Platelet-derived Growth Factor Stimulates Protein Kinase D through the Activation of Phospholipase Cg and Protein Kinase C*

Platelet-derived growth factor (PDGF) stimulates protein kinase D (PKD) in a timeand dose-dependent manner. We have used a series of PDGF receptor mutants that display a selective impairment of the binding of SH2-containing proteins (GTPase-activating protein, SHP-2, phospholipase Cg (PLCg), or phosphatidylinositol 3*-kinase (PI3K)) to show that Tyr-1021, the PLCgbinding site, is essential for PKD stimulation by PDGF in A431 cells. We next investigated whether any one of these four binding sites could mediate PKD activation in the absence of the other three sites. F5, a receptor mutant that lacks all four binding sites for GTPase-activating protein, PLCg, PI3K, and SHP-2, fails to activate PKD. A panel of single add-back mutants was used to investigate if any one of these four sites could restore signaling to PKD. Of the four sites, only the PLCg single add-back receptor restored PDGF-mediated activation of PKD, and only this add-back receptor produced diacylglycerol (DAG) in a PDGF-dependent manner. 1,2Dioctanoyl-sn-glycerol, a membrane-permeant DAG analog, was found to be sufficient for activation of PKD. Taken together, these data indicate that PLCg activation is not only necessary, but also sufficient to mediate PDGF-induced PKD activation. Although the presence of a pleckstrin homology domain makes PKD a potential PI3K target, PKD was not stimulated by selective PI3K activation, and wortmannin, an inhibitor of PI3K, did not inhibit PDGF signaling to PKD. The activation of PKD by DAG or by the wild-type and PLCg add-back PDGF receptors was inhibited by GF109203X, suggesting a role for protein kinase C in the stimulation of PKD by PDGF. PDGF induced a time-dependent phosphorylation of PKD that closely correlated with activation. The PDGF-induced activation and phosphorylation of PKD were reversed by in vitro incubation of PKD with protein phosphatase 1 or 2A, indicating that PDGF signaling to PKD involves the Ser/Thr phosphorylation of PKD. Taken together, these results conclusively show that PDGF activates PKD through a pathway that involves activation of PLCg and, subsequently, protein kinase C. The production of lipid second messengers is a common theme in the signal transduction of growth factors (1–5). An important task of current signal transduction research is to link these messengers to their targets or, vice versa, to find lipid messengers for proteins whose structure predicts potential lipid-binding sites. Recently, two protein kinases were cloned (PKD from mouse and its human homolog, PKCm) (6–9) that contain a kinase domain, a pleckstrin homology domain, a cysteine-rich zinc finger domain, and a putative transmembrane domain. We have demonstrated before that PKD is activated by diacylglycerol and by the tumor promotor PDB (8). The cysteine-rich zinc fingers of the classical and novel PKC isoforms have been shown to bind DAG and phorbol esters (10). However, zinc fingers are also important for protein-protein interactions of PKCl and PKCi with stimulatory proteins (11, 12), and it is noteworthy that the presence of a zinc finger in a kinase is not predictive for the DAG/PDB stimulation of its kinase activity. The c-Raf zinc finger mediates the interaction with phosphatidylserine-containing micelles and 14-3-3 proteins and is required for optimal binding to Ras GTP, but it cannot mediate PDB stimulation of Raf kinase activity (13, 14). Some lipid-stimulated kinases can be activated by multiple lipids: PKCd and PKCe are activated not only by DAG, but also by phosphatidylinositol 3,4,5-trisphosphate (10, 15). Akt/PKB is stimulated by the lipid PtdIns(3,4)P2, which involves binding to the PH domain (16). Therefore, the presence of a PH domain and zinc fingers in the N-terminal region of PKD would suggest several possibilities for regulation of its kinase activity through interaction with lipids or proteins. Given the large range of signaling mechanisms that can possibly impinge upon the different domains of PKD (based on the above-mentioned analogies), we decided to investigate which of several growth factor signaling pathways can induce activation of PKD. We have chosen the b-platelet-derived growth factor receptor (b-PDGFR) as a paradigm for our studies. Binding of PDGF induces dimerization and autophosphorylation of the b-PDGFR at specific tyrosine residues. Through these specific phosphotyrosine motifs, the phosphorylated PDGF receptor binds a large variety of SH2 proteins (for a review, see Refs. 18 and 19). The p85 subunit of phosphatidylinositol 39-kinase (PI3K) binds to tyrosines 740 and 751; the phosphotyrosine phosphatase SHP-2 associates with tyrosine 1009; phospholipase Cg1 binds * This work was supported by grants from the European Community (Inco-Copernicus), the Fonds voor Wetenschappelijk Onderzoek-Vlaanderen (Actie Levenslijn), and the Flemish Government (Geconcerteerde Onderzoeksacties). 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 U.S.C. Section 1734 solely to indicate this fact. ‡ Postdoctoral Research Fellow of the Fund for Scientific Research (Fonds voor Wetenschappelijk Onderzoek-Vlaanderen). To whom correspondence should be addressed. Tel.: 32-16-345719; Fax: 32-16345995; E-mail: Johan.Vanlint@MED.KULeuven.ac.be. § Present address: Lithuanian Academy of Sciences, Inst. of Biochemistry, 2600 Vilnius, Lithuania. ¶ Research Director of the Fonds voor Wetenschappelijk OnderzoekVlaanderen. 1 The abbreviations used are: PKD, protein kinase D; PKC, protein kinase C; PDB, phorbol 12,13-dibutyrate; DAG, diacylglycerol; PtdIns, phosphatidylinositol; PH domain, pleckstrin homology domain; b-PDGFR, b-platelet-derived growth factor receptor; PDGF, plateletderived growth factor; PI3K, phosphatidylinositol 39-kinase; GAP, Ras GTPase-activating protein; PLCg, phospholipase Cg; DMEM, Dulbecco’s modified Eagle’s medium; FBS, fetal bovine serum; diC8, 1,2dioctanoyl-sn-glycerol; PP1C/PP2AC, catalytic subunits of the type 1/type 2A protein phosphatases. THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 273, No. 12, Issue of March 20, pp. 7038–7043, 1998 © 1998 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

The production of lipid second messengers is a common theme in the signal transduction of growth factors (1)(2)(3)(4)(5). An important task of current signal transduction research is to link these messengers to their targets or, vice versa, to find lipid messengers for proteins whose structure predicts potential lipid-binding sites.
Recently, two protein kinases were cloned (PKD 1 from mouse and its human homolog, PKC) (6 -9) that contain a kinase domain, a pleckstrin homology domain, a cysteine-rich zinc finger domain, and a putative transmembrane domain. We have demonstrated before that PKD is activated by diacylglycerol and by the tumor promotor PDB (8).
The cysteine-rich zinc fingers of the classical and novel PKC isoforms have been shown to bind DAG and phorbol esters (10). However, zinc fingers are also important for protein-protein interactions of PKC and PKC with stimulatory proteins (11,12), and it is noteworthy that the presence of a zinc finger in a kinase is not predictive for the DAG/PDB stimulation of its kinase activity. The c-Raf zinc finger mediates the interaction with phosphatidylserine-containing micelles and 14-3-3 proteins and is required for optimal binding to Ras GTP, but it cannot mediate PDB stimulation of Raf kinase activity (13,14). Some lipid-stimulated kinases can be activated by multiple lipids: PKC␦ and PKC⑀ are activated not only by DAG, but also by phosphatidylinositol 3,4,5-trisphosphate (10,15). Akt/PKB is stimulated by the lipid PtdIns(3,4)P 2 , which involves binding to the PH domain (16). Therefore, the presence of a PH domain and zinc fingers in the N-terminal region of PKD would suggest several possibilities for regulation of its kinase activity through interaction with lipids or proteins. Given the large range of signaling mechanisms that can possibly impinge upon the different domains of PKD (based on the above-mentioned analogies), we decided to investigate which of several growth factor signaling pathways can induce activation of PKD.
We have chosen the ␤-platelet-derived growth factor receptor (␤-PDGFR) as a paradigm for our studies. Binding of PDGF induces dimerization and autophosphorylation of the ␤-PDGFR at specific tyrosine residues. Through these specific phosphotyrosine motifs, the phosphorylated PDGF receptor binds a large variety of SH2 proteins (for a review, see Refs. 18 and 19). The p85 subunit of phosphatidylinositol 3Ј-kinase (PI3K) binds to tyrosines 740 and 751; the phosphotyrosine phosphatase SHP-2 associates with tyrosine 1009; phospholipase C␥1 binds * This work was supported by grants from the European Community (Inco-Copernicus), the Fonds voor Wetenschappelijk Onderzoek-Vlaanderen (Actie Levenslijn), and the Flemish Government (Geconcerteerde Onderzoeksacties). 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  to tyrosine 1021; and the Ras GTPase-activating protein (GAP) associates with tyrosine 771. Furthermore, three members of the Src kinase family (Src, Yes, and Fyn), Nck, Shc, and several as yet unidentified proteins are known to associate with the ␤-PDGFR (18,19). Using a panel of ␤-PDGFR mutants that are defective in the binding of certain SH2 domain-containing proteins, it is possible to selectively knock out or turn on specific signaling pathways so that the functional role of a particular pathway to downstream responses can be elucidated (20,24,25). Because of the large range of signaling proteins that can bind to the ␤-PDGFR, this ␤-PDGFR mutant system is well suited to investigate the variety of pathways that may activate newly identified components of the cellular signaling apparatus (e.g. PKD). This report shows that in A431 cells, PDGF activates PKD through the subsequential activation of PLC␥ and PKC.

EXPERIMENTAL PROCEDURES
Materials-PDGF was purchased from Upstate Biotechnology, Inc. GF109203X was obtained from Calbiochem, and G418, Lipofectin, Glutamax, and Opti-MEM from were from Life Technologies, Inc. The Biotrak DAG detection kit was from Amersham Corp. Antihemagglutinin antibodies were from Boehringer Mannheim. Protein A-TSK gel was from Affiland (Sart-Tilman, Belgium). All other materials were from Sigma.
The A431 cell line (ATCC CRL 1555), devoid of endogenous ␤-PDG-FRs, was used to create cell lines that express different ␤-PDGFR mutants. For this purpose, a retroviral infection system was used as described by Miller et al. (23). Retroviruses carrying mutant ␤-PDGFR genes were generated as follows. Plasmids containing ␤-PDGFR mutants (in the pLXSN vector, carrying a neomycin resistance gene) were introduced using Lipofectin into an NIH3T3 packaging cell line (⌿2). After 48 h, a second NIH3T3 packaging cell line (PA317) was infected with the ecotropic virus produced by the ⌿2 cells. The PA317 cells were cultured in the presence of 1 mg/ml G418 from 24 h post-infection. Drug-resistant cells were pooled and propagated for several passages. Amphotropic virus was collected from subconfluent cultures of PA317 cells, and proliferating A431 cells on a 10-cm dish were infected with 5 ml of a viral supernatant supplemented with 4 g/ml Polybrene. After 5 h, 5 ml of DMEM containing 10% FBS and 4 g/ml Polybrene was added, and cells were allowed to grow for another 24 h. After 24 h, the viral supernatant was replaced by DMEM containing 10% FBS, and cells were allowed to grow for another 24 h. Infected cells were then selected for a 2-week period in 1 mg/ml G418. Clones were propagated and screened for ␤-PDGFR expression.
The nature of the different ␤-PDGFR mutants has been extensively described and characterized previously (20,24,25). Briefly, WT is an A431 cell line expressing the wild-type PDGF receptor. pLXSN has been infected with the empty retroviral vector pLXSN. F5 expresses a ␤-PDGFR mutant in which all the tyrosines that bind the SH2 domains of GAP, SHP-2, PLC␥, and PI3K have been mutated to phenylalanine. Starting with the F5 mutant, several binding sites were selectively restored (add-back mutants): GAP ϩ for binding of GAP (Tyr-771), PLC␥ ϩ for binding of PLC␥ (Tyr-1021), PI3K ϩ for binding of PI3K (Tyr-740 and Tyr-751), and SHP-2 ϩ for binding of SHP-2 (Tyr-1009). Another series of mutants (single minus mutants) harbors mutations of one specific binding site in the ␤-PDGFR (the sites mentioned above) with all others left intact: GAP Ϫ , PLC␥ Ϫ , PI3K Ϫ , and SHP-2 Ϫ do not bind GAP, PLC␥, PI3K, and SHP-2, respectively.
For assays involving PKB measurements, A431 cells were transfected with hemagglutinin-tagged PKB-pcDNA3 using Lipofectin as a transfection agent according to the instructions of the manufacturer. Briefly, 6 g of DNA and 12 l of Lipofectin were each diluted in 1 ml of Opti-MEM. After 45 min, DNA and Lipofectin were mixed and incubated for 15 min. A431 cells were washed once with Opti-MEM and then incubated with the DNA/Lipofectin mixture in 5 ml of Opti-MEM for 5 h. The medium was then changed to DMEM supplemented with 10% FBS, and cells were further grown for 72 h and then incubated in serum-free DMEM overnight.
Immunoprecipitations and Kinase Assays-Lysates were incubated for 2 h with an antibody against the C-terminal 15 amino acids of PKD or with hemagglutinin antibodies (for hemagglutinin-tagged PKB). Immunocomplexes were captured with 15 l of protein A-TSK gel for 1 h.
PKD immunoprecipitates were washed twice with lysis buffer A, and PKD was eluted by incubating the immunoprecipitates with lysis buffer containing a 0.5 mg/ml concentration of the immunizing peptide. 15 l of PKD eluate was incubated for 5 min at 30°C with 25 l of a kinase assay mixture, resulting in a final concentration of 20 mM Tris (pH 7.4), 100 M ATP (specific activity of 1000 cpm/pmol), 10 mM MgCl 2 , and 1 mg/ml syntide-2.
Akt/PKB immunoprecipitates were washed once with lysis buffer, once with lysis buffer containing 0.5 M LiCl 2 , and twice with kinase assay buffer. The PKB immunoprecipitation pellet was incubated for 15 min with the same phosphorylation mixture as described above, except that syntide-2 was replaced by the RGRPRTTSFAE peptide corresponding to the site in glycogen synthase kinase 3-␤ that is phosphorylated by Akt/PKB in vivo (26).
All reactions were terminated by spotting 30 l of the reaction mixture on P-81 phosphocellulose paper washed in 75 mM phosphoric acid. The radioactivity incorporated in the respective peptides was measured by liquid scintillation counting.
DAG Production Assay-DAG production was measured using the Biotrak DAG detection kit, which uses [ 32 P]phosphatidic acid yield by DAG kinase as a measure of DAG production, according to the instructions of the manufacturer. Briefly, lipids were extracted according to the method of Bligh and Dyer (27) and incubated in a DAG kinase reaction mixture containing 0.05 M imidazole (pH 6.6), 0.05 M NaCl, 12 mM MgCl 2 , 1 mM EGTA, and 500 M ATP (specific activity of 50 cpm/pmol). 32 P i Labeling of Cells and Analysis of PKD Phosphorylation-Confluent cultures of A431 cells were washed twice with DMEM (phosphate-free) and incubated in this medium containing 500 Ci/ml carrier-free 32 P i overnight (12 h). Cells were then stimulated for the indicated times with PDGF (30 ng/ml) and lysed in buffer A. Lysates were subsequently immunoprecipitated with anti-PKD antibody and analyzed by SDS-polyacrylamide gel electrophoresis followed by autoradiography.
Protein Phosphatase Incubations-PKD eluates were incubated for 30 min at 30°C with 50 units/ml PP1 C or PP2A C in the presence or absence of 1 M microcystin. After this incubation, a PKD kinase assay was performed as indicated above in the presence of 1 M microcystin. For visualization of the dephosphorylation of PKD by serine/threonine protein phosphatases, PKD was immunoprecipitated from lysates of 32 P i -labeled cells that were stimulated with PDGF. Immunoprecipitated PKD was then eluted from the immunocomplexes and incubated for 30 min at 30°C with 50 units/ml PP1 C or PP2A C in the presence or absence of 1 M microcystin.

RESULTS AND DISCUSSION
This work is the first report of a dissection of specific growth factor signaling pathways that activate PKD. PDGF stimulates PKD in a time-and dose-dependent manner, both in Swiss 3T3 cells expressing endogenous PDGF receptor and in A431 cells stably overexpressing a retrovirally introduced PDGF receptor. PKD is stimulated by PDGF doses as low as 5 ng/ml, with a maximum at 30 ng/ml (Fig. 1, A and B), which correlates well with the concentration of PDGF required for a variety of cellular responses such as PLC␥, PI3K, and GAP tyrosine phosphorylation (28). PKD activity reached a maximum after 10 min, but remained elevated even at 90 min after addition of PDGF (Fig. 1, C and D). Equal amounts of PKD were present in immunoprecipitates from Swiss 3T3 cells or A431 cells stimulated for various times with PDGF, as evidenced by Western blotting followed by immunostaining with anti-PKD antibodies (Fig. 1, E and F).
Activation of PLC␥ by the ␤-PDGFR Is Necessary and Sufficient to Activate PKD-Selective mutation of the PLC␥-binding site of the ␤-PDGFR into phenylalanine completely abolished the ␤-PDGFR-induced activation of PKD. Full activation was retained when binding sites for SHP-2, GAP, and PI3K were selectively mutated into phenylalanine. These results suggest that the binding of PLC␥ to the ␤-PDGFR is essential for activation of PKD (Table I, Minus mutants).
We used the single add-back mutants to explore specific pathways that can activate PKD. The F5 mutant, which lacks binding sites for SHP-2, GAP, PI3K, and PLC␥, fails to activate PKD. The PLC␥ ϩ mutant, which binds PLC␥ but not PI3K, GAP, or SHP-2, is the only add-back receptor mutant that can mediate full PKD activation by PDGF (Table I, Add-back mutants). These results strongly suggest that PLC␥ activation by PDGF is both necessary and sufficient for activation of PKD. To further strengthen this hypothesis, we investigated whether DAG production was sufficient to cause full activation of PKD. If PLC␥ activation is sufficient to activate PKD, then addition of DAG, the reaction product of its catalytic activity, should have an identical effect. Indeed, as shown in Fig. 2, the addition of the membrane-permeant DAG analog diC8 caused a full activation of PKD. Moreover, of all add-back mutants, only the PLC␥ ϩ add-back mutant was able to mediate PDGF-induced DAG production (Table I, Add-back mutants, DAG production).
The activation of PLC␥ represents a very important branch of the ␤-PDGFR signal transduction mechanism. It has previously been shown that PLC␥ and PI3K are two independent mediators of PDGF-stimulated DNA synthesis (20) and that PLC␥ is essential for growth and development (21). PDGFinduced activation of phospholipase D (29), translocation of myristoylated alanine-rich C kinase substrate to the membrane (30), and activation of the Na ϩ /H ϩ exchanger (31) all occur through a PLC␥-dependent pathway. Hence, it remains an important task to identify the key enzymes that act in pathways downstream of PLC␥. In this report, we clearly identify PKD as an enzyme acting downstream of PLC␥ in PDGF signaling.
The PI3K Pathway Does Not Signal to PKD-The multidomain structure of PKD prompted us to thoroughly investigate a variety of pathways that may impinge on this enzyme. The presence of a pleckstrin homology domain in PKD may represent a target for modulating enzymatic activity. It has been shown that the PH domain of Akt/PKB is crucial for the PI3Kmediated activation of the enzyme (16,32). Therefore, we investigated whether PKD could be activated by selective PI3K activation (Table II). The PI3K ϩ add-back receptor, which activates PI3K without PLC␥ activation, failed to activate PKD, whereas it activated Akt/PKB. Moreover, the activation of PKD by PDGF is not inhibited by wortmannin, a known inhibitor of PI3K (33). Taken together, these data clearly show that PKD is not a target for PI3K signaling. This in vivo approach is particularly useful since in vitro experiments have demonstrated several problems intrinsic to the presence of inositol lipids in kinase assays (see Ref. 34 for a thorough discussion). In vitro assays often show an equal extent of activation by PtdIns(3,4,5)P 3 or PtdIns(4,5)P 2 , so it is difficult to assess what would be the result of an in vivo increase in PtdIns(3,4,5)P 3 on a particular kinase (in the presence of a high background level of PtdIns(4,5)P 2 in the membrane) (34). Moreover, the in vitro kinase activation by these lipids is inhibited by 10 mM MgCl 2 and is mimicked by high concentrations of inositol hexaphosphate and inositol hexasulfate (34). Therefore, in vitro assays of kinase activation through these lipids seem to be misleading due to the ionic charge effects that these compounds may have. In vitro effects of phosphatidylinositol lipids are likely to be valid only when they are corroborated by in vivo data (combined analysis of wortmannin inhibition and effects of PI3K mutants), as is the case for Akt/PKB (22,32).
Involvement of PKC in PKD Activation-We next investigated whether in vivo PKD activation requires the catalytic activity of PKC. Preincubation of A431 cells with GF109203X, a very potent inhibitor of PKC (35,36) but not of PKD (37), completely abolished activation of PKD by PDGF in the WT and the PLC␥ ϩ mutant cell lines (Table III). The activation of PKD by PDB or the DAG analog diC8 was also completely abolished by preincubation with GF109203X (Table III). These results indicate that PDGF causes activation of PKD through the activation of PKC.
The presence of zinc fingers in the N-terminal region of PKD (highly homologous to the zinc fingers of PKC) and the fact that PKD can be stimulated in vitro by phosphatidylserine/diC8 (8) would suggest that a direct interaction of these lipid molecules with the zinc fingers of PKD is sufficient for in vivo stimulation of the enzyme by the ␤-PDGFR. However, we have now clearly demonstrated that in vivo, the activation of PKD by PDGF or diC8 requires activation of PKC (Table III). The precise reason for this discrepancy still remains to be elucidated, but it could be that prior phosphorylation of PKD by a PKC-dependent mechanism is required in vivo to bring PKD in contact with DAG in the plasma membrane of the cell. This phosphorylation could induce a conformational change in the enzyme, cause its translocation, or promote the dissociation of a binding inhibitor. Hence, the enzyme may need a phosphorylation-induced translocation step to become activated. A similar type of regulation exists for ␤-adrenergic receptor kinase 1, which is also a PH domain-containing protein kinase. After phorbol ester stimulation, ␤-adrenergic receptor kinase 1 is phosphorylated by PKC and translocates from the cytosol to the membrane (39). Furthermore, it has been demonstrated that interaction of the ␤-adrenergic receptor kinase 1 PH domain with membrane lipids and G protein ␤␥-subunits coordinately stimulates ␤-adrenergic receptor kinase 1 activity (17). Hence, one could envision a tight regulation of PKD whereby, under resting conditions, PKD is sequestered in a pool away from the plasma membrane. When PLC␥ is activated, PKC is stimulated and promotes the phosphorylation of PKD (directly or through an intermediary kinase). This phosphorylation could then allow a PH domain-mediated interaction of PKD with the plasma membrane and the subsequent activation of the enzyme. Further work will be needed to test this hypothesis. Several PDGFinduced responses have been reported to be mediated by PKC: expression of immediate-early genes such as c-myc, c-fos, egr-1, junB, and fra-1 (38, 40); ␣ 2 -integrin expression (41); activation of the Na ϩ /H ϩ exchanger (31); and mitogenic signaling in hu-

FIG. 2. Activation of PKD by diC8, a diacylglycerol analog.
A-431 cells expressing wild-type PDGF receptors were incubated for 10 min with increasing concentrations of diC8. Cells were then lysed, and lysates were immunoprecipitated with anti-PKD antiserum. PKD was eluted from the immunoprecipitates with the immunizing peptide, and the eluted PKD activity was measured by a syntide-2 kinase assay as described under "Experimental Procedures." The results shown are representative of three independent experiments.

TABLE I
Activation of PKD kinase activity and DAG production by ␤-PDGFR mutants Cells were incubated for 10 min with 30 ng/ml PDGF or control buffer. For kinase assays, cells were lysed, and the lysates were immunoprecipitated with anti-PKD antiserum or control antiserum. PKD activity was then assayed as described under "Experimental Procedures." For measurement of DAG production, lipids were extracted from the cells, and DAG was measured using a Biotrak DAG kit. Activities are represented as percent maximum response and are representative for three independent experiments.

TABLE II
Activation of Akt/PKB (but not PKD) by PI3K Cells (expressing wild-type, PI3K add-back, or PLC␥ ϩ add-back PDGF receptors) were preincubated for 20 min with 200 nM wortmannin or solvent and subsequently incubated for 10 min with 30 ng/ml PDGF or control buffer. Cells were then lysed, and the lysates were immunoprecipitated with anti-PKD antiserum or anti-hemagglutinin serum (for immunoprecipitation of hemagglutinin-tagged Akt/PKB). PKD and Akt/PKB activities were then assayed as described under "Experimental Procedures." Activities are represented as percent maximum response and are representative for three independent experiments. PKD  man mesangial cells (42). Our experiments have shown that PKD acts downstream of PKC in ␤-PDGFR signaling. Therefore, it is tempting to speculate that some of the so-called PKC-mediated PDGF responses may be mediated ultimately through PKD. PDGF-induced PKD Activation Involves the Ser/Thr Phosphorylation of PKD-PDGF induces phosphorylation of PKD, as evidenced by the incorporation of 32 P in PKD immunoprecipitated from 32 P i -labeled cells that were stimulated for various times with PDGF (Fig. 3B). When comparing Fig. 1D and Fig. 3B, it is clear that the time course of incorporation of phosphate in PKD closely parallels the time course of activation of the enzyme. To show that PDGF-induced phosphorylation of PKD is required for activation of PKD, we incubated the activated PKD with the Ser/Thr-specific protein phosphatase PP1 C or PP2A C . Each phosphatase was able to fully reverse the PKD activation, and this inactivation was blocked by the specific phosphatase inhibitor microcystin (Fig. 3A). To further demonstrate the effect of each of these phosphatases on the phosphorylation status of PKD, we incubated the 32 P i -labeled PKD from PDGF-stimulated cells with PP1 C and PP2A C . As shown in Fig. 3A (insets), both Ser/Thr-specific phosphatases caused dephosphorylation of PKD. These data strongly suggest that a Ser/Thr phosphorylation event is involved in the activation of PKD by PDGF. Similar effects of phosphatases have been reported for other kinases such as extracellular signalrelated kinase and PKB, which are also stimulated in kinase cascades (43,44). The existence of multiple levels of control in kinase activation mechanisms is not without precedence. Akt/ PKB, another PH domain-containing kinase, is activated by both protein phosphorylation (44,45) and inositol lipid binding (16,32). In this respect, a particularly interesting similarity between PKD and Akt/PKB regulation emerges. Both enzymes can be directly stimulated in vitro by a lipid mediator (DAG and PtdIns(3,4,5)P 3 , respectively), and both enzymes can be stimulated in vivo by another upstream kinase (PKC and 3-phosphoinositidedependent protein kinase 1, respectively) (16,32,(43)(44)(45). Hence, the dual regulation of protein kinases by lipid ligands and protein phosphorylation emerges as a new regulatory theme in signal transduction.
This report presents a signaling model for future in vivo studies on PKD regulation. Further research will be required to elucidate the precise mechanistic details of the activation of PKD by PDGF. In particular, the contribution of each of the domains of PKD to the mechanism of PKD activation and function will have to be investigated using PKD mutants.  3. PDGF induces phosphorylation of PKD, which is reversed by PP1 C and PP2A C . A, activated PKD was incubated for 30 min at 30°C in the absence (control (C)) or presence of protein phosphatase (PP1 C or PP2A C ) with (ϩ, solid bars) or without (Ϫ, hatched bars) 1 M microcystin. Subsequently, PKD activity was measured by a syntide-2 kinase assay as described under "Experimental Procedures." The results shown are representative of three independent experiments. Insets, 32 P i -labeled PKD, immunoprecipitated from PDGF-stimulated 32 P i -loaded cells, was incubated for 30 min at 30°C in the absence (first inset) or presence of PP1 C (second inset) or PP2A C (third inset) with (right lanes) or without (left lanes) 1 M microcystin. PKD phosphorylation was then analyzed by SDS-polyacrylamide gel electrophoresis and autoradiography. The results shown are representative of three independent experiments. B, PDGF induces a time-dependent phosphorylation of PKD. Serum-starved 32 P i -loaded A-431 cells expressing wild-type PDGF receptors were incubated for the indicated times (minutes) with 30 ng/ml PDGF. Cells were then lysed, and lysates were immunoprecipitated with anti-PKD antiserum. PKD phosphorylation was analyzed by SDS-polyacrylamide gel electrophoresis and autoradiography. The results shown are representative of three independent experiments.