Phosphorylation of the platelet-derived growth factor receptor-beta by G protein-coupled receptor kinase-2 reduces receptor signaling and interaction with the Na(+)/H(+) exchanger regulatory factor.

G protein-coupled receptor kinase-2 (GRK2) can phosphorylate and desensitize the platelet-derived growth factor receptor-beta (PDGFRbeta) in heterologous cellular systems. To determine whether GRK2 regulates the PDGFRbeta in physiologic systems, we examined PDGFRbeta signaling in mouse embryonic fibroblasts from GRK2-null and cognate wild type mice. To discern a mechanism by which GRK2-mediated phosphorylation can desensitize the PDGFRbeta, but not the epidermal growth factor receptor (EGFR), we investigated effects of GRK2-mediated phosphorylation on the association of the PDGFRbeta with the Na(+)/H(+) exchanger regulatory factor (NHERF), a protein shown to potentiate dimerization of the PDGFRbeta, but not the EGFR. Physiologic expression of GRK2 diminished (a) phosphoinositide hydrolysis elicited through the PDGFRbeta but not heterotrimeric G proteins; (b) Akt activation evoked by the PDGFRbeta but not the EGFR; and (c) PDGF-induced tyrosyl phosphorylation of the PDGFRbeta itself. PDGFRbeta desensitization by physiologically expressed GRK2 correlated with a 2.5-fold increase in PDGF-promoted PDGFRbeta seryl phosphorylation. In 293 cells, GRK2 overexpression reduced PDGFRbeta/NHERF association by 60%. This effect was reproduced by S1104D mutation of the PDGFRbeta, which also diminished PDGFRbeta activation and signaling (like the S1104A mutation) to an extent equivalent to that achieved by GRK2-mediated PDGFRbeta phosphorylation. GRK2 overexpression desensitized only the wild type but not the S1104A PDGFRbeta. We conclude that GRK2-mediated PDGFRbeta seryl phosphorylation plays an important role in desensitizing the PDGFRbeta in physiologic systems. Furthermore, this desensitization appears to involve GRK2-mediated phosphorylation of PDGFRbeta Ser(1104), with consequent dissociation of the PDGFRbeta from NHERF.

The platelet-derived growth factor receptor-␤ (PDGFR␤) 1 plays a pivotal role in development, in the growth and main-tenance of mesenchymal cells, and in pathologic proliferative processes like malignant neoplasia (1, 2) and atherosclerosis (3)(4)(5). Maintaining cellular homeostasis and bridling neoplasia therefore requires precise regulation of signaling through this receptor protein tyrosine kinase. Mechanisms described thus far for long and short term regulation of the PDGFR␤ include degradation and down-regulation of cellular PDGFR␤s (6,7), tyrosyl dephosphorylation (8,9), and phosphorylation of the PDGFR␤ on serine residues (10,11).
We recently reported that PDGFR␤ desensitization can be effected by GRK2-mediated phosphorylation of the PDGFR␤ on serine residues (10). GRK2 is a ubiquitously expressed member of the GRK family of serine/threonine kinases and has previously been characterized by its ability to phosphorylate and desensitize a vast array of heptahelical receptors (12). As an allosteric enzyme, GRK2 is activated by agonist-occupied receptors (12). GRK2 phosphorylates the PDGFR␤ in an agonistdependent manner (10), just as it phosphorylates heptahelical receptors. GRK2-mediated PDGFR␤ desensitization manifests in short term assays for phosphoinositide hydrolysis (13) and phosphatidylinositol 3-kinase activation (14) and in long term assays for [ 3 H]thymidine incorporation, proliferation (13), and cellular migration (14). Moreover, GRK2-mediated PDGFR␤ seryl phosphorylation reduces the activation of the PDGFR␤ itself, as assessed by receptor tyrosyl phosphorylation (10,13). In contrast, GRK2 overexpression does not desensitize the EGFR, even though GRK2 phosphorylates EGFR serine(s) in purified protein and cellular overexpression systems (10,13).
To discern the mechanisms by which GRK2-mediated RPTK phosphorylation could desensitize the PDGFR␤ but not the EGFR, we have focused this investigation on the interaction between the PDGFR␤ and the Na ϩ /H ϩ exchanger regulatory factor (NHERF), a PDZ domain-containing protein known to bind to the PDGFR␤, but not the EGFR, via the C-terminal tetrapeptide Asp-Ser 1104 -Phe-Leu of the receptor (15). The association of the PDGFR␤ with NHERF (also known as EBP50) has been shown to potentiate PDGFR␤ dimerization and signaling (15). NHERF also binds to the C-terminal tetrapeptide Asp-Ser 411 -Leu-Leu of the ␤ 2 -adrenergic receptor, in an agonistdependent manner (16). Association between NHERF and the ␤ 2 -adrenergic receptor affects receptor-mediated regulation of * This work was supported in part by National Institutes of Health Grants HL63288 (to N. J. F.) and HL64744 (to K. P.) as well as a grant-in-aid from the American Heart Association (to N. J. F.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  Na ϩ /H ϩ exchange (16) as well as receptor recycling (17). This ␤ 2 -adrenergic receptor/NHERF association appears to be disrupted by GRK5-mediated phosphorylation of the ␤ 2 -adrenergic receptor on Ser 411 (17) (a site not phosphorylated by GRK2 (18)). By analogy with the action of GRK5 on the ␤ 2 -adrenergic receptor, we hypothesized that GRK2 could phosphorylate the PDGFR␤ on Ser 1104 and thereby decrease the affinity of NHERF/PDGFR␤ interaction, diminish PDGFR␤ dimerization and activation, and reduce downstream signaling. To test this hypothesis, we examined the effect of GRK2-mediated PDGFR␤ phosphorylation on NHERF/PDGFR␤ association. In addition, to determine whether GRK2-mediated PDGFR␤ phosphorylation and desensitization occur in physiologic systems, we examined PDGFR␤ regulation in MEFs derived from GRK2 Ϫ/Ϫ and cognate WT mice.

MATERIALS AND METHODS
Plasmid Constructs-Plasmids encoding the N-terminal FLAGtagged human PDGFR␤ and bovine GRK2, each in pcDNA I (Invitrogen), have been described (10,19). The plasmid encoding an N-terminal hemagglutinin-tagged rabbit NHERF was the kind gift of Randy Hall (15). The bovine GRK2 cDNA was subcloned into pcDNA 3.1/hygro (Invitrogen) by using HindIII sites that flanked the insert (19). To create S1104D and S1104A mutations in the PDGFR␤, we employed cassette PCR with the following mutagenic primers: 5Ј-gggcgcgcggccgcctacaggaagtcatcctctgcttccgcccgaggcgc-3Ј (S1104D) and 5Ј-gggcgcgcggccgcctacaggaaggcatcctctgcttccgcccgaggcgc-3Ј (S1104A) (the underscore indicates the mutation to Asp or to Ala, italics indicate the 3Ј NotI site, and bold type denotes the stop codon). The 5Ј (nonmutagenic) primer comprised nucleotides 3061-3080 of the native PDGFR␤ sequence. After PCR, a BstEII/NotI fragment was subcloned into the FLAGtagged PDGFR␤ construct, and the fidelity of mutagenesis was confirmed by dideoxy sequencing.
Cell Culture-MEFs were derived from embryonic day 10 embryos by the explant outgrowth and 3T3 approaches (20) and propagated in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin (medium A). GRK2 Ϫ/Ϫ and WT littermate embryos were obtained from GRK2 Ϫ/ϩ matings among C57Bl/6/129 hybrid mice and were genotyped by Southern blotting of genomic DNA, as described (21). All of the animal protocols were approved by the Institutional Animal Care and Use Committee and complied with the Guide for the Care and Use of Laboratory Animals.
To increase the number of cell lines expressing physiologic levels of GRK2 and the number of GRK2 ϩ/ϩ MEF lines available for genetically valid comparisons, we stably transfected our two GKR2 Ϫ/Ϫ MEF lines with the GRK2/pcDNA3.1/hygro plasmid (using LipofectAMINE 2000; Invitrogen) according to the manufacturer's protocol. The colonies were selected in medium A containing hygromycin at 250 g/ml.
HEK 293 cells were propagated and transfected as previously reported (10). For each transfection, cell surface PDGFR␤ expression was measured by immunofluorescence and flow cytometry, as described (22). Co-transfected and mutant PDGFR␤ cell lines were used only when cell surface PDGFR␤ density was within 30% of that measured in control cells. GRK2 overexpression levels ranged from 20-to 40-fold over endogenous levels, assessed by immunoblotting serially diluted specimens, and was equivalent for all GRK2-overexpressing cells used within a single experiment.
Phosphoinositide Hydrolysis-MEFs were labeled in medium B (medium A without fetal bovine serum and with 0.1% (w/v) bovine serum albumin) with [ 3 H]myoinositol (PerkinElmer Life Sciences) for 18 h, as described (13), except that PDGF-AA (Calbiochem) was added to the labeling medium at 10 ng/ml to down-regulate the PDGF␣-receptors (23) expressed by the MEFs. MEFs were then challenged with the indicated agonist for 15 min (37°C) before being lysed and processed for isolation of total inositol phosphates by anion exchange chromatography, as described (22). HEK 293 cells were also challenged with the indicated stimuli for 15 min (37°); they were metabolically labeled in medium C (minimal essential medium, 0.1% (w/v) bovine serum albumin, penicillin/streptomycin, 1% (v/v) nonessential amino acids, and 1% pyruvate; Invitrogen) and assayed as described (13,22). Data processing involved normalizing total inositol phosphates to the total amount of [ 3 H]myoinositol taken up by the cells to obtain the percent conversion of 3 H to inositol phosphates, as described (22).
Immunoprecipitations and Immunoblotting-These procedures were performed as described previously (10,13,14). Cells were serumstarved for 16 h in medium B (MEFs) or C (293 cells) before assays. For co-IP experiments, the cell proteins were cross-linked with dithiobis(succinimidyl)propionate (Pierce) as described (10,22). Immunological reagents were those described previously (10,13,24), with the addition of the following IgGs: rabbit anti-EBP50 (NHERF) (Calbiochem, Inc.), mouse anti-PLC␥-1 and rabbit anti-GRK2 (Santa Cruz Biotechnology, Inc.), and rabbit anti-phosphoserine (Chemicon, Inc.). To confirm antibody specificity for endogenous proteins, parallel blots were performed with nonimmune IgG of species and isotype identical to that used for identifying proteins of interest. For sequential immunoblots, anti-PDGFR␤ IgG was always used before anti-phosphoserine or antiphosphotyrosine IgG, because the latter two IgGs failed to desorb from PDGFR␤ bands (data not shown). To quantitate PDGFR␤ tyrosyl and seryl phosphorylation, we performed densitometry on chemiluminescence-exposed BioMax MR TM film (Kodak). Anti-phosphotyrosine-or anti-phosphoserine-scored PDGFR␤ band densities were normalized to cognate PDGFR␤ band densities. (Because phosphotyrosine and phosphoserine signals were negligible in the absence of agonist, only PDGFR␤ bands from PDGF-stimulated cells were used for comparisons among cell lines.) To quantitate NHERF/PDGFR␤ co-IP, NHERF band densities from co-immunoprecipitations were normalized to cognate PDGFR␤ band densities. Phospho-Akt and phospho-ERK1/2 band densities were normalized to cognate actin band densities. Statistical Analyses-Independent means were compared with unpaired t tests, and one-way analysis of variance (with Tukey's post-hoc test for multiple comparisons) was used for comparisons among more than two groups, with Prism 2 software (GraphPad, Inc.). The data are presented as the means Ϯ S.D. in the text and as the means Ϯ S.E. in the figures.

Phosphorylation and Desensitization of the PDGFR␤ by
Physiologic Levels of GRK2-Although we have found that GRK2 phosphorylates and desensitizes the PDGFR␤ both in cellular overexpression and in purified protein systems, we have yet to determine whether physiologically expressed GRK2 regulates the PDGFR␤. To address this question, we began by comparing endogenous PDGFR␤ signaling in GRK2 Ϫ/Ϫ MEFs with that in cognate WT MEFs. Physiologic expression of GRK2 in WT MEFs reduced PDGFR␤-evoked phosphoinositide hydrolysis by 2.3-fold, compared with that observed in GRK2 Ϫ/Ϫ MEFs (Fig. 1A). This effect of GRK2 was receptorspecific, because physiologic GRK2 expression had no effect on phosphoinositide hydrolysis elicited in these cells through the protease-activated receptor-1 (thrombin receptor) (Fig. 1A), which we have previously shown is a relatively poor GRK2 substrate (25). To determine whether random genetic variation could be confounding our comparisons among WT and GRK2 Ϫ/Ϫ MEF lines, we phenotypically characterized each cell line by immunoblotting cell extracts for key proteins in the PDGFR␤-evoked phosphoinositide hydrolysis signaling pathway (Fig. 1B). Although distinctly different with regard to GRK2 expression, both of the WT and both of the GRK2 Ϫ/Ϫ cell lines expressed comparable (or functionally equivalent) levels of PDGFR␤, PLC␥-1, and NHERF. Thus, our inferences about the role of GRK2 on PDGFR␤ signaling were facilitated.
To minimize the effect of undetected genetic drift on our comparisons among GRK2 Ϫ/Ϫ and WT MEFs, we used our two GRK2 Ϫ/Ϫ MEF lines to create MEF clones stably transfected to express either GRK2 (n ϭ 4) or no protein (GRK2 0 cells, n ϭ 4). We selected GRK2-transfected clones that expressed GRK2 at levels comparable with those observed in WT MEFs ( Fig. 2A). In addition, clones were selected on the basis of expressing comparable levels of proteins that could affect PDGFR␤ signaling, including the PDGFR␤ itself, PLC␥-1, GRK5, GRK6, and NHERF ( Fig. 2A). In results averaged from these four GRK2expressing and 4 GRK2 0 clones, physiologic levels of GRK2 expression reduced PDGFR␤-evoked phosphoinositide hydrolysis by 38% (p Ͻ 0.01) but had no effect on phosphoinositide hydrolysis evoked through direct activation of heterotrimeric G proteins (Fig. 2B). Furthermore, physiologic levels of GRK2 expression reduced Akt activation evoked by PDGF (by 38 Ϯ 12% (p Ͻ 0.05)) but not by epidermal growth factor (Fig. 3A). Thus, GRK2-mediated desensitization of the PDGFR␤ was receptor-specific, as it was in the comparison of WT with GRK2 Ϫ/Ϫ MEFs (Fig. 1). In addition, PDGFR␤ desensitization by physiologic levels of GRK2 was also signal pathway-specific, as we have observed previously in GRK2 overexpression systems (14). Whereas PDGFR␤ signaling through PLC␥ and phosphatidylinositol 3-kinase was attenuated by GRK2 (Figs. 2B and 3A, respectively), signaling through Grb2/Sos1/Ras to ERK1/2 was not (data not shown). Lastly, physiologic levels of GRK2 expression reduced activation of the PDGFR␤ itself. Tyrosyl phosphorylation of the PDGFR␤ was 25 Ϯ 5% less in GRK2-expressing cells than in GRK2 0 cells (p Ͻ 0.05; Fig. 3B). Thus, by comparing cells expressing physiologic GRK2 levels with cells expressing no GRK2, we found GRK2-mediated PDGFR␤ desensitization to be remarkably similar to that observed by comparing GRK2-overexpressing with endogenous GRK2-expressing cells (10,13,14).
In cellular overexpression and purified protein systems, we previously found that GRK2 phosphorylated the PDGFR␤ on serine(s), in the short time frame (minutes) congruent with GRK2-mediated PDGFR␤ desensitization (10,13,14). In this regard, GRK2-mediated PDGFR␤ desensitization paralleled GRK2-mediated heptahelical receptor desensitization. To determine whether GRK2 phosphorylates the PDGFR␤ on  Fig. 1 were stably transfected with plasmids containing either no insert (Vector) or the bovine GRK2 cDNA, and clones were selected. A, immunoblot (IB) for protein expression. Fifty g of protein from the indicated MEF lines (5 ng of purified GRK2, Std) were subjected to sequential immunoblotting, with the indicated specific or nonspecific IgGs. Single blots are shown, representative of three performed for each of four vector and four GRK2 MEF clones. GRK5/6 and PLC␥-1 IgGs were murine; results with nonimmune rabbit IgG (not shown) were equivalent to nonimmune mouse. B, phosphoinositide hydrolysis. Four GRK2-null MEF clones (GRK2, None) and 4 GRK2expressing MEF clones (GRK2, Native), including those in A, were exposed (37°C for 15 min) to medium lacking (None, "basal") or containing 2 nM PDGF-BB or fluoroaluminate, which activates heterotrimeric G proteins (22). The results (means Ϯ S.E.) from cognate clones were averaged, and the results from at least two experiments with each clone, performed in triplicate, are displayed. *, p Ͻ 0.01 compared with GRK2 native cells. serine(s) when GRK2 is expressed at physiologic levels, we examined PDGF-induced PDGFR␤ seryl phosphorylation in GRK2-expressing and GRK2 0 MEFs (Fig. 4). Within 10 min of PDGFR␤ activation, seryl phosphorylation of the PDGFR␤ was 2.5 Ϯ 0.6-fold greater in MEFs expressing physiologic GRK2 levels than it was in GRK2 0 MEFs ( Fig. 4; p Ͻ 0.05). Thus, GRK2 serine-phosphorylated the PDGFR␤ in cells expressing physiologic levels of both proteins, in a manner that correlated temporally with GRK2-mediated PDGFR␤ desensitization. Moreover, the magnitude of PDGFR␤ seryl phosphorylation in GRK2-expresssing and -null fibroblasts suggested that GRK2 mediates the majority of agonist-dependent PDGFR␤ seryl phosphorylation in these cells.
GRK2-mediated PDGFR␤ Phosphorylation Attenuates NHERF/PDGFR␤ Association-Could GRK2 phosphorylate the PDGFR␤ on serine 1104 and thereby diminish the affinity of PDGFR␤/NHERF association? If so, such a mechanism could help to explain why GRK2-mediated phosphorylation reduces tyrosyl phosphorylation of the PDGFR␤ but not the EGFR (10). Although the PDGFR␤ is cross-linked by NHERF, the EGFR is not (15). To explore this possibility, we first determined that NHERF overexpression augmented PDGFR␤ tyrosyl phosphorylation in our transfected cell system (2.7 Ϯ 0.3-fold (p Ͻ 0.02); Fig. 5 (15), as GRK2-mediated PDGFR␤ phosphorylation does (Fig. 3B). To determine whether GRK2-mediated PDGFR␤ phosphorylation attenuated the association of Densitometry values for pAkt were divided by cognate values for actin; this ratio was normalized to that obtained for GRK2 native cells stimulated with PDGF. The means Ϯ S.E. of three experiments are depicted at right. *, p Ͻ 0.05 compared with GRK2 native cells. B, GRK2 diminishes PDGFR␤ activation. Quiescent MEFs were exposed to medium containing vehicle (Ϫ) or 2 nM PDGF-BB for 5 min (37°C) and then solubilized and subjected to PDGFR␤ IP. Immunoprecipitates were resolved by SDS-PAGE and immunoblotted sequentially for PDGFR␤ and then phosphotyrosine (pY). A single blot is shown and represents three performed with each of four MEF clones. PDGFR␤ phosphotyrosine band densities were divided by cognate band densities for the PDGFR␤, and these quotients were normalized to that obtained for PDGF-stimulated control (GRK2 native) cells. *, p Ͻ 0.05 compared with control cells.

FIG. 4. Physiologic levels of GRK2 effect serine phosphorylation of endogenous PDGFR␤s in fibroblasts.
The GRK2-expressing and GRK2-null MEF clones from Fig. 2 were exposed to vehicle (Ϫ) or 2 nM PDGF-BB for 10 min (37°C), then solubilized, and subjected to PDGFR␤ IP. Immunoprecipitates were immunoblotted serially for PDGFR␤ (bottom gel) and phosphoserine (pSer, top gel). Shown are immunoblots (IB) from a single experiment, representative of three performed in duplicate. The phosphoserine band densities were divided by cognate PDGFR␤ band densities, and these quotients were normalized to that obtained for GRK2-null cells. These normalized data are plotted (means Ϯ S.E.) in the bottom panel. *, p Ͻ 0.05 compared with results from GRK2-null cells. NHERF with the PDGFR␤, we co-immunoprecipitated NHERF with the PDGFR␤ from cells that either overexpressed GRK2 or expressed endogenous GRK2. With GRK2 overexpression, seryl phosphorylation of the PDGFR␤ increased by 2.1 Ϯ 0.6fold (Fig. 6, A and B; p Ͻ 0.05), much as we found by comparing MEFs expressing physiologic levels of GRK2 with GRK2 0 MEFs (Fig. 4). Correspondingly, there was a 60 Ϯ 30% reduction in the amount of NHERF associated with the PDGFR␤ (Fig. 6, C and D; p Ͻ 0.05). This correlation of reduced NHERF/ PDGFR␤ association with GRK2-mediated PDGFR␤ seryl phosphorylation suggested that serine 1104 of the PDGFR␤ was indeed a GRK2 phosphorylation site.
Identification of PDGFR␤ Ser 1104 as a GRK2 Phosphorylation Site-From the foregoing observations, we reasoned that GRK2 could phosphorylate PDGFR␤ Ser 1104 (of the Cterminal Asp-Ser 1104 -Phe-Leu) and thereby diminish both the affinity of NHERF/PDGFR␤ interaction and, consequently, the degree of PDGF-induced receptor activation. To test this possibility, we sought to determine whether phosphorylation of PDGFR␤ Ser 1104 could effect at least some of the changes in PDGFR␤ behavior we observed consequent to GRK2-mediated PDGFR␤ phosphorylation. To that end, we created S1104D and S1104A PDGFR␤ mutants. Although we expected the S1104D mutant to model the Ser 1104 -phosphorylated PDGFR␤, we expected that both the S1104A and S1104D mutants would demonstrate diminished binding affinity for NHERF, because each of the analogous mutations in the ␤ 2 -adrenergic receptor Cterminal tetrapeptide severely attenuated NHERF binding in blot overlay assays (26). To test these expectations, we first assessed the association of NHERF with the S1104D PDGFR␤. As with the GRK2-phosphorylated PDGFR␤, the S1104D PDGFR␤ demonstrated a 50 Ϯ 10% decrease in NHERF association, compared with the WT PDGFR␤ ( Fig. 7A; p Ͻ 0.05). Moreover, this reduction in NHERF association correlated with a 20 Ϯ 5% reduction in tyrosyl phosphorylation of the S1104D PDGFR␤ ( Fig. 7A; p Ͻ 0.05). In a similar fashion, the S1104A PDGFR␤ also demonstrated reduced tyrosyl phosphorylation (50 Ϯ 20%, p Ͻ 0.05), compared with the WT PDGFR␤ (Fig.  7B).
Like GRK2-mediated PDGFR␤ phosphorylation, mutation of PDGFR␤ Ser 1104 reduced PDGFR␤/NHERF association and PDGFR␤ activation. For this reason, we expected that Ser 1104 mutation would also reduce signaling downstream of the PDGFR␤, as GRK2-mediated PDGFR␤ desensitization does (Figs. [1][2][3]. Indeed, the S1104A and the S1104D PDGFR␤s each demonstrated impairment of PDGFinduced phosphoinositide hydrolysis in cells expressing endogenous NHERF (Fig. 8A), and the magnitude of impairment was similar to that observed with GRK2-phosphorylated WT PDGFR␤s (Fig. 2). Moreover, impairment of signal transduc-FIG. 6. GRK2 overexpression augments PDGFR␤ seryl phosphorylation and reduces PDGFR␤/NHERF association. HEK 293 cells were transfected (or not, Ϫ) with plasmids encoding the PDGFR␤, GRK2, or no protein (Vector, "control"), as indicated. Serum-starved cells were exposed to vehicle (Ϫ) or 2 nM PDGF-BB for 5 min (37°C) and then subjected to PDGFR␤ IP. A, immunoprecipitates were immunoblotted (IB) sequentially for PDGFR␤ and then phosphoserine (pSer). Shown are immunoblots from a single experiment, representative of three performed. B, band densities for pSer were divided by cognate band densities for the PDGFR␤, and these quotients were normalized to that obtained for unstimulated control cells. Plotted are the means Ϯ S.E. of results from three experiments. *, p Ͻ 0.05 compared with control cells. C, immunoprecipitates were immunoblotted sequentially for PDGFR␤ and then endogenous NHERF. The cell lysates were immunoblotted for GRK2 (20 g of protein from control cells and 2 g from GRK2-overexpressing cells) and endogenous NHERF (20 g of protein from all cells). Shown are immunoblots from a single experiment, representative of three performed. D, band densities for NHERF were divided by cognate band densities for the PDGFR␤, and these quotients were normalized to that obtained for unstimulated control cells, to obtain the percentage of control. Plotted are the means Ϯ S.E. of results from three experiments. *, p Ͻ 0.05 compared with control cells. tion by Ser 1104 mutation was signal pathway-specific and analogous to the pattern we observed previously with GRK2-mediated PDGFR␤ phosphorylation ( Fig. 3 and Ref. 14). Whereas Ser 1104 mutant PDGFR␤s failed to activate PLC␥-1 as much as the WT PDGFR␤ did (Fig. 8A), the S1104D PDGFR␤ activated ERK just as much as the WT PDGFR␤ (Fig. 8B). Thus, Ser 1104 mutant PDGFR␤s resemble GRK2-phosphorylated WT PDGFR␤s with regard to NHERF binding, PDGFR␤ activation, and downstream signaling.
If GRK2 does desensitize the PDGFR␤ by phosphorylating PDGFR␤ Ser 1104 , then we would expect the S1104A PDGFR␤ to resist desensitization by GRK2. To test this prediction, we examined phosphoinositide hydrolysis elicited by the WT and S1104A PDGFR␤s, in cells expressing either endogenous or high levels of GRK2. With the WT PDGFR␤, PDGF-induced phosphoinositide hydrolysis was reduced by 40% when GRK2 was overexpressed. Interestingly, this reduction in phosphoinositide hydrolysis was equivalent to that observed with the S1104A mutation of the PDGFR␤ (Fig. 8C). However, overexpression of GRK2 failed to desensitize S1104A PDGFR␤ signaling, even though the levels of GRK2 overexpression tested were equivalent to those achieved in cells expressing WT PDGFR␤s (Fig. 8C). These results further support a model in which GRK2 phosphorylates Ser 1104 of the PDGFR␤ and thereby engenders PDGFR␤ desensitization, at least in part by reducing PDGFR␤/NHERF association. A, phosphoinositide hydrolysis. HEK 293 cells transfected with the indicated PDGFR␤ construct were processed for phosphoinositide hydrolysis as in Fig.  2 (except no PDGFR␣ down-regulation was necessary). Total inositol phosphates produced in response to 2 nM PDGF-BB or fluoroaluminate were normalized to those measured in PDGF-challenged cells expressing WT PDGFR␤s ("control"). Plotted are the means Ϯ S.E. from three experiments performed in triplicate. Basal inositol phosphates were 1.3 Ϯ 0.5, 1.5 Ϯ 0.5, and 1.6 Ϯ 0.8 (percent conversion units) for WT, S1104A-, and S1104D-PDGFR␤ cells, respectively. Cell surface expression of the S1104D-and S1104A-PDGFR␤, respectively, were 120 Ϯ 30 and 110 Ϯ 25% of WT PDGFR␤ levels. *, p Ͻ 0.05 compared with WT. B, PDGFR␤ S1104D mutation fails to alter ERK signaling. The cells from A were exposed to medium containing either vehicle (Ϫ) or 2 nM PDGF-BB for 10 min (37°C) and then solubilized; lysate proteins resolved by SDS-PAGE were immunoblotted (IB) sequentially with IgG against phospho-ERK1/2 (pERK1/2) and endogenous NHERF or in parallel with nonimmune rabbit IgG (bottom gel). Shown are the results of a single experiment, representative of three performed. C, GRK2 fails to desensitize the S1104A PDGFR␤. HEK 293 cells were transfected with plasmids encoding either the WT or the S1104A PDGFR␤, and either GRK2 (GRK2 level, High) or no protein (GRK2 level, Native). Control cells were those transfected with the WT PDGFR␤ and empty vector plasmids. Left panel, cells were assayed for PDGF-induced phosphoinositide hydrolysis, as in A. The values for PDGF-induced total inositol phosphates (percent conversion units) were averaged across three experiments and plotted as the means Ϯ S.E. *, p Ͻ 0.05 compared with control cells. Basal values for inositol phosphates (percent conversion units) were 1.9 Ϯ 0.1, 1.7 Ϯ 0.1, 1.7 Ϯ 0.1, and 1.7 Ϯ 0.3 for cells transfected with WT PDGFR␤/vector, WT PDGFR␤/GRK2, S1104A PDGFR␤/vector, and S1104A PDGFR␤/GRK2, respectively. Right panel, 40 g of "native GRK2" cells and 2 g of "high GRK2" cells were immunoblotted for GRK2. The results are from a single experiment, representative of three performed.