G protein-coupled receptor kinase specificity for phosphorylation and desensitization of alpha2-adrenergic receptor subtypes.

The α2-adrenergic receptor (α2AR) subtype α2C10 undergoes rapid agonist-promoted desensitization which is due to phosphorylation of the receptor. One kinase that has been shown to phosphorylate α2C10 in an agonist-dependent manner is the βAR kinase (βARK), a member of the family of G protein-coupled receptor kinases (GRKs). In contrast, the α2C4 subtype has not been observed to undergo agonist-promoted desensitization or phosphorylation by βARK. However, the substrate specificities of the GRKs for phosphorylating α2AR subtypes are not known. We considered that differential capacities of various GRKs to phosphorylate α2C10 and α2C4 might be a key factor in dictating in a given cell the presence or extent of agonist-promoted desensitization of these receptors. COS-7 cells were co-transfected with α2C10 or α2C4 without or with the following GRKs: βARK, βARK2, GRK5, or GRK6. Intact cell phosphorylation studies were carried out by labeling cells with 32Pi, exposing some to agonist, and purifying the α2AR by immunoprecipitation and SDS-polyacrylamide gel electrophoresis. βARK and βARK2 were both found to phosphorylate α2C10 to equal extents (>2-fold over that of the endogenous kinases). On the other hand, GRK5 and GRK6 did not phosphorylate α2C10. In contrast to the findings with α2C10, α2C4 was not phosphorylated by any of these kinases. Functional studies carried out in transfected HEK293 cells expressing α2C10 or α2C4 and selected GRKs were consistent with these phosphorylation results. With the marked expression of these receptors, no agonist-promoted desensitization was observed in the absence of GRK co-expression. However, desensitization was imparted to α2C10 by co-expression of βARK but not GRK6, while α2C4 failed to desensitize with co-expression of βARK. These results indicate that short term agonist-promoted desensitization of α2ARs by phosphorylation is dependent on both the receptor subtype and the expressed GRK isoform.

The ␣ 2 -adrenergic receptor (␣ 2 AR) subtype ␣ 2 C10 undergoes rapid agonist-promoted desensitization which is due to phosphorylation of the receptor. One kinase that has been shown to phosphorylate ␣ 2 C10 in an agonistdependent manner is the ␤AR kinase (␤ARK), a member of the family of G protein-coupled receptor kinases (GRKs). In contrast, the ␣ 2 C4 subtype has not been observed to undergo agonist-promoted desensitization or phosphorylation by ␤ARK. However, the substrate specificities of the GRKs for phosphorylating ␣ 2 AR subtypes are not known. We considered that differential capacities of various GRKs to phosphorylate ␣ 2 C10 and ␣ 2 C4 might be a key factor in dictating in a given cell the presence or extent of agonist-promoted desensitization of these receptors. COS-7 cells were co-transfected with ␣ 2 C10 or ␣ 2 C4 without or with the following GRKs: ␤ARK, ␤ARK2, GRK5, or GRK6. Intact cell phosphorylation studies were carried out by labeling cells with 32 P i , exposing some to agonist, and purifying the ␣ 2 AR by immunoprecipitation and SDS-polyacrylamide gel electrophoresis.
␤ARK and ␤ARK2 were both found to phosphorylate ␣ 2 C10 to equal extents (>2-fold over that of the endogenous kinases). On the other hand, GRK5 and GRK6 did not phosphorylate ␣ 2 C10. In contrast to the findings with ␣ 2 C10, ␣ 2 C4 was not phosphorylated by any of these kinases. Functional studies carried out in transfected HEK293 cells expressing ␣ 2 C10 or ␣ 2 C4 and selected GRKs were consistent with these phosphorylation results. With the marked expression of these receptors, no agonist-promoted desensitization was observed in the absence of GRK co-expression. However, desensitization was imparted to ␣ 2 C10 by co-expression of ␤ARK but not GRK6, while ␣ 2 C4 failed to desensitize with co-expression of ␤ARK.
These results indicate that short term agonist-promoted desensitization of ␣ 2 ARs by phosphorylation is dependent on both the receptor subtype and the expressed GRK isoform.
During continuous activation by agonists, many G proteincoupled receptors display the property of desensitization, which is manifested as a waning of receptor function over time (1). The ␣ 2 adrenergic receptors (␣ 2 AR) 1 signal via multiple effectors, including the inhibition of adenylyl cyclase by coupling to the inhibitory G protein G i . With the ␣ 2A AR (also denoted as human ␣ 2 C10), this signaling has been shown to undergo such desensitization after brief periods of agonist exposure (2,3). The primary mechanism of rapid desensitization of ␣ 2 C10 appears to be receptor phosphorylation, which ultimately leads to receptor-G protein uncoupling. Several lines of evidence implicate the ␤-adrenergic receptor kinase or a related G protein-coupled receptor kinase (GRK), as the kinase responsible for desensitization of ␣ 2 C10. Desensitization of ␣ 2 C10 occurs after seconds to minutes of agonist exposure, requires saturating concentrations of agonist, is inhibited by the ␤ARK inhibitor heparin, and is ablated in a mutant receptor lacking phosphorylation sites in the third intracellular loop (2). In an in vitro reconstituted phospholipid vesicle system, purified ␣ 2 C10 has been shown to undergo phosphorylation by the ␤-adrenergic receptor kinase (␤ARK) in an agonist-dependent manner similar to that found with the ␤ 2 AR (4). In addition, a peptide derived from the ␣ 2 C10 third intracellular loop has been found to be an excellent substrate for ␤ARK phosphorylation in vitro (5), and intact cell agonist-promoted phosphorylation of the receptor is enhanced by co-expression of ␤ARK (6). Finally, it has been shown recently that removal of specific serines by mutagenesis in a predicted ␤ARK consensus sequence in the ␣ 2 C10 ablates agonist-promoted desensitization and receptor phosphorylation (7). Taken together, these results implicate phosphorylation of ␣ 2 C10 by ␤ARK as one of the key components in short term agonist-promoted desensitization.
However, ␤ARK is only one of several members of a family of kinases, termed GRKs, which phosphorylate the agonist-occupied form of G protein-coupled receptors (8). Other members of this kinase family that have been cloned and found to have diverse tissue distributions include ␤ARK2 (GRK3), GRK5, and GRK6 (9 -11). It is not known whether these other kinases phosphorylate ␣ 2 C10 in a manner similar to that of ␤ARK (GRK2). In regard to ␣ 2 ARs, another subtype, the human ␣ 2C AR (␣ 2 C4), fails to undergo agonist-promoted phosphorylation or functional desensitization in model systems examined thus far (3,6). This is intriguing, because the third intracellular loop of this receptor has a number of serines or threonines that may be potential GRK phosphorylation sites. It is not known whether ␣ 2 C4 could be a substrate for some of these other GRKs, such that the lack of desensitization/phosphorylation reported to date may be due to lack of expression of such kinases in the cell systems utilized.
We therefore undertook the current studies to delineate whether ␣ 2 C10 and ␣ 2 C4 phosphorylation and desensitization is GRK isoform-specific. The studies were designed to answer two questions: 1) which GRKs phosphorylate ␣ 2 C10 and 2) do any of the known GRKs phosphorylate ␣ 2 C4? To approach this, we co-expressed each receptor subtype individually with each of the four kinases and assessed agonist-promoted phosphorylation of the receptor in the intact cell setting. To confirm the significance of such phosphorylation (or lack thereof), we also developed a transient expression system whereby functional desensitization of each of the two ␣ 2 AR subtypes was determined in cells with or without co-transfection with a given GRK. The results indicate that agonist-promoted ␣ 2 AR desensitization is both receptor subtype and GRK isoform selective.

EXPERIMENTAL PROCEDURES
Tissue Culture and Transfections-COS-7 cells were maintained in Dulbecco's modified Eagle's medium and HEK293 cells were maintained in Earle's minimal essential medium, each supplemented with 10% fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin at 37°C, 5% CO 2 . The DEAE-dextran method of transfection was used to introduce plasmid DNA into each cell type for transient expression. The common expression vector utilized for expression of the ␣ 2 ARs and GRKs was pBC12BI as reported previously (7). Therefore, for COS-7 cells, 10 g of pBC-C10 or pBC-C4 and 10 g of pBC12BI containing bovine ␤ARK, bovine ␤ARK2, human GRK5, or human GRK6 cDNAs were transfected for co-expression of the ␣ 2 ARs and the GRKs. A transient expression system was utilized to examine functional desensitization in HEK293 cells. Since HEK293 cells do not contain the large T antigen, they do not transiently express cDNAs under control of the SV40 promoter to an appreciable extent. Therefore, to obtain transient expression in HEK293 cells using the above vectors, we co-transfected the pRSV-T antigen vector (12), which provides for expression of the large T antigen. In addition, an expression vector containing the D 1 A dopamine receptor cDNA, pRKD 1A R (13), was cotransfected in the HEK293 studies. This receptor was used to elicit a stimulatory effect on adenylyl cyclase for subsequent inhibition by the ␣ 2 ARs.
Western Analysis-For detection of receptor protein by immunoblotting, cells were washed with PBS and scraped in 5 mM Tris-HCl, 2 mM EDTA, pH 7.4, containing the protease inhibitors benzamidine (10 g/ml), soybean trypsin inhibitor (10 g/ml), and leupeptin (5 g/ml), and the particulates centrifuged at 40,000 ϫ g for 10 min at 4°C. The supernatant was then aspirated, and the pellets were solubilized directly in Laemmli sample buffer (50 mM Tris, 100 mM dithiothreitol, 2% SDS, 10% glycerol, 0.1% bromphenol blue). For assessment of GRK expression, detached whole cells were homogenized with a Brinkman Polytron and sonicated in sample buffer. Proteins were fractionated on a 10% polyacrylamide gel and then transferred onto a nitrocellulose membrane (Protran, Schleicher & Schuell) using a Bio-Rad Transblot apparatus. The Western blot was performed by blocking the filters with 5% non-fat dry milk and incubating with the following antibodies: 1:2000 of antisera containing polyclonal antibody raised to ␤ARK to identify ␤ARK and ␤ARK2, 1:1000 of antisera containing polyclonal antibody raised to GRK6 to identify GRK5 and GRK6, and 1:8000 dilution of antisera containing polyclonal antibody raised against peptides localized in the third intracellular loops of ␣ 2 C4 or ␣ 2 C10 (14). The blots were incubated with an anti-rabbit horseradish peroxidase-conjugated second antibody and then developed using enhanced chemiluminescence (DuPont NEN).
Phosphorylation and Immunoprecipitation-COS-7 cells transiently co-expressing either ␣ 2 C10 or ␣ 2 C4 along with either ␤ARK, ␤ARK2, GRK5, or GRK6 were incubated with [ 32 P]orthophosphate (ϳ2.4 mCi/ 150-cm 2 plate) for 2 h at 37°C, 5% CO 2 . Cells were then incubated for 20 min in either medium alone or medium plus agonist (10 M UK14304), washed five times with ice-cold PBS, and scraped in buffer containing 10 mM Tris-HCl, pH 7.4, 5 mM EDTA, 5 mM EGTA, 10 mM NaF, and 10 mM sodium pyrophosphate and the aforementioned protease inhibitors, which were included in this and all subsequent steps. Particulates were centrifuged at 40,000 ϫ g for 10 min at 4°C, and the resulting pellet was resuspended in the previously mentioned buffer, sonicated for 15 s, and centrifuged once again. ␣ 2 C10 and ␣ 2 C4 were then purified by immunoprecipitation as described previously (7,15). Briefly, membranes were solubilized by stirring in PBS containing 1% Triton X-100, 0.05% SDS, 1 mM EDTA, 1 mM EGTA for 2 h at 4°C. Unsolubilized material was removed by centrifugation at 40,000 ϫ g for 20 min at 4°C, and the solubilized material was incubated with preimmune serum and protein A-Sepharose beads for 30 min at room temperature. The beads containing nonspecific immunoprecipitant were removed by brief centrifugation, and the remaining supernatant was incubated with a 1:200 dilution of anti-␣ 2 C10 or anti-␣ 2 C4 antisera, as appropriate, and protein A-Sepharose beads for 16 h at 4°C. The beads were washed five times, sonicated in SDS-sample buffer, and removed by centrifugation. Protein containing equal amounts of receptor was fractionated on a 10% SDS-polyacrylamide gel. Autoradiography was used to detect phosphorylation of receptors, and the amount of radioactivity was quantitated on a Molecular Dynamics Phosphor-Imager with ImageQuant software. Using the above method, there was no detectable signal when nontransfected cells were used. For presentation purposes, autoradiograms were produced by exposing the gels to x-ray film for ϳ16 h.
Functional Desensitization and cAMP Determination-In order to delineate the relevance of the results from receptor phosphorylation experiments, a system to assess receptor function and desensitization in HEK293 cells was developed. HEK293 cells were transiently transfected such that either ␣ 2 C4 or ␣ 2 C10 was expressed along with the D 1 A dopamine receptor and a given GRK. Thus, ␣ 2 AR-mediated inhibition of dopamine-stimulated cAMP accumulation was utilized as the assay for assessment of ␣ 2 AR function. Using such an approach, dopamine (0.5 M) stimulated transfected HEK293 cell cAMP levels ϳ5-fold over basal levels. Co-incubation with the ␣ 2 AR agonist UK14304 inhibited dopamine stimulated cAMP accumulation by ϳ70% in a dose-dependent manner. (In nontransfected cells, neither dopamine or UK14304 had any effect on cAMP accumulation). For agonist-promoted desensitization studies, cells were exposed to the same conditions as those of the phosphorylation studies (see above) except that the hydrophilic ␣ 2 AR agonist PGE-6201204 was used. After washing five times in cold PBS, cells were gently tapped to dislodge them from the flask, centrifuged at 400 ϫ g at 4°C, and resuspended in Earle's minimal essential medium containing 100 M ascorbic acid. Cells were then incubated for 5 min in a 200-l volume of medium containing 0.5 M dopamine, 100 M ascorbic acid, and eight concentrations (0 -1 mM) of UK14304. Reactions were terminated by the addition of 20 l of 1.0 N HCl. Cellular cAMP production was then measured using a radioimmunoassay as described previously (16).
Miscellaneous-Protein was determined for all assays using the bicinchoninic acid method (17), with BSA used as standard. ␣ 2 AR expression was determined by radioligand binding using [ 3 H]yohimbine, with phentolamine to determine nonspecific binding, as described previously (3). Dose-response curves for cAMP inhibition were fit by a nonlinear iterative least squares technique using software from Graph-Pad (San Diego, CA).
Materials-Tissue culture supplies were purchased from JRH Biosciences and radioisotopes were from DuPont NEN. The cDNAs and the antisera for the various GRKs were kindly provided by Dr. J. Benovic, Thomas Jefferson University, Jefferson Medical College, and the pRSV-T antigen vector was obtained from Dr. R. Reed, The Johns Hopkins University School of Medicine. UK14304 was from Research Biochemicals International (Natick, MA) and PGE-6201204 was from Procter and Gamble Pharmaceuticals (Cincinnati, OH). All other reagents were purchased from sources listed previously (2,7,15).

RESULTS
To delineate which GRKs phosphorylate the ␣ 2 C10 and ␣ 2 C4 receptors, these receptors were transiently expressed in COS-7 cells along with ␤ARK, ␤ARK2, GRK5, or GRK6. Expression of the receptors as assessed by [ 3 H]yohimbine binding was typically ϳ5 pmol/mg. Expression of the GRKs was assessed by Western blots (Fig. 1). As is shown, using the methods outlined and the antisera directed against ␤ARK (which also recognizes ␤ARK2), ␤ARK was found to be expressed in nontransfected COS-7 cells. Upon transfection with pBC-␤ARK, expression was increased Ͼ15-fold. Under these conditions, ␤ARK2 was not detected in nontransfected cells. However, ␤ARK2 was clearly expressed after transfection with pBC-␤ARK2. The results of Western blots for GRK5 and GRK6, using antisera directed against GRK6, which also recognizes GRK5, are shown in Fig. 1B. A weak band, which could represent either kinase, was noted in nontransfected cells. Upon transfection with the GRK5 or GRK6 constructs, expression of the respective kinases was markedly enhanced. Western blots of ␣ 2 C10 and ␣ 2 C4 in membranes isolated from COS-7 cells following treatment in either medium alone or medium plus agonist for 20 min were performed in order to confirm that the receptors are recognized by the antibodies under both conditions. As seen in Fig. 2, ␣ 2 C10 and ␣ 2 C4 are recognized by their respective antibodies in the agonist occupied and nonoccupied forms of the receptors to the same extent.
To determine whether enhanced expression of the various kinases augments phosphorylation of ␣ 2 C10 and ␣ 2 C4, whole cell phosphorylation studies were carried out using COS-7 cells separately expressing the two receptor subtypes and overexpressing each GRK. Fig. 3A shows an autoradiograph of a representative experiment with ␣ 2 C10, and Fig. 3B summarizes the results from multiple experiments. It is clear that there is detectable agonist-promoted phosphorylation of ␣ 2 C10 in COS-7 cells not transfected with any GRK. Cells overexpressing ␤ARK or ␤ARK2, however, showed higher levels of receptor phosphorylation (220 Ϯ 30% for ␤ARK and 210 Ϯ 30% for ␤ARK2 co-transfections, as compared with non-GRK-transfected, p Ͻ 0.01). In contrast, levels of ␣ 2 C10 phosphorylation did not change upon overexpression of GRK5 or GRK6 (i.e., 90 Ϯ 30% for GRK5 and 80 Ϯ 50% for GRK6, p ϭ not significant). Identical experiments were performed to determine whether ␣ 2 C4 is phosphorylated by agonist under conditions of overexpression of these kinases. Fig. 4A shows a representative experiment. In the absence of overexpression of a GRK, ␣ 2 C4 does not appear to undergo any agonist-mediated phosphorylation. Nor is this receptor phosphorylated under any of the other circumstances tested (overexpression of ␤ARK, ␤ARK2, GRK5, or GRK6). ␣ 2 C10 co-expressed with ␤ARK was used as a positive control for these studies. We have interpreted the above experiments as being consistent with ␣ 2 C10 being a substrate for ␤ARK and ␤ARK2, but not GRK5 or GRK6. For ␣ 2 C4, none of the kinases appear to phosphorylate the receptor under the conditions studied.
From the results of the above phosphorylation experiments, we would predict that overexpression of certain kinases (such as ␤ARK) would augment agonist-promoted desensitization of ␣ 2 C10, while such overexpression of other kinases (such as GRK6) would not. Furthermore, for ␣ 2 C4, which does not undergo agonist-promoted phosphorylation, we would not expect to see a gain of desensitization with overexpression of any GRK. While COS-7 cells provide for the high levels of expression of receptors and kinases required for the phosphorylation studies, functional studies are problematic, since transfected ␣ 2 AR expressed in COS-7 cells mediate a complex pattern of cAMP modulation due to dual coupling to G s and G i (18 -20). A model system was thus developed in HEK293 cells. With transient expression of ␣ 2 AR in these cells, agonist-mediated inhibition of cAMP is easily observed. With overexpression of Following transfection of the various GRKs into COS-7 cells, expression was confirmed by Western analysis of protein isolated from whole cells. A illustrates overexpression of ␤ARK and ␤ARK2 relative to that contained in nontransfected (NT) cells. Antisera directed against ␤ARK (which also recognizes ␤ARK2) was used in these experiments. B is a representative Western blot which shows a weak band in the nontransfected (NT) lane (which could represent either GRK5 or GRK6) and overexpression of both GRK5 and GRK6 after transfection. The antisera used for the blots in B is directed against GRK6, but also recognizes GRK5.
FIG. 2. Western analysis of ␣ 2 C4 and ␣ 2 C10 in membranes isolated from COS-7 cells following treatment with medium alone or medium plus agonist. To confirm that nonagonist and agonist occupied forms of the ␣ 2 ARs are recognized equally by the antisera used for immunoprecipitation in phosphorylation experiments, Western blots were performed with membranes isolated from cells expressing ␣ 2 C4 and ␣ 2 C10 following treatment with medium alone or medium plus 10 M UK14304 for 20 min. The antisera specific for ␣ 2 C10 or ␣ 2 C4 was as described in Methods. Shown is a single experiment representative of three performed.
FIG. 3. Agonist-promoted phosphorylation of ␣ 2 C10 co-expressed with and without the various GRKs. Whole cell phosphorylation experiments were performed using COS-7 cells expressing ␣ 2 C10 and overexpressing each GRK. Cells were incubated with [ 32 P]P i for 2 h, exposed to media alone (Ϫagonist) or media plus 10 M UK14304 (ϩagonist) for 20 min and then purified by immunoprecipitation using anti-␣ 2 C10 antisera. A shows an autoradiograph of a representative experiment, and B summarizes the results from three separate experiments. While there is detectable phosphorylation of ␣ 2 C10 in COS-7 cells not transfected with any GRK, cells overexpressing ␤ARK or ␤ARK2 show higher levels of receptor phosphorylation (i.e. 220 Ϯ 30% for ␤ARK and 210 Ϯ 30% for ␤ARK2). In contrast, levels of ␣ 2 C10 phosphorylation do not change upon overexpression of GRK5 or GRK6 (i.e. 90 Ϯ 30% for GRK5 and 80 Ϯ 50% for GRK6). The stoichiometry of ␣ 2 C10 phosphorylation by ␤ARK has been shown previously to be 4 mol of phosphate/mol of receptor (4,7). * ϭ p Ͻ 0.01 compared with ␣ 2 C10 without GRK transfection. ␣ 2 C10, the endogenous kinases were insufficient to provide for detectable agonist-promoted desensitization as indicated by no shift in the dose-response curve or a change in the maximal degree of inhibition. This then allowed us to assess the effect of a given kinase on ␣ 2 AR desensitization by also overexpressing the kinase. For these studies, two kinases, which gave different results in the phosphorylation experiments, ␤ARK and GRK6, were utilized. In all experiments, the D 1 A dopamine receptor was also expressed to provide for a receptor for stimulation of adenylyl cyclase. As is shown in Fig. 5, transfection of the ␤ARK and GRK6 constructs indeed resulted in an increase in expression of each kinase as determined by Western blots. Under the above conditions, agonist exposure for 20 min had no effect on ␣ 2 C10 function in the absence of GRK overexpression (Table I). Thus, no change in the EC 50 for UK14304-mediated inhibition of cAMP accumulation, nor in the maximal extent of inhibition, was noted (Table I). However, when ␤ARK was overexpressed, a clear desensitization of ␣ 2 C10 was observed. In these cells, the EC 50 increased ϳ6.4-fold after agonist exposure with a ϳ17% decrease in the maximal degree of inhibition (Table I). (It should be noted that the mean basal EC 50 for inhibition with ␤ARK co-expression trended toward being slightly lower than without ␤ARK, but this was not consistent and was not statistically significant). When examined at a submaximal concentration of UK14304 in the assay (10 nM), the observed shift in the dose response curve was found to result in ϳ50% desensitization of the receptor (from ϳ47 to ϳ25% inhibition). In contrast, cells overexpressing GRK6 displayed no agonist-promoted desensitization as indicated by no shift in the dose-response curve or change in the maximal degree of inhibition. These results are in agreement with what was found with the phosphorylation studies, in that agonistpromoted phosphorylation of ␣ 2 C10 was augmented by ␤ARK but not GRK6. For ␣ 2 C4, functional desensitization was not observed when ␤ARK was overexpressed (Table I), also consistent with the phosphorylation studies.

DISCUSSION
Several mechanisms responsible for agonist-promoted desensitization of ␣ 2 AR have been established (reviewed in Refs. 1 and 21). Four different processes have thus far been identified: phosphorylation, sequestration, down-regulation, and decreases in cellular G i . For those ␣ 2 AR subtypes that do display rapid agonist-promoted desensitization, the predominant mechanism responsible for such short term regulation is phosphorylation of the receptor (2,6,22). Down-regulation of receptor expression and changes in cellular G proteins are consequences of prolonged agonist exposure, and sequestration probably serves to dephosphorylate the receptor, but has not been clearly implicated in rapid desensitization.
That ␤ARK, or a related GRK, is the kinase that phosphorylates ␣ 2 C10 and mediates agonist-promoted desensitization was first suggested by in vitro studies with purified platelet ␣ 2A AR and purified bovine ␤ARK. In a reconstituted phospholipid vesicle system, ␣ 2A AR was shown to undergo phosphorylation by ␤ARK in an agonist-dependent fashion (4). However, the relevance of this finding to receptor regulation in intact cells was not clear, and indeed even the presence of agonistpromoted desensitization of ␣ 2 AR was debated. Subsequent studies using recombinant ␣ 2 C10 expressed in Chinese hamster fibroblasts (CHW cells) and CHO cells have clearly established that this receptor undergoes agonist-promoted desensitization (2,3,6). Such desensitization was found to occur rapidly (within minutes of agonist exposure) and correlated with receptor phosphorylation as assessed in intact cell studies. ␣ 2 C10 desensitization in CHW cells is blocked by heparin, a known inhibitor of ␤ARK phosphorylation, but not by inhibitors of protein kinases A or C (2). In vitro phosphorylation studies using synthetic peptides and purified ␤ARK have shown that a peptide with sequence identical to that of a portion of the third intracellular loop of ␣ 2 C10 is an excellent substrate for ␤ARK (5). A mutated ␣ 2 C10, lacking these residues in the third intracellular loop, was found to be functionally intact when expressed in CHW cells, but failed to display agonist-promoted desensitization (2). This mutated receptor also failed to undergo agonist-promoted phosphorylation in intact cells (2). A correlation between receptor phosphorylation and agonist-promoted desensitization of ␣ 2 AR has also been found in CHO cells recombinantly expressing the ␣ 2 AR subtypes. The ␣ 2 C10 was found to display agonist-promoted phosphorylation and functional desensitization, while ␣ 2 C4 did not phosphorylate or desensitize (3,6). In addition, in COS-7 cells it has been reported that co-expression of ␤ARK enhances FIG. 4. Agonist-promoted phosphorylation of ␣2C4 co-expressed with and without the various GRKs. Whole cell phosphorylation experiments were performed using COS-7 cells expressing ␣ 2 C4 and overexpressing each GRK as in Fig. 3. As a control for phosphorylation, ␣ 2 C10 ϩ ␤ARK was included in these experiments. A shows a representative experiment, while B summarizes the results of three independent experiments. In the absence of overexpression of a GRK, ␣ 2 C4 does not appear to undergo any agonist-mediated phosphorylation. Likewise, this receptor is not phosphorylated under any of the conditions tested (i.e. upon overexpression of ␤ARK, ␤ARK2, GRK5, or GRK6).
FIG. 5. Western analysis of ␤ARK and GRK6 expressed in HEK293 cells. To confirm overexpression of ␤ARK and GRK6 in HEK293 cells after transfection, a Western analysis was performed using protein isolated from whole cells as described under "Experimental Procedures." agonist-promoted phosphorylation of recombinantly expressed ␣ 2 C10 (6). Finally, recent studies utilizing site-directed mutagenesis of the predicted ␤ARK phosphorylation sites within the sequence EESSSS of the third intracellular loop of ␣ 2 C10 have been reported (7). In these studies, substitutions of selected serines with alanines were carried out and intact cell phosphorylation and desensitization in CHO cells assessed. The results indicated that all four serines (residues 296 -299) are phosphorylated by ␤ARK during agonist exposure and that such phosphorylation is obligatory for functional desensitization. Taken together, these studies strongly indicate phosphorylation by ␤ARK as a key mediator of short term agonistpromoted desensitization of ␣ 2 C10 (also further reviewed in Refs. 1 and 21).
Whether ␣ 2 C10 or ␣ 2 C4 serve as substrates for other members of the GRK family has not been investigated. Regarding ␣ 2 C10, we wondered whether other GRKs have the potential to phosphorylate this receptor, while with ␣ 2 C4 we were interested in whether any of these kinases could subserve this function. We considered that such kinase specificity might provide for cell-type differences in the desensitization response of ␣ 2 AR. Based on the diversity of primary structure of the known GRKs, differences in their requirement for substrate recognition/phosphorylation is clearly plausible (8). However, to date little is known about substrate specificities of the different kinases using intact receptors expressed in cells. To address this with the ␣ 2 ARs, we studied ␣ 2 C10, which has been shown to undergo desensitization, and ␣ 2 C4, which does not appear to undergo desensitization in the model systems examined thus far. Whole cell intact receptor phosphorylation studies were carried out in COS-7 cells co-expressing ␣ 2 C10 or ␣ 2 C4 with ␤ARK, ␤ARK2, GRK5, or GRK6 in the absence or presence of agonist. Both ␤ARK and ␤ARK2 were found to phosphorylate ␣ 2 C10 in an agonist-dependent manner, while GRK5 and GRK6 did not. ␣ 2 C4 was not phosphorylated by any of the kinases studied. To assess whether such phosphorylation (or lack thereof) had a physiologic correlate, functional studies were carried out in cells with or without co-expression of specific kinases. Consistent with what was found in the phosphorylation studies, ␣ 2 C10 desensitization was imparted by coexpression of ␤ARK, while there was no effect upon coexpression of GRK6. For ␣ 2 C4, ␤ARK overexpression did not induce desensitization. We conclude from these studies that ␣ 2 AR desensitization displays both receptor subtype and GRK isoform specificity. We and others had previously shown in CHO cells or COS-7 cells that ␣ 2 C10 underwent agonist-promoted phosphorylation and desensitization, but that ␣ 2 C4 does not (3,6). Based on the current study, it appears that this lack of desensitization is not due to a lack of expression of a particular kinase in these cells. Rather, it appears that ␣ 2 C4 is not a substrate for any of the kinases tested. For ␣ 2 C10, which was known to be phosphorylated by ␤ARK, it was not clear which of the other GRKs could also subserve this function. Based on the current study, ␣ 2 C10 appears to be a substrate for ␤ARK and ␤ARK2, but not GRK5 or GRK6.
GRK substrate specificity is consistent with a number of key differences among the mammalian kinases cloned to date. For example, the kinases appear to possess distinct mechanisms responsible for their localization to the membrane. ␤ARK and ␤ARK2 contain binding sites within their C terminus for the ␤␥ subunits of heterotrimeric G proteins (23,24), while GRK5 appears to be constitutively associated with the membrane possibly due to a stretch of basic amino acids localized in its C terminus, which interact with phospholipid head groups (25). The mechanism for membrane localization of GRK6 is distinct from the others in that it appears to involve palmitoylation (26). The substrate specificity of the various kinases, as determined by peptide studies, has also been shown to differ. While ␤ARK and ␤ARK2 preferentially phosphorylate serine or threonine residues C-terminal to acidic amino acids (5), GRK5 and GRK6 do not appear to display such a preference. Another difference among the kinases involves their regulation by autophosphorylation (27,28). While GRK5 is autophosphorylated (29), likely leading to its activation, ␤ARK, ␤ARK2, and GRK6 are not (28).
That one of the ␣ 2 AR subtypes does not undergo agonistpromoted phosphorylation by any of these GRKs furthers the notion that this property may be a distinguishing feature between structurally related receptor subtypes. ␣ 2 C10 and ␣ 2 C4 have similar affinities for endogenous catecholamines and both couple to G i and, though less efficiently, to G s as well. Yet ␣ 2 C4 lacks the molecular requirements for GRK phosphorylation, which has turned out to be the most obvious difference between the two receptors reported to date. Similarly, within the ␤AR subtypes, the ␤ 3 AR does not undergo rapid agonist-promoted phosphorylation/desensitization (30), which is in contrast to ␤ 1 and ␤ 2 AR. Thus, receptors such as ␣ 2 C4 and ␤ 3 AR may have evolved to be resistant to this adaptive response. That only some of the GRKs phosphorylate ␣ 2 C10 provides for yet another level of complexity. Based on the current findings it can be predicted that ␣ 2 C10 on cells expressing exclusively a GRK5/GRK6-type kinase would not be expected to undergo rapid desensitization while in another cell which expressed a ␤ARK/␤ARK2-type kinase, ␣ 2 C10 desensitization would occur. As we learn more about the localization of the different GRKs at the cellular level within various organs, these concepts can be assessed in regard to specific physiologic responses.
Reed for providing the pRSV-T vector. We also thank Cheryl Theiss and Elizabeth Donnelly for technical assistance and Katie Gouge for manuscript preparation.
Note Added in Proof-Recently, Cottecchia and colleagues have reported receptor subtype and GRK specificity for agonist-promoted desensitization of the ␣ 1 -adrenergic receptor (31).