The third intracellular loop of alpha 2-adrenergic receptors determines subtype specificity of arrestin interaction.

Nonvisual arrestins (arrestin-2 and -3) serve as adaptors to link agonist-activated G protein-coupled receptors to the endocytic machinery. Although many G protein-coupled receptors bind arrestins, the molecular determinants involved in binding remain largely unknown. Because arrestins selectively promote the internalization of the alpha(2b)- and alpha(2c)-adrenergic receptors (ARs) while having no effect on the alpha(2a)AR, here we used alpha(2)ARs to identify molecular determinants involved in arrestin binding. Initially, we assessed the ability of purified arrestins to bind glutathione S-transferase fusions containing the third intracellular loops of the alpha(2a)AR, alpha(2b)AR, or alpha(2c)AR. These studies revealed that arrestin-3 directly binds to the alpha(2b)AR and alpha(2c)AR but not the alpha(2a)AR, whereas arrestin-2 only binds to the alpha(2b)AR. Truncation mutagenesis of the alpha(2b)AR identified two arrestin-3 binding domains in the third intracellular loop, one at the N-terminal end (residues 194-214) and the other at the C-terminal end (residues 344-368). Site-directed mutagenesis further revealed a critical role for several basic residues in arrestin-3 binding to the alpha(2b)AR third intracellular loop. Mutation of these residues in the holo-alpha(2b)AR and subsequent expression in HEK 293 cells revealed that the mutations had no effect on the ability of the receptor to activate ERK1/2. However, agonist-promoted internalization of the mutant alpha(2b)AR was significantly attenuated as compared with wild type receptor. These results demonstrate that arrestin-3 binds to two discrete regions within the alpha(2b)AR third intracellular loop and that disruption of arrestin binding selectively abrogates agonist-promoted receptor internalization.

G protein-coupled receptors (GPCRs) 1 transduce extracellular stimuli into intracellular signaling via coupling to heterotrimeric guanine nucleotide-binding proteins (G proteins) (1). To ensure that stimuli are translated into signals of appropriate magnitude and specificity, these signaling cascades are tightly regulated. GPCRs are subject to three principal modes of regulation: desensitization, in which a receptor becomes refractory to continued stimuli; endocytosis, whereby receptors are removed from the cell surface; and down-regulation, in which total cellular receptor levels are decreased (2,3). Although multiple mechanisms contribute to these regulatory processes, GPCR phosphorylation by G protein-coupled receptor kinases (GRKs) and subsequent binding of arrestins plays an important role in the regulation of many GPCRs (2).
Four mammalian arrestins have been identified with arrestin-1 and -4 being specific to the visual system and arrestin-2 and -3 (also termed ␤-arrestin-1 and -2) being ubiquitously expressed (4 -7). Arrestin binding to activated-phosphorylated GPCRs results in the physical uncoupling of receptor from G protein, a process that functions to terminate agonist-mediated signaling. The two nonvisual arrestins also directly interact with clathrin (8), the adaptor protein AP2 (9), and phosphoinositides (10) to promote GPCR internalization. Indeed, nonvisual arrestins have been implicated in the desensitization and internalization of a wide variety of GPCRs including members of the class 1 (rhodopsin-like) and class 2 (secretin-like) families (2,11). Nevertheless, despite numerous studies that demonstrate an essential role for arrestins in the regulation of GPCR signaling and trafficking, the precise molecular determinants within GPCRs required for arrestin binding have not been thoroughly characterized.
Previous studies have demonstrated an important role for the third intracellular loops of the ␣ 2 ARs in mediating proteinprotein interaction. These loops are quite large (Ͼ150 amino acids) and include sites for GRK phosphorylation (26), Gi activation (27), and binding of 14-3-3 (28), sphinophilin (29), and arrestin (30). Although arrestin binding to GPCRs is dependent on both the phosphorylation and activation state of the receptor (31), the receptor domains that mediate the agonist-dependent nature of arrestin binding have not been thoroughly characterized. Because third intracellular loops mediate agonist-dependent binding and activation of heterotrimeric G proteins (32,33), it seems likely that specific regions of the third intracellular loop might also confer the agonist dependence and selectivity of arrestin binding. Indeed, the third intracellular loop has been implicated in arrestin interaction for a number of GPCRs including rhodopsin (34), ␣ 2a -adrenergic (30), M 2 and M 3 muscarinic (30), ␦-opioid (35), 5-hydroxytryptamine 2A (36), CXCR4 (37), and the luteinizing hormone/choriogonadotropin (LH/CG) receptor (38).
The subtype-specific differences in arrestin sensitivity for the three ␣ 2 ARs provides a useful model to identify specific regions involved in arrestin binding. To address this issue, we studied arrestin binding to a series of glutathione S-transferase (GST) fusion proteins containing various regions of the third intracellular loops of the three ␣ 2 AR subtypes. Our results revealed arrestin binding specificity that recapitulates the arrestin selectively observed previously in ␣ 2 AR trafficking (25). Truncation and site-directed mutagenesis revealed that arrestin-3 binds to two discrete regions within the ␣ 2b AR third intracellular loop. Moreover, disruption of arrestin binding selectively abrogated agonist-promoted internalization of the ␣ 2b AR. These studies help to address questions of specificity in GPCR/arrestin interaction that ultimately will lead to a better understanding of the role of arrestins in regulating receptormediated signaling.

EXPERIMENTAL PROCEDURES
Plasmid Construction-FLAG-tagged ␣ 2A AR, ␣ 2B AR, and ␣ 2C AR were cloned into pcDNA3 as described previously (25). The third intracellular loops of the ␣ 2a AR (residues 218 -374), ␣ 2b AR (residues 194 -368), and ␣ 2c AR (residues 232-379) were amplified by PCR using the full-length receptors as template. The PCR products were cut with EcoRI/XhoI (␣ 2b AR and ␣ 2c AR) or EcoRI/SalI (␣ 2a AR), gel-purified, inserted into the plasmid pGEX4T-2 in-frame with GST, and sequenced. The following ␣ 2b AR third loop truncation mutants were made and cloned into pGEX4T-2 using the same strategy , and RR/3R). PCR products were cut with EcoRI/XhoI, ligated into EcoRI/XhoI-digested pGEX4T-2, and sequenced. The K200A/R201A, R204A/R205A, and R358A/R359A/R360A mutations were also generated in full-length ␣ 2b AR by two-step PCR using the expand high fidelity PCR system according to the manufacturer's recommendations. Briefly, using pcDNA3-␣ 2b AR as template, an N-terminal fragment of the mutant ␣ 2b AR was created using a cytomegalovirus (5Ј-tgt acg gtg gga ggt-3Ј) sense primer and a mutagenic antisense primer, and a C-terminal fragment of the mutant ␣ 2b AR was generated using a mutagenic sense primer and an Sp6 (5Ј-gat aag ata tca cag tgg att tac-3Ј) antisense primer. Mutagenic primers used were: KR sense (5Ј-ctg atc gcc gca gcc agc aac cgc-3Ј) and antisense (5Ј-gcg gtt gct ggc tgc gta gat cag-3Ј); RR sense (5Ј-cgc agc aac gcc gca ggt ccc agg-3Ј) and antisense (5Ј-cct ggg acc tgc ggc gtt gct gcg-3Ј); 3R sense (5Ј-ggg cag tgg tgg gct gca gcg gcg cag ctg acc cgg-3Ј) and antisense (5Ј-ggt cag ctg cgc cgc tgc agc cca cca ctg ccc-3Ј). PCR products were purified, and N-terminal and C-terminal products (100 ng of each) were then used as template using the cytomegalovirus and Sp6 primers. PCR products were cut with EcoRI/XhoI, purified, and ligated into pcDNA3. An RR/3R mutation in full-length ␣ 2b AR was generated as described above using RR as template for the first round of PCR and then proceeding as specified above for the 3R mutation.
Expression and Purification of GST-␣ 2 AR Fusion Proteins-BL21 (De3) lysS cells transformed with a GST-␣ 2 AR fusion construct were grown overnight at 37°C, diluted 1:100 into LB containing ampicillin, grown for 3 h at 37°C, and then induced with 0.1 mM isopropylthiogalactyl-pyranosidase for 2 h at 30°C. Cells were pelleted (3000 ϫ g for 30 min) and washed with phosphate-buffered saline (PBS) containing 1 mM dithiothreitol and protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 0.2 mg/ml benzamidine, 20 g/ml leupeptin, 10 g/ml pepstatin, 10 g/ml aprotinin). Cells were pelleted, resuspended in 1-3 ml of PBS plus protease inhibitors (per 100 ml of culture), lysed by incubation for 10 min on ice with 1 mg/ml lysozyme, and then aliquoted (0.4 ml), frozen, and stored at Ϫ80°C until needed. Aliquots were thawed on ice, Triton X-100 (2% final) and sarcosyl (0.5% final) were added, and the cells were frozen, thawed, and centrifuged for 1 h (30,000 rpm in TLA-45 rotor). The supernatant (ϳ200 l) was then incubated with 200 l of 50% glutathione-agarose bead slurry for 1 h at 4°C, and the beads were washed twice with PBS containing protease inhibitors and 1% Triton X-100 and resuspended in 200 l of PBS with protease inhibitors. To quantify protein amounts, 20 l of resuspended beads were incubated with SDS sample buffer and centrifuged, and the supernatant was electrophoresed on a 10% SDS-polyacrylamide gel. The gel was stained with Coomassie Blue, and protein levels were quantified using bovine serum albumin as standard.
Cell Culture and Transfection-HEK 293 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 g/ml streptomycin, and 100 units/ml penicillin at 37°C in a humidified atmosphere containing 5% CO 2 . HEK 293 cells grown to 50 -75% confluence in 60-mm dishes were transfected with 3 g of FLAG-tagged wild type or mutant (KR, RR, 3R, RR/3R) ␣ 2b AR using 10 l of FuGENE-6 reagent according to the manufacturer's protocol. Briefly, cells were incubated with the FuGENE-DNA mixture for 5 h and then split into poly-l-lysine-coated 12-well dishes (for ERK1/2 assays) or 24-well dishes (for enzyme-linked immunosorbent assay). Enzyme-linked immunosorbent assays were performed 24 h after transfection as described previously (25), whereas ERK1/2 assays were done 48 h after transfection.
Analysis of Phospho-ERK1/2-HEK 293 cells in 60-mm dishes were transfected as described above and then split into three wells of a poly-l-lysine-coated 12-well dish. The following day, cells were serumstarved overnight in Dulbecco's modified Eagle's medium containing 0.5% fetal bovine serum. The next day, cells were stimulated with 10 M UK14304 at 37°C for 0 -30 min and then rinsed with PBS, lysed by addition of SDS sample buffer, and scraped off the plates. Samples were boiled and then electrophoresed on 10% SDS-polyacrylamide gels. The gels were transferred to nitrocellulose and blocked for 30 min in Trisbuffered saline containing 0.1% Tween 20 and 5% non-fat dry milk. Phospho-ERK1/2 was detected as described previously (25).

Arrestins Differentially Bind to ␣ 2 AR Third Intracellular
Loops-Previous studies have demonstrated subtype-specific differences in arrestin-promoted internalization of ␣ 2 ARs with arrestin-3 enhancing internalization of ␣ 2b AR and ␣ 2c AR and arrestin-2 only acting on the ␣ 2b AR (25). To explore the mechanistic basis of arrestin binding selectivity for ␣ 2 ARs, we initially focused on the third intracellular loop of the ␣ 2 ARs. The first and second intracellular loops of the ␣ 2 ARs share sequence homology; however, the third intracellular loops have very divergent sequences, and this region likely directly contributes to arrestin binding specificity (27,30). The third intracellular loops of the ␣ 2a AR (residues 218 -374), ␣ 2b AR (residues 194 -368), and ␣ 2C AR (residues 232-379) were expressed as GST fusion proteins, purified on glutathione-agarose, and then used in direct binding assays with purified arrestin-1, -2, or -3 (Fig. 1A). Arrestin-1 did not bind to any of the ␣ 2 AR fusion proteins (data not shown), arrestin-2 bound only to the GST-␣ 2b AR third loop, and arrestin-3 bound to both the GST-␣ 2b AR and the GST-␣ 2c AR third loops but not to the ␣ 2a AR (Fig. 1B). A dose-response analysis was next performed to determine whether there were binding differences between arrestin-2 and -3 and the ␣ 2b AR. Arrestin-3 was found to bind much more effectively to the ␣ 2b AR as compared with arrestin-2 with ϳ20fold more binding at the highest concentrations of arrestin (Fig.  1C). Overall, these results largely recapitulate the selectivity of arrestins in promoting internalization of the ␣ 2 ARs (25) and suggest that this selectivity is mediated by differences in arrestin binding to the third intracellular loops of these receptors.
Identification of Arrestin-3 Binding Domains within the ␣ 2B AR Third Loop-In an effort to identify specific arrestin binding domains, we further investigated the interaction of arrestin-3 with the ␣ 2b AR third loop. Initially, the third loop was bisected into NT (residues 194 -292) and CT (residues 293-368) pieces and tested for arrestin binding (Fig. 2A). These studies revealed that both the NT and CT regions of the ␣ 2b AR third loop bind arrestin-3, albeit not as well as the intact third loop (Fig. 2B). Truncation mutagenesis of the NT construct revealed that arrestin-3 binding was primarily localized to the first ϳ20 residues as NT1 (residues 194 -214) bound arrestin-3 as well as NT, whereas NT2 (residues 206 -292) did not bind (Fig. 2B). Similar analysis of CT revealed that arrestin-3 binding was primarily localized to the last ϳ25 amino acids since CT3 (residues 344 -368) bound arrestin-3, whereas CT1 (residues 293-312), which contains a long stretch of acidic residues implicated previously in receptor desensitization (39), and CT2 (residues 313-358) did not (Fig. 2B). These results suggest that the proximal and distal ends of the third intracellular loop of the ␣ 2b AR contain the major arrestin-3 binding domains.
To define specific residues within NT1 and CT3 that contribute to arrestin binding, a series of alanine point mutants were generated and tested for their ability to bind arrestin-3. Point mutations introduced into the NT1 and CT3 are indicated by an asterisk in Fig. 3, A and B, respectively, whereas double mutations are underlined. These studies revealed that mutation of basic residues within the NT1 and CT3 constructs resulted in a dramatic reduction in arrestin-3 binding. Specifically, mutation of Arg-201, Arg-204, and Arg-205 in NT1 effectively disrupted arrestin-3 binding, whereas mutation of Arg-207 partially disrupted binding (Fig. 3A). In contrast, mutation of Lys-200, Ser-202, Asn-203, Pro-205, and Arg-206 in NT1 had no significant effect on arrestin-3 binding. The placement of essential residues involved in arrestin binding was interesting given recent studies demonstrating that mutation of Lys-382 in the third loop of the parathyroid hormone receptor reduced arrestin-promoted internalization (40), whereas residues in the LH/CG receptor important for arrestin-promoted internalization are localized within the analogous region of the third intracellular loop (38). Mutational analysis of CT3 also revealed the involvement of basic residues in arrestin binding with mutation of several arginines (358, 359, 360, 365, and 368) as well as Lys-367, resulting in a significant reduction FIG. 1. In vitro binding of arrestins to GST-␣ 2 AR third loop fusion proteins. In A, ␣ 2 AR third loops were expressed as GST fusions in bacteria and purified on glutathione-agarose beads. Proteins were electrophoresed on a 10% SDS-polyacrylamide gel and visualized by Coomassie Blue staining. Shown are 1 g of purified GST or GST fusions containing the third intracellular loop of the ␣ 2a AR (2A), ␣ 2b AR (2B), or ␣ 2c AR (2C) and purified arrestin-1, -2, and -3. In B, 500 ng of purified GST-␣ 2 AR fusion proteins were incubated with 300 ng of purified arrestin-2 or arrestin-3 as described under "Experimental Procedures." Bound proteins were eluted by addition of SDS sample buffer and then electrophoresed on a 10% SDS-polyacrylamide gel, transferred to nitrocellulose, and probed with an arrestin monoclonal antibody (F4C1). The experiment was repeated a total of three times with similar results. In C, various concentrations of arrestin-2 and arrestin-3 (0.02-0.34 M) were incubated with 250 ng of GST-␣ 2b AR third loop fusion protein as described under "Experimental Procedures." Proteins were separated on a 10% SDS-polyacrylamide gel, transferred to nitrocellulose, and blotted with arrestin-2-or arrestin-3-specific polyclonal antibodies and goat anti-rabbit secondary antibody. Blots were quantified by densitometric scanning and compared with a purified arrestin protein standard. Each point represents the mean Ϯ S.E. of four to five independent experiments.

FIG. 2. In vitro binding of arrestin-3 with specific domains of the ␣ 2b AR third loop.
A, schematic representation of the GST-␣ 2b AR third loop constructs used in binding assays with purified arrestin-3. Relative arrestin-3 binding to ␣ 2b AR third loop segments is represented as ϩ, with ϩϩϩϩ indicating the strongest binding. In B, GST-␣ 2b AR third loop fusions (500 ng) were incubated with 300 ng of purified arrestin-3, and bound proteins were eluted, electrophoresed on a 10% SDS-polyacrylamide gel, transferred to nitrocellulose, and probed with an arrestin-3-specific antibody. Shown is a representative blot from three independent experiments. in arrestin-3 binding (Fig. 3B). In contrast, mutation of Leu-344 and -345 and Gln-355 had a partial effect on binding, whereas Gln-362, Thr-364, and Glu-366 mutations had no effect on arrestin-3 binding. The important role of the proximal and distal ends of the third intracellular loop in arrestin binding is reminiscent of the domains implicated in G protein binding and activation (27). In fact, previous studies have identified a BBXXB motif (where B is a basic residue and X is any residue) in the ␣ 2 AR involved in Gi activation with Lys-367 and Arg-368 contributing to this motif (27,32). This suggests that significant overlap between the arrestin-3 and Gi binding sites on the ␣ 2b AR will likely contribute to the mechanism by which arrestin mediates desensitization (i.e. G protein uncoupling).
To verify that specific arrestin binding mutations made within the NT1 and CT3 constructs were important for arrestin binding in the context of the whole third loop, a series of GST-␣ 2b AR third loop mutants were generated. Four different GST-␣ 2b AR fusions incorporating either the K200A/R201A (KR), R204A/R205A (RR), R358A/R359A/R360A (3R), or RR/3R mutations were used in binding assays with purified arrestin-3 and compared with the wild type third loop. The ability of the KR and 3R mutants to bind arrestin-3 was modestly reduced as compared with the wild type ␣ 2b AR (Fig. 4A). However, RR and RR/3R mutant binding to arrestin-3 was strongly attenuated with an 80 -90% reduction in binding. To detect potential binding differences between the RR and RR/3R mutations, a doseresponse analysis was performed. The RR mutation alone had a very similar binding pattern to RR/3R with the R204A/R205A mutation almost completely disrupting arrestin-3 binding to the ␣ 2b AR third loop (Fig. 4B). Taken together, these results suggest that the ␣ 2b AR third intracellular loop contains two arrestin-3 binding domains with the N-terminal region playing the predominant role.
Internalization of Wild Type and Mutant ␣ 2b ARs in HEK 293 Cells-We next incorporated the various third loop mutations (KR, RR, 3R, and RR/3R) into the holo-␣ 2b AR. Because arrestins are involved in agonist-promoted internalization of the ␣ 2b AR (25), we anticipated that disrupting arrestin binding to the ␣ 2b AR third intracellular loop would attenuate receptor internalization. HEK 293 cells expressing FLAG-tagged wild type or mutant ␣ 2b ARs were incubated with agonist for 30 min and then analyzed for cell surface receptors by enzyme-linked immunosorbent assay (Fig. 5). Internalization of the wild type ␣ 2b AR was ϳ30% after agonist treatment, consistent with previous studies of ␣ 2b AR internalization in HEK 293 cells (41). Internalization of the KR and 3R mutant receptors was very similar to that of the wild type ␣ 2b AR, consistent with the in vitro data showing that these mutations did not severely disrupt arrestin binding. In contrast, internalization of the RR mutant was reduced ϳ50%, whereas the RR/3R mutant was decreased ϳ65% as compared with the wild type receptor. These data suggest that disrupting arrestin binding to the third intracellular loop of the ␣ 2b AR has an inhibitory effect on agonist-promoted receptor internalization. These results also help to confirm the important role of arrestins in mediating internalization of the ␣ 2b AR.
Signaling of Wild Type and Mutant ␣ 2b ARs in HEK 293 Cells-To ensure that the various mutations did not directly affect signaling of the ␣ 2b AR, we next analyzed the ability of the wild type and mutant ␣ 2b ARs to activate ERK1/2. Our previous studies demonstrated that all three ␣ 2 AR subtypes activate ERK1/2 in an agonist-dependent manner via a pathway that is Gi-and Ras-dependent but arrestin-and internalization-independent (25). HEK 293 cells expressing wild type or mutant (RR, 3R, RR/3R) ␣ 2b ARs were incubated with agonist for 0, 5, or 30 min and then analyzed for ERK activation by immunoblotting for phospho-ERK1/2. All receptors activated ERK1/2 from 5-to 7-fold after a 5-min treatment with agonist, suggesting that the mutations that inhibit arrestin binding and receptor internalization have no significant effect on ␣ 2b AR activation of signaling (Fig. 6, A and B).
Conclusions-Our results on the NT1 region of the ␣ 2b AR suggest the importance of a BXXBB binding motif that is essential for arrestin-3 binding. Mutation of Arg-201, Arg-204, or Arg-205 completely disrupted arrestin-3 binding, whereas mutation of surrounding residues had minimal effect on arrestin binding (Fig. 3). Interestingly, the analogous region of the ␣ 2c AR, but not the ␣ 2a AR, can also directly bind arrestin-3 (data not shown). Although the three ␣ 2 ARs share significant homology within the N-terminal 20 residues of the third intracellular loops, a key basic residue present in both the ␣ 2b AR (Arg-201) and the ␣ 2c AR (Arg-234) is replaced with Gln-221 in the ␣ 2A AR (Fig. 7). The absence of this particular basic residue within the ␣ 2A AR may disrupt arrestin binding. It is also interesting to note that mutation of Arg-239 (the last B in the BXXBB motif) within the NT1 region of the ␣ 2c AR completely disrupts arrestin-3 binding (data not shown), further establishing the importance of basic residues within this region for arrestin binding. Several recent studies have also suggested a role for receptor third intracellular loops in arrestin binding (30, 34 -38). For example, the third loops of the ␦-opioid receptor (35) and the LH/CG receptor (38) can directly bind arrestins, and third loop peptides from the LH/CG receptor inhibit receptor desensitization by sequestering arrestin-2 (38). The domains involved in these interactions share similar homology and generally contain several basic residues as well as serines and/or threonines (Fig. 7). Although the ␦-opioid receptor studies suggested a role for two serines in arrestin binding (35), all of these domains contain the BXXBB motif and help to establish the importance of such a motif in arrestin binding.
At least two sites within arrestin are involved in GPCR binding, a phosphorylation recognition site that binds to phosphoserines/threonines on the receptor and an activation recognition site that binds the agonist-bound conformation of the receptor (31). Arrestin binding to phosphoserines/threonines is thought to destabilize the arrestin polar core and promote secondary high affinity binding to the receptor and binding to the phospholipid bilayer (42). In our studies, the third loops were not phosphorylated; however, arrestin binding sites that would normally be inaccessible in the unactivated holo-recep-tor are likely available for arrestin binding in the isolated third intracellular loops. One possible reason why the ␣ 2b AR binds arrestin-2 and -3 with higher affinity than the ␣ 2c AR may be the acidic stretch of amino acids located within the ␣ 2b AR third loop. This acidic stretch is involved in mediating receptor desensitization (39) and may act to destabilize the arrestin polar core and promote binding much like phosphorylation does.
It is important to note that although the RR and RR/3R mutant ␣ 2b ARs demonstrated a defect in agonist-promoted receptor internalization, the inhibition was not complete. This suggests either the incomplete disruption of arrestin binding within the third loop (as seen in our GST binding studies) or a role for additional arrestin binding sites on the receptor. Additional potential arrestin binding sites on ␣ 2 ARs include the second intracellular loop, which contains the DRY motif that has been implicated in arrestin association in several GPCRs (43)(44)(45). Further studies involving ␣ 2 AR domains may reveal insight as to how arrestin regulates both internalization and signaling of receptors as recent studies propose that arrestin binding to different regions of a receptor can regulate different functions (40).
In summary, we have identified two discrete arrestin-3 binding domains within the ␣ 2b AR third intracellular loop. Specifically Arg-201, Arg-204 and Arg-205, and to a lesser extent Arg-358, Arg-359, and Arg-360, are important for arrestin binding in vitro. Mutation of these residues severely impairs agonist-dependent receptor internalization, suggesting the critical nature of arrestin in this process. Interestingly, arrestin-3 directly interacts with the basic residues of the G protein activation motif located in the third loop, suggesting a model by which arrestin binding displaces G protein, thus attenuating signaling in addition to linking the receptor to clathrin-coated pits.