Insulin Activation of Mitogen-activated Protein Kinases Erk1,2 Is Amplified via β-adrenergic Receptor Expression and Requires the Integrity of the Tyr350 of the Receptor*

Insulin activates a complex set of intracellular responses, including the activation of mitogen-activated protein kinases Erk1,2. The counterregulatory actions of insulin on catecholamine action are well known and include phosphorylation of the β2-adrenergic receptor on Tyr350, Tyr354, and Tyr364 in the C-terminal cytoplasmic domain, as well as enhanced sequestration of the β2-adrenergic receptor. Both β-adrenergic agonists and insulin provoke sequestration of β2-adrenergic receptors in a synergistic manner. In the current work, cross-talk between insulin action and β2-adrenergic receptors revealed that insulin activation of Erk1,2 was amplified via β2-adrenergic receptors. In Chinese hamster ovary cells, expression of β2-adrenergic receptors enhanced 5–10-fold the activation of Erk1,2 by insulin and prolonged the activation, the greatest enhancement occurring at 5 min post-insulin. The potentiation of insulin signaling on Erk1,2 was proportional to the level of expression of β2-adrenergic receptor. The potentiation of insulin signaling requires the integrity of Tyr350 of the β2-adrenergic receptor, a residue phosphorylated in response to insulin. β2-adrenergic receptors with a Y350F mutation failed to potentiate insulin activation of Erk1,2. Expression of the C-terminal domain of the β2-adrenergic receptor (Pro323-Leu418) in cells expressing the intact β2-adrenergic receptor acts as a dominant negative, blocking the potentiation of insulin activation of Erk1,2 via the β2-adrenergic receptor. Blockade of β2-adrenergic receptor sequestration does not alter the ability of the β2-adrenergic receptor to potentiate insulin action on Erk1,2. We propose a new paradigm in which a G-protein-linked receptor, such as the β2-adrenergic receptor, itself acts as a receptor-based scaffold via its binding site for Src homology 2 domains, facilitating signaling of the mitogen-activated protein kinase pathway by insulin.

G-protein-linked receptors (GPLRs) 1 and growth factor receptors with intrinsic tyrosine kinase activity (TKR) represent two prominent pathways for cellular signaling (1,2). Study of the integration of signaling between GPLR and TKR pathways has recently revealed the existence of cross-talk at the most proximal point, receptor to receptor interaction with GPLRs acting as substrates for TKRs (3)(4)(5). Insulin stimulates the phosphorylation of the ␤ 2 -adrenergic receptor (␤2AR) on tyrosyl residues Tyr 350 / 354 and Tyr 364 , both in vivo (3,4) and in vitro (5) using recombinant, purified ␤2AR and insulin receptors. Tyrosyl residue 350, a prominent residue for insulin receptor-catalyzed phosphorylation, is embedded in a sequence motif (Tyr-Gly-Asn-Gly) that is similar to the motifs known to interact with Caenorhabditis elegans sem5 Src homology 2 (SH2) domains when phosphorylated (6). Phosphorylation of sites on the ␤2AR by the insulin receptor and the insulin-like growth factor-1 receptor include a motif for TKR at Tyr 364 (7), the Grb2 binding site at Tyr 350 (4,8), and a potential Shc binding site at Tyr 132 (4,5,7). For insulin action, activation of 1-phosphatidylinositol 3-kinase (PI3K) is an early event, following temporally the phosphorylation of the insulin receptor and insulin receptor substrate-1 (9,10). In response to insulin stimulation, the p85 regulatory subunit of PI3K binds the IRS-1 via SH2 domain(s), activating the catalytic p110 subunit, which phosphorylates various phosphoinositides at the 3Ј position of the inositol ring (11). Ample reports support the premise that PI3K and its 3Ј-phosphoinositide products are critical to intracellular trafficking of membrane-bound elements in general (12) and of downstream elements of TKR signaling, particularly insulin (13). We have shown that insulin, much like ␤-adrenergic agonists, provoke rapid sequestration of ␤2AR in a synergistic manner (14,15). We probed for possible cross-talk between insulin and ␤2ARs in the mitogen-activated protein kinase pathway. Remarkably, the activation of Erk1,2 by insulin in Chinese hamster ovary (CHO) cells was found to be amplified by ␤2AR expression, i.e. the higher the level of cellular complement of ␤2AR, the greater was the potentiation of insulin activation of Erk1,2. Further studies reveal that the ability of the ␤2AR to amplify the insulin response was dependent on the integrity of the Tyr 350 residue, which is phosphorylated by the insulin receptor and constitutes a binding site for an SH2 domain to which Grb2, and other proteins, can bind (16). A new paradigm is proposed in which G-protein-linked receptors function as a receptor-based scaffold via a binding site for SH2 domains, amplifying signaling via the mitogenactivated protein kinase pathway.
Assay of Activation of Erk1,2-Cells were stimulated with epidermal growth factor (EGF) (50 ng/ml) or insulin (100 nM) for the indicated times and then lysed (150 mM NaCl, 5 mM EDTA, 50 mM NaF, 40 mM sodium pyrophosphate, 50 mM KH 2 PO 4 , 10 mM sodium molybdate, 2 mM sodium orthovanadate, 20 mM Tris-HCl, pH 7.4, 1% Triton X-100, 0.5% Nonidet P-40, 6 mM dithiothreitol, 10 g/ml aprotinin, 10 g/ml leupeptin, and 0.2 mM phenylmethylsulfonyl fluoride). Samples were subjected to SDS polyacrylamide gel electrophoresis and immunoblotting as described earlier (21). The blots were stained with antibodies purchased from Promega (Madison, WI) that specifically recognize only the dually phosphorylated, active forms of Erk1,2. For inhibitor studies, A431 cells and CHO clones stably transfected to express ␤ 2 -adrenergic receptors were pretreated for 12 h in the absence or presence of one of the following agents: the PI3K inhibitor LY294002 (LY; 20 M), the MEK inhibitor PD98059 (10 M), or the Src inhibitor PP2 (50 nM). After an overnight exposure to an inhibitor, the cells were incubated with or without 100 nM insulin for 5 min, and the amount of Erk1,2 activated was measured.
Radioligand Binding Studies-The number of ␤2AR was determined by radioligand binding. Intact A431 cells were incubated with 0.5 nM [ 125 I]iodocyanopindolol (PerkinElmer Life Sciences) in the presence or absence of 10 M propranolol at 23°C for 90 min. The incubation buffer contained 50 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , and 150 mM NaCl. The cells were collected on GF/C membranes at reduced pressure and washed rapidly. The radioligand bound to the washed cell mass retained by the filter was quantified by use of a ␥-counter (17).
Sequestration of ␤ 2 -Adrenergic Receptor-Receptor sequestration was assayed using the hydrophilic, membrane-impermeable ␤-adrenergic antagonist [ 3 H]CGP-12177 (18). A431 cells were preincubated with isoproterenol (10 M) for periods up to 60 min or preincubated with insulin (100 nM) for 5 to 30 min. The cells were then resuspended in Dulbecco's modified Eagle's medium containing 20 mM HEPES, pH 7.4, and 70 nM [ 3 H]CGP-12177 (PerkinElmer Life Sciences) at 4°C for 6 h. The cells were diluted, collected on GF/C membranes, and washed rapidly. The radioligand bound to the washed cell mass retained on the filter was counted by use of liquid scintillation spectrometry. Nonspecific binding was defined as the radioligand binding insensitive to competition by the unlabeled, ␤-adrenergic antagonist propranolol (10 M).
Confocal Microscopy-A431 cells stably transfected with green fluorescence protein (GFP)-tagged ␤ 2 -adrenergic receptor were treated with or without insulin in the absence or presence (overnight) of the PI3K inhibitor LY294002 (20 M). The cells were incubated with or without 100 nM insulin for 5 min, fixed with 3% paraformaldehyde, and washed 3 times with modified Shields' medium (MSM)-Pipes buffer (18 mM MgSO 4 , 5 mM CaCl 2 , 40 mM KCl, 24 mM NaCl, and 5 mM Pipes, pH 6.8). The cells were analyzed by confocal laser scanning microscopy on an Odyssey instrument (Noran Instruments). The construct pCDNA3-␤ 2 AR-GFP, encoding a ␤ 2 -adrenergic receptor fusion protein with GFP at its carboxyl terminus (22), was a generous gift from Dr. J. L. Benovic. The confocal microscopy was performed at the University Microscopy Imaging Center at Stony Brook.
Data Presentation and Analysis-Unless otherwise noted, the values presented are means Ϯ S.E. The autoradiograms are representative of multiple (at least three) independent experiments. In all figures, an asterisk denotes a mean value with statistical significance (p Յ 0.05) compared with the mean values of the control, time 0, or with the control group as indicated in the legend to the figure.

RESULTS
Initial studies of the time-course for activation of Erk1,2 revealed a biphasic response to insulin stimulation of CHO clones. In CHOK cells, which express very low numbers of endogenous ␤2ARs, the activation of Erk1,2 displayed an initial peak of activation (as measured with phosphospecific antibodies for activated Erk1,2) in response to insulin at 0.5-1.0 min (Fig. 1, inset), followed by a gradual decline in activation observed in analyses of multiple experiments performed on separate occasions (Fig. 1, bar graph). The activation of Erk2 (p42Erk) in response to insulin exceeded that of Erk1 (p44ErK). When examined in CHO cells expressing a modest level of ␤2ARs (18.2 fmol/100,000 cells), the activation of Erk1,2 in response to insulin shows both a biphasic response, as well as an amplified response to stimulation by insulin. The   FIG. 2. The activation of Erk1,2 in response to insulin is increased with increasing expression of ␤2AR. CHO clones were stably transfected and selected to express different levels of ␤2AR, indicated as wild-type (WT, with very low expression of ␤2AR), low (␤L), middle (␤M), and high (␤H) expressing clones. The activation of Erk1,2 was determined, as indicated in the legend to Fig. 1, in clones incubated in the absence (-) or presence of either 100 nM insulin (ϩInsulin) or 50 ng/ml epidermal growth factor (ϩEGF) for 5 min. The results of three independent assays in which the amount of the phosphoactivated forms of p44 and p42 Erk were quantified are summarized in the bar graphs.
FIG. 1. The activation of Erk1,2 in response to insulin is potentiated by ␤2AR. CHOK cells and CHO clones stably transfected to express the ␤2AR (CHO-␤2AR) were challenged with insulin (100 nM) for the time (min) indicated, and the activation of Erk1,2 was determined using antibodies specific for the dually phosphorylated, active forms of p44 and p42 Erk (phospho-Erk). The results of three independent assays in which the amount of the phosphoactivated forms of p44 and p42 Erk were quantified are summarized in the bar graphs.
Erk1,2 response to insulin showed an early (0.5-1.0 min) peak of activation followed by a robust later response, peaking at 5 min, declining to unstimulated levels within 10 min (not shown). The response for activation of p42Erk in the CHO-␤2AR clones was 5-7-fold greater than that in the ␤2AR-deficient CHOK clones. The response for activation of p44Erk by insulin was similar to that observed for p42Erk, i.e. 5-7-fold greater in the CHO-␤2AR clones as compared with the CHOK clones. The simplest interpretation of the data is that the expression of ␤2AR potentiates and prolongs the ability of insulin to activate Erk1,2, a novel hypothesis.
To test the hypothesis that the expressed level of the Gprotein-linked receptor ␤2AR regulates the temporal nature and magnitude of the Erk1,2 activation by insulin, we examined the Erk1,2 response to activation by insulin in cells that stably express varying levels of ␤2AR (Fig. 2). Four clones were included in the analysis that displayed few receptors with CHOK (wild-type; 0.6 Ϯ 0.08 fmol/100,000 cells), low (␤L; 4.5 Ϯ 1.1 fmol/100,000 cells), middle (␤M; 18.3 Ϯ 2.3 fmol/100,000 cells), and high (␤H; 27.9 Ϯ 3.0 fmol/100,000 cells) levels of expressed ␤2AR ([ 125 I]iodocyanopindolol binding; mean values Ϯ S.E., n ϭ 3). The activation of Erk1,2 was measured in response to insulin challenge after 5 min. Immunoblots of the cell lysates were stained with antibodies to the phosphospecific, activated forms of Erk1 and Erk2. In the absence of insulin, levels of activated Erk1,2 were very low. Often, as shown here, the basal levels of phosphorylated, activated Erk1,2 were suppressed in the clones expressing ␤2AR. The greater the expression of ␤2AR, the greater was the apparent suppression of basal levels of Erk1,2 activation. Challenge with EGF provokes an activation of Erk1,2 in wild-type cells with low levels of ␤2AR, whereas the EGF response is attenuated in the clones expressing significant levels of ␤2ARs. The Erk1,2 activation in response to insulin, in sharp contrast, is clearly amplified in cells expressing ␤2ARs. Although not strictly proportional, the activation of Erk1,2 by insulin, but not EGF, was amplified to a greater extent with increasing levels of ␤2AR expression. These data suggest that the ␤2AR appears to facilitate insulin signaling through the mitogen-activated protein kinase cascade to the level of Erk1,2.
It was important to evaluate whether co-stimulation of the insulin-and ␤2AR-stimulated pathways leads to activation of Erk1,2 and whether some adrenergic agonist activity derived from serum might contribute to the insulin response. Treating A431 human epidermoid carcinoma cells with isoproterenol (10 M) or insulin (100 nM) stimulated the activation of both Erk1 and Erk2 (Fig. 3). Insulin and isoproterenol, in combination, do not produce an additive response. Propranolol (10 M) blocked isoproterenol-stimulated activation of Erk1,2 but not the response to stimulation by insulin. These data rule out the possibility that co-stimulation by some serum-derived agent via a ␤2AR-mediated pathway is an explanation for the enhanced insulin-stimulated activation of Erk1,2 in cells expressing ␤2AR. The diterpene activator of adenylyl cyclase forskolin stimulates cyclic AMP accumulation and also provoked a weak activation of p44Erk and an activation of p42Erk similar to that of isoproterenol. These same results were obtained in CHO clones stably transfected to express ␤2ARs (data not shown). The inverse agonist compound ICI118551 was without effect on Erk1,2 activation in these cells (data not shown).
We extended these studies by examining the ability of insu- Aliquots of untreated A431 cells, as well as cells treated with either isoproterenol or with insulin, were also treated with the beta-adrenergic antagonist propranolol (10 M) to block beta-adrenergic stimulation by isoproterenol and any agonist-like activity derived from serum. To evaluate the activation of Erk1,2 in response to elevation of intracellular cyclic AMP, cells were treated with the diterpene forskolin (30 M) for 5 min. The activation of Erk1,2 was determined using antibodies specific for the dually phosphorylated, active forms of p44 and p42 Erk (phospho-Erk). The results of three independent assays in which the amount of the phosphoactivated forms of p44 and p42 Erk were quantified are summarized in the bar graphs.
FIG. 4. Activation of Erk1,2 in A431 cells: dose-response to insulin and to isoproterenol, in combination. A431 cells were treated in the presence or absence of isoproterenol (0, 1, or 10 M) and insulin (0, 1, 10, or 100 nM) or both for 5 min. The activation of Erk1,2 was determined using antibodies specific for the dually phosphorylated, active forms of p44 and p42 Erk (phospho-Erk). The results of three independent assays in which the amount of the phosphoactivated forms of p44 and p42 Erk were quantified are summarized in the bar graphs.
lin to activate Erk1,2 in A431 cells that were challenged simultaneously with either 1 or 10 M isoproterenol (Fig. 4). The response to insulin alone is provided for comparison. Treatment with 1 M isoproterenol alone provokes activation of Erk1,2. Increasing the concentration of insulin leads to further activation of Erk1,2. Increasing the concentration of isoproterenol alone to 10 M leads to additional activation of that produced by 1 M. Challenging the A431 cells with 10 M isoproterenol in combination with increasing concentrations of insulin leads to a dampening of the activation of Erk1,2 especially at the lower concentrations of insulin (1 and 10 nM). These data suggest that the ability of isoproterenol to activate Erk1,2 is intrinsically lower than that of insulin to activate Erk1,2. In combination, insulin and isoproterenol do not activate Erk1,2 in an additive manner but rather appear to compete for the activation process in a weakly competitive manner.
The observations both in vivo (4,23) and in vitro (5) demonstrate that insulin treatment leads to the phosphorylation of the ␤2AR on tyrosyl residues, particularly Tyr 350 . Phosphorylation of Tyr 350 creates a binding site for SH2 domains that can act as a protein module for the docking and regulation of a wide spectrum of adaptor molecules, as well as interesting molecules such as Src and PI3K (6). The ability of insulin to counterregulate the ability of the ␤2AR to activate adenyl cyclase is dependent upon the availability of the Tyr 350 for phosphorylation. Mutation of this tyrosyl residue abolishes the counterregulation of ␤2AR-stimulated cyclic AMP accumulation (4,24). Similarly, the ability of insulin to provoke profound sequestration of the ␤2AR is dependent upon the availability of Tyr 350 , i.e. the Y350F mutant fails to display insulin-induced sequestration (14). We tested whether mutation of the wild-type ␤2AR to Y350F, likewise, might influence the ability of insulin to signal to the level of Erk1,2 activation (Fig. 5). As displayed in the immunoblot of whole-cell extracts stained with antibodies to the phosphospecific, activated forms of Erk1,2, insulin stimulates an enhanced activation of Erk1,2 in CHO cells expressing the wild-type (Tyr 350 ) ␤2AR. To address the nature of the effect of expressing a mutant receptor on the insulin response, we selected clones that express nearly equivalent levels of ␤2AR (3.5 Ϯ 0.7 and 3.0 Ϯ 0.6 fmol/100,000 cells for mutant Y350F and wild-type ␤2AR-expressing cells, respectively). The effect of the Y350F mutation was clear; the activation of Erk1,2 by insulin in the cells expressing the mutant receptors was largely abolished. Analysis of data from multiple experiments confirms the data displayed in the immunoblot. Mutation of Tyr 350 of ␤2AR blocks the ability of insulin to counterregulate ␤2ARmediated activation of adenylyl cyclase (3,4) and to stimulate sequestration of the ␤2AR (14), as well as the ability of the insulin receptor to phosphorylate Tyr 350 both in vivo (4) and in vitro (5). Taken together, these results further implicate the ␤2AR acting as a receptor-based scaffold for insulin signaling in which phosphorylation of the Tyr 350 residue of the ␤2AR creates a binding site for SH2 domains in response to insulin, a key element to activation of Erk1,2.
The Tyr 350 residue is located in the C-terminal, cytoplasmic tail of the ␤2AR. If the ␤2AR is acting as a template or scaffold for insulin signaling to the level of Erk1,2, then expression of the C-terminal domain of the receptor that is not localized to the plasma membrane may influence this novel ability of the ␤2AR to potentiate insulin signaling. Human epidermoid A431 cells were employed because of their relatively high native expression of ␤2AR. A431 clones were stably transfected with FIG. 5. The activation of Erk1,2 in response to insulin is potentiated by ␤2AR and dependent upon the integrity of ␤2AR tyrosyl residue 350. CHOK cells (CHOK) and clones expressing either the wild-type ␤2AR (Y350) or the tyrosine to phenylalanine (Y350F) mutant form of the ␤2AR were challenged with 100 nM insulin for 5 min, and the amount of activated Erk1,2 was determined. The results of three independent assays in which the amount of the phosphoactivated forms of p44 and p42 Erk were quantified are summarized in the bar graphs.

FIG. 6. The activation of Erk1,2 by insulin in epidermoid carcinoma A431 cells is blocked by expression of the C-terminal, cytoplasmic domain (BAC1) of the ␤2AR. Panel A, A431 cells were
stably transfected with an expression vector harboring the C-terminal cytoplasmic domain (BAC1) of the hamster ␤2AR or the empty vector (-). Staining immunoblots of whole-cell extracts of A431 clones with antibodies against the C-terminal cytoplasmic domain of the ␤2AR reveals the expressed 11-kDa molecular mass form of BAC1. Clones were challenged with 100 nM insulin for 5 min, and the phosphotyrosylcontaining species were identified in immunoblots of whole-cell extracts using an anti-phosphotyrosine antibody (Anti-pY). Note that the phosphotyrosine content of BAC1 is increased in A431 cells expressing BAC1 that have been challenged with insulin. Panel B, clones were challenged with 100 nM insulin for 5 min, and the amount of the activated Erk1,2 was determined. Note that expression of BAC1 resulted in a block of the ability of insulin to activate Erk1,2 in these cells.
an expression vector (pCDNA3) harboring the entire cytoplasmic C-terminal domain of the ␤2AR (BAC1). Expression was confirmed for the A431 cells stably transfected to express BAC1 by accumulation of the peptide (M r ϭ 11,000) in sufficient quantities to permit ready detection in immunoblots of the whole-cell extracts from these cells, probed with an antibody to BAC1 (Fig. 6A). Insulin challenge of A431 cells yields a significant activation of Erk1,2 (Fig. 6B). The expression of BAC1 in A431 clones resulted in a total loss of the ability of insulin to signal to the level of Erk1,2. Probing immunoblots of extracts from the A431 clones with antibodies specific for phosphotyrosine (Anti-pY) revealed, indeed, that BAC1 was tyrosinephosphorylated in response to insulin (Fig. 5A). Expression of BAC1 had no effect on the ability of isoproterenol (10 M) to stimulate a normal elevation of intracellular cyclic AMP accumulation by A431 cells (data not shown). Several slower migrating species that are recognized by the BAC1 antibody and are phosphorylated in response to insulin are likely to represent palmitoylated forms of BAC1 (25). Thus, with respect to the activation of Erk1,2 by insulin in cells expressing native ␤2AR, BAC1 acts as a dominant-negative molecule that is phosphorylated in response to insulin stimulation and suppresses the ability of insulin to signal.
PI3K plays an important role in many cellular process, including intracellular trafficking of molecules (26,27). The microbial product wortmannin and the LY294002 compound both inhibit PI3K and many facets of intracellular trafficking. Inhibitors of P13K block various aspects of insulin action, and we evaluated whether the LY294002 compound would influence the ability of the ␤2AR to enhance insulin activation of Erk1,2 and perhaps influence the sequestration of ␤2ARs (Fig. 7). Insulin, as well as isoproterenol, stimulates the sequestration of ␤2AR (14). The cellular localization of the ␤2AR was defined by use of a GFP-tagged version of the ␤2AR, characterized earlier (18,22). Using A431 cells stably transfected with an expression vector harboring the GFP-tagged ␤2AR, in tandem with confocal laser scanning microscopy, we observed the GFPtagged ␤2AR display a pattern of distribution largely confined to the cell membrane (Fig. 7, panel a). Challenge with insulin induces a dramatic sequestration of the ␤2AR from the cell membrane to intracellular locales (Fig. 7, panel b), as previously observed (14). In the presence of the LY294002 inhibitor (20 M), in contrast, the ability of insulin to induce sequestration of the ␤2AR was abolished (Fig. 7, panel c). Analysis of ␤2AR internalization using the hydrophilic, beta-adrenergic antagonist ligand [ 3 H]CGP-12177, confirms in an independent manner that inhibition of PI3K with the LY compound suppresses insulin-stimulated ␤2AR sequestration, measured at 5 min post-insulin (Table I). Taken together, these data suggest that the ability of insulin, but not the ability of isoproterenol, to stimulate sequestration of the ␤2AR was dependent upon PI3K, i.e. inhibition of PI3K by LY compound suppresses sequestration of ␤2AR in response to insulin. These novel data prompted us to probe further the possible role for ␤2AR inter-

FIG. 8. The activation of Erk1,2 by insulin in epidermoid carcinoma A431 cells is blocked by inhibition of MEK with PD98059
but is not altered by the blockade of ␤2AR sequestration by the PI3K inhibitor LY294002. CHOK cells and CHO clones stably transfected to express the ␤2AR (CHO-␤2AR) were challenged with insulin (100 nM) for 5 min, and the activation of Erk1,2 was determined using antibodies specific for the dually phosphorylated, active forms of p44 and p42 Erk (phospho-Erk). 12 h prior to the challenge with insulin, the clones were pretreated with or without one of the following three agents: 10 M PD98059, 20 M LY294002, or 50 nM PP2. The results shown in the bar graphs are representative of three independent assays.  ␤2AR in response to insulin but not in response to isoproterenol A431 cells were treated in the absence (Control) and presence of either 10 M isoproterenol or 100 nM insulin. The extent of ␤2AR sequestration was determined using the hydrophilic, cell-impermeable beta-adrenergic antagonist radioligand [ 3 H]CGP-12177. The incubation with isoproterenol was for 30 min, whereas the incubation with insulin was measured at 5 min, as employed in the experiments shown in Fig.  6. The isoproterenol-stimulated sequestration was blocked by addition of the beta-adrenergic antagonist propranolol (10 M). Aliquots of cells were pretreated with the PI3K inhibitor LY294002 (20 M) and then stimulated with either isoproterenol or insulin. The LY compound suppresses insulin-stimulated sequestration of the ␤2AR. The data are the mean values Ϯ S.E. of three replicate experiments. Total binding was 29.4 Ϯ 3.2 fmol/100,000 cells.

Treatment
Sequestration of ␤2AR To probe further the potentiation of insulin signaling to Erk1,2 by ␤2AR, we tested inhibitors of Src, MEK, and PI3K. The studies were performed in both A431 cells and CHO clones stably transfected to express ␤2AR. Recent data have implicated the non-receptor tyrosine kinase Src in the biology of ␤2AR (28); for this reason the PP2 inhibitor of Src was tested first. Treatment with the PP2 inhibitor (50 nM) failed to influence the ability of cells expressing ␤2AR to potentiate insulin activation of Erk1,2 (Fig. 8). The MEK inhibitor PD98059 (29) (10 M, overnight), in sharp contrast, abolishes all activation of Erk1,2, suggesting that the potentiation of insulin-stimulated Erk1,2 activation by ␤2AR is, in fact, mediated via MEK.
The LY294002 compound, which inhibits 1-phosphatidylinositol 3-kinase, was without effect on the ability of ␤2AR to potentiate Erk1,2 activation by insulin (Fig. 8), although LY294002 was found to suppress insulin-stimulated ␤2AR sequestration (Fig. 7). We wondered whether inhibition of PI3K activity and blockade of ␤2AR sequestration might alter either the timecourse or decay of the activation of Erk1,2 by insulin. A431 cells pretreated with the LY compound for 2 h prior to challenge with insulin display the same onset, activation, and decay of the response as cells not treated with LY294002 (Fig. 9).
The results suggest that insulin and isoproterenol can both activate Erk1,2 independently of each other but that their activation in combination is non-additive. To test further the notion that these two agents may compete on one level with each other for the activation of Erk1,2, we investigated the effects of pretreating A431 cells with isoproterenol for 30 min on the ability of a subsequent challenge of insulin to activate Erk1,2. Cells were challenged with isoproterenol for 30 min and then challenged directly without or with increasing concentrations of insulin (Fig. 10). When challenged sequentially, the ability of isoproterenol to counterregulate the activation of Erk1,2 by insulin was revealed. The activation of Erk1,2 by insulin was diminished in cells challenged 30 min prior with isoproterenol. When challenged simultaneously, this ability of isoproterenol to counterregulate insulin action was not so obvious (Figs. 3 and 4). DISCUSSION The counterregulatory effects of catecholamines on insulin action are well known. We reveal the activation of Erk1,2 to be the target of both insulin and beta-adrenergic agonists, with beta-adrenergic agonists and insulin, at one level, competing for ␤2AR (Fig. 11). Because the extent of Erk1,2 activation is less for beta-adrenergic agonists than for insulin, this competition dictates the final level of Erk1,2 activity, which in turn can modulate other members of the mitogen-activated protein kinase network. On another level, the current work identifies a novel role for the ␤2AR and perhaps other GPLRs, i.e. acting as a receptor-based scaffold enhancing the signaling of other pathways. Expression of the ␤2AR clearly can potentiate the ability of insulin to activate Erk1,2. The evidence to support this notion is as follows: activation of Erk1,2 by insulin is potentiated severalfold and prolonged in cells expressing elevated levels of ␤2AR; the extent of the potentiation of insulin action on Erk1,2 correlates with the amount of ␤2AR expressed; mutation of the ␤2AR tyrosyl residue (Y350F) that is both phosphorylated in response to insulin and creates a binding site for SH2 domains abolishes the potentiation of insulin action; the C-terminal, cytoplasmic domain of the ␤2AR (BAC1), when expressed in cells, is phosphorylated in response to insulin and acts as a dominant negative with respect to enhanced activation of Erk1,2 by insulin; and, blockade of ␤2AR internalization does not suppress the ability of the ␤2AR to potentiate insulin FIG. 9. The time-course and decay of activation of Erk1,2 by insulin in epidermoid carcinoma A431 cells is not altered by the blockade of ␤2AR sequestration by the PI3K inhibitor LY294002. CHOK cells and CHO clones stably transfected to express the ␤2AR (CHO-␤2AR) were challenged with insulin (100 nM) for 0, 3, 5, or 10 min, and the activation of Erk1,2 was determined using antibodies specific for the dually phosphorylated, active forms of p44 and p42 Erk (phospho-Erk). 2 h prior to the challenge with insulin, the clones were pretreated with or without 20 M LY294002. The results shown in the bar graphs are representative of three independent assays.
FIG. 10. Activation of Erk1,2 in A431 cells: analysis of insulin stimulation following a prior challenge with isoproterenol. A431 cells were challenged with either isoproterenol (0 or 10 M) alone or insulin (0, 1, 10, or 100 nM) following a pretreatment with 10 M isoproterenol. The activation of Erk1,2 was determined using antibodies specific for the dually phosphorylated, active forms of p44 and p42 Erk (phospho-Erk). The results of three independent assays in which the amount of the phosphoactivated forms of p44 and p42 Erk were quantified are summarized in the bar graphs. activation of Erk1,2. Several well known GPLRs themselves activate the Erk1,2 pathway (2). In some cases, the activation of Erk1,2 requires internalization of the GPLR, whereas in others the activation of Erk1,2 proceeds in the absence of internalization (30,31). The ability of the ␤2AR to cross-talk to and potentiate insulin action on Erk1,2 also does not require large-scale ␤2AR sequestration.
GPLR-based scaffold functions have been implicated for direct effects of GPLR-agonists action on the mitogen-activated protein kinase network (28). We propose that the potentiation of insulin activation of Erk1,2 is yet another example of a GPLR-based scaffold. The ␤2AR, upon phosphorylation in response to insulin (3-5, 23, 24), possess a binding site for SH2 domains (5) that can interact with a variety of adaptor molecules and enzymes, including PI3K and dynamin that are involved in ␤2AR sequestration (16). The sequestration of ␤2AR in response to agonist requires G-protein-linked receptor kinase-catalyzed phosphorylation, arrestin binding, and the involvement of clathrin, dynamin, and Src. The activation of Erk1,2 by insulin as compared with isoproterenol shares many common features (Fig. 11) and some interesting differences. Insulin, as well as beta-adrenergic agonists, stimulate phosphorylation of the ␤2AR, the former by intrinsic tyrosine kinase activity of the insulin receptor and the latter by G-proteinlinked receptor kinases. Both beta-adrenergic agonists and insulin can sequester ␤2AR into a vesicle-associated form and activate Erk1,2 independently. The sequestration by insulin, but not that by isoproterenol, is sensitive to LY. Blockade of insulin-induced sequestration by LY294002, however, does not block activation of Erk1,2 by insulin. The extent of the activation of Erk1,2 is greater for stimulation by insulin than it is for beta-adrenergic agonists. Pretreatment with isoproterenol for 30 min leads to a diminished capacity of insulin to activate Erk1,2, suggesting that at one level the two pathways compete for ␤2AR.
Taking these observations into account, we speculate that the ␤2AR acts as a receptor-based scaffold for the MEK kinase, MEK, and/or Erk1,2 elements of the pathway, targeting one or more of these elements to the cell membrane. Although insulin clearly sequesters ␤2AR, the activation of Erk1,2 proceeds even when the internalization is blocked, suggesting that the tyrosine-phosphorylated ␤2AR may organize the elements leading to Erk1,2 activation. Scaffold proteins, such as the A-kinase anchoring protein 250 gravin, have been shown only recently to play an integral role in the signaling of GPLRs (16, 18 -20). Expression of the ␤2AR potentiates and prolongs the insulin signaling event, and mutation of the Tyr 350 residue abolishes the binding site for SH2 domains, as well as the potentiation. Insulin stimulates the tyrosine phosphorylation and inactivation of the ␤2AR with respect to activation of adenyl cyclase (and cyclic AMP accumulation), while simultaneously providing a scaffold via a binding site for SH2 domains on the ␤2AR that can facilitate a second wave of signaling to the activation of Erk1,2. This proposal complements the notion that GPLRs can act as scaffolds for transactivation of receptor tyrosine kinases (2, 28), only in this case the tyrosine kinase creates a GPLR-based scaffold via protein phosphorylation. FIG. 11. Activation of Erk1,2 in response to isoproterenol and to insulin. Insulin activates the insulin receptor tyrosine kinase (IRTK) that, in turn, phosphorylates Tyr 350 and Tyr 364 of the beta 2 -adrenergic receptor (␤AR). PI3K and dynamin (Dyn) associate with the ␤AR via Grb2 binding to the binding site for Src homology 2 domains created by phosphorylation of Tyr 350 . Insulin-stimulated sequestration of vesicle-associated (VA) ␤AR assists in counterregulation of ␤AR signaling to Gs and adenylylcyclase. Insulin-stimulated sequestration requires PI3K activity. The LY294002 inhibitor of PI3K (LY) blocks insulin-stimulated sequestration of ␤AR but does not alter ␤AR-mediated activation of the mitogenactivated protein kinase Erk1,2. These data demonstrate that the ␤AR acts as a receptor-based scaffold to facilitate Erk1,2 activation, even in the absence of sequestration. Overexpression of ␤AR amplifies the ability of insulin to activate Erk1,2. In contrast, the ␤-adrenergic agonist isoproterenol (ISO) activates the ␤AR, stimulates serine/threonine (Ser, Thr) phosphorylation of the receptor by a G-protein-linked receptor kinase (GRK). Phosphorylation/sequestration of the ␤AR leads to desensitization of signaling to Gs and adenylylcyclase, but later, activation of Erk1,2. ␤-Arrestin (Arr) binds to the phosphorylated, C-terminal domain of the ␤AR and facilitates sequestration of vesicle-associated ␤AR via clathrincoated pits. The non-receptor tyrosine kinase Src participates in the desensitization process. The inhibition of PI3K via LY294002 does not block isoproterenol-stimulated sequestration of ␤AR but does block the activation of Erk1,2. Thus, the ␤AR can be counterregulated by insulin or desensitized by ␤-adrenergic agonists, or can enhance the signaling of growth factors to activation of Erk1,2 as a receptor-based scaffold or participate more directly in the activation of Erk1,2.