Trafficking of β₂-adrenergic receptors: insulin and β-agonists regulate internalization by distinct cytoskeletal pathways

β₂-Adrenergic receptors (β₂ARs) are members of the superfamily of G-protein-coupled receptors (GPCRs). These prototypic GPCRs display acute homologous desensitization in response to β₂-adrenergic agonists as well as counterregulation by several growth factor receptors with intrinsic tyrosine kinase activity, including insulin (Morris and Malbon, 1999). The trafficking of the β₂AR is central to both agonist-induced and insulin-induced regulation of β-adrenergic signaling (Lefkowitz, 1998; Morris and Malbon, 1999). Sequestration of β₂AR in response to insulin represents a loss of the surface complement of receptors and is one of the mechanisms of the counterregulatory effects of insulin on catecholamine action (Karoor et al., 1998). Agonist-induced sequestration of β₂ARs has been reported using a variety of techniques (Carman and Benovic, 1998; Gagnon et al., 1998). By comparison, little is known about the counterregulation by tyrosine kinases and how insulin influences GPCR trafficking.

β₂ARs are phosphorylated by insulin treatment of cells (Karoor and Malbon, 1998). Studies in vivo have shown that insulin stimulates the phosphorylation of two major tyrosine residues, Y350 and Y364; both residues are located in the C-terminal cytoplasmic domain of the β₂AR (Karoor et al., 1995). The phosphorylation of the Y350 residue in response to insulin creates an SH2-binding site to which Grb2, the p85 catalytic domain of phosphatidylinositol 3-kinase, and the GTPase dynamin can dock (Shih and Malbon, 1998). Purified insulin receptor and recombinant β₂ARs have been used to show that insulin stimulates the insulin receptor-catalyzed phosphorylation of these same residues (Baltensperger et al., 1996; Doronin et al., 2000). Phosphorylation of the β₂AR impairs its ability to signal to its cognate G-protein Gs, a blockade that requires Grb2 with an intact SH2 domain (Shih and Malbon, 1998).

Insulin catalyzes a robust internalization of β₂AR. Insulin thereby suppresses β-adrenergic signaling, precluding access of β-agonist to the β₂AR. In spite of similarities in the ability of β-agonists and insulin to stimulate sequestration of the β₂AR, important differences may exist in the character of the pathways by which these sequestrations occur (Karoor et al., 1998). For example, β₂AR internalization in response to insulin, but not β-adrenergic agonist, can be blocked with inhibitors of phosphatidylinositol 3-kinase (PI3-kinase), such as wortmannin or LY294002 (Wang et al., 2000), as well as by inhibitors of Src activity (Shumay et al., 2002). In the current work, we extend the studies on GPCR trafficking and probe the role of cytoskeletal elements in the trafficking of β₂AR by insulin and by β-agonists. The results provide compelling evidence that insulin and β-adrenergic agonists employ unique cytoskeletal elements in trafficking receptor from the cell surface. Furthermore, the results reveal that the internalization and recycling aspects of receptor trafficking in response to insulin have unique cytoskeletal requirements: sequestration requires an intact actin cytoskeleton, whereas recycling to the cell surface requires intact microtubules.

The plasmid encoding the enhanced-green fluorescent protein (GFP)-tagged human β₂AR (in pCDNA3) was obtained from Jeffrey Benovic (Kimmel Cancer Center, Thomas Jefferson University, Philadelphia, PA). To analyze and characterize an expression of GFP-tagged β₂AR, the following antibodies were used: anti-β2AR [CM 02], anti-peptide antibody to an exofacial domain of the β₂AR (Wang et al., 1989) and anti-GFP rabbit polyclonal antibodies (Quantum Biotechnologies, Montreal, Quebec, CN); goat, anti-rabbit antibody conjugated with horseradish peroxidase and goat anti-mouse antibody conjugated with horseradish peroxidase (both from Kirkegard & Perry Laboratories, Gathesburg, MD); mouse anti-β-tubulin, anti-β-tubulin-Cy3 and anti-actin antibody (Sigma, St Louis, MO).

Human epidermoid carcinoma cells (A431) were maintained in Dulbecco modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (HyClone, Logan, UT), penicillin (60 μg/ml) and streptomycin (100 μg/ml), and grown in a humidified atmosphere of 5% CO₂ and 95% air at 37°C. A431 clones stably expressing the GFP-tagged human β₂AR were cultivated with the addition of G418. Chinese hamster ovary (CHO) cells were obtained from the ATCC collection, propagated and stably transfected with pCDNA3 harboring the GFP-tagged human β₂AR (Liu et al., 1999a; Liu et al., 1999b).

For most studies, A431 cells were serum starved overnight with the following treatment as indicated. Cells were harvested and lysed in a lysis buffer (1% Triton X-100, 0.5% Nonidet-40, 10 mM dithiothreitol, 5 μg/ml aprotinin, 5 μg/ml leupeptin, 100 μg/ml bacitracin, 100 μg/ml benzamidine, 2 mM sodium orthovanadate, 150 mM NaCl, 5 mM EDTA, 50 mM NaF, 40 mM sodium pyrophosphate, 50 mM KH2PO4, 10 mM sodium molybdate and 20 mM Tris-HCl, pH 7.4) at 4°C for 20 minutes. After centrifugation of the cell debris at 14,000 g for 30 minutes, clarified lysates were subjected to immunoprecipitation for 2 hours with antibodies specific for the β₂AR (CM04) linked covalently to agarose beads. Immune complexes were washed three times with RIPA buffer (20 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, 10 mM DTT, 1% Triton X-100, pH 8.0) and separated on 10% SDS-acrylamide Laemmli gels. Immunoblotting and detection of the β-tubulin or actin were performed with anti-β-tubulin and anti-actin antibody (both from Sigma), respectively, as previously described (Fan et al., 2001a; Fan et al., 2001b).

For the confocal microscopy studies, cells expressing β₂-AR and grown in eight-well Nunc chamber slides were serum starved and treated as indicated. Objects were imbedded in SlowFade (Molecular Probes) anti-fade reagent. Images were acquired on the Nikon Eclipse E600 microscope (oil-immersion, 60· objective) using He and Ne lasers. The digital images were exported as TIFF files and analyzed in Adobe Photoshop 5.5.

For immunofluorescent staining, cells were fixed with 3% paraformaldehyde and 0.25% glutaraldehyde in PBS pH 7.4 containing 2% of sucrose, permeabilized with 0.1% of Triton X-100 and incubated with blocking buffer (1% of normal goat serum in PBSTriton X-100) for 1 hour at room temperature. Fixed cells were incubated with primary antibody for 30 minutes (37°C); this was followed by extensive washing and a 30 minute incubation with the Alexa 655 conjugated secondary antibody (Molecular Probes, Eugene, OR) and final wash steps.

A431 cells and stably transfected clones were routinely stimulated with either isoproterenol (10 μM) or insulin (100 nM) for 30 minutes and the trafficking of the GFP-tagged receptor monitored by fluorescence microscopy. Cells were serum-deprived for 18 hours before stimulation to remove growth factors and catechols from the cell media. For studies of the role of cytoskeleton elements in receptor trafficking, drugs were added either 30 minutes (for treatment with latrunculin or taxol) or 15 minutes (for treatment with nocodazole) in advance of the challenge with hormones. For studies with nocodazole only, cells were pre-cooled on ice before addition of the drug, to insure that the microtubules were fully depolymerized. The concentrations at which the drugs were used are as follows: nocodazole, 10 μM; taxol, 1 μM; and latrunculin, 1 μM.

The expression of β₂ARs in stably transfected A431 and CHO clones was quantified using the radiolabeled, high-affinity β-adrenergic antagonist [¹²⁵I]iodocyanopindolol (ICYP) to bind to intact cells. Identical radioligand binding assays were performed using the radiolabeled, water-soluble, membrane-impermeant [³H]CGP-12177 to determine the amount of cell-surface β₂AR in the untreated cells, as well as in the cells treated with either insulin or isoproterenol in the absence or presence of the cytoskeletal inhibitors/stabilizers (Karoor and Malbon, 1998). The data are presented as mean±s.e.m., where the amount of cell-surface receptor in the untreated cells is set as 100%.

Human epidermoid carcinoma A431 cells are a popular model for studying β₂AR biology and regulation (Morris and Malbon, 1999). The regulation of the receptor was studied using human β₂AR tagged on the C-terminus with GFP (Gagnon et al., 1998) and examined by confocal microscopy (Fan et al., 2001a; Fan et al., 2001b). A431 cells respond to the β-adrenergic agonist isoproterenol with activation of the β₂AR, accumulation of cyclic AMP, desensitization, internalization, resensitization and recycling of receptor back to the cell surface (Shih and Malbon, 1994). Activation of the insulin receptor tyrosine kinase leads to counterregulation of the β₂AR, uncoupling from Gs and internalization (Karoor and Malbon, 1998). Treatment of A431 cells with insulin (100 nM) or isoproterenol (10 μM) led to a marked internalization of β₂AR (Fig. 1, control). In the absence of either hormone, β₂ARs were localized largely to the cell membrane (yellow arrows throughout), whereas treatment with isoproterenol or insulin leads to internalization of β₂AR (white arrowheads throughout). These changes were quantified by the equilibrium radioligand binding studies, which made use of the water-soluble, cell-impermeant β-adrenergic antagonist ligand [³H]CGP-12177 to quantify the amount of cell-surface β₂ARs. When compared with the percentage of cell-surface localized β₂ARs observed in untreated cells, the amount of β₂ARs lost to internalization in response to isoproterenol was 42±3.4% (n=6) and in response to insulin, 31±4.0% (n=6), as measured by [³H]CGP-12177 binding to intact A431 cells in suspension. This level of internalization agrees well with previous data on the internalization of β₂ARs in response to isoproterenol and to insulin in other cell lines (Karoor et al., 1998; DeCostanzo et al., 2002; Morris and Malbon, 1999).

It has been known for many years that microtubules have a role in the trafficking of GPCRs, including the β₂AR (Limas and Limas, 1983). We examined what effects disruption of the microtubules with nocodazole would have on the trafficking of β₂AR, observing that this agent, which in vivo binds tubulin and depolymerizes microtubules, blocks isoproterenol-induced internalization of β₂ARs (Fig. 1, +Noc.). Remarkably, nocodazole did not influence the response to insulin; insulin stimulated a robust internalization of β₂AR even in the presence of this microtubule inhibitor. Thus, we gained the first insight that the trafficking of the β₂AR by two potent regulators of internalization had some fundamental differences in mechanism. The role of the other major cytoskeletal system, the microfilaments formed from F-actin, was examined using latrunculin A, which binds the actin monomer and blocks F-actin dynamics. Treating A431 cells with latrunculin A had little influence on the sequestration of β₂ARs in response to isoproterenol, but effectively blocked the ability of insulin to internalize the β₂AR (Fig. 1, +Latr.).

The effects of nocodazole were examined by analyzing the trafficking of the GFP-tagged receptor and by confocal analysis of cells stained with anti-β-tubulin (Fig. 2). Microtubules stained prominently in these epidermoid carcinoma cells (Fig. 2A), the patterns of microtubules being somewhat sensitive to treatment of the cells with either 100 nM insulin (+Ins) or with the β-adrenergic agonist isoproterenol (10 μM, +Iso). In untreated cells, the majority of the β₂ARs were localized to the cell membrane. Merging the two images shows the redistribution of microtubules and marked internalization of β₂ARs that occurs when the cells are treated with either insulin (i.e. counterregulation) or isoproterenol (i.e. agonist-induced sequestration). The β-agonist-induced induction of the β₂ARs, by contrast, was not observed in the nocodazole-treated cells (Fig. 2B). Treatment with nocodazole provoked a profound destabilization of the microtubular network in the untreated and hormone-treated cells alike (Fig. 2B). Microtubules were markedly shortened in the nocodazole-treated cells. Remarkably, in spite of the loss of much of the cytoskeletal architecture by depolymerization of microtubules, the internalization of β₂ARs in response to insulin stimulation was essentially the same as noted in the control cells (Fig. 2A). Treating the cells with latrunculin A did not alter the localization of the β₂ARs at the cell membrane in the absence of hormones (Fig. 2C). Agonist-induced sequestration of β₂ARs in response to isoproterenol proceeded normally in cells treated with latrunculin A (Fig. 2C). The ability of insulin to counterregulate β₂ARs and provoke internalization, by contrast, was essentially blocked in the latrunculin A-treated cells.

Taxol stabilizes microtubules by binding to a pocket of β-tubulin on the inner surface of a microtubule. This negates the effects of GTP hydrolysis that drives the depolymerization occurring on the other side of the monomer (He et al., 2001). Taxol treatment of A431 cells had a counterintuitive effect on the trafficking of β₂ARs. Treatment with taxol stabilized the microtubules, while still permitting the internalization of receptor in response to stimulation of either insulin or isoproterenol (Fig. 3A). Internalization of β₂ARs in response to either insulin or isoproterenol was attenuated modestly by taxol. Well-defined, cell-membrane localization of β₂ARs was evident, although perinuclear accumulation of receptor in response to either hormone was also noted. The distribution of receptors appears to be more homogeneously partitioned between the cell membrane and the perinuclear sites of accumulation, with less receptor found elsewhere in the cytoplasmic compartment. The effects of taxol treatment on the microtubules were profound, as noted by epifluorescence analysis of the β-tubulin (Fig. 3B). In taxol-treated cells, relocalization of the microtubular network at the cell membrane was prominent. Stimulating taxol-treated cells with either isoproterenol or insulin reduced the accumulation of microtubules at the cell membrane and the formation of microtubule arrays elsewhere in the cell (Fig. 3B).

Components of the multivalent, signaling complexes associated with the β₂AR include AKAP250 (gravin), AKAP79, protein kinase A, protein kinase C, Src and protein phosphatase 2B (Shih and Malbon, 1994; Cong et al., 2001; Lin et al., 2000; Fan et al., 2001a; Fan et al., 2001b; Oliveria et al., 2003). β₂AR signaling complexes were isolated by immunoprecipitation to ascertain whether or not actin or β-tubulin could be detected in the complex (Fig. 4). Analysis of the immune precipitations with anti-β₂AR antibodies revealed the presence of both actin and β-tubulin. Treatment with insulin or isoproterenol increased the amount of both cytoskeletal elements associated with the signaling complexes. AKAP250 (gravin), a scaffold protein for the β₂AR, displays an F-actin binding site (Gelman, 2002). Other elements of GPCR signaling complexes have been reported to interact with microtubules (Roychowdhury and Rasenik, 1994; Wang et al., 1990). Immunoprecipitations performed with unrelated antibodies and with unmodified matrix failed to pull down cytoskeletal elements, as determined by staining of immunoblots of the precipitates with either anti-β-tubulin or anti-F-actin antibodies (data not shown).

Compared with what is known about agonist- and insulin-induced internalization of β₂ARs, little is known about the recycling of receptor back to the cell membrane and what role, if any, the cytoskeleton plays in this process. To address the role of the F-actin cytoskeleton and the microtubular network in β₂AR recycling, cells were first stimulated with either insulin or isoproterenol for 30 minutes to induce full receptor sequestration. The cells were then either washed free of insulin to induce recovery from insulin treatment or incubated with 10 μM propranolol – a high-affinity, β-adrenergic antagonist – to block isoproterenol binding to the receptors, and allowed to recover for up to 90 minutes (Fig. 5A). After 30 minutes with either insulin or isoproterenol, the bulk of the receptor was sequestered to perinuclear areas away from the cell membrane. The time-course for recovery reveals a progressive recycling of the β₂ARs to the cell membrane that is largely complete within 60-90 minutes. To test the role of microtubules in the recovery phase and in recycling of β₂ARs, cells were treated with either insulin or isoproterenol for 30 minutes and then co-treated with nocodazole for 30 minutes and finally washed free of insulin (insulin-induced) or treated with propranolol (isoproterenol-induced) in buffer containing nocodazole (Fig. 5B). The disruption of microtubules led to the attenuation of the recycling of receptors from isoproterenol- or insulin-treated cells. Although microtubule dynamics were not important to the ability of insulin to stimulate internalization of the β₂ARs, microtubules appeared essential to the recovery phase and recycling of the receptors back to the cell membrane. For β₂ARs internalized in response to isoproterenol, the recycling also was impaired by nocodazole. Thus, the internalization of β₂ARs in response to β-adrenergic agonist, as well as the recycling of the internalized β₂ARs back to the cell membrane, was blocked by disruption of microtubules.

To test the role of F-actin cytoskeleton in the recovery phase and recycling of β₂ARs, cells were stimulated with either insulin or isoproterenol for 30 minutes and then co-treated with latrunculin A for 30 minutes and finally washed free of insulin (insulin-induced) or treated with propranolol (isoproterenol-induced) in buffer containing latrunculin A (Fig. 5C). Disruption of F-actin dynamics markedly attenuated the recycling of receptors in cells stimulated with isoproterenol. The effects of latrunculin-A treatment on the recycling of receptors internalized in response to insulin stimulation were, by contrast, more modest. Normal F-actin dynamics were important to the ability of insulin to internalize the β₂ARs, but were nonessential to the recovery phase and recycling of the receptors back to the cell membrane (Fig. 5C). For β₂ARs internalized in response to isoproterenol, the recycling was impaired by latrunculin A, whereas the internalization process in response to β-agonist was essentially insensitive to the disruption of F-actin cytoskeleton.

We noted the ability of isoproterenol to stimulate a dramatic change in the organization of the microtubule network (Fig. 2) and wondered whether the change was a result of the receptor activation per se or the most probable downstream response to the β-adrenergic agonist, i.e. increased accumulation of intracellular cyclic AMP. Treating the cells with the plant diterpene forskolin led to robust accumulation of intracellular cyclic AMP by activating the adenylyl cyclase and bypassing the receptor/G-proteins upstream. Treatment with forskolin alone produced a pattern of microtubules somewhat similar to that observed in response to isoproterenol, suggesting that the regulation is probably dependent on both cyclic AMP and protein kinase A (Fig. 6). Nocodazole disrupted the microtubule network, but one could still observe arrays of fine microtubules stimulated by isoproterenol or by forskolin treatment. The effects of taxol on the stabilization of microtubules were profound (Fig. 3B). Treatment with either isoproterenol or forskolin (and presumably elevated cyclic AMP levels), by contrast, led to the appearance of radiant microtubules that were not present in the cells treated with taxol alone (Fig. 6).

Insulin signaling dominates two major cellular events, mitogenesis and metabolic regulation. These events are focused largely on insulin action at the level of skeletal muscle, liver and adipose tissue, but are observed to some extent in all cells (Olefsky, 1999; Czech and Corvera, 1999; Saltiel and Kahn, 2001). β-Catecholamines generally act to oppose the actions of insulin. Catecholamines stimulate glycogen breakdown, protein degradation, gluconeogenesis and lipolysis, whereas insulin acts to counteract each of these important metabolic pathways. A variety of studies has shown that the ability of insulin to counterregulate the β₂AR is an essential element of insulin action (Morris and Malbon, 1999). Insulin provokes the tyrosine phosphorylation of β₂AR on specific tyrosyl residues (Y350, 354 and 364), which are confined to the cytoplasmic, C-terminal tail of the receptor (Doronin et al., 2000; Karoor et al., 1995). Phosphorylation of the Y350 residue of the β₂AR creates a docking site for SH2 domains of a variety of proteins, including Grb2 and dynamin, and the regulatory subunit of PI3 kinase (Baltensperger et al., 1996). The integrity of Y350 and its phosphorylation in response to insulin are essential for insulin to express its counterregulatory control of β₂AR functions and β₂AR sequestration (Shih and Malbon, 1998).

Our current work provides several new insights into the manner in which β-adrenergic agonists differ from insulin in the trafficking of β₂ARs. Most notably, this study reveals differing roles of the microtubule versus F-actin networks in enabling the internalization of β₂ARs for agonist-induced regulation versus counterregulation by insulin. Agonist-induced internalization of β₂ARs follows desensitization and is a hallmark for virtually all GPCRs (Morris and Malbon, 1999). Agonist-induced sequestration of the β₂AR to perinuclear locales occurs within minutes, but requires 15-30 minutes to reach maximal internalization. Nocodazole in vivo binds tubulin monomer and induces the depolymerization of microtubules. We found that nocodazole effectively blocks internalization of β₂ARs in response to β-agonist, but it had no influence on the ability of insulin to counterregulate β₂ARs through sequestration. Thus, we speculate that an intact microtubular network enables agonist-induced sequestration of GPCRs.

The F-actin cytoskeletal network is essential for many cellular functions. Latrunculin A binds the actin-monomer and acts to sequester actin, blocking F-actin dynamics. In the current study, treatment with latrunculin A showed no influence on agonist-induced internalization of β₂ARs, but rather was found to block insulin-induced sequestration of the β₂ARs. These effects of disrupting actin cytoskeleton on insulin action are not a reflection of an effect proximal to receptor, as insulin receptor autophosphorylation, tyrosine phosphorylation of IRS-1,2 and Cbl, and serine/threonine phosphorylation of Akt in response to insulin are unaffected by latrunculin (Kanzaki and Pessin, 2001). Actin microfilaments do enable the translocation of the GLUT4 glucose transporter to the cell surface in response to insulin, a process that has many similarities to the counterregulation of β₂ARs in response to insulin that operates in the reverse orientation (Shumay et al., 2002). Taken together, these results reinforce the notion that insulin-induced internalization and insulin-induced export of GLUT4 to the cell membrane may constitute use of the same cellular network.

One neglected feature of the agonist-induced trafficking of GPCRs is how the internalized receptors recycle to the cell membrane. We explored whether the recovery from desensitization and from counterregulation by insulin in cells made use of different cytoskeletal elements to recycle the β₂ARs. β₂ARs internalized by either β-agonist or insulin displayed similar time-courses for the recycling of receptor back to the cell membrane. Although insulin-stimulated internalization of β₂ARs was insensitive to the disruption of the microtubules, the recovery of β₂ARs to the cell membrane was effectively blocked by nocodazole. Surprisingly, although the internalization of β₂ARs by agonist was blocked by nocodazole, the recycling of β₂ARs back to the cell membrane was only partially influenced by treatment with nocodazole. Treatment with latrunculin, by contrast, was effective at blocking the recycling of the β₂ARs internalized in response to β-agonist. Thus, counterregulation by insulin and agonist-induced internalization both provoke a massive sequestration of these GPCRs from the cell surface to perinuclear locales in cells, but make use of very different cytoskeletal elements both to internalize and to recycle the receptors back to the cell surface.

Analysis of microtubules and actin microfilaments in cells challenged with β-adrenergic agonist or insulin indicated that some level of rearrangement occurs in response to stimulated cells with these agents. For β-agonist, it seems that the elevation of cyclic AMP may be responsible for the changes in microtubule architecture. For the insulin-stimulated cells, the response may be more difficult to define, because the downstream signaling for insulin, unlike β-adrenergic agonist (Morris and Malbon, 1999), is populated with the mitogen-activated protein kinase cascades and many protein kinases and phosphatases whose activities are regulated by insulin (Pessin and Saltiel, 2000; Czech and Corvera, 1999; Saltiel and Kahn, 2001). The β₂AR is a member of a multivalent signaling complex composed of the receptor in combination with AKAP250 (gravin), protein kinases A and C, Src and protein phosphatase 2B, and perhaps transiently with other signaling elements such as the heterotrimeric G-protein Gs. It has been shown that AKAP250 possesses an F-actin binding motif (Gelman, 2002) and that microtubules can bind to Gs (Wang et al., 1990). As tantalizing as these speculations may be, much work will be required to elucidate the molecular details and partners involved in the trafficking of GPCRs in response to agonist, as well as to receptor tyrosine kinases.

We acknowledge the support from United States Public Health Service Grants from the NIDKD, National Institutes of Health.