G protein-coupled receptor endocytosis in ADP-ribosylation factor 6-depleted cells.

The internalization of G protein-coupled receptors is regulated by several important proteins that act in concert to finely control this complex cellular process. Here, we have applied the RNA interference approach to demonstrate that ADP-ribosylation factor 6 (ARF6) is essential for the endocytosis of a broad variety of receptors. Reduction of endogenous expression of ARF6 in HEK 293 cells resulted in a correlated inhibition of the beta(2) -adrenergic receptor internalization previously characterized as being sequestered via the clathrin-coated vesicle pathway. Furthermore, other receptors internalizing via this endocytic route, namely the angiotensin type 1 receptor and the vasopressin type 2 receptor, were also impaired in their ability to be sequestered when levels of endogenous ARF6 in cells were reduced. Interestingly, endocytosis of the endothelin type B receptor, characterized as being internalized via the caveolae pathway, was also markedly inhibited in ARF6-depleted cells. In contrast, internalization of the vasoactive intestinal peptide receptor was unaffected by reduced levels of ARF6. Finally, internalization of the acetylcholine-muscarinic type 2 receptor via the non-clathrin-coated vesicle pathway was also inhibited in ARF6-depleted cells. Taken together, our results demonstrate that ARF6 proteins play an essential role in the internalization process of most G protein-coupled receptors regardless of the endocytic route being utilized. However, this phenomenon is not general. In some cases, another ARF isoform or other proteins may be essential to regulate the endocytic process.

G protein-coupled receptors constitute the largest family of transmembrane proteins. Their main role is to communicate a broad variety of signals to the cell to generate a cellular response. Upon sustained stimulation, G protein-coupled receptors can activate signaling pathways that will lead to their desensitization and, for most receptors, their internalization. Numerous functions have been attributed to the sequestration of receptors from the cell surface. First, this phenomenon is believed to be important for receptor resensitization (1) and receptor degradation (2). Second, it has been more recently appreciated that the internalization process is required for the activation of specific signaling cascades such as the activation of mitogenic signaling events (3). Numerous proteins are in-volved in the regulation of receptor endocytosis. Their nature may vary according to the endocytic route being utilized by the receptors.
The most commonly used and best characterized endocytic pathway is the clathrin-coated vesicle (CCV) 1 pathway. Interestingly, the ␤-arrestin proteins that are recruited to activated G protein-coupled receptors have been shown to be important regulators of this endocytic route. The ␤-arrestins can interact directly with the two main components of the clathrin coat, clathrin and its adaptor protein AP-2 (4,5). Moreover, ␤-arrestin proteins were proposed to act as scaffold proteins and play a major role in the recruitment of the tyrosine kinase c-Src responsible for the phosphorylation of dynamin (6). Furthermore, ␤-arrestin proteins can be found in complex with several other proteins involved in vesicle trafficking (7,8). Our group has shown that the ADP-ribosylation factor 6 (ARF6), a monomeric GTP-binding protein, interacts with the ␤-arrestins upon receptor activation (9). Using purified proteins in a reconstituted system, the interaction of ARF6 with ␤-arrestins served to potentiate its activation by the guanine nucleotide exchange factor, ARNO, which has also been found in complex with the ␤-arrestins. Using mutants of ARF6 that were ineffective in loading GTP (ARF6T27N) or that mimicked the GTP-bound form (ARF6Q67L) as well as ARF regulatory proteins such as ARNO and GIT1, we suggested that these small GTP-binding proteins are involved in regulating the internalization of the ␤ 2 -adrenergic receptor (9). Using these same mutants, ARF proteins have been shown to regulate several trafficking events (10 -12), promote the remodeling of the actin cytoskeleton (13)(14)(15), and alter the composition of membrane lipids via the activation of phospholipase D (16,17) and phosphatidylinositol-(4)-phosphate 5-kinase (18).
Other endocytic routes may be utilized by activated receptors. Interestingly, some receptors have been shown to be present in caveolae structures and were proposed to be processed to intracellular compartments via the sequestration of these membrane microdomains (19,20). The proteins responsible for the formation and the trafficking of caveolae or lipid rafts are poorly characterized. Similar to the clathrin-coated vesicles, caveolae appears to require the GTPase dynamin for the fission of the budding vesicle (21,22). In contrast, the ␤-arrestin as well as the ARF GAP GIT1 do not play an important role in the regulation of this endocytic pathway (19). Furthermore, an alternative route, the non-clathrin-coated vesicle pathway * This work was supported in part by the Canadian Institutes of Health Research Grant MOP-53199 and the Heart and Stroke Foundation of Canada. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
(NCCV) seems to be utilized by other types of receptors. Proteins responsible for the formation of these endocytic vesicles remain to be identified. Internalization via this pathway is independent of ␤-arrestin, GIT1, and dynamin (19,23,24) but was proposed to involve ARF6 (24). Of all the G protein-coupled receptors, only the acetylcholine muscarinic M2 receptor has been shown to be internalized by this endocytic route (25).
To assess the importance of ARF6 proteins in the regulation of the endocytic process of a broad variety of G protein-coupled receptors, we took advantage of the RNA interference approach to specifically inhibit the endogenous expression of this small GTP-binding protein. Previous studies have used mutant forms of ARF6 to determine the distribution of this small GTPase and to assess its role in endocytosis and trafficking (9). Results generated by studying these mutants hinted at a role for ARF6 in many cases. More recently, the use of ARF6T27N and its significance in overexpressing systems was challenged. Macia et al. (26) suggested that this mutant was not a reliable marker of the GDP-bound form of ARF6. When overexpressed in cells, most of the protein aggregated in vivo and mislocalized (26). To definitely prove that endogenously expressed ARF6 is an important regulator of the endocytic process of the ␤ 2 -adrenergic receptor and to determine whether ARF6 is essential for the internalization of other G protein-coupled receptors, we resorted to the RNA interference approach. Gene silencing by small interfering RNA (siRNA) has allowed us to generate cells that express lower levels of ARF6 and study the consequence on receptor endocytosis. This method offers great advantages, because the level of inhibition of ARF6 production can easily be controlled. Using this technique, we were able to demonstrate that ARF6 is essential for the endocytosis of the ␤ 2 -adrenergic receptor. Moreover, we found that the internalization of several other G protein-coupled receptors was impaired in cells expressing reduced levels of ARF6. However, we identified one receptor that did not require ARF6 to be internalized. Taken together, our study demonstrates the essential role of ARF6 in controlling the internalization process of G protein-coupled receptors being internalized via the three main endocytic pathways.

EXPERIMENTAL PROCEDURES
Reagents and Antibodies-Minimal essential medium and fetal bovine serum were purchased from Sigma. All other tissue culture reagents were purchased from Invitrogen. Murine monoclonal anti-FLAG, fluorescein isothiocyanate-conjugated secondary antibodies, and the monoclonal anti-␤ 1 and ␤ 2 -adaptin antibodies were purchased from Sigma. Anti-HA 12CA5 was from Roche Applied Science; the Alexa Fluor 568-and Alexa Fluor 488-coupled anti-mouse IgGs and the transferrin Alexa Fluor 488 conjugate were purchased from Molecular Probes. Anti-clathrin heavy chain antibody was from BD Pharmingen. The anti-ARF6 polyclonal antibody was a generous gift from J. Donaldson (National Institutes of Health). The agonist peptides endothelin-1, arginin-vasopressin, and angiotensin II (Ang II) were obtained from American Peptide. Vasoactive intestinal peptide (VIP) was from Peninsula Laboratories. Isoproterenol and acetylcholine were purchased from Sigma. The Silencer siRNA construction kit was purchased from Ambion, and the Alexa Fluor 546-coupled negative control siRNA was from Qiagen. Paraformaldehyde was from Fisher. Gel/Mount TM was obtained from Biomeda. The CatchPoint TM cAMP fluorescent assay kit was purchased from Molecular Devices. All other reagents were from Sigma.
RNA Interference-Double-stranded siRNAs targeting human ARF6, with 19-nucleotide duplex RNA and 2-nucleotide 3Ј dTdT overhangs were synthesized using the Silencer TM siRNA construction kit from Ambion (Austin, TX). To design ARF6-specific siRNA duplexes, the nucleotide sequence of human ARF6 gene was screened for unique 21-nucleotide sequences starting with two adenosines (AA) and containing a G/C ratio of 30 -50%. Two 21-nucleotide sequences were chosen corresponding, respectively, to the positions 244 -265 and 347-368 on the human ARF6 mRNA relative to the start codon: 5Ј-AAGGUCUCAUCUUCGUAGUGG-3Ј (sequence 1) and 5Ј-AAUC-CUCAUCUUCGCCAACAA-3Ј (sequence 2). The sequences were compared with the human genome data base using BLAST, and no homology was found with other genes. Template DNA oligonucleotides were designed according to instructions from the Silencer TM siRNA construction kit and chemically synthesized (Sigma-Genosys). A control siRNA sequence targeting glyceraldehyde-3-phosphate dehydrogenase was synthesized from a template DNA oligonucleotide provided with the siRNA transcription kit. A non-silencing fluorescently labeled siRNA (Qiagen) was also used in the microscopy experiments. The target sequence of the negative control siRNA Alexa Fluor 546 is 5Ј-AATTCTCCGAACGTGTCACGT-3Ј. The fluorescent tag is added at the 3Ј-end of the sense strand.
Cell Culture and Transfection-HEK 293 cells plated in 15-cm dishes were maintained at 37°C, 5% CO 2 in minimal essential medium supplemented with 1 mM non-essential amino acids, 1 mM sodium pyruvate, 10% fetal bovine serum, and 1% penicillin/streptomycin. 24 h prior to transfection, 2 ϫ 10 5 cells were seeded in 6-well plates or onto glass coverslips in minimal essential medium supplemented with 1 mM nonessential amino acids, 1 mM sodium pyruvate, and 10% fetal bovine serum. Cells were transfected the next day with the different receptor constructs (␤ 2 AR-FLAG, V2R-HA, AT1R-FLAG, ET B R-FLAG, VIPR-FLAG, or M2MR-HA) together with empty vector (pBK(⌬)) or siRNA using Lipofectamine TM 2000 according to the manufacturer's instructions.
Receptor Internalization Assay by Flow Cytometry-Internalization of the different G protein-coupled receptors by cDNA transfection of HEK 293 cells was measured using specific antibodies as described previously (28). Briefly, cells plated in a 6-well dish were co-transfected with either empty pBK(⌬) vector or siRNA #1 and one of the receptor constructs (␤ 2 AR-FLAG, V2R-HA, ET B R-FLAG, VIPR-FLAG, or M2MR-HA). 72 h after transfection, the cells were serum-starved for at least 1 h then stimulated for 30 min with 10 M isoproterenol (␤ 2 AR-FLAG), 1 M arginin-vasopressin (V2R-HA), 0.1 M endothelin-1 (ET B R-FLAG), 0.1 M vasoactive intestinal peptide (VIPR-FLAG), or 100 M acetylcholine (M2MR-HA). The cells were then placed on ice and incubated for 1 h at 4°C with a monoclonal anti-FLAG or anti-HA 12CA5 antibody and then for 45 min with a fluorescein isothiocyanateconjugated anti-mouse IgG as described previously (28). The cells were rinsed three times with PBS then put in suspension in 400 l of PBS-HEPES and fixed by adding 100 l of a 3.8% formaldehyde solution. Cell surface fluorescence was detected using a FACSCalibur flow cytometer (BD Biosciences) and expressed as the percentage of internalized receptors. Each point was analyzed in duplicate, and results are expressed as means Ϯ S.E. of several experiments (as indicated in the figure legend).
Confocal Microscopy-Constitutive internalization of transferrin in control and ARF6-depleted HEK 293 cells was assessed by confocal microscopy using Alexa Fluor 488-coupled transferrin. Briefly, HEK 293 cells were seeded onto glass coverslips and transfected with empty vector (pBK(⌬)) or ARF6 siRNA #1 (60 nM) and a commercially available Alexa Fluor 546-coupled negative control siRNA (5 nM). 72 h after transfection, cells were serum-starved for 1 h, then Alexa Fluor 488-coupled transferrin was added for 15 min (3 g/ml at 37°C). Cells were fixed by incubating in PBS containing 4% paraformaldehyde (15 min at room temperature), then washed three times and mounted onto microscope slides using Gel/ Mount TM . Similarly, internalization of receptors was also visualized by microscopy. Briefly, HEK 293 cells seeded onto glass coverslips were transfected with the different receptor constructs (␤ 2 AR-FLAG, AT1R-FLAG, and VIPR-FLAG) and either empty vector (pBK(⌬)) or ARF6 siRNA #1. 72 h after transfection, cells were serum-starved for 1 h and stimulated for 15 min with 10 M isoproterenol (␤ 2 AR-FLAG), 1 M angiotensin II (AT1R-FLAG), or 0.1 M vasoactive intestinal peptide (VIPR-FLAG). Cells were then fixed with paraformaldehyde and incu-bated at room temperature with a monoclonal anti-FLAG antibody for 1 h in minimal essential medium containing 0.1% bovine serum albumin, 10 mM HEPES, and 0.05% saponin. Cells were washed twice and then incubated with an Alexa Fluor 568-coupled secondary antibody for 45 min. Cells were rinsed three times and mounted onto microscope slides. Staining of native clathrin-coated pits was done in similar conditions where fixed HEK 293 cells (controls and siRNA-transfected) were incubated with the anti-clathrin heavy chain and a specific Alexa Fluor 488coupled monoclonal antibody, respectively. Cellular localization of fluorescently labeled proteins was visualized using a Zeiss confocal microscope (LSM510META). Each figure shows representative results that were observed in the majority of the transfected cells (Ͼ80%) in at least three independent experiments (Ͼ50 cells examined).
Intracellular cAMP Accumulation Measurement-HEK 293 cells previously seeded in 6-well plates were transiently transfected with the ␤ 2 AR-FLAG construct together with empty vector (pBK(⌬)) or siRNA directed against ARF6 (ARF6 siRNA #1). 72 h after transfection, cAMP accumulation was measured using the CatchPoint TM cAMP fluorescent assay kit according to the manufacturer's instructions. Briefly, cells were rinsed once with PBS, then with PBS containing 0.02% EDTA. Cells were suspended in minimal essential medium/10% fetal bovine serum and counted. They were then centrifuged at 1,000 ϫ g for 5 min. The cells were rinsed using Krebs-Ringer bicarbonate buffer solution containing 10 mM glucose (KRBG; 115 mM NaCl, 24 mM NaHCO 3 , 5 mM KCl, 1 mM MgCl 2 , 2.5 mM CaCl 2 , 10 mM HEPES, 10 mM glucose, pH 7.4) and re-suspended in KRBG containing 0.02% EDTA and 0.75 mM isobutylmethylxanthine. 20,000 cells, in suspension, were stimulated with either isoproterenol (0.1 M) or forskolin (20 M) for 15 min at 37°C. Lysis buffer (20 l) was then added, and cell lysates (40 l) were added to a 96-well plate pre-coated with goat anti-rabbit IgG. 40 l of rabbit anti-cAMP antibody and 40 l of cAMP-HRP conjugate were added successively, and the plate was incubated at room temperature for 2 h. Plate content was then aspirated, and the wells were washed four times using wash buffer. 100 l of Stoplight Red substrate was added to every well, and the plate was incubated at least 10 min at room temperature shielded from light. Fluorescence was detected using a Flexstation II system (Molecular Devices). The amount of cAMP was calculated from a cAMP standard curve. Results were expressed as a percentage of the maximum cAMP produced by forskolin stimulation.
Data Analysis-The mean Ϯ S.E. are expressed for values from the number of separate experiments indicated. Statistical significance was determined by performing an unpaired Student's t test (Mann-Whitney test) to compare ARF6 siRNA-transfected and control cells (PRISM software).

Reduction of ARF6 Expression by RNA
Interference-To reduce the expression of ARF6, we designed two different siRNA duplexes that targeted the segments 244 -265 and 347-368 of the human ARF6 open reading frame. HEK 293 cells were transfected with the transfection reagent alone, a control siRNA targeting glyceraldehyde-3-phosphate dehydrogenase, or two different siRNAs designed against ARF6. First, to quantify the level of ARF6 expression, cell lysates were prepared from control and siRNA-transfected cells and analyzed by immunoblotting using specific anti-ARF6 antibodies, 72 h posttransfection. As shown in Fig. 1A, addition of the transfection reagent or the control siRNA directed against glyceraldehyde-3-phosphate dehydrogenase (60 nM) did not alter the levels of ARF6 endogenously expressed. However, transfection of the two siRNA (60 nM each) directed against ARF6 resulted in a marked inhibition of ARF6 expression. The two siRNA designed were equally effective in inhibiting ARF6 expression, and transfection of the two siRNA simultaneously did not further reduce expression of the small GTPase. For the experiments described hereafter, siRNA #1 was used.
The dose-dependent effect of siRNA on ARF6 expression was further investigated by transfecting increasing amounts of siRNA (20, 60, 85, and 285 nM). As illustrated in Fig. 1B, the reduction in ARF6 expression was correlated with the amounts of siRNA transfected. Furthermore, the decrease in cellular ARF6 levels did not affect the expression of other proteins associated with the endocytic process, namely the clathrin heavy chain and the ␤ 2 -adaptin (a subunit of the AP-2 complex) at the three lowest concentrations of siRNA used (Fig. 1B). However, when cells were transfected with 285 nM siRNA, the highest concentration tested, expression of ARF6 together with clathrin and ␤ 2 -adaptin was significantly impaired (Fig. 1B), and cellular viability was severely compromised (data not shown). Further analysis revealed that cells transfected with 20, 60, and 85 nM ARF6 siRNA exhibited a reduction in ARF6 expression of 30%, 55%, and 70%, respectively, compared with control cells (n ϭ 4).
To determine whether depletion of ARF6 in HEK 293 cells affected basal endocytosis, we assessed the constitutive internalization of transferrin coupled to Alexa Fluor 488 in control and ARF6-depleted cells. To visualize which cells had been transfected with the siRNA, we added to our ARF6 siRNA #1 (60 nM) transfection a non-silencing fluorescently labeled siRNA (negative control siRNA coupled to Alexa 546, red) (5 nM). Fig. 1C illustrates that ARF6 depletion does not affect the constitutive endocytosis of fluorescently labeled transferrin.
G Protein-coupled Receptor Endocytosis in ARF6 Knockdown Cells-To demonstrate the role of ARF6 proteins in regulating the endocytic process of G protein-coupled receptors, we first examined the internalization of the ␤ 2 -adrenergic receptor (␤ 2 AR) known to sequester via the clathrin-coated vesicles pathway. Before agonist stimulation, ␤ 2 ARs were found at the plasma membrane as shown by confocal microscopy (Fig. 2A,  upper left panel). Following exposure to isoproterenol, receptors internalized into intracellular compartments ( Fig. 2A, upper  right panel). Inhibition of ARF6 expression by siRNA had no effect on the distribution of the receptor prior to agonist activation ( Fig. 2A, lower left panel). As in the control cells, receptors were found at the cell surface. However, upon agonist stimulation, internalization was prevented in ARF6-depleted cells ( Fig. 2A, lower right panel). These observations were better characterized by FACS analysis (Fig. 2B). In our HEK 293 cells, increasing concentrations of siRNA (20, 60, and 85 nM) inhibited the agonist-promoted internalization of ␤ 2 ARs, in a concentration-dependent fashion. Indeed, transfection of 20 nM ARF6 siRNA led to a 28% inhibition of endocytosis, whereas 60 and 85 nM siRNA inhibited the internalization process by 81 and 91%, respectively. Interestingly, the inhibition of endocytosis can be fully correlated with the inhibition of ARF6 expression by the siRNA as illustrated in Figs. 1C and 2B. Taken together, these experiments clearly demonstrate that ARF6 is an essential protein for the internalization of the ␤ 2 AR, because an inhibition of ARF6 expression results in a significant reduction in agonist-promoted receptor endocytosis.
To verify whether the inhibition of receptor internalization by ARF6 was specific to this cellular event, and did not result from an inhibition of receptor function, we tested whether generation of second messengers was affected by ARF6-depletion. Fig. 2C illustrates that accumulation of cAMP resulting from the activation of the ␤ 2 AR was not affected by transfection of ARF6 siRNA (60 nM). Taken together, these data clearly demonstrate that ARF6 specifically regulates the signaling events leading to the internalization process of the ␤ 2 AR.
We then went on to test the role of ARF6 on the endocytic process of other G protein-coupled receptors. In all experiments, we used the lowest concentration of siRNA that produced a significant effect on the internalization of the ␤ 2 AR (60 nM). First, we examined other receptors known to internalize via the clathrin-coated vesicle pathway, namely the angiotensin type 1 (AT1R), and the vasopressin type 2 (V2R) receptors (27,29,30). As illustrated by confocal microscopy, internalization of the AT1R was triggered by angiotensin II (Ang II) stimulation (Fig. 3A, upper panels). As for the ␤ 2 AR, transfection of the siRNA directed against ARF6 markedly prevented the internalization of the AT1R. In these cells, the angiotensinstimulated receptors remained trapped at the plasma membrane (Fig. 3A, lower panels). Similarly, internalization of the V2R was also markedly reduced in ARF6-depleted cells. Indeed, when cells were stimulated with arginin-vasopressin, 33 Ϯ 4% of the receptors were found internalized inside the cells when assessed by FACS analysis. Transfection of the ARF6 siRNA resulted in a 70% inhibition of receptor endocytosis (Fig. 3B). These experiments clearly demonstrate the importance of ARF6 in the internalization process of G proteincoupled receptors being sequestered via the clathrin-coated vesicle pathway.
To verify whether this inhibition in receptor internalization was due to an effect on the integrity of clathrin-coated pits in general, we examined and compared the distribution of clath-rin in control and ARF6-depleted cells. To identify the cells transfected with the siRNA #1 targeted against ARF6 (60 nM), we co-transfected the cells with a fluorescent negative control siRNA (5 nM). Visualization of clathrin-coated pits was achieved by staining the heavy chain of clathrin and scanning FIG. 2. Internalization of the ␤ 2 AR is markedly impaired in ARF6-depleted cells. A, HEK 293 cells were transfected with FLAG-␤ 2 AR together with empty vector (pBK(⌬)) or ARF6 siRNA #1 (60 nM). 72 h after the transfections, cells were serum-starved for 1 h and subsequently stimulated or not with isoproterenol (iso) for 15 min. Cells were then fixed and incubated with the different antibodies. Distribution in the control and ARF6 siRNA#1-transfected cells was examined by confocal microscopy. Upper left panel illustrates the plasma membrane distribution of the ␤ 2 AR before agonist-activation; upper right panel shows the internalized receptors after isoproterenol application. Lower panels illustrate the distribution of the receptor in ARF6-depleted cells before (left) and after (right) receptor activation. Most ␤ 2 ARs remain present at the cell surface upon agonist stimulation in these conditions. The bar represents 10 m. This figure is representative of the results obtained in three independent experiments. B, agonist-induced FLAG-␤ 2 AR internalization was measured by flow cytometry before and after isoproterenol treatment (30 min) in control and ARF6 siRNA #1-transfected HEK 293 cells (20,60, and 85 nM). Results are expressed as percentage of cell surface internalization, which is defined by the relative difference of fluorescence between non-stimulated and stimulated cells. The data represent the mean Ϯ S.E. of 3-6 independent experiments. Statistical significance was determined using a Mann-Whitney t test (PRISM software) (*, p Ͻ 0.05; **, p Ͻ 0.01). C, agonist-induced cAMP accumulation in control or ARF6-depleted HEK 293 cells was measured. Cells, transiently expressing the ␤ 2 AR-FLAG, were treated with 20 M forskolin or 0.1 M isoproterenol for 15 min at 37°C. cAMP accumulation was expressed as a percentage of the maximal cAMP accumulation produced by forskolin. The data represent the mean Ϯ S.E. of three independent experiments. the bottom section of the cell (attached to the coverslip). As illustrated in Fig. 3C, transfection of the siRNA did not have any significant effect on the number or the integrity of the native clathrin-coated pits at the plasma membrane.
Second, we examined the role of ARF6 in the endocytic process of receptors internalizing mainly via the caveolae path-way: the endothelin type B receptor (ET B R), and the vasoactive intestinal peptide receptor (VIPR) (19). In our HEK 293 cells, 29 Ϯ 5% of the ET B Rs were internalized following endothelin-1 stimulation (Fig. 4A). A 30-min pretreatment of the cells with 500 M methyl-␤-cyclodextrin, a specific inhibitor of the caveolae endocytic pathway, resulted in a 72% inhibition of the ET B R internalization demonstrating the importance of this endocytic pathway in our HEK 293 cells. Transfection of the siRNA directed against ARF6 resulted in a similar inhibition with a 79% reduction of the ET B R internalization. These data illustrate the important role for ARF6 proteins in the internalization of receptors utilizing the caveolae pathway. Similarly, 29 Ϯ 3% of the VIPRs were internalized following vasoactive intestinal peptide stimulation and pretreatment with methyl-␤-cyclodextrin led to a 91% inhibition of receptor internalization. Interestingly, decrease in endogenous levels of ARF6 by   FIG. 3.

ARF6-depletion significantly impairs the clathrin-mediated internalization of other G protein-coupled receptors without affecting the integrity of the native clathrin-coated pits.
A, FLAG-AT 1 R distribution in control and ARF6 siRNA#1-transfected HEK 293 cells was examined by confocal microscopy before and after angiotensin II stimulation (Ang II; 1 M, 15 min) similarly as in Fig. 2A. Upper left panel illustrates the plasma membrane distribution of the AT 1 R before agonist-activation; upper right panel shows the internalized receptors after Ang II application. Lower panels illustrate the distribution of the receptor in ARF6-depleted cells before (left) and after (right) receptor activation. Most AT 1 Rs remain present at the cell surface upon agonist stimulation in these conditions. The bar represents 10 m. This figure is representative of the results obtained in three independent experiments. B, agonist-induced V2R internalization was measured by flow cytometry before and after 1 M arginin-vasopressin treatment (30 min) in control and ARF6 siRNA #1-transfected cells (60 nM) as in Fig. 2B. Results are expressed as percentage of cell surface internalization, which is defined by the relative difference of fluorescence between non-stimulated and stimulated cells. The data represent the mean Ϯ S.E. of six independent experiments. Statistical significance was determined using a Mann-Whitney t test (PRISM software) (**, p Ͻ 0.01). C, the integrity of the clathrin-coated pits was examined by staining clathrin. Control and siRNA-transfected cells (60 nM ARF6 siRNA#1 and 5 nM Alexa Fluor 546-coupled negative control siRNA: red) were fixed and stained with a specific antibody directed against the heavy chain of clathrin and an Alexa Fluor 488-coupled murine secondary antibody (green). Visualization of the pits was achieved by scanning the bottom section of the cell. The The data represent the mean Ϯ S.E. of 4 -7 independent experiments. Statistical significance was determined using a Mann-Whitney t test (PRISM software) (*, p Ͻ 0.05; **, p Ͻ 0.01). B, FLAG-VIPR distribution in control and ARF6 siRNA#1-transfected HEK 293 cells was examined by confocal microscopy before and after vasoactive intestinal peptide stimulation (VIP; 0.1 M, 15 min) similarly as in Fig. 2A. Left panels illustrate the plasma membrane distribution of the VIPRs before agonist activation; the right panel shows the internalized receptors after VIP application in both control (upper panels) and ARF6 siRNA#1 transfected cells (lower panels). Most VIPRs are internalized upon agonist-stimulation even when endogenous ARF6 levels are down-regulated. The bar represents 10 m. This figure is representative of the results obtained in three independent experiments where more than 50 cells were examined. transfection of the siRNA had no effect on the internalization of this receptor (Fig. 4A). Immunofluorescent imaging was used to further characterize this observation (Fig. 4B). Before agonist stimulation, VIPRs were present at the plasma membrane in both control and ARF6-depleted cells (Fig. 4B, left panels). Upon VIP exposure, most receptors were internalized in both control and ARF6-depleted cells (Fig. 4B, right panels). These data suggest that the internalization of the VIPR is not dependent on ARF6. Moreover, we examined whether ARF6 depletion affected second messenger production for this specific receptor.
Our results indicate that transfection of the ARF6 siRNA did not have any effect on the ability of the VIPR to promote generation of cAMP following agonist stimulation (data not shown). Taken together, our results clearly demonstrate the importance of ARF6 for the internalization of receptors utilizing the caveolae pathway. However, there are exceptions: not all receptors require ARF6 to be sequestered from the cell surface following agonist stimulation.
Finally, we examined the role of ARF6 in the internalization process of the acetylcholine muscarinic M2 receptor (M2MR). This receptor is sequestered via an alternative endocytic route, the NCCV pathway. Its internalization cannot be blocked by dominant negatives of dynamin and ␤-arrestin or by overexpression of the ARF GAP GIT1 (19). However, one report has shown that expression of an ARF6 mutant can block the internalization of this receptor (25). Here, 27 Ϯ 7% of the M2MRs were found to internalize following acetylcholine stimulation (Fig. 5). Transfection of the ARF6 siRNA resulted in a significant decrease of receptor endocytosis (70%). Indeed, in ARF6depleted cells, only 8 Ϯ 4% of the receptors were internalized following agonist stimulation.
Taken together, our results suggest that ARF6 is necessary to drive the endocytic process of a variety of G protein-coupled receptors being internalized via the CCV, the caveolae, and the NCCV pathways. However, this cannot be considered a general mechanism, because some receptors, namely the VIPR, do not require ARF6 to internalize. DISCUSSION Agonist-promoted endocytosis is an important feature of most G protein-coupled receptors. This phenomenon allows the activation of downstream signaling events, the resensitization of the receptors or their degradation. This family of receptors has been reported to utilize numerous endocytic pathways that are regulated by different sets of regulatory proteins. We have previously suggested that ARF6, together with its regulatory proteins, play an important role in modulating the endocytosis of the ␤ 2 AR (9,19). Moreover, the possible involvement of ARF proteins was demonstrated by the ability of the ARF GAP GIT1 to modulate the internalization of several receptors being sequestered via the clathrin-coated vesicle pathway. Overexpression of mutant forms of ARF6 or its regulatory proteins have been the most commonly used strategies to delineate the importance of GTP-binding proteins in endocytosis and trafficking. The RNA interference technology has provided a new tool for studying the relevance of endogenously expressed proteins. Transfection of siRNAs has been shown to substantially and specifically knock down targeted proteins. Here, we have designed two siRNAs that proved to be equally effective in inhibiting ARF6 expression in HEK 293 cells. Interestingly, while some groups have observed over 95% inhibition in some cases, we were not able to achieve such high inhibition levels without transfecting drastically high concentrations of siRNA, which affected nonspecifically the expression of other proteins. However, we did achieve significant ARF6 knock-down in our optimal transfection conditions to observe marked physiological effects. Surprisingly, sustained reduction of endogenous ARF6 expression did not prove to be detrimental to the cells. We did observe a very small level of cell death (data not shown), but this could be attributed to the very nature of siRNAs, because the same extend of cell death was observed with the control siRNA. Moreover, cell morphology was not affected by siRNA transfection, because microscopy studies did not reveal any differences in the shape of mock-transfected or siRNA-transfected cells. Most importantly, receptor targeting to the plasma membrane was unaffected by ARF6-specific siRNA transfection. For all the receptors that we tested, cell surface expression as determined by FACS was similar for siRNA-transfected cells and cells transfected with receptors alone.
The study of the internalization profile of several receptors allowed us to elucidate the importance of ARF6 proteins in that complex cellular event. First, we found that depleting ARF6, in HEK 293 cells, severely impaired the agonist-dependent internalization of the ␤ 2 AR via the CCV pathway without affecting the basal endocytosis, the integrity of the native pits, or receptor functionality. This observation supports our previous report where we initially proposed a role for ARF6 GTP-binding proteins in the internalization of the ␤ 2 AR (9). To examine the importance of ARF6 in this particular endocytic route, we tested whether reduced levels of ARF6 could also affect the internalization of the AT1R and the V2R. Our data showed that both receptors were greatly impaired in their ability to internalize following agonist stimulation in ARF6-depleted cells. Taken together, these results provide strong evidence that ARF6 is essential for the endocytosis of several G proteincoupled receptors via the CCV pathway.
We were then intrigued to see whether ARF6 could be directly implicated in the internalization of G protein-coupled receptors via the caveolae pathway. We did not expect this small GTP-binding protein to play an important role, because overexpression of GIT1, an ARF GAP, had been shown to have no effect on the endocytosis of the ET B R, and VIPR, two receptors being internalized via the caveolae pathway (19). The internalization route utilized by these two receptors in our cells was confirmed using a specific inhibitor of the caveolae pathway. Surprisingly, we found that the internalization of ET B R was impaired in cells exhibiting reduced levels of ARF6 suggesting that ARF6 can also modulate receptor internalization via the caveolae pathway. Interestingly, the ARF GAP necessary to promote the hydrolysis of GTP on ARF6 must be different from GIT1. On the other hand, the internalization of the VIPR was not affected by the decrease of endogenous ARF6 expression. These results demonstrate that ARF6 is not essential for the internalization of all receptors. In some cases, another ARF family member may regulate the process. Indeed, some receptors have been found in complex with ARF1 (31)(32)(33). Even if this ARF isoform is mainly present on membranes of intracellular organelles such as the Golgi (34,35), its presence at the plasma membrane has been previously reported (36). In the case of the VIP receptor, ARF1 or another member of this family of proteins may be important to regulate the endocytic process.
Lastly, we examined the internalization of the M2MR known to utilize a poorly described alternative endocytic route, the NCCV pathway. Similarly to most other receptors studied, we observed that reduced levels of ARF6 did inhibit the agonistpromoted internalization of the M2MR. This result, together with other published observation where mutant forms of ARF6 were overexpressed (25), strongly hints at a preponderant role for ARF6 in this lesser known endocytic pathway. Using a similar RNA interference approach in HeLa cells, it was recently demonstrated that ARF6 was also involved in the clathrin-independent endocytosis of Tac, the interleukin-2 receptor ␣-subunit (37).
Taken together, these results convincingly highlight the essential role of ARF6 proteins in the regulation of the endocytic process of most G protein-coupled receptors. However, we have found an exception: the VIP receptor. In some cases, an ARF isoform other than ARF6 may be required. Our laboratory is presently investigating that possibility. Furthermore, the need for ARF6 in regulating the endocytic process of receptors may be determined by the nature of the receptor itself. Indeed, all the receptors tested in this study belong to the rhodopsin-like family of receptors (family 1), with the exception of the VIP receptor. The latter is a member of the glucagon-, secretin-like family of receptors (family 2). These receptors are structurally different from the family 1, they do not contain within their structures the DRY or NPXXY motifs, and the other features that characterize this family of receptors. We are further investigating the role of the different ARF proteins in regulating the internalization process of family 2 and 3 receptor members. The importance of the nature of the receptor in utilizing ARF6dependant signaling events to regulate endocytosis has been highlighted by other studies. In HeLa cells, the transferrin receptor, a classic marker of clathrin pathway, does not require ARF6 for its internalization (37) illustrating that not all clathrindependent internalization events requires this small GTPbinding protein. However, in Chinese hamster ovary cells, it was previously demonstrated that ARF6 was important in controlling the trafficking of this same receptor (10) demonstrating once again the difference in signaling pathways between cell types.
Here, the use of the RNA interference approach has been extremely useful to demonstrate the essential role of endogenous ARF6 in the endocytic process of a broad variety of receptors. This method may offer great advantages over the heterologous expression of mutant proteins, which may not be processed normally, or function as we had previously predicted. The precise molecular mechanism by which endogenous ARF6 regulates G protein-coupled receptor endocytosis remains to be elucidated. Our study highlights the importance of the nature of the receptor in activating specific signaling cascades to regulate its own internalization. In most cases, the signaling events are dependent on ARF6 proteins regardless of the en-docytic route being utilized. A closer look at ARF6 interacting partners might provide new insights into the molecular mechanisms regulated by this small GTPase in CCV, caveolae, and NCCV endocytic pathways.