A highly conserved tyrosine residue in G protein-coupled receptors is required for agonist-mediated beta 2-adrenergic receptor sequestration.

An aromatic residue, tyrosine 326 in the prototypical human beta 2-adrenergic receptor, exists in a highly conserved sequence motif in virtually all members of the G protein-coupled receptor family. The potential role of this conserved aromatic amino acid residue in the cellular processes of sequestration (a rapid internalization of the surface receptor) and down-regulation (a slower loss of total cellular receptors) associated with agonist-mediated desensitization of the beta 2-adrenergic receptor was assessed by replacing tyrosine residue 326 with an alanine residue (beta 2AR-Y326A). This mutation completely abolishes agonist-mediated receptor sequestration without affecting the ability of the receptor to activate maximally adenylyl cyclase, to undergo rapid desensitization, and to down-regulate in response to agonist. The only other major change associated with the mutated receptor is a complete loss of the ability to resensitize following rapid desensitization. These results imply that this tyrosine residue, which is part of a highly conserved sequence motif in G protein-coupled receptors, may be responsible for their agonist-mediated sequestration and that sequestration and down-regulation of the receptor are dissociable phenomena. The lack of resensitization in the sequestration-defective beta 2-adrenergic receptor mutant strongly suggests that the sequestration pathway is an important mechanism by which cells re-establish the normal responsiveness of G protein-coupled receptors following the removal of agonist.

An aromatic residue, tyrosine 326 in the prototypical human P2-adrenergic receptor, exists in a highly conserved sequence motif in virtually all members of the G protein-coupled receptor family. The potential role of this conserved aromatic amino acid residue in the cellular processes of sequestration (a rapid internalization of the surface receptor) and down-regulation (a slower loss of total cellular receptors) associated with agonistmediated desensitization of the &-adrenergic receptor was assessed by replacing tyrosine residue 326 with an alanine residue (&AR-FZ6A). This mutation completely abolishes agonist-mediated receptor sequestration without affecting the ability of the receptor to activate maximally adenylyl cyclase, to undergo rapid desensitization, and to down-regulate in response to agonist. The only other major change associated with the mutated receptor is a complete loss of the ability to resensitize following rapid desensitization. These results imply that this tyrosine residue, which is part of a highly conserved sequence motif in G protein-coupled receptors, may be responsible for their agonist-mediated sequestration and that sequestration and down-regulation of the receptor are dissociable phenomena. The lack of resensitization in the sequestration-defective p2-adrenergic receptor mutant strongly suggests that the sequestration pathway is an important mechanism by which cells reestablish the normal responsiveness of G proteincoupled receptors following the removal of agonist.
Closely related members of the large superfamily of G protein-coupled receptors exhibit high degrees of amino acid conservation, especially in regions implicated in ligand binding and interactions with G proteins. Other residues, whose roles have remained obscure, are also conserved across virtually all members of this superfamily (1, 2). One such residue, tyrosine 326 of the prototypical human Pz-adrenergic receptor is present at the proposed junction of the seventh transmembrane domain and proximal part of the carboxyl terminus of 1 These authors contributed equally to this work. virtually all G protein-coupled receptors. Aromatic residues in various sequence motifs, NPXY or YXRF, have been implicated in the cellular recycling of receptors for the proteins low density lipoprotein, epidermal growth factor, and insulin (3, 4). It has been assumed that these motifs direct the plasma membrane receptors to clathrin-coated pits. Recently, it has been demonstrated that the epidermal growth factor receptor interacts with the amino-terminal domain of a-adaptin, a component of the endocytotic machinery of clathrin-coated pits, in response to the epidermal growth factor activation (5).
G protein-coupled receptors are known to undergo cellular processing, especially in association with the phenomenon of desensitization in response to prolonged agonist occupancy of the receptor. Several temporally distinguishable cellular events have been characterized for &AFt desensitization (6). Sequestration is observable within minutes of agonist occupancy. It is manifested by a loss of cellular binding sites for hydrophilic ligands (plasma membrane receptors) but not hydrophobic ligands (total cellular receptors). Sequestration quickly follows the more rapid (seconds to minutes) uncoupling of the receptor from its ability to activate the G protein. Whereas the rapid uncoupling process is due to phosphorylation of the pzAR by CAMP-dependent protein kinase and specific receptor kinases @-adrenergic receptor kinase), receptor phosphorylation is not required for sequestration; conversely, sequestration is not required for the onset ofrapid desensitization (7,8). Prolonged (>1 h) exposure of cells to agonist results in the progressive loss of total cellular receptors in a process termed down-regulation. I t has been proposed that PzAR sequestration may be a n obligatory process that precedes down-regulation (9, 10). In the present study we sought to evaluate the potential role ofthe tyrosine 326 residue in these processes. This tyrosine residue, which exists in the sequence NPLN in the pzAR or a closely related sequence in other receptors, is conserved in the majority of the G protein-coupled receptors so far characterized (Fig. 1).

EXPERIMENTAL PROCEDURES Materials
12SI-Cyanopindolol and 1261-pindolol were purchased from DuPont NEN. Protein G, propranolol, and isoproterenol were obtained from Sigma. Gpp(NH)p and CGP12177 were from Boehringer Mannheim. Microcystin-LR was purchased from Calbiochem. 12CA5 ascites were purchased from Babco.

Plasmid Constructions
Mutant and epitope-labeled cDNA were modified by site-directed mutagenesis using the polymerase chain reaction (11). A BgZIzIEcoRV cassette of pBC-NAR (12) was used to create the Wz6 -Ala mutation and was re-ligated into the 12CA5 epitope-tagged construct (13,14). This construct was then blunt end-ligated a t the Sac1 and Sal1 sites into the Hind111 polylinker site of the neomycin resistance expression plasmid pRC-CMV (Invitrogen). Constructs were verified by dideoxy sequencing analysis.
Cell Culture a n d Dansfection Chinese hamster ovary (CHO) cells were obtained from ATCC. Human epitope-labeled BAR in pBC or mutant p,AR cDNA in pRC/CMV plasmids were transfected, respectively, with and without pSV-2neo (12) into CHO cells using coprecipitation with calcium phosphate. Clones were selected in the presence of 1 m g / d G418 (Life Technologies, Inc.). Colonies originating from single cells were subcloned and evaluated for receptor expression and homogeneity using 1251-pindolol binding and immunofluorescence. Cells were maintained following clonal selection in Ham's F-12 with 10% fetal bovine serum, 50 mg/ml penicillin/ streptomycin (50 units), and 500 pg/ml Geneticin@ (Life Technologies, Inc.). Experiments reported in this paper were performed on a t least two different clonal cell lines for each of the mutant and wild type receptors. Expression levels varied between 0.5-0.7 pmol of receptodmg of membrane protein for the cells expressing the mutant receptor and O. €-1. 0 pmol of receptodmg of membrane protein for the wild type receptors. This level of receptor expression corresponds to -2,500-5,000 receptodcell.

Binding Studies
Binding assays using *261-cyanopindolol were performed as previously described (12). Levels of functionally expressed receptors were assessed in two ways. First, total cellular receptor expression was assessed by ligand binding with 1251-pindolol which, because of its hydrophobicity, can measure both surface receptors and intracellular receptors (12). Second, relative fluorescence quantitation using flow cytometry with the 12CA5 antibody was performed on the same aliquots of cells on which 1251-pindolol binding was performed. In these experiments a constant and equal ratio of the 12511-pindolol binding/ relative surface fluorescence (0.94 2 0.12; n = 3) was obtained for both mutant and wild type receptor-expressing cells. This suggests that both cell types express proportionally equivalent amounts of functional surface receptor and similar amounts of total receptors. In addition, total cellular wild type and mutant receptors were immunoprecipitated using the monoclonal antibody 12CA5 (Babco) and revealed by Western blotting experiments using an antibody raised against the carboxyl tail of human p2AR. The results show roughly equivalent amounts of the same molecular weight species of wild type and mutant receptors (data not shown).2 Sequestration a n d Down-regulation Radioligand Binding-Whole cell radioligand binding was performed with 1251-pindolol for 3 hours a t 12-13 "C in the presence or absence of either 300-600 rn CGP 12177 or 200 p~ propranolol essentially as described (8). Cells were treated with isoproterenol either in suspension or while still attached to flasks. If first detached using phosphate-buffered saline (PBS) with 5 rn EDTA solution, it was CN-cia1 to keep the cells and buffers ice-cold. Receptor sequestration was defined as the fraction of specific radioligand binding not competed for by CGP 12177 minus the basal level of sequestration as measured without exposure to agonist.
Flow Cytometry-12CA5 antibody was purified on protein G (Sigma) equilibrated with 20 rn phosphate buffer, pH 7.5, and eluted with 100 rn glycine, pH 2.75, into 1-ml fractions that were immediately neutralized by 1 M Tris buffer, pH 8.5. The resulting antibody fractions were dialyzed against PBS and stored for use a t 1.75 mg/ml. Live cell suspensions were labeled, following treatment with isoproterenol, a t 4 "C for 40-60 min with 1:lOO dilutions of antibody in Ham's F-12 medium, washed with cold Dulbecco's PBS by centrifugation and subsequently labeled with a 1:lOO dilution of affinity-purified, Fc-specific, fluoresceinlabeled goat anti-mouse antibody (Sigma). Cells were rewashed and fixed in 3.6% formaldehyde and analyzed within 1 h on a Becton-Dickinson flow cytometer. Baseline cell fluorescence intensity was determined with washed unlabeled cells, and cells incubated only with the fluorescein-labeled goat anti-mouse antibodies.
Photography-Live cells on glass coverslips were pretreated in the absence or presence of 10 p~ isoproterenol for 30 min in Ham's F-12 medium a t 37 "C, chilled to 4 "C, exposed sequentially for 40 min (with an intervening wash) to 1:lOO dilutions of 12CA5 mouse monoclonal  antibody (1.75 mg/ml) and affinity-purified goat anti-mouse Fc-specific antibody (4 mg/ml), washed, and fixed in 4% formaldehyde in phosphate-buffered saline for 10 min a t 4 "C. Photographic exposures of coverslips were done under identical conditions, and fluorescence intensities converted to pseudocolor images over a dynamic range of 1-256 following background subtraction.
Desensitization a n d Resensitization CHO cells expressing either wild type or mutant p,AR were grown to confluence in T-75-cm flasks. Prior to experiments, cells were put in serum-free Ham's F-12 medium for 10-24 h. Cells expressing wild type or mutant p,AR were incubated with 0.1 rn ascorbic acid and 10 p~ isoproterenol for 20 min to induce desensitization. At the end of the incubation, the medium was removed and cells were washed extensively with cold PBS. Cells that were to be assessed for resensitization were washed twice with warm PBS (-30 m l ) to remove the agonist and allowed to continue incubation in fresh serum-free medium for 20 min as previously described (7). The resensitization period was terminated by washing with cold PBS. Cells were scraped into a lysis buffer (10 rn Tris, pH 7.4, 5 rn EDTA, 1 rn microcystin-LR) and centrifuged at 40,000 x g for 25 min at 4 "C. The pellets were resuspended in cold 75 rn Tris (pH 7.4). 2 rn EDTA, and 12.5 rn MgCl, using a Teflon pestle.
Membrane preparations were assayed for agonist-stimulated adenylyl cyclase as previously described (12, 15). Data treatment was done as previously reported (15).

RESULTS AND DISCUSSION
A mutant p2AR was produced by site-directed mutagenesis by replacing tyrosine 326 in the sequence NPLlY by an alanine residue to yield the construct /32AR-Y326A. In order to facilitate the detection of the expressed receptors by means other than ligand binding, both wild type pzAR and p2AR-y326A were epitope-tagged on the amino terminus using the 12CA5 epitope as previously described (Fig. 1) (13, 14,161. As shown in Fig. 2 Red corresponds to regions of highest receptor density decreasing progressively to yellow, green, blue, and black (no receptors). Relative fluorescence intensities were as follows: wild type with no isoproterenol pretreatment = 1, wild type with isoproterenol pretreatment = 0.16, mutant without and with isoproterenol pretreatment = 1 and 0.97, respectively. Photographs were taken at the same magnification. Apparent Merences in cell size relate to the fields chosen. B, sequestration of wild type and mutant & U t in CHO cells exposed to isoproterenol as as- sure of cells to 10 isoproterenol for up to 90 min produced a rapid internalization or sequestration of 3040% of the cell surface wild type p&, while sequestration was totally absent in cells expressing the mutant receptor. Essentially the same lack of sequestration was observable whether surface receptors were assessed by ligand binding using the hydrophilic ligand CGP 12177 or by flow cytometry using the monoclonal antibody 12CA5 (Fig. 2B). As shown in Fig. 2A, the lack of agonistmediated sequestration with the mutant receptor is observable directly by fluorescence microscopy. Cells expressing the mutant receptor do not lose cell surface fluorescence following agonist treatment. In contrast to the effect of the mutation on sequestration, both wild type and mutant p2AR were similarly and efficiently down-regulated upon prolonged exposure (up to 24 h) to the agonist isoproterenol (Fig. 2 0 . These results strongly suggest a role for this region of the receptor in the process of sequestration and demonstrate that sequestration and down-regulation are dissociable phenomena. In order to assess whether abolition of receptor sequestration was due to the specific mutation of the aromatic residue in this sequence motif or indirectly due to a major impairment of other essential receptor functions, we determined whether any of the signal transduction properties of the mutated receptor were significantly altered. As shown in Fig. 3, the p2AR-F26A mutant can interact with a G protein. In agonist competition curves for labeled antagonist (1261-cyanopindolol) binding in membranes, both wild type and mutant receptors display biphasic competition curves that could be shifted to monophasic curves in the presence of the guanine nucleotide Gpp(NH)p. This is the hallmark of receptodG protein interaction. However, whereas the KO values derived for the agonist high agonist isoproterenol were different with 37 and 15%, respectively, for the wild type and mutant receptors. These results indicate that the mutant receptor may have a reduced ability to couple to the G protein.
The mutant p2AR-y326A was as effective as the wild type p2AR in mediating maximal agonist stimulation of adenylyl cyclase. A 5-7-fold stimulation of the enzyme activity was observed at maximal isoproterenol concentrations (wild type: basal = 28 8 pmol/mg.min; maximal activation = 129 f 24 pmoYmg.min; p2AR-y326A mutant: basal = 18 f 3 pmoY mg.min, maximal activation = 124 2 17 pmol/mg.min; mean 2 S.E., n = 8). However, as might be expected from the lower proportion (fraction) of receptors existing in a high affinity state (Fig. 3), the agonist isoproterenol was less potent in stimulating adenylyl cyclase via the mutant versus the wild type receptor (EC50 = 27 f: 3 n~ for wild type p 2 A R ; EC50 = 267 2 33 n~ for p2AR-y326A; mean 2 S.E., n = 8) (Fig. 4). Although functionally coupled, the mutant receptor appears to be less well coupled than the wild type receptor. However, it is unlikely that this impairment in coupling is responsible for the complete absence of sequestration of the mutant receptor. Indeed, previous studies from our laboratories (17,19) have established that respectively. After desensitization both receptors undergo a 2-fold rightward shif€ in their EC50 for agonist stimulation and a decrease in the maximal stimulation of adenylyl cyclase. Desensitized curves are significantly different from their respective control curves ( p < 0.05). After removal of agonist, the wild type p f l undergoes a recovery of responsiveness that results in a significant increase of the maximal stimulation of adenylyl cyclase in comparison to control and desensitized curves ( p < 0.05). During the same period of resensitization, the pfl-Y3z6A mutant was not able to undergo any recovery in the responsiveness to isoproterenol stimulation. The desensitized and resensitized curves were indistinguishable. Receptor levels determined following desensitization or resensitization were not altered in comparison to control levels (data not shown). In all experiments, adenylyl cyclase responses were normalized to the stimulation observed in the same membrane preparations with 10 m~ sodium fluoride. Each curve represents the clonal cell lines, and the mean curves were analyzed using ALLFIT (29).

average of four separate experiments performed with two different
Binding assays were performed using 1z61-cyanopindolol as previously described (12). p2AR that are partially or completely uncoupled from G proteins and adenylyl cyclase stimulation as a result of other mutations are nonetheless totally normal in their agonist-mediated sequestration patterns.
Historically, studies of the mechanism of desensitization have led to the notion that sequestration might play an important role in the overall process of desensitization. Several more recent studies have supplied evidence contrary to this notion (17)(18)(19). The results shown in Fig. 4 corroborate findings that sequestration does not appear to be a major determinant for the onset of rapid desensitization. p2AR-y326A mutants, which are totally incapable of sequestering, are uncoupled, like wild type receptors, by a short exposure to the agonist isoproterenol. Whereas the onset of desensitization does not appear to involve sequestration, other evidence has previously suggested that sequestration might be more important to the process of resensitization of the adenylyl cyclase response (15,20). To test this possibility, we examined the ability of both wild type and mutant receptor-expressing P2AR cells to resensitize following removal of the desensitizing agonist. Fig. 4 reveals a dramatic difference in the behavior of the two receptors. Whereas the wild type PzAR was resensitized 20 min following removal of the agonist, over the same period of time no detectable recovery of responsiveness was observed with the pzAR-F26A mutant. These data strongly suggest an involvement of sequestration in the process of resensitization of the P-adrenergic receptorcoupled adenylyl cyclase response. Our results are also in line with recent observations of Yu et al. (15), who showed that, by selectively inhibiting sequestration of the P z A R in CHO cells or creating a sequestration-defective pzAR mutant by site-directed mutagenesis of serine and threonine residues in the C terminal tail of the receptor, they could essentially block resensitization of responsiveness of adenylyl cyclase.
In this study we show that the presence of the tyrosine residue in the highly conserved sequence motif NPXXY in the P z A R appears to be required for agonist-mediated sequestration of the receptor. Alteration of this sequence by mutagenesis appears to interfere specifically with the process of sequestration since the mutated receptor was able to interact effectively with G proteins and stimulate adenylyl cyclase. Although slight differences in these activation parameters were observed with the mutated receptor, the magnitude of these changes could not explain the total lack of sequestration of the p2AR.
The nature of the cellular compartment in which G proteincoupled receptors may sequester has not been extensively characterized. Recently, however, von Zastrow and Kobilka (13) have shown that P2AR internalized in response to agonist can be identified in the same endosomes as constitutively recycling transferrin receptors. Constitutively recycling receptors such as the transferrin and low density lipoprotein receptors contain tyrosine residues in motifs (YXRF or WXY) that have been proposed to form tight turn conformations recognized by the endocytotic machinery (21,22). By contrast, the secondary structural environment of the conserved tyrosine residue in G protein-coupled receptors (mutated in this study) most likely differs from this paradigm, since this tyrosine residue appears consistently in a motif in which it is 2 or 3 residues removed from conserved Asn and Pro residues (NF'(X)23Y). Moreover, whereas in constitutively recycling receptors the tyrosine containing motifs are usually 15-20 residues removed from the transmembrane domains (23), the conserved tyrosine residue in G protein-coupled receptors is in close proximity to the seventh transmembrane domain of these receptors. The structural differences between the internalization motifs of constitutively recycling receptors and the tyrosine-containing sequences of G protein-coupled receptors might suggest distinct internalization mechanisms for these receptors; it is interesting, therefore, to speculate that such differences might underlie the constitutive or agonist-mediated character of the internalization phenomena.
The highly conserved nature of the sequence surrounding the conserved tyrosine residue in a large number of the G proteincoupled receptors suggests that it may serve as a "sequestration signal" in all these receptors. The prevalence of this conserved sequence also raises the possibility that sequestration may be a common mechanism of G protein-coupled receptor trafficking in response to agonist stimulation. However, mutations in other regions of the receptor, specifically Ser, Thr, and Gln residues in the proximal half of the carboxyl terminus of the P z A R , also lead to a loss of agonist-mediated sequestration (15). In addition, related receptors, all of which possess the tyrosine residue in the conserved sequence motif, have been found to sequester to various degrees as exemplified by the various a2AR subtypes (24,251 and the EP3, and EPBb prostaglandin E receptor subtypes (26). Altogether, these results suggest that while the motif is apparently necessary for sequestration, it may not be sufficient. Illustratively, the human &AR, which contains the tyrosine residue in the conserved motif, fails to sequester. However, when a chimeric construct of the P3 receptor with the C terminal tail of the PzAR is expressed in cells, the resulting chimeric receptor undergoes agonist-mediated sequestration (14).
The results observed with the sequestration-defective p2AR-P Z 6 A mutant point to a specific role of sequestration in the overall process of desensitization. Originally, removal of surface receptors through the process of sequestration was thought to play a major role in desensitization. As shown in this work (Fig. 4) and in other work using selective inhibitors of sequestration (12, 191, cells can desensitize normally without receptor Sequestration. However, the lack of resensitization in the sequestration-defective PzAR-P26A mutant strongly suggests that the sequestration pathway is an important mechanism by which cells reestablish the normal responsiveness of G protein-coupled receptors following removal of the stimulus. Since short term desensitization to agonist is mediated by phosphorylation of the receptor, it is interesting to speculate that the compartment where receptors are sequestered may be associated with dephosphorylation of receptor (15,20). Our data are consistent with the proposal of Sibley et al. (20), who have shown that sequestered receptors in isolated vesicles are less phosphorylated than desensitized receptors still in the plasma membrane. Moreover, these isolated vesicles containing sequestered receptors show higher receptor phosphatase activity than do plasma membranes (20). Thus, rather than representing an important element in the onset of desensitization, sequestration may promote resensitization by transporting uncoupled receptors from the plasma membrane to a compartment in which they can be dephosphorylated and from which they can then recycle back to the cell surface as signalingcompetent receptors.