Involvement of Actin in Agonist-induced Endocytosis of the G Protein-coupled Receptor for Thromboxane A2

The role of actin in endocytosis of G protein-coupled receptors is poorly defined. In the present study, we demonstrate that agents that depolymerize (latrunculin B and cytochalasin D) or stabilize (jasplakinolide) the actin cytoskeleton blocked agonist-induced endocytosis of the β isoform of the thromboxane A2 receptor (TPβ) in HEK293 cells. This suggests that endocytosis of TPβ requires active remodeling of the actin cytoskeleton. On the other hand, disruption of microtubules with colchicine did not affect endocytosis of the receptor. Expression of wild-type and mutant forms of the small GTPases RhoA and Cdc42 potently inhibited endocytosis of TPβ, further indicating a role for the dynamic regulation of the actin cytoskeleton in this pathway. Agonist treatment of TPβ in HEK293 cells resulted in the formation of actin stress fibers through Gαq/11 signaling. Because we previously showed that endocytosis of TPβ is dependent on arrestins, we decided to explore the relation between arrestin-2 and -3 and actin in endocytosis of this receptor. Interestingly, we show that the inhibition of TPβ endocytosis by the actin toxins in HEK293 cells was overcome by the overexpression of arrestin-3, but not of arrestin-2. These results indicate that the actin cytoskeleton is not essential in arrestin-3-mediated endocytosis of TPβ. However, arrestin-3 could not promote endocytosis of the TPβY339A and TPβI343A carboxyl-terminal mutants when the actin cytoskeleton was disrupted. Our data provide new evidence that the actin cytoskeleton plays an essential role in TPβ endocytosis. Furthermore, our work suggests the existence of actin-dependent and -independent arrestin-mediated pathways of endocytosis.

Activation of the thromboxane A 2 receptor (TP) 1 induces platelet aggregation, constriction of vascular and bronchiolar smooth muscle cells, as well as mitogenesis and hypertrophy of vascular smooth muscle cells (1). TP is a member of the class A family of G protein-coupled receptors (GPCRs). Two TP isoforms have been characterized in human: TP␣ (343 amino acids) and TP␤ (407 amino acids), which share the first 328 amino acids but differ in their carboxyl-terminal tail (C-tail) regions (2,3). We previously described that TP␤, but not TP␣, undergoes agonist-induced and constitutive endocytosis, which are mediated by distinct motifs in the TP␤ C-tail (4,5).
TP can activate many G proteins such as G q , G s , G i2 , G 11 , G 12 , G 13 , G 16 , and G z proteins (6 -13). The major signaling pathway of thromboxane A 2 and its receptor is via the activation of the G q/11 family (12). Increasing data indicate that receptor signaling participates in GPCR endocytosis. In this regard, we have recently shown that the first step in the regulation of agonist-mediated endocytosis of TP␤ involves the activation of G␣ q signaling (14). However, the molecular mechanism by which G␣ q regulates TP␤ agonist-induced endocytosis is still unclear and is one of the major research interests of our laboratory. It is known that GPCR-mediated activation of G␣ 12/13 leads to formation of actin stress fibers, through regulation of the small Rho GTPase protein (15). On the other hand, recent studies reported that the activation of the Rho signaling pathway could be achieved by a G␣ q/11 -dependent mechanism and that distinct guanine nucleotide exchange factors for Rho were involved in G␣ q -versus G␣ 12/13 -mediated activation of Rho signaling (16 -18). Experiments performed by Dutt et al. (17) demonstrated that microinjection of activated G␣ q in fibroblasts induced actin stress fiber formation via the small G protein Rho.
An important cellular response to an extracellular signal is the reorganization of the actin cytoskeleton. The actin cytoskeleton has many functions in eukaryotic cells like locomotion by extension of pseudopods, intracellular protein transport, and endocytic processes (19,20). Several studies characterized the role of actin filaments and actin-based motor proteins in endocytosis in yeast (21,22). However, the role of the actin cytoskeleton in GPCR endocytosis in mammalian cells is still controversial. Functional relation between clathrin-mediated endocytosis and the actin cytoskeleton in these cells has been proposed based on recent findings demonstrating that several components of the mammalian endocytic machinery directly or indirectly interact with the actin cytoskeleton (19). Arrestin-2 and arrestin-3 modulate GPCR desensitization, signaling, and act as adapter proteins in GPCR endocytosis (23). These proteins were shown to interact with clathrin and AP-2, two important components of the endocytic coated pits (24 -26). We observed previously that arrestin-2 and -3 promoted agonistinduced endocytosis of TP␤ (4).
Very little is known regarding the relation between arrestins and actin in the regulation of GPCR endocytosis. In the present study, we show that TP␤ endocytosis requires actin dynamics and we characterize the role of the actin cytoskeleton in the arrestin-mediated endocytosis of TP␤. We also demonstrate that TP␤-mediated activation of G␣ q signaling is involved in actin stress fiber formation, which is necessary for agonistinduced endocytosis of TP␤.

EXPERIMENTAL PROCEDURES
Materials-Latrunculin B, cytochalasin D, and colchicine were from BIOMOL, and jasplakinolide from Calbiochem. The hemagglutinin (HA)-specific monoclonal antibody was from Babco. The fluorescein isothiocyanate-conjugated goat anti-mouse and rhodamine phalloidin were purchased from Molecular Probes. The goat anti-mouse alkaline phosphatase-conjugated antibody and the alkaline phosphatase substrate kit were purchased from Bio-Rad. U46619 was from Cayman.
Cell Culture and Transfection-HEK293 cells were maintained in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum (ICN) at 37°C in a humidified atmosphere containing 5% CO 2 . Transient transfections of HEK293 cells grown to 75-90% confluence were performed using FuGENE 6 TM (Roche Applied Science) according to the manufacturer's instructions. Empty pcDNA3 vector was added to keep the total DNA amount added per plate constant. Cells grown on 60-mm plates were transfected using 6 g of total DNA.
Endocytosis Assays-Endocytosis of receptors was measured by ELISA using transiently transfected HEK293 cells as described previously (4). Briefly, 1.2 ϫ 10 6 cells were grown overnight in 60-mm plates. The cells were then transfected with pcDNA3 alone (control) or cotransfected with pcDNA3-HA-TP␤ or either indicated receptor constructs combined with pcDNA3-Arr3WT, pcDNA3-Arr2WT, pcDNA3-Arr3R396A, pcDNA3-Arr3⌬LIEF, pHook2-RhoA, pHook2-Rho(N17), pHook2-Rho(V12), pHook2-Cdc42, pHook2-Cdc42(L61), pHook2-Cdc42(N17), or RGS(GRK2). Overexpression levels of the arrestins, Rho, and Cdc42 constructs were adjusted to be equal between samples and were roughly 5-fold that of endogenous proteins as determined by Western blot analysis with respective specific antibodies. Transfected cells were maintained as described above for 24 h. Thereafter, 2 ϫ 10 5 cells were transferred to 24-well plates pre-coated with 0.1 mg/ml poly-L-lysine (Sigma) and maintained for an additional 24 h. To assess the agonist-induced endocytosis, the transfected cells were washed once with phosphate-buffered saline (PBS) and were pretreated with different cell-permeant actin toxins: latrunculin B (5 mg/ml for 1 h), cytochalasin D (2 mM for 20 min), jasplakinolide (10 mM for 30 min), or Me 2 SO (vehicle), followed by incubation with 100 nM U46619 at 37°C for 2 h in Dulbecco's modified Eagle's medium. Thereafter, the cells were fixed with 3.7% formaldehyde plus TBS (20 mM Tris, pH 7.5, 150 mM NaCl) for 5 min at room temperature followed by washing three times with TBS. Nonspecific binding was blocked with TBS containing 1% bovine serum albumin for 45 min at room temperature. The cells were then incubated with a HA-specific monoclonal antibody (Babco) at a dilution of 1:1000 in TBS/bovine serum albumin for 1 h at room temperature. Three washes with TBS buffer followed, and cells were briefly reblocked for 15 min at room temperature. The cells were incubated with a 1:1000 in TBS/bovine serum albumin of alkaline phosphatase-conjugated goat anti-mouse antibody (Bio-Rad) for 1 h at room temperature. The cells were washed three times with TBS, and a colorimetric alkaline phosphatase substrate (Bio-Rad) was added following the instructions from the manufacturer. The resulting colorimetric reactions were measured using a Titertek MultisKan MCC/340 spectrophotometer. Cells transfected with pcDNA3 were studied concurrently to determine background. Values shown represent the mean Ϯ S.E. of four to six independent experiments performed in triplicate.
Immunofluorescence Microscopy-1.0 ϫ 10 6 HEK293 cells stably transfected or not with the empty pcDNA3 vector (pcDNA3 cells) or pcDNA3-HA-TP␤ (TP␤ cells) were grown overnight in 60-mm plates as described above. The cells were then transiently transfected with the indicated constructs and maintained overnight as described above. Then, 2 ϫ 10 5 cells were transferred on coverslips and further grown overnight. The cells were then stimulated at 37°C with 100 nM U46619 before being fixed with 3% paraformaldehyde plus PBS for 30 min at room temperature. Thereafter, the cells were washed with PBS and permeabilized with 0.1% Triton X-100 plus PBS for 30 min at room temperature. Nonspecific binding was blocked with 0.1% Triton X-100 plus PBS containing 5% nonfat dry milk for 30 min at room temperature. The cells were then incubated with an HA-specific monoclonal (1:500 dilution) for 1 h at room temperature in PBS supplemented with 5% nonfat dry milk. Then, cells were washed three times with 0.1% Triton X-100 plus PBS, followed by incubation with a goat fluorescein isothiocyanate-conjugated anti-mouse secondary antibody (Molecular Probes) at a dilution of 1:200 for 1 h at room temperature. The cells were washed three times with permeabilization buffer and incubated with rhodamine phalloidin (1:200 dilution) in PBS for 30 min at room temperature. Finally, coverslips were mounted using the Vectashield mounting medium (Vector Laboratories) and examined by immunofluorescence microscopy on a Nikon Eclipse TE2000-U microscope using a ϫ60 objective. Images were collected using SimplePCI Camera software and processed with Adobe Photoshop software.
Live Cell Fluorescence Microscopy of Pseudopods Extension-1.0 ϫ 10 6 HEK293 cells stably transfected with the empty pcDNA3 vector (pcDNA3 cells) or pcDNA3-HA-TP␤ (TP␤ cells) were grown overnight in 60-mm plates as described above. The cells were then transiently transfected with pcDNA3-EGFP-TP␤(pcDNA3 cells), pcDNA3-EGFP-Arr2, pcDNA3-EGFP-Arr3, or pcDNA3 (TP␤ cells) and maintained overnight as described above. Then, 2 ϫ 10 5 cells were transferred on coverslips and further grown overnight. The coverslips were mounted on a heating block at 37°C on the microscope. Images were collected each 30 s during 60 min using a Nikon Eclipse TE2000-U microscope with a ϫ60 objective. Agonist stimulation with 1 M U46619 occurred after the capture of two images. Images were collected using SimplePCI Camera software and processed with Adobe Photoshop software.

Agonist-induced Endocytosis of TP␤ Requires Actin Dynamics-
We investigated the role of the actin cytoskeleton in TP␤ agonist-induced endocytosis by ELISA in HEK293 cells transiently expressing HA-tagged TP␤ using two different actin filament-disrupting agents, latrunculin B (Lat B) and cytochalasin D (Cyt D), as well as an actin filament-stabilizing agent, jasplakinolide (Jas). Lat B binds and sequesters actin monomers, whereas Cyt D cups the growing barbed end of actin filaments, both of them preventing actin polymerization by two distinct mechanisms. Cyt D depolymerizes predominantly the actin stress fibers, whereas Lat B disrupts both the cortical actin and actin stress fibers (29 -31). On the other hand, Jas stabilizes actin filaments by binding to their pointed end thereby promoting actin polymerization (32). Microtubules are also important constituents of the cell ultrastructure. Colchicine, a microtubule-disrupting agent, was also tested. As seen in Fig. 1A, disruption as well as stabilization of actin filaments significantly inhibited the agonist-induced endocytosis of TP␤ after 2 h of U46619 stimulation. Endocytosis of TP␤ was inhibited by roughly 70% in the presence of Lat B and Cyt D, whereas Jas decreased it by ϳ60% when compared with vehicle-treated cells. In contrast, colchicine did not affect the ligand-induced endocytosis of TP␤. A detailed time course anal-ysis of agonist-induced endocytosis of TP␤ in the presence of these agents is presented in Fig. 1B. It can be seen that Lat B and Cyt D almost completely abrogated endocytosis of the receptor over the time intervals tested. Jas had a significant inhibitory effect on TP␤ endocytosis as well after 60 min of agonist treatment. Colchicine did not modulate agonist-induced endocytosis of TP␤. These results suggest that actin filaments, but not microtubules, are involved in TP␤ endocytosis. Moreover, our data indicate that TP␤ endocytosis requires actin dynamics because both actin-disrupting and actin-stabilizing agents prevented endocytosis of this GPCR.
The role of actin assembly/disassembly in GPCR endocytosis is still controversial. We thus decided to determine the effects of Lat B, Cyt D, and Jas on endocytosis of the ␤ 2 -adrenergic receptor (␤ 2 AR), the prototypical GPCR for arrestin-and clathrin-mediated endocytosis. Fig. 2 illustrates that the agonistinduced endocytosis of the ␤ 2 AR was not affected by the actininterfering agents used in this study. Thus, contrary to TP␤, agonist-induced endocytosis of the ␤ 2 AR does not appear to require a dynamic actin cytoskeleton, indicating that the role of actin in GPCR endocytosis is specific to some GPCRs. This also showed that the effects of the agents observed on TP␤ endocytosis were not the result of nonspecific cell toxicity.
Small G proteins of the Rho family (Rho, Rac, and Cdc42) are important players in the rearrangement of the actin cytoskeleton. To confirm the implication of actin dynamics in TP␤ trafficking, we analyzed the effect of co-expressing wild-type and mutant RhoA and Cdc42 proteins on the agonist-induced endocytosis of TP␤. The dominant-negative mutants of RhoA (RhoA(V12)) and Cdc42 (Cdc42(L61)), and their respective con-stitutively active mutants, RhoA(N17) and Cdc42(N17), were used. All of the RhoA and Cdc42 constructs significantly affected endocytosis of TP␤, with an inhibition ranging from 43 to 80% (Fig. 3). These results support our idea that TP␤ endocytosis needs a fine regulation of active actin cytoskeleton rearrangement, as will be discussed below.

Stimulation of TP␤ Rapidly Promotes the Formation of Actin
Stress Fibers and Pseudopods-We were then wondering whether we could visualize actin cytoskeleton rearrangement after agonist stimulation of TP␤. For this purpose, HEK293 cells transiently expressing HA-TP treated or not with agonist were incubated with rhodamine-conjugated phalloidin to observe actin filaments by fluorescence microscopy. Phalloidin is a toxin that specifically binds to F-actin. Formation of actin stress fibers induced by activation of TP␤ (Fig. 4A) or TP␣ (data not shown) was detected for different time points analyzed between 2 and 120 min. Interestingly, TP␤-containing endocytic vesicles were aligning themselves on underlying actin filaments (Fig. 4A, middle panel) after 30 min of receptor stimulation. We previously reported that the intracellular compartment where TP␤ accumulated after sustained agonist treatment corresponded to the Rab11-positive recycling endosome (33). Prominent co-localization of actin and receptors was observed on endosomal endocytic compartments (Fig. 4A, bottom panel) after 60 min of incubation with agonist. The fact that both TP␣ and TP␤ induce the formation of actin stress fibers but that only TP␤ undergoes agonist-promoted endocytosis indicates that formation of actin stress fibers is not sufficient for receptor endocytosis. Other elements, such as receptor-specific sequences, must also be involved. To further confirm that the actin cytoskeleton was involved in endocytosis of TP␤, we also performed immunofluorescence microscopy in the presence of Lat B. Cells transfected with pcDNA3-HATP␤ were labeled at 4°C with an anti-HA antibody prior to internalization experiments to allow for detection of only the receptors that are trafficking from the cell surface, as we described previously (5,33,34). The cells were treated with Lat B as described above and were then stimulated or not with 100 nM U46619 for 60 min. As shown in Fig. 4B, treatment with Lat B inhibited both the actin rearrangement and endocytosis of surface receptors following U46619 stimulation when compared with Fig. 4A. The same results were obtained for time points ranging from 0 to 45 min of receptor stimulation (data not shown). These data clearly show that disruption of actin dynamics prevents TP␤ entry from the cell surface into intracellular compartments, supporting our ELISA results from Fig. 1.
It is known that extension of pseudopods requires sophisticated control of the actin cytoskeleton. Ge et al. (35) demonstrated that activation of the GPCR protease-activated receptor 2 promoted ERK1/2-and arrestin-dependent reorganization of the actin cytoskeleton, polarized pseudopodia extension, and chemotaxis. Moreover, arrestins were shown to translocate to pseudopodia following protease-activated receptor 2 activation. Protease-activated receptor 2⅐arrestin⅐ERK1/2 complexes were enriched in pseudopodia after receptor stimulation. So we were curious to see whether U46619 treatment led to pseudopodia extensions in TP␤-expressing cells and to characterize the localization of TP␤ and arrestins with respect to these extensions. HEK293 cells expressing TP␤-GFP, GFP-Arr2, or HA-TP␤ and GFP-Arr2 were stimulated with U46619 and visualization of GFP-tagged proteins carried out by live cell fluorescence microscopy. Images were collected every 30 s for 60 min. Fig. 5 shows that U46619 treatment of TP␤-expressing cells results in the formation of pseudopods (arrows in upper and middle panels). Stimulation of TP␤ triggered an enrichment of arrestin-2 in the pseudopod extensions (Fig. 5, right middle panel) when compared with a diffuse cytoplasmic pattern in the absence of receptor stimulation (Fig. 5, left middle panel) like reported for the protease-activated receptor 2 (35). Distribution of GFP-Arr2 was not altered by agonist treatment in the absence of TP␤ (Fig. 5, bottom panel). Activation of TP␤ also appeared to promote its redistribution into the newly formed pseudopods (Fig. 5, upper panel). Live cell imaging showed that there was extension and retention of pseudopods following the first 60 min of agonist stimulation of TP␤ after which all activity appeared to cease. GFP-tagged arrestin-3 behaved exactly as GFP-Arr2 (data not shown).
TP␤-mediated Activation of G␣ q Signaling Is Involved in the Formation of Actin Stress Fibers-One of the major pathways leading to the formation of actin stress fibers is through activation of RhoA via stimulation of G␣ 12/13 . Recently, Vogt et al. (16) proposed that RhoA activation induced by receptor agonists via G␣ q/11 also occurs although with lower potency than through G␣ 12/13 (16). Because we previously reported that agonist-induced endocytosis of TP␤ requires G␣ q signaling and that we showed in the present study that it also necessitates actin rearrangement, we were then interested in evaluating the role of G␣ q signaling in the reorganization of the actin cytoskeleton in our system. To do so, we used the RGS domain of GRK2, which was shown to specifically inhibit G␣ q/11 signaling (27). HEK293 cells transiently co-transfected with pcDNA3-HATP␤ and pcDNA3, or pcDNA3-HATP␤ and pcDNA3-RGS(GRK2), were stimulated with 100 nM U46619 for 120 min. Fig. 6A demonstrates that the RGS(GRK2) construct drastically inhibited agonist-induced endocytosis of TP␤, further confirming our previous results concerning the role of G␣ q signaling in this process (14). Fig. 6B shows that stimulation with U46619 of HEK293 cells stably expressing TP␤ transiently transfected with pcDNA3 results in the formation of actin stress fibers. On the other hand, transfection of the RGS-(GRK2) construct abrogated the formation of actin stress fibers (right panel). Thus, in HEK293 cells, TP␤ stimulates the formation of actin stress fibers through G␣ q signaling.

Overcoming of Actin Disruption by Arrestin-3, but Not Arrestin-2, in TP␤ Endocytosis-We previously reported that agonist-induced endocytosis of TP␤ is arrestin-dependent (4).
To investigate the role of arrestins in TP␤ endocytosis when the actin cytoskeleton is disrupted, we performed ELISA experiments on HEK293 cells co-transfected with pcDNA3-HATP␤ and pcDNA3, pcDNA3-Arr3, or pcDNA3-Arr2 in the presence or absence of Lat B. Interestingly, results from presence of Lat B (Fig. 7B, right panel). Likewise, measurements of actin polymerization by FACScan analysis of fluorescence associated with rhodamine-conjugated phalloidin failed to demonstrate a role of arrestin-3 overexpression in actin polymerization in the presence or absence of Lat B after TP␤ stimulation (data not shown). These results suggest that agonist-induced endocytosis of TP␤ can involve two distinct mechanisms: an arrestin-3-dependent/actin-independent and an arrestin-2-dependent/actin-dependent process.
The AP2-and Clathrin-binding Domains of Arrestin-3 Are Required in the Promotion of Actin-independent Endocytosis of TP␤-We next wanted to ascertain if the arrestin-3 effect in the presence of Lat B was mediated through AP-2 and clathrin. To this end, we used the LIEF/AAEA and the R396A mutants of arrestin-3, respectively, defective in clathrin-and AP-2-binding, as well as the AAEA/R396A double mutant (28). HEK293 cells were thus co-transfected with pcDNA3-HATP␤ and the constructs as indicated in Fig. 8, and then subjected to TP␤ endocytosis experiments in the presence of Lat B. Data presented in Fig. 8 show that the arrestin-3 mutants failed to promote TP␤ endocytosis in the presence of Lat B, indicating that this is a AP-2 and clathrin-dependent mechanism. All the arrestin-3 mutant constructs co-immunoprecipitated with TP␤ to the same extent as the wild-type arrestin-3 (data not shown), suggesting that our observations are not caused by an impaired interaction between the mutant arrestin-3 proteins and TP␤.

The YXXXI Motif in the C-tail of TP␤ Is Involved in the Promotion of Actin-independent TP␤ Endocytosis by Arrestin-3-
We then wanted to verify if a particular region or motif found in the C-tail of TP␤ could be associated with actin-independent TP␤ endocytosis. With this respect, one particular motif stood out of all the TP␤ mutants that we previously generated and described (5). We demonstrated before that the YX 3 motif in the C-tail of TP␤ was essential for constitutive endocytosis of the receptor. The two TP␤Y339A and TP␤I343A mutants were deficient in constitutive endocytosis but displayed agonist-induced properties similar to the wild-type receptor (5). The coupling of these two TP␤ mutants to G␣ q was shown to be the same as for wild-type TP␤ (33). Moreover, the binding of arrestin-3 with these two TP␤ mutants was equal to the wild-type receptor as evaluated by co-immunoprecipitation studies (data not shown). No function has yet been attributed to the Tyr 339 and Ile 343 residues in agonist-induced endocytosis of TP␤. As illustrated in Fig. 9, we observed that overexpression of arrestin-3 could not rescue agonist-induced endocytosis of the TP␤Y339A (Fig. 9A) and TP␤I343A (Fig. 9B) mutants in the presence of latrunculin B, in contrast to TP␤ (Fig. 7). Thus, the YX 3 motif found in the proximal portion of the TP␤ COOHterminal tail is essential for the arrestin-3-mediated promotion of actin-independent endocytosis following agonist stimulation. DISCUSSION It has been suggested that actin assembly plays a variable, but not obligatory role in receptor-mediated endocytosis in mammalian cells (36,37). A number of studies have looked at actin rearrangement induced by GPCRs, but the role of actin in GPCR endocytosis is poorly characterized. In the present study, we gathered evidence showing that a dynamic actin cytoskeleton was essential for agonist-induced endocytosis of TP␤. In contrast, treatment of cells with actin depolymerizing or stabilizing agents failed to affect agonist-induced endocytosis of the ␤ 2 AR. This is in agreement with Shumay et al. (38) who demonstrated that latrunculin A did not affect ␤ 2 AR endocytosis but that the actin cytoskeleton was rather necessary for recycling of the ␤ 2 AR to the cell surface. With respect to other GPCRs, actin depolymerization was shown to inhibit endocytosis of the CXCR4, A2a-adenosine, ␣1B-adrenergic, bombesin, and endothelin receptors (39 -42). Conversely, endocytosis of the CXCR1 and CXCR2 receptors was promoted by the same treatment (43). Involvement of microtubules in GPCR endocytosis also seems to vary with the receptor studied. Shumay et al. (38) found that nocodazole, another microtubuledepolymerizating agent, effectively blocks internalization of ␤ 2 AR in response to ␤-agonists. In our study, depolymerization of microtubules with colchicine did not modulate agonist-induced endocytosis of TP␤. Thus, the role of actin and microtubules in GPCR endocytosis cannot be generalized and appears to be receptor-specific.
Members of the Rho GTPase family (Rho, Rac, and Cdc42) are key regulatory molecules that link surface receptors to the organization of the actin cytoskeleton. The GTPase function of the Rho family is also required in endocytosis, suggesting that they could represent molecular bridges between cytoskeleton reorganization and vesicular trafficking (44). Rho activation leads to the assembly of contractile actin-myosin filaments (stress fibers) and associated focal adhesion complexes (45). Rac1 is thought to regulate the polymerization of actin to produce lamellipodia (46). On the other hand, activation of Cdc42 was shown to induce actin-rich surface protrusions called filopodia (46). Lamaze et al. (47) demonstrated that activated Rho and Rac inhibited endocytosis of the transferrin and epidermal growth factor receptors when expressed in HeLa cells. However, very little is known regarding the implication of these small G proteins in GPCR endocytosis. A dominant active mutant of RhoA was reported to decrease endocytosis of muscarinic receptors (48). Recently, we demonstrated that activation of Rac1 inhibited agonist-induced endocytosis of TP␤ (49). Here we show that RhoA and Cdc42 are also involved in TP␤ endocytosis. The effects of overexpressing RhoA and Cdc42 on the actin cytoskeleton are well characterized (50 -52). Wildtype and constitutive mutants of RhoA and Cdc42 promote actin rearrangement in the absence of any cellular stimulation. The dominant negative mutants will prevent actin rearrangement following activation of a receptor. Thus all of the RhoA and Cdc42 constructs will affect the dynamics of the actin cytoskeleton: they will push it toward one end of the equilibrium, i.e. polymerized actin or block its polymerization. This may explain why all of the RhoA and Cdc42 constructs significantly inhibited endocytosis of TP␤ in our study. Indeed, the data obtained with the RhoA and Cdc42 proteins support our idea of an essential receptor-mediated dynamic regulation of the actin cytoskeleton in agonist-induced endocytosis of TP␤, as was indicated by the results obtained with the agents interfering with the actin cytoskeleton.
Our results showed that agonist stimulation of TP␤ induced cytoskeleton rearrangement, like actin stress fibers and extension of pseudopods. Cyt D predominantly depolymerizes actin stress fibers. Cortical actin is more resistant to disruption by Cyt D in comparison to actively turning over actin, like the stress fibers (30). Data obtained with the Rho mutants and Cyt D suggests that actin stress fibers are important for TP␤ endocytosis to proceed. Interestingly, we observed that TP␤ endocytic vesicles were aligning with underlying actin stress fibers with both structures converging toward endosomal compartments. It is thus tempting to speculate that actin stress fibers are directing the trafficking of TP␤-containing endocytic vesicles to the target endosomal compartment.
TP␤ can activate G␣ q and G␣ 12/13 among many other G proteins. The major pathway leading to actin stress fiber formation by a GPCR is by activation of G␣ 12/13 . However, G␣ q/11 was also recently demonstrated to promote formation of stress fibers, albeit with less efficiency than G␣ 12/13 (17). We showed in this study that TP␤-mediated formation of actin stress fibers is a G␣ q/11 -dependent process. The inhibition of TP␤ endocytosis and actin stress fiber formation (in light of the discussion above) by RGS(GRK2) further supports our previous demonstration of an absolute requirement for G␣ q signaling in TP␤ endocytosis (14). Interestingly, Barnes et al. (53) recently showed that the concurrent recruitment of arrestin-2 and activation of G␣ q/11 leads to full activation of RhoA and to the subsequent formation of stress fibers following stimulation of the angiotensin II type 1A receptor. Experiments are underway in our laboratory to explore the role of arrestins in the G␣ qmediated stress fiber formation by TP␤.
Perhaps the most intriguing finding that we report in this article is the ability of arrestin-3, but not arrestin-2, to rescue agonist-induced endocytosis of TP␤ when the dynamics of the actin cytoskeleton are disrupted. The use of arrestin-3 mutants showed that this was an AP-2-and a clathrin-dependent mechanism. We do not currently have a clear answer as to why only arrestin-3 displays this particular property. We tested other known arrestin-3 mutants (54, 55) targeting domains distinguishing between arrestin-2 and arrestin-3 to no avail: all the arrestin-3 mutants were rescuing TP␤ endocytosis when the actin cytoskeleton was disrupted. In this FIG. 6. TP␤-mediated activation of G␣ q signaling is involved in the formation of actin stress fibers. A, G␣ q signaling is involved in agonist-induced internalization of TP␤. HEK293 cells were transiently co-transfected with pcDNA3-HATP␤ and either pcDNA3 or pcDNA3-RGS(GRK2), which interferes with G␣ q signaling. Agonist-induced internalization was assessed by ELISA after 2 h of stimulation with 100 nM U46619 at 37°C. Results represent the mean Ϯ S.E. of three independent experiments, each done in triplicate. B, HEK293 cells stably expressing FLAG TP␤ were transiently transfected with pcDNA3 (left panels) or pcDNA3-HA-RGS(GRK2) (right panels). Cells were stimulated with 1 M U46619 for 1 h at 37°C. The cells were then fixed, permeabilized, and incubated with a HA-specific monoclonal antibody. RGS(GRK2) and the actin filaments were visualized by incubating the cells with a fluorescein isothiocyanate-conjugated antimouse antibody and rhodamine-conjugated phalloidin, respectively. Cells were examined by fluorescence microscopy on a Nikon Eclipse TE2000-U microscope using a ϫ60 objective. Images were collected using the Simple PCI Camera software. Results shown are representative of three independent experiments. regard, endogenous arrestin-3, but not arrestin-2, was shown to be rapidly recruited to clathrin-coated pits upon stimulation of the ␤ 2 -adrenergic and m1 muscarinic receptors stably expressed in RBL-2H3 cells (56). Arrestin-2 and -3 share 78% of sequence homology but not exactly the same localization in the cell. Santini et al. (56) demonstrated that arrestin-2 was evenly distributed between the Triton-soluble and -insoluble pool, whereas arrestin-3 was predominantly found in the Triton-insoluble fraction, following or not receptor stimulation. Triton-soluble fractions of cells include cytosolic proteins in comparison to Triton-insoluble fractions, which contain proteins associated with the cytoskeleton or with glycosylphosphatidylinositol-rich membrane domains. Interestingly, Santini et al. (57) demonstrated that recruitment of arrestin-3 to coated pits was not detectably affected by disruption of the actin cytoskeleton with Lat B in RBL cells. Worthy of note, Lamaze et al. (31) noticed that the clustering of the transferrin receptor into coated pits was unaffected by latrunculin B. Thus we could speculate that arrestin-3 over-expression could rescue TP␤ endocytosis when the actin cytoskeleton is disrupted because it is already associated with undetermined membrane domains or cytoskeletal components at the cell membrane, and could thus still target the receptor to coated pits. However, both arrestin-2 and -3 can promote agonist-induced endocytosis of TP␤ when the actin cytoskeleton is intact. This suggests that there are arrestindependent/actin-dependent and arrestin-dependent/actinindependent mechanisms for TP␤ endocytosis. This is supported by the failure of all the actin toxins to completely abolish endocytosis of the receptor. In the same line of ideas, it is interesting that endocytosis of the ␤ 2 AR was unaffected by disruption or stabilization of the actin cytoskeleton when considering a finding by Kohout et al. (58). These authors reported that endocytosis of the ␤ 2 AR was profoundly inhibited (87%) in arrestin-3 knock-out cells, but not in arrestin-2 knockout cells. This could explain in part why endocytosis of the ␤ 2 AR is not altered by treatment with Lat B, Cyt D, or Jas because our data suggest that arrestin-3-mediated endocytosis can be an actin-independent mechanism. Further experiments will be needed to fully understand the relationship between arrestinmediated endocytosis and the actin cytoskeleton.
We previously demonstrated that the C-tail of TP␤ is responsible for endocytosis of the receptor (34). Amino acids 355-366 were identified as being necessary for promotion of TP␤ endocytosis by arrestins. We also reported that a Tyr 339 -X 3 -Ile 343 motif was specifically involved in constitutive endocytosis of TP␤ (5). Surprisingly, this YX 3 I motif appears to be necessary for arrestin-3 to rescue agonist-induced endocytosis of TP␤ in conditions where there is disruption of the actin cytoskeleton. We observed that this YX 3 I motif could be replaced by a YX 2 motif in TP␤ with conservation of the endocytic properties of the receptor (5). So, we thought that perhaps the YX 3 I motif could interact with the 2 subunit of the AP-2 complex known to interact with a YX 2 sequence. Our efforts to show an interaction between the C-tail of TP␤ and the 2 subunit of AP-2 failed to produce conclusive results. Further work will be required to resolve this issue. A, HEK293 cells were transiently co-transfected with pcDNA3-HATP␤ and pcDNA3, pcDNA3-Arr3, or pcDNA3-Arr2 constructs. Cells were pretreated with 5 g/ml latrunculin B or Me 2 SO (DMSO) (vehicle) for 1 h at 37°C. The cells were then stimulated or not with 100 nM U46619 for 2 h in the continuous presence of latrunculin B. Agonist-induced endocytosis was assessed by ELISA. The results represent the mean Ϯ S.E. of three independent experiments, each done in triplicate. B, HEK293 cells stably expressing TP␤ transiently overexpressing GFP-arrestin-3 (right panel) or not (left panel) were treated with Me 2 SO, Lat B, and agonist as above. The cells were processed for immunofluorescence microscopy as described under "Experimental Procedures." Results shown are representative of three independent experiments.
FIG. 8. The AP2-and clathrin-binding domains of arrestin-3 are required in the promotion of actin-independent endocytosis of TP␤. HEK293 cells were transiently co-transfected with pcDNA3-HATP␤ and pcDNA3, pcDNA3-Arr3, pcDNA3-Arr3LIEFAAEA, pcDNA3-Arr3R396A, or pcDNA3-Arr3AAEA/R396A constructs. Cells were pretreated with 5 g/ml latrunculin B for 1 h at 37°C. The cells were then stimulated or not with 100 nM U46619 for 2 h at 37°C in the continuous presence of latrunculin B. The percentage of cell surface receptor loss following agonist stimulation was measured by ELISA. The results represent the mean Ϯ S.E. of three independent experiments, each done in triplicate. Data are presented as the percentage of TP␤ endocytosis occurring in the absence of latrunculin B.
In summary, the results here demonstrated that agonistinduced endocytosis of TP␤ was dependent on a dynamic actin cytoskeleton that could be regulated by the small RhoA and Cdc42 GTPases, and that TP␤-containing endocytic vesicles were aligning on underlying actin stress fibers. TP␤-mediated formation of actin stress fibers was G␣ q -dependent. Our data also showed that promotion of endocytosis by arrestin-3, but not by arrestin-2, can be actin-independent, a process that is AP-2-and clathrin-dependent.