G protein-coupled receptor kinase 5 regulates beta 1-adrenergic receptor association with PSD-95.

We previously reported that the beta(1)-adrenergic receptor (beta(1)AR) associates with PSD-95 through a PDZ domain-mediated interaction, by which PSD-95 modulates beta(1)AR function and facilitates the physical association of beta(1)AR with other synaptic proteins such as N-methyl-d-aspartate receptors. Here we demonstrate that beta(1)AR association with PSD-95 is regulated by G protein-coupled receptor kinase 5 (GRK5). When beta(1)AR and PSD-95 were coexpressed with either GRK2 or GRK5 in COS-7 cells, GRK5 alone dramatically decreased the association of beta(1)AR with PSD-95, although GRK2 and GRK5 both could be co-immunoprecipitated with beta(1)AR and both could enhance receptor phosphorylation in vivo. Increasing expression of GRK5 in the cells led to further decreased beta(1)AR association with PSD-95. Stimulation with the beta(1)AR agonist isoproterenol further decreased PSD-95 binding to beta(1)AR. In addition, GRK5 protein kinase activity was required for this regulatory effect since a kinase-inactive GRK5 mutant had no effect on PSD-95 binding to beta(1)AR. Moreover, the regulatory effect of GRK5 on beta(1)AR association with PSD-95 was observed only when GRK5 was expressed together with the receptor, but not when GRK5 was coexpressed with PSD-95. Thus, we propose that GRK5 regulates beta(1)AR association with PSD-95 through phosphorylation of beta(1)AR. Regulation of protein association through receptor phosphorylation may be a general mechanism used by G protein-coupled receptors that associate via PDZ domain-mediated protein/protein interactions.

PDZ domains, named after the first three proteins in which they were discovered (PSD-95/Dlg/ZO-1), bind to small specific primary sequences in their target proteins, typically a carboxyl-terminal peptide (1). PDZ domain-mediated protein/protein interactions have been identified that target many receptors, ion channels, enzymes, and other structural molecules (2,3). PDZ domain-containing scaffolding proteins have been shown to play very important roles in organizing signaling cascades in a specific spatial order, anchoring proteins in specific subcellular compartments, and regulating ion channel activity (2)(3)(4)(5)(6).
A variety of G protein-coupled receptors (GPCRs) 1 (including the ␤ 1 -adrenergic receptor (␤ 1 AR), ␤ 2 AR, vasoactive intestinal peptide, V2 vasopressin, metabotropic glutamate, serotonin, and somatostatin receptors) contain typical class I PDZ domain-interacting motifs of the form (S/T)X(V/I/L) at their carboxyl termini (7). ␤ 2 AR was the first GPCR reported to interact with a PDZ domain-containing protein (8). Agonist-stimulated binding of the first PDZ domain of the Na ϩ /H ϩ exchanger regulatory factor (NHERF) to the extreme C terminus of ␤ 2 AR provides a mechanism by which ␤ 2 AR controls Na ϩ /H ϩ exchanger NHE3 function in the kidney (8). Furthermore, interaction with NHERF has been reported to regulate the recycling of internalized ␤ 2 AR (9). Recently, additional PDZ domaincontaining proteins interacting with several distinct GPCRs have been identified. MUPP1, a multi-PDZ domain-containing protein, has been demonstrated to interact with three subtypes of serotonin receptors through its PDZ10 domain (10). SSTRIP has been identified as a novel binding partner for somatostatin receptor subtype 2 (11,12). Cortactin-binding protein 1 also has been reported to interact with somatostatin receptor subtype 2 in an agonist-dependent manner (13). Another PDZ domain-containing protein, PICK1 (protein interacting with C kinase), has been reported to interact with protein kinase C (PKC) and to regulate its phosphorylation of G protein-coupled metabotropic glutamate receptor subtype 7a (14). Using yeast two-hybrid screening, we recently identified PSD-95, a postsynaptic density-enriched protein, as a novel binding partner of ␤ 1 AR (15). PSD-95 belongs to the large family of membrane-associated guanylate kinase domain proteins that contain multiple PDZ domains. PSD-95 is known to interact with several proteins, including N-methyl-D-aspartate receptors (16), K ϩ channels (1), the receptor-tyrosine kinase ErbB4 (17), and nitric-oxide synthase (18,19), through its three PDZ domains. PSD-95 is thought to play critical roles in assembling these and other molecules into signaling complexes and regulating their signal transduction (20). Association with PSD-95 decreases agonist-induced ␤ 1 AR internalization, but has no effect on receptor desensitization or receptor-induced cAMP accumulation (15). This association provides a physical linkage between ␤ 1 ARs and other synaptic proteins, such as N-methyl-D-aspartate receptors, which are known to be regulated by ␤ 1 AR stimulation (21). Association with PSD-95 may also provide a molecular mechanism by which ␤ 1 AR may be localized to synapses and regulate synaptic plasticity.
If PDZ domain-containing proteins that interact with the carboxyl terminus of a GPCR via a PDZ domain do indeed serve as multivalent scaffolding proteins to selectively assemble and target receptor-containing signaling complexes to specific cellular areas (10,12,15), this assembly is likely to be highly regulated. The assembly of receptor signaling complexes on the cell surface is a dynamic process, and the complexes must be able to disassemble under certain physiological conditions. Despite the fact that several protein/protein interactions between GPCRs and PDZ domain-containing proteins have been well documented, little is known about how such PDZ domainmediated interactions might be regulated.
G protein-coupled receptor kinases (GRKs) constitute a family of seven serine/threonine-protein kinases that phosphorylate agonist-bound, activated GPCRs. GRK-mediated receptor phosphorylation rapidly initiates receptor desensitization and internalization through clathrin-mediated endocytosis by targeting the receptors for binding by arrestin proteins (22,23). It has been reported that in ␤ 2 AR, the serine/threonine-rich cytoplasmic tail is the target region phosphorylated by GRKs (24). In vitro experiments have indicated that GRK2 and GRK5 preferentially phosphorylate some of these sites (24). Both ␤ 1 AR and ␤ 2 AR have carboxyl-terminal PDZ target motifs that have been reported to bind to PDZ domain-containing proteins (8,15). NHERF binds to the DSLL-COOH motif of ␤ 2 AR, and PSD-95 binds to the ESKV-COOH motif of ␤ 1 AR. In both cases, mutation of the serine at position Ϫ2 eliminates PDZ domaincontaining protein binding. Interestingly, this Ser residue at position Ϫ2 in ␤ 2 AR has been mapped as a phosphorylation site for GRK5, but not for GRK2 (24). Even though nothing is known about the GRK phosphorylation sites in ␤ 1 AR, it has been reported that ␤ 1 AR can be phosphorylated and desensitized by both GRK2 and GRK5 in a manner similar to that for ␤ 2 AR (25). Considering that the protein sequences in ␤ 1 AR and ␤ 2 AR are highly conserved in the carboxyl-terminal tail, it is reasonable to believe that the Ser residue in the ESKV-COOH motif of ␤ 1 AR may also be a phosphorylation site for GRKs, and phosphorylation of this site may affect receptor interaction with its binding partner. Therefore, we investigated whether GRKs might regulate ␤ 1 AR interaction with PSD-95.
Cell Culture and Transfection-All tissue culture media and related reagents were purchased from Invitrogen. COS-7 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and penicillin/streptomycin in a 37°C incubator under 5% CO 2 . Cells in 100-mm dishes were transfected with LipofectAMINE (at a 4:1 ratio with DNA) according to the manufacturer's protocol. HEK293 cells were maintained under the same conditions as COS-7 cells, except that minimal essential medium was used. HEK293 cells were transfected with FuGene 6 (at a 2.5:1 ratio with DNA; Roche Molecular Biochemicals) according to the manufacturer's protocol. After transfection, cells were grown for 36 -48 h before agonist stimulation. Prior to stimulation, appropriately transfected cells were incubated in serum-free media (Dulbecco's modified Eagle's medium (for COS-7 cells) or minimal essential medium (for HEK293 cells) containing 0.1% bovine serum albumin and 10 mM HEPES, pH 7.4) for 60 min at 37°C. Agonist stimulation was performed at 37°C in these serum-free media for the times indicated in the figure legends.
Cellular Immunoprecipitations and Immunoblotting-For most immunoprecipitation experiments, COS-7 cells in 100-mm plates were scrapped into 1 ml of ice-cold lysis buffer A (20 mM HEPES, 150 mM NaCl, 2 mM EDTA, 10% glycerol, 0.5% Nonidet P-40, 5 g/ml aprotinin, 5 g/ml leupeptin, 20 g/ml phenylmethylsulfonyl fluoride, 10 mM NaF, and 1 mM sodium orthovanadate). The cells were lysed by incubation on ice for 30 min and then clarified by centrifugation at 21,000 ϫ g for 12 min at 4°C. The clarified supernatants were used in cellular co-immunoprecipitation experiments. 50 l of each supernatant were diluted into an equal amount of 3ϫ SDS-PAGE sample buffer and served as a whole cell extract control.
For protein cross-linking experiments, transfected COS-7 cells in 100-mm plates were incubated in serum-free medium for 60 min and then exposed to 10 M isoproterenol for 10 min. The cells were rinsed with cold Dulbecco's phosphate-buffered saline and replaced with 1 ml of cross-linking buffer (Dulbecco's phosphate-buffered saline containing 10 mM HEPES, pH 7.4, and a 2.5 mM concentration of the cell-permeable cross-linking reagent dithiobis(succinimidyl propionate) (Pierce)) (30). For the stimulated cells, 10 M isoproterenol was present in the cross-linking buffer. The cells were incubated for 30 min at room temperature with continuous slow rocking. The cross-linking reaction was terminated by quickly removing the cross-linking buffer and replacing it with 1 ml of ice-cold lysis buffer B (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 2 mM EDTA, 10% glycerol, 0.5% Nonidet P-40, and protease inhibitors). The cell lysates were clarified as described above, and the supernatants were used for immunoprecipitation experiments.
Clarified cell extract (1 ml) was then incubated with 25 l of anti-FLAG M2 affinity gel slurry at 4°C with gentle rotation for 4 h. Beads were washed four to five times with ice-cold lysis buffer, and the bound proteins were eluted with 50 l of 3ϫ SDS-PAGE sample buffer. Equivalent amounts of sample in 3ϫ SDS-PAGE sample buffer were resolved on pre-cast 4 -20% polyacrylamide gel (Invitrogen) and then transferred to nitrocellulose membrane by semidry blotting. Nitrocellulose membranes were blocked with 5% fat-free milk in TBST (20 mM Tris, pH 7.4, 500 mM NaCl, and 0.1% Tween 20) and incubated with the appropriate primary antibody at room temperature for 1 h. PSD-95 was detected with rabbit anti-Myc antibody (1:1000 dilution); ␤-arrestin-1 and ␤-arrestin-2 were detected with rabbit anti-His 6 antibody; GRK2 was detected with GRK2/3-specific monoclonal antibody C5/7; and GRK5 was detected with GRK4/5/6-specific monoclonal antibody A16/17. FLAGtagged receptor proteins in the immunoprecipitation were blotted with BioM2 IgG. After washing with TBST, membranes were incubated for 1 h with horseradish peroxidase-conjugated anti-mouse or anti-rabbit IgG secondary antibody (1:2000 dilution) or with horseradish peroxidase-conjugated streptavidin (for BioM2). Protein bands were visualized via SuperSignal chemiluminescence substrate (Pierce). Quantitation of band density was performed with a Bio-Rad Fluor-S MultiImager.
In Vivo Receptor Phosphorylation-COS-7 cells in 100-mm dishes were transiently transfected with FLAG-␤ 1 AR alone or with either GRK2 or GRK5. 48 h after transfection, cells were washed with phosphate-free Dulbecco's modified Eagle's medium and incubated with [ 32 P]orthophosphate to label the intracellular ATP pool as described previously (31). Cells were then stimulated with 10 M isoproterenol for 5 min and washed twice with ice-cold Dulbecco's phosphate-buffered saline. Cells were solubilized in ice-cold lysis buffer A, and receptors were immunoprecipitated with anti-FLAG M2 affinity gel. Immune complexes were resolved on 10% polyacrylamide gels. The gels were stained with Coomassie Blue, dried, and subjected to autoradiography. Receptor phosphorylation was quantitatively analyzed with a Molecular Dynamics PhosphorImager as described previously (25).
In Vitro Glutathione S-Transferase (GST) Fusion Protein Phosphorylation-GRK2 and GRK5 were purified from baculovirus-infected Sf9 insect cells as previously described (28,32). GST fusion proteins (GST, GST-␤ 1 CT (where ␤ 1 CT is the carboxyl terminus of ␤ 1 AR), and GST-␤ 2 CT) were purified from bacteria using glutathione-Sepharose 4B beads (Amersham Biosciences, Inc.) and resuspended in phosphatebuffered saline containing 0.5% Nonidet P-40 and protease inhibitors (15). To phosphorylate GST fusion proteins immobilized on the beads, 10 l of beads with bound fusion protein were washed three times with GRK kinase buffer (20 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 20 mM EDTA, and 1 mM dithiothreitol) and incubated with 0.5 g of purified GRK2 or GRK5 and [␥-32 P]ATP (60 M final concentration, 400 cpm/ pmol) in total of 40 l at 30°C for 30 min. Reactions were stopped with 2ϫ SDS sample buffer, and phosphoproteins were resolved by SDS-PAGE. The gels were dried and processed for autoradiography as described above.
Receptor Internalization-For receptor internalization assays, HEK293 cells in 100-mm dishes were transiently transfected with pcDNA3/FLAG-␤ 1 AR or pcDNA3/FLAG-␤ 2 AR in the absence and presence of GW1/Myc-PSD-95. One day after transfection, cells were split into polylysine-coated six-well plates (Biocoat) and grown overnight at 37°C. Cells were serum-starved for 1 h before stimulation with 10 M isoproterenol for 30 min at 37°C. Cells were placed on ice, and cellsurface FLAG-tagged receptors were detected with anti-FLAG M2 antibody, followed by fluorescein isothiocyanate-conjugated anti-mouse IgG as described previously (26). Receptor internalization is defined as the percentage loss of surface receptors caused by agonist stimulation, as measured by cell flow cytometry.

RESULTS
Expression of GRK5 Decreases PSD-95 Binding to ␤ 1 AR-To test the effect of GRKs on the association of ␤ 1 AR with PSD-95, we transfected COS-7 cells with FLAG-␤ 1 AR and Myc-PSD-95 in the absence or presence of either GRK2 or GRK5. As reported previously (15), in the absence of exogenous GRKs, immunoprecipitation of ␤ 1 AR followed by Western blotting revealed robust co-immunoprecipitation of a PSD-95⅐␤ 1 AR complex (Fig. 1A). Treatment with the ␤-adrenergic receptor agonist isoproterenol had no significant effect on this association in the absence of GRKs. However, expression of GRK5 dramatically decreased the association of PSD-95 with ␤ 1 AR, whereas the association of ␤ 1 AR with PSD-95 was not altered by coexpression of GRK2 (Fig. 1A). With an increased expression level, GRK5 further decreased the amount of PSD-95 binding to the receptor (Fig. 1B), whereas an increased expression level of GRK2 was ineffective (Fig. 1C).
Furthermore, the negative effect of GRK5 on PSD-95/␤ 1 AR binding was potentiated by agonist stimulation of ␤ 1 AR (Figs. 1A and 2). In the cells lacking GRK5, 10 min of stimulation with isoproterenol did not change the association of ␤ 1 AR with PSD-95. However, expression of even a small amount of GRK5 dramatically decreased the association of PSD-95 with ␤ 1 AR (Figs. 1B and 2). In the presence of GRK5, isoproterenol stimulation further decreased the amount of PSD-95 associated with the receptor (Figs. 1A and 2). With extended stimulation by agonist, the amount of PSD-95 bound to ␤ 1 AR was further decreased (Fig. 2). This agonist effect could result from 1) agonist stimulation activating the receptor and changing its conformation, thus presenting a better substrate for GRK5, or 2) activated GPCR enhancing GRK5 kinase activity (23). In both cases, agonist stimulation led to enhanced receptor phosphorylation by GRK5. These results suggest that receptor phosphorylation might be responsible for the GRK5 effect on PSD-95/␤ 1 AR binding. The reason why no agonist effect was observed in the cells in the absence of GRK5 may be because there is not a sufficient amount of endogenous GRK5 (compared with GRK2) in the COS-7 cells. In fact, Western blotting of COS-7 cell lysate suggested that there is a significantly higher level of GRK2 than of GRK5 in these cells (data not shown).
Both GRK2 and GRK5 Can Associate with ␤ 1 AR and Enhance Its Phosphorylation-It has been well documented that many GPCRs are substrates for both GRK2 and GRK5, as they can be phosphorylated by GRK2 or GRK5 both in vivo and in vitro (23). Furthermore, GRK2 and GRK5 have been demonstrated to associate with GPCRs in an agonist-dependent manner (30,31). Here we tested the effect of GRK2 and GRK5 on ␤ 1 AR phosphorylation and their association with ␤ 1 AR in COS-7 cells. As has been previously shown (25), both GRK2 and GRK5 can enhance the phosphorylation of ␤ 1 AR. In the cells expressing ␤ 1 AR alone, agonist stimulation caused little increase in receptor phosphorylation. However, we observed a dramatic increase in agonist-stimulated receptor phosphorylation when ␤ 1 AR was coexpressed with either GRK2 or GRK5 in COS-7 cells (Fig. 3A).
The interaction between GRKs and their substrate receptors is so weak (or transient) that co-immunoprecipitation of GRK5/ ␤ 1 AR is undetectable under standard immunoprecipitation conditions (data not shown). However, GRK5/␤ 1 AR association is detectable following treatment of the stimulated cells with the membrane-permeable cross-linking reagent dithiobis(succinimidyl propionate). As shown in Fig. 3B, when GRK2 or   FIG. 1. GRK5, but not GRK2, regulates the binding of PSD-95 to ␤ 1 ARs. COS-7 cells were transiently transfected with a combination of pcDNA3/FLAG-␤ 1 AR and GW1/Myc-PSD-95, together with pRK5/ GRK5, pRK5/GRK2, or empty vector (3 g) as indicated. For agonist stimulation, 48 h after transfection, cells were incubated with serumfree medium for 60 min and then exposed to 10 M isoproterenol (ISO) for 10 min at 37°C (stimulated cells). The FLAG-tagged receptor was immunoprecipitated (IP) with anti-FLAG M2 beads as described under "Experiment Procedures." Proteins bound to the beads were resolved by SDS-PAGE and Western-blotted with anti-Myc or anti-FLAG BioM2 antibody. GRK2 and GRK5 in the cell lysates were detected with a mixture of monoclonal anti-GRK2 and anti-GRK5 antibodies. Expression of PSD-95 and GRKs in the cell lysate is shown for comparison. The data shown are representative of four independent experiments. A, GRK5, but not GRK2, reduces the association of PSD-95 with ␤ 1 AR; B, an increased expression level of GRK5 decreases the association of PSD-95 with ␤ 1 AR; C, increased expression of GRK2 has no effect on the association of PSD-95 with ␤ 1 AR. IB, immunoblot. GRK5 was coexpressed with ␤ 1 AR in COS-7 cells, immunoprecipitation of the receptor revealed robust association of GRK2 and GRK5 with the receptor (Fig. 3B).
GRK5 Kinase Activity Is Required to Regulate the ␤ 1 AR Association with PSD-95-Since both GRK2 and GRK5 are coimmunoprecipitated with ␤ 1 AR, it is possible that the GRK5 inhibition of PSD-95/␤ 1 AR association could be due to competition between PSD-95 and GRK5 for binding to the receptor. To address this question, we used a GRK5 mutant, K215R (GRK5(DN)), which has no kinase activity. When ␤ 1 AR and PSD-95 were coexpressed with differing amounts of GRK5(DN), the association of PSD-95 with ␤ 1 AR was not significantly decreased (Fig. 4), even though GRK5(DN) still retains the same ability to co-immunoprecipitate with the receptor as wild-type GRK5 (data not shown). These data indicate that the ability GRK5 to reduce receptor association with PSD-95 is not due to competition between GRK5 and PSD-95 for receptor binding, but requires the protein kinase activity of GRK5. This conclusion is also supported by the preceding results that agonist stimulation potentiated the ability of GRK5 to dissociate PSD-95 from ␤ 1 AR (Fig. 2).
To quantitatively determine the effect of GRKs on PSD-95/ ␤ 1 AR association, we analyzed the PSD-95 immunoreactive protein bands using a Bio-Rad Fluor-S MultiImager. We set the amount of PSD-95 associated with ␤ 1 AR in the absence of exogenous GRKs and without agonist stimulation as the basal level (100%), and all other results are expressed as a percentage of this basal level. As shown in Fig. 5, when ␤ 1 AR and PSD-95 were coexpressed with either GRK2 or the GRK5(K215R) mutant (GRK5(DN)), the association of ␤ 1 AR with PSD-95 was not significantly changed. Agonist stimulation of these cells also had no effect on PSD-95/␤ 1 AR association. However, coexpression of GRK5 dramatically decreased the binding of ␤ 1 AR with PSD-95, down to 24% of the control level. Agonist stimulation further decreased the amount of PSD-95 bound to the receptor to 5% of the basal level (Fig. 5).
GRK5 Regulates PSD-95/␤ 1 AR Association by Phosphorylation of the Receptor-GRK5 could phosphorylate either ␤ 1 AR or PSD-95 to regulate the association of PSD-95 with ␤ 1 AR. To ascertain the phosphorylation target of GRK5, we expressed ␤ 1 AR, PSD-95, ␤ 1 AR with GRK5, or PSD-95 with GRK5 in COS-7 cells. Lysate from cells expressing ␤ 1 AR was then mixed with that from cells expressing PSD-95, followed by immunoprecipitation of ␤ 1 AR. As shown in Fig. 6, when GRK5 was coexpressed with PSD-95, it was unable to inhibit PSD-95 binding to ␤ 1 AR. However, when GRK5 was coexpressed with ␤ 1 AR, the association of PSD-95 with ␤ 1 AR was decreased. Since the lysate contains no additional ATP, GRK5 should be catalytically inactive once the cells are lysed, suggesting that GRK5 regulates ␤ 1 AR association with PSD-95 by phosphorylation of ␤ 1 AR, not PSD-95.
PSD-95/␤ 1 AR Association Is Not Affected by ␤-Arrestins-Following phosphorylation of the agonist-activated receptor by GRKs, ␤-arrestins bind to the receptor and promote receptor internalization (22). Thus, ␤-arrestin binding to the receptor might prevent PSD-95/␤ 1 AR association. To investigate whether ␤-arrestin binding is required for regulating PSD-95/ ␤ 1 AR association, we tested the effect of ␤-arrestins on PSD-95/␤ 1 AR association in the presence of GRK5. COS-7 cells were transfected with either ␤-arrestin-1 or ␤-arrestin-2 with ␤ 1 AR, PSD-95, and GRK5. As shown in Fig. 7, the amount of PSD-95 associated with ␤ 1 AR was decreased by expression of GRK5. However, the association of PSD-95 with ␤ 1 AR was not significantly further changed with overexpression of either ␤-arrestin-1 or ␤-arrestin-2 in the presence of GRK5. We also noted that overexpression of ␤-arrestins did not facilitate GRK2 reg-  3. Both GRK2 and GRK5 enhance the phosphorylation of and are associated with ␤ 1 AR. A, GRKs enhance the phosphorylation of ␤ 1 AR in vivo. 2 days after transfection, cells were washed with phosphate-free Dulbecco's modified Eagle's medium and incubated with [ 32 P]orthophosphate to label the intracellular ATP pool. Cells were stimulated with 10 M isoproterenol (ISO) for 5 min and washed twice with ice-cold Dulbecco's phosphate-buffered saline. Cells were then solubilized in lysis buffer A, and the receptor was immunoprecipitated with anti-FLAG M2 affinity gel as described under "Experimental Procedures." Phosphorylation of the receptor was analyzed with a Molecular Dynamics PhosphorImager. B, GRKs co-immunoprecipitate with ␤ 1 AR. COS-7 cells were transiently transfected with FLAG-␤ 1 AR alone or with either GRK2 or GRK5. The cells were incubated in serum-free medium for 60 min before stimulation with 10 M isoproterenol for 10 min and then washed with Dulbecco's phosphate-buffered saline and exposed to the cross-linking reagent dithiobis(succinimidyl propionate) for 30 min. The reaction was quickly quenched by 1 ml of ice-cold lysis buffer B. The cell lysates were clarified, and receptors were immunoprecipitated (IP) with anti-FLAG M2 beads. GRK2 or GRK5 associated with receptors was detected by immunoblotting (IB) with a mixture of the monoclonal anti-GRK2 and GRK5 antibodies. The data shown are representative of four independent experiments. ulation of PSD-95/␤ 1 AR association (data not shown). These findings support the idea that ␤-arrestin binding to the GRKphosphorylated receptor is not required to regulate PSD-95/ ␤ 1 AR association.
The ␤ 1 AR Carboxyl Terminus Is a Better Substrate for GRK5 than for GRK2-To investigate the functional difference between GRK2 and GRK5 that is responsible for their differential regulatory effects on ␤ 1 AR, we compared the ability of GRK2 and GRK5 to phosphorylate the GST-␤ 1 AR carboxyl terminus (GST-␤ 1 CT) fusion protein or GST-␤ 2 CT (Fig. 8). Equal amounts of GST fusion proteins (including GST alone, GST-␤ 1 CT, and GST-␤ 2 CT) and kinase were used in each reaction. GRK5 and GRK2 equally phosphorylated GST-␤ 2 CT. However, under the same conditions, GRK5-mediated GST-␤ 1 CT phosphorylation was substantially better than that mediated by GRK2 (Fig. 8). These results suggest that, at least in vitro, ␤ 1 AR-CT is a better substrate for GRK5 than for GRK2. DISCUSSION Phosphorylation of GPCRs by GRKs is a major mechanism responsible for termination of cellular signaling by these receptors or desensitization (33). Among the somatic GRKs, GRK2, GRK3, and GRK5 have been reported to enhance the phosphorylation of ␤ 1 AR in HEK293 cells (25). Based on sequence and functional similarities, GRK2 and GRK3 belong to one GRK subfamily, whereas GRK5 belongs to a distinct subfamily that includes GRK4, GRK5, and GRK6 (23). Here we studied the ability of GRKs to regulate a PDZ domain-mediated interaction with ␤ 1 AR by using GRK2 and GRK5, two enzymes from different subfamilies. We have demonstrated that phosphorylation of ␤ 1 AR by GRK5, but not by GRK2, appears to regulate its binding to one intracellular protein-binding partner, PSD-95.
Several recent reports suggest that phosphorylation can reg- ulate PDZ domain-mediated protein/protein interactions. Phosphorylation of Ser 880 in the glutamate receptor 2 subunit by PKC reduces its binding affinity for glutamate receptor-interacting protein (34). Furthermore, serine phosphorylation also has been implicated in the differential regulation of the ␣-amino-3-hydroxy-5-methylisoxazole propionate (AMPA) receptor interaction with PDZ domain-containing proteins (35). Phosphorylation of the carboxyl terminus of the inward rectifier K ϩ channel (Kir2.3) by protein kinase A inhibits its association with PSD-95 and SAP97 (36,37). In fact, for all class I PDZ domain-interacting motifs, there is a phosphorylatable serine or threonine at position Ϫ2 that is critical for PDZ domaincontaining protein binding. Phosphorylation of the serine or threonine at this site may provide an efficient way to regulate PDZ domain-mediated protein interactions.
␤ 2 AR interacts with the PDZ domain-containing protein NHERF through the PDZ1 domain (8,38). This PDZ domainmediated interaction of ␤ 2 AR with NHERF family proteins has been shown to control the choice between degradation and recycling of internalized ␤ 2 AR (9). Very interestingly, the recycling of ␤ 2 AR was dependent on the intact PDZ domain-binding motif (DSLL-COOH) in the carboxyl terminus of ␤ 2 AR (Fig. 9), and receptor recycling is modulated by overexpression of GRK5, but not of GRK2 (9). The Ser residue in this motif has been shown to be a phosphorylation site for GRK5, but not for GRK2, in vitro ( Fig. 9) (24). It is likely that phosphorylation of this Ser residue by GRK5 can decrease the binding of NHERF or other proteins to ␤ 2 AR and thereby regulate the recycling of the internalized receptors. Interestingly, one splice variant of GRK6 (GRK6A) can phosphorylate NHERF at Ser 289 as well, also suggesting that GRKs may regulate PDZ domain-mediated interactions (39).
Here we have shown that phosphorylation of ␤ 1 AR by GRK5 decreases the association of ␤ 1 AR with PSD-95. It has been reported that ␤ 1 AR can be phosphorylated by GRKs in a manner analogous to that of ␤ 2 AR (25). The ESKV-COOH motif in the carboxyl terminus of ␤ 1 AR appears to be responsible for the interaction with PSD-95 (15). The Ser residue at position Ϫ2 (Ser 475 ) is the only putative GRK phosphorylation site in this motif (Fig. 9). One explanation for our observation is that this Ser residue is phosphorylated by GRK5, but not by GRK2, and that phosphorylation of this Ser residue decreases the affinity of PSD-95 binding. We have shown that the ␤ 1 AR carboxylterminal tail is a better substrate for GRK5 than for GRK2, but further studies will be required to assess whether phosphorylation of Ser 475 indeed results from the action of GRK5 in vivo and whether this particular phosphorylation underlies all of the GRK5 regulatory effect on PSD-95 association. Thus, for both ␤ 1 AR and ␤ 2 AR, association with PDZ domain-containing partners is specifically regulated by GRK5, not GRK2. In view of the fact that many GPCRs have the similar serine/threoninebased PDZ domain-binding motifs in their carboxyl termini and that they are also putative substrates for GRKs, it is reasonable to suspect that phosphorylation of PDZ domain-binding motifs of receptors by GRK5 or other GRKs may represent a common mechanism for regulation of PDZ domain-mediated interaction in other GPCRs as well. More generally, it is also possible that other protein kinases (such as protein kinase A or PKC) may play similar roles.
The PDZ domain-mediated assembly and turnover of receptor signaling complexes is a dynamic process that is likely to be highly regulated. The regulatory mechanism involving phosphorylation of receptor might be important for receptor functions such as receptor internalization. Agonist-induced receptor internalization, a process in which activated receptors move from the cell surface into intracellular vesicles, is a common feature of GPCRs and is important for receptor regulation (33). Since many known PDZ domain-mediated protein/protein interactions are involved in the anchoring and targeting of proteins in specific membrane areas (such as synapses), it is likely that PDZ domain interaction with GPCRs will inhibit receptor internalization. PDZ domain-mediated PSD-95/␤ 1 AR interaction might be involved in anchoring the receptors in the postsynaptic membrane. Indeed, PSD-95 interaction with ␤ 1 AR decreased agonist-induced receptor internalization, possibly due to the time required for the dissociation of the PSD-95⅐␤ 1 AR complex. When activated by agonist stimulation, GRK5 can phosphorylate agonist-bound receptors (28). This phosphorylation of the receptor by GRK5 decreases the binding affinity of PSD-95, releasing the receptor from the PSD-95 complex and facilitating internalization. Release of receptor from PSD-95 may also make it available for distinct protein interactions and signaling events. A similar regulatory mechanism may also be applicable to other proteins, such as AMPA receptors, in which phosphorylation of the glutamate receptor 2 subunit by PKC is very important in regulation of AMPA receptor internalization during synaptic plasticity (35).
Even though the evidence here strongly supports the idea that phosphorylation of ␤ 1 AR decreases its association with PDZ domain-containing proteins, we cannot rule out the possibility that receptor phosphorylation might also be critical for recruiting proteins important for receptor signal transduction. In this case, phosphorylation of the receptor may serve as an initial signal to assemble receptor signaling complexes. For example, it has been reported that PKC activation can increase the phosphorylation of the glutamate receptor 2 subunit and recruit PICK1 to the excitatory synapses (35). Furthermore, cerebellar long-term depression also requires PKC-regulated interactions between the glutamate receptor 2/3 subunit and PDZ domain-containing proteins (40). Using synthetic peptides derived from the tumor suppressor MMAC/PTEN PDZ-binding domain to coprecipitate proteins from human cell lysates, Adey et al. (41) reported that threonine phosphorylation of this PDZbinding domain both inhibits and stimulates PDZ binding.
FIG. 8. The ␤ 1 AR carboxyl terminus is a better substrate for GRK5 than for GRK2. Equal amounts of GST fusion proteins (GST, GST-␤ 1 CT, and GST-␤ 2 CT) immobilized on glutathione-Sepharose 4B beads were incubated with equal amounts of purified GRK2 or GRK5 in the presence of [␥-32 P]ATP. GRK2-or GRK5-mediated GST fusion protein phosphorylation was analyzed as described under "Experimental Procedures." The data shown are representative of three independent experiments.
FIG. 9. Protein sequence alignment of the extreme carboxyl termini of ␤ 1 AR and ␤ 2 AR. The arrows indicate residues in both receptors that are critical for binding with PDZ domain-containing proteins. The serine at position Ϫ2 is the putative phosphorylation site for GRK5.
In summary, we have provided strong evidence that GRK5, but not GRK2, can regulate the association of PSD-95 with ␤ 1 AR by phosphorylation of the receptor. GRK-mediated receptor phosphorylation may provide a common regulatory mechanism for the association of other GPCRs with PDZ domaincontaining proteins. This regulation may be involved in many important processes such as receptor internalization, receptor localization to specific domains (e.g. synapses), and receptor signaling events.