Insulin-induced β-Arrestin1 Ser-412 Phosphorylation Is a Mechanism for Desensitization of ERK Activation by Gαi-coupled Receptors*

β-Arrestin1 is an adapter/scaffold for many G protein-coupled receptors during mitogen-activated protein kinase signaling. Phosphorylation of β-arrestin1 at position Ser-412 is a regulator of β-arrestin1 function, and in the present study, we showed that insulin led to a time- and dose-dependent increase in β-arrestin1 Ser-412 phosphorylation, which blocked isoproterenol- and lysophosphatidic acid-induced Ser-412 dephosphorylation and impaired ERK signaling by these G protein-coupled receptor ligands. Insulin treatment also led to accumulation of Ser-412-phosphorylated β-arrestin1 at the insulin-like growth factor 1 receptor and prevented insulin-like growth factor 1/Src association. Insulin-induced Ser-412 phosphorylation was partially dependent on ERK as treatment with the MEK inhibitor PD98059 inhibited the insulin effect (62% reduction, p = 0.03). Inhibition of phosphatidylinositol 3-kinase by wortmannin did not have a significant effect (9% reduction, p = 0.41). We also found that the protein phosphatase 2A (PP2A) was in a molecular complex with β-arrestin1 and that the PP2A inhibitor okadaic acid increased Ser-412 phosphorylation. Concomitant addition of insulin and okadaic acid did not produce an additive effect on Ser-412 phosphorylation, suggesting a common mechanism. Small t antigen specifically inhibited PP2A, and in HIRcB cells expressing small t antigen, β-arrestin1 Ser-412 phosphorylation was increased, and insulin had no further effect. Insulin treatment caused increased β-arrestin1 Ser-412 phosphorylation, which blocked mitogen-activated protein kinase signaling and internalization by β-arrestin1-dependent receptors with no effect on β-adrenergic receptor Gs-mediated cAMP production. These findings provide a new mechanism for insulin-induced desensitization of ERK activation by Gαi-coupled receptors.

␤-Arrestin1 is an adapter/scaffold for many G proteincoupled receptors during mitogen-activated protein kinase signaling. Phosphorylation of ␤-arrestin1 at position Ser-412 is a regulator of ␤-arrestin1 function, and in the present study, we showed that insulin led to a timeand dose-dependent increase in ␤-arrestin1 Ser-412 phosphorylation, which blocked isoproterenol-and lysophosphatidic acid-induced Ser-412 dephosphorylation and impaired ERK signaling by these G proteincoupled receptor ligands. Insulin treatment also led to accumulation of Ser-412-phosphorylated ␤-arrestin1 at the insulin-like growth factor 1 receptor and prevented insulin-like growth factor 1/Src association. Insulin-induced Ser-412 phosphorylation was partially dependent on ERK as treatment with the MEK inhibitor PD98059 inhibited the insulin effect (62% reduction, p ‫؍‬ 0.03). Inhibition of phosphatidylinositol 3-kinase by wortmannin did not have a significant effect (9% reduction, p ‫؍‬ 0.41). We also found that the protein phosphatase 2A (PP2A) was in a molecular complex with ␤-arrestin1 and that the PP2A inhibitor okadaic acid increased Ser-412 phosphorylation. Concomitant addition of insulin and okadaic acid did not produce an additive effect on Ser-412 phosphorylation, suggesting a common mechanism. Small t antigen specifically inhibited PP2A, and in HIRcB cells expressing small t antigen, ␤-arrestin1 Ser-412 phosphorylation was increased, and insulin had no further effect. Insulin treatment caused increased ␤-ar-restin1 Ser-412 phosphorylation, which blocked mitogen-activated protein kinase signaling and internalization by ␤-arrestin1-dependent receptors with no effect on ␤-adrenergic receptor G s -mediated cAMP production. These findings provide a new mechanism for insulin-induced desensitization of ERK activation by G␣ icoupled receptors.
Activation of adenylate cyclase and generation of cAMP following ligand binding to some GPCRs coupled to G␣ s is attenuated once ␤-arrestin1 binds to the activated GPCR and sterically uncouples the receptor from further interaction with G␣ s (5). Other GPCRs, however, utilize ␤-arrestin as a scaffold for assembly of a multiprotein signaling complex that leads to activation of ERK1/2 (12)(13)(14). In addition, ␤-arrestin interacts with clathrin, targeting activated receptors to clathrin-coated pits for endocytosis (6,15,16). Therefore, ␤-arrestin performs a dual role during intracellular signaling by GPCRs: desensitizing signals leading to adenylate cyclase activation while coordinating and enhancing signals leading to ERK1/2 activation.
As one would predict from the known functions of ␤-arrestin discussed above, a reduction of ␤-arrestin protein content using antisense oligonucleotide (17), knock-out gene technology (18), small interfering RNA (19) or by insulin treatment (11) is associated with both impaired desensitization of G␣ s signaling (G␣ s signal "supersensitization") and impaired ERK activation.
␤-Arrestin1 function during G protein-mediated ERK signaling is also regulated by phosphorylation of the carboxyl terminus Ser-412 residue. Dephosphorylation of Ser-412 is required for the isoproterenol-stimulated association between ␤-arres-tin1 and Src, a non-receptor tyrosine kinase involved in ␤1and ␤2-adrenergic receptor (AR) mitogenic signaling (12). In addition, Ser-412 dephosphorylation is also required for ␤-arres-tin1/clathrin association, which is essential for ␤2AR internalization (16). Mutants of ␤-arrestin1 that have a serine to aspartic acid substitution (S412D) and thus mimic the phosphorylated state at the 412 residue have reduced affinity for Src and for clathrin and function as dominant negative inhibitors of G protein-mediated MAP kinase signaling (16). The mechanism behind the regulation of ␤-arrestin1 Ser-412 phosphorylation/dephosphorylation has not been described. We have recently shown that insulin treatment leads to ␤-arres-tin1 degradation (10) as well as increased ␤-arrestin1 Ser-412 phosphorylation (11). In these studies, insulin-induced changes in ␤-arrestin1 were associated with increased ␤AR/G␣ s signaling (due to loss of G␣ s signal desensitization) and with impaired G␣ i -mediated MAP kinase activation (due to defective ␤-arrestin/Src and ␤-arrestin/clathrin interaction). However, the relative importance of insulin-induced protein degradation versus insulin-induced Ser-412 phosphorylation in these processes remains unclear.
In addition, the mechanism of regulation of Ser-412 phosphorylation is unknown. Previous studies have described a role for ERK during Ser-412 phosphorylation (16,20), while others have found the protein phosphatase 2A (PP2A) and PP2B in complexes that contain ␤-arrestin (21). In this regard, we have recently found that insulin-induced reduction of PP2A activity is involved in the desensitization of ERK signaling (22), but the role of PP2A in regulating ␤-arrestin1 Ser-412 phosphorylation has not been described.
In the current study, we present a new mechanism of insulin-induced desensitization of GPCR signaling. Thus, insulininduces ␤-arrestin1 Ser-412 phosphorylation, and this disrupts many of the known ␤-arrestin1 functions in GPCR signaling. In addition, we demonstrate a novel role for PP2A in this process.
Immunoprecipitation and Western Blotting-Cells were lysed, and protein concentrations were determined by the Bradford assay as described previously (9). For Western blotting, 25 g of cellular protein were dissolved in Laemmli buffer supplemented with dithiothreitol and boiled for 5 min. For immunoprecipitations, 500 g of total protein were diluted to 1 g/l in PBS to which 3 g of the appropriate antibody were added overnight at 4°C. Immune complexes were captured by adding 40 l of a 50% protein A/G bead mixture for an additional 1 h. Beads were then washed three times with PBS, dissolved in Laemmli buffer plus dithiothreitol, and boiled for 5 min. Proteins were separated by 7.5-10% SDS-PAGE and then transferred to polyvinylidene difluoride membranes. Membranes were blocked with 5% nonfat dry milk in Tris-buffered saline/Tween 20 and probed with the appropriate primary and secondary antibodies. Horseradish peroxidase-conjugated secondary antibody binding was detected by chemiluminescence.
Generation of Small t Antigen-expressing Cell Lines-We have previously described the generation of stable cell lines expressing small t antigen in our laboratory (22). Rat-1 fibroblasts overexpressing human insulin receptors (HIRcB cells) were maintained as described previously (23). A plasmid encoding small t antigen, pCMV5-small t (24), was a gift from Marc C. Mumby (University of Texas Southwestern Medical Center, Dallas, TX). cDNA from pCMV5-small t was digested with HindIII and BamHI and subcloned into pcDNA3.1/Hygro (Invitrogen), which contains the hygromycin B phosphotransferase gene. The resultant plasmid, pcDNA3.1/Hygro-small t, was transfected into HIRcB cells by using Superfect according to the manufacturer's instructions. Stable cell lines were selected in 400 g of hygromycin B/ml. Clonal cell lines were isolated by limiting dilution and then screened by immunoblotting with anti-small t antibody. Parental cells transfected with pcDNA3.1/ Hygro alone (Hyg cells) were used as a control.
Intracellular cAMP Measurement-Intracellular cAMP levels were determined using the cAMP enzyme immunoassay system from Amersham Biosciences as described previously (11) according to manufacturer's instructions.
Ligand Binding Assay-␤AR internalization was determined by ligand binding assay using CGP-12177, a hydrophilic ␤AR ligand, in the manner described previously (11,25). In 12-well plates, 3T3-L1 adipocytes were serum-starved overnight in the presence or absence of insulin (1 ng/ml), washed with PBS at 37°C, and acutely stimulated with isoproterenol (10 M for 15 min) in starvation medium. The reaction was terminated with rapid washing with ice-cold PBS, and 1l/ml [ 3 H]CGP-12177 was added to each well in a binding buffer (Dulbecco's modified Eagle's medium, 20 M HEPES, 10 nM unlabeled CGP-12177, at pH 7.4) and incubated for 4 h at 4°C. Following washing with PBS, plasma membrane proteins were solubilized via two 10-min incubations in acidic buffer, and the resulting supernatant was assayed for 3 H content by scintillation counting. NaOH cell lysis following acid extraction confirmed the retention of CGP-12177 at the cell surface as significant 3 H was not detected in cell lysates. Nonspecific binding was determined by competitive inhibition of CGP-12177 binding with a saturating dose of propanolol (20 M).
Data Quantitation and Statistical Analyses-Densitometry results were quantitated using Kodak Image Station 440 software, and values are expressed as means Ϯ S.E. Data were analyzed using the Student's t test, and p values were assigned with p Յ 0.05 considered significant.

Insulin Treatment Leads to a Time-and Dose-dependent
Increase of ␤-Arrestin1 Ser-412 Phosphorylation-We have previously shown that insulin treatment (100 ng/ml) for 8 -12 h leads to proteasomally mediated degradation of ␤-arrestin1, reducing cellular ␤-arrestin1 function (10). Phosphorylation of Ser-412 can also impair ␤-arrestin1 activity, providing an alternate mechanism to functionally down-regulate this protein.
Insulin Treatment Inhibits GPCR Ligand-mediated Ser-412 Dephosphorylation-Because increased ␤-arrestin1 Ser-412 phosphorylation could lead to desensitization of G␣ i -coupled GPCR activation of MAP kinase, we determined whether insulin treatment prevented Ser-412 dephosphorylation induced by treatment with isoproterenol or LPA. Insulin pretreatment for 6 h prevented isoproterenol-and LPA-induced dephosphorylation of Ser-412 (Fig. 2, A and B) and also inhibited isoproterenol-and LPA-mediated ERK phosphorylation (Fig. 2C). Insulin treatment itself can impair ERK signaling due to Raf-1 hyperphosphorylation and uncoupling of Grb2 and SOS (26 -28), an effect that is fully resolved within 4 -6 h. Consistent with this, the insulin pretreatment protocol did not cause generalized desensitization of ERK signaling as EGF-mediated ERK activation proceeded normally (Fig. 2D).
Isoproterenol is a nonspecific activator of all ␤AR subtypes (␤1, ␤2, and ␤3). The ␤1and ␤2ARs utilize ␤-arrestin during ERK activation, while the ␤3AR activates ERK by a ␤-arrestinindependent but Src-dependent mechanism (29). As insulin treatment causes a partial rather than complete inhibition of isoproterenol-mediated ERK activation, we hypothesized that ␤3AR-mediated ERK signaling, which is ␤-arrestin1-independent, is unaffected by insulin treatment. To test this, we measured ERK phosphorylation following treatment with the ␤3ARspecific ligand BRL 37344 in the presence or absence of insulin.
As shown in Fig. 2E, insulin pretreatment did not impair ␤3AR-mediated ERK activation.
Our preliminary studies showed that, unlike isoproterenol and LPA, IGF-1 treatment did not lead to detectable Ser-412 dephosphorylation in total cell lysates (not shown). We next determined whether the ␤-arrestin1 recruited to the IGF-1R following IGF-1 treatment is Ser-412-phosphorylated. We measured the amount of Ser-412-phosphorylated ␤-arrestin1 in IGF-1R immunoprecipitates and did not detect significant amounts either before or after IGF-1 treatment (Fig. 3A). In contrast, insulin pretreatment greatly increased the amount of Ser-412-phosphorylated ␤-arrestin1 associated with the IGF-1R following IGF-1 treatment. Strikingly the same insulin treatment protocol blocked IGF-1R/Src association (Fig. 3B) and reduced IGF-1-mediated ERK phosphorylation (Fig. 3C). Thus, IGF-1 normally recruits dephosphorylated ␤-arrestin1 to the receptor, which is fully competent to associate with Src and relay signals to ERK. In contrast, after insulin pretreatment, IGF-1 recruits Ser-412-phosphorylated ␤-arrestin1 to the receptor, which cannot couple to Src, and displays impaired downstream signaling.
Insulin-induced Ser-412 Phosphorylation Is Partially Dependent on ERK Signaling-The Ser-412 site is an ERK consensus sequence, and therefore we questioned whether insulininduced Ser-412 phosphorylation proceeds by an ERKdependent mechanism in our system. To assess this, we used the MEK inhibitor PD98059 and found that it substantially reduced insulin-induced ␤-arrestin1 Ser-412 phosphorylation, while the phosphatidylinositol 3-kinase inhibitor wortmannin had no effect (Fig. 4A). We quantified eight independent experiments using PD98059 and found it led to a 62% reduction in insulin-stimulated Ser-412 phosphorylation (Fig. 4B, p ϭ 0.03). Quantification of a similar number of experiments using wortmannin failed to detect a significant effect (9% reduction, p ϭ 0.41, data not shown). Blots showing the effectiveness of PD98059 (preventing MEK-dependent ERK phosphorylation) and wortmannin (preventing phosphatidylinositol 3-kinase-dependent Akt phosphorylation) are shown in Fig. 4, C and D. FIG. 1. Insulin treatment is associated with increased ␤-arres-tin1 Ser-412 phosphorylation. A, serum-starved 3T3-L1 adipocytes were treated with insulin (100 ng/ml) for the indicated time course, and total cell lysates were subjected to Western blotting as described under "Experimental Procedures." B, results from A were quantified by densitometry as described under "Experimental Procedures," and data represent mean Ϯ S.E. of eight experiments. C, cells were treated with the indicated doses of insulin for 6 h. D, results from C were quantified by densitometry as above, and data represent the mean Ϯ S.E. of three experiments. E, serum-starved 3T3-L1 adipocytes were treated with isoproterenol (Iso, 10 M), LPA (10 M), or EGF (10 ng/ml) for the indicated times, and total cell lysates were subjected to Western blotting. *, p Յ 0.05; **, p Յ 0.01. IB, immunoblot; P-, phospho-. PP2A Associates with ␤-Arrestin1, and Treatment with the PP2A Inhibitor Okadaic Acid Leads to Increased Ser-412 Phosphorylation-The multiprotein complex that forms at the activated ␤2AR includes not only ␤-arrestin and other ERK signaling cascade intermediates but also the phosphatases PP2A and PP2B (21). In addition, a recent study has shown that isoproterenol treatment acutely activates PP2A (31). Since two groups, including ours, have previously shown that insulin treatment leads to decreased PP2A phosphatase activity (22,32), we speculated that inhibition of PP2A could contribute to increased ␤-arrestin1 Ser-412 phosphorylation.
First we found that ␤-arrestin1 and PP2A were associated in the basal state and that isoproterenol treatment led to an increase in this association in 3T3-L1 adipocytes (Fig. 5A), suggesting that PP2A is involved in isoproterenol-mediated Ser-412 dephosphorylation. If PP2A can decrease Ser-412 phosphorylation, inhibition of PP2A activity should lead to an increase in Ser-412 phosphorylation. Low (1 M) concentrations of okadaic acid have been shown to specifically inhibit PP2A (33), and when cells were incubated with 1 M okadaic acid, Ser-412 phosphorylation was clearly increased (Fig. 5B). When insulin and okadaic acid were added together, insulin did not significantly increase Ser-412 phosphorylation above the level induced by okadaic acid alone (Fig. 5, C and E), suggesting either that insulin and okadaic acid act through a common mechanism or that okadaic acid treatment alone leads to stoichiometric phosphorylation at Ser-412, which cannot be enhanced further. In addition, treatment of 3T3-L1 adipocytes with okadaic acid prevented isoproterenol-mediated dephos-phorylation of Ser-412 (Fig. 5, D and E), suggesting that isoproterenol treatment requires activation of PP2A to induce Ser-412 dephosphorylation.

Insulin-induced Ser-412 Phosphorylation Is Lost in Small t Antigen-expressing Cells-To further test the hypothesis that
insulin-induced inactivation of PP2A leads to increased Ser-412 phosphorylation, we used an HIRcB cell line stably expressing small t antigen (ST) derived from SV40 (22). The PP2A heterotrimer includes a scaffolding subunit (A), a catalytic subunit (C), and one of several possible regulatory subunits (B), each of which localize PP2A to specific subcellular compartments and direct interaction with specific protein targets (34). In cells infected with SV40, B subunits are displaced from their natural binding to A subunits by ST, resulting in inhibition of PP2A phosphatase activity (35). PP2A is the only protein known to interact with ST, and when expressed in cells, ST inhibits PP2A, but not PP1, activity (22). As shown in Fig.  6, A and B, basal Ser-412 phosphorylation was increased in ST-expressing cells, whereas insulin-induced Ser-412 phosphorylation was prevented.
Insulin-induced ␤-Arrestin1 Ser-12 Phosphorylation Impairs ␤-Adrenergic Receptor Internalization-Down-regulation of ␤-arrestin1 protein levels leads to inhibition of G protein-mediated ERK activation. However, ␤-arrestin1 Ser-412 phosphorylation itself could be an independent mechanism for inhibiting G protein-mediated ERK activation. Therefore, as an alternative approach, we took advantage of the fact that insulin causes ␤-arrestin1 degradation in a dose-responsive manner, and low doses of insulin do not cause proteasomal degradation of ␤-arrestin1, although they are still capable of increasing Ser-412 phosphorylation. As shown in Fig. 7, A and B, treatment with insulin (1 ng/ml) for 6 h had no effect on ␤-arrestin1 protein levels but led to increased Ser-412 phosphorylation. As shown in Fig. 7C, we found that insulin (1 ng/ml) pretreatment had no effect on isoproterenol-mediated cAMP generation, a G␣ s -coupled, ␤-arrestin1-desensitized signaling event that is unaffected by Ser-412 phosphorylation (16). Since ␤-arrestin1 Ser-412 phosphorylation impairs the association of ␤-arrestin1 to Src and also disrupts clathrin-mediated internalization of the ␤2AR, we reasoned that insulin treatment would impair endocytosis of ␤ARs. We measured isoproterenol-induced ␤AR internalization following insulin treatment using the CGP-12177 assay (25). CGP-12177 is a ␤AR agonist that is hydrophilic and not internalized once bound to the receptor, providing a quantitative measure of surface ␤AR before and after ligand stimulation. Pretreatment with insulin (1 ng/ml) to cause ␤-arrestin1 Ser-412 phosphorylation with no change in ␤-arrestin1 protein levels inhibited isoproterenol-mediated ␤AR internalization (75%, Fig. 7D).
Insulin-induced Ser-412 Phosphorylation Leads to Desensitization of ␤-Arrestin-dependent, G Protein-mediated MAP Kinase Signaling-We next hypothesized that Ser-412-phosphorylated ␤-arrestin1 would be inefficient at directing ERK phosphorylation. Again using the lower dose of insulin, we found that insulin (1 ng/ml) pretreatment blocked ERK phosphorylation by three independent G␣ i -coupled ligands (Fig. 8, A  and B). In contrast, insulin treatment had no effect on EGFmediated ERK phosphorylation, which is not ␤-arrestin1/G␣ icoupled but does utilize many of the same MAP kinase signaling intermediates, such as Grb2/SOS, Ras, Raf-1, and MEK. Finally, as a positive control, we measured LPA-mediated ERK phosphorylation in the Hyg and ST cell lines. In ST cells where Ser-412 phosphorylation was increased due to PP2A inactivity, LPA-mediated ERK phosphorylation was impaired (reduced 48% at 2.5 min, p ϭ 0.09; reduced 29% at 5 min, p ϭ 0.31).

DISCUSSION
In the present study, we found that insulin led to ␤-arrestin1 Ser-412 phosphorylation and impaired ␤-arrestin1 function. We demonstrated the roles of a kinase (ERK) and a phosphatase (PP2A) in mediating this effect of insulin on the Ser-412 site. We showed that, by increasing ␤-arrestin1 Ser-412 phosphorylation, insulin treatment led to desensitization of ␤-arrestin1-dependent, G protein-mediated ERK signaling.
This molecular mechanism was demonstrated using the IGF-1R, which is G␣ i -coupled and associates with ␤-arrestin1 and Src following IGF-1 stimulation. When IGF-1 stimulation was preceded by insulin pretreatment, the ␤-arrestin1 associated with the IGF-1R was clearly phosphorylated at Ser-412. Perhaps more importantly, insulin pretreatment led to impaired FIG. 5. PP2A associates with ␤-arrestin1, and treatment with the PP2A inhibitor okadaic acid leads to increased Ser-412 phosphorylation. A, serum-starved 3T3-L1 adipocytes were treated with isoproterenol (Iso, 10 M) for 0 -10 min. Cell lysate proteins (500 g) were immunoprecipitated overnight with either anti-␤-arrestin1 antibody or with anti-PP2A antibody, and the resulting immune complexes were captured on protein A/G beads. Western blotting was then performed using the indicated antibodies. B, serum-starved 3T3-L1 adipocytes were treated with either okadaic acid (OA, 1 M) or insulin (Ins, 100 ng/ml) for the indicated times, and Western blotting of total cell lysates was performed as described under "Experimental Procedures." C, serum-starved 3T3-L1 adipocytes were treated with either okadaic acid (OA, 1 M) alone or with okadaic acid plus insulin (Ins, 100 ng/ml) for 6 h. D, serum-starved 3T3-L1 adipocytes were treated with isoproterenol (Iso, 10 M) for 30 min. Where indicated, cells were pretreated with okadaic acid (OA, 1 M) for 6 h prior to the addition of isoproterenol. Total cell lysates were Western blotted as described under "Experimental Procedures." E, densitometry from at least two separate experiments from B-D as mean Ϯ S.E. IB, immunoblot; P-, phospho-; IP, immunoprecipitation. Ser-412 phosphorylation involves ERK but not phosphatidylinositol 3-kinase. A, serum-starved 3T3-L1 adipocytes were treated with insulin (100 ng/ ml), PD98059 (PD, 20 M), and wortmannin (WM, 100 nM) either alone or in combination for 6 h. Total cell lysates were obtained, and Western blotting was performed as described under "Experimental Procedures." B, data were quantified by densitometry and are displayed as the mean Ϯ S.E. for eight experiments. Data are adjusted for the effect of PD98059 alone (1.1-1.2-fold increase, data not shown). C and D, serum-starved 3T3-L1 adipocytes were treated with PD98059 (B) or wortmannin (C) for 30 min prior to the addition of insulin (100 ng/ml) for the indicated times. Western blotting of total cell lysates was performed as described under "Experimental Procedures," and each blot is representative of four separate experiments. *, p Յ 0.05. IB, immunoblot; P-, phospho-; Ins, insulin.
IGF-1-mediated recruitment of Src to the IGF-1R and inhibition of ERK activation by IGF-1. This insulin-induced increase in ␤-arrestin1 Ser-412 phosphorylation represents a novel mechanism for insulin-induced desensitization of G proteinmediated ERK activation.
The importance of ␤-arrestin1 Ser-412 dephosphorylation during initiation of ERK signaling from the activated ␤-adrenergic receptor has been described previously (16). We definitively demonstrated ligand-mediated Ser-412 dephosphorylation in 3T3-L1 adipocytes using two different ␤-arrestindependent GPCR ligands, isoproterenol and LPA. In contrast FIG. 7. Insulin-induced ␤-arrestin1 Ser-412 phosphorylation impairs ␤-adrenergic receptor internalization. 3T3-L1 adipocytes were serum-starved in the presence or absence of insulin (Ins, 1 ng/ml) for 6 h. A, Western blotting was performed on total cell lysates as described under "Experimental Procedures." B, band densitometry from four separate experiments is shown as mean Ϯ S.E. C, following insulin (1 ng/ml) treatment, cells were washed and then stimulated with isoproterenol (Iso, 10 M) for the indicated times. Intracellular cAMP levels were determined using an enzyme-linked immunoassay system from Amersham Biosciences according to the manufacturer's instructions. The concentration of cAMP is expressed as fmol of cAMP/g of protein.
The solid line (-) represents isoproterenol alone; the dashed line (---) represents insulin (1 ng/ml) pretreatment for 6 h followed by isoproterenol treatment. Insulin pretreatment alone had no effect on [cAMP] (data not shown). Experiments were performed in triplicate, and three independent experiments were performed of which the displayed graph is representative. D, ␤AR internalization was determined by a ligand binding assay using the non-internalized ␤AR ligand [ 3 H]CGP-12177 as described under "Experimental Procedures." Each experiment was performed in triplicate, and the displayed graph is representative of three separate experiments. IB, immunoblot; P-, phospho-. to treatment with GPCR ligands, insulin treatment led to an increase in phosphorylation at Ser-412 and in fact prevented Ser-412 dephosphorylation induced by isoproterenol and LPA. This was associated with desensitization of ERK signaling by both isoproterenol and LPA. Interestingly while stimulation of the IGF-1R, which is G␣ i -coupled and ␤-arrestin1-dependent for MAP kinase signaling, did not lead to detectable reductions in total cellular Ser-412-phosphorylated ␤-arrestin1, receptorbound ␤-arrestin1 was clearly dephosphorylated, suggesting that the IGF-1R interacts preferentially with Ser-412-dephosphorylated ␤-arrestin1. Insulin treatment greatly increased the amount of Ser-412-phosphorylated ␤-arrestin1 at the IGF-1R and blocked the association of the IGF-1R and Src. Therefore, ligand-mediated dephosphorylation of Ser-412 plays a critical role in G protein-mediated ERK signaling, and this was prevented by insulin treatment.
␤-Arrestin1 has an ERK consensus sequence around Ser-412, and a previous study (20) using in vitro phosphorylation and two-dimensional tryptic phosphopeptide mapping identified ERK as the kinase responsible for phosphorylating Ser-412. We found that, using the MEK inhibitor PD98059, insulininduced Ser-412 phosphorylation in 3T3-L1 adipocytes was reduced by 62%. We conclude that ERK, or another MEK-dependent kinase, was responsible for phosphorylating the Ser-412 site in response to insulin treatment. Interestingly other ERK-activating ligands (isoproterenol, LPA, IGF-1, and EGF) failed to increase Ser-412 phosphorylation, suggesting that a non-ERK mechanism also plays a partial role in determining Ser-412 phosphorylation status.
Our findings suggest that changes in phosphatase activity, specifically PP2A, can have a profound effect on Ser-412 phosphorylation and may partially explain the effects of insulin. PP2A is activated upon isoproterenol stimulation (31), and we found that PP2A co-immunoprecipitated with ␤-arrestin1 in 3T3-L1 adipocytes both at base line and in response to isoproterenol treatment. This is consistent with a previous study in which both PP2A and PP2B were recovered from ␤-arrestincontaining ␤2AR immunoprecipitates (21), suggesting a role for PP2A in isoproterenol-mediated Ser-412 dephosphorylation. Consistent with this, we found that inhibition of PP2A activity by okadaic acid treatment enhanced Ser-412 phosphorylation and prevented isoproterenol-mediated dephosphorylation of this residue.
We have previously shown that insulin treatment inhibits PP2A activity (22,32). In the current study, insulin treatment did not lead to further Ser-412 phosphorylation in cells where PP2A was already inactivated by another method. When cells were treated with insulin and okadaic acid together, the effect on Ser-412 phosphorylation was not additive. In addition, we found that insulin treatment had no effect on Ser-412 phosphorylation in HIRcB cells that express the ST, an SV40-derived protein that specifically inhibits PP2A. Interestingly basal phosphorylation of Ser-412 was increased both by okadaic acid treatment and in ST-expressing cells in the absence of ligand stimulation. This suggests that the basal activity of a serine kinase targeting Ser-412 is held in check by PP2A activity and that a balance between the activity of this kinase and PP2A normally regulates Ser-412 phosphorylation levels. It is also possible to conclude from these experiments that okadaic acid treatment and ST expression lead to stoichiometric phosphorylation at Ser-412, and therefore insulin is unable to further increase Ser-412 phosphorylation irrespective of the mechanism involved. However, the results presented here provide a plausible mechanism to explain why ERK activation alone does not fully account for the increase in Ser-412 phosphorylation with insulin treatment.
A previous study examined the stoichiometry of ␤-arrestin1 phosphorylation in human embryonic kidney 293 cells (16). In these cells, ␤-arrestin1 was found to contain 0.84 mol of P i /mol of protein and that Ser-412 accounts for 90% of ␤-arrestin1 phosphorylation (16,20). Therefore, in human embryonic kidney 293 cells, Ser-412 phosphorylation stoichiometry is 0.76 mol of P i /mol of ␤-arrestin1 protein. Our findings in 3T3-L1 adipocytes suggest a similar or slightly lower level of basal Ser-412 phosphorylation as insulin treatment led to a ϳ1.6 -1.8-fold increase in phosphorylation at this site. This level of phosphorylation is fully accommodated if the basal stoichiometry in 3T3-L1 adipocytes is ϳ0.6 mol of P i /mol of ␤-arrestin1 protein.
Treatment with 1 ng/ml insulin, which induced Ser-412 phosphorylation but did not reduce ␤-arrestin1 protein levels, led to impaired MAP kinase signaling by three ␤-arrestin-dependent ligands (isoproterenol, LPA, and IGF-1) and impaired internalization of the ␤AR. Insulin-induced increases in Ser-412 phosphorylation, however, had no effect on ␤AR/G␣ s -mediated cAMP accumulation and also had no effect on MAP kinase signaling by the ␤-arrestin1-independent EGF receptor. These data are consistent with studies using transfection of ␤-arrestin1 mutants to alter the Ser-412 phosphorylation state. Cells expressing the S412D mutant of ␤-arrestin1, which mimics the phosphorylated state of Ser-412, display impaired ␤-ar-restin1 binding to Src and to clathrin, have impaired receptor internalization, and have a defect in ligand-mediated MAP kinase signaling while displaying preserved G␣ s signal desensitization (12,16,21). Thus our findings, in a non-transfected cell system, provide further evidence that desensitization of G␣ s signaling occurs independently of ␤-arrestin1 Ser-412 phosphorylation and show that Ser-412 phosphorylation following insulin treatment is sufficient to cause desensitization of G protein-mediated ERK signaling.
Possible physiologic consequences of increased Ser-412 phosphorylation would be linked to impaired G protein/ERK signaling as G␣ s /cAMP/PKA signals are not affected by Ser-412 phosphorylation status. Therefore it is unlikely that adrenergic receptor-mediated inhibition of leptin secretion, which is mediated by a PKA-dependent mechanism, would be affected (36). However, although classically considered a PKA-dependent event, lipolysis may be partially regulated by adrenergic receptor-mediated ERK signaling (37), and therefore ␤-arrestin1 Ser-412 phosphorylation may play a role in this process. Survival and differentiation of preadipocytes and adipocytes may also depend on GPCR-associated, ERK-mediated signaling. Mitogen-induced proliferation of preadipocytes is ERK-dependent, and norepinephrine-induced ERK signaling prevents apoptosis of mature adipocytes (38).
In summary, we found that increased ␤-arrestin1 Ser-412 phosphorylation following insulin treatment was sufficient to disrupt G protein-mediated MAP kinase signaling and was most likely due to the combination of insulin-induced ERK activation and inhibition of PP2A activity. These data represent a new mechanism whereby insulin treatment causes desensitization of G protein-mediated MAP kinase signaling.