Protein kinase A-mediated phosphorylation of the beta 2-adrenergic receptor regulates its coupling to Gs and Gi. Demonstration in a reconstituted system.

While classically viewed as a prototypic G(s) and adenylyl cyclase-coupled G protein-coupled receptor, recent studies have indicated that some aspects of beta(2)-adrenergic receptor (beta(2)-AR) signaling are inhibited by pertussis toxin, indicating that they are mediated by G(i)/G(o) proteins. These signals include activation of ERK MAPKs and Akt activation, as well as hypertrophic and anti-apoptotic pathways in cardiac myocytes. Studies in cultured cells have suggested the hypothesis that protein kinase A (PKA)-mediated phosphorylation of the beta(2)-AR regulates its coupling specificity with respect to G(s) and G(i). Using a Chinese hamster ovary cell system, we show that mutant beta(2)-ARs with Ala substituted for Ser at consensus PKA sites stimulate robust cyclic AMP accumulation (G(s)) but are unable to activate ERK (G(i)). In contrast, Ser --> Asp mutants are dramatically impaired in their ability to activate adenylyl cyclase but are significantly more active than wild type receptor in activating ERK. Activation of adenylyl cyclase by wild type and Ser --> Ala mutant receptors is not altered by pertussis toxin, whereas adenylyl cyclase stimulated through the Ser --> Asp mutant is enhanced. Activation of ERK by wild type and Ser --> Asp receptors is inhibited by pertussis toxin. To further rigorously test the hypothesis, we utilized a completely reconstituted system of purified recombinant wild type and PKA phosphorylation site mutant beta(2)-ARs and heterotrimeric G(s) and G(i). G protein coupling was measured by receptor-mediated stimulation of GTPgammaS binding to the G protein. PKA-mediated phosphorylation of the beta(2)-AR significantly decreased its ability to couple to G(s), while simultaneously dramatically increasing its ability to couple to G(i). These results are reproduced when a purified recombinant Ser --> Asp mutant beta(2)-AR is tested, whereas the Ser --> Ala receptor resembles the unphosphorylated wild type. These results provide strong experimental support for the idea that PKA-mediated phosphorylation of the beta(2)-adrenergic receptor switches its predominant coupling from G(s) to G(i).

The ␤ 2 -adrenergic receptor (␤ 2 -AR) 1 is a well characterized member of a large, diverse family of cell surface receptors referred to as seven membrane-spanning or G protein-coupled receptors (GPCRs). One of the primary ways in which members of this receptor super-family generate signal specificity is by interacting predominantly with one member or a subset of members of the heterotrimeric G protein family. Prior studies of the ␤ 2 -AR in both reconstituted and cellular systems have shown that it is primarily coupled to G s and thereby linked to stimulation of adenylate cyclase activity and to important physiological processes such as increased cardiac chronotropy and inotropy and relaxation of vascular and bronchial smooth muscle (1)(2)(3)(4). However, certain other aspects of ␤ 2 -AR signaling appear to be mediated via pertussis toxin-sensitive G i proteins. These signals include ERK 1/2 activation, receptor tyrosine kinase transactivation, and Akt activation (5)(6)(7). The important physiological consequences of these signals include both hypertrophic and anti-apoptotic pathways in cultured cardiac myocytes (8).
Recent cellular studies have suggested that the pertussis toxin-sensitive G i -mediated activation of ERK by ␤ 2 -ARs is nonetheless dependent on G s -mediated protein kinase A (PKA) activation, because a chemical inhibitor of PKA (H-89) blocks ERK activation by the ␤ 2 -AR in HEK 293 cells. Furthermore, because a mutant ␤ 2 -AR, which cannot be phosphorylated by PKA, is deficient in ERK activation, it has been suggested that the receptor itself must be phosphorylated by PKA to effectively couple to G i (5). However, definitive assessment of this hypothesis cannot be achieved in the complex environment of an intact cell.
An increasing number of heptahelical receptors are now known to be capable of coupling to more than one type of heterotrimeric G protein (9). But almost nothing is known about the molecular mechanisms that might regulate the relative efficacy of coupling of a particular receptor with different G proteins. Accordingly, here we examine, in a completely reconstituted system, the hypothesis that the coupling of ␤ 2 -ARs to the two G proteins, G s and G i , is regulated by PKAmediated phosphorylation of the receptor. This work, carried out both with purified recombinant wild type ␤ 2 -AR, phosphorylated in vitro by PKA, as well as with purified recombinant mutant receptors that mimic the phosphorylated form of the receptors, provides definitive evidence that PKA phosphorylation of the ␤ 2 -AR can regulate its coupling efficiency to both G s and G i . Moreover, these results expand the previously docu-mented role of PKA-mediated phosphorylation of the ␤ 2 -AR to desensitize the receptor's ability to couple to G s , demonstrating that enhanced coupling of the receptor to G i may be an equally important physiological consequence of this phosphorylation event.

EXPERIMENTAL PROCEDURES
Materials-All tissue culture reagents were from Invitrogen unless otherwise noted. Pertussis toxin was from List Biological Laboratories. ICI-118,551 was from Research Biochemicals International. [ 35 S]GTP␥S and 125 I-cyanopindolol were purchased from PerkinElmer Life Sciences. Recombinant His 6 -TEV protease was purchased from Invitrogen. All other chemicals were purchased from Sigma Chemical Co. except where noted.
Creation of Chinese Hamster Ovary ␤ 2 -AR-expressing Stable Lines-Chinese hamster ovary (CHO) cells were grown and maintained in Ham's F-12 media (Invitrogen) supplemented with 10% (v/v) fetal calf serum and antibiotics (1% penicillin (10 units/ml)/streptomycin (10 mg/ml)) at 37°C in a humidified 5% CO 2 atmosphere. The DNA used to create the stable cell lines was either the N-terminal FLAG-tagged wild type, A-4 mutant, or D-4 mutant ␤ 2 -AR in a modified pBKCMV vector. Cells in 10-cm tissue culture plates were transfected using 5 g of total DNA and a 3:1 ratio of FuGENE (Roche Molecular Biochemicals) reagent according to the manufacturer's protocol. The cells were given fresh media every 3 days in the presence of 0.5 mg/ml Geneticin until single colonies could be isolated. Single colonies were expanded and screened for stable overexpression of the ␤ 2 -AR using 125 I-(Ϫ)-cyanopindolol binding assays to crude membranes (11), with bound and free radioligand separated by filtration over 24-mm GF/C glass microfiber filters (Whatman).
Phospho-ERK Assays-Stably transfected CHO cells in six-well plates were grown in serum starvation media containing 0.1% bovine serum albumin and 10 mM Hepes, pH 7.4, for 16 h prior to agonist stimulation. Where indicated, cells were treated with pertussis toxin (PTX) 100 ng/ml for 16 h prior to agonist stimulation. The cells were stimulated with the appropriate agonist for 5 min at 37°C, and the reactions were terminated by direct addition of 250 l of 2ϫ Laemmli sample buffer to cell monolayers. The cell lysates were sonicated for 10 s, boiled for 10 min, and clarified by brief centrifugation at 20,000 ϫ g. The samples were then electrophoresed on 4 -20% gradient SDS-PAGE gels and transferred to a polyvinylidene fluoride membrane. Phospho-ERK was detected using a 1:1000 dilution of a rabbit polyclonal anti-phospho p44/42 ERK-specific antibody (Cell Signaling Technology). Duplicate gels were used to blot for total ERK expression in the lysate using a rabbit polyclonal anti-ERK 1/2 antibody (Upstate Biotechnology, 1:2000). The blots were washed in 10 mM Tris, pH 7.0/150 mM NaCl/0.05% v/v Tween 20/0.05% (v/v) Nonidet P-40 (TBS-T) and incubated with a 1:7000 dilution of secondary donkey anti-rabbit IgG conjugated to horseradish peroxidase (Jackson Laboratories) for 1 h at room temperature. The blots were visualized by chemiluminescence using SuperSignal reagent (Pierce) according to the manufacturer's instructions. Quantitation of the images was performed using a Fluor-S Multiimager with Quantity One software (Bio-Rad).
Adenylyl Cyclase Assays-Attached cells were washed twice with phosphate-buffered saline, scraped, and lysed by Dounce homogenization. Unlysed cells and nuclei were removed by centrifugation at 500 ϫ g for 5 min at 4°C, and the supernatant was centrifuged at 40,000 ϫ g for 30 min. Pelleted membranes were washed twice and resuspended in 75 mM Tris, pH 7.4, 12.5 mM MgCl 2 , 2 mM EDTA buffer. Adenylyl cyclase activities were determined as previously described (12). Reactions were initiated by the addition of membranes to the assay mixtures and incubated for 15 min at 37°C. Determinations were performed in triplicate.
Cyclic AMP Determination-Cells were grown to ϳ80% confluence in six-well dishes in media containing 0.25 mg/ml Geneticin, and where indicated PTX (100 ng/ml) was added 16 h prior to agonist stimulation. The cells were rinsed once in serum-free media and then treated with 1 mM 3-isobutyl-1-methylxanthine (IBMX) for 5 min prior to stimulation to inhibit phosphodiesterases. Cells were stimulated for 10 min at 37°C with the indicated concentration of (Ϫ)-isoproterenol in the presence of 1 mM IBMX and 0.4 mM ascorbate. Forskolin (10 M) was used as a positive control. To terminate stimulation, the media was aspirated and 200 l of ice-cold lysis buffer (50 mM Tris-HCl, pH 7.4/4 mM EDTA) was added to each well, and the samples were placed on ice. The cells were detached by scraping, and protein concentration was determined by the Bradford assay (Bio-Rad). The samples were boiled for 10 min, centrifuged for 1 min at 14,000 ϫ g, and frozen overnight at Ϫ100°C for subsequent determination of cAMP levels. The levels of cAMP were determined using a 3 H-labeled cAMP assay system (Amersham Biosciences) according to the manufacturer's protocol. The final results were expressed as picomoles of cAMP/mg of protein.
␤ 2 -AR Purification and Reconstitution-The human ␤ 2 -adrenergic receptor (␤ 2 -AR) was expressed and purified from baculovirus-infected Sf9 cells as described previously (13). The wild type and PKA point mutant receptors all contain an N-terminal FLAG epitope tag and a C-terminal hexa-histidine tag to facilitate purification. Briefly, Sf9 cells were harvested by centrifugation 48 h post-infection with the recombinant virus. The cells were lysed by Dounce homogenization, and a membrane fraction was prepared by centrifugation at 33,000 ϫ g for 30 min at 4°C. The membranes were solubilized with 0.25% (w/v) ndodecyl ␤-D-maltoside and purified by sequential chromatography using an alprenolol-Sepharose column (14) with a procedure originally modified from a prior protocol (15), a subsequent nickel affinity column (13), and a final anion exchange Q-Sepharose column. The purified receptor was assayed for its ability to bind 125 I-cyanopindolol and reconstituted into crude soybean phosphatidylcholine (20%) lipid vesicles using a modified protocol (16). Briefly, the receptor and phospholipid were incubated with the detergent n-octyl ␤-D-glucopyranoside (0.95%, w/v) in a buffer containing 100 mM NaCl, 10 mM Tris, pH 7.4, 20 mM MgCl 2 on ice and subsequently eluted over a detergent-removing Extracti-gel column (Pierce). The eluate was then treated with polyethylene glycol 8000 to promote fusion of the nascent vesicles, followed by centrifugation at 150,000 ϫ g for 2 h at 4°C. The pelleted vesicles were resuspended in 10 mM Tris-HCl, pH 7.4/100 mM NaCl and assayed for 125 I-cyanopindolol binding to determine efficiency of reconstitution and to enable equivalent amounts of receptor to be used in the subsequent experiments. Control vesicles were made by the same method but excluding any receptor in the initial incubation period.
Purification of G Protein Subunits-The H 6 TEVpQE-60 plasmids containing either G␣ s or G␣ i-1 were the generous gifts of A. Gilman (University of Texas Southwestern). The recombinant G␣ subunits were grown in Escherichia coli and purified as previously described (17). The G␤␥ subunits were provided by D. Capel and were purified from bovine brain, as previously described (18).
G protein Loading Assay-Stoichiometric amounts of G␣ and G␤␥ subunits were mixed and placed on ice. Vesicles containing the reconstituted ␤ 2 -AR protein were added to the heterotrimeric G protein mixture on ice and incubated for 15-30 min. The [ 35 S]GTP␥S binding buffer consisted of 20 mM Tris-HCl, pH 7.4, 10 mM MgCl 2 , 2 mM EDTA, and 0.1 mM ascorbate. A solution of 1 nM [ 35 S]GTP␥S was made using the binding buffer (19). Typical specific activity of the radiolabeled [ 35 S]GTP␥S mixture was 6,000 -12,000 cpm/pmol. Each binding reaction was performed in a 12-ϫ 75-mm polypropylene tube (Fisher) and consisted of 50 l of the [ 35 S]GTP␥S and 8 l of reconstituted receptor/G protein mixtures in the presence or absence of 10 M (Ϫ)-isoproterenol. Each reaction was done in duplicate at 25°C for the indicated time and quenched by the addition of 5 ml of ice-cold quench buffer (20 mM Tris-HCl, pH 7.4, 100 mM NaCl, 25 mM MgCl 2 , and 1 M GTP). The reactions were then filtered using 0.45-m BA-85 nitrocellulose membranes (Schleicher and Schuell), as described previously (20,21). The filters were washed six times with 5 ml of quench buffer, placed in Lefkofluor scintillation fluid, and counted in a Tri-Carb scintillation counter (Packard). Background counts were determined from reactions lacking G protein in the binding mixture. Control binding was determined by performing the reaction with the heterotrimeric G protein with phospholipid vesicles lacking the receptor to determine the baseline level of G protein loading. The data were plotted using Prism (GraphPad Software Inc.) and fit to a single-site hyperbolic binding curve.
In Vitro Phosphorylation Assays-Purified recombinant ␤ 2 -AR was reconstituted into phospholipid vesicles as described above. A typical reaction used 3-5 pmol of receptor in a final volume of 40 l. The phosphorylation buffer contained 20 mM Tris-HCl, pH 7.4, 2 mM EDTA, 5 mM MgCl 2 , and 60 M [␥-32 P]ATP (ϳ5000 cpm/pmol). For each reaction with PKA, 0.5 g of catalytic PKA subunit (Promega) was included. For GRK2 reactions, 0.4 g of purified recombinant GRK2 was used, and the phosphorylation buffer was supplemented with 10 M isopro-terenol and 0.1 mM ascorbate (22). The reaction was performed at 30°C for 45 min and terminated by the addition of 400 l of ice-cold 100 mM NaCl/10 mM Tris-HCl, pH 7.4. For experiments involving G protein loading using phosphorylated ␤ 2 -AR, the PKA phosphorylation reaction was terminated by the addition of a peptide inhibitor of PKA (22), and the ␤ 2 -AR was then reconstituted with heterotrimeric G proteins and used in the [ 35 S]GTP␥S binding assay. For quantitation of 32 P-phosphorylated receptor, the phospholipid vesicles were reclaimed by centrifugation at 350,000 ϫ g for 40 min at 4°C. The pellet was gently washed once in 100 mM NaCl/10 mM Tris-HCl, pH 7.4, and resuspended in 25 l of Laemmli sample buffer. The entire volume was electrophoresed on 10% polyacrylamide gels, which were then dried and exposed to a PhosphorImager screen and quantitated using a Storm Phosphor-Imager (Amersham Biosciences).

Effect of Mutant ␤ 2 -AR PKA Phosphorylation Sites on G smediated Adenylyl
Cyclase Activation in CHO Cells-We had originally identified Serines 261, 262, 345, and 346 of the ␤ 2 -AR as putative PKA phosphorylation sites, because they occur in classic consensus PKA phosphorylation site motifs RRSS on the intracellular portions of the receptor (10). Based on previously published data, we expected that PKA-mediated phosphorylation of the ␤ 2 -AR would attenuate this receptor's ability to activate adenylyl cyclase, because this is classically a G smediated pathway for the ␤ 2 -AR (22). Therefore, we expected that a ␤ 2 -AR mutant that mimics a constitutively dephosphorylated receptor (four Ser to Ala mutations, or A-4) would exhibit robust agonist-induced signaling to adenylyl cyclase, similar to the wild type receptor. Conversely, a receptor that mimics the PKA-phosphorylated ␤ 2 -AR (four Ser to Asp mutations, or D-4) would be expected to be impaired in its ability to activate adenylyl cyclase. To evaluate this hypothesis, we created stably transfected Chinese hamster ovary (CHO) fibroblast lines expressing either the wild type, D-4, or A-4 ␤ 2 -AR. Individual cell lines with comparable receptor expression levels (roughly 250 fmol/mg as assayed by 125 I-cyanopindolol binding) were used, which was significantly greater than endogenous ␤-receptor expression (5-10 fmol/mg).
We initially measured the ability of isoproterenol to stimulate adenylate cyclase in membranes prepared from each of these cell lines. As shown in Fig. 1, the wild type ␤ 2 -AR produces robust cAMP accumulation. The A-4 ␤ 2 -AR maximally stimulated adenylate cyclase activity to a slightly greater (ϳ20%) extent than wild type receptor, and its dose-response curve was shifted slightly (ϳ3-fold) to the left. In contrast, the D-4 ␤ 2 -AR barely stimulated enzyme activity above basal levels. The level of cAMP production by endogenous ␤-receptors in CHO cells under these conditions was insignificant compared with that obtained from ␤ 2 -AR overexpression.
To assess what effect, if any, ␤ 2 AR coupling to G i would have on cAMP accumulation, we performed dose-response experiments in intact CHO cells expressing the wild type and each of the mutant receptors in the presence and absence of pre-treatment with pertussis toxin (PTX). As shown in Fig. 2, PTX had no effect on wild type and A-4 ␤ 2 -AR-stimulated cAMP production. In contrast, D-4 ␤ 2 -AR mediated cAMP production was significantly increased by PTX pre-treatment, and this effect became more pronounced at higher agonist concentrations (ϳ50% increase in cAMP production). However, even in the presence of PTX, cAMP production by the D-4 ␤ 2 -AR only reached 36% of the level produced by wild type ␤ 2 -AR stimulation. These results suggest that the D-4 mutant may be impaired in cAMP production for two distinct reasons. First, there is a decrease in its ability to activate G s , and second, there is an increase in its ability to activate G i , as compared with the wild type receptor. In the whole cell cAMP assays, which were performed in the presence of the cAMP phosphodiesterase inhibitor IBMX, the D-4 mutant was about 25% as active as the wild type ␤ 2 -AR. Forskolin-mediated cAMP production was not significantly different between the various cell lines (data not shown).
Effect of Mutating ␤ 2 -AR PKA Phosphorylation Sites on G imediated ERK Activation in CHO Cells-Because the ␤ 2 -ARs containing mutations at consensus PKA phosphorylation sites had differing abilities to induce cAMP accumulation in CHO cells, we next examined the ability of isoproterenol to activate the ERK MAPK pathway (a known PTX-sensitive signal) in these different receptor cell lines. As shown in Fig. 3A, stimulation of CHO cells expressing the wild type ␤ 2 -AR with 1 M isoproterenol for 5 min induced significant ERK 1/2 phosphorylation. As reported previously for other cell lines (5, 23), ␤ 2 -AR-mediated ERK activation was inhibited by pre-treatment with PTX in CHO cells (decreased by 75%).
As shown in Fig. 3A, the D-4 ␤ 2 -AR-expressing cell line had a significantly greater ability to activate ERK 1/2, compared with the wild type ␤ 2 -AR, in response to agonist (ϳ2.5-fold enhancement), and this signal is also PTX-sensitive. The PTXinsensitive component of the ERK 1/2 signal is likely due to the additional influence of G s activation and/or cAMP generation on pathways leading to ERK 1/2 activation (24 -26). A cell line expressing the A-4 ␤ 2 -AR at comparable levels did not significantly activate ERK in response to isoproterenol stimulation. There was no significant difference in the ability of phorbol 12-myristate 13-acetate (1 M) to activate ERK between the cell lines, and this signal was not affected by pre-treatment with PTX, demonstrating that ERK pathways are intact in all the cell lines. Quantitation of these data is shown in Fig. 3B. Total ERK protein levels were comparable among the cell lines (data not shown).
Coupling of ␤ 2 -AR PKA Phosphorylation Site Mutants to G s and G i in a Purified, Recombinant Reconstituted System-To determine the G protein coupling specificity of the wild type and PKA mutant receptors, we expressed and purified recombinant receptors from Sf9 cells. We also purified recombinant G␣ s and G␣ i-1 from E. coli and endogenous G␤␥ subunits from bovine brain. We then reconstituted the purified receptors with either purified heterotrimeric G s or G i in phospholipid vesicles. The ability of the receptor to interact with the G protein was assayed by [ 35  to spontaneous G protein nucleotide turnover were determined for each experiment (in the absence of receptor) and subtracted from all subsequent calculations. This system allows for the direct determination of relative coupling efficiencies between receptors and different G proteins. Fig. 4 shows that agonist stimulation of both the wild type and A-4 ␤ 2 -ARs results in robust activation of G s over a 15-min time course. In contrast, agonist stimulation of the D-4 ␤ 2 -AR reveals significant impairment in its ability to couple to G s . The D-4 mutant is roughly 75% impaired in inducing G s activation at the 15-min time point, compared with the wild type ␤ 2 -AR. This correlates well with the decreased cAMP production stimulated by this receptor in CHO cells, as we would expect, because adenylyl cyclase activation is G s -mediated.
Because the D-4 ␤ 2 -AR is better able to activate the PTXsensitive ERK MAPK pathway in CHO cells, we reasoned that the D-4 mutant would be better able to couple to PTX-sensitive G proteins. To evaluate this, we reconstituted the purified D-4 mutant and wild type ␤ 2 -ARs with heterotrimeric G i-1 protein in phospholipid vesicles and assayed the agonist-promoted [ 35 S]GTP␥S binding to the G␣ subunit. Fig. 5 shows that the wild type ␤ 2 -AR has a small, but significant, level of agonistmediated G i coupling. However, the D-4 ␤ 2 -AR has significantly increased coupling to G i even in the absence of agonist, as compared with the wild type ␤ 2 -AR. This high level of basal coupling was further increased by isoproterenol stimulation (up to 6-fold the wild type signal) and was modestly decreased by administration of the inverse agonist ICI-118,551, suggesting receptor-regulated responses.

PKA Phosphorylation of Recombinant ␤ 2 -AR Decreases G s Coupling and Increases G i Coupling in a Reconstituted
System-Because the PKA phosphorylation site mutant ␤ 2 -ARs (serines 261, 262, 345, and 346 mutated) have different signaling characteristics from the wild type ␤ 2 -AR when expressed in CHO cells, we next wanted to confirm that these mutants represented the characteristics of either an endogenously PKAphosphorylated receptor or a constitutively dephosphorylated receptor. To confirm that we had correctly identified the relevant PKA sites within the ␤ 2 -AR, we overexpressed and purified the wild type, D-4, and A-4 ␤ 2 -ARs from baculovirusinfected Sf9 cells. The receptors were purified to similar specific activities (ϳ17,000 pmol/mg) and reconstituted into phospholipid vesicles for in vitro characterization.
The in vitro phosphorylation assays depicted in Fig. 6 show that the wild type ␤ 2 -AR is phosphorylated by both PKA and GRK2 to similar stoichiometries of 2.3 and 2.6 mol of P i /mol of receptor, respectively. When compared with the wild type, the mutant receptors were phosphorylated by GRK2 to a similar stoichiometry of 2.12 mol of P i /mol (consistent with previously published data (27)), but the PKA phosphorylation of the mutants was decreased to 0.75 mol of P i /mol of receptor. Therefore, serines 261, 262, 345, and 346 are relevant phosphorylation sites on the ␤ 2 -AR specifically for PKA. The residual PKA phosphorylation seen in the mutants is most likely due to phosphorylation of non-consensus sites under these in vitro conditions. In support of this interpretation, when expressed in cells labeled with 32 P, the PKA-site mutant ␤ 2 -AR is phosphorylated by GRK2, but has no measurable phosphorylation by PKA (28). We also assayed the purified receptors for affinity for the agonist isoproterenol by 125 I-cyanopindolol competition binding and found no significant difference in the affinities of any of the purified receptors for isoproterenol (data not shown).
The PKA site mutations of the ␤ 2 -AR were initially made because the sensitivity of the ERK MAPK signal in cells to the PKA inhibitor H-89 suggested the idea that PKA phosphorylation of the ␤ 2 -AR was necessary for this pathway's activation. However, there is some controversy over the mechanism of H-89, because at higher concentrations it appears to function as a ␤-receptor antagonist (29). Both the cellular and in vitro data for the D-4 ␤ 2 -AR suggest that it mimics a constitutively PKA-phosphorylated ␤ 2 -AR, because it is impaired in its G s signaling and enhanced in its G i signaling. To test the G protein-coupling properties of a PKA-phosphorylated wild type ␤ 2 -AR directly, we PKA-phosphorylated the wild type ␤ 2 -AR in vitro and reconstituted the receptor with purified heterotrimeric G proteins for coupling studies. Control phosphorylation reactions, carried out in the absence of ATP, were used to determine the wild type ␤ 2 -AR coupling under these conditions. As shown in Fig. 7A, PKA phosphorylation of the wild type ␤ 2 -AR decreases the receptor's ability to cause agonist-promoted G s activation by roughly 50%. In stark contrast, the PKA-phosphorylated receptor is enhanced in its ability to activate G i-1 by ϳ6-fold (Fig. 7B). It is noteworthy that, unlike the D-4 ␤ 2 -AR, the PKA-phosphorylated receptor does not display a high level of constitutive G i coupling but is nonetheless able to reach levels of agonist-induced coupling to G i-1 that are virtually identical to those achieved by the D-4 mutant ␤ 2 -AR. DISCUSSION The ␤ 2 -AR has classically been viewed as a prototypic G scoupled receptor, whose signaling properties are largely mediated by the generation of intracellular cAMP and the subsequent activation of PKA. Recently, it has been shown that, in addition to its classical signaling via cyclic AMP, the ␤ 2 -AR is also capable of activating the ERK MAPK pathway (30,31). Activation of ERK by the ␤ 2 -AR proceeds via multiple pathways, with a subset of these inhibited by pertussis toxin (PTX), indicating that they are mediated by G i /G o (5, 23, 24, 32). Cellular studies suggest that PKA may be able to regulate the G s and G i coupling of the ␤ 2 -AR. For example, ␤ 2 -AR mediated activation of ERK, although PTX-sensitive, appears to require prior activation of PKA, because a chemical inhibitor of PKA (H-89) blocks ␤ 2 -AR-mediated ERK activation (5). These data support the hypothesis that phosphorylation of a PKA substrate is necessary for ␤ 2 -AR-mediated G i activation. However, the precise intracellular targets of PKA remain undefined.
The central hypothesis of this work is that PKA-mediated phosphorylation of the ␤ 2 -AR itself allows for the generation of PTX-sensitive signals from the receptor by both decreasing G s and increasing G i coupling. Previously, the ␤ 2 -AR has been shown to be a substrate for PKA phosphorylation (33). Moreover, in vitro GTPase assays have shown that PKA phosphorylation of the ␤ 2 -AR decreases the ability of the receptor to activate G s by ϳ50% (22). These experiments did not, however, examine the ability of phosphorylation to alter coupling to other G protein isoforms.
Accordingly, to achieve a definitive assessment, we established an entirely reconstituted system of purified, recombinant components and assessed whether phosphorylation of the ␤ 2 -AR by PKA could regulate the receptor's coupling to G s and G i . We undertook three approaches to assess this hypothesis. First, we mimicked PKA phosphorylation by replacing the consensus serine PKA phosphorylation sites of the ␤ 2 -AR with aspartate residues and examined the effect of these mutations on G s -and G i -mediated signals. Second, we validated our mutant receptor-G protein coupling data in an in vitro reconstituted system. Finally, we phosphorylated the ␤ 2 -AR with PKA and assessed the ability of the phosphorylated receptor to couple to G proteins relative to the non-phosphorylated receptor.
Whether assessed with mutant receptors in a cellular system or with purified recombinant mutant receptors or PKA-phosphorylated ␤ 2 -ARs in a reconstituted system, all three approaches provided congruent sets of data consistent with our hypothesis.
The slight differences in the behavior of the A-4 receptor in different assay systems are likely due to two offsetting effects of the mutation of Ser-261 and Ser-262 located within a key G protein coupling region of the receptor (34). On the one hand, the mutations eliminate any desensitization attributable to PKA-mediated phosphorylation of the receptor (10). On the other hand, as demonstrated in the reconstitution studies in which the receptors can be studied in the absence of other regulatory influences, the Ser 3 Ala mutations (Fig. 4) somewhat diminish coupling of the receptors to G s . As a result, such mutants of the ␤ 2 -AR may appear slightly superactive or slightly impaired in their cAMP-generating capacity depending on the details of the assay, such as adenylyl cyclase versus whole cell cAMP, time of stimulation with agonist, cell type, etc.
In our membrane adenylate cyclase assays, the A-4 mutant showed a modest increase in V max (20%) and EC 50 for isoproterenol (3-fold). In our whole cell cAMP assays, A-4 showed no significant difference from wild type in V max and a slight decrease in EC 50 for isoproterenol (3-fold).
These results are consistent with previous studies using PKA site mutants of the ␤ 2 -AR. In this laboratory, both Liggett et al. (35) and Hausdorff et al. (10), working with Chinese hamster fibroblast (CHW) cells, found no significant difference between the A-4 and wild type receptor in either EC 50 or V max for isoproterenol-stimulated cyclase activity or cAMP production, although a statistically insignificant increase in V max for A-4 (ϳ25%) was noted by Liggett et al. These investigators also noted a significant increase in cAMP generated by A-4 in whole cell cAMP assays but only after 15 min of incubation with isoproterenol when desensitization had occurred. Our whole cell cAMP experiments were done at 10 min, and, like Liggett et al., we observed no significant difference in cAMP at that time point. Yuan et al. (36) also examined the adenylyl cyclase-stimulating activity of a related ␤ 2 -adrenergic receptor mutant, in which only serines 261 and 262 were changed to Ala. Using L cells, they found no significant change in the EC 50 , whereas V max was modestly (30%) increased. More detailed examination of the coupling efficiency of the mutant receptor, however, using quantitation of high and low affinity state agonist binding and GTP shifts, indicated that the mutant was somewhat uncoupled, consistent with our data. Thus, our current as well as previous adenylate cyclase and whole cell cAMP data with Ser 3 Ala PKA mutants of the ␤ 2 -adrenergic receptor all indicate that such mutant receptors are modestly impaired in G s coupling and in desensitization.
The most interesting and significant findings of the present studies, however, relate to the characteristics of the novel Ser 3 Asp (D-4) receptor and PKA-phosphorylated receptor. In the adenylyl cyclase and whole cell cAMP assays, D-4 was markedly impaired with respect to G s coupling and enhanced in its G i coupling, as evidenced by the enhancement of D-4 receptor-stimulated cAMP production in PTX-treated cells. The impaired coupling to G s was confirmed in the reconstitution experiments. This effect of receptor phosphorylation to not only diminish coupling to G s but to enhance coupling to G i (which inhibits adenylyl cyclase) provides an additional mechanism by which PKA phosphorylation of the receptor can "desensitize" cAMP generation in response to ␤-agonists.
The requirement for PKA phosphorylation of the ␤ 2 -adrenergic receptor for G i coupling was further revealed in the whole cell ERK activation assays and in the reconstitution experiments. In the CHO cells, D-4 was considerably more efficacious than the wild type receptor in mediating pertussis toxin-sensitive (i.e. G i -dependent) ERK activation, whereas, strikingly, A-4 was completely devoid of activity. Similarly, in the reconstitution experiments D-4 showed markedly enhanced coupling to G i . Particularly striking was the similarity between the D-4 mutant and the PKA-phosphorylated ␤ 2 -AR in the reconstituted system. They both demonstrated a 6-fold increase in G i coupling over the non-phosphorylated wild type ␤ 2 -AR. With respect to G s coupling, the D-4 mutant was impaired 75% and the PKA-phosphorylated ␤ 2 -AR was impaired by 50% compared with non-phosphorylated wild type ␤ 2 -AR.
The somewhat greater constitutive, i.e. agonist-independent, activity of the D-4 mutant compared with the phosphorylated receptor with respect to G i activation, and its somewhat greater impairment with respect to G s activation likely reflect the greater number of added negative charges on the D-4 mutant compared with the phosphorylated receptor. These extra negative charges could alter the D-4 mutant receptor's conformation to further hinder G s activation and to increase basal signaling to G i . We had initially created the D-4 mutant, in part, because the comparable A-4 mutant had already been shown to have no measurable phosphorylation by PKA in cells (28). Thus, we expected no further PKA-mediated phosphorylation of the D-4 mutant. It can be reasonably argued that a D-2 mutant might more accurately reflect the phosphorylation stoichiometry observed in our in vitro assays. However, concerns regarding which two serines to mutate and the fact that the remaining two serines could be phosphorylated would make D-2 mutant analysis less informative.
PKA-mediated "switching" of ␤ 2 -AR coupling from G s to G i provides a mechanism by which G i -mediated signals can be generated by the ␤ 2 -AR. Previous studies using chemical inhibitors and assays of downstream signaling events also support such a mechanism. Prior work has shown that small peptides are capable of interacting with and activating G proteins in vitro (37)(38)(39). Small synthetic peptides (14 -22 amino acids) corresponding to portions of the third intracellular loop of the ␤ 2 -AR have been shown to cause G s activation in vitro. One of these peptides, containing two of the putative phosphorylation sites for PKA, when phosphorylated by PKA, showed a decreased ability to stimulate G s loading and an increased ability to promote G i loading compared with non-phosphorylated peptide (40). These experiments suggest a role for PKA phosphorylation of the ␤ 2 -AR at these sites in mediating G i -dependent signaling. However, the data presented here represent the first direct demonstration of altered ␤ 2 -AR G protein coupling specificity by PKA phosphorylation of the intact receptor alone.
The phenomenon of a classically G s -coupled receptor altering its G protein coupling by "switching" its G protein specificity is a novel finding with numerous, potentially important implications. From the standpoint of GPCR pharmacology, this mechanism provides further evidence that the third intracellular loop and proximal tail of the receptor are important in determining the extent and specificity of G protein coupling (34). Alterations to these regions such as phosphorylation or point mutation have the potential to significantly alter the signaling events downstream of the receptor. The alteration of a receptor's G protein coupling specificity by PKA-mediated receptor phosphorylation may not, however, be unique to the ␤ 2 -AR and may have significant consequences for other GPCRs. A specific example involves the vasoactive intestinal peptide receptor in pancreatic acinar cells, where a typically G s -coupled receptor appears to evoke G i -mediated responses in a manner dependent on PKA activation (41). A more recent example involves the mouse prostacyclin receptor, which is normally G s -coupled but can induce G i -and G q -mediated signals in a manner dependent on PKA activity. Furthermore, this effect is blocked by mutation of a consensus PKA site within the prostacyclin receptor (42).
This switching mechanism represents a novel means of modulating a signal from a G s -coupled receptor. This could represent a form of negative feedback where the PKA that is activated by the G s pathway phosphorylates the receptor and thus causes enhanced G i activation to limit further cAMP production. Of potentially greater interest, this may be a mechanism by which the ␤ 2 -AR is able to activate a number of PTXsensitive signals. These signals include ERK activation, antiapoptotic signaling via Akt in cardiac myocytes, and transactivation of receptor tyrosine kinases (5-7). These G i -mediated signals appear to be critically involved in the physiologically important phenomenon of cardiac myocyte hypertrophy, which is one of the earliest pivotal changes in heart failure of a number of etiologies (8,43,44). Understanding this mechanism may lead to novel therapeutic approaches allowing for blockade of the hypertrophic growth signals in clinical situations of chronic adrenergic receptor stimulation while leaving other important aspects of receptor signaling intact.