Resistance of the Human (cid:1) 1 -Adrenergic Receptor to Agonist-induced Ubiquitination A MECHANISM FOR IMPAIRED RECEPTOR DEGRADATION*

Down-regulation is a classic response of most G pro-tein-coupled receptors to prolonged agonist stimulation. We recently showed that when expressed in baby hamster kidney cells, the human (cid:1) 1 - but not the (cid:1) 2 - adrenergic receptor (AR) is totally resistant to agonist-mediated down-regulation, whereas both have similar rates of basal degradation (Liang, W., Austin, S., Ho-ang, Q., and Fishman, P. H. (2003) J. Biol. Chem. 278, 39773–39781). To identify the underlying mechanism(s) for this resistance, we investigated the role of proteasomes, lysosomes, and ubiquitination in the degradation of (cid:1) 1 AR expressed in baby hamster kidney and human embryonic kidney 293 cells. Both lysosomal and proteasomal inhibitors reduced (cid:1) 1 AR degradation in agonist-stimulated cells but were less effective on basal degradation. To determine whether (cid:1) 1 AR traf- ficked to lysosomes we used confocal fluorescence microscopy. We observed some colocalization of (cid:1) 1 AR and lysosomal markers in agonist-treated cells but much less than that of (cid:1) 2 AR even in cells co-trans- fected with arrestin-2, which increases (cid:1) 1 AR internalization. Ubiquitination of (cid:1) 2 AR readily occurred in agonist-stimulated cells, whereas ubiquitination of (cid:1)

G protein-coupled receptors (GPCRs) 1 upon activation by agonists initiate a variety of signaling cascades that modulate cell function and metabolism (1). For most GPCRs, activation is followed by desensitization and internalization of the receptors, which then are targeted to either recycling or degradation pathways (2,3). Several mechanisms for GPCR internalization have been described, the clathrin-coated pit endocytic pathway being the best understood. GPCR kinase-catalyzed phosphorylation of the agonist-occupied receptors promotes arrestin binding to the receptors. Arrestins also interact with clathrin and AP-2 adapter proteins to form complexes that target the receptors into the coated pits (4,5). Based on differences in the stability and trafficking of arrestin-receptor complexes, GPCRs are separated into two classes (6,7). Class A such as ␤ 2 AR, ␣ 1b AR and opioid receptors bind to arrestin-3 with higher affinity than arrestin-2. The interaction is transient, and upon endocytosis, the arrestin dissociates and returns to the cytosol, whereas the receptor undergoes rapid recycling. Class B such as vasopressin V2 and angiotensin AT1a receptors bind with equal affinity to arrestin-2 and -3 to form stable complexes that internalize together. The receptors recycle more slowly and are degraded more rapidly.
Recent studies indicate the involvement of ubiquitination in the internalization and degradation of GPCRs (for reviews, see Refs. 8 and 9). Conjugation of ubiquitin, a 76-amino acid polypeptide, to lysine residues of endoplasmic reticulum-retained proteins initially was found to function as a signal for their degradation by proteasomes (for a review, see Ref. 10). Subsequently, ubiquitination was shown to regulate the endocytic trafficking of several plasma membrane receptors by sorting them to lysosomes for degradation (11). Some mammalian GPCRs that are ubiquitinated include ␤ 2 AR (12), the chemokine CXCR4 receptor (13), and ␦ opioid receptors (Ref. 14, but see Ref. 15), and the vasopressin V2 receptor (16). Eliminating receptor ubiquitination reduces agonist-mediated receptor degradation but not internalization. Thus, ubiquitination appears to function as a sorting signal for targeting the receptors for degradation in lysosomes.
Although the ubiquitination of proteins is a prerequisite for their degradation by proteasomes, the role of the latter in the degradation of GPCRs is unclear. Agonist-promoted degradation of CXCR4 receptors is blocked by inhibitors of lysosomal but not proteasomal function, and the receptors are sorted to lysosomes (13). Proteasomal inhibitors are reported to either reduce (12) or have no effect (17) on agonist-mediated internalization and degradation of ␤ 2 AR. In one study proteasomal but not lysosomal inhibitors reduced the down-regulation of and ␦ opioid receptors (14), whereas in another, lactacystin, a very specific proteasomal inhibitor, did not impair agonist-induced internalization, lysosomal targeting, and degradation of the ␦ opioid receptor (15). Some of these studies found that steadystate receptor levels increase in unstimulated cells treated with proteasomal inhibitors (14,17). The presence of proteasomal inhibitors also enhances the ubiquitination of some GPCRs (14,16).
␤ 1 AR, one of the three ␤AR subtypes, although widely distributed in various tissues, is the predominant subtype in heart and certain brain regions. ␤ 1 AR has a major role in regulating cardiac output in response to norepinephrine and epinephrine (18), and ␤ 1 AR in synaptic junctions mediates the effects of noradrenergic stimulation on long term potentiation (19). Because ␤ 1 AR is associated with human diseases such as congestive heart failure (18) and depression (20), the regulation of ␤ 1 AR is of considerable interest. In comparison to the human ␤ 2 AR, the human ␤ 1 AR is more resistant to agonist-mediated desensitization (21,22), internalization (21,(23)(24)(25)(26)(27), and downregulation (21,23,26,28,29). We have previously shown that in agonist-stimulated BHK cells, ␤ 1 AR is up-regulated due to its resistance to degradation and its increased synthesis (29). We recently demonstrated that ␤ 1 AR expressed in BHK and HEK 293 cells undergoes agonist-mediated endocytosis by the clathrin-coated pit pathway but traffics to an endosomal compartment distinct from that of ␤ 2 AR (27). Thus, differences in subtype degradation may be due to differences in subtype trafficking. Because ubiquitin serves as a sorting signal for lysosomal degradation of receptors, we investigated its role in the resistance of ␤ 1 AR to degradation. We found that in contrast to ␤ 2 AR, agonist-mediated ubiquitination of ␤ 1 AR was not detected. In addition, when we used chimeric receptors in which the C-tails of the two subtypes were exchanged, we observed ubiquitination of the ␤ 1 /␤ 2 ct-AR but not of the ␤ 2 /␤ 1 ct-AR chimera. The inability of the C-tail of ␤ 1 AR to facilitate receptor ubiquitination may be the basis for its insensitivity to agonistmediated degradation.
Cell Culture and Transfection-BHK (clone tk Ϫ ts13) and HEK 293 cells were obtained from the American Type Culture Collection and grown in Dulbecco's modified Eagle's medium with 10% fetal bovine serum. Clonal lines stably expressing either ␤ 1 AR or ␤ 2 AR or co-expressing both arrestin-2 and either subtype were described previously (27,29). To obtain cells expressing wild type or chimeric HA-␤AR or ␤AR-GFP or co-expressing one of these receptors and either arrestin-2 or -3, transfections were done with the appropriate plasmid(s) using LipofectAMINE Plus. The cells were either used after 24 h or selected with G418 to obtain resistant, uncloned cultures that expressed fairly constant ␤AR levels during the experimental period. All stably transfected cells were maintained in medium containing 0.2 mg/ml G418. HEK 293 cells used for experiments were grown on polylysine-coated plastic ware or glass coverslips.
␤AR Binding Assays-Control and treated cells were assayed for total and surface receptors as described previously (30). Briefly, cells grown in 6-well plates were incubated with 10 M ISO at 37°C for up to 24 h, rinsed with ice-cold Ca 2ϩ -and Mg 2ϩ -free DPBS, and lysed in buffer containing protease inhibitors at 4°C. Portions of the lysates were assayed for protein and for total ␤AR binding activity with 250 pM [ 125 I]iodocyanopindolol for 1 h at 37°C. For surface ␤AR, cells grown in 24-well plates were exposed at 37°C to 1 M ISO for 30 min. The plates were placed on a bed of ice, and the cells were rapidly washed twice with ice-cold DPBS and exposed to 5 nM [ 3 H]CGP-12177 at 4°C for 1 h. Finally, the cells were washed as above and assayed for 3 H and protein. For both assays 10 M propranolol was used to define nonspecific binding.
␤AR Degradation Assay-Basal and agonist-mediated ␤AR turnover were determined by the loss of surface-biotinylated receptors as previously described (28,29). Briefly, cells grown in 6-well plates were rinsed with ice-cold buffer B (40 mM NaHCO 3 /100 mM NaCl, pH 8.6) incubated with 1 mM sulfo-NHS-LC-biotin in buffer B at 4°C for 30 min and rinsed with ice-cold 20 mM glutamine in DPBS. Some of the cells were lysed as described above (control), and the rest of the cells were warmed to 37°C in fresh medium in the absence or presence of 10 M ISO for 4 h. When inhibitors were used, the cells were preincubated with the inhibitors for 30 min and then exposed to ISO. The lysates were extracted in RIPA buffer for 1 h at 4°C and centrifuged at 17,000 ϫ g for 20 min. ␤ARs were immunoprecipitated from the soluble extracts using protein A-Sepharose beads coated with anti-␤ 1 AR or -␤ 2 AR antibodies and eluted from the beads by heating in SDS sample buffer at 60°C for 15 min. The eluted proteins were separated by SDS-PAGE, transferred to Immobilon-P (Millipore), and probed with horseradish peroxidaseconjugated streptavidin (1:10,000). The biotinylated proteins were visualized by enhanced chemiluminescence and, after exposure to Bio-Max MR or Lite 2 film (Eastman Kodak Co.), quantified by scanning densitometry and NIH Image software, version 1.62.
Detection of Ubiquitination of ␤ARs-Cells expressing wt-or chimeric ␤ARs grown in 6-well plates were exposed to 10 M ISO for up to 24 h, washed, and lysed as described above except the buffer included 10 mM NEM to inhibit deubiquitination enzymes. After extraction in RIPA buffer, immunoprecipitation, and SDS-PAGE, the receptors were transferred to Immobilon-P and probed with mouse monoclonal antiubiquitin (1:2,000) and horseradish peroxidase-conjugated goat antimouse (1:10,000) antibodies. In addition, a similar blot was probed with anti-␤ 1 AR or -␤ 2 AR and horseradish peroxidase-conjugated goat antirabbit antibodies (all 1:10,000).
Confocal Fluorescence Microscopy-Previous methods with minor modifications were used to visualize cells with a Zeiss LSM 510 laser confocal microscope (27). Briefly, cells stably expressing HA-␤ 1 AR were stained at 4°C with Alexa Fluor 488-conjugated anti-HA antibody (1 g/ml), washed with DPBS, and incubated at 37°C in medium with or without 10 M ISO for 2 h. Cells stably expressing ␤AR-GFP or coexpressing ␤AR-GFP and arrestin-2 were incubated at 37°C in medium with 10 M ISO for the indicated times in the figure legends. Lyso-Tracker® Red (75 nM) was added to the medium 30 min before the end of the ISO treatment. Cells then were washed and fixed with 4% paraformaldehyde in phosphate-buffered saline. For antibody staining, the cells were permeabilized with 0.2% Triton X-100 in DPBS, blocked with 3% dry milk and 0.05% Triton X-100 in DPBS, and incubated with anti-LAMP-1 (10 g/ml) and then Cy3-conjugated anti-mouse antibody (3 g/ml). The coverslips were mounted on slides with ProLong, and the cells were scanned using the confocal microscope. The images were processed using Adobe Photoshop 5.5 or 6.0.
Data Analysis-Unless otherwise indicated, each experiment was repeated at least three times, and each data point within an experiment was done in triplicate. Data were fitted to curves by nonlinear regression analysis and analyzed for statistical significance by two-way analysis of variance or two-tailed t test using Prism 3 (GraphPad Software).

␤ 1 AR Resistance to Agonist-mediated Down-regulation in HEK 293
Cells-We recently showed that human ␤ 1 AR expressed in BHK cells is up-regulated by persistent agonist stimulation due to a cAMP-mediated increase in ␤ 1 AR mRNA levels and the resistance of ␤ 1 AR to agonist-mediated degradation (29). To further investigate the regulation of human ␤ 1 AR, we used the HEK 293 cell line, which is well characterized and serves as a model for studying the function and regulation of various human proteins. HEK 293 cells stably expressing either ␤ 1 AR or ␤ 2 AR were exposed to 10 M ISO at 37°C for up to 24 h and then assayed for total ␤AR binding activity using [ 125 I]iodocyanopindolol. The number of ␤ 1 AR binding sites remained unchanged, whereas that of ␤ 2 AR decreased to 64% that of control with a t1 ⁄2 of 6.8 h (Fig. 1A). We have shown that overexpressing arrestin-2 and -3 increases agonist-mediated internalization of ␤ 1 AR in BHK and HEK 293 cells (27). Enhancement of ␤ 1 AR internalization by arrestin-2, however, does not lead to down-regulation of ␤ 1 AR in BHK cells (29). Several studies indicated that arrestin-3 has a larger role than arrestin-2 in the regulation of ␤ 2 AR (7,12). Therefore, we investigated the effects of arrestin-3 on prolonged agonist-mediated regulation of ␤ 1 AR in BHK and HEK 293 cells. As shown in Fig. 1B, total ␤ 1 AR binding activity was not reduced by persistent agonist treatment regardless of arrestin-3 co-expression. Thus, although both arrestins facilitate the internalization of ␤ 1 AR, neither has an effect on its degradation.
Degradation or turnover of the receptor proteins in HEK 293 cells was determined by measuring the disappearance of biotinylated cell surface ␤AR using immunoprecipitation and streptavidin overlay as described under "Experimental Procedures." In the absence of agonist, the basal degradation of both ␤-subtypes was similar, with ϳ40% of the biotin-labeled receptors being lost in 4 h (Fig. 2). In cells treated with agonist for 4 h, degradation of ␤ 2 AR increased more than that of ␤ 1 AR. Based on these results, ␤ 1 AR expressed in HEK 293 cells is resistant to agonist-mediated down-regulation as was observed in BHK cells (27). Thus, the regulation of human ␤ 1 AR is similar in both cell lines and is significantly different from that of human ␤ 2 AR.
Effects of Lysosomal and Proteasomal Inhibitors on Degradation of ␤ 1 AR-To determine whether proteasomes and lysosomes, two major sites for protein degradation, are involved in ␤ 1 AR degradation, we used a series of known inhibitors. Chloroquine and NH 4 Cl inhibit protease activity in lysosomes by neutralizing the acidic pH (31). E-64 and leupeptin inhibit lysosomal cysteine proteases, and leupeptin also inhibits serine proteases (32). ALLN, MG132, and lactacystin block proteasomal activity, with lactacystin having the greatest specificity (17). Degradation of ␤ 1 AR expressed in BHK and HEK 293 cells was determined using the surface biotinylation procedure described above. The biotinylated cells were pretreated with the inhibitors and then incubated in the presence or absence of ISO for 4 h. As shown in Table I, agonist treatment only had a significant effect on HEK 293 cells not exposed to inhibitor. All the inhibitors reduced the basal ␤ 1 AR degradation in BHK cells, but only chloroquine, E-64, and ALLN reached statistical significance, with the effect of ALLN being substantial. ALLN, followed by MG132 and chloroquine, significantly lowered the agonist-mediated degradation. The degradation of ␤ 1 AR in agonist-stimulated HEK 293 cells was significantly inhibited by NH 4 Cl, E-64, and lactacystin, but only the latter had an effect on basal degradation. These results suggest that ␤ 1 AR degradation is more sensitive to proteasomal and lysosomal inhibitors in agonist-stimulated cells. They further suggest that basal receptor turnover may involve some other mechanism such as plasma membrane degradation (17). None of the inhibitors totally blocked ␤ 1 AR degradation at the concentrations used. It should be noted, however, that higher concentrations often led to cell death (especially HEK cells).
Differences in Extent of ␤ 1 AR and ␤ 2 AR Trafficking to Lysosomes in Agonist-stimulated Cells-To further establish that ␤ 1 AR undergoes some lysosomal degradation, we visualized the intracellular trafficking of ␤ 1 AR as well as ␤ 2 AR using confocal fluorescence microscopy. HEK 293 cells stably expressing HAtagged ␤ 1 AR were stained at 4°C with fluorescent-tagged anti-HA antibody, then washed, warmed up to 37°C in the absence and presence of agonist for 2 h, and stained with LysoTracker ® Red at the end of the 2-h period. In the unstimulated cells, HA-␤ 1 AR (green) was confined to the cell surface, and LysoTracker ® (red) appeared in cytoplasmic vesicles that are presumably lysosomes (Fig. 3A). Because ␤ 1 AR is resistant to agonist-mediated internalization, most of the receptors remained at the plasma membrane of stimulated cells, and only a small fraction was translocated into the cytosol (Fig. 3B). As a result, very little ␤ 1 AR was detected in the lysosomes (yellow). Further confirming some trafficking of ␤ 1 AR to lysosomes, we observed the colocalization of ␤ 1 AR and the lysosomal marker LAMP1 (lysosomal-associated membrane protein) in agonist-treated cells (Fig. 3C).
We then compared the lysosomal trafficking of ␤ 1 AR-GFP and ␤ 2 AR-GFP stably expressed in HEK 293 cells. As shown in Fig. 4, A and D, small amounts of both ␤-subtypes were present in the lysosomes of the unstimulated cells. When the cells were stimulated with ISO, the amount of ␤ 1 AR-GFP in the lysosomes did not increase at 1 h and slightly increased by 4 h (Fig.  4, B and C). The ␤ 1 AR-containing lysosomes were sparsely and peripherally distributed in the cytosol. In sharp contrast, most of the ␤ 2 AR-GFP translocated from the plasma membrane to large aggregates of lysosomes at 1 h that further coalesced by 4 h (Fig. 4, E and F).
Effect of Arrestin on ␤ 1 AR Trafficking to Lysosomes-Although overexpressing arrestin increases ␤ 1 AR internalization without affecting ␤ 1 AR degradation, it may influence receptor trafficking to lysosomes. In this regard, the limited endocytosis of ␤ 1 AR observed in Figs. 3 and 4 may mask the extent of its trafficking to lysosomes. To explore this possibility, we examined the effects of prolonged agonist treatment on the distribution of ␤ 1 AR-GFP co-expressed with arrestin-2 in HEK 293 and BHK cells and, for comparison, ␤ 2 AR-GFP expressed in BHK cells. In both HEK 293 and BHK cells exposed to agonist for 24 h, we observed a substantial redistribution of ␤ 1 AR-GFP from the plasma membrane to cytoplasmic vesicles, a number of which contained the lysosomal marker (Fig. 5, A-D). The ␤ 1 AR-GFP-containing lysosomes, however, formed smaller, more diffuse aggregates compared with the larger, more compacted aggregates containing ␤ 2 AR-GFP in BHK (Fig. 5F) or HEK 293 (Fig. 4F) cells.
Agonist-stimulated Ubiquitination of ␤ 2 AR but Not ␤ 1 AR-Because ubiquitination was found to be involved in regulating   the endocytosis and sorting of some GPCRs including ␤ 2 AR to lysosomes (12,13,16), we investigated whether ␤ 1 AR also is ubiquitinated. We stripped and re-probed blots similar to the one shown in Fig. 2A using a monoclonal anti-ubiquitin antibody. We were unable to detect any ubiquitinated ␤ 1 AR, whereas we found ubiquitinated ␤ 2 AR (data not shown). Because several studies found that ubiquitination of receptors is more readily detected in cells treated with proteasomal inhibitors, we repeated the biotinylation experiments using cells treated with and without of lactacystin, prepared duplicate blots of the immunoprecipitated biotinylated ␤ARs, and probed one with anti-ubiquitin and the other with streptavidin. Ubiquitinated ␤ 2 AR was detected in HEK 293 cells, and its level was substantially enhanced by lactacystin treatment and further increased by ISO stimulation (Fig. 6). Ubiquitinated ␤ 1 AR was not detected in HEK 293 or BHK cells even in the presence of lactacystin and agonist (top panel). ␤ 1 AR was present in these samples, some of which appeared as dimers (bottom panel). Shenoy et al. (12) found that ␤ 2 AR ubiquitination is transient, increasing upon agonist stimulation, and then decreasing, and the presence of NEM, an inhibitor of deubiquitinating enzymes, during cell lysis and immunoprecipitation enhanced the detection of ubiquitinated proteins. We exposed lactacystintreated HEK 293 cells to ISO for increasing times over a 4-h period and added 10 mM NEM to the lysis and RIPA buffers. Even under these optimal conditions, ubiquitination of ␤ 1 AR still was not detected, whereas ubiquitination of ␤ 2 AR was observed in unstimulated cells and increased upon agonist stimulation (Fig. 7). Taken together, these results clearly demonstrate that ubiquitination of ␤ 1 AR does not occur in comparison with the extensive ubiquitination of ␤ 2 AR in the same cells.
Carboxyl Tails of ␤ARs Dictate Subtype-specific Ubiquitination-We have shown that the C termini of ␤ 1 AR and ␤ 2 AR determine subtype specificity in agonist-mediated phosphorylation, internalization, and prolonged regulation in BHK cells (29). We expressed the same chimeras in which the C-tails of ␤ 1 AR and ␤ 2 AR were exchanged (␤ 1 /␤ 2 ct-AR and ␤ 2 /␤ 1 ct-AR) as well as the wt-␤ARs in HEK 293 cells and determined the extent of agonist-mediated internalization of surface receptors and regulation of total receptors (Fig. 8A and B). The results were similar to those obtained in BHK cells. The ␤ 1 /␤ 2 ct-AR chimera exhibited increased internalization and down-regulation compared with wt-␤ 1 AR, whereas the responses of the ␤ 2 /␤ 1 ct-AR chimera were reduced compared with wt-␤ 2 AR. We then explored the possibility of a role of the C-tails in ␤-subtype ubiquitination. The cells were treated with ISO for up to 24 h and analyzed for ␤AR ubiquitination (Fig. 8C). ISO stimulation resulted in a time-dependent increase in the ubiquitination of wt-␤ 2 AR and chimeric ␤ 1 /␤ 2 ct-AR, which lasted up to 24 h. In contrast, no ubiquitinated wt-␤ 1 AR or chimeric ␤ 2 /␤ 1 ct-AR was detected. These results indicate that the C-tails play a major role in controlling the ubiquitination and degradation of the two subtypes and provide additional evidence that the two processes are closely linked.
Co-detection of ␤ 2 AR but Not ␤ 1 AR Proteins by Anti-ubiquitin and -␤ 2 AR Antibodies-Although anti-␤AR antibodies were used for the immunoprecipitations and the streptavidin overlay confirmed that ␤ARs from biotinylated cells were in the precipitates (Fig. 6), it was possible that the ubiquitin-conjugated proteins detected were not ␤ 2 AR but proteins that coprecipitated with ␤ 2 AR. To address this possibility, we probed blots similar to the one probed with the anti-ubiquitin antibody (Fig. 8C) with anti-␤ 1 AR and -␤ 2 AR antibodies instead (Fig.  8D). The anti-␤ 2 AR antibodies detected immune-reactive proteins between 75 and 200 kDa that corresponded to the proteins revealed by the anti-ubiquitin antibody. Co-detected proteins were only observed in immunoprecipitates from cells expressing wt-␤ 2 AR or the ␤ 1 /␤ 2 ct-AR chimera. Even though the anti-␤ 1 AR antibodies reacted with forms of ␤ 1 AR with high apparent molecular masses, no corresponding proteins were detected in blots probed with the anti-ubiquitin antibody. These observations further confirm that ␤ 2 AR but not ␤ 1 AR undergoes ubiquitination. DISCUSSION In the present study we found that the human ␤ 1 AR stably expressed in HEK 293 cells was resistant to agonist-mediated down-regulation, in agreement with our earlier work using BHK cells (29). In contrast, the human ␤ 2 AR undergoes agonist-mediated down-regulation in both cell lines. We identified another major difference between the two subtypes; ␤ 2 AR but not ␤ 1 AR underwent ubiquitination. Ubiquitination of ␤ 2 AR was observed in unstimulated cells, was increased in a timedependent manner by agonist stimulation, and was enhanced by lactacystin treatment. We were unable to detect ubiquitination of ␤ 1 AR even under optimal conditions for that of ␤ 2 AR. In addition, we found that the C-tails of ␤ 1 AR and ␤ 2 AR were key determinants of ubiquitination, as the chimeras, ␤ 1 /␤ 2 ct-AR and ␤ 2 /␤ 1 ct-AR, in which the C-tails are switched, did or did not undergo ubiquitination, respectively. The C-tails are also major determinants of agonist-mediated phosphorylation, internal-ization, degradation, and down-regulation of the two subtypes (Ref. 29 and the present study). ␤ 1 /␤ 2 ct-AR undergoes these processes to a greater extent than does ␤ 2 /␤ 1 ct-AR, the latter being almost as resistant to down-regulation as is wt-␤ 1 AR. Taken together, these results indicated a strong relationship between ubiquitination and down-regulation. They also are consistent with other studies of GPCRs including ␤ 2 AR (12,13,16).
Receptor ubiquitination requires receptor phosphorylation (12,13,34). For ␤ 2 AR, ubiquitination depends not only on GPCR kinase-catalyzed phosphorylation of the receptor C-tail but also on binding of arrestin, which recruits ubiquitinating enzymes to the complex (12). In cells expressing a phosphorylation-defective ␤ 2 AR or lacking arrestins, ubiquitination of ␤ 2 AR is not observed. We propose that the role of phosphorylation in the ubiquitination of ␤ 2 AR is only to facilitate arrestin binding. This proposal is strongly supported by our current and previous findings as well as those of others. First, agonist stimulation increases the phosphorylation of ␤ 1 AR only slightly compared with that of ␤ 2 AR (27,29,33). Second, arrestins interact more transiently with ␤ 1 AR than ␤ 2 AR in agoniststimulated cells (25). Third, the stability of the interaction of arrestins with ␤ 1 AR is increased when the C-tail is from ␤ 2 AR (25), and agonist-stimulated phosphorylation (29) and ubiquitination (present study) of the two subtypes was determined by the C-tail. As a consequence of the limited agonist-stimulated phosphorylation of ␤ 1 AR, the interaction of arrestin with the receptor may be too weak for arrestin to recruit ubiquitinating enzymes, and thus, ␤ 1 AR is not ubiquitinated. We believe that there is a clear sequential relationship between agonist-stimulated phosphorylation of the receptor C-tail and high affinity arrestin binding, ubiquitination, and trafficking to lysosomes to be degraded.
We recently established that although human ␤ 1 AR and ␤ 2 AR undergo agonist-stimulated endocytosis through clathrin-coated pits, each subtype traffics to a different endosomal compartment (27). Our current results may explain this divergence. Conjugation of membrane receptors with ubiquitin is a signal for sorting of the receptors to lysosomes after endocytosis (11,12,35,36). Sorting involves Hrs, a mammalian ortholog of a yeast vacuolar sorting protein, which has both clathrin and ubiquitin binding domains and associates with flat clathrincoated microdomains on early endosomes (36). These endosomes are morphologically and functionally distinct from clathrin-positive and Hrs-negative early endosomes. Transferrin receptors are not ubiquitinated and, after endocytosis, appear in the Hrs-negative endosomes and rapidly recycle to the plasma membrane. Transferrin receptor-ubiquitin fusion proteins localize to Hrs-positive endosomes and are sorted to the degradative pathway. The chemokine receptor CXCR4, which undergoes agonist-promoted ubiquitination, was shown to colocalize with Hrs-positive endosomes (37). Therefore, we propose that the non-ubiquitinated ␤ 1 AR traffics to Hrs-negative endosomes and is preferentially sorted to the recycling pathway, whereas the ubiquitinated ␤ 2 AR localizes to Hrs-positive endosomes and is targeted to lysosomes for degradation.
Although ubiquitin functions as a signal for sorting a number of GPCRs to lysosomes, other sorting mechanisms have been described (38). In agonist-stimulated cells, the ␦ opioid receptor is rapidly degraded, whereas the opioid receptor is recycled, and exchanging their C-tails generates chimeras that, respectively, are recycled and degraded. GASP, a protein that binds to the C-tail of the ␦ opioid receptor, has been identified as lysosomal-sorting protein (39). A mutated ␦ opioid receptor, in which all the cytoplasmic lysine residues were replaced with arginines is internalized, sorted to lysosomes and degraded similar to the wild type receptor (15). In addition, a recycling sequence has been identified in the C-tail of the receptor, which when deleted results in a mutated receptor that is rapidly degraded and, when fused to the ␦ receptor, confers recycling activity on that subtype (40). Thus, we cannot rule out the possibility that ␤ 1 AR may have an analogous sequence in its C-tail that targets the receptor to the recycling pathway as opposed to recycling by default, i.e. in the absence of a positive lysosomal-sorting signal such as ubiquitination or interaction with a GASP-like protein.
Receptors are constantly being turned over and replaced by newly synthesized receptors at rates that maintain the steady state (41). In the absence of agonist, both subtypes turned over at similar rates (t1 ⁄2 of ϳ6.5 h in BHK and ϳ5 h in HEK 293 cells). In the presence of agonist, the degradation rate of ␤ 1 AR was increased negligibly in BHK and modestly in HEK 293 cells, whereas that of ␤ 2 AR was more than doubled in both cell lines. Both lysosomal and proteasomal inhibitors reduced ␤ 1 AR degradation in agonist-treated BHK and HEK 293 cells. Because ␤ 1 AR is not ubiquitinated, the effects of the proteasomal inhibitors remain to be clarified. The basal degradation of ␤ 1 AR was substantially inhibited only by ALLN, whereas lactacystin, a more specific inhibitor of proteasomal function, was less effective in both cell lines. More likely, basal turnover of ␤ 1 AR occurs elsewhere, such as at the plasma membrane (17). Consistent with this possibility, we observed that fluorescent-labeled anti-HA antibody bound to cell surface HA-␤ 1 AR expressed in HEK 293 cells did not accumulate inside the unstimulated cells after 2 h at 37°C (Fig. 3A). In addition, there was very little accumulation of fluorescence inside unstimulated HEK 293 and BHK cells expressing ␤ 1 AR-GFP ( Fig.  4A; Fig. 5, A and C). Based on a different approach, a similar proposal was made that basal and agonist-stimulated degradation of the vasopressin V2 receptor involves two separate processes (16).
To detect any agonist-promoted trafficking of ␤ 1 AR to lysosomes, we used confocal fluorescence microscopy to visualize both receptors and lysosomes. Colocalization of very small amounts of ␤ 1 AR and two different lysosomal markers was observed in agonist-stimulated HEK 293 cells by 2-4 h. By comparison, agonist-promoted trafficking of ␤ 2 AR to lysosomes in HEK 293 cells was observed by 1 h and was much more robust. In addition, lysosomes containing ␤ 2 AR appeared as large aggregates, whereas ␤ 1 AR-associated lysosomes appeared less aggregated, smaller, and more diffuse. In these experiments, ␤ 2 AR mostly redistributed from the plasma membrane to the cytosol, whereas the distribution of ␤ 1 AR was the opposite. In cells overexpressing arrestin-2, the translocation of ␤ 1 AR from membrane to cytosol increased, as did its accumu-lation in lysosomes. Accumulation of ␤ 2 AR in lysosomes, however, was more extensive even though internalization of the two subtypes was similar. Despite the increase in internalization and lysosomal trafficking, ␤ 1 AR remained resistant to down-regulation.
We previously showed that overexpressing arrestin-2 does not alter degradation of ␤ 1 AR or ␤ 2 AR in agonist-stimulated BHK cells (29). Together with our present results, it appears that increasing ␤AR internalization has little effect on ␤AR degradation. This raises an interesting point, as upon agonist stimulation, a fraction of each subtype is translocated from the site of basal turnover to clathrin-coated pits and subsequent endocytosis. Degradation continues, however, at a similar rate for ␤ 1 AR and at twice the rate for ␤ 2 AR. The reduction in ␤ 1 AR degradation by lysosomal inhibitors, the trafficking of ␤ 1 AR to lysosomes, and the degradation of ␤ 1 AR after clathrin-mediated endocytosis all appear to be contradictory to our finding that ␤ 1 AR is not ubiquitinated. A comparison of the rates of internalization and degradation provides some perspective (Table II). In both cell lines, the limited agonist-promoted internalization of ␤ 1 AR means that most of the receptors continue to undergo basal degradation. Therefore, the rate of ␤ 1 AR degradation through an agonist-mediated pathway is quite low compared with the rate of internalization. For every 1000 receptors internalized, only 24 -40 are degraded, and the rest are recycled. In BHK cells overexpressing arrestin-2, the large increase in ␤ 1 AR endocytosis reduces the number to 6 per 1000. The agonist-mediated component of degradation, however, is almost twice that in BHK cells not overexpressing arrestin-2. Although the ratio is the same for both subtypes expressed in HEK 293 cells, the rates of agonistmediated internalization and degradation of ␤ 2 AR are twice those of ␤ 1 AR. These results are consistent with the observed range of trafficking of the two subtypes to lysosomes. They also suggest that the sorting machinery is not completely efficient. Thus, small amounts of non-ubiquitinated ␤ 1 AR may traffic to lysosomes to be degraded, whereas most will be targeted to the recycling pathway.
In summary, we have identified a major difference in the regulation of human ␤ 1 AR and ␤ 2 AR, namely that the latter is ubiquitinated, and the former is not. This provides a mechanism for understanding the resistance of ␤ 1 AR to degradation and down-regulation and the divergent endosomal trafficking of the two subtypes after clathrin-mediated endocytosis.

TABLE II
Agonist-stimulated internalization and degradation of ␤-subtypes; proportion of internalized receptors degraded Data derived from internalization and degradation assays were fitted to one-phase exponential decay curves as described under "Experimental Procedures." Initial rates (%/min) were extrapolated as previously described (28). Values for ␤AR internalized (% max) have been reported previously (29). The following assumptions were made to calculate the basal and agonist components of the degradation rate in agonist-treated cells; receptors remaining on the cell surface are degraded at the basal rate, and the sum of the basal and agonist components equals the degradation rate. All values are the mean Ϯ S.E. of three to six separate experiments.