(cid:1) -Arrestin-mediated ADP-ribosylation Factor 6 Activation and (cid:1) 2 -Adrenergic Receptor Endocytosis*

(cid:1) -Arrestins are multifunctional adaptor proteins known to regulate internalization of agonist-stimulated G protein-coupled receptors by linking them to endocytic proteins such as clathrin and AP-2. Here we de-scribe a previously unappreciated mechanism by which (cid:1) -arrestin orchestrates the process of receptor endocytosis through the activation of ADP-ribosylation factor 6 (ARF6), a small GTP-binding protein. Involvement of ARF6 in the endocytic process is demonstrated by the ability of GTP-binding defective and GTP hydrolysis-deficient mutants to inhibit internalization of the (cid:1) 2 - adrenergic receptor. The importance of regulation of ARF6 function is shown by the ability of the ARF GTP-ase-activating protein GIT1 to inhibit and of the ARF nucleotide exchange factor, ARNO, to enhance receptor endocytosis. Endogenous (cid:1) -arrestin is found in complex with ARNO. Upon agonist stimulation of the receptor, (cid:1) -arrestin also interacts with the GDP-liganded form of ARF6, thereby facilitating ARNO-promoted GTP loading and activation of the G protein. Thus, the agonist-driven formation of a complex including (cid:1)

␤-Arrestins are multifunctional adaptor proteins known to regulate internalization of agonist-stimulated G protein-coupled receptors by linking them to endocytic proteins such as clathrin and AP-2. Here we describe a previously unappreciated mechanism by which ␤-arrestin orchestrates the process of receptor endocytosis through the activation of ADP-ribosylation factor 6 (ARF6), a small GTP-binding protein. Involvement of ARF6 in the endocytic process is demonstrated by the ability of GTP-binding defective and GTP hydrolysisdeficient mutants to inhibit internalization of the ␤ 2adrenergic receptor. The importance of regulation of ARF6 function is shown by the ability of the ARF GTPase-activating protein GIT1 to inhibit and of the ARF nucleotide exchange factor, ARNO, to enhance receptor endocytosis. Endogenous ␤-arrestin is found in complex with ARNO. Upon agonist stimulation of the receptor, ␤-arrestin also interacts with the GDP-liganded form of ARF6, thereby facilitating ARNO-promoted GTP loading and activation of the G protein. Thus, the agonist-driven formation of a complex including ␤-arrestin, ARNO, and ARF6 provides a molecular mechanism that explains how the agonist-stimulated receptor recruits a small G protein necessary for the endocytic process and controls its activation.
Agonist-triggered G protein-coupled receptor (GPCR) 1 endocytosis is a highly regulated process involved in signaling and resensitization of numerous receptors. Agonist-dependent GPCR phosphorylation by G protein-coupled receptor kinases (GRKs) promotes binding of ␤-arrestins to the receptor, which in turn induces desensitization by preventing further receptormediated G protein activation (1). ␤-Arrestins have been shown to regulate GPCR internalization via their direct interaction with clathrin and the ␤ subunit of the clathrin adaptor protein complex AP-2 (2)(3)(4). However, the exact molecular mechanisms regulating the protein-protein interactions involved in clathrin-coated pit formation remain unknown. We have shown recently that a novel GTPase-activating protein for the ADPribosylation factor (ARF) family of small GTP-binding proteins, GIT1, can regulate internalization of the ␤ 2 -adrenergic receptor (␤ 2 AR), as well as other signaling receptors, suggesting a critical role for ARF function in this process (5,6).
The ARF proteins constitute a group of six ubiquitously expressed small GTP-binding proteins. These GTPases are essential components of the molecular machinery that regulates membrane trafficking along endocytic and biosynthetic pathways (7,8). ARF1, the best characterized subtype, is localized to the Golgi complex and is a regulator of vesicle coat recruitment for both COP1-and clathrin-coated vesicles (9 -11). ARF6 is uniquely associated with the plasma membrane (12,13). Activation of ARF proteins is facilitated by guanine nucleotide exchange factors that promote dissociation of bound GDP from the inactive ARF protein followed by binding of GTP to the nucleotide-free ARF. Inactivation of ARF proteins is regulated by GTPase-activating proteins (GAPs). ARF proteins have been shown to regulate various plasma membrane trafficking events such as the constitutive recycling of the transferrin receptor (12), calcium-stimulated exocytosis in chromaffin cells (14), Fc␥ receptor-mediated phagocytosis in macrophages (15), endocytosis at the apical surface of epithelial Madin-Darby canine kidney cells (16), membrane recycling (17,18), and the endothelin-mediated translocation of the GLUT4 glucose transporter (19,20). In addition, ARF6 activation is involved in the remodeling of the actin cytoskeleton (15,21,22). In this study, we examine the involvement of ARF proteins in the internalization process of the ␤ 2 AR and begin to elucidate the molecular mechanisms by which the function of these small GTPbinding proteins is regulated following agonist-stimulation.
Cell Culture and Transfection-HEK 293 cells were maintained and transiently transfected as described previously (5).
Sequestration and Recycling Assays-Receptor sequestration was determined by flow cytometry as described previously (6,28) and defined as the percent of cell surface receptor no longer accessible to extracellular antibodies. Recycling experiments were done similarly except that the agonist was removed after 30 min of stimulation by washing the cells three times with warm media. The cells were placed back at 37°C, and the percent of initial receptors present at the cell surface was assessed 15, 30, and 60 min after agonist removal.
Immunoprecipitation-Immunoprecipitations were done as described previously (5). Briefly, serum-starved HEK 293 cells were stimulated with isoproterenol for indicated times. Dithiobis(succinimidyl propionate) (DSP; Pierce) (25 mM) in phosphate-buffered saline was added to each dish for 20 min, and the cells were lysed in 1 ml of FLAG-radioimmune precipitation assay buffer containing protease inhibitors. The samples were incubated at 4°C for 1 h and spun at 15,000 rpm for 20 min to pellet the particulate fraction. Twenty l was removed for direct immunoblotting of the lysates. Affinity gel FLAG beads (15 l) (Sigma) were added to each tube, and the samples were rotated for 6 h at 4°C. The proteins were eluted into 40 l of SDS-polyacrylamide gel electrophoresis sample buffer containing 5% ␤-mercaptoethanol by heating to 95°C for 10 min. Proteins were resolved on 14% gels and detected by immunoblot analysis.
GST Fusion Proteins and Pull-down Experiments-pGEX2T plasmids bearing ARNO or ␤-arrestin 1 carboxyl-terminal (319 -418) DNAs were prepared, and fusion proteins were purified as described previously (4). GST fusion proteins on glutathione-Sepharose 4B were incubated in 200 l of binding buffer B (20 mM Tris, 2 mM dithiothreitol, 25 mM NaCl, 2 mM EDTA, 0.2% Triton X-100, 2.5 mM MgCl 2 , 1 mM ATP, pH 7.4) containing protease inhibitors for 4 h at 4°C with purified 6xHis-␤-arrestin 1 or recombinant myristoylated ARF6 protein in the absence or presence of either GDP␤S or GTP␥S. The beads were spun and washed three times with binding buffer followed by two washes with binding buffer without any detergent. Beads were resuspended in 2ϫ SDS sample buffer, incubated at 95°C for 10 min, resolved by electrophoresis on a 14% gel, transferred onto nitrocellulose, and analyzed by Western blot or stained with Coomassie Blue for GST-protein detection.
Co-Immunoprecipitation of Endogenous Proteins from Whole Brain Extract-Co-immunoprecipitations from whole brain extract were performed as described previously (29). ␤-Arrestins were immunoprecipitated from the supernatant using an antibody specific for ␤-arrestins (A1CT) covalently cross-linked to Reactigel beads (Pierce). Immunoprecipitated proteins were detected by immunoblot analysis using a goat anti-ARNO amino-terminal antibody (Santa Cruz Biotechnology) and the ␤-arrestin polyclonal antibody.
GTP␥S-binding Assay-[ 35 S]GTP␥S loading assays were done as described previously (30). Briefly, [ 35 S]GTP␥S binding to purified recombinant ARF6 (myristoylated and nonmyristoylated forms) was assayed in a total volume of 50 l using a rapid filtration procedure. ARF6 (0.36 g), [ 35 S]GTP␥S (4 M: 1 ϫ 10 6 cpm) without or with GST-ARNO (1 g) and ␤-arrestin 1 (5 g) were incubated for the indicated times at 37°C. After the reaction was stopped by the addition of 2 ml of wash buffer and rapid filtration onto nitrocellulose, the amount of proteinbound [ 35 S]GTP␥S was quantified. Data are reported as mean Ϯ S.E. of values from duplicate assays and expressed as the amount of GTP␥S specifically bound to ARF6.
Data Analysis-The mean and standard error of the mean are expressed for values obtained from the number of separate experiments indicated performed in duplicate. Statistical analysis was performed by a two-way analysis-of-variance followed by an unpaired Student's t test or Tukey's statistical test.

RESULTS AND DISCUSSION
To directly investigate the role of ARF proteins in endocytosis of GPCRs, we first expressed wild type and mutant ARF proteins together with the ␤ 2 AR and examined receptor internalization following isoproterenol stimulation. Expression of wild type ARF6 did not significantly affect ␤ 2 AR sequestration from the cell surface at any time point, whereas overexpression of a mutant defective in GTP hydrolysis (ARF6Q67L) or in GTP binding (ARF6T27N) showed marked inhibitory effects (Fig.  1a). In contrast, expression of ARF1 or ARF1 mutants, ARF1T31N or ARF1Q71L, had no effect on ␤ 2 AR endocytosis (data not shown). Presumably, this is because activated ARF1 is mainly found on intracellular membranes, whereas activated ARF6 is mainly localized at the plasma membrane.  1. Effect of ARF6, ARF6Q67L, ARF6T27N, ARNO, and GIT1 on internalization of the ␤ 2 AR. a, HEK 293 cells were transiently co-transfected with Flag-␤ 2 AR together with empty vector, ARF6, ARF6Q67L, or ARF6T27N. Agonist-induced internalization of epitope-tagged ␤ 2 AR was measured before and after treatment with isoproterenol (10 M) for the indicated times. b, cells were transiently co-transfected with Flag-␤ 2 AR together with empty vector, ARF6, or ARF6Q67L. Agonist-induced internalization of epitope-tagged ␤ 2 AR was measured before and after 30 min of isoproterenol treatment (10 M). Agonist was removed by washes and cell surface receptor number was quantified after 15, 30, and 60 min. Results were expressed as the percent of cell surface immunofluorescence compared with nonstimulated cells. The data represent the mean of 4 -6 independent experiments done in duplicate (**, p Ͻ 0.001). c, cells transiently expressing Flag-␤ 2 AR together with either ARNO or GIT1 were stimulated with isoproterenol (10 M) for the indicated times. Results were expressed as the percent loss of cell surface immunofluorescence compared with nonstimulated cells. The data represent the mean of 4 -8 independent experiments done in duplicate (*, p Ͻ 0.05; **, p Ͻ 0.001).
Interestingly, expression of ARF6 mutants prevents constitutive recycling of the transferrin receptor, another process involving cell surface vesicle trafficking (12). Although expression of wild type ARF6 had no effect on the kinetics of recycling of the ␤ 2 AR, expression of the ARF6Q67L mutant totally prevented reappearance of internalized receptors at the cell surface, even after 60 min of recovery when 95% of internalized receptors are normally recycled (Fig. 1b). These data suggest that ARF proteins, namely ARF6, regulate internalization as well as recycling of the ␤ 2 AR, two processes requiring plasma membrane vesicle trafficking.
Like other small GTP-binding proteins, activation of ARF by GTP loading is regulated by nucleotide exchange factors, whereas inactivation by GTP hydrolysis is facilitated by GAPs. We have previously shown that expression of the ARF GTPaseactivating protein GIT1 markedly inhibits internalization of the ␤ 2 AR (6). Therefore, we hypothesized that expression of an appropriate ARF guanine nucleotide exchange factor would facilitate ARF6 activation, thereby leading to increased receptor internalization. In HEK 293 cells, we transiently expressed ARNO, an exchange factor for both ARF1 and ARF6 that is localized to the plasma membrane (26,31), and examined the internalization profile of the ␤ 2 AR. Increased cellular levels of ARNO led to increased agonist-stimulated internalization of receptors, whereas expression of GIT1 was inhibitory (Fig. 1c). These results suggest that expression of ARNO facilitates receptor internalization by promoting ARF6 activation. In contrast, GIT1 expression may reduce receptor internalization by triggering immediate inactivation of ARF6. Our data on the contribution of ARF regulatory proteins support the role of ARF6 in regulating agonist-promoted receptor endocytosis.
To elucidate the molecular mechanisms by which ARF6 regulates receptor internalization, we set out to identify proteins interacting with ARF6 and its regulatory proteins. The ␤-arrestins have been shown to play an important role in receptor internalization through their direct interaction with clathrin and its adaptor protein, AP-2 (2, 3). We hypothesized that ␤-arrestin, a protein recruited to GRK-phosphorylated receptors, might also play a role in the regulation of ARF6 function. Therefore, we examined whether ARF6 could interact with ␤-arrestin 1 and ␤-arrestin 2. Cells expressing HA-␤ 2 AR, Flag-␤-arrestin 1, or Flag-␤-arrestin 2 and ARF6 were left untreated or were stimulated with isoproterenol for 2, 5, and 10 min. Subsequently, Flag-␤-arrestin proteins were immunoprecipitated, and associated ARF6 was detected by immunoblotting. Interestingly, the binding of ARF6 to ␤-arrestin 1 or ␤-arrestin 2 increased following receptor stimulation and could be detected readily after 2 min of agonist stimulation (Fig. 2a). The interaction was found to be maximal after 5 and 10 min of receptor activation. von Zastrow and Kobilka (32) have reported that ␤ 2 AR targeting to clathrin-coated pits occurs after 2 min of agonist activation. Similarly, ␤-arrestin 2 can be found in complex with the ␤ subunit of AP-2 following 2 min of ␤ 2 AR stimulation (3). Therefore, our data suggest that ARF6 activation following receptor stimulation might regulate early events of the endocytic process. Using similar experimental conditions, we were unable to co-immunoprecipitate ARF6 with the ␤ 2 AR (data not shown) suggesting that ␤-arrestin is required to bridge this receptor to ARF-dependent signaling pathways. Mitchell et al. (33) have reported that in 1321N1 cells, ARF proteins can be found in complex with several but not all Ca 2ϩ -mobilizing G protein-coupled receptors, activation of which lead to phospholipase D stimulation. Whether these interactions are direct or mediated via an adaptor protein has not yet been examined.
To confirm that the interactions between ␤-arrestin and ARF6 were direct, we used purified recombinant Flag-␤-arrestin 1 and 2 together with ARF6 and analyzed their ability to interact in vitro. As illustrated in Fig. 2b, ARF6 can be coimmunoprecipitated with both ␤-arrestin 1 and ␤-arrestin 2.
Mutagenesis studies of ␤-arrestin proteins have revealed that the binding sites for both clathrin and AP-2 are present in the carboxyl-terminal portion of the protein (2-4). To determine whether the interaction between ␤-arrestin and ARF6 is also mediated via the carboxyl-terminal part of ␤-arrestin , we used a GST fusion protein of the last 100 amino acids of ␤-arrestin 1 (GST-␤-arrestin 1-(319 -418)) and examined the interaction with ARF6. Pull-down assays revealed that a binding site for ARF6 is present within this region of ␤-arrestin 1 (Fig. 2c). Interestingly, the interaction between GST-␤-arrestin 1-(318 -419) and ARF6 was found to be regulated by the nature of the nucleotide bound to the small G protein, which is purified mainly in its GDP-bound state. The addition of GDP␤S did not alter the interaction, whereas addition of GTP␥S completely abolished the binding of ARF6 to ␤-arrestin 1 (Fig. 2c). Taken together, these results suggest that ␤-arrestin might act as a scaffold protein bringing together the ARF protein in complex with receptors. Furthermore, this interaction is regulated by both agonist activation of the receptor and the nature of the nucleotide bound to ARF6. Once bound to GTP, activated ARF6 dissociates from ␤-arrestin proteins and is available to promote FIG. 2. ␤-Arrestins interact with ARF6. a, Flag-␤-arrestin 1 and 2 were co-expressed in HEK 293 cells together with HA-␤ 2 AR and ARF6. Using Flag-affinity beads, ␤-arrestins were immunoprecipitated (IP) after the indicated time of isoproterenol-stimulation (10 M). The presence of ARF6 in the immunoprecipitate was detected (IB) using specific antibodies (J. Donaldson, National Institutes of Health). Mock cells co-expressed the HA-␤ 2 AR and ARF6. Amounts of immunoprecipitated ␤-arrestins were detected using a Flag-probe antibody (Santa Cruz). b, purified recombinant Flag-␤-arrestin 1 and 2 (2 g) were incubated with purified ARF6 (0.3 g). ␤-Arrestin 1 and 2 were immunoprecipitated with Flag-affinity beads and the associated ARF6 detected by immunoblotting. c, GST-␤-arrestin 1-(319 -418) was incubated with ARF6 in the absence and presence of GDP␤S or GTP␥S (0.1 mM). GST-ARNO was captured using glutathione-Sepharose 4B and pelleted, and associated ARF6 was detected by immunoblotting. Data shown are representative of at least four independent experiments. Similar results were obtained with myristoylated ARF6. the endocytic process.
Next, we examined whether ARNO could also interact with ␤-arrestin proteins. Cells overexpressing HA-␤ 2 AR, Flag-ARNO, and His-␤-arrestin 1 or His-␤-arrestin 2 were left untreated or stimulated with isoproterenol for 5 min. Subsequently, ARNO was immunoprecipitated, and associated ␤-arrestin 1 or ␤-arrestin 2 were detected by Western blotting. Fig. 3a shows that ARNO can be co-immunoprecipitated with both ␤-arrestins. In this regard, it is interesting that a recent study reported that ARNO can promote release of a pool of membrane-associated ␤-arrestin involved in the desensitization process of the luteinizing hormone/choriogonadotropin receptor (34). In addition, ARNO and ARF6 could also be found in a complex in cells expressing HA-␤ 2 AR, Flag-ARNO, and ARF6 (data not shown). Under these experimental conditions, the interactions between ARNO and ␤-arrestin, as well as ARNO and ARF6, did not appear to be regulated by agonist stimulation of the receptor. Furthermore, using a GST fusion protein of ARNO and purified recombinant Flag-␤-arrestin 1 and 2, ARNO and ␤-arrestin proteins were found to interact directly (Fig. 3b). These results demonstrate that ␤-arrestin can be found in complex with ARF6 as well as with its nucleotide exchange factor, ARNO. However, the interaction between ␤-arrestin and ARNO does not require agonist activation of the ␤ 2 AR. Furthermore, ARNO was found to interact simultaneously with ␤-arrestin 1 and ARF6, suggesting that the binding of ␤-arrestin 1 and ARF6 to ARNO is not mutually exclu-sive. Indeed, increasing amounts of ␤-arrestin 1 did not prevent ARF6 binding to ARNO (data not shown).
To verify that these molecular interactions occur with endogenous proteins, ␤-arrestins were immunoprecipitated from bovine brain extracts with a ␤-arrestin-specific antibody covalently cross-linked to beads, and associated proteins were detected by immunoblotting. ARNO was found to be present in the ␤-arrestin immunoprecipitates but not in the pre-immune serum immunoprecipitates (Fig. 3c). However, under several experimental conditions, we were unable to detect ARF6 in the endogenous ␤-arrestin immunoprecipitates (data not shown). This is likely due to the agonist dependence of this interaction, as illustrated above. These findings demonstrate a specific interaction between ARNO and ␤-arrestin at endogenous levels of proteins and confirm the results obtained from cellular coimmunoprecipitations and in vitro experiments.
It is well established that the binding of ARNO to ARF6 serves to catalyze the exchange of GDP for GTP to activate the ARF protein. We next wanted to determine whether ␤-arrestin simply acts as an adaptor protein between activated receptors and the ARF6 complex or, rather, plays a more direct role in the regulation of the ARF6 activation process. To investigate the effect of ␤-arrestin on ARNO-mediated GTP␥S-binding to ARF6, we used recombinant ␤-arrestin 1, ARF6, and ARNO in an in vitro GTP␥S loading assay. The rate of GTP␥S binding to ARF6 was enhanced by increasing concentrations of ARNO (data not shown). Further, the amount of GTP␥S bound to ARF6 in the presence of ARNO was significantly increased in a time-dependent fashion, consistent with its role as an ARF nucleotide exchange factor (Fig. 4). Using an amount of ARNO that leads to submaximal GTP loading, further addition of ␤-arrestin 1 led to a marked potentiation of the rate of activation of ARF6 stimulated by ARNO without affecting maximal loading of GTP␥S. Therefore, the effect of ␤-arrestin 1 was most significant at early time points (5 and 10 min). In the absence of ARNO, ␤-arrestin 1 did not have any effect on the amount of GTP␥S bound to ARF6. These data suggest that by interacting with both ARF6 and ARNO, ␤-arrestin 1 facilitates GTP loading of the ARF protein, thereby leading to a potentiation of the activation of this small GTP-binding protein.
Recent reports have demonstrated that ARF proteins are involved in several intracellular trafficking processes. Here, we show that ARF6 regulates internalization of the ␤ 2 AR, a prototypical G protein-coupled receptor. Expression of the GTP hydrolysis-deficient mutant (ARF6Q67L) or the GTP bindingdefective mutant (ARF6T27N) results in the inhibition of the internalization process. Interestingly, expression of ARF6Q67L or ARF6T27N inhibits Fc␥ receptor-mediated phagocytosis in were left untreated or stimulated with isoproterenol (iso, 10 M) for 5 min. Flag-ARNO was immunoprecipitated (IP), and associated ␤-arrestin 1 or ␤-arrestin 2 was detected by immunoblotting (IB) using a His probe antibody (Santa Cruz). Amounts of immunoprecipitated ARNO were detected using a goat anti-ARNO amino-terminal antibody (Santa Cruz). MOCK cells co-expressed the HA-␤ 2 AR, ␤-arrestin 1, and ␤-arrestin 2. Levels of ␤-arrestin 1 and 2 expression in the cell lysates were verified using the ␤-arrestin 1 A1CT antibody. b, GST or GST-ARNO fusion proteins were incubated with purified Flag-␤-arrestin 1 and 2 (2.5 g). GST or GST-ARNO were captured on glutathione-Sepharose, and the amounts of ␤-arrestin proteins interacting were detected by immunoblotting. Data shown are representative of at least three independent experiments. c, ␤-arrestin was immunoprecipitated from bovine brain lysates with an antibody specific to ␤-arrestin (A1CT) and resolved by SDS-polyacrylamide gel electrophoresis. The presence of ARNO and ␤-arrestin in the immunoprecipitates was detected using antibodies to ARNO (goat anti-ARNO NT) and ␤-arrestin (A1CT). Immunoblot analysis of ARNO and ␤-arrestin in total cell lysates is also shown. Data are representative of three independent experiments. Similar results were obtained in mouse brain lysates. ␤-Arrestin-mediated ARF6 Activation macrophages while stimulating endocytosis at the apical surface of Madin-Darby canine kidney cells (15,16). Several reports have demonstrated that ARF6Q67L accumulates at the plasma membrane where it induces invaginations, whereas ARF6T27N is distributed to an internal tubulovesicular compartment (17,18,35). Although both ARF6 mutants affect internalization, they probably do so by different mechanisms (15,16). Similarly, we have shown that expression of GIT1, an ARF GAP, also inhibits internalization of the ␤ 2 AR. However, expression of ARNO, an ARF guanine nucleotide exchange factor, is stimulatory. By overexpressing GIT1, we might promote rapid and unregulated inactivation of ARF6, thereby leading to inhibition of vesicle formation. In contrast, expression of ARNO would result in further ARF6 activation and therefore promote vesicle formation. At endogenous levels of proteins, both ARNO and GIT probably contribute to the formation of endocytic vesicles by regulating, in a coordinated fashion, the turnover of nucleotide on ARF proteins.
According to our findings, ␤-arrestin appears to serve as an agonist-controlled scaffold bringing together the exchange factor, ARNO, and the GDP-bound form of ARF6, thereby promoting ARF6 activation in proximity to the receptor. The agonist dependence of the ARF6 interaction with ␤-arrestin brings this entire process under the control of the receptor. Once bound to GTP, active ARF6 proteins dissociate from ␤-arrestin and are now available to promote the endocytic process. The biochemical details involved in this process remain to be determined. However, the closely related ARF1 protein is known to drive vesicle budding from donor membranes by promoting coat protein assembly through the recruitment of AP-1 and coatomers (COPI, COPII, and clathrin) (36,37). Furthermore, ARF6 has been shown to activate phospholipase D, thereby increasing the levels of phosphatidic acid, and to activate phosphatidylinositol 4-and 5-kinase activity to increase phosphatidylinositol 1,4,5trisphosphate levels (8,38). Interestingly, ␤-arrestin, GRK, ARNO, and GIT proteins all bind to phosphatidylinositol 4,5bisphosphate and/or phosphatidylinositol 1,4,5-trisphosphate (39 -42). Therefore, activation of ARF6 following GPCR stimulation would presumably lead to the recruitment of vesicle coat proteins (clathrin and AP-2), reorganization of the actin cytoskeleton, and modification of the lipid content of the plasma membrane, all of which promote receptor endocytosis. Subsequently, inactivation of ARF6 by GTP hydrolysis, which is necessary for proper trafficking, is achieved by ARF GAPs such as GIT1, which as previously illustrated are recruited to the receptor through their interaction with GRKs (5).
The results reported here add to the growing list of ␤-arrestin-regulated functions that link these molecules to endocytosis. These include interaction with clathrin, AP-2, N-ethylmaleimide-sensitive fusion protein, and Src (2,3,24,43). Moreover, the ability of ␤-arrestin to spatially localize ARF6, while facilitating its activation, is quite analogous to the recently discovered functions of ␤-arrestin to localize and facilitate the activation of several MAP kinases (29,44). Interestingly, other small GTP-binding proteins have also been shown to regulate receptor internalization and recycling. Understanding how these various protein-protein interactions are dynamically and temporally integrated to regulate G protein-coupled receptor endocytosis and recycling is an important goal for future studies.