Phosphorylation of the Platelet-derived Growth Factor Receptor-β and Epidermal Growth Factor Receptor by G Protein-coupled Receptor Kinase-2

Accumulating evidence suggests that receptor protein-tyrosine kinases, like the platelet-derived growth factor receptor-β (PDGFRβ) and epidermal growth factor receptor (EGFR), may be desensitized by serine/threonine kinases. One such kinase, G protein-coupled receptor kinase-2 (GRK2), is known to mediate agonist-dependent phosphorylation and desensitization of multiple heptahelical receptors. In testing whether GRK2 could phosphorylate and desensitize the PDGFRβ, we first found by phosphoamino acid analysis that cells expressing GRK2 could serine-phosphorylate the PDGFRβ in an agonist-dependent manner. Augmentation or inhibition of GRK2 activity in cells, respectively, reduced or enhanced tyrosine phosphorylation of the PDGFRβ but not the EGFR. Either overexpressed in cells or as a purified protein, GRK2 demonstrated agonist-promoted serine phosphorylation of the PDGFRβ and, unexpectedly, the EGFR as well. Because GRK2 did not phosphorylate a kinase-dead (K634R) PDGFRβ mutant, GRK2-mediated PDGFRβ phosphorylation required receptor tyrosine kinase activity, as does PDGFRβ ubiquitination. Agonist-induced ubiquitination of the PDGFRβ, but not the EGFR, was enhanced in cells overexpressing GRK2. Nevertheless, GRK2 overexpression did not augment PDGFRβ down-regulation. Like the vast majority of GRK2 substrates, the PDGFRβ, but not the EGFR, activated heterotrimeric G proteins allosterically in membranes from cells expressing physiologic protein levels. We conclude that GRK2 can phosphorylate and desensitize the PDGFRβ, perhaps through mechanisms related to receptor ubiquitination. Specificity of GRK2 for receptor protein-tyrosine kinases, expressed at physiologic levels, may be determined by the ability of these receptors to activate heterotrimeric G proteins, among other factors.

The platelet-derived growth factor receptor-␤ (PDGFR␤) 1 mediates signaling important for the proliferation, migration, and survival of mesenchymal cells (1). Like most receptor protein-tyrosine kinases, the PDGFR␤ is thought to exist as a quiescent monomeric protein that, upon binding agonist, dimerizes and consequently phosphorylates itself and other proteins on Tyr residues (1). The significance of regulating PDGFR␤ Tyr kinase activity is highlighted by the receptor's critical role in embryonic development and in the pathogenesis of vascular proliferative diseases like atherosclerosis and neointimal hyperplasia subsequent to vascular injury (1)(2)(3).
A number of mechanisms have been demonstrated to diminish PDGFR␤ signaling, including Tyr dephosphorylation (4,5), removal of receptors from the cell surface (6), degradation of receptors (6,7), and, described most recently, phosphorylation of the PDGFR␤ on serine residues (8). In cells, the PDGFR␤ Tyr kinase activity desensitizes or wanes with persistent exposure to PDGF (9). Desensitization commonly occurs consequent to serine phosphorylation of heptahelical receptors that couple to heterotrimeric G proteins (10). In addition, serine phosphorylation of both the insulin receptor (11) and EGFR (12) has been associated with receptor desensitization.
The serine/threonine kinase G protein-coupled receptor kinase-2 (GRK2) phosphorylates and desensitizes a large number of heptahelical receptors (10). We found recently that overexpression of GRK2 could attenuate phosphoinositide hydrolysis, [ 3 H]thymidine incorporation, chemotaxis, and cellular proliferation evoked via the PDGFR␤ in vascular SMCs (13,14). Importantly, receptor-specific desensitization by GRK2 correlated with a reduction in PDGF-promoted Tyr phosphorylation of the PDGFR␤ (14). The plausibility of GRK2-mediated regulation of the PDGFR␤ in cells is suggested by the caveolar localization of PDGFRs in cells (15) and the ability of caveolin to bind to GRK2 (16). GRK2 also has been demonstrated to associate with phosphoinositide 3-kinase-␣ (17), which translocates to activated PDGFR␤s (1) and is required for PDGFR␤ internalization (18). Moreover, a Tyr kinase that is a principal effector of the PDGFR␤, c-Src, appears to be a critical activator of GRK2 activity in cells (19,20). Thus, by activating c-Src, agonistactivated PDGFR␤s could activate GRK2 indirectly. To address whether GRK2 affected PDGFR␤ desensitization directly, by phosphorylating the PDGFR␤, or indirectly, by phosphorylating other regulatory proteins, we undertook the current study.

EXPERIMENTAL PROCEDURES
Materials-All cell culture products were from Invitrogen. Human PDGF-BB was from Upstate Biotechnology, and LPA, human EGF, and nucleotides were from Sigma. All radionuclides were from PerkinElmer Life Sciences. Sources for antibodies are provided below.
Plasmid and Adenovirus Constructs-Plasmids encoding bovine GRK2, dominant-negative (K220R) mutant GRK2, and the GRK2 Cterminal peptide were described previously (21), as were the empty vector and GRK2-encoding recombinant adenoviruses (14). Plasmids encoding wild type and kinase-dead K634R mutant constructs of the human PDGFR␤ (22) and the human EGFR (23) were the generous gifts of Andrius Kazlauskas and Axel Ullrich, respectively. Cassette PCR was employed to replace the endogenous signal sequence of each cDNA with an influenza virus hemagglutinin signal sequence followed by the FLAG™ epitope (24). The 5Ј (mutagenic) primer for the EGFR was 5Ј-cgcggggcggccgcacc(atgaagaccatcatcgcc-ctgagctacatcttctgcctggtgttcgcc)[gactacaaggacgatgacgacaag]ctggaggaaaagaaagtttgc-3Ј. Italics indicate a NotI site, parentheses surround the signal sequence, brackets surround the epitope sequence, boldface type highlights the initiator methionine, and underlined type denotes nucleotides 259 -279 of the native EGFR sequence. The mutagenic primer for the PDGFR␤ was similar, except that EcoRI replaced NotI, and nucleotides 452-473 of the PDGFR␤ (25) were used. To create the FLAG-tagged EGFR (F-EG-FR) construct, a 770-bp NotI/XmaI-cut PCR fragment was subcloned into the native EGFR cDNA, which had been subcloned previously into pcDNA I (Invitrogen). For the F-PDGFR␤ construct, a 108-bp EcoRI/XmaI-cut PCR fragment was subcloned into the native PDGFR␤, subcloned previously into pcDNA I. Sequence fidelity was confirmed by dideoxy sequencing. To create the F-PDGFR␤ (K634R) mutant, a 1.2-kb FspI/BstEII fragment of the mutant construct was subcloned into the cognate site in the F-PDGFR␤. Throughout the text, the F-PDGFR␤ and F-EGFR constructs are referred to as PDGFR␤ and EGFR, respectively.
Cell Culture, Transfection, Adenoviral Infection, and Gene Expression-HEK 293 cells (21) (from the American Type Culture Collection) and rabbit aortic smooth muscle cells (SMCs) (14) were cultured as described previously. Transfections of 293 cells with 10 g of total plasmid DNA per 100-mm dish were performed by calcium phosphate/ DNA co-precipitation, as described (21). Adenoviral infections of SMCs were also performed as described previously (14). All transfected or infected cell populations were pooled before aliquoting for both assays and determinations of plasmid expression. Assays were performed 2 days after transfection of 293 cells or 3 days after infection of SMCs. Target expression levels for GRK2 and GRK2(K220R) were 40-fold higher than endogenous levels in 293 cells or SMCs (ϳ12 pmol/mg of cell protein (14)), assessed by immunoblotting of serially diluted cell extracts. Efficiency of plasmid or adenoviral vector expression, respectively, was typically 35-50% in 293 cells (assessed by cell surface immunofluorescence and flow cytometry) (21) or Ն95% in SMCs, assessed by fluorescence microscopy (14). Transfected 293 cell lines cotransfected with various plasmids typically demonstrated receptor expression levels (24) within 30% of that measured in empty vector-cotransfected control cells. Cells with receptor expression levels outside this range were not used.
To determine the effect of pertussis toxin (PTX) on agonist-stimulated ERK activation, serum-starved SMCs were pretreated with PTX (100 ng/ml; List Laboratories) or vehicle for 16 h (26), challenged with the indicated agonist for 10 min, and then lysed in Laemmli buffer (24). Multiple sample aliquots were subjected to replicate SDS-PAGE and immunoblotting (14), and replicate blots were probed for either phospho-ERK1/2 or total ERK1/2, with antibodies from New England Biolabs. Graded amounts of sample were loaded, so as to create standard curves from which to interpolate relative signal intensity by densitometry. Phospho-ERK signals were normalized to cognate total ERK signals.
Receptor Immunoprecipitations-To assess the effects of GRK2 on PDGFR␤ and EGFR total and Tyr phosphorylation, and to determine the association of GRK2 with the PDGFR␤ and EGFR, cells were serum-starved overnight, exposed to vehicle-or agonist-containing medium for the indicated times, and subjected to IP as described previously for the FLAG-tagged endothelin B receptor (24), except that M2-agarose beads (Sigma) were used. For co-immunoprecipitation of GRK2 with the PDGFR␤ or EGFR, the cell-permeant cross-linker dithiobis(succinimidyl propionate) (Pierce) was used as described (24). For the purpose of immune complex kinase assays (see Figs. 5-7), IP was performed similarly, except that we used IP buffer II (50 mM Hepes, pH 7.5, 10% (v/v) glycerol, 1% (v/v) Triton-X-100, 150 mM NaCl, 1.5 mM MgCl 2 , 1 mM sodium orthovanadate, and protease inhibitors).
Intact Cell Phosphorylation Assays-After transfected 293 cells were serum-starved overnight, these assays were performed and quantitated with a PhosphorImager™ (Molecular Dynamics) as described previously (24).
Immunoblotting-Receptor expression levels and the protein concentration of cell lysates were used to load equivalent amounts of receptor per lane for SDS-PAGE as described (24). Proteins were transferred to either nitrocellulose or PVDF membranes, immunoblotted, stripped, and reprobed as described (24), with the following IgGs: for phospho-Tyr, pY20/HRP (Transduction Laboratories); for the PDGFR␤ and EGFR C-terminal domains, respectively, rabbit IgGs sc-432 and sc-03 (Santa Cruz Biotechnology); for ubiquitin, monoclonal Ub(P4D1) (Santa Cruz); and for equivalent identification of GRK2 and the GRK2 Cterminal peptide (GRK2ct), monoclonal E23/8 (14). Horseradish peroxidase-conjugated anti-IgGs for rabbit and mouse came from Jackson ImmunoResearch. For anti-ubiquitin blots, peroxidase-conjugated antimouse IgG came from Amersham Biosciences. For sequential immunoblotting, anti-phospho-Tyr blotting always followed anti-receptor blotting (because our nitrocellulose stripping procedure could not remove pY20; data not shown).
Immune Complex Kinase Assays-Bovine GRK2 was produced in Sf9 cells by recombinant baculovirus-mediated expression and purified as described (21). Receptors immunoprecipitated from 100-mm dishes of 293 cells (challenged or not with agonist) were washed three times in IP buffer III (20 mM Hepes, pH 7.5, 150 mM NaCl, 10% (v/v) glycerol, 0.1% (v/v) Triton-X-100, 1 mM sodium orthovanadate) and three times with kinase buffer (20 mM Tris-Cl, pH 8.0 (25°C), 2 mM EDTA, 10 mM MgCl 2 , 1 mM dithiothreitol). Immune complex beads were aspirated to dryness with a 28-gauge needle and resuspended in 15 l of kinase buffer. Reactions were performed (35°C, 30 min) in a total volume of 30 l of kinase buffer with 200 nM GRK2 or vehicle and 0.1 mM ATP containing 10 Ci of [␥-32 P]ATP. Reactions were terminated by the addition of 30 l of 2ϫ Laemmli buffer (21), and samples were heated to 65°C for 10 min to dissociate immune complexes. Samples of equivalent receptor mass (determined from protein assay of solubilized cells) were loaded onto each lane of 6% SDS-polyacrylamide gels. For each sample, 90% was subjected to SDS-PAGE for subsequent phosphoamino acid analysis, and 10% was subjected to SDS-PAGE for subsequent immunoblotting for receptor and for phospho-Tyr. Receptor phosphorylation was quantitated with a PhosphorImager™ (Molecular Dynamics).
PDGFR␤ Down-regulation-SMCs infected with either the empty vector or GRK2-encoding adenovirus were serum-starved overnight and then exposed to serum-free medium containing either 2 nM PDGF-BB or vehicle for the indicated times. To terminate incubations, SMCs were transferred to ice, washed twice with cold phosphate-buffered saline, and lysed for membrane preparation, as described above. Membrane pellets were then resuspended in solubilization buffer (50 mM Hepes, pH 7.4, 0.5% (v/v) Nonidet P-40, 250 mM NaCl, 10% (v/v) glycerol, 2 mM EDTA, 10 mM NaF, with protease inhibitors) and mixed gently for 45 min (4°C). Insoluble debris was pelleted at 20,000 ϫ g for 10 min (4°), and the supernatant was subjected to protein assay (24). For each sample, 35 g of membrane protein were loaded in duplicate lanes of a 4 -12% gradient SDS-polyacrylamide gel and subjected to immunoblotting for PDGFR␤.
Data Analysis-A paired t test was used to compare GRK2-overexpressing or PTX-treated with cognate control cells, and two-sided p values were calculated with Excel™ software (Microsoft). With Prism™ software (Graphpad, Inc.), repeated measures one-way analysis of variance with a post-hoc Tukey's test was used to compare GTP loading values among SMC membrane preparations treated with various agonists. Data in the text are mean Ϯ S.D., whereas means Ϯ S.E. are depicted in figures.

RESULTS
Coupling of the PDGFR␤ to G␣ i -We found recently that signaling through the PDGFR␤ was desensitized, in a receptorspecific manner, in SMCs overexpressing GRK2 (13,14). Unlike the PDGFR␤, all other receptors known to be desensitized by GRK2 are heptahelical receptors that activate heterotrimeric G proteins allosterically (10). We therefore asked whether the PDGFR␤, like heptahelical receptors, could also activate G proteins allosterically. To answer this question, we tested the ability of the PDGFR␤ to promote GTP turnover in G␣ subunits. In rabbit SMC membranes (which express only the PDGFR␤) (29), we could find no evidence that PDGF stimulated GTP␥S binding to G␣ q , whereas such binding was stimulated by endothelin-1 (data not shown). However, PDGF stimulated a 1.7-fold increase in GTP␥S binding to G␣ i1-3 (Fig. 1A). LPA, which is known to activate G i -coupled, heptahelical endothelial differentiation gene (Edg) receptors (30), stimulated a 2.8-fold increase in GTP␥S binding to G␣ i1-3 , whereas EGF failed to promote GTP turnover on G␣ i1-3 (Fig. 1A). Thus, the PDGFR␤ appeared to activate G␣ i isoforms in SMC membranes, but the allosteric nature of the PDGFR␤/G␣ i interaction remained to be established. PDGFR␤s may induce the production of reactive oxygen species (ROS) (31), and ROS may activate G␣ i isoforms (32). We therefore sought to ascertain whether the PDGFR␤ activated G␣ i isoforms allosterically, as heptahelical receptors do (33), or only indirectly via the production of ROS.
To demonstrate that the PDGFR␤ activated G␣ i via a mechanism(s) different from ROS, we employed two approaches. First, in SMC membrane preparations (Fig. 1A) we found no activation of G␣ i upon stimulation of the EGFR, which, like the PDGFR, is also known to promote production of ROS (34). Second, in intact SMCs, we used PTX to inhibit G i /G o -mediated activation of ERK1/2, because H 2 O 2 -induced ERK activation via G i /G o has been shown to be PTX-insensitive (32). We found that PDGF-induced ERK activation in SMCs was reduced 30 Ϯ 15% (p Ͻ 0.02) by PTX (Fig. 1B), consistent with G i /G o activation that is not mediated by ROS. By comparison, G i /G o -mediated (35) ERK activation induced by LPA was abolished by PTX treatment, and EGF-induced ERK activation of comparable magnitude was not affected significantly by PTX (Fig. 1B). Thus, at least a fraction of PDGFR␤-dependent ERK activation in vascular SMCs can proceed via PTX-sensitive G proteins. Moreover, the PTX sensitivity of the PDGF-promoted ERK activation suggests that the PDGFR␤ can couple to G i directly, as heptahelical receptors do, rather than just indirectly via the production of ROS. Interestingly, the selectivity of GRK2 for desensitizing the PDGFR␤ and not the EGFR in SMCs (14) is mirrored by the selectivity of PTX-sensitive G protein coupling to the PDGFR␤, and not the EGFR, in these cells.
Functional Effects of GRK2 on PDGFR␤ Activation-To inhibit the possible effects of GRK2 on activation of the PDGFR␤ (14), we employed a kinase-dead mutant of GRK2 (K220R), which we (21,24) and others (36) have used extensively to inhibit GRK-mediated serine phosphorylation of heptahelical receptors. In Fig. 2, reciprocal effects on PDGFR␤ activation were observed with overexpression of either K220R-GRK2 or wild type GRK2. K220R-GRK2 overexpression effected a 69% increase in PDGFR␤ Tyr phosphorylation (p Ͻ 0.02), whereas GRK2 overexpression effected a 28% decrease in agonist-promoted PDGFR␤ Tyr phosphorylation (p Ͻ 0.02). In contrast, neither K220R-GRK2 nor GRK2 itself had a significant effect on agonist-promoted Tyr phosphorylation of the EGFR. In this heterologous overexpression system, PDGFR␤-mediated G protein activation (Fig. 1) did not appear to be important for GRK2 membrane recruitment (10) and activity on the PDGFR␤. Expression of a G␤␥-sequestering GRK2 C-terminal polypeptide (14), at levels equivalent to those of the co-overexpressed GRK2, did not alter GRK2-mediated PDGFR␤ desensitization (data not shown). These findings in 293 cells confirm and extend our findings with GRK2 overexpression in SMCs (14). Because potentiating and inhibiting GRK2 activity in cells affected PDGFR␤ activation reciprocally, and in a receptorspecific fashion, a direct interaction between GRK2 and the PDGFR␤ seemed likely.
To determine whether GRK2 interacted with the PDGFR␤ itself, we performed co-immunoprecipitation experiments, using the EGFR as a negative control (because its activation (Fig.  2) and signaling (14) were not altered by changes in GRK2 activity). In cells expressing both the PDGFR␤ and GRK2 (Fig.  3), GRK2 associated with the PDGFR␤ in an agonist-dependent manner, much like that we observed previously for heptahelical receptors desensitized by GRK2 (24). Surprisingly, we obtained similar results with the EGFR (data not shown). This agonist-promoted association between GRK2 and either the PDGFR␤ or the EGFR suggested that GRK2 could phosphorylate both of these receptors but with functionally distinct consequences for each receptor.
GRK2-mediated Phosphorylation of the PDGFR␤ and EGFR-If endogenous GRK2 could phosphorylate the PDGFR␤ in 293 cells, we would expect to observe agonist-dependent phosphorylation of the PDGFR␤ on serine residues. In 293 cells labeled metabolically with 32 P i , we found the PDGFR␤ to be phosphorylated tonically on serine residues and to increase its degree of serine phosphorylation substantially with PDGF challenge (Fig. 4A). These findings accorded with those of Bioukar et al. (8) in Rat2 fibroblasts.
As we found in Fig. 2, overexpression of GRK2 in these experiments diminished agonist-promoted PDGFR␤ Tyr phos-phorylation, by 37 Ϯ 8%, as assessed by phospho-Tyr immunoblotting (Fig. 4B). However, GRK2 overexpression had no effect on the total level of PDGFR␤ phosphorylation, assessed by receptor immunoprecipitation and autoradiography (Fig. 4B). Thus, a GRK2-mediated increase in PDGFR␤ serine phosphorylation could have counterbalanced the GRK2-engendered decrease in PDGFR␤ Tyr phosphorylation, yielding no net change in total PDGFR␤ 32 P incorporation. Data supporting GRK2mediated Ser/Thr phosphorylation of the EGFR in intact cells were more straightforward. Overexpression of GRK2 in 293 cells increased the agonist-induced incorporation of 32 P into the EGFR without altering the agonist-induced Tyr phosphorylation of the EGFR (Fig. 4C).
To confirm that GRK2 overexpression could increase serine phosphorylation of the agonist-stimulated PDGFR␤ or EGFR, we adapted a system (8) that allowed us to preserve agonistmediated activation of the receptors but enabled us to minimize the effect of endogenous 293 cellular GRK2 (21, 37) on our phosphorylation reactions. With purified GRK2, we phosphorylated receptors immunoprecipitated from cells that had been exposed to vehicle-or agonist-containing medium. Receptor phosphorylation was evaluated after resolution of immune complexes by SDS-PAGE, and proteins were transferred to PVDF membrane to facilitate subsequent phosphoamino acid analysis. As depicted in Fig. 5A and Fig. 6A, incubation with GRK2 augmented agonist-induced phosphorylation of the PDGFR␤ and EGFR, compared with vehicle controls. As is characteristic of GRK2-mediated heptahelical receptor phosphorylation, GRK2-mediated phosphorylation of the PDGFR␤ and EGFR was agonist-promoted. Furthermore, GRK2 catalyzed the phosphorylation of the PDGFR␤ and EGFR on serine(s), as revealed by hydrolysis and phosphoamino acid analysis of the receptor (see Fig. 5C and Fig. 6B). Phosphoserine was detected only in PDGFR␤s and EGFRs incubated with GRK2. Interestingly, GRK2 also demonstrated some serine phosphorylation of PDGFR␤s and EGFRs not challenged with agonist, to the extent that receptor Tyr phosphorylation was induced by IgG-mediated receptor cross-linking during IP (22). These findings suggest that GRK2-mediated serine phosphorylation of these receptor protein-tyrosine kinases depended upon the extent of receptor activation, as has been demonstrated with the ␤ 2 -adrenergic (heptahelical) receptor (38). Taken together, these results demonstrate the PDGFR␤ and EGFR to be novel receptor substrates for GRK2.
GRK2-mediated phosphorylation of heptahelical receptors requires agonist-mediated stabilization of the active conformation of the receptor (10). For the PDGFR␤, agonist dimerizes receptor monomers, with consequent activation of receptor ty-  3. GRK2 associates with the PDGFR␤ in an agonist-dependent manner. HEK 293 cells transfected with plasmids encoding GRK2 and either no protein (GRK2) or the N-terminal epitope-tagged PDGFR␤ (PDGFR ϩ GRK2) were stimulated with 2 nM PDGF-BB for 5 min. Cell proteins were cross-linked with dithiobis(succinimidyl propionate), and the PDGFR␤ was immunoprecipitated. PDGFR␤ IPs and solubilized cell extracts (lysate) underwent SDS-PAGE and serial immunoblotting for GRK2 (top) and total receptor (middle). Shown are results from a single experiment, representative of three experiments performed. By densitometry, the amount of GRK2 co-immunoprecipitating with the PDGFR␤ increased 5 Ϯ 3-fold with agonist. rosine kinase activity (1). Is agonist binding sufficient to transform the PDGFR␤ into a GRK2 substrate, as would be expected of a heptahelical receptor? Alternatively, does allosteric activation of GRK2 by the PDGFR␤ require a receptor conformational change that depends on the Tyr kinase activity of the receptor? To differentiate these possibilities, we undertook immune complex kinase assays with both wild type and kinasedead (K634R) PDGFR␤ constructs (Fig. 7). Whereas GRK2mediated serine phosphorylation increased overall phosphorylation of the wild type PDGFR␤ by ϳ25%, consistent with data in Fig. 5, A and B, GRK2 failed to phosphorylate the kinase-dead PDGFR␤ mutant. It therefore seems that Tyr kinase activity of the PDGFR␤ is required for the receptor to be recognized as a substrate by GRK2.
Effects of GRK2 on PDGFR␤ Ubiquitination and Down-regulation-Myriad possible molecular consequences of GRK2mediated phosphorylation might distinguish the PDGFR␤ from the EGFR and explain the differences we observed in GRK2-mediated receptor desensitization. To explore these possibilities, we first examined agonist-induced receptor ubiquitination, a post-translational modification requiring receptor phosphorylation (7), believed to be involved in the degradation of both the PDGFR␤ (7) and the EGFR (39). Although GRK2 overexpression enhanced agonist-promoted PDGFR␤ ubiquitination (by 90%, p Ͻ 0.04), it had no effect on agonist-promoted EGFR ubiquitination (Fig. 8). Thus, in GRK2-overexpressing cells, augmentation of receptor protein-tyrosine kinase ubiquitination correlated with reduction of receptor activation, assessed by Tyr phosphorylation.
Could GRK2-mediated PDGFR␤ phosphorylation, by enhancing PDGFR␤ ubiquitination, subsequently augment PDGFR␤ down-regulation? This possibility is suggested by the association between PDGFR␤ ubiquitination and degradation (7). Moreover, such a mechanism could help explain GRK2mediated desensitization of PDGF-elicited, long-term responses that transpire over hours to days, such as primary SMC chemotaxis (13), [ 3 H]thymidine incorporation, and proliferation (14). In primary SMCs, GRK2 overexpression also desensitizes the PDGFR␤ within minutes (14), as it does in 293 cells (Fig. 2). In SMCs overexpressing GRK2 at levels comparable with those achieved in the transfected 293 cell system (data not shown), PDGFR␤ down-regulation proceeded at a rate indistinguishable from that observed in control cells (Fig.  9). Thus, although GRK2-related enhancement of PDGFR␤ ubiquitination correlates with receptor desensitization, it fails to correlate with receptor down-regulation. Therefore, the GRK2-mediated PDGFR␤ desensitization observed in longterm SMC assays would likely result from an impaired ability of the PDGFR␤ to activate effector enzymes, such as phospholipase C-␥ (14) and phosphoinositide 3-kinase-␣ (13), rather than from changes in the cellular complement of PDGFR␤s. DISCUSSION Several lines of evidence from this study support the inference that GRK2 phosphorylates and desensitizes the PDGFR␤, the only receptor protein-tyrosine kinase yet demonstrated to be regulated by a GRK. First, as would be expected of any GRK2 substrate, the PDGFR␤ demonstrated agonist-promoted serine phosphorylation both in cells and in the presence of purified GRK2. Second, GRK2 associated with the PDGFR␤ in cells in an agonist-dependent manner. Moreover, Tyr phospho-

FIG. 4. Agonist-induced PDGFR␤ and EGFR phosphorylation in 293 cells: effects of GRK2 overexpression.
HEK 293 cells transfected or not (Ϫ) with the indicated plasmids were labeled metabolically with 32 P i , exposed to medium-containing vehicle, 2 nM PDGF-BB, or 1.7 nM EGF for 5 min, and subjected to receptor IP. IPs resolved by SDS-PAGE (6% gels) were transferred to PVDF membrane and processed for autoradiography, followed by either phosphoamino acid analysis or immunoblotting. A, agonist-induced serine phosphorylation of the PDGFR␤ in 293 cells expressing endogenous levels of GRK2. Shown is a thin-layer electrophoresis of phosphoamino acids derived from PDGFR␤ bands. The positions of ninhydrin-stained phosphoamino acid standards (pS, pT, and pY) are indicated with circles. Similar results were obtained in three independent experiments. B, GRK2 overexpression reduces PDGFR␤ Tyr but not total phosphorylation induced by agonist. A single PVDF membrane is imaged either by PhosphorIm-ager™ ( 32 P) or after serial immunoblotting for the PDGFR␤ and phospho-Tyr (pY). Both the immature (160-kDa) and mature (180-kDa) glycosylated forms of the PDGFR␤ (40) are seen by immunoblotting, but only the mature form demonstrated PDGF-induced phosphorylation assessed by 32 P. Similar results were obtained in three independent experiments. C, GRK2 overexpression augments EGFR phosphorylation. A single PVDF membrane is imaged either by PhosphorImager™ ( 32 P) or after serial immunoblotting for the EGFR and phospho-Tyr. Results are from a single experiment, representative of three performed. rylation of the PDGFR␤ demonstrated receptor-specific desensitization or sensitization, respectively, when cellular GRK2 activity was either augmented or inhibited. Third, allosteric G␣ activation by the PDGFR␤ provided evidence that the PDGFR␤, like heptahelical receptors, can assume a conformation recognizable not only by G␣ subunits but also by GRK2. Furthermore, by activating and promoting the dissociation of heterotrimeric G proteins, agonist-stimulated PDGFR␤s could supply G␤␥ subunits that appear to be required for recruiting cytosolic GRK2 to plasma membrane receptors (under conditions of physiologic expression levels) (10). Last, augmenting cellular GRK2 activity not only diminished PDGFR␤ Tyr phosphorylation but also enhanced PDGFR␤ ubiquitination, thereby suggesting multiple levels of functional significance for GRK2-mediated phosphorylation of the PDGFR␤.
GRK2-mediated PDGFR␤ Phosphorylation and Desensitization-Potential regulation of the PDGFR␤ by serine phosphorylation was first demonstrated with casein kinase I-␥2 (8), which, like GRK2, can phosphorylate peptides more efficiently when they contain serines located C-terminal to acidic amino acids (10), and Ն14 such serines exist in the cytoplasmic domain of the PDGFR␤ (40). As with GRK2 in our study, overexpression of casein kinase I-␥2 in cells reduced agonist-promoted PDGFR␤ Tyr phosphorylation. However, inhibition of cellular casein kinase I-␥2 also reduced the overall level of PDGFR␤ phosphorylation and perhaps even reduced PDGFR␤ Tyr phosphorylation, as suggested by a decrease in the number of phosphorylated proteins co-immunoprecipitated with the PDGFR␤ (8). In contrast to inhibiting cellular casein kinase I-␥2, inhibiting cellular GRK2 activity increased PDGFR␤ activation, assessed by receptor Tyr phosphorylation. Thus, whereas in-FIG. 5. GRK2 mediates agonist-promoted serine phosphorylation of the PDGFR␤. HEK 293 cells transfected or not (None) with the PDGFR␤ plasmid were exposed to medium-containing vehicle or 2 nM PDGF-BB for 10 min at 37°C, solubilized, and subjected to PDGFR␤ IP. immunoprecipitated PDGFRs were used to perform immune complex kinase assays with [␥-32 P]ATP in the absence (Ϫ, Control) or presence (ϩ) of purified GRK2; SDS-PAGE and blotting to PVDF membrane followed. Parallel PVDF membranes were processed either for phosphoamino acid analysis or for immunoblotting total PDGFR␤ or phospho-Tyr (not shown). A, autoradiogram from a single representative PVDF membrane. With parallel immunoblots (not shown), GRK2 and control samples showed equivalent levels of PDGFR␤ phospho-Tyr. B, receptor phosphorylation data (mean Ϯ S.E.) from five independent experiments are summarized. Radioactivity in PDGFR␤ bands was first normalized to the relative amount of PDGFR␤ quantitated by immunoblotting; receptor-normalized 32 P counts were then normalized to those obtained from IPs of unstimulated cells incubated without GRK2 (control basal). *, p Ͻ 0.05 compared with cognate control value. C, receptor bands from panel A were hydrolyzed and subjected to phosphoamino acid analysis as in Fig. 4. Shown are results from a single experiment representative of five experiments performed. The positions of ninhydrin-stained phosphoamino acid standards (pS, pT, and pY) are indicated with arrows.
FIG. 6. GRK2 mediates agonist-promoted serine phosphorylation of the EGFR. HEK 293 cells transfected or not (None) with the EGFR plasmid were exposed to medium-containing vehicle or 1.7 nM EGF for 10 min at 37°C, solubilized, and subjected to EGFR IP. Immunoprecipitated EGFRs were used to perform immune complex kinase assays as in Fig. 5 hibition of cellular GRK2 activity abrogated PDGFR␤ desensitization assessed by receptor Tyr phosphorylation, inhibition of cellular casein kinase I activity did not.
Consequences of PDGFR␤ phosphorylation by casein kinase I-␥2 and GRK2 differed, as well, in immune complex kinase assays. When immunoprecipitated PDGFR␤ was serine-phosphorylated by purified GST-casein kinase I-␥2, Tyr phosphorylation of the PDGFR␤ was reduced (8). Casein kinase I-␥2mediated phosphorylation might therefore have reduced PDGFR␤ Tyr kinase activity, as Bioukar et al. (8) propose. By contrast, no decrease in PDGFR␤ Tyr phosphorylation obtained when immunoprecipitated PDGFR␤ was serine-phosphorylated by purified GRK2 (see legend to Fig. 5). How then might GRK2-mediated phosphorylation diminish PDGFR␤ Tyr phosphorylation in cells? Likely molecular mechanisms would seem to involve accessory cellular proteins, whose binding to the PDGFR␤ could be affected by GRK2-mediated serine phosphorylation of the receptor. Such accessory proteins might include ␤-arrestin isoforms, important to GRK-mediated desensitization of heptahelical receptors (10), or SHP-2, important in dephosphorylating the PDGFR␤ (4), among other possibilities.
Because PDGFR␤ Tyr phosphorylation is not altered by GRK2-mediated receptor serine phosphorylation in immune complex kinase assays (Fig. 5), we can make reasonable inferences about the stoichiometry of GRK2-mediated PDGFR␤ serine phosphorylation in these assays. GRK2-mediated serine phosphorylation increases total PDGFR␤ phosphorylation by ϳ20% in the immune complex kinase assay (Fig. 5B). Because the expected stoichiometry of agonist-induced PDGFR␤ Tyr phosphorylation is ϳ9 (9), an incremental 20% phosphorylation from GRK2 would correspond to ϳ2 mol of phosphate (on serines) per mol of PDGFR␤. (This analysis assumes equal probability of phosphate exchange with 32 P, of course, for PDGFR␤ Tyr and serine residues phosphorylated in 293 cells before IP.) Thus, the magnitude of GRK2-mediated PDGFR␤ serine phosphorylation seems congruent with that observed for heptahelical receptors in cells (21,24) and in purified protein systems validated by saturation binding with ␤-arrestins (41).
GRK2-mediated Phosphorylation of the EGFR without EGFR Desensitization-Data from this study also demonstrate for the first time that GRK2 can phosphorylate the EGFR, both in cells and in purified protein preparations. This finding was unexpected, because neither augmentation nor inhibition of cellular GRK2 activity affected EGFR activation (see Fig. 2 and Ref. 14), and GRK2 overexpression failed to affect EGFR ubiquitination (Fig. 8 (14), or SMC chemotaxis (13) promoted by EGF. Thus, there is an apparent paradox between GRK2-mediated phosphorylation of EGFRs and a lack of demonstrable GRK2-mediated EGFR desensitization in certain cells. Of course, it is possible that GRK2mediated EGFR phosphorylation does desensitize as yet unexplored signaling mechanisms downstream of the EGFR. To explain our current findings, however, it remains to be determined whether GRK2-phosphorylated EGFRs fail to undergo subsequent desensitizing modifications, such as receptor Tyr dephosphorylation, in the cell models we have examined.

), SMC [ 3 H]thymidine incorporation
Although GRK2 can phosphorylate the EGFR in a 293 cell overexpression model, the ability of GRK2 to phosphorylate the EGFR under more physiologic conditions remains uncertain. At physiologic levels of expression, GRK2 appears to require G protein ␤␥ subunits to facilitate its translocation from the cytosol to plasma membrane receptors (10), and the EGFR, FIG. 8. GRK2 overexpression augments agonist-induced ubiquitination of the PDGFR␤ but not the EGFR. HEK 293 cells were transfected with N-terminal FLAG-tagged constructs of either the PDGFR␤ or the EGFR and either vector (Empty) or GRK2-encoding plasmids. Exposed to medium containing vehicle (basal) or agonist for 30 min at 37°C, cells were then solubilized, and lysates were subjected to receptor IP/immunoblotting, as in Fig. 2. Serial immunoblotting was performed first for ubiquitin (top panels) and then for either the PDGFR␤ or EGFR (bottom panels). A, immunoblots from single experiments, representative of three performed in duplicate for each receptor. B, ubiquitinated receptor band densities were first normalized to cognate total receptor band densities. Next, agonist-stimulated ubiquitination signals from each cell line were divided by the corresponding values obtained for the appropriate unstimulated, vector-co-transfected (control) cells to obtain -fold/(control basal). Data are summarized (mean Ϯ S.E.) graphically from three experiments performed in duplicate. *, p Ͻ 0.04 compared with control.
FIG. 9. The time course of PDGFR␤ down-regulation is unaltered in GRK2-overexpressing SMCs. SMCs infected with vector (Empty) or GRK2-encoding adenoviruses were exposed to medium containing vehicle (control) or 2 nM PDGF-BB for the indicated times and then lysed mechanically. Membrane fractions from these SMCs were subjected to SDS-PAGE and immunoblotting for the PDGFR␤. A, measurements of PDGFR␤ band density from each cell line were normalized to those obtained from cognate unstimulated (control) cells and expressed as % of control: 100 ϫ (stimulated/unstimulated). Each point depicts the mean of three independent experiments performed in duplicate. B, representative immunoblot from naïve and PDGF-challenged SMCs harvested after 4 h of incubation. unlike the PDGFR␤, does not appear to activate heterotrimeric G proteins (at least not in mesenchymal cells like SMCs) (Fig.  1). Moreover, the action of GRK2 is inhibited by calcium/calmodulin, in a manner that is relieved by protein kinase Cmediated phosphorylation of GRK2 (42). Because EGFR stimulation is known to increase cytosolic calcium without inducing significant protein kinase C activation in mesenchymal cells (12), net inhibition of GRK2 by calcium/calmodulin might be expected to result from isolated EGFR activity in these cells. In contrast, the PDGFR␤ in SMCs stimulates protein kinase C activity significantly (43). GRK2-mediated EGFR phosphorylation seems more plausible under physiologic conditions in cells like rat hepatocytes, in which the EGFR does activate heterotrimeric G proteins (44).
PDGFR␤ Coupling to Heterotrimeric G Proteins-Whether assessed by PDGF-promoted guanine nucleotide turnover in G␣ i1-3 or by PTX-mediated inhibition of signaling (Fig. 1), the PDGFR␤ in SMCs demonstrated coupling to heterotrimeric G proteins. Alderton et al. (45) also found PTX to inhibit PDGFR␤-, but not EGFR-evoked ERK phosphorylation in 293 cells; however, the [EGF] used in their experiments (100 nM) was far greater than in ours. In Fig. 1B, although PTX inhibited ERK activation by PDGF, it had no effect on comparable levels of ERK activation evoked by EGF. In contrast, using PDGF concentrations Ն2-fold higher than ours, Luttrell et al. (35) found that PTX inhibited neither PDGF-nor EGF-induced ERK activation in Rat1 fibroblasts. With higher PDGF concentrations, we too found that PTX failed to reduce PDGF-promoted ERK activation (data not shown). This result is to be expected, because most PDGFR␤-mediated ERK phosphorylation proceeds through Tyr kinase-dependent pathways producing Ras activation (1). At higher PDGF concentrations, high levels of PDGFR␤-promoted, Ras-mediated (and saturable) ERK activation could easily obscure the effect of abrogating G i /G o -mediated ERK activation. It should be noted that the PDGFR␤ is not unique among receptor Tyr kinases in signaling through heterotrimeric G proteins. Similar findings have been described for the insulin receptor (26), insulin-like growth factor-1 receptor (35), fibroblast growth factor receptors-1 and -2 (46,47), and even the EGFR in hepatocytes (48).
Effect of GRK2 on PDGFR␤ Ubiquitination-Agonist-induced ubiquitination of the PDGFR␤ requires PDGFR␤ Tyr kinase activity (7), much as GRK2-mediated serine phosphorylation of the receptor does. How GRK2-mediated phosphorylation of the PDGFR␤ might lead to enhanced receptor ubiquitination remains obscure. However, serine or threonine phosphorylation of proteins is known to trigger protein ubiquitination (44,49,50). It is therefore possible that GRK2mediated serine phosphorylation of the PDGFR␤ (but not the EGFR) could alter the interactions of the receptor with other proteins so as to promote the binding of c-Cbl (51) or other potential ubiquitin E3 ligases to the receptor. Alternatively, because allosterically activated GRK2 can phosphorylate nonreceptor substrates (10), it could, after PDGFR␤ activation, act like casein kinase 2 in serine-phosphorylating the ubiquitinconjugating (E2) enzyme CDC34 (49) and thereby potentially alter receptor ubiquitination. These and other possibilities remain to be explored. Likewise, whether receptor ubiquitination can contribute to PDGFR␤ desensitization, because ubiquitination has only modest effects on PDGFR␤ down-regulation (7), remains an open question.