Protein Kinase C (cid:1) Regulation of the JNK Pathway Is Triggered via Phosphoinositide-dependent Kinase 1 and Protein Kinase C (cid:2) *

The protein kinase C (PKC)-related enzyme PKC (cid:1) / PKD (protein kinase D) is activated by activation loop phosphorylation through PKC (cid:3) . Here we demonstrate that PKC (cid:1) is activated by the direct phosphorylation of PKC (cid:2) . PKC (cid:1) colocalizes with PKC (cid:2) in HEK293 and MCF7 cells as shown by confocal immunofluorescence analyses. PDK1, known as the upstream kinase for several PKC isozymes, associates intracellularly with PKC (cid:2) and PKC (cid:3) . PKC (cid:3) is phosphorylated by PDK1 in vitro , leading to kinase activation as similarly reported for PKC (cid:2) activation by PDK1. Coexpression of PDK1, PKC (cid:2) and PKC (cid:1) in HEK293 cells results in PKC (cid:1) activation. In contrast, the coexpression of PDK1 and PKC (cid:3) with PKC (cid:1) does not activate PKC (cid:3) or consequently PKC (cid:1) . PDK1/PKC (cid:2) -triggered activation of PKC (cid:1) inhibits JNK, a downstream effector of PKC (cid:1)

The activation of protein kinases through growth factor receptors is achieved via a complex network of intracellular signal processes involving second messengers, protein-protein interaction, and the phosphorylation/dephosphorylation of interacting proteins. Protein kinase C (PKC) 1 family has been shown to be involved in the signal transduction of a wide range of biological responses, triggering changes in cell morphology, proliferation, and differentiation (1,2). The PKCs comprise a family of intracellular serine/threonine-specific kinases that are, depending on the isoform typically activated by Ca 2ϩ , lipid second messengers and/or protein activators (1,3). Recently, a more detailed activation model involving PDK1 as a PKC activation loop kinase preceding lipid activation has been established (4).
A group of kinases with PKC-like functional structures has been described. This kinase family consists of PKC (5), its mouse homologue termed PKD (6), PKC (7), and PKD2 (8). These kinases share structural homology to PKCs with respect to the catalytic domain and to the presence of amino-terminal cysteine fingers, defining the structural basis for lipid binding. PKC/PKD differs from PKC isozymes by an acidic domain (9), a PH domain within the regulatory region (10) and the lack of a typical pseudo-substrate site.
PKC/PKD activation occurs through several mediators via a PKC-dependent pathway (11). PKC and PKC⑀ have been implicated in the activation of PKC/PKD through binding to the PKC PH domain (12). Both enzymes associate and, in the case of PKC, directly phosphorylate the activation loop of PKC (13). Although a precise positioning of the PKC/PKC-PKD module in signal transduction pathways is currently unknown, it seems to be part of a kinase cascade that is triggered by multiple cell surface receptors. In the case of PDGF-initiated signal processes, PKC/PKD has been placed downstream of phospholipase ␥ (14). This predicts that phospholipase ␥-derived diacylglycerol generation results in PKC and consequently PKD/PKC activation, triggering p42 MAPK-modulated gene expression (15).
The discovery that PKC activation depends upon an ordered series of phosphorylation events before lipids activate has led to a search for an upstream kinase identifying PDK1 phosphorylating the PKC activation loop (4,16). PDK1 is a constitutive active and mainly cytosolic localized kinase that translocates upon growth factor-induced synthesis of 3Ј-phosphorylated phosphoinositides via its PH domain to the cytoplasm membrane in which it phosphorylates downstream kinases such as Akt (17). The discovery that PKCs are activated through PDK1 and that PKC activates PKC/PKD implicates that PDK1 is involved in PKC activation. We have carried out this study to analyze in detail the activation of PKC by upstream kinases and to identify potential downstream pathways. Here we demonstrate that PKC⑀ colocalizes with, phosphorylates, and activates PKC. PDK1 activates PKC and PKC⑀ in vitro, but upon transient expression in HEK293 cells, PDK1 acts only via PKC⑀ and not via PKC to activate PKC, leading to an inhibition of JNK activity.

MATERIALS AND METHODS
Cloning of Expression Constructs-The PDK1 expression constructs and the recombinant enzyme were gifts of Alex Toker (Harvard Medical School, Boston, MA). The construction of the PKC and PKC-GFP expression vectors have been described previously (13,18). PKC-GFP migrates in SDS-PAGE at ϳ110 kDa, and PKC K/R and PKC A/E migrate at 80 kDa. The PKC⑀ expression vector was a gift of Angelika Hausser (University of Stuttgart, Stuttgart, Germany). The construction of the PKC-GFP expression construct has been described previ-ously (19). Recombinant expression and purification of PKC and Histagged PKC K612W either in Sf158 or Sf9 cells were described previously (13,20).
Antibodies and Reagents-Rabbit antibodies (Santa Cruz Biotechnology) were used for PKC (D-20), PKC⑀ (C-15), and PKC (C-15) detection. The monoclonal 9E10 anti c-myc antibody was used for the detection of c-myc-tagged proteins (Roche Diagnostics). Western blot detection was carried out using an alkaline phosphatase-based detection system (Dianova). The cell lines used (HEK293, MCF7) were obtained from ATCC. SF9-produced affinity-purified PKCs were obtained from Calbiochem.
Transfection of HEK293 and MCF7 Cells-600,000 HEK293 cells were seeded in 6-cm plates 24 h before transfection. Cells were transfected using 5 g of DNA with 30 l of Superfect (Qiagen) and incubated for 2 h. The transfection mixtures were replaced by fresh medium, and cells were lysed 48 h after transfection (see below). For MCF7 cells, the FuGENE 6 transfection reagent (Roche Diagnostics) was used according to the manufacturer's instructions.
Immunofluorescence Analysis-PKC-GFP, PKC⑀, and PDK1 were transfected in HEK293 and MCF7 cells. 24 h after transfection, cells were washed twice with PBS and fixed in 4% paraformaldehyde (in PBS) for 15 min at 37°C. Permeabilization and blocking of the cells were performed by incubation with 0.05% Tween 20 and 5% fetal bovine serum in PBS for 30 min. The cells were rinsed three times with PBS and then incubated with the primary antibody against PKC⑀, PDK1, and PKC (0.2 g/ml) for 2 h (0.05% Tween 20 and 5% fetal bovine serum in PBS). For immunofluorescence detection of PKC, cells were incubated with a secondary anti-rabbit antibody (Alexa-46-conjugated, 1:1000, MobiTec) for 2 h. PDK1 was stained with an anti-c-myc monoclonal antibody and visualized with Cy3-labeled anti-mouse antibody. After staining, the cells were rinsed four times with PBS and prepared for microscopic analysis. Images were acquired using a confocal laser scanning microscope (TCS SP2, Leica) equipped with a ϫ100/1.4 HCX PlanAPO oil-immersion objective. GFP was excited with an argon laser (488-nm line), whereas Alexa-546 was excited using a helium-neon laser (543-nm line). Each image represents a two dimensional parallel projection of sections in the z-series taken at 0.8 -0.9-m intervals across the cell.
Immunoprecipitation Assays and in Vitro Kinase Assays-Cells were lysed at 4°C in lysis buffer (20 mM Tris-HCl, pH 7.4, 1% Nonidet P-40, 150 mM NaCl, 5 mM MgCl 2 , 1 g/ml leupeptin, 1 g/ml aprotinin, 1 mM phenylmethanesulfonyl fluoride, 2 mM sodium orthovanadate, 1 mM sodium fluoride). After 30-min incubation on ice, the lysates were centrifuged at 10 000 ϫ g for 15 min at 4°C, and immunoprecipitation was performed as described previously (21). 1 ml of lysate was incubated for 1.5 h at 4°C with 5 l of the antibody. Immunocomplexes were harvested by adding 30 l of protein G-Sepharose, incubated for 1 h, and washed twice with lysis buffer and one time with PBS. The samples were fractionated by SDS-PAGE followed by transfer to a nitrocellulose membrane. For in vitro kinase assays, the immunoprecipitates were washed additionally in 1 ml of phosphorylation buffer (50 mM Tris, pH 7.5, 10 mM MgCl 2 , 2 mM dithiothreitol). Kinase reaction with or without 2.5 g of syntide 2 as a substrate was started by adding 4 Ci of [␥-32 P]ATP (Amersham Biosciences) in 20 l of kinase buffer and incubated for 15 min at 37°C. For in vitro phosphorylation, reactions 450 units of recombinant protein PKCs were used in a volume of 40 l of phosphorylation buffer. Enzyme assays were carried out in the presence of lipid micelles as described previously (20). In vitro activation of autophosphorylation and substrate phosphorylation of 10 g of myelin basic protein (MBP, Sigma) and 2.5 g of syntide 2 (Sigma) by recombinant PKC in the presence or absence of lipid micelles was tested (data not shown). JNK assays were performed as described previously (22). Equal amounts of glutathione S-transferase-fusion protein used were verified by Ponceau staining. Phosphorylation reactions were stopped by adding 5ϫ concentrated sample buffer subsequently fractionated on a SDS-PAGE, transferred to a nitrocellulose membrane, and exposed to autoradiography. Syntide 2 phosphorylation was monitored by autoradiography after fractionation on 20% SDS-PAGE followed by drying of the gel. Autoradiographs were analyzed by using phosphorimaging analysis (Amersham Biosciences). In vitro kinase experiments and coimmunoprecipitations were performed at least three times. Shown are representative autoradiographs and immunoblots.

RESULTS
PKC⑀ Is a Direct Upstream Kinase for PKC-We have recently shown that PKC is phosphorylated and activated by PKC (13). These findings together with independent studies showing association of PKD with PKC and to lesser degree with PKC⑀ (12) allow us to postulate that PKC⑀ might also act as a direct upstream kinase for PKD/PKC. Earlier studies were in favor of this hypothesis, showing that PKC⑀ similar to PKC is located at the Golgi compartment (23,24). Because PKC has been demonstrated to translocate from the cytoplasm to the Golgi compartment thus leading to the appearance of an activated enzyme (19,25), PKC⑀ can be considered as an upstream kinase phosphorylating the activation loop of PKC, which has already been demonstrated for PKC (13).
Using in vitro kinase assays, we have tested whether recombinant PKC, PKC⑀, and PDK1 are able to phosphorylate recombinant Sf9-produced purified PKC, leading to kinase activation. As shown in Fig. 1, left-hand panels, recombinant PKC displays autophosphorylation leading to substrate (syntide 2) phosphorylation, whereas kinase-dead His-tagged PKC K612W shows no evidence of autophosphorylation at all (Fig. 1). PKC K612W phosphorylation is enhanced by adding purified SF9-expressed PKC (positive control) and also, to a significantly higher degree, by adding purified PKC⑀ ( Fig. 1, upper panel), thus indicating a direct phosphorylation as shown for PKC. In contrast to the phosphorylation of kinasedead PKC by PKC⑀, the phosphorylation of wild-type PKC leads to a significant enhancement of syntide 2 phosphorylation ( Fig. 1, lower panel). SF9-expressed purified PKC and PKC⑀ do not phosphorylate syntide 2 significantly but do show autophosphorylation (Fig. 1). As a control for nonspecific phosphorylation, purified PDK1 was used to phosphorylate PKC. No enhancement of PKC phosphorylation by recombinant PDK1 was detected, demonstrating the specificity of PKCand PKC⑀-mediated PKC activation.
PKC Colocalizes with PKC⑀-PKD has been shown to be associated with PKC⑀ using biochemical techniques such as in vitro pull-down assays using glutathione S-transferase fusion proteins and coimmunoprecipitations (12). Based on these data, we questioned whether both enzymes colocalize in HEK293 cells. Immunofluorescence analysis was performed after transient coexpression of a PKC-GFP fusion protein with PKC⑀ in HEK293 cells. As shown in Fig. 2 PKCs are activated through a complex cascade involving lipids and the phosphoinositide-dependent kinase PDK1 (4, 16, 26). Therefore, we were interested in analyzing the role of PDK1 in PKC/PKC activation. First, we tested whether cel-FIG. 1. PKC⑀ phosphorylates and activates PKC in vitro. The indicated purified Sf9-or Sf158-produced proteins were in vitro phosphorylated, fractionated by SDS-PAGE, transferred to a membrane, and exposed to overnight autoradiography (upper panel). Aliquots from each reaction were incubated with the PKC substrate syntide 2, fractionated by 20% SDS-PAGE, dried, and exposed to autoradiography. lular colocalization can be demonstrated upon transient expression of PDK1 and PKC⑀. As shown in Fig. 2, second row, the colocalization of PKC⑀ and PDK1 can be seen albeit to a lesser extent. These findings correlated with published data (4) demonstrating a complex between PDK1 and several PKC isotypes. This prompted us to test whether PKC colocalizes with PDK1. Upon transient coexpression of PDK1 and PKC-GFP, a weaker colocalization can be demonstrated compared with PDK1 and PKC⑀ (Fig. 2, upper and center panels). Similarly, as demonstrated in HEK293 cells, colocalization could be shown in the mammary epithelial cell line MCF7 upon transient coexpression of PKC-GFP with PKC⑀ and, to lesser degree, of PKC⑀ with PDK1 (Fig. 2, fourth and fifth row). These data demonstrate an intracellular complex between PDK1 and PKC with PKC isozymes.
Previous studies demonstrated phosphorylation of selected PKC isotypes by PDK1 at the activation loop as well as the presence of a complex between PKC and PDK1 (4,16). Therefore, we further analyzed the association between PDK1 and PKC using biochemical techniques. PKC wild type and mutants were coexpressed with c-myc-tagged PDK1 and a kinasedead mutant PDK1 K112W . From the obtained cell lysates, 90% were used to immunoprecipitate PDK1 with an anti c-myc antibody to detect PDK1 (Fig. 3, center panel) and to detect associated PKC (Fig. 3, upper panel). 10% of the lysates were used to estimate PKC expression (Fig. 3, lower panel). Immunoprecipitation efficiency of PKC was monitored using the same amount of cell lysate as used for coimmunoprecipitations (Fig. 3, right lane, upper panel). There was no endogenously expressed PKC detectable in anti-c-myc immunoprecipitates, whereas kinase-dead PKC K/R , constitutive active PKC A/E , as well as transiently expressed wild-type PKC were readily detected in PDK1 immunoprecipitates (Fig. 3, upper panel). Apparently, the association of PKC/PDK1 is independent of PDK1 kinase activity. In immunoprecipitates of PDK1 K112W as well as in wild-type PDK1, PKC could be detected. Interestingly, kinase-dead PDK1 showed a faster migration in SDS-PAGE compared with wild-type PDK1, which may be attributed to the lack of autophosphorylation (Fig. 3, center panel). In addition, PKC migration showed a shift in PDK1 immunoprecipitates, which is probably because of PDK1-dependent phosphorylation and which is absent in PDK1 K112W immunoprecipitates (Fig. 3, upper panel). PDK1/PKC association is independent of phosphorylation/autophosphorylation as coimmunoprecipitates of kinase-dead mutants coprecipitate with the same efficiency as the wild-type enzymes. Expression levels of PKC WT and of the mutants used were monitored by Western blot detection (Fig. 3, lower panel).
PDK1 Phosphorylates and Activates PKC-As shown in Fig.  3, the differential migration of PKC in PDK1 immunoprecipitates (upper panel) indicated a phosphorylation of PKC through PDK1. Therefore, we further analyzed these findings using in vitro kinase assays of PKC immunoprecipitates. PKC was expressed as a GFP-fusion protein in HEK293 cells, migrating at a different size in SDS-PAGE as PDK1. This expression clearly enables a distinction to be made between PKC phosphorylation from the PDK1 autophosphorylation signal. PKC was immunoprecipitated using a COOH-terminal antiserum and in vitro phosphorylated with SF9-produced pu- . Intracellular colocalization was analyzed using confocal immunofluorescence microscopy. 24 h after transfection, cells were fixed and stained for PKC⑀ and PDK1 with rabbit antiserum followed by incubation with fluorescence-labeled secondary antibodies. PKC and PKC were expressed as carboxyl-terminal fusion protein (center panels) and visualized by direct fluorescence. Both stains were combined (right-hand panels). The overlay is indicated by the yellow color. The indicated expression constructs were transiently expressed in MCF7 cells and stained for immunofluorescence as indicated (center panels).
FIG. 3. PDK1 coprecipitates with PKC. HEK293 cells were cotransfected with c-myc-tagged PDK1, kinase-dead PDK1 (PDK1 K112W ), and the indicated PKC mutants. 40 h upon transfection, cells were lysed. PDK1 was immunoprecipitated using an anti-c-myc monoclonal antibody and proceeded to Western blot analysis using an anti-PKC rabbit antiserum for detection (upper panel). Aliquots of the PDK1 immunoprecipitates were stained with an anti-c-myc monoclonal antibody as a control (center panel). PKC expression is shown by Western blot analysis of total cell lysates (TCL). All blots shown were visualized through an alkaline phosphatase detection system. IP, immunoprecipitate.
rified PDK1. As shown in Fig. 4A, upper panel, PDK1 incubation leads to an enhancement of the PKC phosphorylation band demonstrating a direct phosphorylation by PDK1. Phosphorylation of immunoprecipitates from kinase-dead PKC K/R by PDK1 was also monitored but could not be clearly distinguished from the strong autophosphorylation signal of PDK1, as both enzymes migrate at similar size in SDS-PAGE (data not shown). As a control, PKC expression levels are shown using Western blot detection (Fig. 4A, lower panel).
We further estimated whether PKC-mediated PDK1 phosphorylation leads to an enhanced substrate phosphorylation efficiency of PKC. PKC immunoprecipitates were incubated with the in vitro substrate MBP. As shown in Fig. 4B, the enhancement of MBP phosphorylation could be shown approximately upon 5-min incubation, decreasing within a 1-h period. Under the same experimental conditions, the addition of purified PDK1 significantly enhances the substrate phosphorylation capacity of PKC, which shows high activation levels after just 3 min decreasing to approximate basal levels within a period of 60 min (Fig. 4B, lower panel). Using similar amounts of PKC immunoprecipitates, a significant enhancement in substrate phosphorylation efficiency can be demonstrated upon incubation with PDK1, which demonstrates an in vitro activation of PKC through PDK1 in accordance with the phosphorylation data.
PDK1 Differentially Affects PKC⑀ and PKC-triggered PKC Activation-PDK1 acts as an upstream kinase for PKC (Fig.  4) and PKC⑀ (4). Both enzymes activate PKC, predicting that a three-level activation cascade amplifies input signals and consequently enhancing PKC substrate phosphorylation. To test this hypothesis, we coexpressed PDK1, PKC/⑀, and PKC to measure PKC activation. The coexpression of PKC or PKC⑀ with PKC leads to stronger syntide 2 phosphorylation of PKC immunoprecipitates (Fig. 5A, lower panel)  Immunoprecipitates of PKC were in vitro phosphorylated with purified recombinant PDK1 from Sf9 cells, fractionated by SDS-PAGE, transferred to a membrane, and exposed to autoradiography (upper panel). PKC immunoprecipitates were visualized by Western blot detection. As a negative control, PDK1 was incubated with immunoprecipitates from vector-transfected cells. As a positive control, PDK1 was incubated with purified PKC form SF9 cells. B, time course of PKC activation by PDK1. Immunoprecipitates of PKC were in vitro autophosphorylated with (lower panels) or without (upper panels) purified PDK1. Loading controls were performed by Western blot analysis. MBP was included to measure substrate phosphorylation. Western blots were visualized by alkaline phosphatase. PKC-transfected HEK293 cells. Coexpression of PDK1, PKC⑀, and PKC also leads to the enhancement of syntide 2 phosphorylation of PKC immunoprecipitates and similarly of immunoprecipitates from PKC⑀/PKC-and PKC-transfected cells. The expression of kinase-dead PDK1 with PKC⑀ and PKC reversed this effect, showing an inhibition of PKC substrate phosphorylation (Fig. 5A, lower panel). PDK1 acts completely differently on PKC-triggered PKC activation in HEK293 cells. The coexpression of PDK1 with PKC and PKC leads to a significant reduction in syntide 2 phosphorylation of PKC immunoprecipitates (Fig. 5A, lower panel). Coexpressing PKC with kinase-dead PDK1 K112W largely does not affect PKC activation. These data suggest a PDK1-and PKC⑀-dependent pathway leading to PKC activation. The coexpression of PDK1 and PDK1 K112W does not notably influence the substrate phosphorylation efficiency of PKC immunoprecipitates (data not shown).
Under the experimental conditions used here, i.e. coexpression of PKC with the indicated upstream kinases, the phosphorylation state of PKC immunoprecipitates only showed weak changes (Fig. 5A, upper panel), representing most probably a basal autophosphorylation. Expression levels of PKC immunoprecipitates (Fig. 5A, center panel) and transgenic expression (data not shown) were verified using Western blot analysis. Kinase activity and protein expression of all expression constructs used for cotransfection experiments were analyzed through in vitro kinase assays and immunoblots (Fig. 5A,  right-hand panels).
PDK1 is able to phosphorylate and activate PKC in vitro (Fig. 4) in a similar way as PKC⑀ (4). In contrast, upon transient expression of PDK1 with PKC/PKC⑀ and PKC, only PKC⑀ is activated, thus leading to PKC phosphorylation. Therefore, we measured the activation state of PKC⑀ and PKC upon coexpression with PDK1 and/or PKC. As shown in Fig.  5B by the autophosphorylation and MBP phosphorylation of PKC immunoprecipitates, PDK1 activates PKC⑀ and PKC. Upon additional coexpression of PKC, PKC⑀ kinase activity is unaffected, whereas PKC kinase activity is reduced. Expression levels of PKC immunoprecipitates used for in vitro kinase assays as well as transgenic expression were monitored using Western blot analysis (data not shown).
Furthermore, we analyzed whether PKC phosphorylates PKC at amino-terminal regulatory serine residues, serving as a binding domain for regulatory proteins as 14-3-3 (27). This could potentially result in the inhibition of PKC kinase activity, e.g. by binding of 14-3-3 proteins. The amino-terminal domain of PKC was transiently expressed as a GFP fusion protein and immunoprecipitated. The immunoprecipitates of constitutive active PKC A/E , kinase-dead PKC K/R (as a negative control), and PKC were used to phosphorylate PKC 1-325GFP. As shown in Fig. 5C, strong phosphorylation of PKC 1-325 GFP could be obtained using PKC immunoprecipitates but not with PKC A/E immunoprecipitates. Immunoprecipitates were analyzed by Western blot for the presence of the respected enzymes and fusion proteins (data not shown). In accordance with experiments showing phosphorylationdependent shifts (19) upon coexpression of PKC or PKC with PKC 1-325 GFP (data not shown), our data indicate that PKC does not phosphorylate amino-terminal serine residues of PKC. These findings show that the inhibition of PKC upon coexpression of PDK1 and PKC must occur via a different mechanism.
Coexpression of PDK1, PKC⑀, and PKC Inhibits JNK Activation-Because in living cells, both novel PKC isozymes act differentially on the activation state of PKC, downstream effectors may be similarly influenced. Several laboratories have shown that PKC/PKD activation negatively interferes with the activation of the JNK pathway (13,28,29). JNK has been shown to colocalize with PKD upon overexpression in COS cells (30).
Therefore, we tested whether the PKC regulation of JNK activity is triggered via PDK1/PKC or PDK1/PKC⑀. JNK activation was measured upon transfection of the indicated expression constructs (Fig. 6) by the in vitro phosphorylation of glutathione S-transferase c-jun using JNK immunoprecipitates. As a positive control for JNK activity, cells were stimulated with tumor necrosis factor. As shown in Fig. 6, JNK activity is largely unaffected by the transfection of PKC. Transient expression of PKC⑀ leads to a moderate effect on JNK activity, which can be strongly enhanced through the coexpression of PKC⑀ with PDK1. These findings correspond with earlier studies showing PKC⑀ activation of JNK in myocytes (31). PDK1 expression leads to some enhancement of JNK activity. Interestingly, the coexpression of PDK1 with PKC⑀ and PKC leads to a significant inhibition of JNK activity. This correlates with the activation state of PKC kinase activity (Fig. 5). Using the same experimental approach, PKC leads to some JNK activity, whereas the expression of PKC and PKC or PDK1/PKC with PKC does not significantly affect JNK activity.
As shown by Western blot analyses, endogenous PKC⑀ is significantly expressed, but endogenous PKC cannot be detected in HEK293 cells (Fig. 6, center panels). These findings clearly indicate PKC⑀ as the biological relevant upstream kinase in HEK293 cells, triggering PKC downstream signaling pathways such as the JNK activation. At the same time, a PKC⑀-triggered pathway exists in HEK293 cells independently of PKC triggering JNK activation. The results of this study including published data (13) are shown in diagrammatic form in Fig. 7.

DISCUSSION
In this study, we have demonstrated that PKC⑀ acts as an upstream kinase in a similar way as previously shown for PKC, directly phosphorylating PKC in vitro and enhancing its substrate phosphorylation efficiency. Using confocal analysis, we have also shown an intracellular colocalization of PKC⑀ and PKC in different cell types and a colocalization of the upstream kinase PDK1 with PKC⑀ and PKC. The latter results were independently confirmed by biochemical coprecipitation. Although we are able to show a direct phosphorylation In a recent study, we have demonstrated that PKC acts as an activation loop kinase for PKC (13). These findings in accordance with earlier data from other laboratories (11,12) indicate a direct phosphorylation of PKC by PKC and PKC⑀. In consequence, using purified PKC⑀/PKC and PKC enzymes, we are able to demonstrate the direct phosphorylation of PKC, leading to an enhanced enzymatic activity ( Fig. 1) (13). Similarly to PKC, PKC⑀ also colocalizes intracellularly with PKC (Fig. 2), indicating a functional relationship with the Golgi compartment as both enzymes were reported to be localized and involved in Golgi-specific functions (23,24).
The discovery of the intracellular colocalization of both PKC isotypes with PDK1 is of interest in terms of their functional relationship. For PKC⑀ and PDK1, colocalization and activation has already been published previously (4). In this study, we demonstrate the biochemical colocalization for PKC and PDK1 that is independent of either PDK1 or PKC kinase activity (Fig. 3). The finding that kinase-dead PKC K/R migrates with a slower relative molecular weight in PDK1 immunoprecipitates already indicates a shift attributed to PDK1 phosphorylation. Consequently, we have demonstrated a direct phosphorylation of PKC by PDK1, which in turn activates PKC (Fig. 4).
PKC activation is thought to be triggered by the initial phosphorylation of PDK1 at the activation loop followed by autophosphorylation in the hydrophobic motif, finally allowing cellular activation through lipids (for review see Ref. 17). The data published until now predict that PDK1 acts as the activation loop kinase either for conventional PKCs (16), atypical PKC (32), or novel PKCs (4), which is corroborated by our data showing activation loop phosphorylation of PKC (Fig. 4). These findings are not unexpected as the primary structure of the activation loop of PKC isotypes is significantly conserved (17).
The spatio temporal aspects of PKC activation are currently unclear. As activated enzyme is located at the Golgi, activation could occur by recruiting PKCs and PDK1 to the Golgi. PKC⑀ has been reported to be localized via the C1 domain at the Golgi (23). PDK1 could be recruited via direct binding to PKC (Fig. 4) (4) or by binding to Golgi-generated phosphatidylinositol phosphate 2 (33), which has been shown to bind the PH domain of PDK1 (34).
Taking into account the in vitro data showing activation of PKC by PDK1, it is of interest to note that upon transient expression of PDK1 with PKC⑀ or PKC and PKC, activation of PKC occurs only through PKC⑀ (Fig. 5). Upon coexpression of PDK1 and PKC with PKC, PKC is inhibited (Fig. 5B), which points to a negative regulation through PDK1 and PKC. As PKC is downstream PKC, a negative feedback loop regulated by PKC can be postulated. We could further exclude the inhibition of PKC via direct phosphorylation of amino-terminal serines through PKC (Fig. 5C). These serines are considered as binding domains for negatively regulatory proteins as 14-3-3 (27,35).
Selective activation of PKC through PKC⑀ further acts on downstream signaling cascades such as JNK. The PDK1 dependence of PKCs explains our earlier findings of expressing a constitutive active mutant of PKC with PKC and thus inhibiting JNK activity. This mutant (PKC A/E ) most probably bypasses the need for PDK1 activation, leading to PKC activation and inhibition of JNK (13). The finding that PDK1 and PKC⑀ coexpression in HEK293 cells leads to enhanced JNK activation indicates a PKC⑀-dependent JNK activation, which has to be distinguished from the opposite effect, namely the PKC⑀/PKC mediated inhibition of JNK activity. PKC⑀ is involved in several cellular functions in which the PKC-mediated effect seems to be dominant via PKC⑀-mediated JNK activation.
A recent study reported a negative regulation of JNK through PKD by the physical interaction of activated PKD with JNK, leading to the inhibition of JNK and c-jun phosphorylation by the sterical blocking of substrate access (30). In an earlier study, direct phosphorylation of Thr-654/Thr-669 of the EGF receptor via PKC has been suggested. This probably leads to negative interference with EGF-induced JNK activation through the PDGF-mediated activation of PKC (see Fig.  7) (29). Both models complement each other, implicating a PDGF-induced activation of PKC/PKD via PDK1/PKC⑀ and leading to the phosphorylation of the EGF receptor and, in parallel, the complexing of JNK with PKC/PKD. This would provide a potential parallel switch effectively shutting down JNK via PKC activation.
PKC activation through distinct PKC subtypes probably reflects cell-type/tissue-specific functions. PKC expression has been associated with the cellular differentiation of keratinocytes (36), which do strongly express PKC (37). In contrast, PKC is not detected in HEK293 cells, whereas endogenous PKC⑀ is expressed significantly higher (Fig. 6, center panel), pointing to an endogenous PKC activation via PKC⑀. Keratinocyte differentiation is dependent upon PKC (38) and is accompanied by the down-regulation of EGF receptor signaling (39). These findings point to a role played by PKC activated via PDK1, thus, phosphorylating the activation loop of PKC and resulting in the phosphorylation of the EGF receptor and to a role in initial signal events preceding cellular differentiation in this cell type.
Acknowledgments-We thank Angelika Hausser and Eva Behrle in performing immunofluorescence analysis and Klaus Pfizenmaier for critical discussions. We are grateful to Helen Schlie␤er for critical reading of the paper.