Regulation and Intracellular Trafficking Pathways of the Endothelin Receptors*

The effects of endothelin (ET) are mediated via the G protein-coupled receptors ETA and ETB. However, the mechanisms of ET receptor desensitization, internalization, and intracellular trafficking are poorly understood. The aim of the present study was to investigate the molecular mechanisms of ET receptor regulation and to characterize the intracellular pathways of ET-stimulated ETA and ETB receptors. By analysis of ETA and ETB receptor internalization in transfected Chinese hamster ovary cells in the presence of overexpressed βARK, β-arrestin-1, β-arrestin-2, or dynamin as well as dominant negative mutants of these regulators, we have demonstrated that both ET receptor subtypes follow an arrestin- and dynamin/clathrin-dependent mechanism of internalization. Fluorescence microscopy of Chinese hamster ovary and COS cells expressing green fluorescent protein (GFP)-tagged ET receptors revealed that the ETA and ETB subtypes were targeted to different intracellular routes after ET stimulation. While ETA-GFP followed a recycling pathway and colocalized with transferrin in the pericentriolar recycling compartment, ETB-GFP was targeted to lysosomes after ET-induced internalization. Both receptor subtypes colocalized with Rab5 in classical early endosomes, indicating that this compartment is a common early intermediate for the two ET receptors during intracellular transport. The distinct intracellular routes of ET-stimulated ETA and ETB receptors may explain the persistent signal response through the ETAreceptor and the transient response through the ETBreceptor. Furthermore, lysosomal targeting of the ETBreceptor could serve as a biochemical mechanism for clearance of plasma endothelin via this subtype.

The multiple physiological effects of the vasoactive peptide hormone endothelin (ET) 1 (1) are mediated via the G protein-coupled receptors (GPCRs) ET A and ET B (2). ET-1 acts directly on ET A receptors expressed on vascular smooth muscle cells to mediate a long lasting vasoconstrictive response. The prolonged response through agonist-stimulated ET A receptors has also been demonstrated in cultured cells expressing this receptor (3). Although the ET B receptor may also mediate vasoconstriction (4,5), this subtype, which is primarily situated at the plasma membrane of endothelial cells, is first of all considered to cause transient NO-mediated vasodilatation (6).
Circulating ET-1 is rapidly removed from the plasma, and current evidence suggests that the ET B receptor is important for clearance of plasma ET-1 both under normal conditions (7,8) and in pathological conditions with increased plasma levels of ET-1 (9). The mechanism of ET B receptor-mediated clearance of ET-1 is not known, but one hypothesis is that receptorbound ET-1 is targeted to degradation in the cell.
The general model for regulation of GPCR desensitization and internalization is primarily based on studies of the prototypic ␤ 2 -adrenergic receptor (10 -15). According to this model, agonist-activated receptors are phosphorylated by GPCR kinases (GRKs), followed by binding of arrestin and initiation of receptor internalization through a clathrin/dynamin-dependent mechanism (16 -18). Both the ET A and the ET B receptors have been shown to undergo rapid desensitization (19,20) and to internalize with similar kinetics (3,21) in transfected cell models. However, neither the molecular mechanisms of ET A and ET B receptor regulation nor their intracellular trafficking pathways are known. Addressing these issues may not only elucidate the mechanisms of ET receptor regulation but may also provide the molecular basis for the distinct physiological responses mediated by the two receptors.
We have used transfected Chinese hamster ovary (CHO) and COS cells as model systems for studying ET receptor internalization and intracellular trafficking. To facilitate the latter, we have constructed chimeric ET receptors with a C-terminal fusion of green fluorescent protein (GFP). The chimeric ET receptors were found to have similar functional properties as their wild type counterparts, making them appropriate tools for this study. We found that although both the ET A and the ET B receptors utilize an arrestin and dynamin/clathrindependent mechanism of internalization, they follow different intracellular pathways upon ET stimulation. Based on our data, we suggest that different intracellular trafficking routes may be an important mechanism underlying the distinct physiological responses mediated by the ET A and ET B receptors.

EXPERIMENTAL PROCEDURES
Plasmid Constructs-The human ET A and ET B receptors in pME18sf (22) were provided by Prof. T. Masaki (Kyoto University, Japan). An EcoRI-NotI fragment of pME18sf-ET A was subcloned into pcDNA1 for expression in CHO cells. For subcloning of the ET B receptor, PCR-directed mutagenesis was used to introduce a HindIII site in front of the start codon, followed by cloning of the fragments HindIII-EcoRV and EcoRV-XbaI into pcDNA1. The cDNAs of hemagglutinin (HA) epitopetagged dynamin WT and dynamin K44A mutant (23) in pRK5 were gifts from Dr. C. van Koppen (University of Essen, Germany). Rat ␤-arrestin-1 and ␤-arrestin-2 were in pCMV5 (24). Their respective mutants V53D and V54D in pcDNA1 were from Dr. M. G. Caron (Duke University, Durham, NC). Rat ␤ARK1 in pCMV5 and ␤ARK1-CT in pRK5 were gifts from Dr. R. J. Lefkowitz and Dr. W. Koch, respectively (Duke University). MycRab5 Q79L in pcDNA1 and the human transferrin receptor (TfR) in pGEM1-T7 were from Dr. H. Stenmark (the Norwegian Radium Hospital, Oslo). An EcoRI digest of pGEM1-hTfR was subcloned into pcDNA1 for expression in CHO cells.
Construction of ET A -GFP and ET B -GFP Chimeric Receptors-pEG-FPN1 (CLONTECH), an expression vector containing the coding sequence of a modified form of GFP from Aequorea victoria was chosen for cloning and expression of the ET receptor fusion proteins. For construction of ET A -GFP, PCR-directed mutagenesis was performed on the template pBS-ET A using the oligonucleotide primers 5Ј-ctaagcaatcatgtggatg-3Ј and 5Ј-gcggatcccggttcatgctgtccttatggctgc-3Ј to remove the stop codon of ET A and introduce a BamHI restriction site. The HindIII-KpnI and KpnI-BamHI fragments of ET A were subsequently inserted into pEGFPN1 to create a receptor fusion protein with GFP linked to the C-terminal cytoplasmic tail. Similarly, for construction of ET B -GFP, PCR was performed on the template pcDNA1-ET B with the primers 5Ј-gatcccaatagatgtgaac-3Јand 5Ј-cgcggatcccgagatgagctgtatttattactgg-3Ј to remove the stop codon and introduce a BamHI site. The HindIII-BamHI fragment of ET B was subsequently cloned into the pEGFPN1 expression vector.
Cell Lines and Transfections-CHO K1 cells (ATCC no. CCL-61) were maintained in Ham's F-12 Kaighn's modified medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum (Bio Whittaker) and 50 g/ml garamycin in a humidified atmosphere containing 5% CO 2 at 37°C. COS-7 cells (ATCC no. CRL-1651) were propagated in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and garamycin as above. Cells were grown to 40 -60% confluency in 35-mm wells before transient transfection using LipofectAMINE reagent (Life Technologies) according to the manufacturer's instructions. For immunofluorescence experiments, cells were plated onto glass coverslips in 24-well plates before LipofectAMINE transfection. Rab5 Q79L was expressed in CHO cells using the vaccinia T7 expression system as described by Stenmark et al. (25).
Kinetics of 125 I-ET-1 Internalization-Internalization of ET A and ET B receptors was determined using 125 I-ET-1 (Amersham Pharmacia Biotech). Twenty-four hours after transfection, the cells were placed on ice, washed, and preincubated in binding buffer (Hanks' balanced salt solution with 20 mM Hepes, pH 7.4, 0.2% BSA, and 0.1% glucose) before binding of 50 pM 125 I-ET-1 in binding buffer for 3 h at 4°C. The cells were washed extensively in ice-cold PBS to remove unbound radioligand before incubation in prewarmed binding buffer at 37°C for various time points. The cells were subsequently treated for 2 ϫ 10 min at 4°C with 0.5 M acetic acid, 0.15 M NaCl, pH 2.5 (ET A ), or 50 mM glycine-HCl, 0.5 M NaCl, pH 2.5 (ET B ) to strip off surface-bound radioligand. Acid-stripped cells were then lysed for 10 min in 1 M NaOH before the fraction of surface-bound ET-1 in the acid wash and the fraction of internalized ET-1 in the lysate were determined by ␥-spectrometry. Internalized receptor was calculated as the fraction of radioactivity not associated with the cell surface after acid wash. Nonspecific binding was determined using excess of unlabeled ET-1 (Peninsula Laboratories).
Western Blot Analysis-To verify the expression of the different GPCR regulators and their respective mutants, we performed Western blot analysis of samples from the transiently transfected CHO cells. Protein samples were prepared by harvesting cells in 10 mM Tris-HCl, pH 7.4, 1% SDS, 1 mM Na 3 VO 4 . 25 g of protein was loaded per well for SDS-polyacrylamide gel electrophoresis followed by electroblotting onto polyvinylidene difluoride membranes. For ␤ARK1 and ␤ARK1-CT, we used TBS (50 mM Tris-HCl, pH 8.0, 100 mM NaCl) with 0.1% Tween 20, 2.5% dry milk powder, and 2.5% BSA for blocking and antibody incubations. Primary antibody was affinity-purified polyclonal anti-GRK2 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) used at 1:300. As secondary antibody, we used horseradish peroxidase-conjugated donkey anti-rabbit IgG (Amersham Pharmcia Biotech) diluted at 1:3000. For ␤-arrestins and dynamin, we used PBS with 0.05% Tween 20, 2.5% dry milk powder, and 2.5% BSA for blocking and antibody incubations. To detect ␤-arrestins, we used polyclonal antisera raised against the respective proteins (24) and horseradish peroxidase-conjugated donkey anti-rabbit secondary antibody. Hemagglutinin-tagged dynamin was detected with a monoclonal anti-HA (12CA5) antibody (Babco) diluted 1:1000 and a horseradish peroxidase-conjugated sheep anti-mouse Ig secondary antibody (1:3000). The blots were developed using enhanced chemiluminescence (ECL) as recommended by the manufacturer (Amersham Pharmacia Biotech).
Phosphorylation of GFP Fusion Proteins-Transiently transfected CHO cells (35-mm wells) expressing ET A -GFP or ET B -GFP with or without coexpression of ␤ARK1 or ␤ARK2 were labeled with 500 Ci/ well of [ 32 P i ] (NEN Life Science Products) in phosphate-free Dulbecco's modified Eagle's medium for 1 h at 37°C. To promote receptor activation/phosphorylation, the cells were subsequently incubated with ET-1 (1 M) for 10 min. After ET-1 stimulation, the cells were rapidly chilled on ice, washed twice in PBS, and lysed for 30 min in ice-cold radioimmune precipitation buffer containing protease inhibitors (150 mM NaCl, 50 mM Tris-HCl, pH 8.0, 5 mM EDTA, 1% Nonidet P40, 0.5% deoxycholate, 0.1% SDS, 10 mM NaF, 10 mM Na 2 pyrophosphate, and protease inhibitors (0.1 mM phenylmethylsulfonyl fluoride, 10 g/ml benzamidine, 10 g/ml leupeptin, 5 g/ml aprotinin, and 1 g/ml pepstatin A)). Cell lysates were cleared by centrifugation for 10 min at 10,000 ϫ g (4°C), and immunoprecipitation of labeled receptors was performed using a rabbit polyclonal peptide antibody against GFP (CLONTECH) and protein A-agarose (Upstate Biotechnology Inc.). Beads with antibody-antigen complexes were washed three times in radioimmune precipitation buffer and subsequently denatured at 65°C in 2ϫ SDS loading buffer (Laemmli) containing 5% urea and 5% ␤-mercaptoethanol. Radioligand Binding-Radioligand binding was performed on membrane preparations according to Elshourbagy et al. (26) with the modifications outlined below. CHO cells transiently transfected with receptor were washed in ice-cold PBS and harvested in 25 mM Tris-HCl, 2 mM EDTA, pH 8.0, containing protease inhibitors (30 g/ml benzamidine, 100 M phenylmethylsulfonyl fluoride). The cells were homogenized on ice using a Dounce glass homogenizer, followed by centrifugation at 50,000 ϫ g for 15 min. The membrane pellets were resuspended in 25 mM Tris-HCl, pH 7.5, 2 mM EDTA, 100 mM NaCl, 1 mg/ml BSA, and protease inhibitors as above, and saturation binding analysis was performed for 1 h at 37°C using 2-1000 pM 125 I-ET-1 in the absence (total binding) or presence (nonspecific binding) of a 100-fold excess of unlabeled ET-1. For the binding experiments, we used 1 g of membrane protein in a total volume of 75 l per sample. The samples were filtered through Whatman GF/C filters presoaked with 0.05% BSA to remove unbound ligand. After three washes in 0.5 M NaCl, the filters were subjected to ␥-spectrometric analysis.

Analysis of Internalized ET A -GFP and ET B -GFP by Fluoresence
Microscopy-For analysis of ET receptor trafficking, transfected cells grown on glass coverslips were placed on ice, washed, and preincubated for 10 min with Hanks' balanced salt solution-Hepes before binding of 0.4 M ET-1 in binding buffer (Hanks' balanced salt solution-Hepes with 0.2% BSA and 0.1% glucose) for 1 h at 4°C. Unbound ligand was removed by washing in Hanks' balanced salt solution-Hepes, and the cells were subsequently incubated at 37°C for various periods of time. The cells were fixed in 4% paraformaldehyde before mounting onto object glasses using Mowiol (Hoechst). The specimens were analyzed using either a Zeiss Axiovert-100 fluorescence microscope or a Leica TCS laser-scanning microscope with a 100 ϫ 1.3 oil immersion objective. Images were processed and overlaid using Photoshop 4.0 and the Metamorph Imaging System.
Uptake of Fluorescent Probes and Staining of Intracellular Antigens for Fluorescence Microscopy-For the transferrin (Tf) uptake experiments, cells were cotransfected with ET receptor and the human TfR. Cells were pretreated with 10 g/ml cycloheximide for 30 min before ET-1 binding was performed as described above. During ET-1 internalization at 37°C, tetramethylrhodamine-conjugated Tf (Molecular Probes, Inc., Eugene, OR) was added to the cells in binding buffer at a concentration of 25 g/ml for labeling of the pericentriolar recycling compartment. The cells were washed in ice-cold PBS before fixation in 4% paraformaldehyde and mounting onto object glasses for fluorescence microscopy analysis.
Labeling of lysosomes in CHO cells was done overnight with 0.2 mg/ml Texas Red-conjugated dextran (Molecular Probes) followed by a 2-h chase period at 37°C to wash the probe out of early and late endosomes. To analyze receptor trafficking in this experiment, ET-1 was internalized for various periods of time during the 2-h chase period. The cells were fixed and mounted as described above and subjected to fluorescence microscopy.
For staining of intracellular antigens, the cells were permeabilized in PBS containing 0.1% Triton X-100 after paraformaldehyde fixation. Free aldehyde groups were quenched by adding a few drops of 1 M glycine, pH 8.5, to the PBS wash prior to the permeabilization step. 10% fetal bovine serum in PBS was added for 5 min at room temperature for blocking before antibody incubation. Primary and secondary antibodies were diluted in PBS to a concentration recommended by the manufacturer or as experienced to give optimal results. Between antibody incubations, the cells were extensively washed in PBS. Immunofluorescence staining of endogenous ␤COP in CHO cells was done using a rabbit antiserum and a Texas Red-conjugated goat anti-rabbit secondary antibody (Southern Biotechnology). Late endosomes and lysosomes in COS cells were visualized using a rabbit antiserum against human LAMP-2 and an Alexa594-conjugated goat anti-rabbit secondary antibody (Molecular Probes). Staining of MycRab5 Q79L expressed using the vaccinia T7 system was performed using a mouse anti-Myc antibody followed by a rhodamine-labeled goat anti-mouse secondary antibody (both from Dr. H. Stenmark, the Norwegian Radium Hospital, Oslo).

Mechanisms of ET A and ET B Receptor Internalization in CHO Cells-To investigate the effects of different regulators of GPCR on ET receptor internalization, CHO cells were tran-
siently transfected with ET A or ET B receptors alone or together with wild type or mutant forms of ␤ARK1 (GRK2), ␤-arrestin-1, ␤-arrestin-2, and dynamin. Receptor internalization was determined by 125 I-ET-1 uptake as described under "Experimental Procedures." Internalized receptor at each time point was calculated as the fraction of radioactivity left after removing surface-bound radioligand with low pH wash. ET A and ET B receptors were both rapidly internalized upon agonist stimulation, reaching 25% within 10 min (Fig. 1), which is consistent with previous reports on ET receptor internalization (3,21).
␤ARK1-CT, a C-terminal fragment of ␤ARK1 containing a G␤␥ binding domain (pleckstrin homology domain) (27), was cotransfected with the ET receptors to competitively inhibit the actions of endogenous ␤ARK. We found that ␤ARK1-CT significantly decreased the rate of ET A receptor internalization (Fig.  1A). No effect was observed on internalization of the ET B subtype in this experimental condition (Fig. 1B), although a small inhibition was discernible at higher levels of ␤ARK1-CT overexpression (not shown). Coexpression of ␤ARK1 wild type had no effect on the internalization kinetics of any of the receptor subtypes (Fig. 1, A and B). However, as shown in the Western blots of Fig. 1, nontransfected CHO cells also express significant amounts of ␤ARK1.
␤-Arrestin-1 and ␤-arrestin-2 are nonvisual arrestins shown to interdict signaling through GPCRs phosphorylated by GRK (24). Arrestin subsequently targets the receptor to clathrinmediated endocytosis by binding to the clathrin heavy chain and the ␤-adaptin subunit of the AP-2 adaptor (15,28). As shown in Fig. 1, C and D, internalization of both ET receptor subtypes increased when overexpressing the wild type ␤-arrestin-1 and ␤-arrestin-2. The effect appeared to be more pronounced on internalization of the ET A receptor than the ET B receptor. This observation was substantiated by ET receptor internalization experiments in CHO cells where transfection of increasing amounts of ␤-arrestin-1 or ␤-arrestin-2 cDNA was associated with smaller increments of internalization of ET B than of ET A (data not shown). Nevertheless, these results show that both receptors are recognized by arrestins and can be recruited to an arrestin-mediated pathway of internalization. ␤-Arrestin-1 V53D and ␤-arrestin-2 V54D are dominant negative mutants of ␤-arrestin-1 and ␤-arrestin-2, respectively, which competitively inhibit endogenous arrestin binding to clathrin (14). When the V53D and V54D mutants were cotransfected with the ET receptors in CHO cells, the internalization rate was significantly reduced for both receptor subtypes (Fig.  1, E and F), demonstrating that endogenous arrestins of the CHO cells are involved in regulation of the transfected ET receptors.
Cotransfection of the GTPase-deficient mutant dynamin K44A (23,29) substantially inhibited internalization of the ET A and ET B receptors, suggesting that both receptors are subjected to endocytosis through clathrin-coated pits (Fig. 1, G  and H). As shown, overexpression of wild type dynamin did not affect ET receptor internalization. Impaired endocytosis of both ET receptor subtypes was also observed after incubation in hyperosmotic sucrose and upon potassium depletion (not shown), i.e. methods demonstrated to inhibit clathrin-mediated endocytosis (30,31). Taken together, our results implicate a role for arrestins and a dynamin/clathrin-dependent pathway in regulation of ET receptor endocytosis.
Functional Characteristics of the ET A -GFP and ET B -GFP Fusion Proteins-Redistribution of ET receptors from the plasma membrane to intracellular locations has previously been observed (21,32), but neither the receptor-containing compartments nor the intracellular trafficking pathways of the receptors have been characterized. To delineate the intracellular trafficking pathways of the ET A and ET B receptors, we employed ET receptor-GFP fusion proteins for visualization of receptors during endocytosis. Chimeric ET A and ET B receptors with GFP fused to the cytoplasmic tail were constructed as described under "Experimental Procedures." Radioligand binding studies of membrane preparations from CHO cells transiently transfected with ET A -GFP or ET B -GFP demonstrated pharmacological properties very similar to the wild type receptors (Table I). Furthermore, internalization of 125 I-ET-1 in CHO cells expressing ET A -GFP or ET B -GFP showed that both subtypes were rapidly internalized with kinetics similar to the wild type ET A and ET B receptors (Fig. 2, A and B). The capacity of ET A -GFP and ET B -GFP to mediate ET-1-induced phosphatidyl inositol hydrolysis and inositol phosphate accumulation was also examined, and as shown in Fig. 2, C and D, the chimeric receptors displayed similar efficacies as compared with their wild type counterparts.
To investigate the ability of ET A -GFP and ET B -GFP to undergo agonist-induced phosphorylation, we performed whole cell phosphorylation assays. Transfected CHO cells metabolically labeled with inorganic phosphate, 32 P i , were incubated for 10 min in the absence or presence of 1 M ET-1. Immunoprecipitation of the receptors using an antibody directed against GFP was followed by SDS-polyacrylamide gel electrophoresis and detection of phosphorylated receptors by laser scan Phos-phorImager analysis as described under "Experimental Procedures." As shown in Fig. 3, phosphorylation of both ET A -GFP (migrating at M r 80,000 -110,000) and ET B -GFP (migrating at M r 70,000 -85,000) was induced by ET-1 stimulation and could be detected both in the absence and presence of coexpressed ␤ARK1 or ␤ARK2. Two distinct bands underwent significant agonist-dependent phosphorylation both in ET A -GFP-and ET B -GFP-transfected cells in accordance with previous findings for wild type ET receptors (20). Altogether, our results show that attachment of GFP to the cytoplasmic tail of the ET receptors does not alter the functional characteristics of the receptors.

Characterization of the Intracellular Trafficking Pathways of ET A -GFP and ET B -GFP-ET
A -GFP and ET B -GFP were transiently transfected into CHO cells to investigate the intracellular routes of the receptors after agonist stimulation. In the absence of agonist, both receptors were localized at the plasma membrane (Fig. 4, A and C). To study the intracellular ET receptor trafficking pathways, CHO cells were incubated with ET-1 (0.4 M) for 1 h at 4°C. Unbound ligand was removed by extensive washing, and internalization of the receptors was subsequently studied at 37°C for different periods of time. Since endocytosis does not occur at 4°C, receptor internalization is synchronized once the medium is changed to 37°C. Within a few minutes, both ET A -GFP and ET B -GFP could be detected in intracellular compartments. As shown by immunofluorescence microscopy, strikingly different patterns were observed for the two receptor subtypes. While the ET B receptor was localized to punctate endosome-like structures throughout the cytoplasm 15 min after initiation of internalization, the ET A receptor was found to accumulate in a distinct spotlike structure near the nucleus after 15 min (Fig. 4, B and D). The perinuclear localization of ET A -GFP is typical for the pericentriolar recycling compartment, a tightly clustered accumulation of tubules and vesicles located near the centrioles and the Golgi apparatus (33), which serves as a compartment for controlling recycling of membrane proteins (34). To investigate the nature of the compartment containing ET A -GFP, CHO cells transfected with ET A -GFP were stained for immunofluorescence using an antiserum against ␤COP, a subunit of the Golgi-stack coatomer complex (35), and a Texas Red-conjugated secondary antibody after 15 min of ET-mediated receptor internalization. Cycloheximide was included during the incubations to prevent accumulation of newly synthesized ET A -GFP in the Golgi apparatus. As shown in Fig. 5, the red fluorescent ␤COP staining (Fig. 5B) was distinct from the intracellular green fluorescence representing ET A -GFP (Fig.  5A), since no yellow color due to colocalization could be seen in the overlaid pictures (Fig. 5C). Repeated analysis always demonstrated that the perinuclear ET A -GFP-containing compartment was close to, but not overlapping with, the Golgi stacks.
At steady state, transferrin has been shown to accumulate in the recycling compartment, since the rate-limiting step in recycling is transport from this location (36). To investigate to what extent the perinuclear compartment containing ET A -GFP overlapped with the recycling compartment, tetramethylrhodamine-labeled transferrin (Rh-Tf) was internalized into CHO cells cotransfected with ET A -GFP and TfR. After 15 min of ET-mediated internalization in the presence of Rh-Tf, we found extensive colocalization of ET A -GFP and Rh-Tf in the pericen-

FIG. 5. Localization of ET A -GFP but not ET B -GFP to the pericentriolar recycling compartment in CHO cells.
Upper panels, A-C, after binding at 4°C, ET-1 was internalized at 37°C for 15 min in CHO cells expressing ET A -GFP. After paraformaldehyde fixation and permeabilization in 0.1% Triton X-100, Golgi stacks were visualized by staining with an anti-␤COP antiserum followed by a Texas Red-conjugated goat anti-rabbit secondary antibody. The overlay (C) shows no colocalization between ET A -GFP and ␤COP. Middle and lower panels, D-I, ET A -GFP (D-F) or ET B -GFP (G-I) was expressed in CHO cells together with the TfR. At the start of the 15-min ET-1 internalization period, Rh-Tf was added to the cells for labeling of the pericentriolar recycling compartment. The overlays show that ET A -GFP (F) but not ET B -GFP (I) colocalized with Rh-Tf in the recycling compartment. triolar recycling compartment (Fig. 5, middle panels). ET A -GFP was observed intracellularly for up to 30 min after the initiation of internalization and subsequently reappeared at the plasma membrane (see below). These data indicate that the recycling pathway is the dominant transport route of agoniststimulated ET A receptors in CHO cells.
When CHO cells cotransfected with ET B -GFP and TfR were investigated, we could not detect ET B -GFP in the recycling compartment. Fig. 5 (lower panels) shows ET B -GFP after 15 min of ET-mediated internalization in the presence of Rh-Tf. The spread vesicular structures containing ET B -GFP do not overlap with the central spot representing the Tf-containing recycling compartment, indicating that the ET B receptor follows another transport pathway in the cell. To investigate whether ET B receptors were transported to lysosomes for degradation, we studied ET B receptor internalization in cells where the lysosomal compartments had been labeled with red fluorescent dextran as described under "Experimental Procedures." After 15 min of internalization, there was no significant overlap between the ET B -containing vesicles and the dextranlabeled lysosomes (Fig. 6, upper panels), but after 90 min (middle panels) the vesicular pattern overlapped extensively (see FIG. 6

. Lysosomal transport of ET B -GFP and recycling of ET A -GFP. CHO cells expressing ET B -GFP (upper and middle panels) or ET A -GFP (lower panels)
were incubated overnight with Texas Red-conjugated dextran (0.2 mg/ml) to label the lysosomal compartments. A 2-h chase without dextran was subsequently performed in order to wash the probe out of early and late endosomes. During the same time span, ET-1 was internalized for different time points as described under "Experimental Procedures" before paraformaldehyde fixation and fluorescence microscopy. ET B -GFP colocalized with dextran-Texas Red-labeled lysosomes after 90 min (middle panels; F is overlay) but not after 15 min (upper panels; C is overlay) of ET-1 internalization. ET A -GFP did not colocalize with lysosomes but reappeared at the plasma membrane after 90 min of ET-1 internalization (lower panels). Incubations were performed in the presence of 10 M cycloheximide to block de novo synthesis of ET receptors.

FIG. 7. Intracellular trafficking of ET A -GFP and ET B -GFP in COS cells.
Upper panels, A-C, after binding at 4°C, ET-1 was internalized at 37°C for 15 min in COS cells expressing ET A -GFP as described under "Experimental Procedures." At the start of the 15-min internalization period, Rh-Tf was added to the cells for labeling of the recycling compartment (COS cells express high levels of endogenous transferrin receptor). The overlay (C) shows colocalization of ET A -GFP and Rh-Tf in the recycling compartment. Middle and lower panels, D-I, ET-1 was bound at 4°C and internalized at 37°C for 10 min (middle panels, D-F) or 60 min (lower panels, G-I) in COS cells expressing ET B -GFP. After paraformaldehyde fixation and permeabilization in 0.1% Triton X-100, late endosomes and lysosomes were identified by immunostaining with a rabbit anti-human LAMP-2 antiserum and an Alexa594-conjugated goat anti-rabbit secondary antibody. As shown, ET B -GFP did not overlap with LAMP-2-positive compartments after 10 min of internalization (middle panels; F is overlay). However, distinct overlapping vesicles were identified after 60 min of internalization (lower panels; I is overlay). Fig. 6C, overlay). These findings suggest a degradative transport pathway for the ET B -receptor, which is also consistent with our observation that ET B -GFP did not reappear at the plasma membrane in the presence of cycloheximide. Conversely, ET A -GFP did not colocalize with lysosomes after prolonged ET-mediated internalization, but it was found to recycle back to the plasma membrane independently of cycloheximide (Fig. 6, lower panels). The intracellular trafficking pathways of ET A -GFP and ET B -GFP were also analyzed in COS cells, where the same characteristic differences in intracellular routing were observed. As shown in Fig. 7, ET A -GFP colocalized with transferrin in recycling endosomes (upper panels), whereas ET B -GFP was found to enter LAMP-2 positive late endosomes and lysosomes upon prolonged periods of internalization (middle and lower panels).
Colocalization of ET A -GFP and ET B -GFP with Rab5 in Early Endosomes-Different members of the Rab family of small GTPases have been shown to regulate distinct transport steps in eucaryotic cells (37). Rab5 is localized to early endosomes and is a rate-limiting component in membrane docking and fusion in the early endocytic pathway (38). Rab5 Q79L is a GTPase-deficient mutant of Rab5 shown to increase homotypic fusion events between early endosomes and retard transport out of these compartments (39). When ET A -GFP and ET B -GFP were expressed together with Rab5 Q79L in CHO cells, both receptors could be detected in early endosomes (Fig. 8, A-F).
This experiment demonstrates that the intracellular transport routes of both ET A -GFP and ET B -GFP go through classical early endosomes before they separate into the recycling pathway and the degradative pathway, respectively.

DISCUSSION
The purpose of this study was to investigate the mechanisms of internalization and intracellular trafficking of the ET A and ET B receptors. We report for the first time the involvement of arrestin and dynamin in the regulation of the ET receptors, and we describe the previously unknown intracellular routes of agonist-stimulated ET A and ET B receptors.
Both ET A and ET B receptors have been shown to undergo agonist-dependent phosphorylation catalyzed by GRK. In transfected HEK293 cells with overexpression of either ␤ARK1 (GRK2), ␤ARK2 (GRK3), or GRK5, all of the kinases have been shown to phosphorylate agonist-activated ET receptors (20). However, the specificity of the different GRK isoforms as to the substrates ET A and ET B is not known. In the present study, increased levels of ␤ARK1 did not augment internalization of the ET receptors. However, ␤ARK1-CT substantially inhibited internalization of the ET A receptor and to a lesser degree that of ET B , demonstrating that at least ET A receptor internalization is a ␤ARK1-mediated process. The less pronounced effect of ␤ARK1-CT on internalization of ET B may be due to phosphorylation of the ET B receptor by other endogenously ex- FIG. 9. Model for regulation and intracellular trafficking of the ET receptors. Based on the data of the present study, a model for regulation and intracellular trafficking of the ET A and ET B receptors is proposed. As indicated in the diagram, both ET A and ET B are rapidly internalized upon agonist stimulation, a process that depends on GRK, arrestin, dynamin, and clathrin. Internalized ET A and ET B are subsequently directed to Rab5 positive early endosomes (sorting endosomes). However, from this location, the two receptor subtypes are targeted to different intracellular fates. Whereas the ET A receptor is directed to the pericentriolar recycling compartment and subsequently reappears at the plasma membrane, the ET B receptor is directed to lysosomes for degradation.

FIG. 8. Colocalization of ET A -GFP and ET B -GFP with Rab5 in early endosomes. ET A -GFP (upper panels)
or ET B -GFP (lower panels) was expressed together with MycRab5 Q79L in CHO cells using T7 vaccinia transfection. Five hours after transfection, ET-1 was bound to the cells at 4°C and subsequently internalized for 30 min at 37°C. The cells were fixed in paraformaldehyde and permeabilized with 0.1% Triton X-100. Overexpressed MycRab5 Q79L was identified by immunostaining with a mouse monoclonal anti-Myc antibody and a rhodamine-conjugated goat anti-mouse secondary antibody. Overlays (right) show extensive colocalization of both ET A -GFP (C) and ET B -GFP (F) with Rab5 in early endosomes.
pressed receptor kinases in the CHO cells not subject to competitive inhibition by ␤ARK1-CT. The lack of effect of ␤ARK1 wild type overexpression on internalization of the ET A receptor indicates that ␤ARK1-mediated phosphorylation is not ratelimiting for endocytosis in CHO cells, a cell line that displays substantial levels of endogenously expressed ␤ARK1. Although both receptor subtypes have cytoplasmic tails containing several serine and threonine residues (40,41), only ET B has motifs similar to those previously demonstrated to be phosphorylated by GRK (42).
The ability of the ␤-arrestin-1 V53D and ␤-arrestin-2 V54D mutants to inhibit both ET A and ET B receptor internalization provides strong evidence that arrestin isoforms regulate both receptor subtypes. These mutant arrestins compete with endogenous arrestins in the cell by binding to clathrin, but they display reduced affinity for the receptor (14). Our data therefore demonstrate that the endogenous arrestins of the CHO cells contribute to ET receptor endocytosis. However, our data do not indicate the specific isoforms of arrestin involved in ET receptor regulation. Arrestin appears to be a rate-limiting factor for internalization, since overexpression of ␤-arrestin-1 or ␤-arrestin-2 resulted in a significant increase in the rates of internalization of both the ET A and the ET B receptors. Consistent with these data, it has earlier been demonstrated that arrestin binding is rate-limiting in the process of desensitization of the GPCR rhodopsin in Drosophila (43).
As shown, the GTPase-deficient mutant dynamin K44A significantly impaired ET receptor internalization. The involvement of dynamin is considered to be a strong indicator for a clathrin-dependent mechanism of internalization, although evidence also indicates that dynamin may regulate other mechanisms of endocytosis (44). However, our observations that potassium depletion of CHO cells (31) or pretreatment in hyperosmotic sucrose (30) both impaired internalization of ET A and ET B receptors provide further evidence for the involvement of clathrin-coated pits in endocytosis of these receptors. Previously, Chun et al. (45) have proposed a role for caveolae in ET receptor signaling and regulation, based on colocalization of ET A receptors and caveolin in COS cells as assessed by immunofluorescence. Although other GPCRs and related proteins involved in signal transduction have also been shown to be localized to caveolae (46,47), the role of these subcellular structures in GPCR internalization is still poorly understood.
Recently, it was shown that although internalized ␤ 2 adrenergic receptor colocalizes with the transferrin receptor in early endosomes, the receptors were located to different subpopulations of clathrin-coated vesicles at earlier stages of endocytosis (48). We have not investigated whether the ET A and ET B receptors follow the same initial steps of endocytosis as the transferrin receptor, but since both subtypes depend on classical regulators of GPCRs, it is likely that they follow the same pathway as the ␤ 2 adrenergic receptor into early endosomes. The classical early endosome, also called the sorting endosome, is a tubulovesicular structure where membrane proteins destined for degradation are sorted away from recycling proteins and packed into multivesicular bodies for further transport. Membrane proteins destined for recycling are, on the other hand, rapidly transported from the sorting endosome to the pericentriolar recycling compartment and are subsequently directed back to the plasma membrane (49,50). We found that the transport routes of both the ET A and the ET B receptors go through early endosomes, which is probably the final common compartment of the two receptor subtypes.
The information for receptor targeting either to recycling or to lysosomal degradation may reside in the cytoplasmic tail of the receptors (51). Furthermore, data from a recent report suggest that the presence of a cluster of serines in the tail of a GPCR appears to mediate stable binding of arrestin and dictate internalized receptors to lysosomes (52). According to the hypothesis presented in the latter report, configurations of the cytoplasmic tail causing less stable binding to arrestin lead to recycling and resensitization of the receptor. Interestingly, the ET B receptor has a C-terminal cluster of serines, which is not present in the tail of the ET A receptor. Lysosomal targeting of the ET B receptor could therefore be due to more stable interactions with arrestin, while less stable binding to the tail of the ET A receptor may lead this subtype to the recycling pathway. Consistent with this hypothesis, it has been demonstrated that the ET A receptor is not able to translocate ␤-arrestins to intracellular compartments (53). The hypothesis that arrestin may target GPCRs to lysosomes is also supported by ultrastructural data showing subcellular location of ␤-arrestin in multivesicular bodies (24).
ET-1 has been shown to dissociate very slowly from the ET receptors, even at the low pH values corresponding to those of endosomes (pH 5.5-6.0) and lysosomes (pH 4.4 -5.5) (21,54,55). Indeed, ET-1 has been reported to remain bound to the ET A receptor after internalization (21). Thus, it has been hypothesized that ligand-occupied ET A receptors may continue to activate the G protein after endocytosis. In this respect, internalization of GPCRs has been reported to be a prerequisite for transactivation of the epidermal growth factor receptor (56). However, rapid receptor endocytosis, resensitization, and recycling back to the plasma membrane in order to initiate repeated cycles of signaling may also provide the basis for a prolonged signal response. In fact, several reports have shown that receptor internalization is obligatory for resensitization of GPCRs (57)(58)(59). Thus, the evidence provided in this report that ET A receptors are targeted to recycling may explain the long lasting signal response observed through this receptor subtype.
ET-1 binds almost irreversibly to the ET B receptor at physiological conditions (55). Thus, in context of our findings, direct transport of receptor-associated ET-1 to lysosomes for degradation could serve as the mechanism for clearance of plasma ET-1 via the ET B receptor subtype. The high affinity between ET B receptor and ET-1 ensures efficient capture of ET-1 at the low (picomolar) levels in plasma but also implicates that the receptor is simultaneously sacrificed in the clearance process. This hypothesis correlates, however, well with the transient nature of ET B receptor-mediated signaling. Moreover, we have shown in the present study that the reappearance of ET B receptors at the plasma membrane is sensitive to cycloheximide. This implicates that the supply of de novo synthesized receptor molecules to the cell surface would be a limiting factor for the ability of ET B receptors to mediate efficient clearance of plasma ET-1.
In conclusion, the present study demonstrates that the ET receptor subtypes ET A and ET B are rapidly internalized in an agonist-dependent manner, a process that appears to depend on GRK, arrestin, dynamin, and clathrin. Furthermore, we have demonstrated that internalized ET A and ET B receptors both enter Rab5 positive early endosomes (sorting endosomes). However, the two receptor subtypes are subsequently targeted to different intracellular fates. While the ET A receptor follows a recycling pathway through the pericentriolar recycling compartment and reappears at the plasma membrane, the ET B receptor is directed to lysosomes for degradation. Based on the data presented in this study, we propose a model for the intracellular trafficking pathways of the ET receptors, which is illustrated in the schematic diagram of Fig. 9. The distinct intracellular routes of ET A and ET B may explain the characteristic physiological responses mediated by these receptors.
Thus, the rapid recycling of ET A may provide a basis for the prolonged contractile response mediated through this receptor, whereas lysosomal targeting of ET B supports current evidence suggesting a role for this receptor in clearance of ET from the circulation. The aim of future studies will be to delineate the mechanisms of the sorting process in early endosomes where the intracellular routes of the two receptor subtypes separate.