A primate-dominant third glycosylation site of the beta2-adrenergic receptor routes receptors to degradation during agonist regulation.

beta(2)-adrenergic receptors (beta(2)AR) of all species are N-linked glycosylated at amino terminus residues approximately 6 and approximately 15. However, the human beta(2)AR has a potential third N-glycosylation site at ECL2 residue 187. To determine whether this residue is glycosylated and to ascertain function, all possible single/multiple Asn --> Gln mutations were made in the human beta(2) AR at positions 6, 15, and 187 and were expressed in Chinese hamster fibroblast cells. Substitution of Asn-187 alone or with Asn-6 or Asn-15 decreased the apparent molecular mass of the receptor on SDS-PAGE in a manner consistent with Asn-187 glycosylation. All receptors bound the agonist isoproterenol and functionally coupled to adenylyl cyclase. However, receptors without 187 glycosylation failed to display long term agonist-promoted down-regulation. In contrast, loss of Asn-6/Asn-15 glycosylation did not alter down-regulation. Cell surface distribution and agonist-promoted internalization of receptors and recruitment of beta-arrestin 2 were unaffected by the loss of 187 glycosylation. Furthermore, acutely internalized wild-type and Gln-187 receptors were both localized by confocal microscopy to early endosomes. During prolonged agonist exposure, wild-type beta(2)AR co-localized with lysosomes, consistent with trafficking to a degradation compartment. However, Gln-187 beta(2)AR failed to co-localize with lysosomes despite agonist treatments up to 18 h. Phylogenetic analysis revealed that this third glycosylation site is found in humans and other higher order primates but not in lower order primates such as the monkey. Nor is this third site found in rodents, which are frequently utilized as animal models. These data thus reveal a previously unrecognized beta(2)AR regulatory motif that appeared late in primate evolution and serves to direct internalized receptors to lysosomal degradation during long term agonist exposure.

␤ 2 -adrenergic receptors (␤ 2 AR) of all species are Nlinked glycosylated at amino terminus residues ϳ6 and ϳ15. However, the human ␤ 2 AR has a potential third N-glycosylation site at ECL2 residue 187. To determine whether this residue is glycosylated and to ascertain function, all possible single/multiple Asn 3 Gln mutations were made in the human ␤ 2 AR at positions 6, 15, and 187 and were expressed in Chinese hamster fibroblast cells. Substitution of Asn-187 alone or with Asn-6 or Asn-15 decreased the apparent molecular mass of the receptor on SDS-PAGE in a manner consistent with Asn-187 glycosylation. All receptors bound the agonist isoproterenol and functionally coupled to adenylyl cyclase. However, receptors without 187 glycosylation failed to display long term agonist-promoted down-regulation. In contrast, loss of Asn-6/Asn-15 glycosylation did not alter down-regulation. Cell surface distribution and agonistpromoted internalization of receptors and recruitment of ␤-arrestin 2 were unaffected by the loss of 187 glycosylation. Furthermore, acutely internalized wild-type and Gln-187 receptors were both localized by confocal microscopy to early endosomes. During prolonged agonist exposure, wild-type ␤ 2 AR co-localized with lysosomes, consistent with trafficking to a degradation compartment. However, Gln-187 ␤ 2 AR failed to co-localize with lysosomes despite agonist treatments up to 18 h. Phylogenetic analysis revealed that this third glycosylation site is found in humans and other higher order primates but not in lower order primates such as the monkey. Nor is this third site found in rodents, which are frequently utilized as animal models. These data thus reveal a previously unrecognized ␤ 2 AR regulatory motif that appeared late in primate evolution and serves to direct internalized receptors to lysosomal degradation during long term agonist exposure.
Like many other G-protein-coupled receptors the ␤ 2 -adrenergic receptor (␤ 2 AR) 1 undergoes a loss of cellular signaling during prolonged agonist exposure. This phenomenon, termed desensitization, serves to integrate ␤ 2 AR function within the context of a cell that is receiving many signals (1). In pathologic conditions, this adaptive response can also contribute to the pathophysiology of the disease and can limit therapeutic effectiveness of agonists administered over prolonged periods of time (1). Early molecular events during short term agonist activation include phosphorylation of the ␤ 2 AR by G-proteincoupled receptor kinases and protein kinase A, which act to rapidly regulate receptor function along a minute-by-minute time frame. With long term agonist exposure, an additional series of events leads to a net loss of cellular receptors, termed down-regulation, which markedly reduces the cellular response (such as cAMP production). The basis of receptor down-regulation includes degradation of the ␤ 2 AR protein, which is linked to early desensitization events, because G-protein-coupled receptor kinase-mediated phosphorylation of ␤ 2 AR leads to recruitment and binding of ␤-arrestin, which, via its scaffolding functions, facilitates internalization (2). A sorting of internalized receptors can result in reinsertion into the cell membrane or movement of a receptor into a degradative pathway with an ultimate loss of the intact receptor (3). The structural features of the ␤ 2 AR that appear to be involved in internalization, recycling, and/or down-regulation, include G-protein-coupled receptor kinase phosphorylation sites in the intracellular tail (4), the third intracellular loop protein kinase A phosphorylation site (5), the NPXXY motif of the seventh transmembrane spanning domain (6), and the PDZ-binding domain in the distal carboxyl-terminal tail of the receptor (7). Expression or trafficking of some G-protein-coupled receptors, such as the ␤ 2 AR, gastrin-releasing peptide receptor, and the V2-vasopressin receptor is dependent on post-translational modifications including N-and O-linked glycosylation (8 -10). All ␤ 2 AR genes identified to date from multiple species have two sites for N-linked glycosylation within the amino terminus of the receptor at amino acid positions ϳ6 and ϳ15. Removal of these sites in the hamster ␤ 2 AR by mutagenesis results in receptors that have an impaired capacity to insert into the membrane. Nevertheless, ϳ50% of the total receptor compliment is ultimately expressed on the cell surface (8).
Interestingly, the human ␤ 2 AR has an additional potential N-glycosylation site, localized to the second extracellular loop, at amino acid 187; this consensus sequence is not found in a wide variety of non-primate species reported to date (see Fig.  1). This prompted us to examine the role of Asn-187 of the human ␤ 2 AR with the idea that this modification may represent a specific mechanism for receptor trafficking that occurred late in mammalian evolution and serves a unique function necessary for primate homeostasis.

MATERIALS AND METHODS
Constructs and Transfections-Site-directed mutagenesis of the human ␤ 2 AR cDNA was carried out so as to substitute Gln for Asn at * This work was supported by National Institutes of Health Grants HL45967 and HL65899. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ʈ  1 The abbreviations used are: ␤ 2 AR, ␤ 2 -adrenergic receptor; 125 I-CYP, amino acids 6, 15, and 187, both singly and in combination. This resulted in eight cDNAs encoding receptors with all possible glycosylation combinations, denoted as NNN (wild-type), QNN, NQN, NNQ, QQN, QNQ, NQQ, and QQQ. These cDNAs were subcloned into a pFLAG-CMV construct, which contained the 5Ј-FLAG epitope (DYKD-DDDK) in-frame with the ␤ 2 AR. Chinese hamster fibroblasts (CHW-1102 cells) were stably transfected by a calcium precipitation technique as described, and selection was performed in 300 g/ml G418. Cells were maintained in 100 units/ml penicillin and 100 g/ml streptomycin in Dulbecco's modified Eagle's medium with 10% fetal calf serum in a 5% CO 2 atmosphere at 37°C. For ␤-arrestin recruitment studies, human embryonic kidney-293 cells were transiently co-transfected by a calcium precipitation method with the ␤ 2 AR construct of interest and a construct consisting of a ␤-arrestin 2-green fluorescent protein fusion gene as described (11). Radioligand Binding-Cells were washed twice with phosphatebuffered saline and detached by scraping in 5 mM Tris, 2 mM EDTA, pH 7.4, at 4°C. This and subsequent buffers contained 5 g/ml of the protease inhibitors leupeptin, soybean trypsin inhibitor, aprotinin, and benzamidine, and 0.1 mg/ml phenylmethylsulfonyl fluoride. Cell particulates were homogenized with a Polytron and centrifuged at 400 ϫ g for 10 min, and the supernatant was pelleted by centrifugation at 30,000 ϫ g for 10 min and resuspended in 75 mM Tris, 12 mM MgCl 2 , 2 mM EDTA, pH 7.4, at 37°C. Competition and saturation radioligand binding studies with these membranes were carried out using 125 Icyanopindolol ( 125 I-CYP) as described previously (12). To quantitate only the cell surface ␤ 2 AR, radioligand binding was carried out using adhered intact cells in 12-well dishes with the hydrophilic radioligand 3 H-CGP12177 (6 nM) in the absence or presence of 1.0 M propranolol used to define nonspecific binding as described (13). Reactions were performed at 4°C for 4 h. Then cells were washed three times in ice-cold phosphate-buffered saline, solubilized in 1% SDS in phosphate-buffered saline, and the contents were counted in a liquid scintillation counter. Additional experiments to ascertain internal versus cell surface ␤ 2 AR utilized intact cell binding with 125 I-CYP, in competition with 0.3 M unlabeled CGP12177 (identifies cell surface receptors) and 1.0 M propranolol (identifies total cellular pool of receptors) as described elsewhere in detail (14). For the 125 I-CYP binding experiments, reactions were terminated by a dilution in 5 mM Tris 2 mM EDTA buffer at 4°C, and bound radioligand was separated from free by vacuum filtration over glass fiber filters.
Adenylyl Cyclase Activities-Membranes were incubated with 30 mM Tris, pH 7.4, 2.0 mM MgCl 2 , 0.8 mM EDTA, 120 M ATP, 60 M GTP, 2.8 mM phosphoenolpyruvate, 2.2 g of myokinase, 100 M cAMP, and 1 Ci of [␣-32 P]ATP for 30 min at 37°C as described (15). Confocal Microscopy-To localize ␤ 2 AR, ␤-arrestin 2, and various intracellular organelles, confocal microscopy was performed with a Zeiss 510 laser scanning microscope and a 63/1.4 oil immersion lens (␤-arrestin 2-green fluorescent protein) or 63/1.2 water immersion lens (others). For ␤ 2 AR localization, studies were performed on fixed cells. In a typical experiment, attached cells maintained in Dulbecco's modified Eagle's medium were treated with agonist for the indicated times at 37°C in 5% CO 2 , washed three times with cold phosphate-buffered saline, and fixed with Ϫ20°C methanol. ␤ 2 AR receptors were identified with a mouse anti-FLAG-M2 primary antibody (Sigma) and the goat anti-mouse secondary antibody Alexa Fluor 488 (Molecular Probe).
Lysosome compartments were stained with a rabbit anti-cathepsin-D primary antibody (Santa Cruz) (16), and the secondary antibody was goat anti-rabbit Alexa Fluor 543. The early endosome compartment was stained with a rabbit anti-EEA1 primary antibody (Affinity BioReagents) (17), with the secondary antibody being Alexa Fluor 543. Images were obtained with excitations of 488 or 543 nm and emission spectra of 505-530 or 560 nm, respectively. For ␤-arrestin recruitment studies in live cells, cells transfected with ␤ 2 AR and ␤-arrestin 2-green fluorescent protein (11) were transferred onto collagen-coated glass bottom dishes. Forty-eight hours after transfection, cells were treated with the indicated agonist, and real time confocal microscopy was carried out using excitation at 488 nm and a 515-540 nm emission filter.
Other-For Western blots, cell membranes were solubilized in radioimmune precipitation assay buffer, and the proteins were fractionated on 10% SDS-PAGE and transferred to nitrocellulose membranes. They were blotted with primary antibody (mouse anti-FLAG-M2) at a titer of 1:750 and a secondary antibody (Amersham Biosciences) for chemiluminescence detection. For fractionation experiments, membranes were separated using a self-generating Percoll (Amersham Biosciences) gradient as described (18). Six 1.5-ml fractions were collected and probed by Western blot with the cathepsin-D and EEA1 antibodies or were used for radioligand binding studies to quantitate ␤ 2 AR. Protein concentrations were determined by the copper bicinchoninic acid method (19). Radioligand binding and kinetic data were fit to standard models using the program Prism (GraphPad Software).

RESULTS AND DISCUSSION
Western blots of membranes from transfected cells expressing the wild-type and seven mutated ␤ 2 ARs are shown in Fig.  2. As expected, an individual removal of position 6 or 15 glycosylation sites resulted in decreases in apparent molecular mass of the major high molecular mass form (from ϳ60 to ϳ56 kDa), and the loss of both sites reduced the molecular mass proportionately. Consistent with position 187 also being glycosylated, the NNQ receptor also had a lower apparent molecular mass like that of the QNN and NQN receptors. Removal of the position 187 site, in combination with those at 6 and/or 15, had the expected additive decrease in apparent molecular mass of the major receptor species as shown.
Each of the expressed receptors had similar binding affinities for isoproterenol, and functionally stimulated adenylyl cyclase (Table I). The consequences of long term (18 h) exposure of cells expressing each of the eight receptors on down-regulation are shown in Fig. 3. A striking and consistent effect imparted by the removal of the glycosylation site at position 187 was observed. As shown, wild-type ␤ 2 AR underwent a 34 Ϯ 3.4% down-regulation. A similar extent of down-regulation was also observed for the single and doubly substituted QNN, NQN, and QQN receptors. In contrast, single removal of the 187glycosylation site (NNQ receptor) resulted in a complete loss of agonist-promoted down-regulation, and in fact an increase in expression was noted (15 Ϯ 6.7%, p Ͻ 0.01 versus wild-type). This same phenotype was observed with all mutant ␤ 2 ARs Gln substitution at positions 6, 15, and 187 is shown (ntf is nontransfected Chinese hamster fibroblast cell lysates, see "Materials and Methods" for other abbreviations). The wild-type ␤ 2 AR appears as a multiply glycosylated receptor with the major high molecular mass form at ϳ60 kDa. Each substitution reduced the apparent molecular mass proportionally, including those with the Asn-187 3 Gln substitution in isolation or with other substitutions.
(NNQ, QNQ, NQQ, and QQQ) where the position 187 was mutated, regardless of the presence or absence of the other glycosylation sites (Fig. 3). The QQQ receptor displayed the greatest increase in expression during agonist exposure. This may be because of the fact that the QQN receptor has somewhat less down-regulation than wild-type and thus less impediment to the up-regulation events imposed by QQQ, as compared with NNQ. Additional studies were also performed using the glycosylation inhibitor tunicamycin. Chinese hamster fibroblast cells expressing wild-type ␤ 2 AR at 50% confluency were treated for 12 h with 0.25 g/ml tunicamycin in the media and then for an additional 12 h with 10 M isoproterenol. Tunicamycin-treated cells underwent only 46 Ϯ 1.3% of the isoproterenol-promoted down-regulation observed with non-tunicamycin-treated cells. These results further support the notion that it is the loss of glycosylation imposed by the mutation at 187 that leads to the altered down-regulation phenotype of the NNQ receptor.
The basis for the marked phenotype imposed by the glycosylation deficiency at position 187 was further explored with the NNQ mutant, thereby maintaining wild-type glycosylation at positions 6 and 15. At base line (i.e. absence of agonist exposure) both wild-type and NNQ ␤ 2 AR were predominantly expressed at the cell surface, as qualitatively shown by confocal microscopy (Fig. 4) and quantitatively determined by wholecell radioligand binding with hydrophilic and hydrophobic li-gands (95 Ϯ 8.3 versus 90 Ϯ 2.4% of the total receptor complements were at the cell surface for wild-type versus NNQ, respectively, n ϭ 4). One of the earliest events in internalization of ␤ 2 AR upon agonist binding is the recruitment of ␤-arrestin from the cytosol to the cell-surface-expressed receptors. ␤-arrestin-green fluorescent protein translocation was ascertained by confocal microscopy in live cells (Fig. 5). Agonist (10 M isoproterenol) exposure resulted in a time-dependent recruitment of ␤-arrestin in cells expressing either receptor. As shown, in wild-type (NNN)-expressing cells, the predominantly intracellular distribution of ␤-arrestin becomes mostly cell surface by 5 min of exposure to isoproterenol. The translocation of NNQ was qualitatively indistinguishable from the wild-type ␤ 2 AR. Consistent with these observations, agonist-promoted receptor internalization as determined by quantitative hydrophilic radioligand binding with intact cells revealed identical kinetics both in terms of the maximal response (45 Ϯ 4.4 versus 49 Ϯ 6.2% internalized) and the rate constant (k ϭ 0.16 Ϯ 0.029/min versus 0.12 Ϯ 0.022/min, n ϭ 4) for the NNN and NNQ receptors, respectively (Fig. 6). Taken together, these data indicate that the initial events in agonist-promoted downregulation, which leads to the redistribution of cell surface ␤ 2 AR to the cell interior, are not perturbed by the absence of glycosylation at Asn-187 of the receptor.
We considered then that the agonist-promoted up-regulation observed with the glycosylation site 187-deficient ␤ 2 AR could be the result of an altered (enhanced) synthesis of new receptors during agonist exposure, or internalized receptors that are not targeted in a wild-type manner to the degradation pathway. New receptor synthesis in the presence of agonist was quantitated by radioligand binding after the exposure of the cells to an irreversible receptor-alkylating agent (Pindobind, Sigma). Cells were treated with 10 nM Pindobind for 2 h (which alkylated ϳ70% of the receptors), washed, and placed back in  . It thus appears that the accumulation of NNQ receptors evoked by the agonist is not based on decreased receptor internalization or increased de novo receptor synthesis, suggesting that the NNQ receptor has altered entry into the degradation pathway(s) after internalization.
The fate of internalized receptors was assessed by confocal microscopy of fixed cells, with an emphasis on co-localization of NNN or NNQ with early endosomes (a short term event) and lysosomes (a long term event). Results of studies with the EEA1 antibody, which identifies early endosomes, are shown in Fig. 7. For both NNN and NNQ receptors, a 15-min exposure to isoproterenol resulted in internalization (Fig. 7, d and j) and co-localization with early endosomes (f and l). Similar findings were observed with incubations up to 4 h (data not shown). Taken together with the other short term agonist data shown in Figs. 5 and 6, it appears that NNQ internalizes and is localized to the early endosomes in a wild-type manner. However, as shown in Fig. 8, NNQ fails to co-localize with lysosomes during prolonged agonist incubation. For the NNN wild-type ␤ 2 AR, one can clearly observe a loss of cell surface and total receptor expression (a versus d) and intracellular accumulation of receptor after 4 h of agonist exposure (d). The intracellular wild-type receptors co-localized with the lysosomal marker cathepsin-D (Fig. 8f). In contrast, overall expression of the NNQ receptor was maintained under the same agonist exposure conditions (Fig. 8, g versus j), and although internalized receptors were observed, they failed to co-localize with the lysosomal  6. Agonist-promoted internalization of the position 187 glycosylation-deficient ␤ 2 AR is unimpaired. Cells were exposed to 10 M isoproterenol for the indicated times and then washed, and whole-cell radioligand binding with the hydrophilic radioligand 3 H-CGP12177 was carried out as described under "Materials and Methods." Results are from 4 experiments. marker (l). Four hours of agonist exposure represents an optimal time point for co-localization of wild-type receptor with lysosomes, as longer exposures result in degradation and a decreased signal. Nevertheless, even with more prolonged agonist exposure (up to 18 h), we never observed the co-localization of NNQ with lysosomes (data not shown). To further support the confocal findings, cells expressing NNN or NNQ were treated with vehicle or isoproterenol for 4 h, whole-cell homogenates were prepared, and the proteins were fractionated on a Percoll gradient. Consistent with the findings of others (25), one of the six fractions (fraction 5) was enriched in lysosomes as determined by Western blots using the cathepsin-D antibody. 125 I-CYP radioligand binding was performed on the proteins from this lysosomal fraction to ascertain whether agonist exposure increased receptors in this pool for NNN-expressing cells and had no effect on the pool of NNQ receptors. Although this fraction undoubtedly contains other compartments/proteins in addition to the specific lysosomes identified by confocal microscopy, we indeed observed increased agonist-promoted receptors in the fraction from NNN cells (from 123 Ϯ 24 to 218 Ϯ 42 fmol/mg, p ϭ 0.03, n ϭ 8), but not NNQ cells (from 65 Ϯ 12 to 65 Ϯ 17 fmol/mg). Although the magnitude of the agonist effect with the wild-type is modest in this assay, the lack of any change with the NNQ mutant is consistent with the confocal results.
The phenotype of the NNQ receptor, then, appears to be one in which the agonist-promoted down-regulation process is perturbed at the recycling step because of a lack of glycosylation at this third extracellular site. The net effect is a lack of downregulation and, in fact, a net increase in cellular receptors due to the continued synthesis of new receptors in an environment of decreased degradation. Thus a previously unrecognized site necessary for agonist-promoted down-regulation has been identified. However, it is noted that the ␤ 2 ARs of other species that lack Asn-187 (e.g. hamster, guinea pig, rat ␤ 2 AR) (20 -24) nevertheless have been reported to undergo agonist-promoted down-regulation in vivo, although the phenomenon has been minimally explored in recombinant cells expressing non-human ␤ 2 ARs (23). This may suggest that other sites/mechanisms are in place in these receptors that abrogate the necessity of Asn-187. As noted earlier, this site is absent in a wide variety of diverse species despite the high degree of surrounding sequence homology (Fig. 1). However, of the 11 ␤ 2 AR genomic sequences reported from primates, 5 have the Asn glycosylation site at the second intracellular loop residue, as is observed in the human receptor (Table II). These five Asn-187-containing primates are higher order primates (human, greater and lesser apes), whereas those without this glycosylation site are all lower order primates (monkey, lemur, tamarin, and galago). This suggests that perhaps Asn-187 provides a specific role for receptor regulation in the larger higher order primates. This mechanism may have evolved because of (i) pressure for this specific type of glycosylation-dependent trafficking or (ii) as a compensatory consequence to retain receptor down-regulation after the evolutionary loss of an alternate down-regulatory motif. ␤ 2 AR sequence comparisons of primate species that carry Asn-187 with those that do not and comparisons with non-primates are hampered by the lack of full-length sequences in primates. Thus scenario (ii) above cannot be adequately explored at this time. Nevertheless, the current study has identified a previously unrecognized site that is essential for long term agonist-promoted down-regulation and has apparently appeared late in primate evolution.