The β3-Adrenergic Receptor Activates Mitogen-activated Protein Kinase in Adipocytes through a Gi-dependent Mechanism*

Promiscuous coupling between G protein-coupled receptors and multiple species of heterotrimeric G proteins provides a potential mechanism for expanding the diversity of G protein-coupled receptor signaling. We have examined the mechanism and functional consequences of dual Gs/Gi protein coupling of the β3-adrenergic receptor (β3AR) in 3T3-F442A adipocytes. The β3AR selective agonist disodium (R,R)-5-[2[[2-(3-chlorophenyl)-2-hydroxyethyl]-amino]propyl]-1,3-benzodioxole-2,2-dicarboxylate (CL316,243) stimulated a dose-dependent increase in cAMP production in adipocyte plasma membrane preparations, and pretreatment of cells with pertussis toxin resulted in a further 2-fold increase in cAMP production by CL316,243. CL316,243 (5 μm) stimulated the incorporation of 8-azido-[32P]GTP into Gαs (1.57 ± 0.12; n = 3) and Gαi (1.68 ± 0.13;n = 4) in adipocyte plasma membranes, directly demonstrating that β3AR stimulation results in Gi-GTP exchange. The β3AR-stimulated increase in 8-azido-[32P]GTP labeling of Gαi was equivalent to that obtained with the A1-adenosine receptor agonist N 6-cyclopentyladenosine (1.56 ± 0.07; n = 4), whereas inclusion of unlabeled GTP (100 μm) eliminated all binding. Stimulation of the β3AR in 3T3-F442A adipocytes led to a 2–3-fold activation of mitogen-activated protein (MAP) kinase, as measured by extracellular signal-regulated kinase-1 and -2 (ERK1/2) phosphorylation. Pretreatment of cells with pertussis toxin (PTX) eliminated MAP kinase activation by β3AR, demonstrating that this response required receptor coupling to Gi. Expression of the human β3AR in HEK-293 cells reconstituted the PTX-sensitive stimulation of MAP kinase, demonstrating that this phenomenon is not exclusive to adipocytes or to the rodent β3AR. ERK1/2 activation by the β3AR was insensitive to the cAMP-dependent protein kinase inhibitor H-89 but was abolished by genistein and AG1478. These data indicate that constitutive β3AR coupling to Gi proteins serves both to restrain Gs-mediated activation of adenylyl cyclase and to initiate additional signal transduction pathways, including the ERK1/2 MAP kinase cascade.

Long before the discovery of the ␤ 3 AR and its recognition as a unique, adipocyte-specific receptor controlling lipolysis and thermogenesis, Rodbell and colleagues (1) made the observation that there was an unusual, biphasic stimulation of cAMP production in adipocytes in response to the ␤-adrenergic receptor agonist isoproterenol. Depending upon the concentration of GTP in the assay, isoproterenol could either stimulate or inhibit adenylyl cyclase activity in adipocyte plasma membranes. Murayama and Ui (2) showed that this inhibitory phase could be relieved by pretreatment of adipocytes with pertussis toxin (PTX). 1 This curious observation lay fallow until studied later in greater detail by Bégin-Heick (3)(4)(5). However, it was not until the cloning and characterization of the ␤ 3 AR gene and the development of selective ␤ 3 AR agonists (6, 7) that it was postulated that this novel adipocyte-specific ␤AR may be responsible for the biphasic adenylyl cyclase response in adipocytes (8). We have previously noted that despite the relatively high level of expression of the ␤ 3 AR in adipocytes, the efficiency of coupling of the ␤ 3 AR to stimulation of adenylyl cyclase is low (9). However, there has been no clear biochemical demonstration of physical coupling of the ␤ 3 AR to G i , other than comparative functional experiments in the presence or absence of PTX (10), nor has there been any indication of what additional second messenger pathway may be activated as a consequence of this putative coupling of ␤ 3 AR to G i .
Recently, many G protein-coupled receptors have been shown to mediate cellular growth or differentiation responses through the activation of MAP kinase cascades (11). Receptors signaling via PTX-sensitive G i/o proteins, as well as PTX-insensitive G q/11 proteins may activate the ERK1/2 MAP kinase cascade through a mechanism involving tyrosine protein phosphorylation and the activation of the low molecular weight G protein p21 ras (12)(13)(14). Little is known about the potential role of ␤ARs in the regulation of the MAP kinase pathway. Recently, we have found that in fibroblasts the ␤ 2 AR mediates Ras-dependent ERK1/2 activation through its ability to couple to a PTX-sensitive G i protein (15). ␤ 2 AR coupling to G i occurs as a result of PKA-dependent phosphorylation of the receptor, which effectively "switches" receptor coupling from G s to G i proteins. In contrast, ␤ 2 AR-mediated ERK1/2 activation in S49 lymphoma cells is an entirely G s -dependent process (16). In this system, PKA-mediated phosphorylation of the low molec-* This work was supported by National Institutes of Health Grants DK46793 and DK53092 (to S. C.) and DK02352 (to L. M. L.), National Institutes of Health Training Program Grants T32ES0731 (to K. J. S.) and T32GM07105 (to S. K. S.), and National Institutes of Health Medical Scientist Training Program Grant T32GM07171 (to G. J. D. R.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Here, we demonstrate that stimulation of the ␤ 3 AR in adipocytes induces the direct activation of both G s and G i . In these cells, G i activation results in both the attenuation of ␤ 3 ARmediated stimulation of adenylyl cyclase and the activation of the ERK1/2 MAP kinase pathway. Unlike the ␤ 2 AR signal in fibroblasts, ␤ 3 AR activation of the ERK1/2 pathway is independent of cAMP and PKA. These data suggest that the promiscuous coupling of the ␤ 3 AR in adipocytes permits the simultaneous transduction of two independent signaling pathways. This property of the ␤ 3 AR may be responsible, in part, for the unique physiological effects of selective ␤ 3 AR agonists in vivo, such as their potency for stimulating lipolysis (17,18) and their ability to prevent or reverse obesity (6, 19 -22).
Cell Culture and Transfections-3T3-F442A preadipocytes were maintained in Dulbecco's modified Eagle's medium supplemented with 10% calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin (Life Technologies, Inc.) at 37°C in a humidified 5% CO 2 atmosphere. Upon reaching 90% confluence, the cells were stimulated to differentiate into adipocytes by culturing in a differentiation media composed of Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, 100 g/ml streptomycin, and 5 g/ml insulin (23). HEK-293 cells were grown in six-well dishes and transfected with 1.2 g of pBC␤3 or pBC␤2 DNA (24) by calcium phosphate co-precipitation (25). Ten h later, cells were washed in 1 mM EGTA in phosphate-buffered saline and incubated in growth medium for 24 h. The cells were then serum-starved for 24 h and treated with PTX (100 ng/ml), propranolol (10 M), SR59230A (25 M), CL316,243 (10 M), or isoproterenol (100 M), and MAP kinase assays were performed as described below. Cells that were pretreated with PTX were incubated with the toxin for 16 -20 h at a concentration of 100 ng/ml.
Adenylyl Cyclase Assays-Plasma membranes were prepared from 3T3-F442A adipocytes or Sprague-Dawley epididymal white adipose tissue as follows. For cultured cells, the monolayers were gently rinsed with ice-cold phosphate-buffered saline. All subsequent steps were performed at 4°C. Lysis buffer (5 mM Tris/2 mM EDTA, pH 7.2, containing 10 g/ml soybean trypsin inhibitor, 10 g/ml benzamidine, 2 g/ml aprotinin, and 0.1 mM phenylmethylsulfonyl fluoride) was added, and the cells were released from the dish with a rubber policeman. Following gentle disruption with a glass/Teflon homogenizer, nuclei were removed by centrifuging for 5 min at 800 ϫ g. The supernatant was collected and centrifuged at 40,000 ϫ g for 20 min. The pellet was resuspended in fresh lysis buffer with a Teflon pestle, centrifuged again, and then resuspended in a small volume of lysis buffer. For rat adipocytes obtained from epididymal adipose tissue, the same procedure was followed except that first the tissue was finely minced, and adipocytes were isolated by incubating the tissue pieces with collagenase (Worthington) (26). Adenylyl cyclase activity in plasma membranes was measured as described previously (9). Control incubations in the absence or presence of isobutylmethylxanthine (0.25 mM) found no evidence of particulate cyclic nucleotide phosphodiesterase activity in these plasma membranes. In some experiments, cAMP production was measured in intact 3T3-F442A adipocytes. For these whole cell assays, growth media was replaced with serum-free medium 3 h prior to stimulation. The serum-free medium consisted of Dulbecco's modified Eagle's medium supplemented with 1 g/liter fraction V bovine serum albumin, 10 mM HEPES, pH 7.4, 100 units/ml penicillin, and 100 g/ml streptomycin. Then cells were incubated with the indicated concentra-tions of ␤-agonist in fresh serum-free medium for 10 min at 37°C in the presence of 0.25 mM isobutylmethylxanthine. Reactions were stopped by the addition of ice-cold 5% trichloroacetic acid, and particulate material was removed by centrifugation. The cAMP concentrations from both assay methods were determined by radioimmunoassay using a polyclonal antiserum to iodinated cAMP (27). Protein concentrations were determined by the method of Bradford (28).
Western Blotting and Photolabeling of Heterotrimeric G Proteins-Western blotting was performed with antibodies specific for individual G␣ subunits. Adipocyte membranes (30 g of protein) were solubilized in SDS-polyacrylamide gel electrophoresis sample buffer and resolved on 4 -20% Novex gradient gels (San Diego, CA). A recombinant G protein standard mix containing G␣ i1 , G␣ i2 , G␣ i3 , and G␣ o was included as a positive control (gift of Dr. Pat Casey). The proteins were then transferred to nitrocellulose membranes. The membranes were incubated for 16 h at 4°C with rabbit anti-G␣ i1ϩ2 , anti-G␣ i3 , or anti-G␣ s , 1:1000 dilution, followed by alkaline phosphatase-conjugated goat antirabbit antisera (1:10,000 dilution, Amersham Pharmacia Biotech) as secondary antibody. Immunoreactive bands were visualized by Storm PhosphorImager (Molecular Dynamics).
Synthesis and purification of the 8-azido-[ 32 P]GTP was performed as described (29,30) with minor modifications (31). For some experiments, 8-azido-[ 32 P]GTP was purchased from Andotek Life Sciences. The photoaffinity labeling procedure of Offermanns (32) was followed with modifications (31). For labeling of the G␣ i subunits, adipocyte membranes (50 g of protein) were incubated with 8-azido-[ 32 P]GTP (0.5 Ci) for 10 min at 30°C in a volume of 60 l containing 50 mM HEPES (pH 7.6), 1 mM EDTA, 50 mM NaCl, 500 M MgCl 2 , 50 M GDP, 100 M ATP, and 1 g/ml adenosine deaminase in the presence or absence of the adenosine A 1 receptor agonist N 6 -cyclopentyladenosine or the ␤ 3 ARselective agonist CL316,243. Labeling of the G␣ s subunit was performed as above at a free Mg 2ϩ concentration of 2-5 mM. The binding reactions were stopped by placing tubes on ice, and all subsequent steps were performed at 4°C. The membranes were collected by centrifugation for 10 min at 14,000 ϫ g and resuspended in a buffer containing 50 mM HEPES (pH 7.6), 1 mM EDTA, 50 mM NaCl, 500 M MgCl 2 , and 2 mM dithiothreitol. Following irradiation of the samples with a UV lamp (4 watts, 254 nm) at a distance of 3 cm for 5 min, the samples were collected by centrifugation as before and solubilized in Laemmli sample buffer. Proteins labeled with 8-azido-[ 32 P]GTP were resolved by SDSpolyacrylamide gel electrophoresis using a 10% polyacrylamide gel and visualized by a Storm PhosphorImager (Molecular Dynamics).
MAP Kinase Assays-Activation of MAP kinase was determined by measuring the phosphorylation state of ERK 1 and ERK 2 (33). For this assay, 3T3-F442A cells were grown and differentiated in six-well culture plates and serum-starved for 24 h prior to stimulation with ␤-adrenergic agonists or growth factors. The serum-free medium consisted of Dulbecco's modified Eagle's medium supplemented with 1 g/liter fraction V bovine serum albumin, 10 mM HEPES, pH 7.4, 100 units/ml penicillin, and 100 g/ml streptomycin. Five min after the addition of ␤-agonists or growth factors, the medium was removed, and the cells were solubilized by the direct addition of Laemmli sample buffer (34). These cell lysates were sonicated for 5 s, and aliquots (30 g protein/ lane) were resolved by SDS-polyacrylamide gel electrophoresis. ERK1/2 phosphorylation was detected by protein immunoblotting using a 1:3000 dilution of rabbit polyclonal phospho-MAP kinase-specific antisera (NEBiolabs) with a 1:10,000 dilution of alkaline phosphataseconjugated goat anti-rabbit antisera (Amersham Pharmacia Biotech) as secondary antibody. Quantitation of ERK1/2 phosphorylation was performed using a Storm PhosphorImager (Molecular Dynamics). After quantitation of ERK1/2 phosphorylation, nitrocellulose membranes were stripped of immunoglobulin and reprobed using rabbit polyclonal anti-ERK 2 IgG (Upstate Biotechnology, Inc.) to confirm equal amounts of ERK 2 protein.

RESULTS AND DISCUSSION
Stimulation of 3T3-F442A adipocyte plasma membranes with the selective ␤ 3 AR agonist CL316,243 resulted in a 3-4fold increase in adenylyl cyclase activity Fig. 1A). Membranes prepared from cells pretreated with PTX exhibited a more robust increase in cAMP production, which was at least 2-fold above that observed in membranes from naive cells. Whereas basal levels of cAMP generation were elevated following PTX treatment, the effect of PTX on the -fold activation of adenylyl cyclase by forskolin was minimal (Fig. 1B). We obtained similar results from Chinese hamster ovary cells, which heterologously express the mouse ␤ 3 AR (data not shown) (35). These results suggest that the ␤ 3 AR is constitutively coupled to both G s and G i , because inhibition of G i function leads to increased tonic production of cAMP.
Rat white adipocytes have been shown to contain the two splice variants of G␣ s and the three isoforms of G␣ i (␣ i1 , ␣ i2 , ␣ i3 ) (5,36). As shown in Fig. 2, 3T3-F442A adipocytes express the same repertoire of G proteins as found in rat white adipocytes: two splice variants of G␣ s , G␣ i3 , and G␣ i1 /G␣ i2 . Others have previously shown that adipocytes do not contain the PTXsensitive G␣ o protein (36 -39). To determine whether the PTX effects on cAMP production resulted from constitutive coupling of the ␤ 3 AR to both G s and G i , the photolabile GTP analog 8-azido-[ 32 P]GTP was used to measure ␤ 3 AR agonist-dependent GTP loading of G␣ s and G␣ i . This method has been used by several investigators to demonstrate physical coupling between G protein-coupled receptors and individual G protein ␣ subunits, based upon several well defined criteria for receptor-G protein interaction (40 -42). Because the binding of GTP analogs to G␣ subunits is highly dependent upon free magnesium concentrations (31), we conducted our experiments in adipocyte plasma membranes under separate conditions favorable for 8-azido-[ 32 P]GTP labeling of each G␣ family. Because the ␤ 3 AR is coupled to G␣ s and activation of adenylyl cyclase, Fig. 3A shows that under high magnesium conditions, which favor receptor-stimulated GTP loading of G␣ s , CL316,243 could stimulate specific 8-azido-[ 32 P]GTP labeling of G␣ s (1.57 Ϯ 0.12; n ϭ 3; p Ͻ 0.001). Under low magnesium conditions, which are optimal for determining receptor-stimulated GTP loading to G␣ i , there was a similar 1.68 Ϯ 0.13-fold increase (n ϭ 4; p Ͻ 0.001) in specific 8-azido-[ 32 P]GT G protein labeling of the 40 -42-kDa G␣ i species (Fig. 3B). We obtained similar data from 3T3-F442A membranes (not shown). Fig. 3B also shows that the ability of CL316,243 to stimulate G i -GTP exchange in adipocyte membranes was equivalent to that obtained with the A 1 -adenosine receptor agonist N 6 -cyclopentyladenosine (1.56 Ϯ 0.07; n ϭ 4; p Ͻ 0.001). N 6 -Cyclopentyladenosine served as a positive control because the A 1 -adenosine receptor is expressed in both primary adipocytes (43,44) and differentiated 3T3-F442A adipocytes (45) and couples to all three G␣ i species (46,47).
A number of G i -coupled receptors have been shown capable of activating MAP kinase (11). Recently, Daaka and colleagues (15) reported that the ␤ 2 AR could couple to both G␣ s and G␣ i , with consequent activation of MAP kinase. We therefore determined whether one of the functional consequences of the coupling of the ␤ 3 AR to G␣ i in 3T3-F442A adipocytes was the stimulation of MAP kinase. Intact 3T3-F442A adipocytes treated with CL316,243 (5 M) exhibited approximately a 2.5fold stimulation of ERK1/2 phosphorylation (Fig. 4). Activation of the endogenous ␤ 2 AR in 3T3-F442A adipocytes with the ␤ 2 AR-selective agonist salbutamol also yielded a similar 2-fold stimulation of ERK1/2 phosphorylation. Both ␤AR responses were less robust than that observed following activation of endogenous receptors for EGF, consistent with previous findings (13). Also shown in Fig. 4, pretreatment of 3T3-F442A adipocytes with PTX completely blocked ERK1/2 phosphorylation in response to either ␤ 2 AR or ␤ 3 AR stimulation, but it did not affect stimulation by EGF. The addition of propranolol (0.1 M) eliminated ERK1/2 phosphorylation in response to salbutamol but had no effect on the activation of MAP kinase induced by CL316,243 (not shown), consistent with a CL-mediated effect exclusively through the ␤ 3 AR (7).
The relative sensitivity of the ␤ 3 AR to activate MAP kinase versus cAMP accumulation in adipocytes was assessed by treating intact cells with increasing concentrations of CL316,243 (Fig. 5). Because the MAP kinase assays involve the stimulation of intact cells, we included cAMP dose-response data also from intact cells. The EC 50 for cAMP production in these experiments was 12 nM, whereas parallel measurements of MAP kinase activation (EC 50 ϭ 280 nM) indicated that the PTXsensitive coupling of the ␤ 3 AR to MAP kinase activation is less potent under these conditions. Nevertheless, the dose-response curves for both measurements yielded a unit slope, indicative of the high selectivity of CL316,243 for the ␤ 3 AR (7,48). Note that the EC 50 for cAMP production in these whole cell experiments is significantly less than that found in plasma membranes (Fig.  1). However, it is generally recognized that dose-response curves as well as ligand binding data from intact cells are shifted "to the left," when compared with data obtained using isolated membrane preparations (49,50). These comparative studies shown in Fig. 5 suggest that cAMP production is the FIG. 1. ␤ 3 AR agonist stimulation of cAMP production in 3T3-F442A adipocytes. A, dose-response curve for agonist stimulation of cAMP production in response to CL316,243. Membranes of untreated or PTX-pretreated adipocytes were prepared and incubated with increasing agonist concentrations. Cyclic AMP production was measured by radioimmunoassay using polyclonal antisera to iodinated cAMP. The data shown are representative of two dose-response experiments. B, basal (Bas) and forskolin (FSK)-stimulated cAMP production in untreated and PTX-pretreated membranes. The data shown are the average of four experiments.
FIG. 2. G protein ␣ subunit determination. Plasma membranes from 3T3-F442A adipocytes and rat epididymal white adipose tissue (EWAT) were resolved by SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. The complement of G protein ␣ subunits present was assessed by Western blotting as described under "Materials and Methods." Std, G protein standard. favored pathway in response to activation of the ␤ 3 AR in adipocytes. It must be remembered, however, that the catalytic activity of the subsequent kinases and their juxtaposition to substrate targets will ultimately determine the relative importance of these two pathways. Future studies will address these issues.
To determine whether the dual G s /G i coupling of the ␤ 3 AR and activation of MAP kinase was an adipocyte-specific event or a unique feature of the rodent ␤ 3 AR, we attempted to reconstruct these effects by transiently transfecting the human ␤ 3 AR into HEK-293 cells. As shown in Fig. 6A, transient transfection of the h␤ 3 AR resulted in a low basal level of ERK1/2 phosphorylation (lane 2), and stimulation with the ␤ 3 AR agonist increased ERK1/2 phosphorylation ϳ4-fold above basal level (lane 3), as compared with the 7-8-fold activation achieved by EGF (lane 7). There was no effect of the ␤ 3 AR agonist on mock-transfected cells (lane 1). This activation of MAP kinase by the h␤ 3 AR was completely blocked by the selective ␤ 3 AR antagonist, SR59230A (lane 4) (51). In addition, the nonselective ␤AR agonist isoproterenol at a concentration capable of activating the ␤ 3 AR (100 M), in the presence of propranolol, also led to an ϳ4-fold stimulation of ERK1/2 phosphorylation (lane 5). Finally, PTX completely eliminated MAP kinase activation (lane 6), as observed in 3T3-F442A adipocytes (Fig. 4). Together these data suggest that the dual coupling of the ␤ 3 AR to G s /G i , along with its activation of MAP kinase, is a general property of this receptor.
Mechanistically, ␤ 2 AR-mediated MAP kinase activation in fibroblasts requires sequential receptor coupling to G s and G i , because cAMP-dependent, PKA-mediated phosphorylation of the ␤ 2 AR is a prerequisite for receptor-G i coupling (15). To compare the role of cAMP and PKA in ␤AR-mediated MAP kinase activation in adipocytes, we compared the effects of the PKA inhibitor H-89 on ␤ 2 AR-and ␤ 3 AR-stimulated MAP kinase activation. As shown in Fig. 6B, H-89 blocked ␤ 2 ARstimulated MAP kinase activation (lane 8), consistent with the results of Daaka et al. (15). In contrast, the ␤ 3 AR signal was insensitive to H-89 (lane 3). Fig. 6B also shows that MAP kinase activation by the ␤ 3 AR was sensitive to low micromolar concentrations of the tyrosine kinase inhibitor genistein (lane 4) and to the EGR receptor tyrphostin AG1478 (lane 5). These results indicate that, like other G␣ i -coupled receptors, the ␤ 3 AR employs a tyrosine kinase receptor in its mechanism of recruiting MAP kinase (52,53). In other experiments, treatment with dibutyryl cAMP enhanced ␤ 2 AR-stimulated ERK1/2 phosphorylation by ϳ70%, consistent with the model of PKAdependent ␤ 2 AR-G i coupling. However, stimulation of MAP kinase by the ␤ 3 AR was unaffected by dibutyryl cAMP (data not shown). Together, these results indicate that MAP kinase activation via the ␤ 3 AR, unlike the ␤ 2 AR, is insensitive to modulation of the PKA pathway. Thus ␤ 3 AR-mediated MAP kinase activation can be distinguished in several ways from the ␤ 2 AR pathway. First, the ␤ 3 AR is not a substrate for PKA (24). Second, the ␤ 3 AR appears to be constitutively coupled to both G s and G i . Third, ␤ 3 AR-stimulated activation of cAMP is not required for activation of the MAP kinase cascade in adipocytes. In contrast to the finding of PKA-dependent ␤ 2 AR-mediated MAP kinase activation in S49 lymphoma cells (16), the ␤ 3 AR pathway in adipocytes is completely PTX-sensitive.
One of the remarkable features of the ␤ 3 AR is that treatment with ␤ 3 AR-selective agonists in vivo can prevent or reverse obesity due to either congenital or diet-induced etiology (6, 19 -22, 54). These agents are also efficacious over prolonged periods of administration (21), which is quite distinct from the rapid desensitization and down-regulation that is characteristic of chronic treatment with ␤ 1 -and, particularly, ␤ 2 AR-selective drugs (55,56). These unique properties of the ␤ 3 AR have been suggested to be related to the fact that ␤ 3 AR is resistant to desensitization (57), because it is not a substrate for either PKA or the G protein-coupled receptor kinases (24). Instead, our finding that the ␤ 3 AR is capable of activating both the PKA pathway and the MAP kinase cascade raises another possibility. The simultaneous recruitment of both signaling networks may result in a more potent stimulation of lipolysis and/or may promote the growth and differentiation of brown adipocytes, which is observed in all ␤ 3 AR agonist-treated animals (20,21,58). It will now be important for us to test the consequence of impairing one or the other of these signal transduction pathways to determine their contribution to these ␤ 3 AR-mediated responses.