Transfer of M2 Muscarinic Acetylcholine Receptors to Clathrin-derived Early Endosomes following Clathrin-independent Endocytosis*

Upon agonist stimulation, many G protein-coupled receptors such as β2-adrenergic receptors are internalized via β-arrestin- and clathrin-dependent mechanisms, whereas others, like M2 muscarinic acetylcholine receptors (mAChRs), are internalized by clathrin- and arrestin-independent mechanisms. To gain further insight into the mechanisms that regulate M2 mAChR endocytosis, we investigated the post-endocytic trafficking of M2 mAChRs in HeLa cells and the role of the ADP-ribosylation factor 6 (Arf6) GTPase in regulating M2 mAChR internalization. Here, we report that M2 mAChRs are rapidly internalized by a clathrin-independent pathway that is inhibited up to 50% by expression of either GTPase-defective Arf6 Q67L or an upstream Arf6 activator, Gαq Q209L. In contrast, M2mAChR internalization was not affected by expression of dominant-negative dynamin 2 K44A, which is a known inhibitor of clathrin-dependent endocytosis. Nevertheless, M2 mAChRs, which are initially internalized in structures that lack clathrin-dependent endosomal markers, quickly localize to endosomes that contain the clathrin-dependent, early endosomal markers early endosome autoantigen-1, transferrin receptor, and GTPase-defective Rab5 Q79L, which is known to swell early endosomal compartments. These results suggest that M2mAChRs initially internalize via an Arf6-associated, clathrin-independent pathway but then quickly merge with the clathrin endocytic pathway at the level of early endosomes.

Endocytosis is an important mechanism that is used to regulate the activity of a variety of cell surface receptors, including G protein-coupled receptors (GPCRs) 1 (1,2). Defining the cellular mechanisms by which GPCR endocytosis is regulated is important for understanding how cells attenuate GPCR signaling as well as understanding the role of receptor endocytosis in cell signaling. Over the past several years, it has become clear that different GPCRs utilize multiple, distinct pathways of internalization that display differential sensitivities to either pharmacological inhibitors or dominant-inhibitory mutants that are pathway-selective. The mechanisms that regulate the internalization of some receptors, such as ␤ 2 -adrenergic receptors (␤ 2 ARs), have been well characterized. However, the mechanisms that regulate both internalization and sorting of other receptors, such as the M 2 muscarinic acetylcholine receptor (mAChR), are not as well understood (3).
The sequence of events leading to GPCR endocytosis is shared by many receptors. Upon agonist binding, these heptahelical GPCRs activate heterotrimeric G protein signaling pathways, which in turn regulate a variety of intracellular processes (4). Many GPCRs then become rapidly phosphorylated either by second messenger kinases such as protein kinase A or by specific G protein receptor kinases on serine/threonine residues present in cytoplasmically exposed domains (5). In the case of ␤ 2 ARs, the critical serine/threonine residues reside in the cytoplasmic tail, whereas in mAChRs, the relevant residues reside in the third intracellular loop (6). Phosphorylation facilitates the binding of a ␤-arrestin family member, which inhibits further receptor-G protein coupling and thus attenuates receptor signaling. In many instances, the binding of ␤-arrestin targets the GPCR to clathrin-coated pits for rapid endocytosis (7)(8)(9). Dominant-inhibitory mutants of either ␤-arrestins, the 100-kDa GTPase dynamin, or clathrin heavy chain (Hub mutants) inhibit the internalization of GPCRs such as ␤ 2 ARs, which use clathrin-dependent mechanisms (1,10).
In contrast to the ␤ 2 AR, several observations indicate that the G i -linked, M 2 mAChR is internalized via a poorly characterized clathrin-independent endocytic pathway (3,(11)(12)(13)(14)(15)(16). First, although ␤-arrestin binding to this receptor is essential for signal attenuation, it is not required for M 2 mAChR endocytosis in HEK 293 cells (11,14). Second, dominant-negative clathrin Hub mutants do not inhibit the agonist-induced M 2 mAChR internalization (14). Finally, M 2 mAChR internalization shows a differential sensitivity to mutants of dynamin, which is required for clathrin-and caveolae-mediated internalization (17)(18)(19). Mutant dynamin K44A potently inhibits the clathrin-dependent internalization of GPCRs but has little effect on M 2 mAChR internalization. However, recent studies have shown that M 2 mAChR internalization is strongly inhibited by mutants of dynamin that lack the N-terminal GTPbinding domain (⌬1-272 dynamin) or the K535M dynamin mutant, which cannot be stimulated by phosphatidylinositol 4,5-bisphosphate (17). This suggests that although M 2 mAChR internalization is clathrin-and ␤-arrestin-independent, dynamin is still required. These findings raise the question: what is the endocytic pathway by which M 2 mAChRs are internal-ized? One possible regulator of M 2 mAChR trafficking might be the ADP-ribosylation factor 6 (Arf6) GTPase, which has been shown to influence both clathrin-independent and clathrin-dependent endocytic trafficking.
The Arf family of Ras-related GTPases is known to regulate intracellular trafficking processes in both the endocytic and secretory pathways (20,21). The Arf6 GTPase, which is located at the plasma membrane (PM) and on endosomal structures, has been shown to influence the actin cytoskeleton (22)(23)(24) as well as clathrin-independent and clathrin-dependent endocytic processes (25). Activation of Arf6 can initiate cortical actin rearrangements to form either protrusive structures or lamellar structures in cells (22,23). It has recently been shown that the effects of Arf6 on membrane traffic and actin rearrangements in cells involves the localized elevation of cellular phosphatidylinositol 4,5-bisphosphate levels through the stimulation of phosphatidylinositol 4-phosphate 5-kinase ␣ (26,27).
Arf6 is also involved in the regulation of membrane trafficking in the endocytic pathway and has been shown to influence endosomal membrane recycling via clathrin-independent mechanisms (25), clathrin-dependent trafficking of transferrin receptors (28), Fc ␥ receptor-mediated phagocytosis in macrophages (29), apical endocytosis of polymeric IgA receptors in Madin-Darby canine kidney cells (30), and exocytosis of chromaffin granules (31). More recently, Arf6 has been implicated in the ␤-arrestinand clathrin-dependent endocytic trafficking of ␤ 2 ARs (32,33). It was shown that agonist-bound ␤ 2 ARs stimulated the ␤-arrestin-mediated activation of Arf6, which was required for the efficient internalization of these receptors (32). In addition, it has been shown that overexpression of the ARF6 exchange factor ARNO stimulates the release of ␤-arrestin from membrane-docking sites, which then binds to the luteinizing hormone/choriogonadotropin receptor to mediate desensitization (34).
In this study, we investigated the intracellular trafficking of the M 2 mAChR in HeLa cells. Our studies suggest that M 2 mAChRs are rapidly internalized by a clathrin-independent pathway, which is sensitive to mutants of Arf6. After internalization, M 2 mAChRs, which are initially observed in structures that lack clathrin-dependent endosomal markers, localize to early endosomes of the clathrin pathway that contain the early endosome autoantigen 1 (EEA-1) and internalized transferrin receptors. Thus, the M 2 mAChR appears to utilize a novel endosomal trafficking pathway whereby it is transferred from a clathrin-independent endocytic pathway to endosomal compartments of the clathrin-dependent pathway.

EXPERIMENTAL PROCEDURES
Reagents and Antibodies-Rat monoclonal antibodies against human M 2 mAChRs were purchased from Chemicon International (Temecula, CA), and mouse antibodies against the early endosomal marker, EEA-1, were obtained from Transduction Laboratories (Burlingame, CA). Mouse antibodies against the human transferrin receptor (clone B3/25) were purchased from Roche Biochemicals. Alexa 594-and Alexa 488conjugated goat anti-mouse, goat anti-rat, and goat anti-rabbit IgG and Alexa 594-labeled transferrin were purchased from Molecular Probes, Inc. (Eugene, OR). 3 H-Labeled N-methylscopolamine ([ 3 H]NMS) was purchased from PerkinElmer Life Sciences. All other reagents were obtained from Sigma.
Cell Culture and DNA Transfections-HeLa cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 IU/ml penicillin, and 100 g/ml streptomycin (complete medium) at 37°C with 5% CO 2 . The HeLa cells were either grown on glass coverslips (for immunolocalization) and transfected in 6-well dishes or grown in 12-well dishes (for [ 3 H]NMS binding) using FuGENE 6 (Roche Biochemicals) according to the manufacturer's directions. Plasmids encoding M 2 mAChRs were transiently transfected into HeLa cells at 1 g/well (6-well dish). Wild type and mutant Arf6 plasmids were transiently transfected at 0.5 g of plasmid DNA/well (6-well dish), whereas G␣ q plasmids and GFP-Rab5 plasmids were transiently transfected into 6-well dishes using 2.5 g of DNA; wild type and mutant dynamin plasmids were transfected using 10 g of plasmid/well.
To generate stable HeLa cell transfectants, wild type M 2 mAChR plasmid was transfected into HeLa cells using the calcium phosphate co-precipitation method (25). Thirty-six hours after transfection, the cells were detached and replated at a 1:25 dilution into complete medium containing 600 g/ml G418 (Invitrogen). Approximately 2 weeks later, G418-resistant clones were amplified and tested for M 2 mAChR expression by indirect immunofluorescence microscopy. The positive clones were expanded, and one of these (clone 17) was found to express moderate levels of M 2 mAChRs, as judged by relative fluorescence intensity, and was chosen for further studies.
Indirect Immunofluorescence-The cells were treated as described in the figure legends at 30 -36 h following transfection, fixed in 2% formaldehyde in phosphate-buffered saline (PBS) for 10 min, and rinsed with 10% fetal bovine serum and 0.02% azide in PBS (PBS-serum). Fixed cells were incubated with primary antibodies diluted in PBSserum containing 0.2% saponin for 45 min and then washed (three times, 5 min each) with PBS-serum. The cells were then incubated in fluorescently labeled secondary antibodies diluted in PBS-serum plus 0.2% saponin for 45 min, washed with PBS-serum (three times, 5 min each) and once with PBS, and mounted on glass slides.
For Alexa 594-transferrin internalization, transfected cells expressing M 2 mAChRs were briefly rinsed three times with 0.5% bovine serum albumin in Dulbecco's modified Eagle's medium and incubated in the same medium for 30 min at 37°C. The cells were then incubated with both 1 mM carbamoylcholine chloride (carbachol) and Alexa 594-transferrin (50 g/ml) for various times, briefly rinsed with a acid wash (0.5% acetic acid, 0.5 M NaCl, pH 3.0) and complete medium, and then fixed and processed for immunofluorescence localization of M 2 mAChRs. The samples were observed using an Olympus BX40 epifluorescence microscope equipped with a 60ϫ Plan pro lens, and photomicrographs were prepared using a Spot RT monochrome C digital camera (Diagnostic Instruments, Inc., Sterling Heights, MI) or were observed and photographed with a Zeiss LSM 510 laser scanning confocal microscope.
Loss of Cell Surface Receptor Assay-1.5 ϫ 10 5 transfected HeLa cells were grown in 4-or 12-well dishes for 24 h in complete Dulbecco's modified Eagle's medium and treated with or without 1 mM carbachol as described in the figure legends and then rinsed three times with ice-cold PBS, pH 7.4, on ice. Surface M 2 mAChRs were detected as described (14,35) by incubating cells with the cell-impermeant muscarinic ligand, [ 3 H]NMS (2 nM), for 2 h at 4°C. The cells were washed three times (5 min each) with ice-cold PBS and solubilized with 1% Triton X-100 in PBS for 10 min. The cell extracts were transferred to microcentrifuge tubes, and the insoluble material was pelleted by microcentrifugation at 14,000 rpm for 15 min at 4°C. The supernatant was collected, and the protein concentration was determined from an aliquot using a micro-BCA protein assay (Pierce). The radioactivity present in the remaining sample was determined by scintillation counting. Nonspecific binding of [ 3 H]NMS to untransfected HeLa cells was subtracted from the transfected samples. The mass of [ 3 H]NMS bound to cells (in fmol) was calculated using the bound radioactive counts/min and the specific activity of the [ 3 H]NMS (70 Ci/mmol); this was normalized to the protein content of the samples. Receptor internalization is defined as the percentage of surface M 2 receptors not accessible to [ 3 H]NMS at each time relative to non-carbachol-treated cells.
Statistical Analysis-The data are expressed as the means Ϯ S.E. from the indicated number of independent experiments performed in either triplicate or quadruplicate. Statistical analysis was performed using a single-factor analysis of variance followed by a Tukey's statistical test.

Agonist-induced Internalization and Recycling of M 2 mAChRs in HeLa
Cells-To investigate the intracellular trafficking of the M 2 mAChR, HeLa cells were transiently transfected with a plasmid encoding the wild type human M 2 mAChR. Using a rat monoclonal antibody against M 2 mAChRs and confocal microscopy, we examined the effects of agonist stimulation on the cellular distribution of M 2 mAChRs (Fig.  1a). In untreated cells, M 2 mAChRs were distributed in a diffuse pattern along the PM. The addition of 1 mM carbachol induced a rapid redistribution of surface M 2 mAChRs into numerous punctate endosomal structures within 15 min at 37°C. These endosomal structures were initially dispersed near the cell periphery but became clustered in the juxtanuclear region of cells within 20 -30 min. This distribution did not noticeably change even up to 60 min of incubation with carbachol.
Next, we quantified the kinetics of M 2 mAChR internalization in stably transfected HeLa cells for comparison with the behavior of M 2 mAChRs in stably transfected HEK 293 cells (35). M 2 mAChRs exhibit rapid internalization kinetics in response to agonist stimulation but slow and incomplete recycling behavior in HEK 293 cells. Stably transfected HeLa cells expressing M 2 mAChRs were incubated with 1 mM carbachol for various times at 37°C and then chilled to 4°C, and the loss of surface receptor-binding sites for the cell-impermeant mAChR ligand [ 3 H]NMS was determined (Fig. 1b). Within 10 min after the addition of carbachol, ϳ80% of the initial surface [ 3 H]NMS-binding sites is lost. This coincides with the time of appearance of M 2 mAChRs within punctate endosomal structures (Fig. 1a). The kinetics of M 2 internalization was also determined in transiently transfected cells (see Fig. 5, diamonds), where we observed that ϳ50% of surface M 2 mAChRs were internalized after 15 min of agonist treatment. This is consistent with the immunofluorescence observations in Fig.  1a showing both PM and endosomal staining in transiently transfected cells after 15 min of agonist treatment. Thus, M 2 mAChRs are rapidly internalized in HeLa cells, which is similar to what has been reported for M 2 mAChR internalization in HEK 293 cells (35).
Next, we examined whether internalized M 2 mAChRs recycled back to the cell surface upon agonist removal (Fig. 1c). Stably transfected cells were treated with 1 mM carbachol for 60 min at 37°C, washed, and warmed for various times in growth medium that did not contain carbachol, prior to quantifying the reappearance of surface [ 3 H]NMS-binding sites. After 2 h of agonist removal, we observed a recovery of surface M 2 mAChRs to ϳ50% of the level observed in unstimulated cells. This was not due to new synthesis of M 2 mAChRs, because the observed increase in [ 3 H]NMS-binding sites after 2 h of agonist removal was the same when measured in the presence or absence of 20 g/ml cycloheximide (data not shown). These results indicated that M 2 mAChRs were rapidly internalized in HeLa cells in response to agonist stimulation, whereas M 2 mAChR recycling was a relatively slow and incomplete process. The kinetics of internalization and recycling were similar to those described for M 2 mAChRs expressed in HEK 293 cells (35).
In the studies below, we used a transient co-transfection approach to investigate the effects of different mutant proteins, known to affect endocytosis, on M 2 mAChR internalization. Although the overall transfection efficiency in our experiments ranged from 30 to 50%, the co-transfection efficiency in these experiments was ϳ95-100% (as judged by indirect immunofluorescence staining for both M 2 mAChRs and a given mutant protein) (data not shown). This permitted us to quantify the effects of endocytic mutants on M 2 mAChR internalization without background from cells that expressed M 2 mAChRs but not an endocytic mutant.
Previous studies have shown that agonist-induced internalization of M 2 mAChRs in HEK 293 cells occurs via clathrinindependent endocytosis (14). M 2 mAChR internalization is insensitive to inhibition by the K44A mutant of the 100-kDa GTPase, dynamin (dyn) (10), which mediates the scission of clathrin-coated pits at the PM and Golgi complex and also the scission of caveolae (18,19,36). To investigate whether M 2 mAChR internalization involved clathrin-independent mechanisms in HeLa cells, we transiently co-transfected HeLa cells with plasmids encoding green fluorescent protein (GFP)-tagged dynamin 2 K44A (dyn2-GFP K44A) and M 2 mAChR or M 2 mAChR plasmid alone and determined the effects of agonist stimulation on M 2 receptor internalization (Fig. 2). In cells expressing M 2 mAChR alone or M 2 mAChR and dyn2-GFP K44A, treatment with 1 mM carbachol for 30 min induced M 2 mAChR internalization into numerous punctate endosomal structures (Fig. 2, A and B). In contrast, expression of dyn2-GFP K44A in HeLa cells completely blocked the clathrin-dependent internalization of Alexa 594-labeled transferrin (Fig.  2C). Thus, agonist-induced internalization of M 2 mAChRs in HeLa cells appears to be mediated by clathrin-independent mechanisms.
Internalized M 2 mAChRs Merge with Early Endosomes of the Clathrin-dependent Endocytic Pathway-Given the findings above, we sought to determine the identity of the endosomal structures to which internalized M 2 mAChRs localized following agonist treatment. We performed double labeling immunofluorescence experiments using antibodies against known markers of endosomal compartments along with antibodies against M 2 mAChRs. We compared the distribution of inter-nalized M 2 with that of the early endosome marker EEA-1, with transferrin receptors, and with the lysosomal marker, LAMP2. We observed that the M 2 mAChR ϩ endosomal structures extensively co-localized with the early endosomal marker, EEA-1, and to a lesser extent with transferrin receptors (Fig. 3). This suggested that once M 2 mAChRs are internalized via a clathrin-independent pathway, they are transferred to early endosomes of the clathrin endocytic pathway. We also examined the localization of internalized M 2 mAChRs in cells expressing GTPase-defective Rab5 Q79L, which is known to stimulate homotypic early endosomal fusion and results in the swelling of early endosomes (37). Treatment of cells co-expressing M 2 and GFP-Rab5 Q79L with 1 mM carbachol resulted in the localization of M 2 mAChRs to very large swollen endosomal structures that extensively co-localized with GFP-Rab5 Q79L. This indicated that M 2 mAChRs, once internalized by clathrin-independent mechanisms, indeed become localized to early endosomes of the clathrin-mediated endocytic pathway. In contrast to early endosomal markers, internalized M 2 mAChRs did not significantly co-localize with the lysosomal marker, LAMP2 (data not shown).
To further explore this apparent transfer of M 2 mAChRs between the clathrin-independent and clathrin-dependent pathways, we compared the intracellular localization of M 2 mAChRs and Alexa 594-labeled transferrin over time in cointernalization experiments (Fig. 4). Transiently transfected

FIG. 3. Internalized M 2 mAChRs co-localize with clathrindependent early endosomal markers. A, transiently transfected
HeLa cells expressing M 2 mAChRs were treated with 1 mM carbachol at 37°C for 30 min, fixed, and processed for immunofluorescence colocalization of M 2 mAChRs, using rat anti-M 2 mAChR monoclonal antibodies and using mouse monoclonal Abs against EEA-1 or transferrin receptors. Primary antibodies were visualized using fluorescently labeled, non-cross-reactive secondary antibodies. The arrows indicate endosomal profiles that co-label with antibodies against both M 2 mAChRs and EEA-1. B, transiently transfected HeLa cells expressing M 2 mAChRs and GFP-Rab5 Q79L were treated with 1 mM carbachol for 30 min prior to fixation and indirect immunofluorescence. Bar, 10 M.
HeLa cells expressing M 2 mAChRs were simultaneously incubated with 1 mM carbachol and with 50 g/ml Alexa 594labeled transferrin for various times at 37°C, and the distributions of the two proteins were determined. After 5 min of internalization, M 2 mAChRs localized to large endosomal structures, whereas Alexa 594-transferrin localized to distinct endosomal structures that lacked M 2 mAChRs. After 15-30 min of internalization, both M 2 mAChRs and Alexa 594-transferrin extensively co-localized to large endosomal structures in the juxtanuclear region of cells. These results indicated that M 2 mAChR and transferrin are initially internalized in discrete endosomal carriers but then converge in common endosomal structures.
Arf6 Involvement in M 2 mAChR Internalization-To date, very little is known about the regulation of the endocytic pathway followed by the M 2 mAChR. We have previously shown that the Arf6 GTPase regulates endocytic trafficking along a clathrin-independent pathway and delivers cargo to tubular endosomal structures that emanate from the centrosomal region of cells (25). Using the interleukin-2 receptor ␣ subunit Tac as an endocytic marker of this pathway, we observed that expression of the GTPase-deficient Arf6 mutant, Q67L, inhibits internalization of Tac into cells when assessed after 40 h of transfection. In contrast, expression of the GTP binding-defec-tive Arf6 mutant T27N inhibits the recycling of internalized Tac back to the cell surface. Recent studies have also shown that both Arf6 Q67L and Arf6 T27N inhibit the clathrin-mediated internalization of ␤ 2 ARs (32,33). However, nothing is known about the effects of these Arf6 mutants on clathrinindependent GPCR internalization.
We first investigated the effects of wild type (WT) Arf6 and GTP binding-defective Arf6 T27N on M 2 mAChR internalization in transiently transfected HeLa cells by measuring the agonist-induced loss of surface M 2 mAChRs with the [ 3 H]NMS binding assay (Fig. 5). We observed that neither co-transfection with WT Arf6 nor Arf6 T27N significantly affected M 2 mAChR internalization after either 15 or 30 min of agonist treatment. After 30 min of carbachol treatment, ϳ60% of surface M 2 receptors were internalized in cells expressing either M 2 mAChR alone or M 2 mAChR co-transfected with either WT Arf6 or Arf6 T27N.
We next examined the effects of persistent activation of Arf6 in cells by expressing the GTPase defective mutant of either Arf6 itself (Arf6 Q67L) or of an upstream activator of Arf6, HA-tagged G␣ q Q209L (HA-G␣ q Q209L), by measuring the loss of surface M 2 mAChRs with the [ 3 H]NMS binding assay. A recent study by Boshans et al. (23) showed that co-expression of wild type Arf6 and a constitutively activated mutant of G␣ q could mimic bombesin-induced activation of Arf6 in Chinese hamster ovary cells. We have previously observed that cotransfection of WT Arf6 and HA-G␣ q Q209L together induces cell surface protrusions, 2 a phenotype that is indicative of Arf6 activation (22). Fig. 6 shows that in cells expressing M 2 mAChR alone, treatment with 1 mM carbachol for 30 min induced the internalization of ϳ78% of surface M 2 mAChRs. In contrast, only 36 and 39% of surface M 2 mAChRs were internalized in cells co-expressing either Arf6 Q67L or G␣ q Q209L, respectively. This suggested that persistent activation of Arf6 inhibited M 2 mAChR internalization.
In these experiments, we also observed that cells co-transfected with Arf6 Q67L expressed significantly fewer surface M 2 mAChRs (116 Ϯ 57 fmol [ 3 H]NMS bound/mg protein or 7 Ϯ 2%) when compared with cells expressing M 2 mAChRs alone (1977 Ϯ 979 fmol [ 3 H]NMS bound/mg protein). This suggested that M 2 receptors may be sequestered within intracellular compartments even in unstimulated Arf6 Q67L-transfected cells. To further investigate these effects of Arf6 Q67L and G␣ q Q209L on M 2 internalization, we examined the effects of these mutants on the intracellular localization of M 2 mAChR using indirect immunofluorescence microscopy (Figs. 7 and 8).  untreated cells, which co-expressed M 2 mAChR and Arf6 Q67L, M 2 mAChRs were localized both at the cell surface and also to large endosomal clusters where M 2 mAChRs co-localized with Arf6 Q67L (Fig. 7B). These results indicated that there indeed was significant intracellular accumulation of M 2 mAChRs in unstimulated Arf6 Q67L-transfected cells, which is consistent with the reduced levels of surface M 2 mAChRs observed in these cells. Recent studies by Brown et al. (27) showed that these large Arf6 Q67L ϩ endosomal clusters are enriched in phosphatidylinositol 4,5-bisphosphate and F-actin. These cells also exhibited an altered morphology with many protrusive structures, which we have previously shown to be a phenotype of cells expressing Arf6 Q67L (22,27). Following carbachol treatment, the M 2 receptors remained in a localization pattern that was indistinguishable from untreated Arf6 Q67L-transfected cells (Fig. 7B) and did not show the punctate vesicular structures observed in cells expressing M 2 mAChR alone (Fig.  7A). The results of the [ 3 H]NMS binding experiments above (Fig. 6) further indicate that Arf6 Q67L inhibits the internalization of the M 2 mAChRs that are present at the cell surface in these cells. Expression of Arf6 Q67L did not affect the clathrin-mediated endocytosis of Alexa 594-Tfn, which is consistent with our previous studies showing that Arf6 Q67L does not inhibit transferrin internalization in HeLa cells (25).
In cells co-expressing M 2 mAChR and HA-G␣ q Q209L, the M 2 receptors were primarily localized at the cell surface (Fig.  8B). Following carbachol treatment, cells expressing HA-G␣ q Q209L contained far fewer M 2 mAChR-labeled endosomal structures than cells expressing M 2 mAChR alone. As observed with Arf6 Q67L, expression of HA-G␣ q Q209L did not affect the clathrin-mediated internalization of transferrin. Taken together, these results indicated that persistent activation of Arf6 via expression of constitutively activated Arf6 Q67L or the upstream activator, G␣ q Q209L, inhibited M 2 mAChR internalization but that co-expression of WT Arf6 and GTP bindingdefective Arf6 T27N did not. DISCUSSION In this study, we investigated the intracellular trafficking of the M 2 mAChR via a clathrin-independent endocytic pathway in HeLa cells. The M 2 receptor has been shown to be internal-ized in HEK293 cells by a poorly characterized, clathrin-and arrestin-independent mechanism (10 -14, 38). Our present results indicate that persistent activation of the small GTPase Arf6 inhibits M 2 mAChR internalization but does not affect the clathrin-mediated internalization of transferrin receptors. Interestingly, following internalization in structures that initially lack clathrin endocytic markers, M 2 mAChRs rapidly localize to early endosomes of the clathrin endocytic pathway where they co-localize with the endosomal marker EEA-1 and with internalized Alexa 594-transferrin. This raises the intriguing possibility that although the M 2 mAChR is initially internalized via an Arf6-associated pathway, it is quickly transferred to endosomes of the clathrin-dependent pathway.
We observed that dominant-inhibitory dynamin 2 K44A did not interfere with the internalization of M 2 receptors in HeLa cells but completely blocked the clathrin-dependent internalization of transferrin. This suggested that M 2 internalization was clathrin-independent in HeLa cells and is consistent with published reports that the K44A mutant of dynamin does not inhibit M 2 mAChR internalization in HEK293 cells (10). Recent reports also indicate that M 2 mAChRs and angiotensin AT 1A receptors exhibit a differential sensitivity to mutants of dynamin (17) and suggest that although M 2 mAChR internalization is clathrin-independent, it still requires dynamin.
To gain some insight into the nature of the endocytic pathway followed by M 2 mAChR, we investigated the identity of the endosomal structures to which internalized M 2 receptors localized. To our surprise, there was extensive overlap in staining between internalized M 2 mAChR and the early endosomal marker EEA-1. EEA-1 is a Rab5 effector that is recruited to the cytosolic leaflet of early endosomes in a phosphatidylinositol 3-phosphate-dependent manner (39). It is involved in both the tethering of vesicles and their subsequent fusion through the assembly of oligomeric complexes that mediate vesicle fusion. Our observations that internalized M 2 mAChRs also co-localize to the swollen early endosomal structures induced by expression of constitutively activated Rab5 Q79L lend further support to the early endosomal localization of internalized M 2 receptors. A recent study has shown that internalized M 4 mAChRs also localize to EEA-1 ϩ endosomes and with internalized transferrin receptors in PC12 cells (40). Furthermore, M 4 mAChRs were shown to localize to swollen multivesicular endosomes formed in cells expressing constitutively activated Rab5 Q79L. In other studies, M 4 mAChRs have been shown to internalize via arrestin-and clathrin-dependent mechanisms (14). This suggests that different mAChRs that are internalized by distinct mechanisms can meet up in and transit through common endosomal compartments. In support of this hypothesis, we also observed that M 2 mAChRs and fluorescently labeled transferrin are initially internalized in distinct endosomal structures but converge into common endosomal structures within 15 min after internalization.
But what is the clathrin-independent pathway by which the M 2 receptor is initially internalized from the PM? Our results implicate a pathway that is regulated by the Arf6 GTPase. We have previously shown that the Arf6 GTPase regulates a clathrin-independent endocytic pathway between the PM and tubulovesicular endosomes in HeLa cells (25). Arf6 itself cycles between these two cellular locations according to its GTP cycle (41). We and others have shown that GTP-bound Arf6 preferentially localizes to the PM, whereas GDP-bound Arf6 localizes to a juxtanuclear tubular endosomal compartment (25,41,42). Constitutively activated Arf6 Q67L inhibits the internalization of proteins that transit the Arf6-regulated pathway, such as the major histocompatability complex I, whereas GTP bindingdefective Arf6 T27N inhibits recycling of such proteins from tubulovesicular endosomes back to the PM (25,27).
Two observations suggest that the M 2 mAChR is also internalized via an Arf6-associated endocytic pathway. First, Arf6 Q67L strongly inhibited the agonist-induced internalization of M 2 receptors (Fig. 6) but did not perturb the clathrin-mediated internalization of fluorescently labeled transferrin (Fig. 7). We also observed that M 2 mAChRs were sequestered into large endosomal clusters even in the absence of agonist stimulation, where they co-localized with Arf6 Q67L (Fig. 7), suggesting that M 2 mAChRs were internalized earlier in the transfection. Brown et al. (27) recently reported that major histocompatability complex I co-localizes with Arf6 Q67L after 20 h of transfection in tightly clustered endosomal structures, which accumulate in cells with time after transfection and are highly enriched in phosphatidylinositol 4,5-bisphosphate and F-actin. In contrast, after 40 h of transfection, major histocompatability complex I internalization was strongly inhibited by Arf6 Q67L. One explanation for the different temporal effects of Arf6 Q67L on M 2 mAChR internalization, which we have observed, is that as endosomal membranes accumulate in cells, further internalization is inhibited because of a depletion of either membrane and/or critical components necessary for vesicle formation.
To independently test the effects of persistent Arf6 activation on M 2 mAChR endocytosis, we examined the effects of GTPasedefective G␣ q Q209L on M 2 internalization. We observed that co-transfection of M 2 mAChRs with constitutively activated G␣ q Q209L greatly inhibited M 2 mAChR internalization and to the same extent as Arf6 Q67L (ϳ50% inhibition). Constitutively activated G␣ q mutants have been shown to enhance Arf6 activation and to mimic, in part, the phenotypic effects of Arf6 Q67L in cells (23). We have observed that co-expression of wild type Arf6 and G␣ q Q209L recreates the surface protrusions FIG. 8. Constitutively active G␣ q Q209L inhibits M 2 mAChR but not transferrin internalization. HeLa cells were transiently transfected with plasmids encoding M 2 mAChR alone (A) or M 2 mAChR and HA-G␣ q Q209L (B) and then treated for 30 min with or without 1 mM carbachol, fixed, and processed for indirect immunofluorescence localization of M 2 mAChRs and G␣ q . Bar, 10 m. C, HeLa cells were transiently transfected with a plasmid encoding HA-G␣ q Q209L and incubated with Alexa 594-labeled transferrin for 15 min prior to processing for indirect immunofluorescence (see "Experimental Procedures"). Bar, 10 m. observed following aluminum fluoride stimulation of HeLa cells expressing WT Arf6 alone. 2 A recent study showed that stimulation of the G␣ q -coupled bombesin receptor increased the proportion of GTP-bound Arf6 in Chinese hamster ovary cells. The precise mechanism by which G␣ q Q209L enhances Arf6 activation remains to be determined. The observation that persistent activation of G␣ q can inhibit M 2 mAChR internalization raises the intriguing possibility that G␣ q -coupled receptors may influence the behavior of M 2 receptor trafficking.
We also observed that neither WT Arf6 nor the GTP bindingdefective Arf6 T27N mutant affected M 2 mAChR internalization (Fig. 5). These observations contrast with those of Claing et al. (32), who recently showed that both Arf6 Q67L and Arf6 T27N inhibit the clathrin-and arrestin-mediated internalization of ␤ 2 ARs. These authors also showed that isoproteronol stimulation of ␤ 2 ARs results in the ␤-arrestin-dependent stimulation of GTP exchange onto Arf6. Previous work has shown that overexpression of the Arf6-specific GTPase-activating protein GIT1 also inhibits ␤ 2 AR internalization but not M 2 receptor internalization (33,43). One major difference between ␤ 2 ARs and the M 2 mAChR is that the former requires ␤-arrestins for internalization and the latter does not (11,14). One possible explanation for the differential sensitivities of these two receptors to Arf6 mutants is that ␤-arrestin stimulation of Arf6 activation is not critical for M 2 mAChR internalization but is important for ␤ 2 AR internalization. These results raise the exciting possibility that Arf6 can differentially affect GPCR internalization via both clathrin-dependent and clathrin-independent mechanisms. The significance of such a dual role for Arf6 in endocytic trafficking along distinct pathways is unknown.
Taken together, our results indicate that the clathrin-independent internalization of the M 2 mAChR initially follows an Arf6-associated endocytic pathway but then quickly merges with early endosomes of the clathrin-dependent pathway. From these EEA-1 ϩ endosomes, M 2 mAChRs either can recycle back to the PM, can remain sequestered in endosomes by unknown mechanisms, or eventually could be sorted to lysosomes for degradation. Given the differential effects of Arf6 on the clathrindependent and clathrin-independent internalization of GPCRs, determining the mechanisms by which Arf6 exerts its effects on these two endocytic pathways will be an important goal for future studies.