Properties of Secretin Receptor Internalization Differ from Those of the β2-Adrenergic Receptor*

The endocytic pathway of the secretin receptor, a class II GPCR, is unknown. Some class I G protein-coupled receptors (GPCRs), such as the β2-adrenergic receptor (β2-AR), internalize in clathrin-coated vesicles and this process is mediated by G protein-coupled receptor kinases (GRKs), β-arrestin, and dynamin. However, other class I GPCRs, for example, the angiotensin II type 1A receptor (AT1AR), exhibit different internalization properties than the β2-AR. The secretin receptor, a class II GPCR, is a GRK substrate, suggesting that like the β2-AR, it may internalize via a β-arrestin and dynamin directed process. In this paper we characterize the internalization of a wild-type and carboxyl-terminal (COOH-terminal) truncated secretin receptor using flow cytometry and fluorescence imaging, and compare the properties of secretin receptor internalization to that of the β2-AR. In HEK 293 cells, sequestration of both the wild-type and COOH-terminal truncated secretin receptors was unaffected by GRK phosphorylation, whereas inhibition of cAMP-dependent protein kinase mediated phosphorylation markedly decreased sequestration. Addition of secretin to cells resulted in a rapid translocation of β-arrestin to plasma membrane localized receptors; however, secretin receptor internalization was not reduced by expression of dominant negative β-arrestin. Thus, like the AT1AR, secretin receptor internalization is not inhibited by reagents that interfere with clathrin-coated vesicle-mediated internalization and in accordance with these results, we show that secretin and AT1A receptors colocalize in endocytic vesicles. This study demonstrates that the ability of secretin receptor to undergo GRK phosphorylation and β-arrestin binding is not sufficient to facilitate or mediate its internalization. These results suggest that other receptors may undergo endocytosis by mechanisms used by the secretin and AT1Areceptors and that kinases other than GRKs may play a greater role in GPCR endocytosis than previously appreciated.

The importance of phosphorylation as a regulatory mechanism for desensitizing receptor signaling has been well estab-lished for class I G protein-coupled receptors (GPCRs), 1 such as the ␤ 2 -adrenergic receptor (␤ 2 -AR) (1,2), as well as for class II GPCRs such as the secretin receptor (3). The role of phosphorylation in restoring GPCR signaling ability is less clear. GPCR phosphorylation is predominantly directed by two different classes of kinases, the G protein-coupled receptor kinases (GRKs) and the second messenger-dependent protein kinases. GPCR phosphorylation by cAMP-dependent protein kinase (PKA) appears to directly diminish the ability of the receptor to signal. In contrast, GRK phosphorylation increases receptor affinity for arrestin proteins, and it is this interaction of the receptor with arrestin that uncouples the receptor from the G protein (4). Additionally, arrestin directs some class I GPCRs, like the ␤ 2 -AR, to a clathrin-mediated endocytic pathway in which receptors are internalized, dephosphorylated, and subsequently recycled to the plasma membrane as competent receptors (5,6). However, whether arrestins are required for endocytosis or resensitization of class II GPCRs has not been examined.
␤-Arrestin 1 and 2 are members of the arrestin protein family that regulate both the signaling and trafficking of GPCRs. ␤-Arrestin translocates from the cytosol to agonistactivated GPCRs where it subsequently targets the GPCR⅐␤arrestin complex to clathrin-coated pits (CCPs), presumably by its ability to bind the clathrin heavy chain and the ␤-adaptin subunit of the AP2 clathrin adaptor protein complex (7,8). The roles of ␤-arrestin and CCPs in GPCR endocytosis have been established by fluorescence colocalization studies and through functional cellular assays that employ mutants of ␤-arrestin and dynamin (9 -11). These studies demonstrate that endocytosis through CCPs is dependent upon the ability of dynamin, a GTPase, to catalyze the pinching off of the CCP (12). This process is blocked by a dominant negative mutant, dynamin K44A, that is defective in GTP binding. Similarly, ␤-arrestin mutants that have lost the ability to support GPCR trafficking to CCPs also function as dominant negative inhibitors of sequestration. In a series of experiments, Zhang et al. (9) demonstrated that expression of either dynamin K44A or ␤-arrestin V53D prevented endocytosis of the ␤ 2 -AR but not the angiotensin II type 1A receptor (AT 1A R), despite the ability of the AT 1A R (also a class I GPCR) to interact with ␤-arrestin. Thus, class I GPCRs that desensitize by ␤-arrestin binding can internalize with distinct properties.
Secretin receptors internalize following agonist stimulation (13,14), but the endocytic pathway utilized remains uncharac-terized. The ability of secretin receptors to robustly desensitize in the presence of GRKs suggests that they may internalize through a mechanism similar to that of the ␤ 2 -AR. In this study we investigate the role of receptor kinases and ␤-arrestins in the regulation of secretin receptor endocytosis. The internalization of wild-type and COOH-terminal truncated secretin receptors was measured in the presence of exogenously overexpressed GRKs, dominant negative mutants of dynamin and ␤-arrestin, or in the presence of PKA inhibitors. Our results demonstrate that, unlike the ␤ 2 -AR, secretin receptor internalization is not inhibited by ␤-arrestin-V53D or dynamin-K44A. Moreover, they suggest a dependence of secretin receptor internalization on PKA activity. These results imply that ␤-arrestin-mediated desensitization is separable from its GPCR trafficking function and that kinases other than GRKs may regulate GPCR endocytosis.

EXPERIMENTAL PROCEDURES
Materials-General chemicals and reagents were from Sigma. Secretin was obtained from Peninsula Labs. Human embryonic kidney cells (HEK 293 cells) were obtained from the American Tissue Culture Collection. Tissue culture supplies were obtained from Life Technologies. Labeled secretin ( 125 iodine) was prepared and purified by high performance liquid chromatography (3,15). [2,  Plasmid Preparation-The FLAG epitope-tagged rat secretin receptor, the HA epitope-tagged ␤ 2 -AR, and ␤-arrestin-green fluorescent protein (␤-arrestin-GFP) have been described previously (3,16,17). The COOH-terminal truncated secretin receptor was prepared from the NH 2 -terminal, FLAG-tagged secretin receptor by polymerase chain reaction to include amino acids 1-398. The cDNAs were inserted into the pcDNA 1/Amp plasmid (Invitrogen) using HindIII and BamHI. Fidelity was demonstrated by dideoxy sequencing. GRK cDNAs were produced as described previously: GRK 2 (18) and GRK 5 (19). Plasmid purification was performed with Qiagen reagents.
Secretin receptor-green fluorescent protein (secretinR/GFP) was prepared by inserting the rat secretin receptor into pEGFP-N3 vector (CLONTECH) via BamHI and XhoI restriction sites by the method previously described (17). Fidelity was demonstrated by dideoxy sequencing and function was confirmed by cAMP accumulation in response to agonist (secretin) by whole cell cyclase (see methods below). AT 1A R/GFP was prepared in a manner similar to that described by Barak et al. (17).
Cell Culture-HEK 293 cells were grown in modified Eagle's medium (modified Eagle's medium: 10% fetal bovine serum, 50 mg/liter gentamicin) at 37°C in 95% air, 5% CO 2 . One day after transfection cells were split into appropriate plates following trypsin dissociation. Experiments were performed 24 -48 h after transfection.
Transfection-Transient transfections were performed with calcium phosphate co-precipitation. One to 10 g of vector DNA was transferred into a 6-ml Falcon tube with 450 l of sterile water and 50 l of 2.5 M CaCl 2 . Then 500 l of 2 ϫ HEPES-buffered saline (0.28 M NaCl, 0.05 M HEPES, 1.5 mM Na 3 PO 4 , pH 7.1) was added to the tube and mixed well. This mixture was added dropwise to the 100-mm dish of cells. Cells were plated to a density of approximately 2-3 ϫ 10 5 cells to each well for cAMP accumulation experiments and 1-1.5 ϫ 10 6 cells per well for phosphorylation and sequestration experiments.
Membrane Preparation/Binding-All steps were performed at 4°C. Plates were placed on ice, media was aspirated, and the cells were washed with 10 ml of ice-cold PBS. Five to 10 ml of lysis buffer (10 mM Tris, 5 mM EDTA with protease inhibitors: 10 g/ml aprotinin, 5 g/ml leupeptin, 0.7 g/ml pepstatin A, 10 g/ml benzamidine, 0.2 mM phenylmethylsulfonyl fluoride) were added to each plate. With a cell lifter, cells were scraped off the plate and placed in 15-ml conical tubes on ice. Cell fragments were homogenized with a Polytron PT 3000 for 20 -30 s at 14,000 -16,000 cps. Material was centrifuged at 300 -400 ϫ g for 10 min to remove unlysed cells and nuclei. Supernatant was transferred to 13 ϫ 100-mm tubes on ice and centrifuged at 18,000 rpm (40,000 ϫ g) (Sorval SM24 rotor) for 30 min at 4°C. Supernatant was discarded and the membrane pellet was resuspended in binding buffer, for immediate assay, or lysis buffer and stored at Ϫ80°C.
Membrane binding was performed as published (3). Briefly, using a constant amount of HEK 293 cell membrane protein, competition displacement (porcine secretin, Peninsula Labs) of 125 I-porcine secretin binding was performed in triplicate tubes. Nonspecific binding was defined in the presence of 1 M unlabeled porcine secretin. Data was analyzed using Graph Pad-Prism and LIGAND software as described (3,15).
Adenylyl Cyclase Assays-The accumulation of cAMP in intact cells was quantitated chromatographically by the method of Salomon (20). Cells were labeled with [ 3 H]adenine (1 Ci/ml) in modified Eagle's medium, 5% fetal bovine serum, 50 mg/liter gentamicin (1 ml/well) 12 to 16 h prior to experimentation. To assay the accumulation of cAMP, labeling media was aspirated, cells were washed with 1 ml of PBS, and preincubated in 1 ml of media per well (modified Eagle's medium, 0% fetal bovine serum, 10 mM HEPES, 1 mM IBMX; assay medium) for 15-30 min. Cells were stimulated with appropriate agonist, and at the end of the experimental duration, media was aspirated and 1 ml of ice-cold stop solution (0.1 mM cAMP, 4 nCi/ml [ 14 C]cAMP, 2.5% perchloric acid) was placed in each well. Plates remained on ice for 20 -30 min, after which solution was transferred to 12 ϫ 75 tubes containing 100 l of 4.2 M KOH. Tubes were vortexed and stored overnight, at 4°C, prior to cAMP determination by column chromatography (20). Data is normalized for total cellular uptake using [ 14 C]cAMP as described previously (20).
Receptor Expression-In plasmid co-transfection experiments, receptor expression was determined by flow cytometry analysis of a sample from each transfection group (3). The fluorescence was determined by incubation for 1 h at 4°C with monoclonal IgG-M2-FLAG (1:600 dilution, Kodak), three washes with PBS, and detection with Fc-specific, fluorescein-labeled goat anti-mouse (1:200 dilution, Sigma). Cells were then washed, removed from the plate with 10 mM Tris, pH 7.4, 5 mM EDTA and fixed with 3.6% formaldehyde. Samples were analyzed on a Becton-Dickson flow cytometer. Baseline fluorescence was determined from a sample of HEK 293 cells untransfected and/or a sample of HEK 293 cells transfected with the secretin receptor not exposed to primary antibody (IgG-M2-FLAG). Baseline fluorescence was subtracted from each sample.
Receptor Internalization/Sequestration-Sequestration is defined as the number of receptors removed from the cell surface after agonist exposure, as determined by flow cytometry. Cells are incubated with agonist for the appropriate time. After washing with iced PBS, cells on ice are exposed for 1 h at 4°C to monoclonal IgG-M2-FLAG (1:600 dilution, Kodak) or 12CA5 (1:500 dilution, Roche Molecular Biochemicals), three washes with PBS, and detection with Fc-specific, fluorescein-labeled goat anti-mouse (1:200 dilution, Sigma). Cells were then washed, removed from the plate with 10 mM Tris, pH 7.4, 5 mM EDTA and fixed with 3.6% formaldehyde. Samples were analyzed on a Becton-Dickson flow cytometer. Baseline fluorescence was determined from a sample of HEK 293 cells transfected with the secretin receptor not exposed to agonist and another sample not exposed to primary antibody (IgG-M2-FLAG). Baseline fluorescence was subtracted from each sample.
Secretin Receptor Internalization by Immunofluorescence-HEK 293 cells transiently transfected with 2.5 g of cDNA for secretin wild-type receptor or 3.5 g of cDNA for COOH-terminal truncated secretin receptor were plated onto 35-mm dishes containing a central glass well as described (16). Cells were maintained at 4°C to prevent receptor internalization while incubating with agonist (100 nM porcine secretin), primary antibody (IgG-M2-FLAG, Kodak, Inc), and secondary antibody (Fabs conjugated with fluorescein isothiocyanate, Organo Teknica). Incubations were 30 min in duration and occurred in the sequence listed. Cells were washed 3 times with cold PBS after each antibody application. Immediately following the last PBS wash, cells were viewed using confocal microscopy (basal time point), while a second plate of identically treated cells was warmed at 37°C for 1 h prior to imaging (60 min treatment).
Immunofluorescent Colocalization of Secretin Receptor with Angiotensin Receptors, ␤-Arrestin-GFP, or ␤ 2 -ARs-HEK 293 cells transiently transfected with 3.0 g of cDNA for secretin wild-type receptor and 4.8 g of cDNA for angiotensin II type 1A-GFP receptor (AT 1A R/GFP) were plated onto glass coverslips contained in 6-well plates. After an initial wash, cells were stimulated with 100 nM secretin and 100 nM angioten-sin II in a 37°C incubator. After 30 and 60 min, cells were fixed in 4% paraformaldehyde for 25 min. Cells were successively incubated for 1 h at room temperature, with primary antibody (IgG-M2-FLAG) and secondary antibody (Texas Red conjugated goat anti-mouse, Molecular Probes Inc.) dissolved in solubilization buffer consisting of 0.2% Triton X-100 and 1% bovine serum albumin in phosphate-buffered saline. Cells were washed for 20 min in solubilization buffer after each antibody incubation. Immediately following the last wash, glass coverslips containing the cells were mounted onto glass slides for viewing using a Zeiss laser scanning confocal microscope. ␤-Arrestin-GFP (1.7 g of cDNA) was coexpressed with secretin wild-type receptor (2.4 g of cDNA) and an identical protocol was followed except that cells were stimulated with secretin alone for only 2 min. SecretinR/GFP (250 ng of cDNA) and HA epitope-tagged ␤ 2 -ARs (4 g of cDNA) were coexpressed in HEK 293 cells, stimulated with 100 nM secretin and 10 M isoproterenol and treated according to the protocol above with the exception that IgG-12CA5-HA was the primary antibody used.
␤-Arrestin-GFP Translocation-Transfected HEK-293 cells were plated onto 35-mm dishes containing a central glass well as described (16). Cells in Dulbecco's modified medium buffered with 20 mM HEPES were stimulated with 1 M agonist and viewed using a Zeiss laser scanning confocal microscope.

Characterization of Secretin Receptor Constructs-In order
to demonstrate receptor movement from the plasma membrane to intracellular compartments, epitope-tagged wild-type and COOH-terminal truncated secretin receptor cDNAs were prepared. Quantifying cell surface epitopes has been used with other receptors and represents a non-radioactive method of monitoring receptor sequestration (9,16,18). The truncated receptor contains amino acids 1-398, and is devoid of 10 serines and threonines in the COOH-terminal tail, which eliminates over 50% of the total intracellular sites for potential serine/ threonine kinase-mediated phosphorylation. Binding studies were performed on cell membranes prepared from HEK 293 cells overexpressing the NH 2 -terminal FLAG epitope-tagged wild-type and COOH-terminal truncated secretin receptors. When compared with the membranes prepared from cells transfected with native (non-epitope tagged) secretin receptor, competition binding resulted in similar K D values. K D values ranged from 2.5 nM for the FLAG-truncated secretin receptor to 5.3 nM for the native receptor (Fig. 1A).
In order to investigate the signaling of these receptor constructs, dose-response and time course of cAMP production were studied. Signaling experiments did not demonstrate any significant difference between the wild-type and truncated secretin receptors (Fig. 1B). The EC 50 for cAMP accumulation was approximately 0.1 nM for both the wild-type and the truncated receptors. Our previous data for the native receptor (nonepitope tagged) revealed an EC 50 of 0.4 nM. Truncation of the COOH terminus did not produce any difference in the kinetics of cAMP accumulation compared with the wild-type (Fig. 1C). Having demonstrated that the addition of an NH 2 -terminal FLAG or the truncation of the COOH-terminal tail did not appreciably alter the binding or the signaling of the secretin receptor, we employed these FLAG epitope-tagged receptor tools to investigate the mechanism of secretin receptor internalization.
Secretin Receptor Sequestration-Fluorescence and corresponding interference and fluorescence overlay micrographs (Fig. 2, A and B) demonstrate that the FLAG epitope-tag on the secretin receptor does not hinder endocytosis of the wild-type The half-time for complete desensitization of wild-type or truncated receptor signaling was 5.7 or 7.8 min, respectively (range 4.9 to 6.8 and 6.9 to 9.0 min, respectively, data analyzed by Graph-Pad Prism, one phase exponential association, R 2 ϭ 0.94 and 0.97, respectively). Each point in the wild-type or COOH-terminal truncated curve is the mean of four or three, respectively, independent experiments, each done in triplicate for each time point. The curve was generated using nonlinear regression analysis with Graph-Pad Prism.
receptor, or the COOH-terminal truncated receptor, into vesicles. As measured by flow cytometry, the wild-type secretin receptor sequestered to a maximum of 58 Ϯ 4% after 20 to 30 min. The half-time of sequestration was 8 min (Fig. 3A). In order to determine the role of the COOH-terminal portion of the secretin receptor on sequestration, we studied the sequestration of the truncated secretin receptor during a 1-h period. Truncation did not appear to alter the kinetics of receptor sequestration, however, total surface receptor internalization at all time points beyond 30 min (Fig. 3A) was decreased 10 -15% compared with wild-type (p Ͻ 0.02, nine experiments, each done in duplicate). The COOH-terminal region of the secretin receptor contains multiple potential phosphorylation sites. Others have shown an association between receptor phosphorylation and receptor internalization (18). We next investi-gated the effect of truncation (loss of COOH-terminal phosphorylation sites) on secretin receptor phosphorylation and internalization.
Effect of GRK-mediated Phosphorylation on Secretin Receptor Sequestration-Secretin receptor phosphorylation is increased by specific GRKs (3). In order to investigate the effect of increased receptor phosphorylation on receptor sequestration, we overexpressed GRK 2 and GRK 5 in HEK 293 cells. As shown in Fig. 3, B (for wild-type receptor) and C (for C-terminal truncated receptor), kinases known to augment phosphorylation of the secretin receptor did not increase receptor internalization for either the wild-type or truncated receptor. Although overexpression of GRKs enhances secretin receptor desensitization (3), it does not alter receptor sequestration. At all time points studied (up to 60 min), no significant effect of overexpressed GRKs on sequestration was observed. Expression of GRKs was documented by Western blotting using GRK specific antisera (not shown) as described under "Experimental Procedures." Thus, enhanced GRK-mediated phosphorylation of the secretin receptor, unlike the ␤ 2 -AR (18), does not appear to augment receptor sequestration.
Effect of the PKA Inhibitors, H89 and Staurosporin, on Secretin Receptor Sequestration-G protein-coupled receptor desensitization can be mediated by PKA and/or PKC. We have shown that PKA inhibitors reduce phosphorylation of the secretin receptor (3). However, these inhibitors had no effect on secretin receptor signaling in HEK 293 cells (3). Therefore, we tested the effect of PKA inhibition on secretin receptor sequestration. The PKA inhibitors, H89 and staurosporin, significantly decreased sequestration of both the wild-type and truncated secretin receptors (Fig. 4). Interestingly, sequestration of the COOH-terminal truncated secretin receptor was more profoundly affected by H89 than the wild-type receptor. Staurosporin (1 M) blocks the activity of both PKA and PKC (24). However, compared with the inhibition of sequestration by H89, additional PKC inhibition did not promote further inhibition of sequestration. These data suggest that there may be an association between cAMP-dependent protein kinase phosphorylation and sequestration of the secretin receptor.
Effect of ␤-Arrestin and GRK-dependent Phosphorylation on Sequestration of the COOH-terminal Truncated Receptor in the Presence of PKA Inhibition-Given that inhibition of PKA activity inhibits secretin receptor internalization, we attempted to reverse this by overexpressing ␤-arrestin, GRK 2, or ␤-arrestin and GRK 2 combined. Overexpression of either ␤-arrestin or GRK 2 increased receptor internalization (Fig. 5) in the presence of H89 but the effects were not additive. This result demonstrates that ␤-arrestin and GRK 2 can enhance secretin receptor sequestration; however, their involvement only becomes apparent when PKA dependent phosphorylation is prevented.
Effect of Hypertonic Sucrose on Secretin Receptor Sequestration-Hypertonic medium reduces the clustering of surface receptors into endosomes (25). Hypertonic sucrose similarly inhibited sequestration of the ␤ 2 -AR, the secretin wild-type, and the COOH-terminal truncated receptors (Fig. 4) suggesting that, like other GPCRs, the secretin receptor does internalize by clustering surface receptors into endosomes.
The Role of ␤-Arrestin and Dynamin in Secretin Receptor Sequestration-␤-Arrestin translocates to agonist activated ␤ 2 -AR (26). Using a recently developed ␤-arrestin translocation assay (16), we found that agonist activation of wild-type or truncated secretin receptors initiated rapid movement of ␤-arrestin-GFP from the cytosol to the plasma membrane and that this translocation did not require overexpression of GRKs (Fig.  6). Addition of agonist to cells not overexpressing the secretin receptor caused no translocation of ␤-arrestin-GFP (micrograph not shown). H89-mediated inhibition of PKA activity did not alter the translocation of ␤-arrestin-GFP to the agonistactivated wild-type or COOH-terminal truncated secretin receptor (micrograph not shown). Fluorescence micrographs show co-localization of ␤-arrestin-GFP and secretin wild-type receptors at the plasma membrane (Fig. 7). Thus, ␤-arrestin interacts with the agonist-activated secretin receptor. How-

FIG. 3. Influence of GRKs on the agonist-promoted sequestration kinetics of secretin wild-type and COOH-terminal truncated receptors as assessed by flow cytometry.
A, secretin wildtype or truncated receptors were transiently transfected in HEK 293 cells. Cells were exposed to 0.1 M porcine secretin for the times indicated on the x axis. The half-time for complete sequestration of wildtype and truncated receptors was 7.6 min (range 5.6 to 12.2) and 8.2 min (range 5.6 to 15.3), respectively. The data was analyzed by Graph-Pad Prism, one phase exponential association, R 2 ϭ 0.74 for wild-type and R 2 ϭ 0.64 for truncated receptors. The data represent the mean Ϯ S.E. of four independent experiments done in duplicate. Cells expressing the wild-type (B) or truncated (C) secretin receptor, in conjunction with either GRK2 or GRK5, were exposed to 0.1 M porcine secretin for the times indicated on the x axis. The half-time for complete sequestration of wild-type receptors with GRK2 or GRK5 were 6.4 min (range 4.7 to 10.2) and 8.1 min (range 6.1 to 12.2), respectively. The half-time for complete sequestration of truncated receptors with GRK2 or GRK5 were 7.6 min (range 5.6 to 12.1) and 11.8 min (range 8.8 to 18.1), respectively. The data was analyzed by Graph-Pad Prism, one phase exponential association, R 2 ϭ 0.74 for wild-type receptors with GRK2, R 2 ϭ 0.79 for wild-type receptors with GRK5, R 2 ϭ 0.75 for truncated receptors with GRK2, and R 2 ϭ 0.83 for truncated receptors with GRK5. The data represent the mean Ϯ S.E. of four independent experiments done in duplicate. ever, whether or not ␤-arrestin plays a role in regulating sequestration of the secretin receptor is unknown.
␤-Arrestin helps in directing agonist-activated ␤ 2 -ARs to clathrin-coated pits for internalization (7,9,10). To determine if ␤-arrestin also participates in trafficking the secretin receptor into endosomes, we measured sequestration in cells overexpressing ␤-arrestin or the functionally impaired ␤-arrestin mutant, V53D. Overexpressed ␤-arrestin or V53D did not alter internalization of wild-type or COOH-terminal truncated secretin receptor, whereas sequestration of the ␤ 2 -AR was inhibited in the presence of the sequestration dominant negative ␤-arrestin mutant (Fig. 8A). Western blot with specific antiserum documented V53D overexpression in HEK 293 cells coexpressing the secretin receptor and ␤ 2 -AR (data not shown). Although ␤-arrestin translocates to the agonist-activated secretin receptor, the presence of ␤-arrestin V53D does not interfere with the trafficking of this receptor to endocytic vesicles.
Prior evidence has shown dynamin to be necessary for clathrin-coated vesicle formation (12), the primary endocytic pathway for agonist-activated ␤ 2 -ARs (9). To determine if the secretin receptor internalizes through a dynamin-dependent mechanism, we measured sequestration of secretin receptors and ␤ 2 -ARs in the presence of overexpressed dynamin or its dominant-negative protein, K44A. In corroboration of previous results (9) we showed that overexpression of dynamin-K44A inhibits sequestration of the ␤ 2 -AR (Fig. 8B). However, the functionally impaired dynamin-K44A failed to inhibit internal-ization of either the wild-type or COOH-terminal truncated secretin receptors (Fig. 8B). Furthermore, in HEK 293 cells overexpressing dynamin, no effect on sequestration of the secretin receptor was evident (Fig. 8B). Western blotting with specific antisera confirmed dynamin and dynamin-K44A overexpression in HEK 293 cells co-expressing the secretin receptor and ␤ 2 -AR (data not shown). Thus, the secretin receptor, like the AT 1A R, internalizes via a dynamin-independent mechanism. We hypothesized that AT 1A R and secretin receptors might utilize the same cellular mechanism for internalization and should, therefore, co-localize in the same endocytic vesicles.
Co-localization of Secretin Receptor and AT 1A R in Endocytic Vesicles-HEK 293 cells co-transfected with secretin wild-type and AT 1A R/GFP were stimulated for 30 and 60 min prior to fixation. Secretin receptors were immunofluorescently stained with Texas Red for confocal microscope imaging. Fluorescence micrographs show that secretin wild-type and AT 1A R/GFP are co-localized in endocytic vesicles at 30 min (Fig. 9A) and at 60 min (Fig. 9B). These micrographs show that secretin and AT 1A receptors indeed reside in the same endocytic vesicles.
Co-localization of Secretin Receptor and ␤ 2 -AR in Endocytic Vesicles-Unlike ␤ 2 -ARs, secretin receptor internalization is not affected by the ␤-arrestin dominant negative mutant, V53D, or the dynamin dominant negative protein, K44A. To determine if the secretin receptor internalizes through an endocytic mechanism distinct from that of the ␤ 2 -AR, we cotransfected secretinR/GFP and ␤ 2 -AR-HA in HEK 293 cells. Cells were stimulated for 30 and 60 min, fixed, and ␤ 2 -ARs were immunofluorescently stained for confocal microscope imaging (Fig. 10, A and B). Fluorescence micrographs show that the majority of vesicles observed contain either secretin receptor or ␤ 2 -AR, but not both. Thus, secretin receptors and ␤ 2 -ARs do not principally employ the same endocytic pathway.

DISCUSSION
In this paper we provide direct evidence demonstrating internalization of agonist-activated secretin receptors. Although agonist-activated secretin and ␤ 2 -ARs are each phosphorylated by GRKs and subsequently recruit ␤-arrestin, secretin receptors internalize with different properties, implying that the aforementioned characteristics are not sufficient to predict the sequestration pathway of a GPCR. Furthermore, the majority of secretin receptors internalize into endosomes which do not contain ␤ 2 -ARs, and vice versa. In addition, we demonstrate co-localization of secretin and AT 1A Rs in endocytic vesicles, suggesting that the pathway or mechanism used by these two FIG. 6. ␤-Arrestin-GFP translocation upon agonist stimulation. HEK 293 cells, overexpressing the wild-type secretin receptor, ␤-arrestin-GFP, and GRK 5 were exposed to agonist (0.1 M secretin) and followed with confocal microscopy. As early as 60 s after agonist stimulation, ␤-arrestin-GFP is shown translocating from the cytosol to the plasma membrane. The times after agonist exposure are indicated in the top left corner of each panel (s, seconds). HEK 293 cells expressing the secretin wild-type receptor without overexpressed GRKs demonstrated similar ␤-arrestin-GFP translocation (not shown). Comparable ␤-arrestin-GFP translocation was also observed in cells overexpressing the secretin COOH-terminal truncated receptor (not shown), whereas no translocation occurred in cells not overexpressing the secretin receptor (not shown).
FIG. 7. ␤-Arrestin-green fluorescent protein co-localization with secretin receptor. HEK 293 cells overexpressing the secretin wild-type receptor and ␤-arrestin-GFP protein were exposed to agonist (0.1 M secretin) and fixed. Confocal microscopy was used to produce fluorescence micrographs. Panels A and B show secretin receptors and ␤-arrestin-GFP, respectively, clustered at the cell membrane 2 min after addition of agonist. Panel C is an overlay of panels A and B. The overlay shows many red and green clusters overlapping to form yellow clusters; evidence that ␤-arrestin translocates from the cytosol to secretin receptors. Prior to addition of agonist, ␤-arrestin-GFP was distributed throughout the cytosol (similar to Fig. 6, time 0) and not localized at the membrane (not shown). receptors may also be common to other GPCRs. Finally, we present the unique finding that PKA activity regulates sequestration of a GPCR.
The role of ␤-arrestin in GPCR regulation has been extensively studied, primarily using the prototypical ␤ 2 -AR (6). Agonist activation and GRK phosphorylation of the ␤ 2 -AR trigger the translocation of ␤-arrestin from the cytosol to the plasma membrane where it binds and uncouples the receptor from its heterotrimeric G protein (1,2,16). The receptor is thus desensitized and must be internalized, processed through endocytic vesicles, and dephosphorylated before it is returned to the plasma membrane as a functional receptor (5,27). ␤-Arrestin plays a role in regulating this internalization/resensitization process by directing the ␤ 2 -AR to CCPs (7,10). ␤-Arrestin interacts directly with clathrin; however, the importance of this reaction for receptor internalization is not clear, as ␤-arrestin constructs lacking the inherent clathrin binding motif can still support ␤ 2 -AR internalization (7,8). The ␤ 2 -AR⅐␤-arrestin complex also recruits and binds the ␤ 2 -adaptin subunit of AP-2 (8). AP-2 is a protein complex which serves as an adaptor between receptors and the clathrin lattice during clathrin-coated pit formation (28). Through its interaction with AP-2 and clathrin, ␤-arrestin may serve as a docking protein which links the ␤ 2 -AR to the clathrin lattice.
The clathrin lattice of coated pits contains a uniform distri- FIG. 8. Effect of overexpression of wild-type and mutant ␤-arrestin 1 and dynamin on the agonist-promoted sequestration of secretin and ␤ 2 -ARs as assessed by flow cytometry using receptor-specific antiserum. A, wild-type or truncated secretin receptors, or ␤ 2 -ARs, were transiently transfected in HEK 293 cells together with 5 g of each of the following: empty pCMV5 vector (control), pCMV rat ␤-arrestin-1, or pcDNA1-Amp rat ␤-arrestin-1-V53D (10). Expression of mutant and wild-type ␤-arrestin-1 was monitored by immunoblot using an antibody for ␤-arrestin-1 (23). The data represent the mean Ϯ S.E. of five, five, and four independent experiments done in duplicate for secretin wild-type, secretin truncated, and ␤ 2 -ARs, respectively. B, wild-type or truncated secretin, or ␤ 2 -ARs were transiently transfected in HEK 293 cells together with 8 g of empty pCMV5 vector (control), 8 g of pCB1 rat dynamin I, or 8 g of rat dynamin I-K44A as indicated. Expression of mutant and wild-type dynamin was monitored by immunoblot using an antibody for dynamin I (22). The data represent the mean Ϯ S.E. of four, five, and three independent experiments done in duplicate for secretin wild-type, secretin truncated, and ␤ 2 -ARs, respectively. bution of non-activated dynamin (12). Conformational changes in dynamin, resulting from GTP-binding, promote dynamindynamin interactions that ultimately lead to constriction of coated pits and their subsequent release from the plasma membrane (29). Dynamin K44A has reduced affinity for GTP, rendering its ability to promote clathrin-coated vesicle fission ineffective (12).
Overexpression of the dominant negative proteins dynamin K44A or ␤-arrestin V53D, which inhibit sequestration of ␤ 2 -ARs (9), did not reduce sequestration of the secretin receptor. Given these results, one might conclude that in contrast to the ␤ 2 -AR, secretin receptor internalization is not directed by ␤-arrestin, nor is it internalized in CCPs. The possibility remains, however, that secretin receptors may ultimately traffic in CCPs through a dynamin-independent mechanism or through utilization of another isoform of dynamin. The ability of ␤-arrestin to avidly bind agonist-activated secretin receptors, paired with the recent observation that ␤-arrestin can interact with multiple sites on various GPCRs to affect the characteristics of internalization (26), precludes inferring that ␤-arrestin does not play a role in directing secretin receptors to endocytic vesicles. The possibility remains that ␤-arrestin, when complexed with secretin receptor, may have a different conformation than when complexed with the ␤ 2 -AR. Therefore a ␤-arrestin⅐secretin complex may regulate sequestration through distinct interactions with known adaptor proteins or with as yet unknown proteins. However, our only evidence suggesting that ␤-arrestin enhances sequestration of the secretin receptor comes from study of a mutant in which phosphorylation of the receptor is severely compromised through carboxyl-terminal tail truncation, and after inhibition of PKAmediated phosphorylation (Fig. 5). The contribution of ␤-arrestin to secretin receptor sequestration under normal cellular conditions would likely be minor, given that ␤-arrestin involvement in receptor trafficking is usually associated with GRK-mediated phosphorylation, whereas this study shows that PKA-mediated phosphorylation is more important for secretin receptor sequestration, at least under the conditions examined.
PKA, a cAMP-dependent serine/threonine kinase, phosphorylates and uncouples GPCRs from G proteins. To date, research regarding PKA regulation of GPCRs has focused on GPCR signaling, not sequestration (30). Perhaps the paucity of studies investigating an association between second messengerdependent phosphorylation and sequestration of GPCRs results from the prominent role of GRKs in regulating the internalization of prototypical GPCRs, such as the ␤ 2 -AR (1,18). A broader examination of the role of PKA in vesicular trafficking has shown that PKA regulates internalization of some non-GPCRs and is involved in cellular vesicular transport (31). Goretzki and Mueller (31) recently showed that H-89-mediated inhibition of PKA activity reduced the internalization of receptors for low density lipoprotein receptor-related protein and transferrin (non-GPCRs) in a receptor-specific manner. Thus, there is evidence to indicate that PKA activity regulates receptor internalization. In addition, our prior study, in which H-89 reduced the phosphorylation, but not the signaling, of the secretin receptor, provided the impetus for exploring the possibility that PKA-mediated phosphorylation may be associated with other functions of the secretin receptor, such as sequestration.
In this paper, we demonstrate that the PKA inhibitors H-89 and staurosporin markedly reduced sequestration of the secretin receptor. These data suggest that, like GRK-mediated phosphorylation of the ␤ 2 -AR, PKA-mediated phosphorylation of the secretin receptor may facilitate its sequestration. Although a GRK-mediated contribution to secretin receptor sequestration cannot be totally excluded, GRK-mediated enhancement of truncated receptor sequestration was small when PKA was inhibited (Fig. 5) and overexpressed GRKs had no effect on wild-type or truncated secretin receptor sequestration under whole cell conditions. Given the present results and our previous finding that PKA-mediated phosphorylation of the secretin receptor does not affect signaling (3), we suggest that the primary role of PKA-mediated phosphorylation of the secretin receptor in HEK 293 cells is to promote receptor sequestration. Further studies delineating PKA phosphorylation sites on the secretin receptor, followed by mutagenesis studies, should determine if the effect of PKA activity on receptor sequestration is direct. Putative PKA phosphorylation sites are located at two sites within the intracellular regions of the secretin receptor. One potential site of PKA-mediated phosphorylation is removed by truncation of the secretin receptor. Presently, though, we cannot rule out the possibility that PKA may promote sequestration of the secretin receptor through phosphorylation of some adaptor-like protein that targets the receptor to endocytic vesicles. Regardless of the mechanism, the fact that secretin receptor sequestration is associated with PKA activity raises the possibility that secretin receptor internalization may be subject to heterologous regulation.
In conclusion, regulation of internalization of the secretin receptor, a class II GPCR, differs substantially from that of the prototypical class I ␤ 2 -AR. Like the AT 1A R, the dominant negative mutants ␤-arrestin-V53D and dynamin-K44A do not inhibit the sequestration of the secretin receptor. These results imply that an endocytic pathway or mechanism alternative to that of the ␤ 2 -AR is common to other GPCRs and may have unique implications for receptor recycling and/or down-regulation. The secretin receptor should represent an important tool for delineating this mechanism. This study also provides the novel finding that internalization of secretin receptors is dependent upon PKA activity.