Molecular Analysis of β₂-Adrenoceptor Coupling to G_s-, G_i-, and G_q-Proteins

Experimental Procedures

Materials.

The cDNAs of G_iα2 and G_iα3 in pGEM-2 were kindly provided by Dr. R. Reed (Howard Hughes Medical Institute, Johns-Hopkins-University, Baltimore, MD) (Jones and Reed, 1987). The cDNA of G_16α in pCMV was a gift from Dr. D. Wu (Department of Pharmacology and Physiology, University of Rochester Medical Center, Rochester, NY) (Amatruda et al., 1991). The cDNA of G_qα in pVL1392 was kindly provided by Dr. E. M. Ross (Department of Pharmacology, University of Texas Southwestern Medical Center, Dallas, TX) (Biddlecome et al., 1996). Recombinant baculovirus encoding the unmodified versions of the G-protein subunits β₁γ₂was a kind gift of Dr. P. Gierschik (Abteilung für Pharmakologie und Toxikologie, Universität Ulm, Ulm, Germany). The G_iα2 baculovirus was kindly provided by Dr. A. G. Gilman (Department of Pharmacology, University of Texas Southwestern Medical Center, Dallas, TX). Antibodies recognizing G_iα3(anti-G_iα3, AS 86, C terminal) (Leopoldt et al., 1997), all G_iαsubunits (anti-G_iαcommon, AS 266) (Leopoldt et al., 1997), G_16α(anti-G_16α, AS 339) (Spicher et al., 1994), and G_qα (anti-G_qα, AS 369) (Spicher et al., 1994) were generously provided by Drs. B. Nürnberg and G. Schultz (Institut für Pharmakologie, Freie Universität Berlin, Germany). The antibody recognizing G_iα1/2 was from Calbiochem (La Jolla, CA). [³⁵S]Guanosine 5′-O-(3-thiotriphosphate (GTPγS; 1000–1500 Ci/mmol) was from NEN Life Science Products (Boston, MA). [³H]Dihydroalprenolol (DHA; 85–90 Ci/mmol) was from Amersham Pharmacia Biotech (Piscataway, NJ). Unlabeled GTPγS and GDP were obtained from Roche Diagnostics (Indianapolis, IN). ICI 118,55 ([erythro-dl-1(7-methylindan-4-yloxy)-3-isopropylaminobutan-2-ol]) (ICI) was from Research Biochemicals International (Natick, MA). The M1 monoclonal antibody (detecting the FLAG epitope), (−)-isoproterenol (ISO), salbutamol (SAL), (−)-ephedrine (EPH), and (±)-alprenolol were from Sigma (St. Louis, MO). Dichloroisoproterenol (DCI) was from Aldrich (Milwaukee, WI). All restriction enzymes, DNA polymerase I, and T4 DNA ligase were from New England Biolabs (Beverly, MA). Glass fiber filters (GF/C) were from Schleicher & Schuell (Dassel, Germany). All other reagents were of the highest purity available and from standard suppliers.

Construction of the cDNAs Encoding β₂AR-G_xα Fusion Proteins.

The cDNA of β₂AR-G_αsLin pGEM-3Z (Seifert et al., 1998a,b) was used as a template to amplify the C-terminal portions of the β₂AR. In this construct the β₂AR is tagged at the 5′ end with a DNA sequence encoding the cleavable signal peptide (S) from influenza hemagglutinin, which facilitates correct insertion of the receptor into the plasma membrane, followed by the FLAG epitope (F), which can be recognized by the M1 antibody. The C terminus of the receptor is tagged with a hexahistidine tag. Fusion of the β₂AR with different G_xα-subunits was achieved by sequential overlap-extension polymerase chain reactions (PCRs) usingPfu polymerase (Stratagene, La Jolla, CA). In PCR 1A, the DNA sequence of the C terminus of β₂AR was amplified with pGEM3Z-SF-β₂AR-G_αsLas a template by using a sense primer 5′ of the SacI site in the C terminus (sense SacI primer) and an antisense primer encoding the hexahistidine tag. The cDNAs of the four different G_xα-subunits were amplified in four different PCR reactions (PCR 1B_1–4) using pGEM-2-G_iα2, pGEM-2-G_iα3, pCMV-G_16α, and pVL 1392-G_qα, respectively, as templates. The sense primers annealed with the first 18 bp of the 5′ end of G_xα and included the 18 bp of the hexahistidine tag in their 5′ extensions. The antisense primers encoded the last five amino acids of the C terminus of the G_xα, followed by the stop codon and an extra XbaI site for cloning purposes in the 3′-end extension. In the case of G_qα, a BamHI site instead of anXbaI was included in the 3′-end extension of the antisense primer. In PCR 2 the cDNA fragments from PCR 1A and PCR 1B_1–4 were annealed and amplified using the sense SacI primer and the antisense primers of PCR 1B_1–4. In this way, fragments encoding the C terminus of β₂AR, a hexahistidine tag, and the G_xα followed by an XbaI orBamHI site were obtained. The fragment for β₂AR-G_iα2was digested with EcoRV and XbaI and cloned into pGEM3Z-β₂AR digested with SacI andSalI together with an oligonucleotide linker encoding (5′→3′) an XbaI site, a BamHI site, and aSalI site. The fragments for β₂AR-G_iα3, β₂AR-G_16α, and β₂AR-G_qα were digested with EcoRV plus XbaI or EcoR V plusBamHI, respectively, and cloned into pGEM-3Z-β₂AR-G_iα2digested with EcoRV plus XbaI or EcoRV plus BamHI, respectively. PCR-generated DNA sequences were confirmed by enzymatic sequencing using Sequenase version 2.0 Sequencing kit (USB, Cleveland, OH). For cloning into the baculovirus expression vector pVL 1392, the cDNAs encoding β₂AR-G_xα-fusion proteins in pGEM-3Z were digested with HindIII at the 5′-end of the SF region, blunted with DNA polymerase I (Klenow fragment), and then digested with XbaI or BamHI at the 3′-end of G_xα. Digested fusion protein DNAs were then ligated into pVL 1392 that had been digested with BglII, blunted with Klenow fragment, and subsequently digested withXbaI or BamHI.

Generation of Recombinant Baculoviruses and Cell Culture and Membrane Preparation.

Recombinant baculoviruses encoding the β₂AR-G_xα fusion proteins were generated in Sf9 cells using the BaculoGOLD transfection kit (Pharmingen, San Diego, CA) according to the manufacturer's instructions. After initial transfection, working virus stocks were generated by three sequential virus amplifications. Sf9 cells were cultured in 250-ml disposable Erlenmeyer flasks at 28°C under rotation at 125 rpm in SF 900 II medium (Life Technologies, Grand Island, NY) supplemented with 5% (v/v) fetal calf serum (Gemini, Calabasa, CA) and 0.1 mg/ml gentamicin (Roche Diagnostics). Cells were maintained at a density of 0.5 to 6.0 × 10⁶cells/ml. For infection, cells were sedimented by centrifugation and suspended in fresh medium. Cells were seeded at 3.0 × 10⁶ cells and infected with a 1:100 dilutions of high-titer baculovirus stocks encoding β₂AR-G_xα fusion proteins or nonfused β₂AR plus G_iα2. Except for the experiments shown in Fig. 5, all cultures were also coinfected with a baculovirus encoding β₁γ₂-subunits at a 1:100 dilution. Cells were cultured for 48 h before membrane preparation.

Sf9 membranes were prepared as described (Seifert et al., 1998a), using 1 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride, 10 μg/ml benzamidine, and 10 μg/ml leupeptin as protease inhibitors.

DHA Binding.

Before experiments, membranes were pelleted by a 15-min centrifugation at 4°C and 15,000g and resuspended in binding buffer (12.5 mM MgCl₂, 1 mM EDTA, and 75 mM Tris-HCl, pH 7.4). B_max values were determined by incubating Sf9 membranes (10–40 μg of protein/tube, depending on the specific expression level) with a single saturating concentration of DHA (10 nM). Nonspecific binding was determined in the presence of DHA (10 nM) plus 10 μM (±)-alprenolol. Incubations were performed for 90 min at 25°C and shaking at 200 rpm. Competition binding experiments were carried out with 1 nM DHA in the presence of ISO at various concentrations with or without GTPγS (10 μM). Bound DHA was separated from free DHA by filtration through GF/C filters and washed three times with 2 ml of binding buffer (4°C). Filter-bound radioactivity was determined by liquid scintillation counting.

GTPγS Binding and GTPγS Dissociation.

Membranes were thawed, pelleted by a 15-min centrifugation at 4°C and 15,000g, and resuspended in binding buffer. For GTPγS saturation binding studies, reaction mixtures (500 μl) contained membranes (10–81 μg of protein/tube) in binding buffer supplemented with 0.05% (w/v) bovine serum albumen, 1 μM GDP, and 0.2 to 1 nM [³⁵S]GTPγS plus unlabeled GTPγS at increasing concentrations to give the desired final ligand concentrations. Reaction mixtures additionally contained distilled water (control) and β₂AR ligands at a saturating concentration (ISO, 10 μM; ICI, 1 μM). Incubations were performed at 25°C and shaking at 200 rpm for various periods, depending on the specific properties of the fusion protein. For time-course studies, Sf9 membranes were suspended in 1500 μl of binding buffer supplemented with 1 to 2 nM [³⁵S]GTPγS plus 9 to 48 nM unlabeled GTPγS, 1 μM GDP, and distilled water (control) or β₂AR ligands at a saturating concentration (ISO, 10 μM; ICI, 1 μM). Aliquots of 200 μl (containing 15–70 μg of protein) were taken at seven different time points. In the experiments shown in Fig. 5B, the [³⁵S]GTPγS concentration was 0.4 nM. Assays were conducted in the absence of GDP or in the presence of GDP at 1 nM to 10 μM. In the experiments shown in Tables 2 and 3, reaction mixtures contained 0.4 nM [³⁵S]GTPγS, 1 μM GDP, and different β₂AR ligands at increasing concentrations. Nonspecific [³⁵S]GTPγS binding was determined in the presence of 10 μM GTPγS and was less than 0.1% of total binding. Bound [³⁵S]GTPγS was separated from free [³⁵S]GTPγS by filtration through GF/C filters, followed by three washes with 2 ml of binding buffer (4°C). Filter-bound radioactivity was determined by liquid scintillation counting.

For assessing GTPγS dissociation, GTPγS (0.5 nM [³⁵S]GTPγS plus 9.5 nM unlabeled GTPγS) was allowed to associate to membranes for 60 min in the absence of ligand. [³⁵S]GTPγS dissociation was initiated by the addition of 20 μM unlabeled GTPγS in the absence or presence of ISO (10 μM). Aliquots of 150 μl (containing 20–36 μg of protein) were taken from reaction mixtures at different time points, and bound [³⁵S]GTPγS was separated from free [³⁵S]GTPγS as described above.

SDS-Polyacrylamide Gel Electrophoresis (PAGE) and Immunoblot Analysis.

SDS-PAGE and immunoblotting were performed as described (Wenzel-Seifert et al., 1998a). Solubilized membrane proteins were separated on gels containing 10% (w/v) acrylamide. Proteins were then transferred onto nitrocellulose filters (Fig.1, A and C) or Immobilon-P transfer membranes (Millipore, Bedford, MA) (Fig. 1B). Membranes were reacted with M1 antibody (1:1000) or the antibodies directed against different G-protein subunits (1:500 each). Immunoreactive bands were visualized by sheep anti-mouse IgG (M1 antibody) and donkey anti-rabbit IgG (G-protein antibodies), respectively, coupled to peroxidase, usingo-dianisidine and H₂O₂ as substrates.

Analysis of β₂AR-G_xa Fusion Proteins by Reverse Transcription (RT)-PCR and Digestion with Restriction Enzymes.

mRNA from Sf9 cells infected with recombinant baculoviruses was isolated with the RNeasy kit from Qiagen (Valencia, CA) and treated with RNase-free DNase. mRNA was reverse-transcribed using the First Strand cDNA synthesis kit from Amersham Pharmacia Biotech. The cDNAs of G-protein α-subunits in β₂AR-G_xα fusion proteins were amplified with Taq polymerase (Sigma) using appropriate primer pairs. PCR products were digested with various restriction enzymes, separated on gels containing 2% (w/v) agarose, and visualized by ethidium bromide staining.

Miscellaneous.

Protein was determined using the DC protein assay kit (Bio-Rad, Hercules, CA). Data were analyzed by nonlinear regression, using the Prism program (GraphPad, San Diego, CA).

Results and Discussion

Analysis of the Expression of β₂AR-G_xαFusion Proteins by Receptor Antagonist Saturation Binding.

The expression levels of β₂AR-G_xα fusion proteins as determined by DHA saturation binding (lowest and highest value and mean ± S.D.) were as follows: β₂AR-G_sαL (6.1–10.6 pmol/mg, 7.3 ± 1.8 pmol/mg); β₂AR-G_sαS (2.1–4.6 pmol/mg, 2.9 ± 1.4 pmol/mg); β₂AR-G_iα2 (3.5–18.6 pmol/mg, 7.6 ± 5.9 pmol/mg); β₂AR-G_iα3 (4.5–20.2 pmol/mg; 9.6 ± 7.3 pmol/mg); β₂AR-G_qα(2.7–13.6 pmol/mg, 7.7 ± 5.1 pmol/mg); β₂AR-G_16α (2.1–10.7 pmol/mg, 4.8 ± 4.0 pmol/mg). These data show that β₂AR-G_xαfusion proteins can be expressed at comparable levels, with β₂AR-G_sαS being the fusion protein that expresses the least well.

Analysis of the Expression of β₂AR-G_xαFusion Proteins by Immunoblotting.

In our previous studies we had already documented that structurally intact β₂AR-G_sαLand β₂AR-G_sαSfusion proteins can be expressed in Sf9 cells (Seifert et al., 1998a,b). In immunoblots of membranes expressing β₂AR-G_iα2, β₂AR-G_iα3, β₂AR-G_16α, and β₂AR-G_qα, the M1 antibody recognized proteins with apparent molecular masses of ∼90 kDa (Fig. 1A). These values correspond to the sum of the molecular mass of the β₂AR (∼50 kDa) and G_xα (∼40 kDa). Note that there were no immunoreactive bands of lower molecular mass than ∼90 kDa, indicating that β₂AR-G_xα fusion proteins were not degraded by insect cell proteases. The anti-G_iα1/2 Ig reacted with β₂AR-G_iα2but not with β₂AR-G_iα3(Fig. 1B), whereas the anti-G_iα3 Ig recognized β₂AR-G_iα3but not β₂AR-G_iα2(Fig. 1C). However, the anti-G_iα3 Ig also reacted with proteins of lower molecular mass than ∼90 kDa. Because the immunoblot with the M1 antibody clearly demonstrated that fusion proteins were not degraded (see Fig. 1A), the results obtained with the anti-G_iα3 Ig suggest that the additional bands recognized by this antibody represent nonspecific reactions.

We also quantified the expression of nonfused G_iα2 in membranes coexpressing β₂AR (2.3 pmol/mg) and G_iα2 (Fig. 1B). By using membranes expressing β₂AR-G_iα2(100 μg of protein per lane) at a defined level (15 pmol/mg) as standard, we estimated that the expression level of nonfused G_iα2 is ∼300 pmol/mg. Thus, the stoichiometry of the β₂AR to G_iα2 in the coexpression system is ∼1:100. This ratio is similar to the ratio of the β₂AR to G_sα and of the FPR to G_iα2 expressed in Sf9 cells and most likely represents the in vivo β₂AR/G_iα2ratio (Ransnäs and Insel, 1988; Gierschik et al., 1991; Seifert et al., 1998a; Wenzel-Seifert et al., 1998a, 1999).

We tried to detect β₂AR-G_16α with anti-G_16α Ig (AS 339) (Spicher et al., 1994) and β₂AR-G_qα with anti-G_qα Ig (AS 369) (Spicher et al., 1994). However, the signals detected in the 90-kDa area were quite weak, and there were numerous nonspecific reactions with proteins of molecular mass lower than 90 kDa (data not shown). In addition, we could not convincingly detect β₂AR-G_iα fusion proteins with the anti-G_iαcommon Ig (AS 266) (Leopoldt et al., 1997; data not shown). Evidently, the sensitivity of these antibodies is too low to detect the corresponding β₂AR-G_xα fusion proteins at the expression levels achievable. This interpretation is supported by the fact that anti-G_iαcommon Ig detected nonfused G_iα2expressed at ∼300 pmol/mg quite well (Wenzel-Seifert et al., 1998a) but β₂AR-G_iα fusion proteins are expressed at levels that are ∼20 to 100 times lower than those of nonfused G_iα2.

Analysis of the Expression of β₂AR-G_xαFusion Proteins by RT-PCR and Digestion with Restriction Enzymes.

To overcome the difficulties associated with the use of certain G-protein antibodies, we analyzed the correct expression of fusion proteins by RT-PCR and restriction enzyme digestion using mRNA from infected Sf9 cells. The G_xα portions of fusion protein cDNAs were amplified by PCR and digested with various restriction enzymes (Fig. 1D). As expected, G_iα2-, G_16α-, and G_qα cDNAs were digested by PstI. Because of its small size, the 83-bp fragment of G_qα is not visible. Digestion withBamHI gave the expected fragments with G_iα2-, G_iα3-, and G_16α-cDNA, and digestion with EcoRI resulted in cleavage of the cDNA of G_16α and G_qα. Thus, the RT-PCR data confirm the specific expression of β₂AR-G_xαfusion proteins in Sf9 cell membranes.

Agonist-Competition Studies with β₂AR-G_xα: Differential Ternary Complex Formation with Fusion Proteins.

One of the most stringent tests of receptor/G-protein coupling is the formation of the ternary complex, i.e., the complex consisting of agonist-occupied receptor and guanine nucleotide-free G_xα (De Lean et al., 1980;Seifert et al., 1998b). This complex possesses high affinity for agonists and can be detected in radioligand binding studies in which unlabeled agonist competes with radiolabeled antagonist. Upon binding of GTP or its GTPase-resistant analog GTPγS, the ternary complex is disrupted, and the receptor converts into a state of low agonist affinity (De Lean et al., 1980; Seifert et al., 1998b).

In membranes expressing β₂AR-G_sαLand β₂AR-G_sαSplus β₁γ₂-complex, 42.1 ± 3.4 and 51.2 ±4.9% of the receptors, respectively, displayed high affinity for the full agonist ISO (K_i values, 0.7 ± 0.1 and 1.3 ± 0.3 nM, respectively) (Fig. 2, A and B). GTPγS converted the β₂ARs into a single population of receptors displaying low agonist affinity (K_i values, 155 ± 19 and 182 ± 22 nM, respectively). These data are similar to our previous data on β₂AR-G_sα fusion proteins expressed without the β₁γ₂-complex (Seifert et al., 1998b) and demonstrate that mammalian βγ-complex is not required for efficient ternary complex formation in β₂AR-G_sα fusion proteins. In membranes expressing β₂AR-G_iα2and β₂AR-G_qα there was no detectable high-affinity agonist binding (Fig. 2, C and F). G_16α conferred to 11.4 ± 2.9% of the β₂ARs the ability to bind agonist with high affinity (K_i 2.3 ± 0.4 nM) (Fig. 2E). The ternary complex formed with G_α16 was GTPγ S sensitive. In membranes expressing β₂AR-G_iα3, some high-affinity binding of ISO (14.9 ± 3.5%;K_i, 4.0 nM) could be detected as well, but this high-affinity agonist binding was GTPγS-insensitive (Fig. 2D). These data show that, with regard to ternary complex formation, the β₂AR couples much more efficiently to G_s- than to G_i- and G_q-proteins and that the β₂AR discriminates between different members of the G_i- and G_q-family. The GTPγS insensitivity of the ternary complex formed with G_iα3 reflects the inability of GTPγS to promote dissociation of this G-protein from the β₂AR. Indeed, GTPγS-insensitive ternary complex formation has been repeatedly observed (Szele and Pritchett, 1993; Gürdal et al., 1997; Seifert et al., 1998a) and points to permanent physical interaction of the receptor with the G-protein during the entire G-protein cycle (see also discussion below on receptor agonist-regulated GTPγS dissociation from G-proteins).

General Considerations for GTPγS Binding Studies with β₂AR-G_xα Fusion Proteins and Advantages and Disadvantages of the Sf9 Cell System.

Receptors catalyze GDP release from G_xα and subsequently promote the binding of GTP or its hydrolysis-resistant analog GTPγS to the G-protein (Gilman, 1987; Gierschik et al., 1991; Iiri et al., 1998;Wenzel-Seifert et al., 1998a, 1999). Because the sensitivity of the GTPγS binding assay surpasses the sensitivity of the steady-state GTPase assay (Gierschik et al., 1991; Seifert et al., 1998a), the GTPγS binding assay has become the most widely used assay to monitor receptor-G-protein coupling directly at the G-protein level. The GTPγS binding assay allows for the quantitative comparison of the coupling of a given receptor to different G-proteins and of different receptors to a given G-protein because the measurement of GTPγS binding is independent of an effector system.

A unique property of fusion proteins is the 1:1 stoichiometry of receptor and G-protein (Seifert et al., 1999c; Milligan, 2000). If each receptor interacts only with its fused G_xαpartner, the B_max values of ligand-regulated GTPγS binding and radioligand antagonist binding should be similar. Ligand-regulated GTPγS binding is the difference between maximum agonist-stimulated GTPγS binding and minimum GTPγS binding in the presence of an inverse agonist (Wenzel-Seifert et al., 1998a, 1999). The ratio of the B_max of ligand-regulated GTPγS binding and theB_max of radioligand antagonist binding is defined as the coupling factor and should be approximately 1 for fusion proteins if the receptor interacts only with its fused G_xα partner (Wenzel-Seifert et al., 1999). In fact, in Sf9 insect cells, the coupling of the β₂AR and FPR to endogenous G-proteins is minimal or not detectable, rendering fusion proteins expressed in Sf9 cells a suitable system for analyzing the coupling of a given receptor to different G-proteins and of different receptors to a given G-protein in terms of GTPγS binding under defined conditions (Seifert et al., 1998a; Wenzel-Seifert et al., 1998a, 1999).

To quantitate β₂AR coupling to insect cell G-proteins, we performed GTPγS saturation binding studies with Sf9 cell membranes expressing nonfused β₂AR. TheB_max of ISO-stimulated GTPγS binding in Sf9 membranes expressing β₂AR was extremely low. The coupling factor of 0.01 implies that only one G-protein per 100 expressed β₂AR molecules was activated upon agonist stimulation (Table 1). The poor coupling of the β₂AR to insect cell G-proteins underlines the feasibility of the Sf9 cell system for GTPγS binding studies.

However, it should be emphasized that, despite poor coupling of the β₂AR to insect cell G-proteins, there is a high concentration of as yet poorly defined GTPγS binding sites in Sf9 cell membranes. This results in high basal GTPγS binding rates (Fig.3). Although those GTPγS binding sites are irrelevant with respect to G-protein coupling of the β₂AR (Table 1), these binding sites reduce, nonetheless, the sensitivity of the GTPγS binding assay.Grünewald et al. (1996) reported high basal GTPγS binding in Sf9 membranes, too. To eliminate this background GTPγS binding and to analyze GTPγS binding to the different G-proteins under comparable conditions, we focused our attention on ligand-regulated GTPγS binding to fusion proteins. The validity of this approach and the assumption that basal GTPγS binding is irrelevant to fusion proteins is substantiated by the finding that the coupling factor for most fusion proteins was, as expected, ∼1 (Table 1). We realize that for determination of absolute K_d values for GTPγS binding to G-proteins, it would have been more appropriate not to subtract basal GTPγS binding values, but the considerable background (in particular for β₂AR-G_iα and β₂AR-G_qα fusion proteins) prevented us from doing so. To take into consideration this limitation of our studies, we use the term apparentK_d value where appropriate. We already adopted the background subtraction approach to the analysis of GTPγS binding to G_iα-proteins coupled to the FPR in the fused and nonfused state (Wenzel-Seifert et al., 1998a, 1999). Thus, our approach allows for relative comparison of apparentK_d values for GTPγS binding to different G-proteins coupled to the β₂AR and to the same G-protein coupled to the β₂AR and FPR.

Time Course of GTPγS Binding to β₂AR-G_xα: Comparison with the FPR/G_iα2 Pair and Possible Physiological Implications.

We studied the time course of GTPγS binding to β₂ARG_Xα fusion proteins in the presence and absence of a saturating concentration of ISO. The rate of GTPγS association to G-proteins is determined by the rate of GDP release (Gilman, 1987; Higashijima et al., 1988, 1990). In membranes expressing β₂AR-G_sαL, ISO decreased the apparent t_1/2 of GTPγS association about 7-fold (Fig. 3A). However, GTPγS binding at late time points of the reaction was no longer stimulated by ISO. In our previous study we showed that the β₂AR coupled to G_sαL possesses constitutive activity, i.e., even the agonist-free receptor can efficiently promote GDP release from G_sαL (Seifert et al., 1998b). Thus, we assumed that the lack of agonist effect on GTPγS binding to β₂AR-G_sαLat late time points of the binding reaction reflects the ability of the agonist-free β₂AR to efficiently promote GTPγS binding to the G-protein. To validate this assumption, we studied the effect of the inverse agonist ICI on GTPγS binding. As expected, ICI substantially inhibited GTPγS binding to β₂AR-G_sαL, particularly at late time points of the reaction (Fig. 3A).

ISO also accelerated GTPγS binding in membranes expressing β₂AR-G_sαS, but in contrast to membranes expressing β₂AR-G_sαL, a stimulatory effect of ISO was evident even at late time points of the binding reaction (Fig. 3B). We also noted that thet_1/2 of ISO-stimulated GTPγS binding to G_sαS was about twice as high as for G_sαL. These data can be interpreted in that the agonist-occupied β₂AR catalyzes GDP release from G_sαS with a slower rate than from G_sαL (Seifert et al., 1998b). ICI had only a minimal inhibitory effect on GTPγS binding to β₂AR-G_sαS(data not shown).

With respect to β₂AR-G_iα2, β₂AR-G_iα3, β₂AR-G_qα, and β₂AR-G_16α, a significant stimulatory effect of ISO on GTPγS binding was detected at all time points of the reaction. In contrast to β₂AR-G_sα fusion proteins, there was no significant (β₂AR-G_iα2, β₂AR-G_qα; Fig. 3, C and F) or only a small decrease (β₂AR-G_iα3, β₂AR-G_16α; Fig. 3, D and E) of the t_1/2 of basal GTPγS association by ISO. In addition, the t_1/2values for ISO-stimulated GTPγS binding to membranes expressing β₂AR-G_iα and β₂AR-G_qα fusion proteins were at least 3-fold higher than for β₂AR-G_sα fusion proteins. These data indicate that the β₂AR promotes guanine nucleotide exchange at G_i- and G_q-proteins much more slowly than at G_s-proteins.

The t_1/2 of FPR agonist-stimulated GTPγS binding to fused and nonfused G_iα2 is much lower (∼5 min) (Wenzel-Seifert et al., 1998a, 1999) than thet_1/2 of β₂AR-stimulated GTPγS binding to fused G_iα2 (∼45 min) (Fig.3C), indicating that the FPR promotes guanine nucleotide exchange at G_iα2 much more rapidly than the β₂AR. Based on these findings, one can assume that cellular responses mediated by the G_i- and G_q-protein-coupled β₂AR are slower in onset than responses mediated by the G_s-coupled β₂AR and the G_i-coupled FPR. The different kinetics of receptor-G-protein interaction could thus generate intracellular signals in a timely, ordered fashion.

GTPγS Saturation Binding to β₂AR-G_sαS and β₂AR-G_sαL: The β₂AR Bound to Inverse Agonist May Actively Reduce the Apparent GTPγS Affinity of G_sα.

Figure4A shows a typical GTPγS saturation experiment for β₂AR-G_sαL, and Table 1 provides a summary of the GTPγS saturation binding experiments for all fusion proteins studied. Because of the high constitutive activity of the β₂AR coupled to G_sαL (see Fig. 3A) (Seifert et al., 1998b), GTPγS saturation binding studies for β₂AR-G_sαLwere performed in the presence of ISO and ICI. Moreover, reactions were conducted for 45 min only to detect both agonist and inverse agonist effects on GTPγS binding. We are aware of the fact that after an incubation time of 45 min, an equilibrium of the binding reaction is not yet reached, but at later time points, it becomes increasingly difficult to analyze the effect of an agonist on GTPγS binding to G_sαL. The apparentK_d values of ISO-stimulated GTPγS binding to β₂AR-G_sαLand β₂AR-G_sαSwere in the subnanomolar range (0.4–0.7 nM). ICI reduced the apparent affinity of G_sαL for GTPγS by about 10-fold (apparent K_d, 4.2 nM). For both β₂AR-G_sαLand β₂AR-G_sαS, the coupling factor was ∼1, indicating that the fused β₂AR interacts efficiently with its fused G_sα partner.

To address the question whether the differential regulation by agonist and inverse agonist of the apparent GTPγS affinity of G_sα was an artifact induced by the specific incubation time chosen (45 min), we also determined the apparentK_d values for GTPγS after a 3-h incubation, i.e., when the binding reaction had reached a plateau. As already indicated above, ISO did not stimulate GTPγS binding to G_sαL under these conditions but ICI was still inhibitory (apparentK_d value of GTPγS binding, 4.9 ± 1.0 nM, mean ± S.D., n = 3). The apparentK_d value for ISO-stimulated GTPγS binding to G_sαS after a 3-h incubation was 0.5 ± 0.4 nM (mean ± S.D., n= 3). These values compare very favorably with the values obtained after a 45-min incubation (Table 1) and show that the differences in the apparent affinities of G_sα for GTPγS in the presence of the β₂AR bound to agonist and inverse agonist are observed at short and long incubation times. The differential regulation of the apparent GTPγS affinity of G_sα by the β₂AR bound to inverse agonist and agonist is intriguing because it indicates that the β₂AR bound to an inverse agonist actively reduces the apparent GTPγS affinity of G_sα. The conclusion that the β₂AR bound to ICI actively regulates nucleotide affinities of G_sαis supported by the findings that ICI increases theK_m value of the steady-state GTPase of β₂AR-G_sαL(Seifert et al., 1998a) and that ICI stimulates the binding of xanthosine 5′-triphosphate to G_sαL (Seifert et al., 1999a). Thus, our data suggest that the inverse agonist stabilizes a specific conformation in the β₂AR that actively regulates nucleotide-affinities of G_s-proteins. Evidence for the existence of specific active states of receptors bound to inverse agonists was also obtained for cannabinoid receptors (Bouaboula et al., 1997, 1999).

Partial Agonists Reduce the Apparent GDP Affinity of G_sα Less Efficiently Than Full Agonists.

The differential regulation of the apparent GTPγS affinity of G_sα by inverse agonists and full agonists raised the intriguing question of whether partial agonists increase the apparent GTPγS affinity of G_sα to a lesser extent than a full agonist. To address the question, we determined the apparent K_d value of GTPγS binding to G_sαS after a 3-h incubation, using the partial agonist dobutamine (DOB). The apparentK_d value of DOB-stimulated GTPγS binding to G_sαS was 0.7 ± 0.5 nM (mean ± SD, n = 3) and not significantly different from the apparent K_d value obtained for ISO (0.5 ± 0.4 nM). These data show that full and partial agonists do not differ from each other in their ability to alter the apparent GTPγS affinity of G_sα. In agreement with our data, partial agonists of the α₂-adrenoceptor also do not differ from full agonists with respect to the K_m of high-affinity GTP hydrolysis, i.e., there is no differential regulation of GTP affinity of G-proteins by full versus partial agonists (Wise et al., 1997).

To identify differences in the regulation of GTPγS binding stimulated by full and partial agonists, we studied the effects of DOB and DCI (Seifert et al., 1998b) on the time course of GTPγS binding to β₂AR-G_sαS. The rank order of efficacy of ligands at decreasingt_1/2 of GTPγS binding to G_sα and at saturating G_sα with GTPγS was ISO > DOB > DCI (Fig. 5A). Our data clearly show that partial agonists promote guanine nucleotide exchange less efficiently than full agonists, but differential effects of ligands on the apparent GTPγS affinity of G_sα do not explain these differences.

A recent study of cannabinoid receptors demonstrated that a major difference between full and partial agonists is that partial agonists decrease the apparent GDP affinity of G-proteins less efficiently than full agonists (Breivogel et al., 1998). In fact, with increasing GDP concentrations, the efficacy of partial agonists at promoting GTPγS binding decreases relative to the efficacy of a full agonist. Using a very similar experimental protocol as reported for cannabinoid receptors (Breivogel et al., 1998), we found that the efficacy of the partial agonists DOB and DCI at promoting GTPγS binding to G_sαS decreased with increasing GDP concentration (Fig. 5B). These data support the concept that agonist efficacy is related to the ability of ligands to reduce the apparent GDP affinity of G_xα.

Differential Regulation of the Apparent GTPγS Affinity of G-Proteins by Receptors.

The K_d value of GTPγS binding to purified G_sα is ∼175 to 1750 higher (0.7 μM) (Northup et al., 1982) than the apparentK_d values of GTPγS binding to receptor-coupled G_sα (0.4–4.2 nM). These data suggest that a receptor can dramatically increase the GTPγS affinity of G_sα. By analogy, theK_d value for GTPγS binding to purified G_i-proteins is ∼50 to 100 nM (Carty et al., 1990), whereas the apparent K_d value for FPR-regulated GTPγS binding to G_i-proteins is ∼0.7 to 1.8 nM (Table 1) (Wenzel-Seifert et al., 1998a, 1999). In addition, the β₂AR substantially increases the apparent affinity of G_qα for GTPγS compared with purified G_qα (Table 1) (Hepler et al., 1993; Chidiac et al., 1999). Thus, the data obtained for various receptors and classes of G-proteins suggest that receptors can induce a conformational change in the G_xα that increases the affinity of the G-protein for GTPγS (and presumably for the natural nucleotide GTP) considerably. Our data are in agreement with the concept that GTP/GTPγS binding does not passively follow GDP release but that receptors actively promote GTPγS binding to G-proteins (Iiri et al., 1998). Another factor that can contribute to the large differences in apparent GTPγS affinities in various systems is that purified G-proteins and G-proteins in membranes can exhibit quite different properties (Gierschik et al., 1991).

The above-discussed data raise the question whether different receptors coupled to the same G-protein alter its apparent GTPγS affinity in the same way. To address this question, we compared coupling of the β₂AR and the FPR, a prototypical G_i-protein-coupled receptor (Wenzel-Seifert et al., 1998a, 1999), to fused and nonfused G_iα2. The apparentK_d value of FPR agonist-stimulated GTPγS binding to fused and nonfused G_iα2 is ∼1 nM (Table1). The agonist-occupied β₂AR also catalyzed GTPγS binding to fused and nonfused G_iα2, but the apparentK_d values of agonist-stimulated GTPγS binding to G_iα2 were ∼25 to 70 times higher for the β₂AR than for the FPR (Fig. 4B and Table 1). The apparentK_d values of ISO-stimulated GTPγS binding to β₂AR-G_iα fusion proteins are similar to the GTPγS affinity of purified G_i-proteins (Table 1) (∼50–100 nM) (Carty et al., 1990). These data suggest that the FPR efficiently increases the GTPγS affinity of G_i-proteins, whereas the β₂AR does not. Thus, our data raise the intriguing hypothesis that receptor-specific G-protein conformational states exist that differ from each other in their GTPγS affinity. The molecular basis for such a receptor memory of G-proteins could be differences in the G-protein-coupling domains of various receptors. In fact, the G-protein-coupling domains of the β₂AR and FPR are quite different (Kobilka, 1992; Miettinen et al., 1999).

It is unknown why various receptors differ from each other with respect to regulation of the apparent GTPγS/GTP affinity of a given G-protein. One might assume that, because of the high intracellular GTP concentration (∼50 μM) (Otero, 1990), GTP can readily saturate all G-proteins, even if they are in a state of low GTP affinity. There is, however, evidence for constrained access of GTP to G_xα in native membrane systems (Wieland and Jakobs, 1992; Klinker et al., 1994). Thus, it is possible that in vivo the GTP affinity of G-proteins critically determines their efficiency as signal transducers.

Stoichiometry of β₂AR/G_i-Protein Coupling.

Inefficient β₂AR-induced increase in the apparent GTPγS affinity of G_i-proteins does not imply that the β₂AR is inefficient at activating G_i-proteins. Indeed, the coupling factor in β₂AR-G_iα2 and β₂AR-G_iα3 was ∼1, indicating that all β₂AR molecules activated their fused G_iα partner (Table 1).

One nonfused FPR molecule activates ∼1.0 to 1.5 G_i-proteins, i.e., there is rather linear signal transfer from the receptor to the G-protein (Wenzel-Seifert et al., 1999). These findings raise the question about the number of G_i-proteins activated by the nonfused β₂AR. The stoichiometry of receptor to G_i-proteins was ∼1:100 for both the β₂AR- and FPR- Sf9 cell coexpression systems (Fig. 1B) (Seifert et al., 1998a; Wenzel-Seifert et al., 1998a, 1999). This stoichiometry reflects the in vivo expression stoichiometry of signaling components (Ransnäs and Insel, 1988; Gierschik et al., 1991; Seifert et al., 1998a; Wenzel-Seifert et al., 1998a, 1999). Of interest, one nonfused β₂AR molecule activated approximately one G_i-protein molecule (Table 1). These data show that the nonfused β₂AR and FPR activate a similar number of G_i-proteins, i.e., linear G_i-protein activation is not restricted to the FPR. Additionally, our data on β₂AR/G_i-protein coupling support the recent conclusion that the 1:1 stoichiometry of receptor and G-protein in fusion proteins reflects the in vivo stoichiometry of receptor-G-protein coupling more closely than was previously assumed (Seifert et al., 1999c; Milligan, 2000). These results also underline the usefulness and relevance of fusion proteins as systems to analyze receptor-G-protein coupling.

Incomplete GTPγS Saturation Binding to β₂AR-G_qα Fusion Proteins Can Be Explained by Rapid GTPγS Dissociation: Role of Guanine Nucleotide Dissociation as Mechanism of G-Protein Deactivation and Implications for Receptor/G-Protein Coupling.

The coupling factor in β₂AR-G_16α and particularly in β₂AR-G_qα was much lower than ∼1 (Table1). An explanation for these findings could be that GTPγS dissociates from G_q-proteins much more rapidly than from G_s- and G_i-proteins. In fact, GTPγS and other GTPase-resistant guanine nucleotides can dissociate from various classes of G-proteins, including G_q-proteins (Cassel and Selinger, 1977; Higashijima et al., 1990; Berstein et al., 1992;Kupprion et al., 1993; Breivogel et al., 1998; Chidiac et al., 1999). To address this issue, fusion proteins were loaded with [³⁵S]GTPγS for 1 h in the absence of ISO to avoid interference with agonist-induced dissociation. [³⁵S]GTPγS dissociation was then stimulated by the addition of unlabeled GTPγS at a large molar excess in the absence or presence of ISO to reaction mixtures. Using this protocol, we could clearly detect time-dependent basal [³⁵S]GTPγS dissociation in membranes expressing β₂AR-G_sαL, β₂AR-G_16α, and β₂AR-G_16α (Fig.6). Of interest, basal [³⁵S]GTPγS dissociation proceeded about three times faster in membranes expressing β₂AR-G_16αand β₂AR-G_qα (Fig. 6, B and C) than in membranes expressing β₂AR-G_sα (Fig. 6A). We did not observe any effect of ISO on [³⁵S]GTPγS dissociation in membranes expressing β₂AR-G_16α and β₂AR-G_qα, presumably because the basal GTPγS dissociation rate from these fusion proteins is already high. These data support the notion that rapid dissociation of [³⁵S]GTPγS from β₂AR-G_16αand β₂AR-G_qα prevents these G-proteins from binding GTPγS in stoichiometric amounts.

It is generally assumed that GTPγS binding to G-proteins is quasi-irreversible (Gilman, 1987). Thus, on first glance, it may seem most unexpected that ISO decreased the t_1/2of GTPγS dissociation from β₂AR-G_sαLby about 6-fold. Intriguingly, in very early studies it had already been observed that ISO induced dissociation of [³H]guanylyl imidodiphosphate from G_sα in turkey erythrocyte membranes (Cassel and Selinger, 1977). However, those early studies were not followed up later. The findings that 1 mol of β₂AR-G_sα bound 1 mol of GTPγS in the GTPγS saturation binding studies and that the affinity of β₂AR-coupled G_sα for GTPγS is very high (Fig. 4A and Table 1) are in stark contrast to the rapid agonist-induced GTPγS dissociation. The fact that thet_1/2 of ISO-stimulated GTPγS binding was only moderately lower than the t_1/2 of ISO-stimulated GTPγS dissociation suggests that GTPγS dissociation occurs already in the initial phase of the GTPγS association experiment and, therefore, delays net GTPγS association.

The GTPγS dissociation studies bear important implications for the mechanism by which G-proteins are deactivated. It is generally accepted that the hydrolysis of GTP to GDP and inorganic phosphate determines the transition of the G-protein from the active to the inactive state (Gilman, 1987; Iiri et al., 1998). However, in a recent study we observed dissociations in the efficacies of β₂AR agonists at supporting adenylyl cyclase activation in the presence of inosine 5′-triphosphate and their efficacy at hydrolyzing inosine 5′-triphosphate (Seifert et al., 1999a). In addition, xanthosine 5′-triphosphate supports β₂AR-mediated adenylyl cyclase activation, but xanthosine 5′-triphosphate is not hydrolyzed (Seifert et al., 1999a). Taken together, all these data indicate that nucleotide dissociation is an important mechanism of G-protein deactivation.

The observation of highly efficient β₂AR-stimulated GTPγS dissociation from G_sαL helps us understand the physical interaction of receptors and G-proteins. These data suggest that G_sα bound to GTPγS is in physical contact with the β₂AR, despite the reduction of agonist affinity of the β₂AR by GTPγS (Figs. 2 and 6). Evidence for continuous physical contact between receptor and G-protein during the entire G-protein cycle was already obtained for muscarinic acetylcholine receptors (Matesic et al., 1989). GTPγS-insensitive ternary complex formation also argues for contact of the β₂AR with its G-protein partner during the entire G-protein cycle (see Fig. 2) (Szele and Pritchett, 1993; Gürdal et al., 1997; Seifert et al., 1998a). Thus, in contrast to the generally held opinion (Gilman, 1987; Iiri et al., 1998), G-protein deactivation may be a step in the G-protein cycle that is under the direct control of the receptor. Evidence for direct regulation of nucleoside 5′-triphosphate dissociation and hydrolysis of G_sα by the β₂AR was also provided by two previous studies from our group (Wenzel-Seifert et al., 1998b; Seifert et al., 1999b).

Pharmacological Profile of β₂AR-G_xαFusion Proteins: Evidence for Ligand-Specific Conformations of the β₂AR with Different G-Protein Coupling.

It has been shown that in some systems the pharmacological profile of a receptor depends on the specific G-protein to which the receptor is coupled (Eason et al., 1994; Gettys et al., 1994; Gurwitz et al., 1994). These findings can be interpreted to mean that specific ligands stabilize ligand-specific receptor conformations that differ from each other in their ability to activate different G-proteins. To address this hypothesis, we determined the effects of the β₂AR agonists ISO, SAL, DOB, EPH, and DCI and the inverse agonist ICI on GTPγS binding to β₂AR-G_xα fusion proteins. We assessed both ligand efficacies and ligand potencies.

Efficacies of ligands at β₂AR-G_xα fusion proteins were analyzed in two ways (Table2). First, we analyzed the efficacies of a given ligand at the various fusion proteins. Second, we analyzed the rank order of efficacies of agonists at the different fusion proteins. The efficacies of a given ligand varied largely at the different fusion proteins (0.55–1.03 for SAL; 0.40–0.98 for DOB; 0.26–0.98 for EPH; 0.08–0.67 for DCI and −0.15–0.00 for ICI). The efficacies of SAL and EPH at β₂AR-G_qα were considerably higher than at β₂AR-G_sαS, β₂AR-G_iα2, and β₂AR-G_iα3, whereas the efficacy of DCI was lower at β₂AR-G_qα than at β₂AR-G_sαS, β₂AR-G_iα2, and β₂AR-G_iα3. Moreover, the rank order of efficacy of ligands at activating GTPγS binding to β₂AR-G_iα2was ISO ≫ SAL > DOB ≈ EPH ≈ DCI, whereas the rank order of efficacy at β₂AR-G_iα3was ISO ≫ SAL ≈ DOB > EPH > DCI. We also observed differences in agonist efficacies of the β₂AR coupled to G_q-proteins. Specifically, at β₂AR-G_16α, ligands activated GTPγS binding in the order of efficacy ISO > SAL > DOB ≈ EPH > DCI. In contrast, the order of efficacy of ligands at β₂AR-G_qα was ISO ≈ SAL ≫ DOB ≈ EPH ≫ DCI. These data show that the efficacies of typical β₂AR ligands differ from each other, depending on to which G-protein the receptor is coupled. There are even differences in the pharmacological profile of the β₂AR coupled to different members of the same G-protein family, be it G_s-, G_i-, or G_q-proteins.

Differences in the pharmacological profile of the β₂AR coupled to different G-proteins were also evident upon analysis of ligand potencies (Table3). We observed large variations in the potencies of a given ligand at stimulating GTPγS binding to the different fusion proteins. The variation of the EC₅₀ values was 1.7 to 143 nM for ISO, 39 to 809 nM for SAL, 90 to 1860 nM for DOB, and 917 to 5990 nM for EPH. As was observed for agonist efficacies, the potencies of ligands did not vary systematically for different G-proteins. For example, at β₂AR-G_16α, ISO and EPH exhibited a particularly low potency, whereas at β₂AR-G_iα2, DOB showed a very low potency. We also observed variations in the rank order of potency of ligands at the different fusion proteins. At β₂AR-G_sα fusion proteins, the rank order of potency was ISO > SAL > DOB ≫ EPH. Intriguingly, the rank order of potency of ligands at β₂AR-G_iα3was different from the corresponding rank order at β₂AR-G_iα2(ISO > DOB ≫ SAL ≈ EPH at β₂AR-G_iα3versus ISO ≫ SAL > DOB ≈ EPH at β₂AR-G_iα2). Taken together, our data clearly show that the pharmacological properties of the β₂AR depend on to what G-protein the β₂AR is coupled. These data are compatible with a model in which ligand-specific receptor conformations exist. Those ligand-specific receptor conformations differ from each other in their ability to activate different G-proteins. Our conclusions for the β₂AR are supported by data regarding the pharmacological profiles of the 5-hydroxytryptamine_1A receptor, α₂-adrenoceptor, and muscarinic acetylcholine receptors coupled to different G-proteins (Eason et al., 1994; Gettys et al., 1994; Gurwitz et al., 1994).

Pertussis toxin uncouples agonist-free and agonist-occupied receptors from G_i-proteins (Gierschik et al., 1991;Wenzel-Seifert et al., 1998a). By analyzing the effect of pertussis toxin on the coupling of the β₂AR to G_i-proteins in cardiac myocytes, Xiao et al. (1999) concluded that, in this coupling situation, the β₂AR is not constitutively active. At β₂AR-G_iα2and β₂AR-G_iα3, the efficacies and potencies of partial agonists were, in general, considerably smaller than at β₂AR-G_sαL(Tables 1 and 2). These findings corroborate the conclusion by Xiao et al. (1999) that the G_i-protein-coupled β₂AR is not constitutively active.

The analysis of the effects of agonists and inverse agonists at β₂AR-G_16α did also not provide evidence for constitutive activity of the β₂AR in this coupling setting (Tables 2 and 3). For β₂AR-G_qα, we observed an unexpectedly high efficacy (but not potency) of SAL at activating GTPγS binding. However, this finding does not allow the conclusion that the β₂AR coupled to G_qα is constitutively active, because the analysis of the effects of partial agonists and inverse agonists did not reveal additional evidence for constitutive activity of the β₂AR in this coupling situation.

Conclusions

The fusion protein technique provides a rigorous approach for comparing the coupling of a given receptor to various G-protein α-subunits (Seifert et al., 1999c). It ensures a defined 1:1 stoichiometry of receptor and G-protein and promotes efficient coupling. By using this approach, we have shown that the β₂AR couples to G_s-, G_i-, and G_q-proteins as assessed by ternary complex formation and ligand-regulated GTPγS binding. The combined analysis of ternary complex formation, the kinetics of GTPγS binding, and agonist potencies and agonist efficacies revealed substantial differences in the interaction of the β₂AR with the various classes of G-proteins. Our data suggest the existence of ligand-specific receptor conformations that differ from each other in their ability to activate various G-proteins. In addition, our data on differential coupling of the β₂AR and FPR to G_i-proteins suggest that G-proteins can adopt receptor-specific conformations, i.e., G-proteins possess a receptor-memory. The 1:1 stoichiometry of receptor to G-protein in fusion proteins reflects the in vivo stoichiometry of receptor/G-protein coupling more closely than was previously assumed. Finally, GTPγS dissociation may be a much more important factor in G-protein deactivation than is generally appreciated.