Selective Regulation of Endogenous G Protein-coupled Receptors by Arrestins in HEK293 Cells*

Arrestins play an important role in regulating desensitization and trafficking of G protein-coupled receptors (GPCRs). However, limited insight into the specificity of arrestin-mediated regulation of GPCRs is currently available. Recently, we used an antisense strategy to reduce arrestin levels in HEK293 cells and characterize the role of arrestins on endogenous G s -coupled receptors (Mun- dell, S. J., Loudon, R. B., and Benovic, J. L. (1999) Biochemistry 38, 8723–8732). Here, we characterized GPCRs coupled to either G q (M 1 muscarinic acetylcholine receptor (M 1 AchR) and P2y 1 and P2y 2 purinergic receptors) or G i (somatostatin and AT1 angiotensin receptors) in wild type and arrestin antisense HEK293 cells. The agonist-specific desensitization of the M 1 Ach and somatostatin receptors was significantly attenuated in antisense-ex-pressing cells, whereas desensitization of P2y 1 and P2y 2 purinergic and AT1 angiotensin receptors was unaffected by reduced arrestin levels. To further examine arrestin/ GPCR specificity, we studied the effects of endogenous GPCR activation on the redistribution of arrestin-2 epitope tagged with the green fluorescent protein (arres-tin-2-GFP). These studies revealed a receptor-specific movement of arrestin-2-GFP that mirrored the arrestin-receptor specificity observed in the antisense cells. Thus, agonist-induced activation of endogenous b 2 -adrenergic, prostaglandin E 2 , M 1 Ach, and

Arrestins mediate the desensitization and internalization of a number of G protein-coupled receptors (GPCRs) 1 (1). Agonistdependent phosphorylation of GPCRs by G protein-coupled receptor kinases promotes the high affinity binding of arrestins (1), which in turn sterically inhibits G protein interaction with the receptor, thereby terminating agonist-mediated signaling (2,3). Arrestins are recruited to multiple GPCRs after agonist activation (4), highlighting the important role of these proteins in receptor regulation.
Studies to date have employed overexpression of either wild type (5,6) or dominant negative (6 -8) arrestins with heterologously expressed GPCRs to elucidate many of the functions of these proteins. Although providing important insight into arrestin function and receptor specificity, one of the inherent drawbacks with this approach is the potential for nonspecific effects associated with protein overexpression. In a recent study we demonstrated that an antisense strategy can be successfully employed to reduce endogenous arrestin levels and effect changes in the internalization, desensitization, and resensitization of the G s -coupled ␤ 2 -adrenergic receptor (16). Moreover, the regulation of two other endogenous G s -coupled receptors, the A 2 adenosine and prostaglandin E 2 , was also shown to be affected by reductions in arrestin levels (16). These antisense-expressing cells therefore represent a novel system in which to further explore the involvement of arrestins in the desensitization of native receptors in intact cells.
Although numerous studies have examined the role of arrestins in regulating the responsiveness of the ␤ 2 -adrenergic receptor and other G s -coupled receptors, considerably less attention has been paid to GPCRs coupled to G i (which inhibit adenylyl cyclase activity) and G q (which stimulate inositol phosphate production and Ca 2ϩ release). Here we utilized HEK293 cells expressing arrestin-2 and arrestin-3 antisense constructs to perform a systematic analysis of arrestin specificity for regulating signaling by endogenous G q -coupled P2y 1 and P2y 2 purinergic (17) and M 1 Ach receptors (18) and G i -coupled somatostatin and AT1 angiotensin receptors. In addition, arrestin/GPCR specificity was further characterized by examining changes in arrestin-2green fluorescent protein (GFP) redistribution induced by endogenous receptor activation. Products. Rhodamine-conjugated transferrin, fura-2, and lysosomal tracker red were purchased from Molecular Probes. All other reagents were from Sigma.

Materials-Human
Cell Culture and Transfection-HEK293 cells were maintained in DMEM supplemented with 10% fetal bovine serum, 100 units/ml penicillin G, and 100 g/ml streptomycin sulfate (complete media) at 37°C in a humidified atmosphere of 95% air, 5% CO 2 . For culture of HEK293-EBNA cells stably transfected with pcDNA3 or pcDNA3 containing antisense cDNAs for arrestin-2 or arrestin-3, media was supplemented with 400 g/ml hygromycin and 200 g/ml Geneticin to maintain plasmid expression (16). For transient transfections of arrestin-2-GFP, HEK293 cells were grown in 100-mm dishes to 80 -90% confluence and transfected with 1-5 g of DNA using Fugene-6 following the manufacturer's instructions. Cells were incubated with a DNA/Fugene mixture for 24 h, the media was replaced, and the cells analyzed 48 h after transfection.
Inositol Phosphate Determination-HEK293 cells were seeded at a density of 80,000 cells/well in 12-well plates and labeled the following day for 18 -24 h with myo-[ 3 H]inositol (4 Ci/ml) in DMEM (high glucose, without inositol) as described previously (19). After labeling, cells were washed once in phosphate-buffered saline (PBS) and incubated in prewarmed DMEM containing 0.5% bovine serum albumin, 20 mM Hepes, pH 7.5, and 20 mM LiCl for 10 min at 37°C. Cells were then stimulated with either ATP (100 M), UTP (100 M), 2-methyl-ADP (100 M) or carbachol (1 mM). Reactions were terminated by removing the stimulation media and adding 0.8 ml of 0.4 M perchloric acid. Samples were harvested in Eppendorf tubes, and 0.4 ml of 0.72 N KOH, 0.6 M KHCO 3 were added. Tubes were vortexed and centrifuged for 5 min at 14,000 rpm in a microcentrifuge. Inositol phosphates were separated on Dowex AG 1-X8 columns. Total labeled inositol phosphates were counted by liquid scintillation.
Measurement of Intracellular Calcium ([Ca 2ϩ ] i )-The intracellular free Ca 2ϩ concentration was determined using the fluorescent Ca 2ϩ indicator fura-2 acetoxymethyl ester (fura-2/AM) as previously reported (20). Briefly, HEK-293 cells were loaded with fura-2/AM to a final concentration of 3 M in complete medium and incubated at 37°C for 50 min with stirring. After the loading, the cells were pelleted and washed twice with Locke's solution (154 mM NaCl, 5.6 mM KCl, 1.2 mM MgCl 2 , 2.2 mM CaCl 2 , 5 mM HEPES, 10 mM glucose, pH 7.4) to remove the extracellular dye. For the fluorimetric measurement of the [Ca 2ϩ ] i , 1 ϫ 10 6 cells were placed into a quartz cuvette in a thermostatically controlled cell holder at 37°C with continuous stirring. Fluorescence was measured at 340-and 380-nm excitation and 510-nm emission. Results were expressed as the ratio of the fluorescence at the two excitation wavelengths, which is proportional to the [Ca 2ϩ ] i . To generate concentration response curves, cells were exposed once to varying concentrations of ATP, ADP, or carbachol until a maximum response was attained (Ͻ30 s). To evaluate GPCR desensitization, cells were incubated with either ATP (0.01 nM-100 M) or carbachol (0.1 nM-1 mM) for 30 min, washed, and then rechallenged with either compound.
Adenylyl Cyclase Assays-Where required, drugs were added directly to the culture medium for varying times. Cells were harvested in 10 ml of ice-cold phosphate-buffered saline and pelleted by centrifugation at 200 ϫ g for 1 min. The resulting pellets were washed twice in 10 ml of ice-cold phosphate-buffered saline and frozen at Ϫ70°C until use. Adenylyl cyclase activity was measured using a protein binding assay (21). Cell pellets were thawed and homogenized in a glass Dounce homogenizer containing ice-cold homogenization buffer (0.3 M sucrose, 25 mM Tris-HCl, pH 7.4). A 40-l sample of homogenate was then added to 30 l of premix buffer (final assay concentration, 50 mM Tris-HCl, pH 7.5, 5 mM Mg 2ϩ , 1 mM ATP, 1 M GTP, 250 M Ro201724 (4-(3-butoxy-4methoxybenzyl) imidazolidin-2-one) as phosphodiesterase inhibitor, 20 mM creatine phosphate, and 130 units/ml creatine phosphokinase) and 30 l of drug at the relevant concentration. The tubes were incubated at 37°C for 10 min, and the reaction was terminated by the addition of 20 l of 100% trichloroacetic acid; the tubes were placed on ice for 10 min. Precipitated protein was pelleted by centrifugation at 2900 ϫ g for 20 min at 4°C, and 50 l of the resulting supernatant was added to 50 l H]cAMP in TE buffer (ϳ20,000 cpm), and 100 l of binding protein in TE buffer (to give a final concentration of ϳ750 g of protein/ml; prepared from bovine adrenal cortex). Tubes containing 50 l of standard concentrations of cAMP (0.125-20 pmol) were used to construct a standard curve. After a 2-h incubation at 4°C, 200 l of TE buffer containing charcoal (Norit GSX; 50 mg/ml final concentration) and bovine serum albumin (2 mg/ml final concentration) were added, and 15 min later, the tubes were centrifuged at 2900 ϫ g for 20 min at 4°C. The resulting supernatant was transferred into vials for liquid scintillation counting. Standard curve data were fitted to a logistic expression (GraphPAD Software, San Diego, CA), and the unknowns were read off. Protein content of homogenates was determined (22), and adenylyl cyclase activity was expressed as pmol of cAMP/min/mg of protein.
Fluorescence Microscopy and Single Cell Imaging-To assess the distribution of arrestin-2 in living cells, HEK293 cells (60-mm dish) were transfected with 0.25 g of arrestin-2-GFP. Cells were grown on poly-L-lysine-coated coverslips and mounted on an imaging chamber (Warner Instrument Corp) equipped with an inlet port through which media and drugs could be perfused. For experiments analyzing transferrin distribution, cells were incubated for 15 min with 200 g/ml rhodamine-conjugated transferrin on glass coverslips. Cells were washed 3 times with PBS before imaging then examined by microscopy on a Nikon Eclipse E800 fluorescence microscope using a Plan-Apo 60 ϫ 1.40 NA oil immersion objective. For experiments analyzing rab-5 distribution, arrestin-2 GFP-transfected cells were incubated with a rab-5-specific antibody (Molecular Probes) for 1 h at 4°C in DMEM supplemented with 1% bovine serum albumin. Cells were washed twice with PBS and then treated with drug for 30 min at 37°C in DMEM. The cells were then fixed with 3.7% formaldehyde, PBS for 15 min at room temperature, washed with PBS, and permeabilized with 0.05% Triton X-100/PBS/CaCl 2 for 10 min at room temperature. Nonspecific binding was blocked with buffer A (0.05% Triton X-100/PBS/CaCl 2 containing 5% nonfat dry milk) for 30 min at 37°C. Goat anti-mouse rhodamineconjugated secondary antibody (Molecular Probes) was then added at a dilution of 1:150 in buffer A for 1 h at 37°C. The cells were then washed 6 times with permeabilization buffer, with the last wash left at 37°C for 30 min. Finally, the cells were fixed with 3.7% formaldehyde as described. Coverslips were mounted using Slow-Fade mounting medium (Molecular Probes) and examined by microscopy as described above. All images were collected using QED Camera software and processed with Adobe Photoshop. For confocal microscopy, cells were prepared in the same manner, and images were obtained on a Bio-Rad MRC-Zeiss Axiovert 100 confocal microscope using a Zeiss Plan-Apo 63 ϫ 1.40 NA oil immersion objective.
Experimental Design and Statistics-Dose-response curves were analyzed by the iterative fitting program GraphPAD Prism (GraphPAD Software). Log concentration-effect curves were fitted to logistic expressions for single-site analysis. t1 ⁄2 values for agonist-induced desensitization were obtained by fitting data to a single exponential curve. Where appropriate, statistical significance was assessed by Student's t test or by two-way ANOVA using GraphPAD Prism.

RESULTS
Previously, we used stable expression of arrestin antisense constructs to successfully reduce endogenous arrestin levels in HEK293 cells (16). These cells revealed an important role for arrestins in the desensitization of endogenous G s -coupled receptors such as the ␤ 2 -adrenergic, adenosine A 2b , and prostaglandin E 2 (16). Three of these previously characterized cell lines with demonstrated reductions in arrestin-2 and/or arrestin-3 were used in the present study. The line AS 37, stably transfected with an arrestin-2 antisense construct, exhibits an ϳ50% reduction in arrestin-2 levels, whereas the lines AS 83, transfected with an arrestin-3 antisense construct, and AS 108, transfected with both antisense constructs, exhibit an ϳ50% decrease in arrestin-2 and ϳ75% decrease in arrestin-3 levels. Endogenous arrestin levels in vector-transfected cells were comparable with those in wild type HEK293 cells. To further define the arrestin specificity in regulating GPCRs, we next focused on endogenous G q -coupled receptors. HEK293 cells have been reported to contain a number of G qcoupled receptors including muscarinic (18), thrombin, lysophosphatidic acid, sphingolipid (23), endothelin, bradykinin (24), and P2y 1 and P2y 2 purinergic (17). The addition of carbachol (M 1 Ach), ADP (P2y 1 ), and ATP (P2y 2 ) to HEK293 cells resulted in significant stimulation of inositol phosphate accumulation (Fig. 1). In contrast, incubation with bombesin, bradykinin, thrombin, serotonin, lysophosphatidic acid, or endothelin resulted in Ͻ2-fold increases in inositol phosphate production compared with basal (data not shown). Thus, we focused our studies on the G q -coupled M 1 Ach, P2y 1 and P2y 2 purinergic receptors.
GPCR-mediated phospholipase C activation was assessed in intact cells by measuring [ 3 H]inositol phosphate accumulation following the addition of agonist. M 1 AchR-stimulated inositol accumulation by the muscarinic receptor agonist carbachol (1 mM) was ϳ70% greater 30 min after agonist addition to antisense versus control cells (2.1 Ϯ 0.3, 2.8 Ϯ 0.2-and 2.9 Ϯ 0.4-fold basal for wild type, AS 37, and AS 108 cells, respectively) (Fig. 1A). In contrast, stimulation of P2y 1 receptors with ADP (100 M) or P2y 2 receptors with ATP (100 M) produced equivalent increases in inositol phosphate production among the various cell lines tested (Fig. 1, B and C). Similarly, incubation with 2-methyl-ADP (100 M) or UTP (100 M), selective agonists for the P2y 1 and P2y 2 purinergic receptors, respectively, also produced equivalent increases in inositol phosphate production in the various lines (Fig. 1D). These studies demonstrate that the agonist-specific desensitization of the M 1 AchR was significantly attenuated in cells with reduced arrestin levels, whereas the desensitization of the P2y 1 and P2y 2 purinergic receptors was unaffected. To confirm that each of these receptors was coupled to phospholipase C via G q , as previously reported (17,18), cells were incubated with pertus-sis toxin (100 ng/ml, 12 h). Pertussis toxin pretreatment had no effect on inositol phosphate production by any of the endogenous receptors under investigation, confirming that each of these responses was via G q activation (data not shown).
To further investigate arrestin/receptor specificity, we characterized Ca 2ϩ mobilization induced by activation of each of these G q -coupled GPCRs. In agreement with previous studies we found that ADP, ATP, and carbachol all triggered a rapid and transient intracellular mobilization of Ca 2ϩ (17,18). Complete concentration-effect curves were subsequently generated for the M 1 Ach, P2y 1 , and P2y 2 purinergic receptors in antisense and vector transfected control cells (Fig. 2). The magnitude of the calcium response and agonist potency was unchanged in antisense-expressing cells compared with vectortransfected controls. In addition, the duration of the transient Ca 2ϩ responses was similar in antisense and control cell lines for each of the agonists tested (data not shown). These findings appear to contrast with our M 1 AchR-mediated inositol phosphate analysis, where increased inositol phosphate accumulation was observed in the antisense lines (Fig. 1). However, these differences are likely due to the transient nature of the Ca 2ϩ responses (measured ϳ30 s after agonist addition) versus the longer incubations used for inositol phosphate accumulation (Ͼ10 min).
Our studies next focused on the desensitization of P2y 2 and M 1 AchR-mediated calcium responses in vector-transfected cells. The concentration dependence of agonist-mediated desensitization was examined by pretreating cells with ATP (0.01 nM-100 M) or carbachol (0.1 nM-1 mM) for 30 min before subsequent determination of intracellular calcium mobilization by a maximally activating concentration of each of these agonists (Fig. 3, A and B). Carbachol pretreatment produced a homologous, concentration-dependent desensitization of the subsequent carbachol response, whereas the response to ATP remained largely unaffected. In contrast, ATP pretreatment led to homologous desensitization of the P2y 2 receptor but also resulted in heterologous desensitization of the M 1 AchR. Note that the duration of the transient Ca 2ϩ response was unaffected by agonist pretreatment, with only the magnitude of response affected.
To investigate the involvement of arrestins in the desensitization of these responses, antisense and control cells were pretreated with a fixed concentration of agonist for 30 min, and then intracellular Ca 2ϩ mobilization was measured in response to ATP or carbachol. As previously shown, carbachol pretreatment caused homologous desensitization of the M 1 AchR, whereas the response to ATP was not altered in wild type cells (Fig. 4A). The extent of M 1 AchR desensitization was significantly lower in the various antisense cell lines, further confirming a role for arrestins in M 1 AchR desensitization (Fig. 4A). Conversely, the homologous desensitization of the P2y 2 receptor by ATP was unaffected by reductions in arrestin levels, again suggesting that arrestins do not play a role in desensitization of this receptor (Fig. 4B). Interestingly, the heterologous desensitization of the M 1 AchR by ATP pretreatment was also unchanged in antisense cells, indicating that arrestins are involved only in the homologous desensitization of the M 1 AchR (Fig. 4B).
Since the endogenous arrestins appeared to differentially regulate G q -coupled receptors in HEK293 cells, we were also interested in characterizing G i -coupled receptors in these cells. Screening for multiple G i -coupled receptors (somatostatin, angiotensin, purinergic, ␣ 2 -adrenergic, lysophosphatidic acid, and muscarinic) revealed that the AT1 angiotensin and somatostatin receptors are functionally expressed in HEK293 cells. SRIF-14 (Fig. 6A) and angiotensin II (Fig. 6B) both produced a concentration-dependent inhibition of forskolin-stimulated adenylyl cyclase activity in whole cell homogenates. Both the IC 50 and the magnitude of inhibition of forskolin-stimulated adenylyl cyclase activity by each agonist was similar in vector-transfected and antisense-expressing cells.
We next constructed time courses of AT1 angiotensin and somatostatin receptor desensitization in control and antisense cells. In vector-transfected cells, pretreatment with SRIF-14 and angiotensin II resulted in rapid desensitization of receptor activity (Fig. 6, C and D). Interestingly, the SRIF-14-promoted desensitization was significantly attenuated in AS 37 and AS 108 lines (Fig. 6C). The rate of somatostatin receptor desensitization was significantly slower in both antisense cell lines (t1 ⁄2 ϭ 4.1 Ϯ 0.6 and 5.4 Ϯ 0.6 min for AS 37 and 108, respectively, compared with 2.3 Ϯ 0.4 for vector-transfected cells) (Fig. 6C). Interestingly, although the t1 ⁄2 of desensitization was slower in both antisense cell lines, both lines exhibited significant levels of receptor desensitization at later time points. Conversely, angiotensin II-promoted desensitization was no different in vector-transfected control and antisense cell lines.
The AT1 angiotensin receptor can also couple to activation of G q and phospholipase C in a number of cell lines. Unfortunately, when we measured inositol phosphate accumulation in vector-transfected and antisense cell lines in response to angiotensin II, we found a very weak response with Ͻ2-fold stim-ulation of inositol phosphate accumulation over basal (data not shown). This poor response precludes an accurate assessment of the effects of reduced arrestin levels on AT1 angiotensin receptor coupling to G q .
In addition to their role in regulating GPCR desensitization, arrestins have also been implicated in mediating receptor endocytosis (5)(6)(7)(8). Since our antisense studies suggested a role for arrestins in desensitization of selective GPCRs in HEK293 cells, we also wanted to address the potential role of arrestins in internalization of these receptors. Indeed, our previous studies revealed that arrestins mediate both desensitization and internalization of endogenous ␤ 2 -adrenergic receptors in HEK293 cells (16). Unfortunately, adequate reagents to study the trafficking of most endogenous GPCRs are unavailable. Thus, to further investigate the potential role of arrestins in GPCR internalization, we transfected HEK293 cells with an arrestin-2-GFP chimera and visualized arrestin trafficking in living cells. Arrestin-2-GFP has been demonstrated to redistribute from the cytosol to the plasma membrane upon receptor stimulation (24) and colocalize with selective internalized GPCRs (25,26).
Transfected HEK293 cells expressing arrestin-2-GFP were initially stimulated with G s -coupled receptor agonists. Before agonist stimulation, arrestin-2-GFP displayed a diffuse cytoplasmic distribution (Fig. 7, A and B). Initial attempts to visu- alize arrestin translocation from the cytosol to the membrane following endogenous receptor stimulation (0 -10 min) proved unsuccessful. However, it was noted that at later time points (Ͼ15 min), stimulation with the ␤-agonist isoproterenol (Fig. 7, C and E) or prostaglandin E 2 (Fig. 7, D and F) resulted in significant redistribution of arrestin-2-GFP into a distinct punctate pattern. A similar series of studies revealed that carbachol (Fig. 8, B and C) and SRIF-14 (Fig. 8, E and F) treatment also promoted a punctate distribution of arrestin-2-GFP. In contrast, stimulation with angiotensin II (Fig. 8H), ADP (data not shown), or ATP (data not shown) for up to 60 min had no effect on arrestin-2-GFP localization. Thus, these results reveal a receptor-specific movement of arrestin-2-GFP that mirrored the arrestin-receptor specificity observed in the antisense cells.
To further characterize the localization of these punctate vesicles containing arrestin-2-GFP, lysosomal tracker red, which is a marker for lysosomes (27), and rhodamine-labeled transferrin, which is a marker for early endosomes and the recycling centriolar compartment (28), were utilized. Before agonist stimulation, arrestin-2 displayed a diffuse cytoplasmic distribution (Fig. 9A), whereas transferrin displayed its typical punctate endosomal pattern (Fig. 9B). After carbachol stimulation for 30 min, arrestin-2-GFP redistributed into a distinct punctate pattern (Fig. 9C), exhibiting a significant degree of colocalization with labeled transferrin outside the central recycling centriolar compartment (note arrows in Fig. 9, C and D). Arrestin-2-GFP punctate vesicles also did not colocalize with lysosomal tracker red even after 2 h of carbachol treatment (data not shown). To further define the localization of arrestin-2 GFP puncta we utilized rab-5, a small G protein that is a marker for early endosomes (29). After carbachol stimulation, arrestin-2-GFP again redistributed into a distinct punc-tate pattern (Fig. 9E), exhibiting a significant degree of colocalization with labeled rab-5 (note arrows in Fig. 9, E and F).

DISCUSSION
In this study, two separate approaches were used to investigate arrestin/GPCR specificity in HEK293 cells. First, using previously characterized antisense arrestin-expressing cells (16), we investigated the effect of reduced arrestin levels on the regulation of endogenous G q -coupled (P2y 1 and P2y 2 purinergic, M 1 AchR) and G i -coupled (somatostatin and AT1 angiotensin) GPCRs. Studies on the G q -coupled GPCRs (examining both total inositol phosphate production and intracellular Ca 2ϩ mobilization) revealed that the agonist-specific desensitization of the M 1 AchR was significantly attenuated in cells with reduced arrestin levels, whereas the desensitization of the P2y 1 and P2y 2 purinergic receptors was unaffected. In the intracellular calcium mobilization experiments, heterologous desensitization of the M 1 AchR by ATP was unaltered by reductions in arrestins, providing further evidence that arrestins are mainly involved in the homologous desensitization of agonist-occupied GPCRs.
There are conflicting reports regarding the involvement of arrestins in MAchR regulation. A recent report suggested that overexpressed M 1 , M 3 , and M 4 AchRs undergo arrestin-2 and dynamin-dependent sequestration in HEK-293 tsA201 cells (30), whereas a previous report indicated that sequestration of these receptors is largely arrestin-independent (31). The disparity between these studies may reflect the potentially problematic nature of receptor overexpression in nonnative cell lines. In addition, neither of these studies investigated the consequence of arrestin co-expression on receptor signaling. Although the low endogenous levels of M 1 AchR expression precludes assessment of receptor internalization in HEK293 cells, both inositol phosphate and [Ca 2ϩ ] i data from the present study suggest that arrestins play a role in desensitization of the M 1 AchR. The involvement of arrestins in the regulation of purinergic receptor responses, meanwhile, has remained largely unexplored. Previous studies have indicated that both the P2y 1 and P2y 2 are regulated by protein kinase C (32) and that the carboxyl terminus of the P2y 2 plays a critical role in receptor desensitization and sequestration (33). Although these receptors are subject to rapid homologous desensitization (32), our study indicates that arrestins do not regulate endogenous P2y 1 and P2y 2 purinergic receptor responses in HEK293 cells.
Arrestin receptor specificity was further demonstrated for endogenous G i -coupled receptors in HEK293 cells. Agonistmediated desensitization of the somatostatin receptor was significantly attenuated in cells containing reduced arrestin levels, implicating arrestin involvement in desensitization of this response in HEK293 cells. Conversely, the desensitization of the AT1 angiotensin receptor was unaffected by reductions in arrestin expression. Previous studies have indicated that in heterologous expression systems in the somatostatin receptor family, both the sst 2A and sst 3 receptors are phosphorylated upon agonist stimulation (34,35). Both recombinant sst 2a receptors and endogenous sst 2 -like receptors are internalized in response to agonist activation, although apart from agonist binding the trigger for endocytosis is still unknown (35,36). A recent study suggested that somatostatin receptor desensitization may be dependent upon receptor internalization, although this process was shown to be G protein-coupled receptor kinase-independent (36). As is the case with the M 1 AchR, low levels of endogenous sst receptors limit our ability to assess agonist-induced changes in receptor distribution. Our study does, however, provide the first evidence of arrestin involvement in the desensitization of somatostatin receptors.
Previous studies have shown that AT1 angiotensin receptor internalization is independent of dynamin and arrestin-2, although arrestin overexpression can augment internalization (37). An association between protein kinase C-induced receptor phosphorylation and desensitization by angiotensin II has been demonstrated (38). In agreement with these studies, our results suggest that arrestins are not involved in the desensitization of endogenously expressed AT1 angiotensin receptor responses.
To further investigate arrestin-receptor specificity we used an arrestin-2-GFP conjugate. Several studies have utilized this approach to provide real-time analysis of receptor-arrestin interactions in single cells overexpressing various GPCRs (24 -26). Unlike these studies, however, we failed to see a rapid translocation (0 -120 s) of arrestins from the cytosol to the membrane upon activation of any of the endogenous GPCRs under investigation in this study (24). We attribute this to the low levels of endogenous receptor expression, which may limit the ability to recruit a sufficient amount of arrestin that is visible by conventional immunofluorescent microscopy. We did, however, observe an agonist-specific redistribution of arrestins into a punctate pattern following more prolonged activation of endogenous GPCRs. The pattern of GPCR-activated arrestin-2-GFP redistribution mirrored the arrestin-receptor specificity identified above. Only agonist-induced activation of ␤ 2 -adrenergic, prostaglandin E 2 , M 1 Ach, and somatostatin receptors induced arrestin-2-GFP redistribution, whereas the P2y 1 and P2y 2 purinergic and AT1 angiotensin receptor did not. Of note, activation of overexpressed AT1 angiotensin receptors has recently been shown to trigger a clear time-dependent redistribution of arrestins to an intracellular vesicular compartment where they colocalize with internalized receptors (26). The discrepancy between these results and the present study may be explained by differences in receptor expression levels. However, since our results suggest that endogenously expressed AT1 angiotensin receptors are not regulated by arrestins in HEK293 cells, one needs to be cautious when interpreting results using overexpressed receptors and arrestins.
Previous studies have indicated that for some GPCRs, arrestins can traffic with internalized receptors into early endosomes (25,26). Unfortunately, we were unable to visualize the trafficking of endogenous receptors, but studies using rhodamine labeled-transferrin, rab-5, and lysosomal tracker red indicate that arrestin-2-GFP-containing vesicles colocalize with transferrin receptors and rab-5 in early endosomes (28,29). The function of this prolonged association of arrestin-2 in an endosomal compartment is still unclear, although a recent study using heterologously overexpressed arrestin and GPCR, suggested that this may dictate the rate of receptor dephosphorylation, recycling, and resensitization (39).
In summary, this study demonstrates that arrestins selec- Before stimulation and viewing, coverslips were mounted in a chamber as described under "Experimental Procedures." The initial localization of arrestin-2-GFP and transferrin are shown in panels A and B, respectively. Carbachol (1 mM) was added, and the redistribution of arrestins and transferrin was monitored in real time as described under "Experimental Procedures." Rab-5 localization was determined in fixed cells after carbachol addition using a rab-5-specific primary antibody and a rhodamine-labeled secondary antibody as described under "Experimental Procedures." Images shown (C and E, arrestin-2-GFP; D, transferrin; F, rab-5) were obtained 30 min after agonist addition. The arrows indicate overlap of punctate pattern for arrestin-2-GFP and transferrin (C and D) or arrestin-2-GFP and rab-5 (E and F). All images were processed using Adobe Photoshop software. tively regulate endogeneously expressed GPCRs from the G q (M 1 Ach)-and G i (somatostatin)-coupled families in HEK293 cells. Future studies will investigate the structural determinants of GPCRs that contribute to arrestin selectivity.