alpha 2A/alpha 2C-adrenergic receptor third loop chimera show that agonist interaction with receptor subtype backbone establishes G protein-coupled receptor kinase phosphorylation.

The alpha(2A)-adrenergic receptor (AR) undergoes rapid agonist-promoted desensitization due to phosphorylation by G protein-coupled receptor kinases (GRKs) 2 and 3 at serines in the third intracellular loop of the receptor. In contrast, the alpha(2C)AR fails to display such desensitization or phosphorylation, which has been presumed to be due to this receptor lacking GRK phosphorylation sites. However, the alpha(2C)AR has multiple serines and threonines in putative favorable motifs within its third intracellular loop. We considered that the conformation of the third intracellular loop imposed by agonists binding to the transmembrane-spanning domains could be the basis of this subtype-specific property, rather than the presence or absence of phosphoacceptors per se. To address this, alpha(2A)/alpha(2C) third loop chimeric receptors were constructed. In whole cell phosphorylation studies, the alpha(2A) with the alpha(2C) third loop receptor underwent agonist-promoted phosphorylation while the alpha(2C) with the alpha(2A) third loop receptor did not, indicating that the agonist interaction with the parent receptor backbone establishes the phosphorylation phenotype. We postulated then that agonists with diverse structures that distinctly interact with alpha(2)AR should display different degrees of phosphorylation independent of receptor activation. Indeed, several full and partial agonists were identified, which evoked phosphorylation that was not related to intrinsic activity as established by [(35)S]guanosine 5'-3-O-(thio)triphosphate binding. Taken together, it appears that phosphorylation of the alpha(2)AR evoked by agonist is highly sensitive to the conformation of the third intracellular loop induced/stabilized by agonist to such an extent that these properties dictate the extent of phosphorylation of the loop when phosphoacceptors are present, and are the basis for subtype-specific phosphorylation.

The ␣ 2A -adrenergic receptor (AR) undergoes rapid agonist-promoted desensitization due to phosphorylation by G protein-coupled receptor kinases (GRKs) 2 and 3 at serines in the third intracellular loop of the receptor. In contrast, the ␣ 2C AR fails to display such desensitization or phosphorylation, which has been presumed to be due to this receptor lacking GRK phosphorylation sites. However, the ␣ 2C AR has multiple serines and threonines in putative favorable motifs within its third intracellular loop. We considered that the conformation of the third intracellular loop imposed by agonists binding to the transmembrane-spanning domains could be the basis of this subtype-specific property, rather than the presence or absence of phosphoacceptors per se. To address this, ␣ 2A /␣ 2C third loop chimeric receptors were constructed. In whole cell phosphorylation studies, the ␣ 2A with the ␣ 2C third loop receptor underwent agonistpromoted phosphorylation while the ␣ 2C with the ␣ 2A third loop receptor did not, indicating that the agonist interaction with the parent receptor backbone establishes the phosphorylation phenotype. We postulated then that agonists with diverse structures that distinctly interact with ␣ 2 AR should display different degrees of phosphorylation independent of receptor activation. Indeed, several full and partial agonists were identified, which evoked phosphorylation that was not related to intrinsic activity as established by [ 35 S]guanosine 5-3-O-(thio)triphosphate binding. Taken together, it appears that phosphorylation of the ␣ 2 AR evoked by agonist is highly sensitive to the conformation of the third intracellular loop induced/stabilized by agonist to such an extent that these properties dictate the extent of phosphorylation of the loop when phosphoacceptors are present, and are the basis for subtypespecific phosphorylation.
Desensitization, defined as a waning of a signal despite the presence of stimulus, is a common biological phenomenon. Many receptors that signal through guanine nucleotide-binding proteins (G proteins) display desensitization in the presence of continuous agonist exposure (1). Such regulation of receptor function is key to the cell being able to integrate the myriad of received signals in order to maintain homeostasis under a variety of physiologic and pathologic conditions. Receptor desensitization can be adaptive or maladaptive within the context of disease (2). Additionally, the process can lead to clinical tachyphylaxis, which limits the therapeutic effectiveness of administered agonists (1).
The most rapid form of desensitization of G protein-coupled receptors is due to phosphorylation, leading to a decrease in coupling of the receptor to its cognate G protein. A family of serine/threonine protein kinases, termed G protein-coupled receptor kinases (GRKs), 1 serves to phosphorylate the agonistoccupied forms of these receptors, and are the kinases responsible for agonist-promoted (homologous) desensitization (3,4). Within the adrenergic receptor family, the ␤ 2 -adrenergic (␤ 2 AR) and ␣ 2A -adrenergic (␣ 2A AR) receptors have been extensively studied in regards to GRK-mediated phosphorylation and desensitization. For the ␤ 2 AR, serines and threonines in the carboxyl-terminal tail of the receptor have been shown to be phosphorylated by one or more GRKs (5,6). For the ␣ 2A AR, four serines in the mid-portion of the third intracellular loop within the EESSSS motif, are the residues phosphorylated (7,8). For both receptors, the critical step to initiation of GRK-mediated phosphorylation is the attainment of the agonist-occupied or stabilized conformation, also termed the "active" conformation. Indeed, in the current model of GRK-mediated phosphorylation, the optimal conformation required for G protein coupling is considered the same as that required for GRK-mediated phosphorylation (3,4). Partial agonists, then, evoke receptor phosphorylation of the ␤ 2 AR in direct proportion to their intrinsic activities (9). The ␣ 2 ARs consist of three receptor subtypes, denoted ␣ 2A , ␣ 2B , and ␣ 2C . Each couple to the inhibitory G protein G i , and thus serve to decrease adenylyl cyclase and intracellular cAMP levels in the cell (10). While all three subtypes display similar G i coupling, they differ in their propensity to undergo agonist-promoted phosphorylation and desensitization (11). We and others have shown in recombinant expressing cells that while ␣ 2A and ␣ 2B receptors undergo short term agonist-promoted phosphorylation and desensitization, the ␣ 2C receptor does not (12)(13)(14). This does not appear to be due to the lack of cellular expression of a particular GRK, as overexpression of GRKs 2, 3, 5, and 6 fail to evoke agonist-promoted phosphorylation of the ␣ 2C AR (14). Comparison of the sequences of the third intracellular loops of the ␣ 2A and ␣ 2C receptors ( Fig. 1) shows that both have a number of serines and threonines in similar locations. While the ␣ 2C does not have the ␣ 2A sequence EESSS, which has been shown in vivo (15) and in vitro (16) to be an excellent substrate for GRK phosphorylation, it does have DESS within the mid-portion of the loop. We have considered that while the residues that might be phosphorylated by GRKs are present in both subtypes, the agonist-promoted intracellular milieu might be different between the two subtypes. This could thus be the basis of the differential phosphorylation of the two subtypes by agonist, rather than the presence or absence of phosphoacceptors per se. To address this, chimeric ␣ 2 ARs were constructed which were composed of the ␣ 2A with substitution of the ␣ 2C third intracellular loop, and the ␣ 2C with substitution of the ␣ 2A third intracellular loop. These two chimera, along with the wild-type ␣ 2A and ␣ 2C receptors, were expressed in cells and agonist-promoted phosphorylation and desensitization studies undertaken.

EXPERIMENTAL PROCEDURES
Constructs and Mutagenesis-The wild-type human ␣ 2A AR and ␣ 2C AR cDNAs were subcloned into the expression vector pCDNA1 modified to contain in-frame the sequence encoding the hemagglutinin epitope tag at the amino terminus. To construct the chimeric receptors, site-directed mutagenesis (17) was carried out on each wild-type template to create the restriction endonuclease sites NheI (transmembrane domain 5) and BamHI (carboxyl terminus of the third intracellular loop) to afford exchange of fragments encompassing the third loop ( Fig.  1). The former reactions were also designed so as to maintain ␣ 2A or ␣ 2C sequence in the transmembrane domains of the chimera. Thus the chimera consisted of ␣ 2A AR with the ␣ 2C AR third intracellular loop, denoted ␣ 2A (3L C ), and ␣ 2C AR with the ␣ 2A AR third intracellular loop, denoted ␣ 2C (3L A ).
Transfection and Cell Culture-Chinese hamster fibroblasts (CHW-1102 cells) were permanently transfected by a calcium phosphate precipitation technique as described (15). Individual clonal lines were selected in 0.3 mg/ml G418, and ␣ 2 AR expression determined by radioligand binding (see below). Cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 100 g/ml streptomycin and 100 units/ml penicillin, in a 95% air, 5% CO 2 atmosphere at 37°C. For desensitization experiments, cells were washed three times with phosphate-buffered saline and incubated in fresh media without serum in the presence of 10 M epinephrine and 0.1 M ascorbic acid, or ascorbic acid alone, for 15 min at 37°C. COS-7 cells were transfected by a DEAE-dextran method (14) for phosphorylation and [ 35 S]GTP␥S binding (18).
Whole Cell Phosphorylation-These experiments require high receptor expression; thus, COS-7 cells were transiently transfected to achieve 5-7 pmol/mg expression. Cells in culture were incubated for 2 h with [ 32 P]orthophosphate (3.0 mCi/ml). They were then exposed to vehicle or agonist for 5 min, placed on ice and washed with cold phosphate-buffered saline. Cells were then solubilized in 50 mM Tris (pH 7.5), 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 1 mM EDTA, the phosphatase inhibitors NaF and sodium pyrophosphate (both 10 mM), and the protease inhibitors aprotinin (10 g/ml), leupeptin (10 g/ml), and PMSF (1 mM). Receptor was immunoprecipitated using 1 g/ml hemagglutinin antibody from Roche Molecular Biochemicals essentially as we have described in detail elsewhere (19), and subjected to 10% SDS-polyacrylamide gel electrophoresis. Gels were imaged using a PhosphorImager and the results quantitated using Image Quant software (Molecular Dynamics). For presentation purposes gels were exposed to x-ray film and autoradiograms produced.
Radioligand Binding-CHW cells were washed with phosphate-buffered saline, detached and lysed with a rubber policeman in 5 mM Tris, pH 7.40, 2 mM EDTA buffer, and the particulates centrifuged for 10 min at 37,000 ϫ g. Membranes were resuspended in 75 mM Tris, pH 7.40, 12 mM MgCl 2 , 2 mM EDTA buffer. Radioligand competition and saturation binding assays were carried out on these membranes using [ 3 H]yohimbine as described previously (20). Receptor expression in the CHW cells typically ranged from ϳ500 to 700 fmol/mg and were matched for a given set of experiments.
Adenylyl Cyclase Assays-Membranes from CHW cells were prepared as above and were resuspended in a buffer which provided for a final concentration of 16 mM HEPES, 0.32 mM EDTA, and 0.64 mM MgCl 2 , pH 7.4, in the assay. The reaction consisted of (final concentrations) 2.7 mM phosphoenolpyruvate, 50 nM GTP, 0.1 mM cAMP, 0.12 mM ATP, 50 g/ml myokinase, 0.05 mM ascorbic acid, and 1.0 Ci of [␣-32 P]ATP. Reactions were carried out for 30 min at 37°C in the presence of vehicle (basal), 5 M forskolin, or forskolin and the indicated concentrations of agonist. They were terminated by adding ice-cold stop solution containing [ 3 H]cAMP and excess ATP and cAMP. [ 32 P]cAMP was separated by chromatography over alumina columns. For desensitization studies, cells were exposed to 10 M epinephrine for 15 min, washed five times with cold PBS, and membranes prepared as above.
[ 35 S]GTP␥S Binding-To assign intrinsic activities for various agonists, [ 35 S]GTP␥S binding was carried out. This approach was utilized to avoid any potential problems associated with spare receptors that may be encountered in functional assays. COS-7 cells were transiently transfected with the ␣ 2A AR construct and a construct encoding rat G␣ i2 as described (18). Cell membranes (ϳ20 g) were incubated in buffer containing 25 mM Tris-HCl, pH 7.4, 5 mM MgCl 2 , 1 mM EDTA, 1 mM dithiothreitol, 100 mM NaCl, 1 M GDP, and 2 nM [ 35 S]GTP␥S in a 100-l reaction volume for 15 min at room temperature. Incubations were terminated by dilution with four volumes of ice cold 10 mM Tris-HCl, pH 7.4, buffer and vacuum filtration over Whatman GF/C glass fiber filters. Nonspecific binding was measured in the presence of 10 M GTP␥S.
Statistical Analysis-Radioligand binding and adenylyl cyclase doseresponse studies were analyzed by an iterative least squares technique as described previously using Prizm software (GraphPad, San Diego, CA). Comparisons of results were by paired or unpaired t tests with significance imparted when p Ͻ 0.05. Data are reported as mean Ϯ S.E.

RESULTS
The results of radioligand binding studies are shown in Table  I. The two wild-type receptors and the two chimeric receptors all bound [ 3 H]yohimbine with high affinity. There were no differences between the K D for the ␣ 2A versus ␣ 2A (3L C ), or the ␣ 2C versus ␣ 2C (3L A ) receptor. Competition studies with the antagonist phentolamine revealed an ϳ10-fold lower affinity for ␣ 2C AR compared with ␣ 2A AR. The respective chimeric receptors had affinities that were the same as the parent receptor, i.e. ␣ 2A (3L C ) had the same affinity for phentolamine as ␣ 2A AR, and ␣ 2C (3L A ) had the same affinity as ␣ 2C AR. These data indicate that substitution of the third intracellular loop had no significant conformational effect on the antagonist binding domains, which are localized within the transmembrane-   ) chimera had an epinephrine binding affinity that was not different than wild-type ␣ 2C AR (ϳ125 nM). A similar pattern was found with the agonist UK14304, except that only a ϳ2-fold increase in K i occurred with the ␣ 2A (3L C ) receptor (data not shown). The functional integrity of the receptors was determined in adenylyl cyclase assays, assessing the inhibition of forskolin-stimulated activity by epinephrine. Each chimeric receptor inhibited adenylyl cyclase by Ͼ65%, (Table II, (Ϫ) agonist column). Interestingly, the EC 50 for epinephrine with the ␣ 2C (3L A ) was lower than that for the other receptors. The above studies indicated that each of the chimeric receptors bound agonist with high affinity and functionally coupled to the inhibition of adenylyl cyclase. Whole cell receptor phosphorylation studies were then undertaken to test our hypotheses concerning agonist-promoted phosphorylation. Cells expressing each hemagglutinin-tagged receptor were incubated with [ 32 P]orthophosphate, treated with vehicle or the agonist epinephrine, and receptor-purified by immunoprecipitation using an epitope tag-specific antibody. Results of these studies are shown in Fig. 2. We obtained results similar to those that we previously reported for the two wild-type ␣ 2A AR and ␣ 2C AR (see Introduction). Thus, the ␣ 2A AR underwent agonist-promoted phosphorylation while the ␣ 2C AR did not. However, substitution of the ␣ 2C AR third intracellular loop into the ␣ 2A AR resulted in a receptor that underwent agonist-promoted phosphorylation equivalent to that of wild-type ␣ 2A AR. This strongly indicates that the third loop of the ␣ 2C has residues that can be phosphorylated, but that the intracellular milieu of the ␣ 2C outside of the third loop and/or the conformational changes in the loop, induced by agonist binding within the pocket established by the transmembrane-spanning domains, do not favor phosphorylation. In either case, one might predict that the ␣ 2A AR third loop would not undergo phosphorylation within the context of the ␣ 2C AR. This turned out to be so, as in parallel studies the ␣ 2C (3L A ) receptor failed to undergo agonistpromoted phosphorylation (Fig. 2).
We next considered whether these gain/loss phosphorylation phenotypes were also evident in functional (inhibition of adenylyl cyclase) desensitization studies. Cells in culture were exposed to vehicle or epinephrine, washed, membranes prepared, and agonist-mediated inhibition of forskolin-stimulated adenylyl cyclase activities determined. These results are shown in Table II and Fig. 3. We have previously shown in recombinantly expressing CHW and Chinese hamster ovary cells that ␣ 2A AR undergo such desensitization, manifested by a 4 -5-fold increase in the EC 50 with little or no significant change in the maximal extent of adenylyl cyclase inhibition (12). On the other hand, desensitization of ␣ 2C AR in Chinese hamster ovary cells is not observed under identical conditions. As shown, these phenotypes are also true in recombinantly expressing CHW cells. Of note, despite the fact that phosphorylation occurs with the ␣ 2A (3L C ) receptor, no agonist-promoted desensitization of this receptor was detected (Fig. 3C). Additionally, substitution of the ␣ 2A AR third loop into the ␣ 2C AR had no effect on the latter's lack of functional desensitization (Fig. 3D).
The ␣ 2A and ␣ 2C receptors have highly similar first and second intracellular loops. The major difference in the short cytoplasmic tail is the lack of a palmitoylated cysteine of ␣ 2C AR. We have previously shown that this difference in the tail does not alter agonist-promoted desensitization (20). We thus considered that the conformation established by the transmembrane-spanning backbone of the parent receptor dictates phosphorylation status of the third intracellular loop. This hypothesis was further explored using agonists with substantial differences in structure. Since the extent of GRK2-mediated phosphorylation of the ␤ 2 AR has been shown to be directly related to the agonist intrinsic activity (9), we first delineated the intrinsic activities of each agonist employed by measuring [ 35 S]GTP␥S binding. The agents ranged from being full ago-FIG. 2. Agonist-promoted phosphorylation of wild-type and chimeric ␣ 2 ARs. In A, a representative autoradiogram from a whole cell phosphorylation experiment is shown. In B, mean results from four independent experiments are shown. For the ␣ 2C and ␣ 2C (3L A ) receptors, no receptor band was identified under basal or agonist-exposed conditions; thus, no quantitation is shown.

TABLE II
Functional coupling and desensitization of wild-type and chimeric receptors Cells were exposed to carrier or 10 M epinephrine for 15 min, washed, membranes prepared, and adenylyl cyclase activities determined as described under "Experimental Procedures." Results are from four to six independent experiments. *, p Ͻ 0.01 compared to untreated values. nists to weak partial agonists, with the lowest intrinsic activity being 0.06. For para-aminoclonidine, SKF43315, BHT920, and oxymetazoline, the extent of agonist-promoted phosphorylation appeared to be related to the intrinsic activity of the agonist (Fig. 4). On the other hand, the ␣ 1 AR agonist SKF89748, which displays weak partial agonist activity at the ␣ 2A AR (intrinsic activity ϭ 0.06), evoked phosphorylation that was greater than the phosphorylation evoked by oxymetazoline, which has an intrinsic activity of 0.30. Additionally, UK14304, which was a full agonist, failed to promote phosphorylation to an extent similar to the other full agonist para-aminoclonidine, or the other near full agonists (Fig. 4). Taken together, it appears that phosphorylation of the ␣ 2 AR evoked by agonist is highly sensitive to the conformation of the third intracellular loop induced/stabilized by agonist to such an extent that these properties dictate the extent of phosphorylation of the loop when phosphoacceptors are present, and are the basis for subtypespecific phosphorylation.

DISCUSSION
Although short term agonist-promoted desensitization has been observed with many G protein-coupled receptors, about one-third of those tested appear not to display this regulatory behavior. One might consider, then, that certain cellular events controlled by some receptors are so critical that even under circumstances of prolonged agonist exposure receptor function is maintained.
For the three ␣ 2 AR subtypes, initial studies revealed similar binding affinities for endogenous catecholamines and similar functional coupling to the inhibition of adenylyl cyclase via G i coupling. Subsequent studies have revealed important differences between the subtypes in regards to coupling to G s (10), stimulation of intracellular calcium (21), membrane insertion and trafficking (22), agonist-promoted down-regulation (12), and agonist-promoted phosphorylation and desensitization (12)(13)(14). Desensitization of the ␣ 2A AR was found to be the most robust of the three subtypes, followed by the ␣ 2B AR (12). The ␣ 2C AR failed to display functional desensitization (12) or phosphorylation (13) due to agonist exposure in recombinant cells. Site-directed mutagenesis of the human ␣ 2A AR has shown that agonist-promoted phosphorylation is confined to the third in-tracellular loop (7) at four serines (15) (residues 296 -299, see Fig. 1). Each serine contributes ϳ25% of the total phosphorylation. However, all four serines must be phosphorylated for agonist-promoted functional desensitization to occur (15). In these studies, the finding that partial phosphorylation did not cause "partial" desensitization is consistent with the notion that the non-visual arrestins have strict requirements for binding to GRK-phosphorylated substrates leading to decreases in coupling to G i . This notion supports the findings of the current work, where we observed phosphorylation of ␣ 2A (3L C ), but a FIG. 3. Agonist-promoted desensitization of wild-type and chimeric ␣ 2 ARs. Cells expressing the indicated receptor were treated with vehicle or epinephrine (Epi.) for 15 min, washed, membranes prepared, and agonist-promoted inhibition of forskolin-stimulated adenylyl cyclase activities determined. Only the ␣ 2A AR displayed desensitization, manifested by a rightward shift in the doseresponse curve. Shown is a representative set of experiments. See Table II  failure of this receptor to display agonist-promoted desensitization. Presumably, the third loop conformation attained is adequate for GRK phosphorylation but not subsequent arrestin binding.
Using the current approach, it might be expected that substitution of the ␣ 2A third loop into the ␣ 2C receptor would confer phosphorylation, based on the assumption that the primary structure of this loop imparts the relevant features for the process. Similarly, substitution of the ␣ 2C third intracellular loop into the ␣ 2A would be expected to result in a receptor that failed to undergo agonist-promoted phosphorylation. Instead, we found that the presence or absence of phosphorylation was dependent on the subtype backbone rather than the third intracellular loop. The fact that agonists with different structures promoted markedly divergent degrees of phosphorylation of ␣ 2A AR (which was unrelated to intrinsic activity) supports the notion that the conformation stabilized by agonist is highly variable and has an important impact on third loop phosphorylation. Interestingly, although we observed these unexpected gain and loss phenotypes, and the differential phosphorylation of ␣ 2A AR by various agonists, we failed to observe phosphorylation of ␣ 2C AR due to exposure to any agonist tested (data not shown). Thus, although the ligand binding pocket of the ␣ 2A AR appears to be amenable to modification based on agonist structure, a similar scenario may not exist for ␣ 2C AR. However, since the highly phosphorylatable ␣ 2A AR third intracellular loop fails to undergo phosphorylation in the context of the ␣ 2C AR, the importance of residues outside of the loop is nonetheless clear.
These results support the concept that agonists can stabilize receptor conformation in highly specific ways. This is illustrated in Fig. 5. Here, agonist 1 is shown stabilizing a receptor in a specific R* 1 conformation. Such a conformation may allow for effective coupling to G protein, but not for phosphorylation, or some other process. These processes might include other events that are agonist-promoted, such as coupling to a second G protein, internalization, or down-regulation. In this model, other conformations, such as R* 2 -R* 4 , might have such properties, and depending on the agonist, specific conformations and thus pathways or properties can be realized. Indeed, we have previously shown that ␣ 2 AR coupling to G s (a somewhat inefficient process) is highly dependent on agonist structure (23). For drug development, this is particularly important in that it appears to be possible to design agonists with sufficient intrinsic activities that fail to undergo desensitization. Our findings also suggest that it is overly simplistic to consider agonist-based function to be strictly based on a "universal" active conformation.
In summary, using chimeric-mutagenesis, we have shown that the third intracellular loops of both the ␣ 2A AR and ␣ 2C AR have residues that can be phosphorylated during short term agonist exposure. However, such phosphorylation is present only in the context of the ␣ 2A AR, and is indeed absent when these loops are in the context of the ␣ 2C AR backbone. These results indicate that the subtype ligand binding pockets, which direct third loop conformational changes, have a major influence on subtype-specific phosphorylation of the ␣ 2 AR. The agonist-stabilized conformations (R* 1--R* 6 ) can be different for each agonist, such that differential G protein coupling, phosphorylation or other events are realized. In this example, agonist 2 does not evoke phosphorylation or coupling to G protein B.