (Micromass). Table1 shows the sequence of all peptides used in this study.

The Gα_s and Gα_i1/2peptides were dissolved in buffer C (50 mM Tris-HCl, pH 7.5, 10 mM MgCl₂, 1 mM EDTA) for radioligand binding assays or in buffer B for the adenylyl cyclase assay. In the assay mixtures, the pH ranged between 7.3 and 7.5. Peptide stock solutions were 2.5 mM. All peptide stock solutions were centrifuged at 11,000g for 4 min at room temperature, and supernatants were collected. The concentrations of the Gα_s peptides in solutions were determined spectrophotometrically using the molar extinction coefficient for tyrosine at 280 nm (ε_280nm = 1400 M⁻¹ cm⁻¹) as described by Rarick et al. (1994).

As we might face solubility problems with some peptides or degradation by tissue proteases, we performed either HPLC titration studies and spectrophotometric measurements to verify whether the final concentration of the peptide in the test solution was affected by precipitation, aggregation, nonspecific adsorption to striatal membranes, or proteolysis. Accordingly, Gα_s(374–394)C³⁷⁹A (300 μM) was incubated in 0.5 ml of buffer C in the presence and absence of striatal membranes (∼130 μg of proteins) for 10, 30, 60, and 120 min at 25 and 30°C. Samples were centrifuged at 11,000gfor 4 min at room temperature, and supernatants were collected. As a control, the peptide was also incubated in buffer C in the absence of membranes without any final centrifugation. The concentrations of Gα_s(374–394)C³⁷⁹A in supernatant and control samples were compared by analytical HPLC or spectrophotometric measurement at 280 nm using the conditions described above. The peak areas or absorbances of supernatant and control samples, each assayed in triplicate, were found to vary within the range of ±5–10%, i.e., within the accuracy range of the methods.

Radioligand Binding Assays.

Routine [³H]CGS21680 binding assays were performed as described previously (Mazzoni et al., 1993). Striatal membranes (∼130 μg of proteins) were incubated with 3 nM [³H]CGS21680 in buffer C containing protease inhibitors (1 mM benzamidine, 100 μM PMSF, and 10 μM leupeptin) and adenosine deaminase (1 U/ml) in the presence and absence of Gα_i(344–354) or Gα_speptides (∼300 μM) for 2 h at 25°C. Equivalent experiments were performed in the presence of 100 μM GTPγS. Binding was terminated by vacuum filtration over Whatman GF/C glass fiber filters (Whatman Int. Ltd, Springfield Mill, Maidstone, UK), with five washes (3 ml each) of ice-cold buffer C. Nonspecific binding was determined in the presence of 100 μM NECA. In the presence and absence of GTPγS, specific binding represented 70% and 90% of total binding, respectively.

The dose-dependent effect of peptide Gα_s(374–394)C³⁷⁹A was evaluated by incubating membranes (∼110 μg of proteins) in buffer C with 3.6 nM [³H]CGS21680 in the presence of 100 μM GTPγS and seven different concentrations of the peptide ranging from 0 to 480 μM.

For saturation studies, membranes (∼75 μg of proteins) were incubated in buffer C containing protease inhibitors (as above) with eight different concentrations of [³H]CGS21680 (0.5–108 nM) in the presence or absence of Gα_s(374–394)C³⁷⁹A (350 μM). When experiments were carried out in the presence of GTPγS, the concentrations of [³H]CGS21680 ranged between 0.5 and 250 nM.

[³H]SCH58261 binding assays were performed essentially as described by Zocchi et al. (1996) with some modifications. Striatal membranes (∼70 μg of proteins) were incubated with 0.2 nM [³H]SCH58261 in buffer C containing protease inhibitors and adenosine deaminase (as above) for 1 h at 25°C. The binding reaction was terminated by vacuum filtration over Whatman GF/C glass fiber filters (Whatman Int. Ltd.) with three washes (5 ml each) of ice-cold buffer C. Nonspecific binding was determined in the presence of 100 μM NECA. Specific binding represented 90% of total binding.

To determine equilibrium binding parameters, membranes were incubated with seven different concentrations of [³H]SCH58261 ranging from 0.03 to 5 nM. Saturation curves were fitted using the nonlinear regression equation of the GraphPad Prism program (GraphPad Software, San Diego, CA). The average values (n = 2) of dissociation constant (K_D) and maximal number of binding sites (B_max) were 0.6 nM and 750 fmol/mg of proteins, respectively.

Displacement experiments were performed by incubating membranes and [³H]SCH58261 (0.2 nM) in the presence and absence of different concentrations of either SCH58261 (0.025–5 nM) or NECA (0.5–1000 nM). Control assays were carried out with and without GTPγS (100 μM), whereas the effect of Gα_s(374–394)C³⁷⁹A (300 μM) was evaluated in the presence and absence of the guanine nucleotide.

Adenylyl Cyclase Assay.

Adenylyl cyclase activity was assayed by monitoring the conversion of [α-³²P]ATP to [α-³²P]cAMP, using a previously reported method (Johnson et al., 1994). The method involved addition of [α-³²P]ATP to membranes in the presence of an A_2A adenosine receptor agonist (NECA or CGS21680) and GTP or forskolin (FSK) to stimulate adenylyl cyclase and papaverine as a phosphodiesterase inhibitor. Briefly, to study G_s-mediated adenylyl cyclase activation, the enzymatic activity was routinely assayed in a 100-μl reaction mixture containing 50 mM HEPES/NaOH buffer, pH 7.5, 5 mM MgCl₂, 1 mM DTT, 1 mM EGTA, 0.1 mg/ml creatine phosphokinase, 0.1 mg/ml bacitracin, 0.5 mg/ml creatine phosphate, 0.1 mM ATP, 0.05 mM cAMP, 2 U/ml adenosine deaminase, 0.9 μCi of [α-³²P]ATP, and 0.1 mM papaverine. The concentration of GTP and A_2A adenosine receptor agonists were 100 and 10 μM, respectively. Gα peptides were used at a concentration of 300 μM. The incubation was started by the addition of membranes (∼25 μg of proteins) and was carried out at 30°C. To study FSK-mediated activation of adenylyl cyclase, membranes (∼10 μg of proteins) were incubated at 23°C in 100 μl of medium containing the same buffer and reagents as above except that EGTA was omitted and MgCl₂ was 1 mM. The reactions were stopped after 10 min by placing assay tubes into an ice bath and adding 0.6 ml of a cold stop solution containing 120 mM Zn(C₂H₃O₂)₂/[³H]cAMP (∼12,000 cpm/sample) and then 0.5 ml of 144 mM Na₂CO₃. The total radiolabeled cAMP was isolated on columns of Dowex 50 ion-exchange resin and alumina as described previously (Johnson et al., 1994).

Data Analysis.

A nonlinear multipurpose curve-fitting computer program (EBDA/LIGAND; Elsevier-Biosoft, Cambridge, UK) was used for analysis of saturation data. A partial F test was utilized to determine whether the binding data were best fitted by a one- or two-site model. Saturation curves were also fitted by a nonlinear regression analysis of the GraphPad Prism version 3.0 program (GraphPad Software). Data from concentration-response or displacement curves were analyzed by a least-squares curve-fitting computer program (GraphPad Prism, version 3.0), and the EC₅₀ values were derived. The K_i values for competition binding assays were calculated from the EC₅₀values by the Cheng and Prusoff equation (Cheng and Prusoff, 1973). Values represent the mean ± S.E. of at least three experiments except when otherwise stated. The statistical significance of the differences between means was determined by Student's ttest using GraphPad Prism 3.0.

NMR and Structure Calculations.

The sample for NMR spectroscopy was prepared by dissolving the appropriate amount of Gα_s(374–394)C³⁷⁹A in 0.5 μl of ¹H₂O phosphate buffer (pH 6.6) to obtain a 1 mM solution. The sample was lyophilized and redissolved in aqueous solution with 50% (v/v) hexafluoroacetone trihydrate (HFA).

NMR spectra were recorded on a spectrometer (DRX-600; Bruker Instruments Inc., Billerica, MA). One-dimensional NMR spectra were recorded in the Fourier mode with quadrature detection, and the water signal was suppressed by a low-power-selective irradiation in the homogated mode. Double quantum filter correlation spectroscopy (DQF-COSY) (Piantini et al., 1982), total correlation spectroscopy (TOCSY) (Bax and Davis, 1985), and nuclear Overhauser spectroscopy (NOESY) (Jenner et al., 1979) experiments were run in the phase-sensitive mode using quadrature detection in ω₁ by time-proportional phase increase of initial pulse (Marion and Wüthrich, 1983). Data block sizes were 2048 addresses in t₂ and 512 equidistantt₁ values. Before Fourier transformation, the time domain data matrices were multiplied by shifted sin² functions in both dimensions. A mixing time of 70 ms was used for the TOCSY experiments. NOESY experiments were run at 275 K with mixing times in the range of 100 to 300 ms. The identification of spin systems in the DQF-COSY and TOCSY spectra, assignments of intra- and inter-residual NOESY cross peaks, and their integration were obtained using the interactive program package XEASY (Bartels et al., 1995). The necessary pseudoatom corrections were applied for equivalent and nonstereospecifically assigned protons. Conversion of NOESY cross peak volumes into upper distance bounds was obtained using the CALIBA program incorporated into the program package DYANA (Guntert et al., 1997). Models of the Gα_speptide were generated with the structure calculation algorithm of DYANA (Guntert et al., 1997).

The conformer with the lowest value of the target function was subjected to a restrained energy minimization (EM) using the Amber-all force field (Weiner et al., 1984) with neglect of electrostatics available within the SYBYL software package (version 6.2; TRIPOS Inc., St. Louis, MO). A root mean square (RMS) derivative of 0.5 kcal mol⁻¹ Å⁻¹ was adopted as a convergence criterion. The resulting conformation was fully relaxed through unrestrained EM (RMS deviation about nonhydrogen atoms ≅ 0.1 Å).

Results

Synthetic peptides of various length from the Gα_s carboxyl terminus were used as probes of contact regions between the A_2A adenosine receptor and G_s. For other receptor-G protein systems, peptides corresponding to the last 11 residues of the Gα carboxyl terminus can serve as inhibitors of signal transduction (Hamm et al., 1988; Rasenick et al., 1994; Gilchrist et al., 1998) and stabilize the high-affinity state of the receptor (Hamm et al., 1988). However, the ability of these peptides to stabilize receptor affinity for agonists is not observed for all receptor G protein systems (Gilchrist et al., 1998). We examined the importance of Gα_s peptide size in relation to its competence to affect A_2A receptor-G_sprotein interaction. Table 1 shows the amino acid sequences of the synthetic peptides used in this study. In Gα_s(378–394), Gα_s(376–394), and Gα_s(374–394), Cys³⁷⁹ was substituted with Ala to prevent peptide dimerization.

Effects of Gα_s Carboxyl-Terminal Peptides on Agonist Binding to A_2A Adenosine Receptors.

Gα_s carboxyl-terminal peptides stimulated specific binding of [³H]CGS21680 to A_2A adenosine receptors in rat striatal membranes (Fig. 1), whereas nonspecific binding was not altered. Both in the presence (Fig. 1b) and absence (Fig. 1a) of GTPγS, all peptides tested caused an evident increase of specific binding as compared with respective controls. However, in the presence of the GTP analog, this effect was generally more evident (Fig. 1b). The most effective peptides were Gα_s(378–394)C³⁷⁹A, Gα_s(376–394)C³⁷⁹A, and Gα_s(374–394)C³⁷⁹A, whereas the shortest peptide, Gα_s(384–394), was less active. In accordance to the crystal structure of Gα_s (Sunahara et al., 1997), the bioactive conformation of its carboxyl-terminal portion is an α-helix (α5) spanning from Asp³⁶⁸ to Leu³⁹⁴. Thus, 17-, 19-, and 21-residue peptides may have a stronger propensity to assume an α-helical conformation than do the shortest Gα_s peptides.

We also investigated whether a Gα_i1/2carboxyl-terminal peptide, Gα_i1/2(344–354), modulated agonist binding to A_2A adenosine receptors in rat striatal membranes. The Gα_i1/2peptide had no significant effect in the absence of the guanine nucleotide (Fig. 1a), but it was as effective as Gα_s(384–394), Gα_s(382–394), and Gα_s(380–394) in the presence of GTPγS (Fig.1b).

The dose dependence of agonist binding stimulation was examined for the most active peptide, Gα_s(374–394)C³⁷⁹A. The effects of various concentrations of the peptide on [³H]CGS21680 binding to rat striatal membranes were measured in the presence of 100 μM GTPγS. The dose-response curve is shown in Fig. 2. The derived EC₅₀ value for stimulation of specific binding was 14.9 ± 1.6 μM (n = 3), whereas the maximal efficacy was 250 ± 5% of control at a concentration of 480 μM.

Effects of Gα_s(374–394)C³⁷⁹A on [³H]CGS21680 Binding Parameters to A_2AAdenosine Receptors.

To evaluate whether peptide Gα_s(374–394)C³⁷⁹A was able to stabilize the high-affinity state of the A_2A adenosine receptor and thus to mimic G_s, saturation binding studies were carried out in the presence and absence of a fixed concentration of the peptide. Analysis of data using nonlinear, curve-fitting programs revealed that both saturation curves were better represented by one-site models (Fig.3a). In addition, Scatchard transformation of these data produced linear plots (Fig. 3b). This suggested the existence of a single class of binding sites in both conditions at the concentrations of ligand used. In the absence of the Gα_s peptide, the K_Dand B_max values were 10.3 ± 0.3 nM and 611 ± 77 fmol/mg of protein (n = 3), whereas in the presence of peptide Gα_s(374–394)C³⁷⁹A theK_D and B_maxvalues were 9.1 ± 0.02 nM and 897 ± 4 fmol/mg of protein (n = 3), respectively. The addition of the Gα_s peptide caused an increase of the maximal number of binding sites (P < .05) without significantly affecting the receptor affinity state.

In the presence of 100 μM GTPγS, Gα_s(374–394)C³⁷⁹A was responsible for a 2- to 3-fold increase of agonist binding affinity as compared with its control (K_D, 79.2 nM;n = 2), whereas the maximal number of binding sites showed a decrease (data not shown). However, the Gα_s peptide was not able to shift back the receptor in a high-affinity state for agonist ligands.

Influence of Gα_s(374–394)C³⁷⁹A on Displacement of [³H]SCH58261 Binding by NECA.

To detect receptor affinity states and test the hypothesis that the 21-residue Gα_s peptide stabilized an intermediate affinity state of rat A_2A adenosine receptors, we examined the effect of the peptide on the ability of an agonist, NECA, to compete specific binding of an A_2A-selective antagonist, [³H]SCH58261, in the presence and absence of GTPγS.

In the absence of the guanine nucleotide, the NECA displacement curve was shallow and better represented by a two-site rather than one-site model (P < .01), indicating the presence of two affinity states of the receptor (Fig.4a). The Hill coefficient was 0.7 ± 0.07 (n = 4) (i.e., significantly different from unity,P < .005). The K_i value for the high-affinity sites was 4.7 ± 0.8 nM, representing 40% of total binding sites, whereas the K_ivalue for the low affinity sites was 40.8 ± 1.7 nM. In the presence of 100 μM GTPγS, the NECA competition curve was right-shifted and better fitted by a one-site model (Hill coefficient, 1.5 ± 0.3, n = 3) (Fig. 4b). TheK_i value was 100.7 ± 0.9 nM, which was significantly different (P < .001) from the values calculated for the low and high-affinity sites in basal displacement conditions.

In the absence of GTPγS the addition of peptide Gα_s(374–394)C³⁷⁹A did not shift the displacement curve (Fig. 4a) but modified the curve slope. A one-site model represented better the curve than a two-site model (Hill coefficient, 0.9 ± 0.1, n = 3), whereas a K_i value of 17.9 ± 0.9 nM was obtained. This K_i value was significantly different (P < .001) from the values derived for the low and high-affinity sites in the absence of the peptide. In the presence of GTPγS, the addition of the Gα_s peptide caused a left-shift of the displacement curve (Fig. 3b), which was still better represented by a one-site rather than a two-site model (Hill coefficient, 1.2 ± 0.2, n = 3). The K_i value (48.1 ± 0.8 nM) was 2-fold lower than that obtained in the presence of GTPγS alone (P < .001) but similar to the value for the low affinity state of the receptor in the absence of the nucleotide. Thus, the peptide appeared to modulate receptor affinity states for agonist ligands.

Because modulation of NECA displacement curves could result as a consequence of a direct effect on [³H]SCH58261 binding to the receptor, the influence of the Gα_s peptide on SCH58261 competition was evaluated. In the absence of both GTPγS and peptide, SCH58261 displaced [³H]SCH58261 binding with a monophasic pattern that was fitted by a one-site model (K_i = 0.2 ± 0.06 nM,n = 3). Neither GTPγS nor GTPγS plus the peptide caused any shift or modification of SCH58261 competition curves (Fig.4c). This result proved that Gα_s(374–394)C³⁷⁹A did not change or influence the receptor affinity for antagonist ligands.

Functional Effects of Gα_s(374–394)C³⁷⁹A on A_2A Adenosine Receptor Signal Transduction.

The effects of peptides Gα_s(374–394)C³⁷⁹A, Gα_s(384–394), and Gα_i1/2(344–354) on adenylyl cyclase stimulated by agonist activation of A_2A adenosine receptors were evaluated (Fig. 5a). The addition of peptide Gα_s(374–394)C³⁷⁹A decreased the basal cAMP production, whereas Gα_s(384–394) and a control peptide, Gα_i1/2(344–354), had no effect. However, statistical analysis showed that, in the presence of Gα_s(374–394)C³⁷⁹A, cAMP production was not significantly different from basal values.

The addition of 100 μM GTP caused a significant increase of cAMP production over the basal activity. Both CGS21680 (10 μM) and NECA (10 μM) in the presence of GTP significantly stimulated adenylyl cylcase activity as compared with basal and GTP-stimulated enzymatic activity (Fig. 5a). The addition of peptide Gα_s(374–394)C³⁷⁹A inhibited agonist-stimulated adenylyl cyclase activity by approximately 35% (Fig. 5a). In fact, the production of cAMP decreased to the same level of that obtained incubating membrane with GTP alone. On the other hand, peptides Gα_s(384–394) and Gα_i1/2(344–354) had no major effects on NECA-stimulated adenylyl cyclase activity leading to 7% inhibition and 9% stimulation of cAMP production, respectively.

Because peptide Gα_s(374–394)C³⁷⁹A inhibited basal adenylyl cyclase activity, we tested its effect on FSK-stimulated enzymatic activity to verify a possible direct modulation of the catalytic unit. Neither Gα_s(374–394)C³⁷⁹A, Gα_s(384–394), nor Gα_i1/2(344–354) significantly modified FSK-induced cAMP production (Fig. 5b). Overall, these results suggest that peptide Gα_s(374–394)C³⁷⁹A disrupted the signal transduction mechanism, which leads from agonist-activated A_2A receptor to increase of cAMP production.

NMR and Structure Calculations.

The conformational properties of Gα_s(374–394)C³⁷⁹A were extensively studied by NMR spectroscopy and molecular modeling. The usual high conformational freedom of linear short peptides in water solution leads to inextricable mixtures of isoenergetic confomers. The use of a solvent system having suitable viscosity and polarity properties allows the most energetically stable confomers to prevail, thus making the NMR spectra more interpretable (Amodeo et al., 1991). Therefore, we recorded spectra in a mixture of water:HFA (50:50, v/v). HFA, which is a typical structure stabilizing cosolvent, acts by favoring the intramolecular hydrogen bonds and consequently folded conformations (Rajan et al., 1997).

NMR analysis was made using one- and two-dimensional protonic homonuclear techniques. DQF-COSY (Piantini et al., 1982), TOCSY (Bax and Davis, 1985), and NOESY (Jenner et al., 1979) experiments were recorded on a Bruker 600 MHz at 300 K. The complete¹H chemical shift assignment of the 21-residue peptide was achieved using the interactive program package XEASY (Bartels et al., 1995).

The low region of the NOESY spectrum is shown in Fig.6. The unusually high number of cross peaks indicated a clear tendency of Gα_s(374–394)C³⁷⁹A to assume a folded conformation. In particular, the qualitative evaluation of sequential and medium connectivities was suggestive of a helical conformation in the carboxyl-terminal region.

Structure calculations were performed using the DYANA software package (Guntert et al., 1997). The NOE effects were translated into interprotonic distances, which were used as constraints in subsequent annealing procedures to produce 20 conformations. The geometry whose interprotonic distances best fitted NOE-derived distances was then refined through successive steps of restrained and unrestrained EM calculations to yield a conformer consistent with the NOE distance constraints (deviation less than 1.2 Å) and an RMS deviation from the starting geometry (backbone of nonhydrogen atoms) of 0.12 Å. As expected from the qualitative analysis of the NOESY spectrum, Gα_s(374–394)C³⁷⁹A exhibited a pattern of backbone torsion angles consistent with an α-helical secondary structure. The α-helix involved residues from Arg³⁸⁰ to Leu³⁹³ with small deviations from canonical patterns around Gln³⁸⁴. In contrast, the lack of well defined patterns of NOE connectivity for 374–380 residues indicated that the amino terminus was structurally disordered.

Recently, we have reported NMR analysis and structure calculation of Gα_s(384–394) (Albrizio et al., 2000). In Fig. 7a a superimposition of Gα_s(374–394)C³⁷⁹A and Gα_s(384–394) structures is shown. In accordance with its longer length, the 21-residue peptide presented a more folded conformation than that of the 11-residue peptide. The most significant alignment was possible for the five carboxyl-terminal Cα atoms. In both peptides, this region displayed an α-helical torsion angle pattern. The degree of folding depended on the peptide size, but both fragments retained greater compactness peculiarities in the carboxyl terminus independently from their length. Figure 7b shows an overlay of NMR-derived conformation of Gα_s(374–394)C³⁷⁹A and the 374–393 fragment extracted from the crystallographic structure of Gα_s parent protein retrieved from the Bookhaven Protein Data Bank (Sunahara et al., 1997).

Discussion

The structural determinants of receptor-G protein interactions have recently received significant attention. Because the A_2A adenosine receptor appears to be loosely coupled to G_s in rat striatal membranes (Luthin et al., 1995), a study of their interaction mechanisms is important in understanding certain peculiar characteristics of this G protein-coupled receptor system. Therefore, we synthesized and used a series of peptides corresponding to various segments of the Gα_s carboxyl terminus in an attempt to identify specific requirements for interaction of the subunit with the A_2A adenosine receptor.

This study shows that synthetic peptides corresponding to progressively longer segments of the Gα_s carboxyl terminus modulate agonist binding to the A_2A adenosine receptor in rat striatal membranes. The ability of these peptides to stimulate agonist binding is related to their size, i.e., amino acid numbers. Thus, the 11-residue peptide, Gα_s(384–394), is one of the least effective, whereas peptides containing 17 to 21 residues, Gα_s(378–394)C³⁷⁹A, Gα_s(376–394)C³⁷⁹A, and Gα_s(374–394)C³⁷⁹A, are the most active. The specificity of longer peptide effects, at the level of the receptor-G_s interface, is supported by the evidence that they show a greater ability to stimulate agonist binding in the presence of GTPγS.

A role of the carboxyl-terminal α-helix (α5) in specific receptor recognition has been indicated by the crystal structure resolution of Gα_s·GTPγS showing a continuous helix from Asp³⁶⁸ to Leu³⁹⁴ (Sunahara et al., 1997). Our NMR analysis demonstrates a marked propensity of the 11- (Albrizio et al., 2000) and 21-residue peptides to form an α-helical structure in solution. Although for the shortest peptide, a defined structure is represented by a turn of α-helix between Arg³⁸⁹ and Leu³⁹⁴, for the longest peptide, the α-helical structure spans from Asp³⁸¹ to Leu³⁹⁴. This shows good overlapping with α-5 of the Gα_ssubunit (Sunahara et al., 1997). These results are in agreement with our observation that the longest peptides have a higher ability to stimulate [³H]CGS21680 binding and support the important role of carboxyl-terminal conformation for Gα_s interaction with the A_2A adenosine receptor. Indeed, the importance of this α-helix is also pointed out by the notion that Gα_s is the only Gα subunit so far crystallized that shows a defined structure at the level of the last carboxyl-terminal residues (Noel et al., 1993; Coleman et al., 1994;Sunahara et al., 1997). An 11-residue synthetic peptide from the Gα_t carboxyl terminus does not acquire an α-helical structure, but it displays a β-turn around Gly³⁴⁸ (Glu³⁹² in Gα_s) in the presence of the inactive receptor, dark-rhodopsin (Rh) (Dratz et al., 1993). This conformation is lost when the peptide is bound to light-activated rhodopsin (metarhodopsin II or Rh*) (Dratz et al., 1993). Thus, an α-helix appears to be a specific bioactive determinant for interaction of Gα_s with its cognate receptors.

In the presence of GTPγS, which uncouples G proteins from receptors, a Gα_i1/2 carboxyl-terminal peptide, Gα_i1/2(344–354), is as effective as the shortest Gα_s peptides in stimulating agonist binding. This finding may suggest that in some conditions the A_2A adenosine receptor is able to interact with G_i proteins. Daaka et al. (1997) have shown that a classical G_s coupled receptor, the β₂-adrenergic receptor, switches to activate G_i proteins after phosphorylation by protein kinase A. A similar event may also occur for agonist-activated A_2A adenosine receptors in rat striatum.

In rat striatal membranes, agonist-activated A_2Aadenosine receptors interact tightly with heterotrimeric G_s proteins in the presence of 5 to 10 mM MgCl₂ and absence of guanine nucleotides (Mazzoni et al., 1993). This interaction stabilizes the high-affinity state of A_2A adenosine receptors. However, it is now evident (Luthin et al., 1995) that in this condition there are still some A_2A adenosine receptors uncoupled from G_s. The Gα_scarboxyl-terminal peptides may bind to these receptors in the low affinity state and stimulate [³H]CGS21680 binding (Fig. 1a). In the presence of GTPγS (Mazzoni et al., 1993), most receptors are uncoupled from G_s and are, therefore, in the low affinity state. The addition of the Gα_s peptide may restore the high-affinity state, mimicking the effect of G_s. This possibility has been tested for the longest Gα_speptide, Gα_s(374–394)C³⁷⁹A, which shows dose dependence and saturability of its stimulatory activity. Indeed, Gα_s(374–394)C³⁷⁹A evokes a larger stimulation of specific agonist binding in the uncoupling condition, but it does not stabilize the high-affinity state of the receptor either in the presence or absence (Fig. 3) of the guanine nucleotide. An intermediate receptor affinity state for the agonist may be reached.

When the effect of Gα_s(374–394)C³⁷⁹A on agonist displacement of [³H]SCH58261 binding has been examined, a stabilization of an intermediate affinity state of A_2A adenosine receptors is observed. The addition of the Gα_s peptide appears to modulate the low and high-affinity states of the receptor leading to a single intermediate affinity state. In the presence of GTPγS, the agonist competition curve is shifted to the right and fitted by a one-site model as compared with that in the absence of the nucleotide (Fig. 4b), whereas the addition of the Gα_s peptide causes a left shift of the displacement curve. The peptide is responsible for a 2-fold decrease of the K_i value. ThisK_i value is similar to that calculated for the low affinity state of the receptor in basal displacement conditions. The peptide appears to affect the receptor conformation in a subtle way. The observation also implicates the existence of multiple distinct receptor conformations, which differ in their interaction with G_s proteins. A similar possibility has been suggested for rhodopsin (Arnis and Hofmann, 1995), the human thyroid-stimulating hormone (Gudermann et al., 1996), and A₁ adenosine receptors (Gilchrist et al., 1998). Our data suggest that G_s has an important role in modulating receptor affinity for agonist ligands. The A_2A adenosine receptor affinity states seem to be regulated by the type of interaction with the Gα_s subunit. However, an alternative or complementary explanation is possible. Disruption of G_s activation by the 21-residue peptide (see below) compromises agonist-induced GTPγS binding to Gα_s and, consequently, the shift of receptor affinity without affecting receptor-G_s coupling. The indirect effect is that the binding affinity for agonist ligands does not decrease while A_2A receptors are still coupled to heterotrimeric G_s proteins. In support of this interpretation is a recent observation (Wade et al., 1999) that basic residues within the carboxyl-terminal portion of the third intracellular loop of the α_2a-adrenergic receptor are important for G_i activation but not required for receptor-G_i coupling. Similarly, different determinants of the Gα subunits can be involved in receptor-G protein coupling and activation.

The Gα_s(374–394)C³⁷⁹A peptide effectively inhibits receptor-mediated adenylyl cyclase activation in rat striatal membranes. The inhibition appears to be specific because the peptide shows no significant effect on basal and FSK-stimulated cAMP production. This observation supports the notion that the carboxyl-terminal region of Gα_s is critical for signal transduction from the activated A_2A adenosine receptor to G_s, but it is not so decisive for mimicking the effects of G_s on the receptor. In addition, disruption of signal transduction appears to require a 21-residue segment rather than the extreme carboxyl terminus. Thus, the peptide secondary structure is a determinant factor in supporting its activities. Recently, Gilchrist et al. (1998) have examined the effects of 11-residue Gα_i carboxyl-terminal peptides and analogs on agonist binding to A₁ adenosine receptors and activation of the effector pathway. Although the Gα_i peptide and some analogs completely block receptor-activated K⁺ current, they do not stabilize the high-affinity state of the receptor. The similarity is obvious and may underlie a specific molecular mechanism of interaction between adenosine receptors and G proteins.

Some studies have pointed out the importance of other regions of the Gα subunits beside the extreme carboxyl terminus in directing the G protein coupling to their cognate receptors. Both tryptic digestion experiments (Mazzoni and Hamm, 1996) and a site-directed mutagenesis investigation (Onrust et al., 1997) have demonstrated that the α4-β6 loop of Gα_t is a point of rhodopsin contact. Furthermore, using various Gα_t/Gα_i1 chimeric proteins, Bae et al. (1997) have shown that the α4-helix and α4-β6 loop are critical for specific 5-HT_1Breceptor-G_i1 interaction and are required for G protein activation by the receptor. In addition, this study has also involved a secondary role for the amino-terminal half of Gα_i1 in 5-HT_1B receptor coupling.

Our study provides additional evidence that the molecular mechanisms of interaction have similar features in different receptor-G protein systems, but the role and importance of each contact site change depending on the receptor type. In the case of the A_2A adenosine receptor, the α-helix conformation of the Gα_s carboxyl terminus seems to be important for signal transduction. However, other parts of the Gα_s molecule are probably involved in determining the allosteric modulation of receptor affinity for agonists. Thus, as suggested for the A₁ adenosine receptor, the interacting site on Gα subunits may be a mosaic with each piece playing a distinct role depending on the type of receptor and G protein. Whereas the Gα_s carboxyl terminus is pivotal for receptor-mediated activation of G_s, multiple interactions may be required to stabilize the high-affinity state of the receptor.

In conclusion we have found that the α-helical conformation of the Gα_s carboxyl terminus is involved in supporting its ability to interact with the A_2A adenosine receptor, but this part of the molecule does not stabilize the high-affinity state of the receptor. The 21-residue carboxyl-terminal peptide, which is able to disrupt signal transduction, causes a direct conformational change of the receptor with stabilization of an intermediate affinity state or induces an indirect effect by preventing GTPγS binding to Gα_s.