Regulation of the Platelet-derived Growth Factor Receptor-β by G Protein-coupled Receptor Kinase-5 in Vascular Smooth Muscle Cells Involves the Phosphatase Shp2*

Smooth muscle cell (SMC) proliferation and migration are substantially controlled by the platelet-derived growth factor receptor-β (PDGFRβ), which can be regulated by the Ser/Thr kinase G protein-coupled receptor kinase-2 (GRK2). In mouse aortic SMCs, however, we found that prolonged PDGFRβ activation engendered down-regulation of GRK5, but not GRK2; moreover, GRK5 and PDGFRβ were coordinately up-regulated in SMCs from atherosclerotic arteries. With SMCs from GRK5 knock-out and cognate wild type mice (five of each), we found that physiologic expression of GRK5 increased PDGF-promoted PDGFRβ seryl phosphorylation by 3-fold and reduced PDGFRβ-promoted phosphoinositide hydrolysis, thymidine incorporation, and overall PDGFRβ tyrosyl phosphorylation by ∼35%. Physiologic SMC GRK5 activity also increased PDGFRβ association with the phosphatase Shp2 (8-fold), enhanced phosphorylation of PDGFRβ Tyr1009 (the docking site for Shp2), and reduced phosphorylation of PDGFRβ Tyr1021. Consistent with having increased PDGFRβ-associated Shp2 activity, GRK5-expressing SMCs demonstrated greater PDGF-induced Src activation than GRK5-null cells. GRK5-mediated desensitization of PDGFRβ inositol phosphate signaling was diminished by Shp2 knock-down or impairment of PDGFRβ/Shp2 association. In contrast to GRK5, physiologic GRK2 activity did not alter PDGFRβ/Shp2 association. Finally, purified GRK5 effected agonist-dependent seryl phosphorylation of partially purified PDGFRβs. We conclude that GRK5 mediates the preponderance of PDGF-promoted seryl phosphorylation of the PDGFRβ in SMCs, and, through mechanisms involving Shp2, desensitizes PDGFRβ inositol phosphate signaling and enhances PDGFRβ-triggered Src activation.

The pathogenesis of atherosclerosis (1) and neointimal hyperplasia after vascular injury (2) fundamentally involves the platelet-derived growth factor receptor-␤ (PDGFR␤) 5 expressed on smooth muscle cells (SMCs) (3). As a receptor protein-tyrosine kinase, the PDGFR␤ autophosphorylates on tyrosyl residues upon binding PDGF. Subsequently, the PDGFR␤ activates intracellular signaling cascades by tyrosinephosphorylating and/or scaffolding multiple proteins critical for cellular proliferation and migration (4). Until they are destroyed in the lysosome, activated PDGFR␤s appear to continue signaling (5). Thus, regulation of PDGFR␤ signaling prior to receptor degradation attains considerable significance.
GRK2 belongs to a seven-member family of Ser/Thr kinases (11,12), each with a central catalytic domain flanked by aminoand carboxyl-terminal domains that serve membrane-localizing, protein association, and other regulatory functions (11). Allosterically activated by agonist-occupied heptahelical (seven-membrane-spanning) receptors, GRKs characteristically phosphorylate these activated receptors and thereby initiate desensitization that is "receptor-specific" (i.e. that affects only the receptor whose activation prompted GRK activity). Only GRKs 2, 5, and 6 are widely expressed at substantial levels in mammalian tissues. With only 58.6% similarity to GRK2 (13), GRK5 has demonstrated receptor substrate specificity both overlapping with (13) and distinct from (14,15) GRK2. GRK subtype-specific phosphorylation sites (16) have been shown to result in distinct downstream molecular consequences for receptors phosphorylated by both GRK2 and GRK5 (17)(18)(19)(20). In light of the role of GRK2 in regulating PDGFR␤ function in fibroblasts, we initiated this investigation to determine whether the PDGFR␤ can be regulated in SMCs by GRK5 and whether GRK5 and GRK2 employ similar or distinct mechanisms for PDGFR␤ regulation.

MATERIALS AND METHODS
Recombinant Adenoviruses and Plasmids-The cDNA encoding bovine GRK5 was inserted into the plasmid pSKAC, and adenoviruses were produced in 293 cells, purified on CsCl gradients, and titered as we described previously (9). Plasmids encoding the N-terminal FLAG TM -tagged human PDGFR␤ (FPDGFR␤) (7), bovine GRK2 (21), and bovine GRK5 (21) have been described previously. The Y1009F mutant FPDGFR␤ was created from its WT congener by cassette PCR, using the following primers: 5Ј-cgcgggccatggcctccgatctcccctggacaccagctccgtcctctttactgccgtgcagcccaatg-3Ј (underscore denotes the Tyr 3 Phe mutation, and italic type denotes the 5Ј NcoI site) and 5Ј-cgcggggcggccgcaagcttctacaggaagctatcctctgc-3Ј (boldface type denotes the stop codon; underscore and italic type denote HindIII and NotI sites, respectively). Subcloning employed pBluescript II KS ϩ (Stratagene) as a shuttle vector, but both the WT and Y1009F mutant PDGFR␤ constructs were ultimately subcloned into pcDNA I (Invitrogen). PCR fidelity was verified by dideoxy sequencing.
Atherosclerosis Studies in Mice-All animal care conformed to Ref. 59. C57Bl/6J mice without (wild type, WT) or with targeted deletion of the apolipoprotein E gene (apoe Ϫ/Ϫ ; Jackson stock number 002052) were purchased from Jackson Laboratories and fed normal mouse chow. Nine-month-old mice were sacrificed, and the circulatory system was perfused with lactated Ringer solution at 80 mm Hg pressure. The common carotid arteries were excised, embedded in OCT compound, frozen at Ϫ150°C, and sliced at 5 m on a cryotome to obtain sections of the distal common carotid, just proximal to the carotid bifurcation (a site commonly involved with atherosclerosis in the 9-month-old apoe Ϫ/Ϫ mouse) (22). Sections were fixed and permeabilized with methanol/acetone (50:50) at room temperature for 2 min, washed twice in Dulbecco's PBS for 1 min, and then incubated (25°C) for 30 min in "blocking buffer": 3% (w/v) bovine serum albumin in TTBS (0.02% (v/v) Tween 20, 10 mM Tris-Cl, pH 7.4, 140 mM NaCl). Next, sections were incubated (25°C, 60 min) in blocking buffer containing 1 g/ml nonimmune rabbit IgG (nonspecific staining) or rabbit IgG specific for GRK5 (sc-565) or the PDGFR␤ (sc-432) (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). After three washes in TTBS, sections were then incubated (25°C, 60 min) in blocking buffer containing the DNA-binding dye Hoechst 33342 (10 g/ml) and 3 g/ml anti-rabbit IgG labeled with either Alexa 546 (for the PDGFR␤) or Alexa 488 (for GRK5) (Molecular Probes, Inc., Eugene, OR). After three additional washes in TTBS, sections were mounted in Gel/Mount TM (Biomeda Corp.) with a glass coverslip. For specimens stained for ␣-SMC actin, we incubated fixed specimens in blocking buffer contain-ing 0.5 g/ml Cy3-conjugated 1A4 (Sigma) as well as Hoechst 33342 (as above). Fluorescence photomicrographs were taken with Chroma TM narrow band-pass filters and a Spot CCD camera, with a fixed shutter speed for all specimens. Nuclear and protein fluorescence images were obtained from individual microscopic fields by rotating fluorescence filters. Protein fluorescence was normalized to DNA fluorescence within each image, as we described previously (23). Atherosclerotic and nonatherosclerotic specimens were always processed in pairs, so that staining and imaging could be equivalent for both phenotypes.
Cell Culture and Adenovirus Infections-New Zealand White rabbits, grk5 Ϫ/Ϫ and littermate WT mice (14) were sacrificed to make thoracic aortic SMCs, as previously described (9). HEK 293 cells and embryo fibroblasts (MEFs) from grk2 Ϫ/Ϫ and littermate WT mice were grown and transfected as described (8). The grk5 Ϫ/Ϫ mice used in this study were hybrids of the 129/SvJ and C57Bl/6J lines and were generated by mating grk5 Ϫ/ϩ mice. Littermate grk5 Ϫ/Ϫ and WT mice were sacrificed simultaneously for SMC production, and all comparisons were made among littermate-derived SMCs. SMCs were used only through passage 7. Infections of SMCs with recombinant adenoviruses were performed at equivalent multiplicities of infection (from 50 to 100), and cells were assayed 48 -72 h after infection, as we described previously (9). In GRK5 "add-back" experiments with grk5 Ϫ/Ϫ SMCs, assays were performed 6 h after adenovirus infection (when GRK5 expression levels were equivalent to those obtaining in WT SMCs).
To assay cell surface PDGFR␤ expression levels in transfected HEK cells, we used cell surface immunofluorescence and flow cytometry, as we described previously (7). Cell surface PDGFR␤ expression levels of all cell lines were within 30% of control cell values; cell lines outside of this range were not used to generate data. Compared with HEK cells co-transfected with empty vector plasmid, HEK cells overexpressing either GRK2 or GRK5 demonstrated equivalent (ϳ35%) reductions in PDGF-promoted phosphoinositide hydrolysis; GRK expression was 20 -40-fold over endogenous levels (data not shown).

SMC Migration and [ 3 H]Thymidine
Incorporation-SMC migration was assayed with a protocol modified from one we reported previously (10). SMCs were serum-deprived for 72 h after adenovirus infection, trypsinized, transiently (Ͻ5 min) treated with 8% fetal bovine serum (to neutralize the trypsin), and washed with low mitogen medium. Next, SMCs were plated onto Transwell TM membranes (8-m pores; Costar) in 24-well dishes and allowed to attach for 4 h. PDGF-BB or vehicle was then delivered outside of the Transwell TM membranes, and SMCs were allowed to migrate for 16 h before fixation with methanol. (Pilot studies demonstrated that originally quiescent SMCs do not divide during this time period with PDGF stimulation (data not shown).) SMC nuclei were stained with Hoechst 33342, imaged by fluorescence microscopy, and counted electronically, as we reported previously (10). SMC [ 3 H]thymidine incorporation was assessed during the final 4 h of a 24-h agonist stimulation, as we reported previously (9). Parallel aliquots of SMCs were subjected to lysis and IB and demonstrated that control and GRK5-overexpressing SMCs expressed equivalent levels of the PDGFR␤ (data not shown).
SMC Proliferation-SMC proliferation was quantitated by enzyme-linked immunosorbent assay (ELISA) for the nuclear scaffolding protein lamin (25). SMCs were plated at 5 ϫ 10 3 / well in low mitogen medium (9) on day 0 in replicate 96-well plates. On day 1, one plate was washed with PBS, fixed, and permeabilized with methanol/acetone (1:1) for 2 min, washed with PBS, and frozen at Ϫ80°C. SMCs on the second plate were refed with low mitogen medium lacking ("basal") or containing agonists at the indicated concentrations. SMCs were refed on day 6 and then harvested, washed, permeabilized, and frozen as above on day 12. Replicate frozen plates were thawed and incubated (25°C) for 30 min in "blocking buffer": 3% bovine serum albumin in PBS. Next, wells were incubated for 1 h (25°C) in blocking buffer with 1 g/ml of either an irrelevant murine IgG 2b (nonspecific signal) or an anti-lamin A/C monoclonal IgG 2b (Santa Cruz Biotechnology) (total signal). Wells were washed with PBS twice and then incubated for 1 h with antimouse/horseradish peroxidase in blocking buffer. After two PBS washes, wells were incubated in 50 l/well substrate solution: 0.1 mg of o-phenylenediamine (Sigma)/ml of 0.03% H 2 O 2 . Reactions were terminated when adequate color development was achieved (ϳ30 min), with 50 l of 1 M H 2 SO 4 . The color of each well was read at 490 nm. Specific A 490 was calculated as total Ϫ nonspecific; nonspecific signal constituted 20 -25% of total signal. Assays were performed in triplicate. The number of SMCs in agonist-stimulated wells was normalized to that in basal wells on day 12 (which, in turn, was ϳ10% above the number measured on day 1).
In Vitro Phosphorylation with Purified GRK5-Recombinant bovine GRK5 was synthesized in baculovirus-infected Sf9 insect cells, as we have previously reported (13). GRK5 was purified as before (13), except that elution from the heparin-Sepharose column was with a 100-ml linear gradient of NaCl that was 150 -1200 mM in buffer A (20 mM HEPES, 2 mM EDTA, pH 7.2) and included 0.02% (v/v) Triton X-100. Fractions at ϳ800 mM NaCl were pooled and diluted with buffer A to reduce the [NaCl] to Ͻ150 mM (buffer B). Subsequently, the diluted, purified GRK5 was concentrated by ultrafiltration and stored at 0.5 mg/ml in 50:50 (v/v) glycerol/buffer B at Ϫ20°C. By Coomassie Blue staining of SDS-polyacrylamide gels, the GRK5 preparation was ϳ95% pure.
Phosphorylation reactions were carried out with 300 nM GRK5 exactly as we described previously for GRK2 (7), except that the source of PDGFR␤ was grk5 Ϫ/Ϫ SMCs. After PDGF (or vehicle) challenge for 5 min (37°C), SMCs were solubilized and PDGFR␤s were immunoprecipitated. The PDGFR␤ for each reaction was immunoprecipitated from a confluent 100-mm plate.
Statistical Analyses-Results from multiple experiments were averaged for independent groups but analyzed pairwise, within experiments, by repeated measures analysis of variance and Tukey's post hoc test for multiple comparisons (Prism 2 TM Software, GraphPad, Inc.). Data are presented in the text as mean Ϯ S.D. and in the figures as mean Ϯ S.E.

RESULTS
In cell systems in which GRK2 mediates most of the PDGFinduced seryl phosphorylation of the PDGFR␤, persistent PDGFR␤ signaling results in down-regulation of both GRK2 (27) and the PDGFR␤ (7). To determine which widely expressed GRK (GRK2 or GRK5) mediates most of the PDGFR␤ regulation in SMCs, we first tested whether the expression level of either of these GRKs was regulated coordinately with the PDGFR␤. Indeed, with prolonged PDGF stimulation that down-regulates the PDGFR␤ (7), we observed down-regulation of GRK5, but not GRK2 (Fig. 1). To resolve the apparent paradox between these results and our data showing that endogenous GRK2 regulates PDGFR␤s in fibroblasts (8), we compared GRK expression levels in SMCs and fibroblasts.
If vascular SMC GRK5 expression is regulated coordinately with the PDGFR␤ in a physiologically meaningful way, then we should expect GRK5 to be up-regulated under pathological conditions that promote PDGFR␤ up-regulation too. To test this expectation, we examined GRK5 and PDGFR␤ expression in atherosclerosis, a pathologic process involving SMC proliferation and migration in response to myriad cytokines and growth factors (28). The "fibrous cap" of atherosclerotic lesions in the mouse comprises largely SMCs, identified in Fig. 2 by staining for ␣-SMC actin. Although atherosclerotic and normal arteries demonstrate equivalent ␣-SMC actin expression per cell, atherosclerotic arteries demonstrate substantially more PDGFR␤ and GRK5 expression per ␣-SMC actin-expressing cell (Fig. 2). Interestingly, unlike GRK5, GRK2 was not up-regulated in these atheroma SMCs (data not shown). The coordinate up-regulation of GRK5 and the PDGFR␤ in SMCs of atherosclerotic arteries, along with coordinate down-regulation of GRK5 and the PDGFR␤ in cultured SMCs, suggests that GRK5 and the PDGFR␤ in SMCs may be functionally related.
To determine possible effects of GRK5 on PDGFR␤ activity in SMCs, we began by overexpressing GRK5 in primary rabbit SMCs with a recombinant adenovirus. The prevalence of GRK overexpression was assessed by immunofluorescence microscopy to be 90 -100%, as described previously (9). In GRK5overexpressing SMCs, GRK5 was ϳ20 -30-fold overexpressed, FIGURE 1. Prolonged PDGFR␤ stimulation down-regulates GRK5, but not GRK2. Quiescent WT mouse SMCs were incubated without (control) or with 2 nM PDGF-BB for the indicated times, and then lysed. Twelve g of SMC protein were subjected to SDS-PAGE and IB (sequentially) for GRK5, GRK2, and actin. A, blots from a single experiment, representative of two performed in duplicate with independent SMC lines. Probing parallel blots with nonimmune IgG yielded no bands in the areas of interest. B, GRK band densities were divided by the cognate actin band densities, and these ratios were normalized to those obtained in control samples at each time point. The means Ϯ S.E. of two independent SMC lines are displayed. C, 20 g of protein from MEFs or SMCs of the indicated genotype were subjected to SDS-PAGE and serial IB for the indicated GRK and actin. Results are representative of three independent experiments with three cell lines of each genotype. FIGURE 2. GRK5 and the PDGFR␤ are coordinately up-regulated in SMCs within atherosclerotic lesions. The distal carotid arteries from 9-month-old C57Bl/6J WT and congenic apoe Ϫ/Ϫ (atherogenic) mice were harvested, embedded, and sectioned as described under "Materials and Methods." Serial sections were stained immunofluorescently for either ␣-SMC actin, GRK5, or the PDGFR␤, as indicated; all sections were counterstained with Hoechst 33342 for nuclear DNA. Scale bar, 50 m; original magnification was ϫ1,100. Shown are samples from two specimens stained in parallel, representative of three such specimen pairs. Specimens stained with nonimmune primary IgG yielded only DNA fluorescence (data not shown).
relative to endogenous GRK5, and PDGFR␤ expression was 100 Ϯ 15% of that seen in control SMCs ( Fig. 3A and data not shown).
GRK5 Desensitizes Heptahelical and PDGF Receptors-Phosphoinositide hydrolysis elicited through SMC heptahelical receptors was clearly desensitized by GRK5 overexpression (Fig. 3B). As we have observed in HEK cells (24), GRK5 inhibited phosphoinositide hydrolysis elicited by endothelin and PAR1 (protease-activated receptor-1). However, contrary to results in HEK cells overexpressing GRK5 (29) or SMCs overexpressing GRK2 (9), GRK5 overexpression also blunted phosphoinositide hydrolysis evoked through thromboxane A 2 receptors. This inhibition of G q -coupled receptor signaling could have been mediated at the level of G␣ q/11 subunits or by the GRK5 RGS (regulator of G protein signaling) domain (29,30). Supporting this hypothesis, GRK5 overexpression inhibited thromboxane-evoked phosphoinositide hydrolysis to a degree equivalent to that observed with fluoroaluminate (ϳ40%), which activates G proteins independently of receptors (31). In contrast, GRK5 overexpression inhibited endothelin-and PAR1-evoked phosphoinositide hydrolysis to a greater extent (ϳ70%, p Ͻ 0.05). Thus, these levels of GRK5 overexpression appeared to reduce signaling with both receptor-specific and G proteinrelated mechanisms.
Overexpression of GRK5, like GRK2 (9), also inhibited phosphoinositide hydrolysis effected by the PDGFR␤ (the only PDGFR expressed in rabbit aortic SMCs) (32) by 60% (Fig. 3B). Thus, the ability to desensitize both PDGF and heptahelical receptors appears to extend across GRK subtypes. Importantly, this inhibition of PDGFR␤-evoked phosphoinositide hydrolysis did not involve heterotrimeric G proteins. In rabbit SMCs, we found that the PDGFR␤ activates G␣ i , but not G␣ q/11 (7). Although G␤␥ i subunits can activate PLC-␤ (33), we found no evidence of such activation by the PDGFR␤ in our SMCs. Treatment of SMCs with pertussis toxin (to inactivate G i/o ) failed to affect PDGF-induced phosphoinositide hydrolysis, but eliminated G i/o -dependent (34) activation of extracellular signal-regulated kinase (ERK) by lysophosphatidic acid (data not shown). In light of these data, it seemed that overexpression of GRK5 inhibited PDGFR␤-mediated activation of PLC-␥, a tyrosine kinase-dependent event (4).
GRK5 Diminishes PDGFR␤-evoked SMC Thymidine Incorporation and Proliferation-Although it results from PDGFR␤ signaling distinct from that required for migration (4, 10, 38 -40), PDGFR␤-evoked SMC thymidine incorporation was also diminished 65-70% in GRK5-overexpressing SMCs, in response to PDGF alone or in synergistic combination with G q -coupled receptors (Fig. 4A), just as we observed with GRK2 FIGURE 3. SMC phosphoinositide hydrolysis and migration are reduced by GRK5 overexpression. SMCs were infected with adenoviruses encoding GRK5 or no protein (control, vector, or Ϫ), and assayed or solubilized for IB 3 days later. A, IB were performed on 10 g of cell lysate, with nonimmune (Neg) or anti-GRK5/6 IgG (58) recognizing the 68-kDa GRK5. In parallel, 35 g of SMC lysate was immunoblotted with nonimmune IgG or serially for PDGFR␤ and actin. Results are representative of Ն10 blots, performed with each assay. B, SMCs were metabolically labeled and exposed (37°C) to low-mitogen medium (9)  (9, 10). GRK5, like GRK2 (9), also blunted thymidine incorporation induced by the myriad agonists in fetal bovine serum (in which PDGF plays a critical role) (41). Whereas GRK5 overexpression substantially attenuated thymidine incorporation evoked by the combination of PDGF and G q -coupled receptor agonists, it failed to affect comparable thymidine incorporation evoked by PDGF plus EGF. Thus, GRK5-mediated desensitization demonstrated substrate specificity for receptor proteintyrosine kinases in a manner very similar to that observed with GRK2 (10).
To determine whether the same levels of GRK overexpression that inhibited signaling, migration, and thymidine incorporation would also inhibit SMC proliferation, aliquots of the SMC lines used for the former assays were subjected to 12-day FIGURE 4. GRK5 overexpression attenuates PDGF-promoted SMC proliferation in a receptor-specific manner. A, SMC thymidine incorporation. Quiescent SMCs infected with the indicated adenovirus were exposed to lowmitogen medium containing vehicle (basal), the indicated agonists at concentrations specified above, 5% fetal bovine serum (FBS), or PDGF plus 0.17 nM EGF. [ 3 H]Thymidine incorporation after 24 h of agonist exposure is plotted as the mean Ϯ S.E., from at least four experiments performed in triplicate. *, p Ͻ 0.05 compared with vector-infected SMCs. Uninfected and vector-infected SMCs showed indistinguishable stimulus-induced thymidine incorporation (data not shown). B, lamin ELISA. The indicated number of quiescent SMCs were plated in low mitogen medium and subjected the next day to lamin ELISA. Shown are the means Ϯ S.D. of a single experiment performed in triplicate, representative of eight performed. For A 490 versus SMC number, R 2 ϭ 0.957. C, SMC proliferation. Quiescent, adenovirus-infected SMCs (5 ϫ 10 3 /well) were exposed to low mitogen medium containing vehicle (basal), 1.5 nM fibroblast growth factor-2 (FGF), or other agonists as specified in Fig. 3. SMC proliferation after 12 days of agonist exposure was assessed by lamin ELISA and plotted as 100 ϫ ((stimulated/basal) Ϫ 1), means Ϯ S.E. from at least three independent experiments performed in quadruplicate. *, p Ͻ 0.05 compared with vector-infected SMCs (by repeated measures analysis of variance).  DECEMBER 8, 2006 • VOLUME 281 • NUMBER 49

JOURNAL OF BIOLOGICAL CHEMISTRY 37763
proliferation studies with an ELISA for the nuclear scaffolding protein lamin, which we used as a surrogate for counting nuclei. (We found an excellent correlation between SMC number and results from the lamin ELISA (Fig. 4B).) GRK5 inhibited PDGFpromoted proliferation by ϳ35% (p Ͻ 0.05) and inhibited G qcoupled receptor-promoted proliferation less consistently (Fig.  4C). Importantly, in inhibiting PDGFR␤-evoked SMC proliferation, GRK5 also demonstrated specificity for regulating receptor protein-tyrosine kinases, in that GRK5 overexpression failed to reduce SMC proliferation elicited through SMC fibroblast growth factor receptors (Fig. 4C).
In these SMC lines, the importance of GRK5 in regulating the SMC PDGFR␤ manifested itself clearly. Endogenous GRK5 reduced PDGFR␤-evoked phosphoinositide hydrolysis by 35% (p Ͻ 0.05), but had no effect on fluoroaluminate-induced (G protein-mediated) phosphoinositide hydrolysis (Fig. 6B). In  (24)) for WT and KO SMCs, respectively. C, quiescent SMCs were stimulated (or not; basal) with 2 nM PDGF-BB or 1.7 nM EGF and subjected to [ 3 H]thymidine incorporation assay, as in Fig. 4. Shown are the means Ϯ S.E. from four experiments performed with two independent pairs of WT and KO SMC lines. Basal [ 3 H]thymidine incorporation values were 2 Ϯ 1 ϫ 10 3 cpm for both WT and KO SMCs. *, p Ͻ 0.05 compared with WT (paired analysis). D, SMCs exposed for 5 min (37°C) to low mitogen medium containing vehicle or 2 nM PDGF-BB were solubilized, and anti-PDGFR␤ or cognate nonimmune IgG (Ϫ) was used for IP. Immunoprecipitates were resolved by SDS-PAGE and immunoblotted serially for the PDGFR␤ and then phosphotyrosine (pY). Relative densities of phosphotyrosine bands were quantitated as in Fig. 5B and plotted as means Ϯ S.E. from three independent experiments. Shown are blots from a single experiment, representative of three performed with at least two WT and at least two KO SMC lines each. *, p Ͻ 0.05 compared with GRK5 KO. addition, endogenous GRK5 diminished thymidine incorporation evoked by the PDGFR␤ (by 56%), but not the EGF receptor (Fig. 6C). Finally, physiologically expressed GRK5 also reduced PDGFR␤ tyrosyl phosphorylation, by 35 Ϯ 10% (Fig. 6D). Thus, physiologically expressed GRK5 mediated receptor-specific PDGFR␤ desensitization at the level of PDGFR␤ autophosphorylation/activation, second messenger signaling, and signaling further downstream from the receptor, and all in a manner congruent with that observed by comparing GRK5-expressing with GRK5-overexpressing SMCs (Figs. 3-5).
To ascertain that GRK5 itself was responsible for the excess PDGFR␤ seryl phosphorylation we observed in GRK5-expressing SMCs, we used purified GRK5 to phosphorylate the partially purified PDGFR␤ in vitro (Fig. 7B). In the absence of purified GRK5, we found some PDGF-dependent PDGFR␤ seryl phosphorylation (Fig. 7B, lane 2). This PDGFR␤ seryl phosphorylation, however, could be attributed to intracellular Ser/Thr kinases, acting before PDGFR␤ IP (as in grk5 Ϫ/Ϫ SMCs, in Fig.  7A). As a result of purified GRK5 activity in vitro, this agonistdependent PDGFR␤ seryl phosphorylation increased ϳ2-fold (Fig. 7B). Purified GRK5 activity showed an even larger relative increase in seryl phosphorylation with PDGFR␤s obtained from unstimulated SMCs (ϳ4-fold). However, it should be noted that this "agonist-independent" effect was not independent of PDGFR␤ activation. IgG in our immune complex kinase assay dimerizes immunoprecipitated PDGFR␤s, and thereby promotes PDGFR␤ autophosphorylation/activation (Fig. 7B), which is a prerequisite for GRK-mediated PDGFR␤ phosphorylation (7).
In these experiments with vascular SMCs, physiologically expressed GRK5 phosphorylated and desensitized the PDGFR␤ in a manner resembling that of GRK2 expressed physiologically in fibroblasts (8). Indeed, although GRK2 appeared to mediate most of the agonist-induced PDGFR␤ Ser phosphorylation in fibroblasts (8), GRK5 mediated most of the agonist-induced FIGURE 7. GRK5 serine-phosphorylates the PDGFR␤ in intact SMCs and in purified protein preparations. A, phosphorylation in intact SMCs. SMCs were stimulated and processed for PDGFR␤ IP and IB just as in Fig. 6D, except that serial IB was performed for the PDGFR␤ and then phosphoserine (pSer), not phosphotyrosine. Shown are blots from a single experiment, representative of three performed with at least two WT and at least two KO SMC lines each. Relative densities of pSer bands were quantitated as in Fig. 5B and plotted as means Ϯ S.E. from three independent experiments. *, p Ͻ 0.05 compared with GRK5 KO. B, PDGFR␤ phosphorylation with purified GRK5. PDGFR␤s were immunoprecipitated from grk5 Ϫ/Ϫ SMCs that had been stimulated and processed just as in A. After IP, purified GRK5 was added to PDGFR␤ immune complexes, and phosphorylation proceeded (35°C) for 30 min. Immune complexes were then pelleted; supernatant GRK5 and pelleted PDGFR␤s were resolved by separate SDS-PAGE procedures and subjected to IB. The PDGFR␤ sample was divided and probed in parallel for PDGFR␤ and then pSer (sequentially) or phosphotyrosine. Shown are blots from a single experiment, representative of four performed with two grk5 Ϫ/Ϫ SMC lines. Relative densities of phosphoserine bands were normalized to cognate PDGFR␤ band densities ("arbitrary units"), averaged across four independent experiments, and plotted as means Ϯ S.E. *, p Ͻ 0.02 compared with PDGFR␤s from cells treated without PDGF. #, p Ͻ 0.01 compared with PDGFR␤s incubated without purified GRK5.
PDGFR␤ Ser phosphorylation in SMCs. Furthermore, GRK5 augmented PDGFR␤ seryl phosphorylation as it does with heptahelical receptors (12), in an agonist-dependent manner and on a rapid time scale congruent with desensitization of second messenger production (seen in Fig. 6B).
Does the predominance of SMC GRK5 in PDGFR␤ seryl phosphorylation correlate with a predominance of GRK5 in PDGFR␤ desensitization? To address this question, we sought to determine the relative contributions of GRK5 and GRK2 to PDGFR␤ regulation in SMCs. To that end, we used siRNA to reduce SMC expression of either GRK2 or GRK5, and assessed the effect of GRK knock-down on phosphoinositide hydrolysis. As we observed with GRK5-null and WT SMCs (Fig. 6B), G protein (fluoroaluminate)-evoked phosphoinositide hydrolysis was unaffected by changes in GRK expression (Fig. 8A). In contrast, PDGF-evoked phosphoinositide hydrolysis was enhanced (ϳ35%) by reduction in the expression of just GRK5, and not GRK2 (Fig. 8A). Moreover, this GRK-specific difference obtained even though the siRNA-mediated knock-down of GRK2 was somewhat more efficacious than that for GRK5 (Fig.  8, B and C). Consequently, the GRK isoform that regulates PDGFR␤ signaling in SMCs predominantly is GRK5, and not GRK2.
How could GRK5-mediated Ser phosphorylation of the PDGFR␤ augment Shp2 recruitment to the PDGFR␤? To address this question, we asked whether GRK5 activity affected phosphorylation of PDGFR␤ Tyr 1009 , since phospho-Tyr 1009 is the primary PDGFR␤ docking site for Shp2 (4). With IgG specific for the Tyr 1009 -phosphorylated PDGFR␤, we found that Tyr 1009 was hyperphosphorylated in GRK5-expressing as compared with GRK5-null SMCs (Fig. 9B). Thus, there was a greater prevalence of Tyr 1009 -phosphorylated PDGFR␤s in GRK5expressing SMCs, and consequently a greater prevalence of PDGFR␤s capable of recruiting Shp2.
This finding demonstrates that the GRK5-mediated reduction in overall PDGFR␤ tyrosyl phosphorylation is site-specific. Indeed, although GRK5 activity enhanced phosphorylation of PDGFR␤ Tyr 1009 , it substantially diminished phosphorylation of PDGFR␤ Tyr 1021 (Fig. 9C). PDGFR␤s from GRK5-expressing SMCs demonstrated 9 Ϯ 3-fold less phospho-Tyr 1021 than PDGFR␤s from GRK5-null SMCs (p Ͻ 0.05). This GRK5-associated reduction in PDGFR␤ phospho-Tyr 1021 would be expected to reduce PLC-␥1/PDGFR␤ association and consequent PLC-␥1-mediated phosphoinositide hydrolysis (4), just as we observed with these SMCs in Fig. 6B. Such site-specific reduction in PDGFR␤ tyrosyl phosphorylation probably explains why overall PDGFR␤ tyrosyl phosphorylation is only modestly reduced in GRK5-expressing (as compared with GRK5-null) SMCs (Fig. 6D). FIGURE 8. The PDGFR␤ is regulated in SMCs predominantly by GRK5 and not GRK2. WT SMCs were treated with siRNA that targeted the mRNA for no known protein (Control), for GRK2, or for GRK5. Subsequently, the SMCs were metabolically labeled with [ 3 H]inositol and, 72 h after siRNA treatment, stimulated as in Fig. 6 or processed for IB. A, inositol phosphates expressed as stimulated/basal were averaged among two independent SMC lines (with each siRNA-treated group assayed in triplicate); means Ϯ S.E. are plotted. *, p Ͻ 0.05 compared with control SMCs. Basal inositol phosphate values were 2.3 Ϯ 0.8, 2.6 Ϯ 0.4, and 2.3 Ϯ 0.3 (percent conversion units (24)) for SMCs treated with control, GRK2, and GRK5 siRNA, respectively. B, extracts (35 g of protein) from SMCs treated with the indicated siRNA were subjected to SDS-PAGE and IB. Blots were probed serially for GRK2, GRK5, and actin. Shown are the results of a single experiment, representative of three performed. Serial IB for PDGFR␤ showed equivalent PDGFR␤ expression in all SMC groups (not shown). C, densitometry of GRK bands was normalized to corresponding actin bands on each blot, and these ratios were normalized to cognate ratios obtained from SMCs transfected with control siRNA, to obtain the percentage of control. Shown are the mean Ϯ S.E. of three experiments. *, p Ͻ 0.05 compared with control cells.
To confirm that physiologic GRK5 expression was responsible for the differences in PDGFR␤ phosphorylation and signaling we observed between GRK5-expressing and -null SMCs, we took two approaches. First, as described above, we obtained congruent results from five pairs of WT and cognate GRK5-null SMC lines. Second, we used our GRK5 adenovirus to express GRK5 at 104 Ϯ 8% of WT levels in GRK5-null SMCs, to test whether "rescuing" GRK5 expression would convert a GRK5-null to a WT SMC phenotype (Fig. 10). For this purpose, we assayed PDGF-induced phosphorylation of the PDGFR␤ Tyr 1021 and found that GRK5 "rescue" expression in GRK5-null SMCs reduced phosphorylation of PDGFR␤ Tyr 1021 by 7 Ϯ 3-fold (p Ͻ 0.05) (Fig. 10), and enhanced PDGFR␤/Shp2 association by ϳ3-fold (Fig. 10). These results were remarkably congruent, of course, with those obtained by comparing WT and grk5 Ϫ/Ϫ SMCs in Fig. 9. Thus, whether expressed endogenously or heterologously, physiologic levels of GRK5 expression mediate PDGFR␤ desensitization.
Thus far, we have correlated GRK5-mediated seryl phosphorylation of the PDGFR␤ with diminished PDGFR␤ tyrosyl phosphorylation, desensitization of PDGFR␤-evoked SMC signaling, and enhancement of PDGFR␤/Shp2 association. To demonstrate more directly that GRK5-mediated PDGFR␤ regulation involves Shp2, we compared GRK5-mediated desensitization of the WT PDGFR␤ and Y1009F mutant PDGFR␤, which recruits Shp2 poorly (42) (Fig. 11, A and B). To compare these PDGFR␤s under conditions of comparable GRK5 levels (Fig. 11D), we used HEK cells (which lack endogenous PDGFR␤s) (27). We used phosphorylation of PDGFR␤ Tyr 1021 and Tyr 740 as read-outs for PDGFR␤ activation. Correlating again with enhancement of Shp2/PDGFR␤ association (Fig.  11B), increased cellular GRK5 activity substantially reduced  phosphorylation of Tyr 1021 in the WT PDGFR␤, but not in the Y1009F PDGFR␤ (Fig. 11, A and C). Likewise, increased cellular GRK5 activity approximately halved phosphorylation of Tyr 740 in the WT, but not in the Y1009F PDGFR␤ (Fig. 11, A and C). Thus, GRK5-mediated desensitization of the PDGFR␤ appears to require the PDGFR␤ (a) to recruit Shp2 normally and (b) to enhance this recruitment consequent to GRK5-mediated PDGFR␤ phosphorylation.
If Shp2 is required for GRK5-promoted desensitization of the PDGFR␤, then deficiency of Shp2 should diminish differences observed between GRK5-expressing and -null SMCs. To test this expectation, we reduced SMC Shp2 expression with RNA interference and assessed PDGF-induced phosphoinositide hydrolysis (Fig. 12). The reduction in Shp2 expression achieved with siRNA in these experiments was only ϳ40% (at least in part because of the long half-life of Shp2 (43)) (Fig. 12B). Nonetheless, by augmenting PDGF-induced phosphoinositide hydrolysis in WT SMCs, Shp2 knock-down did diminish the difference in PDGFR␤ signaling between WT and grk5 Ϫ/Ϫ SMCs (Fig. 12C), and thereby attenuated GRK5-mediated PDGFR␤ desensitization. Thus, Shp2 does appear to be an effector of GRK5-promoted PDGFR␤ desensitization.
PDGFR␤/Shp2 Association Is Augmented by the Activity of GRK5, but Not GRK2-To determine whether an Shp2-based mechanism for PDGFR␤ desensitization was specific for GRK5, we tested whether physiologic expression of GRK2 also augments recruitment of Shp2 to the PDGFR␤. For this purpose, we used GRK2-null MEFs, stably transfected to express physiologic levels of (or no) GRK2 (8). We found that the association of the PDGFR␤ and Shp2 was indistinguishable in the absence and presence of GRK2 activity (Fig. 13A), despite the fact that GRK2 activity effected a ϳ25% reduction in overall PDGFR␤ tyrosyl phosphorylation and a ϳ50% reduction in PDGF-induced phosphoinositide hydrolysis (8) (data not shown). To compare GRK2 with GRK5 activity in the same cellular milieu, we overexpressed GRK2 or GRK5 in HEK cells expressing equivalent levels of PDGFR␤s. Although GRK5 enhanced PDGFR␤/Shp2 association by 180 Ϯ 60%, GRK2 reduced this association by 60 Ϯ 20% (p Ͻ 0.05 for each) (Fig. 13B). In this same system, GRK2 overexpression diminished the association of the PDGFR␤ with the Na ϩ /H ϩ exchanger regulatory factor, as we observed before (8); however, despite its effect on PDGFR␤/Shp2 association, GRK5 overexpression had no effect on PDGFR␤/Na ϩ /H ϩ exchanger regulatory factor association (data not shown). Together, our results in SMCs, MEFs, and 293 cells suggest that GRK5-and GRK2-mediated PDGFR␤ desensitization result from discrete molecular mechanisms. FIGURE 11. GRK5-mediated desensitization of the PDGFR␤ requires intact Shp2/PDGFR␤ association. HEK 293 cells were transfected with plasmids encoding a human N-terminal FLAG-tagged PDGFR␤ construct (WT or Y1009F), GRK5 ("high" GRK5 level), or no protein (Vector, "native" GRK5 level). Cells were exposed to medium containing vehicle (Ϫ) or 2 nM PDGF-BB (ϩ) for 5 min (37°C), and then lysed and subjected to IP of the indicated PDGFR␤ construct. Divided IP samples were subjected to parallel SDS-PAGE and sequential IB (with intervening membrane stripping) for the PDGFR␤ and then either the PDGFR␤ phosphorylated at Tyr 1021 (pY-1021), the PDGFR␤ phosphorylated at Tyr 740 (pY-740), or Shp2. A, blots from a single experiment are displayed and represent three experiments performed with similar results. HEK cells transfected without a PDGFR␤ construct yielded no signals on these blots, and all cell lines expressed equivalent levels of Shp2 (data not shown). B, quantitation of Shp2/PDGFR␤ association. Shp2 band densities were normalized to cognate PDGFR␤ band densities; each ratio was normal-ized to that obtained from IPs of PDGF-stimulated HEK cells transfected with the WT PDGFR␤ and empty vector ("control" cells) to obtain the "percentage of control." *, p Ͻ 0.05 compared with cognate cells expressing native GRK5 levels. C, quantitation of PDGFR␤ pY-1021 and pY-740 data. Band densities for pY-1021 and pY-740 were normalized to cognate PDGFR␤ band densities, and data (mean Ϯ S.E. from three independent experiments) were processed as in B. *, p Ͻ 0.05 compared with cognate cells expressing native GRK5 levels. D, WT and Y1009F PDGFR␤ cells express equivalent levels of GRK5. Lysates from control (40 g of protein) and GRK5-overexpressing cells (10 g of protein) were immunoblotted with A16/17 anti-GRK5 or nonimmune mouse IgG 1 (non) (58).

GRK5 Potentiates PDGFR␤-induced Activation of Src, but
Not ERK-The association of Shp2 with the PDGFR␤ is believed to activate Shp2 (4). Consequently, since GRK5 activity in SMCs augments Shp2/PDGFR␤ association, we expected to observe not only reduced PDGFR␤ tyrosyl phosphorylation but also other evidence of enhanced PDGF-induced Shp2 activity. To test this expectation, we examined PDGF-promoted ERK1/2 and Src activation, which can be mediated by Shp2 (4,44). Despite large differences in PDGFR␤/Shp2 association (Fig. 9), GRK5-expressing and -null SMCs demonstrated equivalent activation of ERK-1 and -2 within 5 min of PDGF stimulation (data not shown). This finding is consonant with data from Shp2-deficient fibroblasts. Even the absence of Shp2 does not diminish PDGFR␤-promoted ERK activation within 5 min of PDGF stimulation (44). However, GRK5-expressing SMCs demonstrated 1.9 Ϯ 0.3-fold more Src activation than GRK5null SMCs (Fig. 14, A and B), even though GRK5 activity did not affect PDGFR␤ autophosphorylation on its docking site for Src (Tyr 579 ) (Fig. 14C). Thus, GRK5-mediated seryl phosphorylation of the PDGFR␤ desensitizes signaling selectively. Although PLC-␥1 and perhaps other pathways promoting SMC migration and proliferation are desensitized, Src signaling is augmented. In this way, GRK5-mediated phosphorylation of the PDGFR␤ mirrors GRK-mediated phosphorylation of heptahelical receptors, a process which desensitizes signaling through  Fig. 6 or processed for IB. A, extracts (30 g of protein) from SMCs treated with the indicated siRNA were subjected to SDS-PAGE and IB. Blots were probed serially for Shp2 and tubulin. Shown are the results of a single experiment, representative of three performed. Subsequent PDGFR␤ IB demonstrated that Shp2 siRNA did not affect PDGFR␤ expression (not shown). B, densitometry of Shp2 bands was normalized to corresponding tubulin bands on each blot, and these ratios were normalized to cognate ratios obtained from WT SMCs transfected with control siRNA, to obtain the "percentage of WT control." Shown are the mean Ϯ S.E. of three experiments. *, p Ͻ 0.05 compared with WT control cells. C, inositol phosphates obtained from stimulated SMCs were divided by inositol phosphates obtained from corresponding unstimulated SMCs (-fold/basal); these ratios were normalized within each experiment to the cognate -fold/basal value obtained for PDGF-stimulated WT SMCs treated with control siRNA (6 Ϯ 2-fold/basal), to obtain the "percentage of WT control." Plotted are means Ϯ S.E. from six independent experiments (two with each of the three WT/KO SMC pairs). *, p Ͻ 0.05 compared with cognate control siRNAtreated SMCs. Basal inositol phosphate values were 5 Ϯ 3, 5 Ϯ 2, 7 Ϯ 1, and 7 Ϯ 2 (percentage of conversion units (24)) for WT and KO SMCs treated with control and Shp2 siRNA, respectively. FIGURE 13. GRK2 activity fails to enhance Shp2/PDGFR␤ association. A, GRK2 KO MEFs stably transfected with vector (KO) or GRK2 plasmid expressing physiologic levels of GRK2 (WT) were processed as in Fig. 9 for co-IP of Shp2 with the PDGFR␤. B, HEK 293 cells transiently transfected with the indicated plasmids were processed as in A, and IP blots were probed sequentially for Shp2 and PDGFR␤. All transfected HEK cell lines expressed equivalent levels of Shp2 (data not shown). Results in A and B are from single experiments, representing at least three performed with independent cell lines.

DISCUSSION
This study demonstrates for the first time that the PDGFR␤ is phosphorylated and desensitized by GRK5, a widely expressed kinase previously known only to regulate a multitude of heptahelical receptors (11). Moreover, GRK5 mediates the preponderance of PDGF-induced seryl phosphorylation and desensitization of the PDGFR␤ in SMCs, and reduces overall PDGFR␤ tyrosyl phosphorylation in a manner that is highly site-specific. Whether assessed as second messenger production, migration, thymidine incorporation, or proliferation, PDGFR␤-promoted SMC activity was reduced in a receptorspecific manner by GRK5 activity. Although GRK5-mediated FIGURE 15. Proposed scheme for GRK5-mediated regulation of the PDGFR␤. A, the agonist-activated, dimerized, and autophosphorylated PDGFR␤ is schematically depicted in a caveola, into which it appears to migrate after activation (51). The PDGFR␤, which can activate heterotrimeric G i (7), allosterically activates GRK5, which may be localized to caveolae by its binding to caveolins-1 (49). Once activated, the GRK5 phosphorylates the activated PDGFR␤ on seryl residues. B, by as yet unknown mechanisms, GRK5mediated phosphorylation of PDGFR␤ seryl residues leads to enhanced phosphorylation of PDGFR␤ Tyr 1009 , the docking site for Shp2 on the PDGFR␤; consequently, recruitment of Shp2 to the PDGFR␤ is enhanced. Shp2 then dephosphorylates selected phosphotyrosyl residues on the PDGFR␤ (like phospho-Tyr 1021 and phospho-Tyr 740 (Figs. 9 -11)); the resulting decrease in PDGFR␤ activity is symbolized by removal of the asterisk from R*. C, because GRK5 activity on the PDGFR␤ augments Shp2/PDGFR␤ association and Shp2 activation without affecting the ability of PDGFR␤ to recruit Src (Fig. 14C), PDGF-induced Src activation is enhanced in GRK5-expressing SMCs. With increased levels of activated Shp2 recruited to the PDGFR␤, there is greater Shp2-mediated dephosphorylation of Src at its (autoinhibitory) phospho-Tyr 527 , and consequently greater Src activation (4). Thus, overall, GRK5 activity desensitizes PDGFR␤ signaling through phospholipase C␥-1 and phosphatidylinositol 3-kinase (but not ERK), and promotes PDGFR␤ signaling via Src. Activated molecules are indicated by an asterisk and/or by shading; pS, phosphoserine; pY, phosphotyrosine. A, quiescent GRK5-null and WT SMCs were stimulated (or not) with 2 nM PDGF-BB for 10 min (37°C) and lysed; 20 g of protein from each cell group underwent parallel SDS-PAGE and IB for either activated Src, phosphorylated on Tyr 416 (pY-416), or total Src (bottom). B, Src pY-416 band density was normalized to cognate Src band density, and ratios were normalized to those obtained for PDGF-stimulated GRK5-null SMCs to obtain the "percentage of GRK5 KO." Shown are means Ϯ S.E. from eight experiments performed with four paired KO and WT SMC lines. *, p Ͻ 0.05 compared with GRK5-null. C, WT and GRK5 KO SMCs were stimulated (or not) with PDGF-BB and subjected to PDGFR␤ IP, SDS-PAGE, and IB as in Fig. 9. Blots were probed serially for the PDGFR␤ and then the PDGFR␤ phosphorylated on Tyr 579 (one of two Src docking sites (4)). Shown are the results of a single experiment, representative of three performed.