Receptor-specific Ubiquitination of β-Arrestin Directs Assembly and Targeting of Seven-transmembrane Receptor Signalosomes*

Angiotensin II type 1a (AT1a), vasopressin V2, and neurokinin 1 (NK1) receptors are seven-transmembrane receptors (7TMRs) that bind and co-internalize with the multifunctional adaptor protein, β-arrestin. These receptors also lead to robust and persistent activation of extracellular-signal regulated kinase 1/2 (ERK1/2) localized on endosomes. Recently, the co-trafficking of receptor-β-arrestin complexes to endosomes was demonstrated to require stable β-arrestin ubiquitination (Shenoy, S. K., and Lefkowitz, R. J. (2003) J. Biol. Chem. 278, 14498–14506). We now report that lysines at positions 11 and 12 in β-arrestin2 are specific and required sites for its AngII-mediated sustained ubiquitination. Thus, upon AngII stimulation the mutant β-arrestin2K11,12R is only transiently ubiquitinated, does not form stable endocytic complexes with the AT1aR, and is impaired in scaffolding-activated ERK1/2. Fusion of a ubiquitin moiety in-frame to β-arrestin2K11,12R restores AngII-mediated trafficking and signaling. Wild type β-arrestin2 and β-arrestin2K11R,K12R-Ub, but not β-arrestin2K11R,K12R, prevent nuclear translocation of pERK. These findings imply that sustained β-arrestin ubiquitination not only directs co-trafficking of receptor-β-arrestin complexes but also orchestrates the targeting of “7TMR signalosomes” to microcompartments within the cell. Surprisingly, binding of β-arrestin2K11R,K12R to V2R and NK1R is indistinguishable from that of wild type β-arrestin2. Moreover, ubiquitination patterns and ERK scaffolding of β-arrestin2K11,12R are unimpaired with respect to V2R stimulation. In contrast, a quintuple lysine mutant (β-arrestin2K18R,K107R,K108R,K207R,K296R) is impaired in endosomal trafficking in response to V2R but not AT1aR stimulation. Our findings delineate a novel regulatory mechanism for 7TMR signaling, dictated by the ubiquitination of β-arrestin on specific lysines that become accessible for modification due to the specific receptor-bound conformational states of β-arrestin2.

Receptor endocytosis is an important mechanism for the movement of cell surface receptors into internal cellular compartments, facilitating receptor down-regulation and signal desensitization (1). On the other hand, recent studies of several seven transmembrane receptors (7TMRs) 1 and receptor tyrosine kinases suggest that endocytic processes may also lead to initiation or prolongation of signal transduction by the activation of MAPKs (2)(3)(4). For the mammalian 7TMRs, agonist stimulation leads not only to initiation of G protein-dependent signaling pathways but also to receptor phosphorylation on serine-threonine residues by G protein-coupled receptor kinases (5). Receptor activation and phosphorylation facilitate recruitment of the multifunctional adaptor proteins ␤-arres-tin1 and ␤-arrestin2 (also called arrestin2 and arrestin3) (6). ␤-Arrestins desensitize G protein-mediated signaling by sterically blocking further receptor G protein interaction. ␤-Arrestins have also been demonstrated to bind proteins of the cellular trafficking machinery, such as clathrin, AP2 (adaptin protein subunit 2), NSF (N-ethylmaleimide-sensitive factor), and ARF6 (ADP-ribosylation factor 6), and a variety of signaling proteins such as c-Src, ERK1/2, and JNK3, thereby mediating 7TMR internalization and some aspects of signaling (7)(8)(9)(10)(11)(12)(13). Accordingly, ␤-arrestin serves as an endocytic and signaling adaptor for many 7TMRs. However, it is unknown how the endocytic and signaling roles of ␤-arrestin are integrated and regulated.
Two patterns of 7TMR interaction with ␤-arrestins have been described (14). "Class A" receptors such as the ␤2-adrenergic receptor (␤2AR) bind ␤-arrestins after agonist stimulation and traffic with them to clathrin-coated pits. However, because the interaction is of relatively low affinity, ␤-arrestin dissociates from the receptor as the coated pit pinches off as a coated vesicle, and the receptor internalizes without the ␤-arrestin. In contrast, for "Class B" receptors such as the angiotensin II type 1a receptor (AT1aR) and the vasopressin V2 receptor (V2R), agonist-activated receptors bind ␤-arrestins with high affinity, do not dissociate from them in the plasma membrane, and internalize together into endosomes where they remain stably associated for prolonged periods. Moreover, receptors that bind ␤-arrestin stably (Class B) stabilize cytosolic pERK while inhibiting nuclear pERK activity (12,15,16).
Here we set out to investigate whether ␤-arrestin ubiquitination might serve as a molecular link between its endocytic and signaling functions. Not only does this appear to be the case, but we uncovered the existence of a previously unsuspected heterogeneity of receptor-specific activated and ubiquitinated ␤-arrestin conformations. Our findings suggest a mechanism that may explain the conundrum of how only two forms of ␤-arrestin can mediate and modulate the disparate actions of many 7TMRs.

MATERIALS AND METHODS
Cell lines, Reagents, and Plasmids-COS-7 and HEK-293 cells were obtained from American Type Culture Collection. COS-7 cells were maintained in Dulbecco's modified Eagle's medium (Sigma) supplemented with 10% fetal bovine serum and 1% penicillin streptomycin and transiently transfected with Lipofectamine reagent (Invitrogen). HEK-293 cells were maintained in minimal essential medium supplemented with fetal bovine serum and transiently transfected with Fu-GENE 6 reagent (Roche Diagnostics). M2 anti-FLAG affinity agarose beads, isoproterenol, angiotensin peptide, arginine-vasopressin peptide, substance P, and N-ethylmaleimide were from Sigma. FK2 Ub antibody was from Biomol. Ubiquitin antibody Ub-P4D1 was from Santa Cruz Biotechnology. Monoclonal antibody 12CA5 to HA epitope was from Roche Diagnostics. Alexa-594-and Alexa-633-conjugated secondary antibodies were from Molecular Probes. Horseradish peroxidase-conjugated secondary antibodies were from Amersham Biosciences. Detection of active ERK was with a rabbit polyclonal antiphospho-p44/42 MAPK (Cell Signaling Technology; 1:2000 for Western blot and 1:200 for immunostaining). Total ERK was detected with anti-MAPK1/2 (Upstate Technology Inc). [ 125 I](Ϫ)Iodocyanopindolol, 125 I-labeled angiotensin, and 3 H-labeled AVP were purchased from PerkinElmer Life Sciences.
␤-Arrestin2 was cloned in-frame into the pEGFPN1 and pEGFPC2 vectors (Clontech) to yield ␤-arrestin2-GFP and GFP-␤-arrestin2, respectively. Using wild type ␤-arestin2 plasmid as a template, multiple lysine residues were mutated to arginine residues by utilizing the QuikChange multi-site-directed mutagenesis kit (Stratagene). For single site alterations, a QuikChange site-directed mutagenesis kit (Stratagene) was used. All DNA constructs were confirmed by sequencing. HA-␤2AR plasmid was a gift from Dr. Neil Freedman (Duke University), and HA-AT1aR and HA-V2R plasmid were provided by Dr. Marc Caron (Duke University). FLAG-NK1R was from Dr. Thue Schwartz (University of Denmark).
Immunoprecipitation and Immunodetection-To detect ubiquitinated ␤-arrestin2, ␤-arrestin2-FLAG/pCDNA3, or ␤-arrestin2 K11,12R -FLAG/pCDNA3.1 was transiently transfected into COS-7 cells or HEK-293 cells along with the receptor plasmid. To detect active ERK in ␤-arrestin immunoprecipitates, ␤-arrestin2-FLAG or ␤-arrestin2 K11R,K12R -FLAG was co-expressed with RFP-ERK2 along with HA-AT1aR or HA-V2R. Cells were serum-starved for 2 h and then stimulated or not for the indicated times with the appropriate agonists. Cells were solubilized in a lysis buffer containing 50 mM HEPES (pH 7.5), 0.5% Nonidet P-40, 250 mM NaCl, 2 mM EDTA, 10% (v/v) glycerol, 1 mM sodium orthovanadate, 1 mM sodium fluoride, 1 mM phenylmethylsulfonyl fluoride, leupeptin (5 g/ml), aprotinin (5 g/ml), pepstatin A (1 g/ml), benzaminidine (100 M) and 10 mM N-ethylmaleimide. The use of N-ethylmaleimide in lysis buffers in all co-immunoprecipitation procedures is an important technical feature because it stabilizes ubiquitinated species by preventing their deubiquitination. Soluble extracts were mixed with FLAG M2 affinity beads and rotated at 4°C overnight. Nonspecific binding was eliminated by repeated washes with lysis buffer, and bound protein was eluted with sample buffer containing SDS. The proteins were separated on a gradient gel (4 -20%, Invitrogen) and transferred to nitrocellulose membrane for Western blotting. Chemiluminescent detection was performed using SuperSignal West Pico reagent (Pierce). pERK and ␤-arrestin signals were quantified by densitometry with a Fluor-S MultiImager (Bio-Rad). In all experiments, receptor expression levels were concurrently determined by radioligand binding for the ␤2AR, the V2R, and the AT1aR. Surface expression of FLAG-NK1R was measured by flow cytometry.
Confocal Microscopy-HEK-293 cells have a favorable morphology for examining sections of cytoplasm and nucleus simultaneously and hence were used in these experiments. HEK-293 cells on 10-cm dishes were transiently transfected with HA-␤2AR, HA-AT1aR, HA-V2R, or FLAG-NK1R along with the respective ␤-arrestin2-GFP construct. Twenty-four h post-transfection, cells were plated on collagen-coated 35-mm glass-bottom plates. On the following day, cells were starved for at least 2 h in serum-free medium prior to stimulation. After stimulation, cells were fixed with 5% formaldehyde diluted in phosphatebuffered saline containing calcium and magnesium. Fixed cells were permeabilized with 0.01% Triton in phosphate-buffered saline containing 2% bovine serum albumin for 60 min and incubated at room temperature with the appropriate primary antibody. The secondary antibody incubations were done for 1 h followed by repeated washes using phosphate-buffered saline. Confocal images were obtained on a Zeiss LSM-510 laser-scanning microscope using multitrack sequential excitation (488, 568, and 633 nm) and emission (515-540 nm GFP; 585-615 nm, Texas Red; 650 nm, Alexa-633) filter sets.

Angiotensin II-stimulated Endosomal Recruitment of ␤-
Arrestin2 Requires ␤-Arrestin2 Ubiquitination at Lysines 11 and 12-When HEK-293 cells overexpressing the HA-AT1aR and ␤-arrestin2-FLAG were stimulated with 1 M AngII, we could detect an agonist-dependent increase in ␤-arrestin ubiquitination within 1 min (Fig. 1A). As predicted for a Class B receptor, this ubiquitination signal did not decrease after 15 min of agonist, suggesting that ␤-arrestins do not become rapidly deubiquitinated, thus facilitating their association and co-trafficking with the AT1aRs to endocytic vesicles. The pattern of transient ␤-arrestin2 ubiquitination resulting from the activation of the ␤2ARs is shown in Fig. 1B.
␤-Arrestin2 contains 31 lysine residues dispersed through its primary sequence, which could serve as potential ubiquitination sites. To screen for lysines critical for ubiquitin attachment and hence for endosomal recruitment, we created five different ␤-arrestin2 mutants, ␤-arrestin2-A-E ( Fig. 2A) with a GFP tag at the C terminus. These five mutants collectively contained alterations of all the 31 lysines of ␤-arrestin2 primary sequence to arginine residues. We tested the various lysine mutants for their recruitment to the ␤2AR (Class A, Fig. 2B) and the AT1aR, a prototypic Class B receptor (Fig. 2C). Prior to receptor activation, the wild type and mutant ␤-arrestins remained cytosolic and uniformly distributed (data not shown). All of the ␤-arrestin2 mutants (A, B, C, D, and E) were recruited to the plasma membrane as robustly as the wild type upon stimulation of the ␤2AR with 1 M isoproterenol for 30 min (Fig. 2B). Within 5 min of AT1aR stimulation, the WT and all mutants translocated to cell surface receptors (not shown). After 30 min of AT1aR stimulation the WT and mutants A, C, D, and E were recruited to endosomes but ␤-arrestin2-B was not (Fig. 2C). ␤-Arrestin2-B mutant displays a Class A trafficking pattern with the AT1aR (a Class B receptor). This suggests that lysine(s) located within this group of seven residues (lysines 11, 12, 78, 171, 286, 398, and 401) and possibly their modification with ubiquitin can be crucial in mediating the stable association of activated AT1aRs and ␤-arrestin2.
␤-Arrestin2 K11R,K12R Displays Transient Ubiquitination Pattern upon AngII Stimulation-We next tested the agonist-induced ubiquitination of the ␤-arrestin2 K11R,K12R to determine whether the change in the trafficking pattern also affected its ubiquitination kinetics. The AT1aR was co-expressed with ␤-arrestin2 K11R,K12R -FLAG or ␤-arrestin2-FLAG in HEK-293 cells. After stimulation for 0, 1, or 15 min with AngII, ␤-arrestins were immunoprecipitated and analyzed for ubiquitination by Western blotting. As seen in Fig. 4A, ␤-arrestin2 K11R,K12R was ubiquitinated within a minute of AngII stimulation, but the signal decreased to basal levels at 15 min of AngII stimulation. In contrast, the ubiquitination of ␤-arrestin2 WT remained sustained at 15 min of AngII treatment. Thus, ␤-arrestin2 K11R,K12R displays a transient ubiquitination pattern with AT1aR stimulation similar to ubiquitination induced by the ␤2AR, a prototypic Class A receptor that does not recruit ␤-arrestin2 to endosomes ( Fig. 1B and Refs. 17 and 8). These data suggest that AT1aR stimulation causes transient ubiquitination of as yet undesignated lysines on ␤-arrestin2 as well as sustained ubiquitination on lysines 11 and/or 12.
␤-Arrestin2 K11R,K12R Is Impaired in pERK Scaffolding upon AngII Stimulation-Two independent pathways (G protein-and ␤-arrestin2-mediated) lead to ERK activation upon AT1aR stimulation (19). Additionally in COS-7 cells, AT1aR stimulation markedly increases the phosphorylation of the ␤-arrestin-scaffolded GFP-ERK2 (15,20). To determine whether the changes in ␤-arrestin ubiquitination pattern and trafficking correlate with AngII-stimulated pERK scaffolding activity of ␤-arrestin, we isolated ␤-arrestin immunoprecipitates and determined the amount of ␤-arrestin-bound pERK. HA-AT1aR was expressed in COS-7 show membrane-recruited ␤-arrestin2. All other mutants displayed endosomal distribution of ␤-arrestin2. Confocal microscopy was performed as described in the legend for Fig. 2. B, the translocation patterns of ␤-arrestin2-GFP carrying point mutation at position 11 or 12 after a 30-min stimulation of AT1aR are portrayed. As a comparison the membrane distribution of ␤-arrestin2 K11R,K12R is also shown. These data are representative of similar results from four separate experiments. cells along with RFP-ERK2 and either the ␤-arrestin2-FLAG or ␤-arrestin2 K11R,K12R -FLAG; cells were stimulated with 100 nM AngII for 5 min, and FLAG immunoprecipitates were isolated and probed for the pERK content. ␤-Arrestin2 K11R,K12R immunoprecipitates contained significantly less pERK, averaging only 45% of the levels obtained with ␤-arrestin2 WT (Fig. 4, B and C). This indicates that decreasing the stability of ␤-arrestin ubiquitination can result in a corresponding reduction in the amount of ␤-arrestin-scaffolded pERK.
Stable ␤-Arrestin2 Ubiquitination Is a Prerequisite for Formation of Stable Receptor Signalosomes-We utilized immunostaining and confocal microscopy to determine whether sustained ␤-arrestin2 ubiquitination plays any role in the subcellular distribution of pERK stimulated by AngII. We transfected HEK-293 cells with either ␤-arrestin2-GFP (Fig.  5A) or ␤-arrestin2 K11R,K12R -GFP (Fig. 5B) along with HA-AT1aR and detected the presence of endogenous pERK by utilizing an antibody that specifically reacts to the activated forms of ERK1/2 (Thr-202/Tyr-204-phosphorylated ERK1/2). As shown in Fig. 5A, quiescent cells did not contain active ERK, with receptors occupying the membrane (blue), and ␤-arrestin2 was evenly distributed in the cytosol (green). A 5-min stimulation with AngII led to the recruitment of ␤-arrestin2 to the membrane and robust activation of ERK in the membrane and cytosol with comparatively lesser ERK activity in the nucleus (Fig. 5A, middle row). Longer stimulation (30 min) resulted in the redistribution of receptor, pERK, and ␤-arrestin to endosomal vesicles (Fig. 5A, bottom row). In the presence of ␤-arrestin2 K11R,K12R , we could detect robust ERK phosphorylation at the plasma membrane as well as in the cytosol and the nucleus after a 5-min stimulation (Fig. 5B, 2nd row), but the ERK activation was transient and no signaling endosomal complexes were detectable after 30 min of agonist treatment (Fig.  5B, 3rd row). These experiments suggest that stable ubiquitination of ␤-arrestin, besides being necessary for stable binding to receptors, is also pivotal for the formation of stable signaling receptor complexes and the targeting of sustained ERK activity to the endosomes.
Fusion of the Ub Moiety to ␤-Arrestin2 K11R,K12R Restores AngII-dependent Trafficking and Signaling Properties-To test whether the endocytic and signaling deficiency seen in ␤-arrestin2 K11R,K12R is because of its inability to be stably ubiquitinated, we created ␤-arrestin2 K11R,K12R with C-terminally fused ubiquitin, which mimicked a stably ubiquitinated form and determined the AngII-stimulated trafficking and signaling. Because fusion of ubiquitin is practical only at the C terminus for mammalian expression, we created N-terminal GFP fusions and compared ␤-arrestin2, ␤-arrestin2 K11R,K12R , and ␤-arrestin2 K11R,K12R -Ub K48R (henceforth designated as ␤-arrestin2 K11R,K12R -Ub) for AngII-stimulated trafficking. The lysine 48 in ubiquitin was altered to arginine to prevent Lys-48 modification within ubiquitin, which has been reported to efficiently target most proteins for degradation via the proteasomes (21). As portrayed in Fig. 6A, fusion of Ub restores the endosomal recruitment for ␤-arrestin2 K11R,K12R upon stimulation of the AT1aR for 30 min. Furthermore, when we analyzed the AngII-stimulated pERK distribution with the co-expression of ␤-arrestin2 K11R,K12R -Ub, we found a very robust activation of ERK, which was stabilized on endosomes containing both the AT1aR and ␤-arrestin2 K11R,K12R -Ub (Fig. 6B, bottom panels). These experiments clearly demonstrate that stable ubiquitination of ␤-arrestin2 leads to AngII-stimulated endosomal recruitment of ␤-arrestin while facilitating the sustained MAPK activation on endosomes.
Alteration of ␤-Arrestin Ubiquitination Status Leads to Changes in Immediate Early Gene Expression-Translation of MAPK activation to a specific cell response (transformation, growth, differentiation, etc.) depends on the duration of coordinated activities of the induced immediate early genes including egr-1, c-fos, and c-jun (22)(23)(24). For example, AT1aR stimulation can cause an increase in egr-1 mRNA levels in human airway smooth muscle cells under conditions mimicking the hypertrophic stress response (25). Additionally, AngII-stimulated EGR-1 expression in vascular smooth muscle cells and Chinese hamster ovary cells requires nuclear ERK activation (26,27). Our studies indicate that AT1aR-induced stable ␤-ar-restin2 ubiquitination is specifically responsible for the endosomal co-localization of ␤-arrestin2 and pERK as well as for the prevention of pERK translocation to the nucleus. To determine whether the spatial and temporal changes induced by the ␤-arrestin2 K11R,K12R on AngII-mediated ERK activity affect immediate early gene expression, we compared the levels of EGR1 induced by AngII stimulation to those obtained upon expression of WT ␤-arrestin or ␤-arrestin2 K11R,K12R -Ub. Stim- The immunoprecipitate was also probed for the total ERK content (2nd panel). The expression levels of pERK, ERK, and ␤-arrestin in the lysates as detected by the respective antibodies are displayed in the lower three panels. C, the bar graph presented is a quantification of the amount of pERK normalized to total ERK and ␤-arrestin levels from three independent experiments. *, less than WT; p ϭ 0.04, paired t test.
ulation of the AT1aR in HEK-293 cells expressed without ␤-arrestin2 or with ␤-arrestin2 or ␤-arrestin2 K11R,K12R -Ub led to a modest increase in EGR-1 at 30 min, which did not change up to 2.5 h (Fig. 7). In stark contrast, transfection of ␤-arrestin2 K11,12R led to robust EGR-1 expression at 30 min. EGR-1 level remained significantly higher even 2.5 h after AngII stimulation. These differences appeared to reflect the varying abilities of the different ␤-arrestin constructs to retain pERK in the cytosol, thus preventing nuclear translocation and EGR induction. Taken together, these data suggest that altering the pattern of ␤-arrestin ubiquitination and trafficking can lead to significant changes in terms of transcriptional patterns and gene expression stimulated by the 7TMR.
Lysines 11 and 12 in ␤-Arrestin2 Are Specific Ubiquitination Sites for the AT1aR-To determine whether lysines 11 and 12 on ␤-arrestin2 are the invariant sites of modification required for promoting endosomal recruitment, we tested the trafficking of the mutant ␤-arrestin2 K11R,K12R with two representative Class B receptors, V2R and NK1R. Both V2R and NK1R can stably bind ␤-arrestin2 similar to the AT1aR (Fig. 8A, lower panels). Quite surprisingly, ␤-arrestin2 K11R,K12R -GFP was impaired in endosomal recruitment only to the AT1aR but not to either the V2R or the NK1R (Fig. 8A, upper panels). The Nterminally tagged GFP-␤-arrestin2 K11R,K12R also showed endosomal recruitment upon activation of either the V2R or the NK1R (data not shown). When ␤-arrestin2 K11R,K12R was tested for its ubiquitination time course upon V2R stimulation, we found that it exhibited stable ubiquitination kinetics matching that of the WT ␤-arrestin2 (Fig. 8B). Moreover, ␤-arrestin2 K11R,K12R bound pERK with similar efficacy as the WT ␤-arrestin upon activation of V2Rs (Fig. 8, C and D). Confocal images presented in Fig. 8E demonstrate that both WT ␤-arrestin2 and ␤-arrestin2 K11,12R co-internalized with activated V2Rs and harbored pERK on endosomes. Qualitatively similar endosomal pERK staining was observed with co-expression of GFP-␤-arrestin2 K11R,K12R when NK1R was stimulated (data not shown). The levels and the time course of EGR-1 induction by NK1R and V2R were very similar upon co-expression of either ␤-arrestin2 or ␤-arrestin2 K11R,K12R (not shown).
AVP-stimulated Endosomal Recruitment of ␤-Arrestin2 Is Regulated by a Distinct Group of Lysines-The data presented above strongly suggest a correlation of angiotensin receptorinduced trafficking and signaling roles of ␤-arrestin with ubiquitination at specific sites. We sought to determine whether a similar correlation could be identified for other Class B receptors such as the V2R. It is possible that receptors other than the AT1aR are not linked with ubiquitination on specific lysines on ␤-arrestin but can lead to modification of any available lysine(s). To address this issue we determined the trafficking patterns of the various ␤-arrestin2 lysine mutants ( Fig. 2A) as FIG. 5. Stable ␤-arrestin ubiquitination leads to persistent signaling on endosomes. HEK-293 cells transiently expressing the HA-AT1aR along with ␤-arrestin2-GFP (A) or ␤-arrestin2 K11R,K12R (B) were starved in serum-free media for at least 2 h and stimulated with 1 M AngII at 37°C for the indicated periods of time. Fixed and permeabilized cells were then incubated with pERK polyclonal antibody followed by secondary Alexa-633-conjugated anti-rabbit IgG. This was followed by treatment with 12CA5 monoclonal antibody, which recognizes the HA epitope on the receptor and Alexa-594-conjugated anti-mouse IgG. Fluorescent confocal images were obtained on a Zeiss LSM-510 microscope using multitrack sequential excitation (488, 568, 633 nm) and emission filter sets at 515-540 nm for detecting GFP (␤-arrestin, green), at 585-615 for Alexa-594 (receptor, blue), and at 650 nm for the Alexa-633 (pERK, red). These experiments were reproduced four times with identical results. NS, nonstimulated. induced by the V2R. HEK-293 cells expressing the V2R and either ␤-arrestin2 WT or a ␤-arrestin2 mutant were stimulated with 1 M AVP for 30 min, and the translocation of ␤-arrestin2-GFP was examined by confocal microscopy. As depicted in the confocal images in Fig. 9, ␤-arrestin2-A, -B, -D, and -E mutants displayed endosomal distribution identical to ␤-arrestin2 WT. Unlike its endosomal localization with the AT1aR (Fig. 2C), the ␤-arrestin2-C mutant, however, displayed only membrane translocation after V2R stimulation. In contrast to the situation with the AT1aR and the individual mutants within Group B lysines, mutations within the Group C lysines (i.e. ␤-arrestin2 K18R , ␤-arrestin2 K107R,K108R , ␤-arrestin2 K207R , ␤-arrestin2 K296R ) did not lead to ablation of AVP-stimulated endosomal trafficking of ␤-arrestin2 (data not shown), indicating that two or more lysines in this group may be ubiquitinated upon V2R stimulation. These results imply that agonist stimulation of different 7TMRs can lead to distinct ␤-arrestin2 conformations, thus exposing specific lysine residues for ubiquitination. Thus, ␤-arrestin ubiquitination at lysines 11 and 12 is specific for a conformation of ␤-arrestin associated with activated AT1aRs. Activation of another stable ␤-arrestin binder, V2R, not only does not induce ubiquitination at the same lysines but also appears to target other distinct sites. Our results suggest that conformational changes in ␤-arrestin2 induced by V2R or NK1R interaction are different from those due to AT1aR binding. DISCUSSION In this study, we have demonstrated that stable association of ␤-arrestin2 with Class B receptors is possible only when specific lysines are ubiquitinated, e.g. lysines 11, and 12 for the AT1aR. Sustained ␤-arrestin ubiquitination is required for its co-trafficking with activated receptors and for the generation of stable compartmentalized pERK signals on endosomes. Lysines 11 and 12, however, are not crucial for stable ␤-arrestin2 ubiquitination mediated by either the V2R or the NK1R, although these receptors, like the AT1aR, respond to peptide agonists by co-trafficking with WT ␤-arrestin2 to endosomes. Thus, ubiquitination of ␤-ar-restin2 not only functions as a receptor-activated molecular switch that regulates the assembly and targeting of 7TMR signalosomes but also displays conformational specificity with respect to different 7TMR partners.
The formation of stable AT1aR-␤-arrestin complexes promotes the compartmentalization and stabilization of ␤-arrestin-scaffolded pERK, prevents nuclear translocation of pERK, and suppresses transcriptional activation (15). The AngII-induced cytosolic and endosomal ERK pools appear to be insensitive to PKC inhibitors and independent of G protein activation, but are completely sensitive to ␤-arrestin2 small interfering RNA (19). Hence, a ␤-arrestin2-dependent, G protein-independent pathway mediates the persistent cytosolic and endosomal pERK activation resulting after AT1aR stimulation. The data presented in this study further indicate that the AT1aR-mediated, ␤-arrestin-dependent ERK activation is elicited only when lysines 11 and 12 in ␤-arrestin2 can be ubiquitinated upon receptor stimulation.
Tohgo et al. (16) demonstrated that the plasma membrane recruitment pattern evoked by stimulation of Class A receptors such as the ␤2AR or the ␣ 1bR leads to weak binding between ␤-arrestin2 and pERK, whereas the endosomal pattern of recruitment displayed by Class B receptors such as the AT1aR or the V2R leads to stronger complex formation of ␤-arrestin2 and pERK (16). Mutation of lysines 11 and 12 leads to reversal of the pattern of ␤-arrestin2 trafficking from endosomal (Class B) to plasma membrane (Class A) with respect to the AT1aR. After 5 min of AngII stimulation, both the WT and ␤-arrestin2 K11R,K12R translocate to the cell surface where they co-localize with the activated AT1aR and initiate the formation of active MAPK complexes (Fig. 5). Theoretically, both ␤-arres-tin2 and ␤-arrestin2 K11R,K12R should have comparable pERK bound at 5 min of AngII stimulation. However, the ␤-arrestin2 K11R,K12R immunoprecipitates isolated at 5 min of AngII stimulation contain significantly less pERK, suggesting that the stability of the MAPK scaffolds with respect to WT and ␤-arrestin2 K11R,K12R are different (Fig. 4, B and C). Because ␤-arrestin2 K11R,K12R can be transiently ubiquitinated and behaves as a Class A binder with the Class B AT1aR, it dissociates from the internalizing receptor, most likely because of deubiquitination events (18). This leads to the disassembly of receptor-␤-arrestin signaling complexes. This suggests that the differential binding of pERK to ␤-arrestin with respect to Class A and Class B receptors is because of the specific ubiquitination status of ␤-arrestin2 correlated with the receptor-induced conformational change in ␤-arrestin2.
Sustained ␤-arrestin ubiquitination appears to regulate the movement of endocytic cargo through the cellular microcompartments and in the process to link and regulate receptor endocytosis and ERK activation. Such segregation of ERK activity in the endosomal domains can have at least two effects: 1) depletion of nuclear pERK and attenuation of specific transcriptional responses (e.g. via EGR-1); and 2) phosphorylation of specific cytosolic proteins that may initiate alternate cellular responses such as apoptosis, chemotaxis, and cytoskeletal changes or in turn relay signals to the nucleus by phosphorylating other nucleocytoplasmic shuttling kinases (e.g. p90Rsk). Several cytosolic ERK substrates have been identified and include cytoskeletal and microtubule-associated proteins, proteins of the apoptotic pathway such as Bcl-2 (28), and proteins in signal transduction pathways such as G protein-coupled receptor kinase 2 (GRK2) (29,30), ␤-arrestin1 (31), phospholipase A2 (32), Raf-1 (33), MEK1 (34), phosphodiesterase 4D (35), etc. The exact nature of substrates for the pERK compartmentalized with Class B receptors remains to be elucidated. However, by virtue of its key functions in stabilizing ␤-arrestinscaffolded pERK, ubiquitination of ␤-arrestin2 likely plays an important role in the as yet unidentified mechanisms that protect the endosomal pool of pERK from phosphatases.
Our data suggest that by stabilizing pERK in the endosomal compartments (␤-arrestin-dependent pathway), ␤-arrestin ubiquitination suppresses the expression of EGR-1 and nuclear transcription. In contrast, receptor stimulation of the G proteindependent pathway leads to nuclear translocation of pERK and initiation of transcription. Apparently, in the absence of stable ubiquitination the ␤-arrestin-dependent pathway is not initiated. Moreover, the weak binding of ␤-arrestin2 K11R,K12R to the AT1aR appears to be inefficient in suppressing the G proteinmediated pathway (induction of EGR-1, Fig. 7). It is thus possible that impairment of stable ␤-arrestin2 ubiquitination due to the malfunctioning of the cellular ubiquitination machinery could derange the transcriptional events occurring upon receptor activation and result in unwarranted growth and differentiation or hypertrophy.
␤-arrestin2 appears to be polyubiquitinated according to the SDS-PAGE profile observed in our experiments. ␤-Arrestin ubiquitination as induced by various 7TMRs (␤2AR, V2R, and AT1aR) is detectable by the FK1 ubiquitin antibody that preferentially detects polyubiquitinated proteins (data not shown). Currently our Western blot assays do not rule out the possibility of monoubiquitination versus polyubiquitination at lysines 11 and 12. Nevertheless, ubiquitination of ␤-arrestin2 at lysines 11 and 12 serves to promote its efficient functioning as a 7TMR-regulated signaling scaffold upon AT1aR stimulation. It is thus tempting to speculate that ␤-arrestin ubiquitination serves as a nexus that links receptor activation to the initiation of the MAPK cascade. These findings further indicate that ubiquitination of ␤-arrestin2 empowers not only its endocytic function but its ability to scaffold and activate ERK1/2 as well. In analogy, interleukin-1 receptor-stimulated polyubiquitination of the adaptor protein TRAF6 (tumor necrosis factor receptor-associated factor 6) activates TAK (transforming growth factor-␤-activated kinase), the upstream kinase in the NF-B activation pathway (36). Thus, ubiquitination, originally discovered in the context of degradation of cellular proteins (37,38), is now emerging as an important mechanism for regulating cell signaling as well.
The lysines 11 and 12 in ␤-arrestin2 identified as AngIIdirected ubiquitination sites are conserved among the arrestin family members (visual and nonvisual arrestins). Structural details of bovine arrestin predict lysine 15 (lysine 12 of rat ␤-arrestin2) to be in contact with phosphorylated rhodopsin residues (39). Additionally lysines 11 and 12 (because of their positive charge) are reported to function as phosphate sensors for recognition and guidance of phosphorylated domains of 7TMRs to reach the polar core residues of ␤-arrestin (40). Our mutagenesis protocol (lysine to arginine) does not cause a charge shift and is therefore unlikely to perturb the ␤-arrestin phosphate-sensing role. Perhaps these lysines play a general role in phosphate sensing while remaining as specific sites for ubiquitination upon AngII stimulation.
In contrast to only two isoforms of nonvisual arrestins, ␤-ar-restin1 and ␤-arrestin2, hundreds of potential 7TMR partners exist in the human genome. Even after the characterization of a small sector of the 7TM superfamily, as done here, it is apparent that the 7TM adaptor role of ␤-arrestin is not generic but rather is highly specialized, displaying a "custom-fit" role in regulating different aspects of receptor biology. The divergent roles played by ␤-arrestins are only beginning to be appreciated. For example, PAR1 receptors internalize via ␤-arrestin independent pathways but require ␤-arrestin for receptor desensitization (41). In contrast, ␤-arrestin is indispensable for desensitization as well as internalization for the prototypic ␤2AR (42). N-Formyl peptide receptor recycling but not internalization requires the adaptor role of ␤-arrestins (43). For CXCR4-induced chemotaxis of HEK-293 cells, ␤-arrestin2dependent activation of p38 MAPK pathway is employed (44). On the other hand, PAR2-activated actin cytoskeletal changes, pseudopodial extension, and chemotaxis of NIH3T3 cells requires ␤-arrestin-dependent ERK activation (45).
The mechanism of arrestin-receptor interaction is thought to involve the guidance of phosphate moieties on the receptor (mostly on the C-tail) to reach the three-element polar core of arrestin that holds the arrestin molecule in a basal state (46). Upon receptor interaction and the disruption of the polar core, "activating" conformational changes occur in ␤-arrestin, which poise it to carry out its downstream roles such as movement of the receptor complex to clathrin-coated vesicles, engaging of relevant MAPK cascades, etc. If this scenario is universal, then all 7TMRs should evoke identical cellular responses through a ␤-arrestin-dependent mechanism. However, this is not the case. Our data suggest that there may be multiple receptor-induced ␤-arrestin conformations. Thus, activation of ␤-arrestin has the potential to lead to many diverse cellular responses such as mitogenesis, apoptosis, chemotaxis, or morphological changes. Our data suggest that even similarly grouped receptors such as the Class B V2R, AT1aR, and NK1R, all of which induce stable ␤-arrestin ubiquitination, nonetheless target different residues on ␤-ar-restin2. The functional activities of ␤-arrestins presumably correspond to conformational changes induced by the active state of its receptor partner. Such dynamic conformational changes may or may not expose specific lysine residues of ␤-arrestin, thus making them available for ubiquitination and/or deubiquitination. A variety of such "actively modified" ␤-arrestin molecules might thereby be created, especially given the large number of combinations of lysines available in both ␤-arrestins (35 in ␤-arrestin1 and 31 in ␤-arrestin2).
Receptor-specific ubiquitination of ␤-arrestin could thus serve as a platform hosting the differential display of a variety of activated ␤-arrestin conformations with different functional outcomes.