Protein Kinase A and G Protein-coupled Receptor Kinase Phosphorylation Mediates (cid:1) -1 Adrenergic Receptor Endocytosis through Different Pathways*

Agonist-induced phosphorylation of (cid:1) -adrenergic receptors ( (cid:1) ARs) by G protein-coupled receptor kinases (GRKs) results in their desensitization followed by internalization. Whether protein kinase A (PKA)-mediated phosphorylation of (cid:1) ARs, particularly the (cid:1) 1 AR subtype, can also trigger internalization is currently not known. To test this, we cloned the mouse wild type (cid:1) 1 AR (WT (cid:1) 1 AR) and created 3 mutants lacking, respectively: the putative PKA phosphorylation sites (PKA (cid:2) (cid:1) 1 AR), the putative GRK phosphorylation sites (GRK (cid:2) (cid:1) 1 AR), and both sets of phosphorylation sites (PKA (cid:2) / GRK (cid:2) (cid:1) 1 AR). Following agonist stimulation, both PKA (cid:2) (cid:1) 1 AR and GRK (cid:2) (cid:1) 1 AR mutants showed comparable increases in phosphorylation and desensitization. Saturating concentrations of agonist induced only 50% internalization of either mutant compared with wild type, suggesting that both PKA and GRK

internalization (12). Although mechanisms of phosphorylation, desensitization, and internalization by GRKs have been well studied for the ␤ 2 AR (4 -8), little is known of the role that PKA-mediated phosphorylation plays in the internalization of ␤ARs, particularly the ␤ 1 AR.
GPCRs can internalize via at least two distinct pathways, namely clathrin-coated pits and caveolae. Although very different structurally, clathrin-coated pits and caveolae both serve as microdomains, which, in addition to functioning as transport machinery, also serve as platforms for integrating the cell's signal-transduction pathways (13,14). These membrane domains serve to facilitate cross-talk between different proteins from a particular signaling pathway contained within these distinct regions (15). Proteins to be transported are now thought to have "molecular addresses" located in the amino acid sequences of their cytoplasmic tail regions or to contain a binding site for a particular adaptor protein that directs the molecule to a particular cellular domain (16,17). In this regard, although a number of studies have demonstrated a critical role for GRK phosphorylation and ␤-arrestin binding in the process of clathrin-mediated internalization of the ␤ 2 AR (5), the molecular mechanisms that are involved in the internalization of the ␤ 1 AR are less known. The aim of the present study was to determine the specific role of PKA-and GRK-catalyzed phosphorylation of the ␤ 1 AR in determining the cellular pathway for agonist-promoted receptor internalization.
Plasmid Constructs-We generated mouse wild type ␤ 1 AR (WT␤ 1 AR) as previously described (18). Three different mutants lacking, respectively: the putative PKA phosphorylation consensus sites (PKA Ϫ ␤ 1 AR), the putative GRK phosphorylation sites (GRK Ϫ ␤ 1 AR), and both sets of sites (PKA Ϫ /GRK Ϫ ␤ 1 AR) (Fig. 1A) were generated using a combination of primers (for details see Supplemental Material). All recombinant DNA-containing plasmids were verified for sequence authenticity and subcloned into mammalian expression vectors.
Mammalian Cell Culture and Transfection-HEK 293 cells were maintained as previously described (18). The evening before transfection, 4 ϫ 10 6 cells were plated per 75-cm 2 flasks. These cells were transfected on day 1 by FuGENE6 TM method (Roche Applied Science). For all experiments, each plate received 5 g of total DNA, comprising 0.25 g of the mutant ␤ 1 AR or 0.5 g of the WT␤ 1 AR, with the balance comprising just pRK5 DNA. For the ␤-arrestin2 recruitment studies, each plate received 0.5 g of ␤ 1 AR along with a 10-fold molar excess GFP-␤-arrestin2 plasmid. Cells were split on day 2 into assay dishes as follows: for phosphorylation assays, and cAMP assays, 3 ϫ 10 6 cells/ 100-mm dish, for ␤-arrestin2 recruitment, and confocal studies 1 ϫ 10 6 cells/9.6-cm 2 well. Assays were performed on day 3.
HEK 293 cells were transfected with cDNA containing WT␤ 1 AR, PKA Ϫ ␤ 1 AR, GRK Ϫ ␤ 1 AR, and PKA Ϫ /GRK Ϫ ␤ 1 AR to create stable cell lines. 48 h after transfection cells were selected by the addition of Geneticin at a concentration of 0.5 mg/ml. Expression of receptor was determined by radioligand binding assays. WT␤ 1 AR 551 Ϯ 63, These stable cells were used in ELISA assays, confocal experiments, and immunoblotting studies.
Intact Cell Phosphorylation-Intact cell phosphorylation was per-formed as previously described (19). Briefly, assays were performed at 37°C, in phosphate-free Dulbecco's modified Eagle's medium, 20 mM HEPES, pH 7.4. Labeling was conducted for 1 h in medium containing 100 Ci of 32 P i /ml (inorganic phosphate, P i ). In PKA inhibition experiments, the labeling medium contained 0.04% Me 2 SO followed by incubation with ISO (10 M) for 10 min in the appropriate dishes. The ␤ 1 AR density (pmol/mg of whole cell protein) of each transfected cell line was determined on a non-radioactive aliquot of each transfected cell population by 125 I-cyanopindolol. Equivalent amounts of ␤ 1 AR were immunoprecipitated from each sample and were resolved by SDS-PAGE in 10% gels. Dried gels were subjected to autoradiography and analyzed quantitatively with an Amersham Biosciences PhosphorImager. Radioligand Binding Assays-␤ 1 AR expression was evaluated by 125 I-cyanopindolol binding studies done at 13°C for 3 h as previously described (18). Bound ligand was separated on glass fiber filters (Whatman, GF/C) by vacuum filtration. The filters were washed three times with cold wash buffer (10 mM Tris, 5 mM EDTA (pH 7.4)) and counted in a scintillation counter. Protein concentration was measured with a Bio-Rad DC protein assay kit with bovine serum albumin as the standard. Receptor sequestration after a 30-min exposure to agonist (10 M ISO) was defined as the loss of 125 I-cyanopindolol binding displaced by CGP-12177.
Receptor Binding by ELISA-To determine agonist concentration dependence for internalization of the wild type and three mutants, the receptor number was measured by the ELISA method, as described previously (20,21). Briefly, stable cell lines expressing the WT␤ 1 AR or the PKA Ϫ or GRK Ϫ mutants were plated onto 24-well tissue culture dishes. To improve adhesion of cells to plastic, the wells were treated with 20 g/ml poly-D-lysine in PBS before plating. 24 h after seeding, cells were incubated with serum-free minimum essential media (MEM) for 10 min at 37°C. Different concentrations of isoproterenol (0.01-10 M) were added for 30 min at 37°C. Reactions were stopped by removing the culture medium followed by fixing cells in 4% formaldehyde/PBS for 5 min at room temperature. Cells were washed three times with PBS followed by blocking with 1% BSA in PBS for 45 min. Mouse monoclonal M2 anti-FLAG IgG was added at a dilution of 1:1000 for 1 h, followed by three subsequent washes with PBS. The samples were then briefly re-blocked, incubated with goat anti-mouse-conjugated alkaline phosphatase at a concentration of 1:1000 in PBS/BSA for 1 h, and washed three times with PBS before the addition of a colorimetric alkaline phosphatase substrate. When adequate color change was reached, 100-l samples were taken for colorimetric readings at 405 nm using a scanning multiwell spectrophotometer. Non-transfected cells were studied concurrently to determine background signal and all experiments were done in triplicate.
GFP-␤-arrestin2 Translocation in Live Cells-GFP-␤-arrestin2 translocation was visualized in real-time on a 37°C heated stage Zeiss laser scanning microscope (LSM-510) as previously described (22). Cells expressing either the WT␤ 1 AR or one of the mutants and GFP-␤-arrestin2 were stimulated with 10 M ISO in serum-free medium buffered with 10 mM HEPES. Images were collected sequentially every minute for a period of 10 min using a single line excitation filter 488 nm and emission filters at 505-550 nm. For each single experiment, quantitation was performed in the image in which maximal agonist-dependent ␤-arrestin2 translocation occurred. ␤-Arrestin2 translocation was calculated and expressed as agonist-promoted percent loss of the green fluorescent color from the cytosol due to GFP-␤-arrestin2 expression.
Intact Cell ␤-AR Desensitization: cAMP Assay-HEK 293 cells were incubated for 1 h in serum-free MEM (10 mM HEPES, pH 7.4) prior to the assay. Two identical sets of 100-mm plates were set up; one set was used as controls and the other for desensitization assays. Cells used for desensitization were exposed to 10 M dobutamine for 5 min at 37°C and were then washed with serum-free MEM (10 mM HEPES, pH 7.4). Control cells were washed in an identical manner. Both sets of cells were then replaced with assay buffer (serum-free MEM, 10 mM HEPES, 1 mM IBMX, 100 nM ICI 118,551) and exposed to ␤ 1 -agonist, dobutamine (0.1-10 M) for 20 min at 37°C. Reactions were stopped by aspiration of the assay buffer and addition of 200 l of Tris-EDTA solution (0.05 M Tris, 4 mM EDTA, supplied with cyclic AMP ( 3 H) assay kit) to each plate. Plates were frozen at Ϫ80°C for subsequent processing of cells. Cells were scraped and collected, boiled for 10 min and placed on ice. Samples were then centrifuged at 15,000 ϫ g for 15 min, and supernatant was transferred to a tube for cAMP measurement. cAMP was measured using the procedure outlined in the assay kit. The amount of protein per fraction was determined using a dye-binding protein assay kit.
Confocal Microscopy-Confocal microscopy was carried out as previously described (18). HEK 293 cells were transfected with the plasmids containing cDNAs encoding either the FLAG-WT␤ 1 AR, FLAG-␤ 2 AR, or one of the mutant FLAG-␤ 1 ARs as well as pRK5. Live cells were incubated in the absence or presence of filipin, ␤-CD, cholesterol, MDC along with sucrose, H-89, and GFX for the indicated times and stimulated with ISO (10 M) for 30 min. All incubations were carried out at 37°C. Staining of FLAG-tagged receptor was carried out as previously described (23). Transferrin uptake was carried out as described previously (24). All samples were visualized under the Olympus 1X70 laser scanning confocal microscope, using single sequential line excitation filters of 568 nm and emission filters of 585 nm for Texas Red detection. Images were viewed using Fluoview software and processed using Adobe Illustrator 9.0.1 and Adobe Photoshop 6.0.1.
Immunoblotting-Pretreatment of cells with inhibitors and stimulation with agonist were carried out at 37°C in serum starvation medium as described in the figure legends. After stimulation, cells were lysed directly with 100 l/well Laemmli sample buffer and proteins (30 g/ lane) were resolved by SDS-PAGE. Phosphorylation of Erk1/2 was detected by protein immunoblotting using a 1:1000 dilution of rabbit polyclonal phospho-specific mitogen-activated protein kinase IgG (New England BioLabs) with horseradish-peroxidase conjugated donkey antirabbit IgG as secondary antibody. Vasodilator-and A kinase-stimulated phosphoprotein (VASP) was detected by using 1:1000 anti-FLAG M2 antibody with horseradish-peroxidase conjugated anti-mouse IgG as secondary antibody. Blots were developed in ECL reagents for 1 min.

Agonist-induced Phosphorylation of the ␤ 1 AR in Intact 293
Cells-To study the role of PKA-and GRK-mediated phosphorylation of the ␤ 1 AR in an agonist-dependent manner, we used HEK 293 cells transiently transfected with plasmids containing the WT␤ 1 AR, PKA Ϫ ␤ 1 AR, GRK Ϫ ␤ 1 AR, or PKA Ϫ / GRK Ϫ ␤ 1 AR cDNAs. In unstimulated cells, the WT␤ 1 AR exists as a phosphoprotein migrating with a molecular mass of ϳ70 kDa (Fig. 1B). Upon stimulation, phosphorylation of the WT␤ 1 AR as well as PKA Ϫ ␤ 1 AR and GRK Ϫ ␤ 1 AR mutants increased ϳ2-fold above basal levels (Fig. 1B). However, when both the PKA and GRK sites were mutated (PKA Ϫ / GRK Ϫ ␤ 1 AR), no agonist-dependent phosphorylation of the receptor was observed.
Agonist-induced Desensitization of the ␤ 1 AR in Intact 293 Cells-To determine the role of PKA and GRK phosphorylation in the desensitization of the ␤ 1 AR we measured cAMP accumulation in HEK 293 cells transiently transfected with the various ␤ 1 AR mutants. The PKA and GRK phosphorylation mutants caused a 6-to 7-fold increase in cAMP, similar to the wild type receptor, indicating that they were fully coupled to Gs (Fig. 2). In contrast, stimulation of the PKA Ϫ /GRK Ϫ ␤ 1 AR mutant produced only a 3-fold increase in cAMP, suggesting that removal of all of the sites resulted in a general impairment of receptor function (Fig. 2). Furthermore, cells overexpressing the WT␤ 1 AR, PKA Ϫ ␤ 1 AR, or GRK Ϫ ␤ 1 AR showed ϳ70% desensitization measured as a reduction in catecholamine-induced cAMP generation on repeated exposure to the ␤1-selective agonist dobutamine. In contrast to the wild type receptor, which fully desensitized even at agonist concentrations that produced a less than maximal increase in intracellular cAMP, cells transfected with the PKA Ϫ /GRK Ϫ ␤ 1 AR were not significantly desensitized by pretreatment with agonist. Consistent with the ␤ 2 AR model of PKA-mediated heterologous desensitization and GRK-mediated homologous desensitization, these data suggest that ␤ 1 ARs could become fully desensitized by either phosphorylation of PKA and/or GRK sites and only when all the phosphorylation sites were removed did a dramatic reduction in agonist promoted desensitization occur.
Both PKA and GRKs Mediate Agonist-induced ␤ 1 AR Internalization-As opposed to desensitization, agonist-induced internalization of ␤ 2 ARs is thought to be mediated predominantly through GRK phosphorylation and ␤-arrestin binding.
To determine the contribution of PKA and GRK phosphoryla-tion to agonist promoted ␤ 1 AR internalization, we measured the loss of ␤ 1 ARs from the cell surface in response to isoproterenol by radioligand binding. As shown in Fig. 3A, agonist stimulation resulted in a marked loss of WT␤ 1 ARs from the cell surface. In contrast, at saturating agonist concentrations the extent of agonist-induced sequestration of both the PKA Ϫ ␤ 1 AR and GRK Ϫ ␤ 1 AR was about half that of the WT␤ 1 AR. The double mutant exhibited minimal agonist-induced internalization. These results contrast with the desensitization data (Fig. 2), which indicate that either PKA or GRK sites alone were sufficient for full receptor desensitization, and suggest that both PKA and GRK phosphorylation are required for full internalization of the ␤ 1 AR. To further examine the role of PKA and GRK phosphorylation in ␤ 1 AR receptor endocytosis, we determined the dose dependence of isoproterenol-stimulated sequestration of the wild type and mutant receptors (Fig. 3B). Whereas WT␤ 1 AR and PKA Ϫ ␤ 1 AR exhibited half-maximal internalization at similar agonist concentrations, the EC 50 for GRK Ϫ ␤ 1 AR internalization was ϳ10-fold higher (EC 50 values for WT␤ 1 AR ϭ 56 nM, PKA Ϫ ␤ 1 AR ϭ 46 nM, GRK Ϫ ␤ 1 AR ϭ 632 nM, and PKA Ϫ /GRK Ϫ ␤ 1 AR ϭ 534 nM). Thus, two distinct mechanisms appear to contribute additively to ␤ 1 AR sequestration, one a GRK site-dependent mechanism that predominates at lower agonist concentrations, and the other a PKA site-dependent mechanism that accounts for approximately half of the agonist-dependent ␤ 1 AR sequestration at higher agonist concentrations. Importantly, the reduced efficiency of sequestration for the ␤ 1 AR mutants was observed over a broad range of agonist concentrations (Fig. 3B).
PKA-mediated ␤ 1 AR Internalization Does Not Involve ␤-Arrestin Recruitment to the Membrane-Previous studies have shown that GRK-mediated desensitization involves recruitment of ␤-arrestin to the phosphorylated ␤ 2 AR (5). We therefore studied the ability of the wild type and mutant receptors to recruit GFP-␤-arrestin2 to the membrane. As shown in Fig. 4A, agonist stimulation of the WT␤ 1 AR resulted in marked translocation of ␤-arrestin from the cytosol to the plasma membrane. In the absence of PKA sites (PKA Ϫ ␤ 1 AR), ␤-arrestin recruitment was significantly increased. In contrast, cells expressing the GRK Ϫ ␤ 1 AR mutant showed marked impairment of ␤-arrestin recruitment to the membrane (Fig. 4, A and B). Removal of all phosphorylation sites in the double mutant resulted in very low ␤-arrestin recruitment comparable to that of the GRK Ϫ ␤ 1 AR (Fig. 4, A and B). Importantly, the ability of the GRK Ϫ ␤ 1 AR (PKA sites intact) to internalize (Fig. 3), despite the marked reduction of ␤-arrestin recruitment (Fig. 4), suggests that internalization of the ␤ 1 AR can proceed via multiple mechanisms.

Phosphorylation of the ␤ 1 AR Directs Internalization via Two
Separate Pathways-␤-Arrestin recruitment has been shown to be important for targeting the ␤ 2 AR to clathrin-coated pits for internalization (25). Because the GRK Ϫ ␤ 1 AR mutant showed marked impairment of ␤-arrestin recruitment compared with the PKA Ϫ ␤ 1 AR mutant, we sought to determine the pathway(s) of internalization following PKA-and GRK-mediated phosphorylation of the ␤ 1 AR. In these experiments cells expressing FLAG epitope-tagged WT␤ 1 AR, PKA Ϫ ␤ 1 AR, GRK Ϫ ␤ 1 AR, or ␤ 2 AR (used as a control) were used for the internalization studies assessed by laser scanning confocal microscopy. Because caveolae have been reported to play a significant role in both the signaling and internalization of several GPCRs, cells were pretreated with two different caveolae pathway inhibitors: filipin, a sterol-binding agent, and ␤-CD, which causes disruption of the caveolae by cholesterol depletion (26). Agonist treatment alone resulted in the formation of intracellular aggregates with complete loss of membrane fluorescence indicating the accumulation of Texas Red-labeled receptors in endosomal vesicles (Fig. 5, panels b, e, h, and k). Pretreatment with filipin did not block internalization of the ␤ 2 AR, WT␤ 1 AR, or the PKA Ϫ ␤ 1 AR (Fig. 5, panels c, f, and i). In marked contrast, filipin prevented redistribution of membrane fluorescence and blocked the formation of intracellular aggregates for the GRK Ϫ ␤ 1 AR mutant (PKA sites intact) (Fig. 5, panel l). Using ␤-CD, another inhibitor of caveolar function, we observed similar results. Again, only internalization of the GRK Ϫ ␤ 1 AR mutant was blocked (Fig. 6A, panel l). Importantly, addition of cholesterol to the cells treated with ␤-CD was able to reverse the inhibitory effect on the GRK Ϫ ␤ 1 AR mutant upon agonist stimulation (Fig. 6A, panel m).
Because caveolae contain many of the proteins involved in GPCR signal transduction, they may function as signaling microdomains as well as locations for receptor internalization. Therefore to determine whether the caveolae inhibitors filipin and ␤-CD indirectly block internalization of the GRK Ϫ ␤ 1 AR mutant by preventing PKA activation, we used a highly sensi- For each experiment, an average of 3-7 cells was used for the quantitative analysis. In unstimulated cells expressing both the ␤ 1 AR and GFP-␤-arrestin2, fluorescence is distributed uniformly throughout the cytoplasm. In presence of 10 M ISO, 53% of GFP-␤-arrestin2 translocates to WT␤ 1 AR displaying a punctated distribution at the plasma membrane. PKA Ϫ ␤ 1 AR, which shows impaired agonist-promoted sequestration, recruited 85% of the cytosolic GFP-␤-arrestin2. The GRK Ϫ ␤ 1 AR and PKA Ϫ /GRK Ϫ ␤ 1 AR showed impairment in GFP-␤-arrestin2 demonstrating that ␤-arrestin recruitment was only due to phosphorylation of the putative GRK sites. *, p Ͻ 0.0001 versus GRK Ϫ ␤ 1 AR and PKA Ϫ /GRK Ϫ ␤ 1 AR; †, p Ͻ 0.0001 versus WT␤ 1 AR (one-factor ANOVA).

FIG. 5. Inhibition of caveolae pathway in cells by filipin prevents internalization of the ␤ 1 AR. Each panel shows representative images from 3-4 experiments.
Confocal microscopy was used to visualize HEK 293 cells transfected with the plasmids containing cDNAs that were FLAG-tagged encoding for ␤ 2 AR, WT␤ 1 AR, PKA Ϫ ␤ 1 AR, and GRK Ϫ ␤ 1 AR. Cells were serum-starved for 3 h and incubated in the absence or presence of 1 g/ml filipin, for 1 h. Following stimulation with ISO (10 M) for 30 min, cells were fixed and stained with Texas Red. Unstimulated cells in panels a, d, g, and j showed distribution of ␤AR at the plasma membrane. Following agonist stimulation, all ␤ARs internalized into cytoplasmic puncta with complete loss of membrane fluorescence (panels b, e, h, and k). Pretreatment with filipin, a sterolbinding agent that disrupts caveolae, prevented internalization of the GRK Ϫ ␤ 1 AR (PKA sites intact) (panel l). Similar treatment of the ␤ 2 AR, WT␤ 1 AR, and PKA Ϫ ␤ 1 AR had no effect indicating internalization by a caveolae-independent pathway (panels c, f, and i). Treatment by filipin alone had no effect on unstimulated, transfected cells (data not shown).
tive functional assay to determine the endogenous activity of cyclic nucleotide-dependent protein kinases in response to agonist. VASP (vasodilator-and A kinase-stimulated phosphoprotein) is a major substrate of cyclic nucleotide-dependent protein kinases that undergoes an electrophoretic mobility shift from 46 to 50 kDa upon phosphorylation (27,28). As shown in Fig.  6B, agonist stimulation by ISO induced marked phosphorylation of VASP. Pretreatment of cells with either filipin or ␤-CD showed a similar shift to that of ISO alone indicating that the caveolae inhibitors have no effect on the activation of PKA. Specificity of PKA activity was confirmed by pretreatment of cells to the PKA inhibitor H-89 that blocked PKA activity and prevented phosphorylation of VASP (Fig. 6B).
To test whether the ␤ 1 AR utilizes a clathrin-coated pit mechanism for internalization, we treated cells with clathrin inhibitors. Pretreatment with the clathrin inhibitors MDC and sucrose (29, 30) resulted in 80 -90% inhibition of internalization for WT␤ 1 ARs and PKA Ϫ ␤ 1 ARs (Fig. 7A, panel c and f) without an apparent effect on internalization of the GRK Ϫ ␤ 1 AR mutant (Fig. 7A, panel i). To confirm the specificity of MDC and sucrose to inhibit clathrin-mediated internalization, we assessed their effect on internalization of transferrin, a receptor that was shown to internalize via clathrin-coated vesicles (24). As shown in Fig. 7B, pretreatment with MDC and sucrose abolished transferrin internalization confirming that clathrin-mediated processes were effectively inhibited. The finding that internalization of the GRK Ϫ ␤ 1 AR mutant can be blocked by two different caveolae pathway inhibitors but not by clathrin inhibitors supports our hypothesis that phosphorylation of the ␤ 1 AR directs the receptor to internalize via a caveolae-dependent pathway (Fig. 5, panel l, and Fig. 6, panel l).
To further demonstrate a role for PKA phosphorylation in internalization of the ␤ 1 AR, we pretreated cells with the selective PKA inhibitor H-89 followed by agonist stimulation. Importantly, FIG. 6. Inhibition of caveolae pathway in cells by ␤-CD prevents internalization of the ␤ 1 AR. A, each panel shows representative images from 3-4 experiments. Confocal microscopy was used to visualize HEK 293 cells transfected with the plasmids containing cDNAs that were FLAG-tagged encoding for ␤ 2 AR, WT␤ 1 AR, PKA Ϫ ␤ 1 AR, and GRK Ϫ ␤ 1 AR. Cells were serum-starved for 3 h and incubated in the absence or presence of 2% ␤-CD for 1 h. In a separate set of plates, 160 g/ml cholesterol was added for 1 h after treatment with ␤-CD. Following stimulation with isoproterenol (10 M) for 30 min, cells were fixed and stained with Texas Red. Unstimulated cells in panels a, d, g, and j show distribution of ␤AR at the plasma membrane. Following agonist stimulation, all ␤ARs are internalized into cytoplasmic puncta with complete loss of membrane fluorescence (panels b, e, h, and k). Pretreatment with ␤-CD disrupts caveolae and prevents internalization of the GRK Ϫ ␤ 1 AR (PKA sites intact) (panel l). Addition of cholesterol to the cells treated with ␤-CD reversed the inhibitory effect on the GRK Ϫ ␤ 1 AR mutant and shows internalization upon agonist stimulation (panel m). Similar treatment by ␤-CD on ␤ 2 AR, WT␤ 1 AR, and PKA Ϫ ␤ 1 AR shows no effect on internalization (panels c, f, and i). Treatment by ␤-CD alone had no effect on unstimulated transfected cells (data not shown). B, filipin and ␤-CD do not affect endogenous PKA activity. HEK 293 cells stably expressing WT␤ 1 ARs, GRK Ϫ ␤ 1 ARs, and PKA Ϫ ␤ 1 ARs were transiently transfected with FLAG-VASP cDNA (2 g/10-cm culture dish). Cells were serum-starved for 24 h in MEM and 0.1% BSA and incubated in the absence or presence of the caveolae inhibitors filipin and ␤-CD for indicated times and stimulated with ISO (10 M) for 5 min. After agonist stimulation, cell lysis was followed by immunoblotting of cell with FLAG antibody (27,28). PKA phosphorylates VASP and is visualized as a phosphorylation-induced electrophoretic mobility shift from 46 to 50 kDa by SDS-PAGE (27,28). PKA-mediated phosphorylation of VASP was confirmed by preincubation with the PKA specific inhibitor H-89 (10 M) for 15 min. the GRK Ϫ ␤ 1 AR mutant (PKA sites intact) showed no internalization upon agonist stimulation (Fig. 8, panel i), whereas robust agonist-mediated internalization of the PKA Ϫ ␤ 1 AR occurred (Fig. 8, panel f). In the case of the WT␤ 1 AR, puncta were visible on the membrane, indicating partial inhibition likely due to the presence of both phosphorylation sites (Fig. 8, panel c). The lack of an H-89 effect on the internalization of PKA Ϫ ␤ 1 AR indicates that other potential PKA phosphorylation sites were not present on the PKA Ϫ ␤ 1 AR mutant. Taken together these data demonstrate that PKA-mediated phosphorylation of the ␤ 1 AR directs internalization via a caveolae pathway, whereas GRK-mediated phosphorylation of the ␤ 1 AR directs internalization through the clathrin-coated pit machinery.
PKC-mediated Phosphorylation Does Not Play a Role in ␤ 1 AR Internalization-To determine whether PKC-mediated phosphorylation directs ␤ 1 AR internalization through a mechanism of heterologous desensitization, cells stably expressing either the FLAG epitope-tagged WT␤ 1 AR, PKA Ϫ ␤ 1 AR, or GRK Ϫ ␤ 1 AR were used in the internalization studies. Pretreatment of WT␤ 1 AR-, PKA Ϫ ␤ 1 AR-, or GRK Ϫ ␤ 1 AR-expressing cells with the PKC inhibitor GFX, followed by agonist stimulation resulted in the formation of intracellular aggregates with complete loss of membrane fluorescence (Fig. 9, panels d, h, and l), similar to cells treated with agonist alone (Fig. 9, panels c, g,  and k). The lack of a GFX effect on cells expressing the wild type or the phosphorylation mutant receptors indicates that PKC-mediated phosphorylation does not play a role in ␤ 1 AR internalization under these conditions. To confirm the specificity of GFX to inhibit PKC-mediated phosphorylation (Fig. 9B), we assessed its effect on cells expressing the WT␤ 1 AR by immunoblotting for phospho-ERK alongside confocal microscopy studies. Whereas pretreatment with PMA alone caused a large increase in the activation of ERK, the pretreatment of GFX with PMA completely blocked the ERK activation. No differences in ERK activation were observed between treatment of ISO alone or ISO with GFX. Taken together the results demonstrated that, first, GFX was able to inhibit endogenous PKC in the cells, and, second that PKC-mediated phosphorylation does not play a role in ␤ 1 AR signaling. DISCUSSION In this study we demonstrate that, in addition to the established role of GRKs in the process of receptor internalization, PKA-mediated phosphorylation plays a critical role in agonistinduced internalization of the ␤ 1 AR. Furthermore, although GRK-mediated phosphorylation directs internalization through a clathrin-coated pit pathway, PKA-mediated phosphorylation directs internalization via a caveolae pathway. These data contrast dramatically with previous findings obtained using PKA and GRK mutants of the ␤ 2 AR. Although both PKA and GRK phosphorylation contribute to desensitization of the ␤ 2 AR (4, 12), PKA phosphorylation does not play a significant role in endocytosis of this receptor (12). Our data suggest that at maximally efficacious concentration of agonist, ␤ 1 AR endocytosis occurs via both clathrin-coated pits and caveolae. Each pathway contributes approximately to half of the observed response, and the two pathways are additive. Thus, endocytosis through clathrin-coated pits cannot compensate for loss of the caveolar pathway, and vice versa.  ). B, pretreatment with MDC and sucrose abolishes transferrin internalization confirming that clathrin-mediated processes are effectively inhibited. Treatment by MDC combined with sucrose alone had no effect on unstimulated transfected cells (data not shown). Treatment by either MDC or sucrose alone inhibited internalization of the WT␤ 1 AR and PKA Ϫ ␤ 1 AR by ϳ40 -50% without an effect on GRK Ϫ ␤ 1 AR internalization (data not shown). Whether this indicates that ␤ 1 ARs naturally exist in two distinct pools, inside and outside of caveolae, and that endocytosis of each pool is independently regulated, is unclear. Because the EC 50 for isoproterenol-stimulated endocytosis of the wild type and PKA Ϫ ␤ 1 AR is ϳ10-fold lower than that for the GRK Ϫ ␤ 1 AR, our data suggest that clathrin-mediated endocytosis is the preferred mechanism for ␤ 1 AR internalization at low agonist concentrations, whereas at higher agonist concentrations, both pathways contribute equally to endocytosis. Importantly, however, the GRK Ϫ ␤ 1 AR still undergoes significant internalization at low agonist concentrations, suggesting that the clathrin-independent mechanism of ␤ 1 AR internalization does contribute to receptor endocytosis over a wide range of agonist concentrations. Data from this study also highlights other differences in the internalization mechanism for the two ␤AR subtypes in that PKCmediated phosphorylation does not play a role in the internalization of the ␤ 1 AR. This is in contrast to previous findings that showed contribution of PKC-mediated phosphorylation to internalization of the ␤ 2 AR (10).
A number of previous studies have found that other GPCRs, including bradykinin B1, cholecystokinin, endothelin subtype A, and muscarinic acetylcholine receptors localize in caveolae (31)(32)(33)(34)(35). Indeed, both ␤ 1 ARs and ␤ 2 ARs have also previously been shown to localize in caveolar microdomains (36 -41), including endogenous ␤ 1 ARs in neonatal rat cardiomyocytes (37,38). Our study adds to this literature by showing that ␤ 1 ARs lacking phosphorylation sites for GRK internalize through caveolae. Our results are also in accordance with findings obtained for the endothelin receptor, where agonist-induced internalization was found to proceed either via clathrin-coated pits or caveolae, depending on the oxidative state of cell surface cholesterol (42). A similar scheme has also been shown for the cholecystokinin receptor, which can utilize either pathway for internalization but internalizes predominantly via the clathrin-coated pit pathway (32).
Both caveolae and clathrin-coated vesicles serve as scaffolds that integrate signal-transduction complexes, providing microdomains for cross-talk between specific signaling molecules (15,16,43). Several studies indicate that proteins involved in adrenergic signaling (e.g. G␤␥, G␣ s , adenylate cyclase, ␤ 1 ARs, and ␤ 2 ARs) co-localize within the same microdomains (37,39), possibly because they contain an "address" for that specific domain (16). The cytoplasmic tail of the receptor is a potential region that likely contains an address site (16). Indeed, scaffold proteins containing PDZ domains have been shown to associate with the C termini of ␤ARs (17), and these interactions can be disrupted through phosphorylation by GRK5 (44,45). Thus cellular signaling and trafficking is precisely controlled by phosphorylation-dependent regulation of the receptor and its association with a variety of scaffold proteins (17). For example, a recent study in cardiac neonatal myocytes has shown that the PDZ motif at the C terminus of the ␤ 1 AR is responsible for its limited internalization and that the mutation of this domain increases internalization to levels similar to those observed with the ␤ 2 AR (31). Indeed, we show here that differential phosphorylation of the ␤ 1 AR plays a critical role in determining its internalization pathway, indicating that the site of phosphorylation may serve as molecular address that directs receptor internalization.
Determination of the internalization pathway utilized by each receptor in this study was based on specific inhibitors for either caveolae or clathrin-coated pits. Specifically, cholesterol depletion by ␤-CD was one of the reagents used to demonstrate caveolae-mediated internalization. In contrast, several studies have suggested that disruption of cholesterol by this agent can interfere with clathrin-coated pit formation (46 -48). However, our data supporting internalization of the GRK Ϫ ␤ 1 AR mutant through caveolae are based on: 1) the use of two different caveolae inhibitors, filipin that binds sterols, and ␤-CD that depletes cholesterol (26); 2) reintroduction of cholesterol to cells depleted of cholesterol by ␤-CD rescued internalization of the GRK Ϫ ␤ 1 AR mutant (49); and 3) transferrin receptor internalization insensitivity to filipin (46). Finally, we showed that the use of the caveolae inhibitors filipin and ␤-CD did not affect the ability of the various mutant receptors to increase endogenous PKA activity in response to agonist.
In conclusion, we demonstrate that PKA-mediated phosphorylation plays an important role in agonist-induced internalization of the ␤ 1 AR in addition to the recognized role of GRKs in receptor internalization. Furthermore, we have shown that, although either set of phosphorylation sites is sufficient to induce desensitization of the ␤ 1 AR, both PKA and GRK sites are required to accomplish complete internalization. These sites may serve as "molecular addresses" that target the receptor into microdomains thus allowing different signal transduction pathways to be triggered. FIG. 9. PKC-mediated phosphorylation does not play a role in ␤ 1 AR internalization. A, each panel shows representative images from 3-4 experiments. Confocal microscopy was used to visualize HEK 293 cells stably expressing the FLAG-tagged WT␤ 1 AR, GRK Ϫ ␤ 1 AR, and PKA Ϫ ␤ 1 AR. Live cells were serum-starved for 3 h and incubated in the absence or presence the PKC-selective inhibitor GFX (2 M) for 30 min. Following stimulation with isoproterenol (10 M) for 30 min, cells were fixed and stained with Texas Red. Unstimulated cells in panels a, e, and i show distribution of receptors at the plasma membrane. Pretreatment with GFX had no effect on agonist-promoted internalization of the ␤ 1 AR mutants as shown by the formation of intracellular aggregates and complete loss of membrane fluorescence (panels d, h, and l). B, to confirm activity and specificity of GFX, HEK 293 cells stably expressing the FLAG-tagged WT␤ 1 AR were plated in a 12-well tissue culture dish and incubated with GFX prior to stimulation with either PMA (0.1 M) or ISO (10 M, 5 min) (50). After agonist stimulation, cells were lysed and resolved by SDS-PAGE and immunoblotted with the phospho-ERK antibodies.