Protein Kinase C Phosphorylates Protein Kinase D Activation Loop Ser 744 and Ser 748 and Releases Autoinhibition by the Pleckstrin Homology Domain*

Persistent activation of protein kinase D (PKD) via protein kinase C (PKC)-mediated signal transduction is accompanied by phosphorylation at Ser 744 and Ser 748 located in the catalytic domain activation loop, but whether PKC isoforms directly phosphorylate these residues, induce PKD autophosphorylation, or recruit intermediate upstream kinase(s) is unclear. Here, we ex-plore the mechanism whereby PKC activates PKD in response to cellular stimuli. We first assessed in vitro PKC-PKD transphosphorylation and PKD activation. A PKD738–753 activation loop peptide was well phosphorylated by immunoprecipitated PKC isoforms, consistent with similarities between the loop and their known substrate specificities. A similar peptide with glutamic acid replacing Ser 748 was preferentially phosphorylated by PKC (cid:1) , suggesting that PKD containing phosphate at Ser 748 is rapidly targeted by this isoform at Ser 744 . When incubated in the presence of phosphatidylserine, phorbol 12,13-dibutyrate and ATP, intact PKD slowly auto-phosphorylated in the activation loop but only at Ser 748 . In contrast, addition of purified PKC (cid:1) to the incubation mixture induced rapid Ser 744 and

Protein kinase C (PKC) 1 is an enzyme family of central importance in signal transduction (1,2). The ten distinct PKC isoforms thus far identified by molecular cloning comprise three subfamilies differing in structure and regulation (1,3,4). Classical (␣, ␤I, ␤II, and ␥) and novel (␦, ⑀, , and ) PKCs are allosterically stimulated by diacylglycerol, produced rapidly in response to increased activity of phospholipase C/D enzymes (5,6), and are cellular targets of the tumor-promoting phorbol esters (7). Classical PKCs also possess functional Ca 2ϩ -binding sites that mediate sensitivity to Ca 2ϩ signals (1). In contrast, atypical (/ and ) PKCs respond to 3Ј-phosphorylated inositol phospholipids (8) but do not bind diacylglycerol/phorbol esters or Ca 2ϩ .
PKCs are implicated as mediators of a diverse array of biological functions, including both short term alterations in cellular activities and the long term determination of cell fate (9). PKC isoforms exhibit distinct expression patterns, and in response to signaling events, become dynamically targeted to discrete subcellular locations and anchored in positions adjacent to substrates (10,11). This differential dynamic localization, as well as intrinsic substrate selectivity (12), may confer separate signaling roles to the individual PKC isoforms. However, despite their recognized importance in signal transduction, few links have been established between individual PKC isozymes and the direct targets that specify their individual biological outcomes.
Protein kinase D (PKD; the murine homologue of human PKC) (13,14) and two recently identified serine protein kinases termed PKC (15) and PKD2 (16) constitute a new protein kinase subfamily separate from the previously identified PKCs. Salient features of PKD structure and function include an N-terminal regulatory domain comprising a putative transmembrane region, a diacylglycerol/phorbol ester binding cysteine-rich domain, and a pleckstrin homology (PH) domain and a C-terminal catalytic domain with a primary sequence and substrate specificity divergent from those of PKCs. Like the classical and novel PKCs, PKD catalytic activity is stimulated in vitro by diacylglycerol or biologically active phorbol esters (17). The cysteine-rich domain and PH domain both play a role in the negative regulation of PKD catalytic activity, because deletion of either of these domains produces a constitutively active enzyme (18,19).
Physiological activation of intact PKD within cells occurs via a phosphorylation-dependent mechanism first identified in our laboratory (20). In response to cellular stimuli, PKD is converted from a low activity form into a persistently active form that is retained during isolation from cells. Engagement of specific G protein-coupled receptors either by multiple peptides (21)(22)(23)(24) or lysophosphatidic acid (23,25), activation of receptor tyrosine kinases such as the platelet-derived growth factor receptor (21), signaling via heterotrimeric and monomeric G proteins (26,27), and oxidative stress (28) were demonstrated to induce PKD activation in a wide variety of cell types including Swiss 3T3 fibroblasts, small cell lung cancer and pancreatic cancer, normal epithelial and smooth muscle cell lines, cardiocytes, and lymphocytes (23, 25, 29 -34).
Throughout all these studies, multiple lines of evidence have indicated that PKC activity is indispensable for PKD activation by cellular stimuli. Cell treatments with phorbol esters or other agents that bypass surface receptors and directly stimulate PKCs are potent triggers of cellular PKD activation (20,35). Cotransfection of PKD, together with active mutant forms of PKC⑀ or PKC, also dramatically activates PKD in the absence of cell stimulation (20,36). Preincubation of cells with the specific PKC inhibitors GF 109203X (37) or Ro 31-8220 (38) that do not directly inhibit PKD (20) impairs PKD activation by all stimuli, reducing it by 70 -80% when used at maximally effective concentrations (28). Furthermore, PKD interacts preferentially with PKC, forming complexes involving the PH domain (36). These findings imply the existence of PKC-PKD protein kinase cascade(s), and we have postulated that some PKC-dependent biological effects involve PKD acting either in parallel or as a downstream intermediate. In particular, PKD has been reported recently to mediate several important cellular activities and processes, including function and organization of the Golgi apparatus (39), metastatic tumor cell invasion (40), epidermal growth factor receptor signaling (41,42), Na ϩ /H ϩ antiporter (43), mitogen-activated protein kinase activation (44), proliferation (45), and adenomatous transformation (46). Moreover, critical downstream targets of PKD signaling are beginning to emerge (47)(48)(49). Therefore, the mechanism whereby PKC activates PKD has been attracting intense interest.
Our previous studies identified Ser 744 and Ser 748 in the PKD activation loop as phosphorylation sites critical for PKC-mediated PKD activation. Whereas a PKD mutant with Ser 744 and Ser 748 mutated to alanines (PKD-S744A/S748A) could not be activated by cellular stimuli, a mutant with these residues mutated to glutamic acid residues (PKD-S744E/S748E) possessed dramatically increased constitutive activity (50). We have shown that Ser 744 and Ser 748 are phosphorylated in vivo during PKD activation in response to phorbol ester stimulation, in a manner blocked by preincubation with GF 109203X (50,51). Specifically, in two-dimensional 32 P tryptic phosphopeptide maps, individual spots were selectively eliminated when PKD forms point-mutated at Ser 744 or Ser 748 were used in transfections (50). Because kinase-deficient PKD, which retains Ser 744 and Ser 748 , also became phosphorylated during stimulation with phorbol ester, we concluded that transphosphorylation at these sites by an upstream kinase, e.g. novel PKCs, was a critical factor responsible for in vivo PKD activation (50,51).
A different group of researchers (52) have questioned the role of activation loop transphosphorylation in the PKD activation mechanism. Consequently, we reinvestigated PKD activation loop phosphorylation using phosphospecific antisera generated against phosphorylated Ser 744 and Ser 748 (34). In these studies, we demonstrated that Ser 744 and Ser 748 phosphorylation were concomitant with activation, induced by either a variety of receptor-mediated or -independent stimuli or by transfection of PKC⑀ or PKC in the absence of cell stimulation. These approaches also produced phosphorylation of kinase-deficient PKD forms at both Ser 744 and Ser 748 , supporting the view that PKC, but not PKD, activity induced phosphorylation of these residues. However, the precise mechanisms involved remain unclear (20,36,(51)(52)(53). In the present study, we use peptides, full-length PKD, and mutant PKD proteins to examine whether PKC mediates direct PKD activation loop Ser 744 and Ser 748 phosphorylation and activation.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfections-Stock cultures of COS-7 cells were maintained in 10-cm tissue culture plates by subculturing every 3-4 days in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum in a humidified atmosphere containing 10% CO 2 and 90% air at 37°C. Confluent stock cells were reseeded at a density of 6 ϫ 10 4 cells/ml in 6-cm dishes, 8 -18 h prior to transfections. All transfections and cotransfections were carried out with equivalent amounts of DNA (5 g/dish for single transfections, 3 g/dish of each DNA for cotransfections), using vector pcDNA3 as the control DNA added to single transfections. Transfections were carried out in Opti-MEM (Invitrogen) using Lipofectin (Invitrogen) at 10 l/dish. DNA-Lipofectin complexes were formed according to the protocol provided by the manufacturer and then cell cultures were layered over with these DNA complexes in a final volume of 2.5 ml/dish in the absence of serum and incubated at 37°C in a humidified atmosphere containing 5% CO 2 , for 5-6 h or overnight to allow uptake of complexes. Fetal bovine serum (10% final concentration) in Opti-MEM was then added to dishes to yield a final volume of 5 ml/dish. Cells were used for experiments after a further 72 h of incubation.
Plasmid Constructs and Fusion Proteins-Plasmid constructs encoding PKCs and wild-type and mutant PKD forms used in this study have been described previously (17,18,20,36,50). The construct pcDNA3-PKD⌬PH/S744A/S748A was generated by standard subcloning procedures. Thus, a Bsu36I restriction fragment within pcDNA-PKD⌬PH (nucleotides 2305-2827 of PKD) was replaced with a corresponding Bsu36I fragment excised from pcDNA3-PKD-S744A/S748A. The GST-Cat and GST-Cat/S744A/S748A fusion proteins were constructed in pGEX-4T by standard subcloning procedures. Thus, C-terminal BamHI fragments (corresponding to amino acids 559-918) of either wild-type PKD or PKD-S744A/ S748A were fused in-frame at the C-terminal of the sequence encoding glutathione S-transferase in the construct pGEX-4T (Amersham Biosciences). Fusion proteins were isolated by affinity chromatography using glutathione-Sepharose 4B (Roche Molecular Biochemicals) and eluted with 10 mM glutathione.
Immunoprecipitations-COS-7 cells transfected either with wildtype or mutant PKD or PKC isoforms were treated as indicated in the figures and then lysed in 1 ml of lysis buffer (1% Triton X-100, 2 mM EDTA, 2 mM EGTA, 2 mM dithiothreitol, 1 g/ml aprotinin, 10 g/ml leupeptin, 1 mM 4-(2-aminoethyl)-benzenesulfonyl fluoride hydrochloride in 50 mM Tris-HCl, pH 7.4). Small amounts (typically 1 ⁄10) of these total lysates were saved and combined with equal volumes of 2ϫ SDS-PAGE sample buffer (0.2 M Tris-HCl, pH 6.8, 6% SDS, 2 mM EDTA, 4% 2-mercaptoethanol, 10% glycerol) for Western blot analysis. To generate PKC or PKD immunoprecipitates, medium was removed, and lysis buffer was added to cells on ice. These lysates were cleared by centrifugation at 15,000 rpm for 10 min, and the cleared supernatants were combined with antibodies and protein A-agarose (30 l) and placed on a rotator at 4°C for 3 h. Immune complexes were collected by brief spinning and washed thoroughly before subsequent kinase assays. Antibodies used for immunoprecipitations were either the isoform-specific antisera from Santa Cruz Biotechnology (for PKC⑀, PKC, or PKC) or the previously described PA-1 antiserum (17) for PKD (1:100 dilution).
Peptide Phosphorylation Assays-For assays of peptide phosphorylation by PKD or PKCs, immune complexes were washed twice with lysis buffer and then twice with kinase buffer consisting of 30 mM Tris-HCl, pH 7.4, 10 mM MgCl 2 , 2 mM dithiothreitol. Substrate peptides (⑀-peptide, syntide-2, PKD738 -754, PKD-738 -754/S744E, PKD-738 -754/S748E, or the PKD C-terminal peptide EEREMKALSERVSIL used to generate PA-1 antiserum, termed PKD CT, each at a final concentration of 2.0 mg/ml), were then added in the presence of [␥-32 P]ATP (2 Ci/reaction diluted with cold ATP to give a final concentration of 100 M) in kinase buffer (final reaction volume, 30 l) and transferred to a water bath at 30°C for 10 min. Reactions were terminated by adding 100 l of 75 mM H 3 PO 4 , and 75 l of the mixed supernatant was spotted to Whatman P-81 phosphocellulose paper. Papers were washed thoroughly in 75 M H 3 PO 4 and dried, and radioactivity incorporated into peptides was determined by detection of Cerenkov radiation in a scintillation counter.
PS/PDB phospholipid vesicles used for assays of peptide or protein phosphorylation by PKD or PKC⑀ were prepared by dehydrating 300 g of PS under ethanol (200 l) in a Speedvac dehydrator and then sonicating the dried lipid into kinase buffer (typically 300 l) in the presence of added PDB and 0.05% nonionic detergent (Triton X-100). For assays of syntide-2 phosphorylation by purified PKD or PKC⑀, fresh aliquots (typically 2 g) of enzyme were thawed on ice, combined with PS/PDB vesicles (final concentrations, 200 g/ml and 200 nM, respectively, in kinase buffer), and aliquoted to tubes on ice containing kinase buffer with or without inhibitors as indicated in the figures. Reactions were initiated by adding a mixture containing syntide-2 (final concentration, 2 mg/ml), together with 100 M ATP (including [␥-32 P]ATP at 2 Ci/assay), and then transferring to a water bath at 30°C for 7 min. Reactions were terminated by adding H 3 PO 4 as above, spotted to P-81 paper, and washed, and radioactivity was counted in a scintillation counter as above.
PKC-PKD Transphosphorylation/PKD Activation Assays-For transphosphorylation of full-length wild-type or kinase-defective PKD proteins by PKC⑀, we transfected COS-7 cells with PKD or PKD-K618N. After 72 h, cells were lysed, and PKD protein was isolated in immune complexes using PA-1. In each experiment, parallel dishes were included to examine the persistence of PKD activation throughout immunoprecipitation, elution from immunocomplexes, and subsequent reimmunoprecipitation. PKD or PKD-K618N immunoprecipitates from unstimulated cells were washed twice with lysis buffer and then twice with kinase buffer. The PKD-PA-1 immune complexes were then incubated with immunizing peptide (15 g in ϳ15 l of kinase buffer) overnight on ice to elute PKD proteins. For transphosphorylation reactions, the supernatant containing eluted PKD protein was transferred to fresh tubes on ice, adjusted to 20 l, ATP (100 M) was added on ice, and then reactions were initiated by mixing in purified PKC⑀ (Calbiochem) for a final concentration of 5.8 g/ml, combined with PS/PDB vesicles (either with or without 3 M Gö6983), and transferred to a 37°C water bath. Reactions were terminated at 2.5 min by adding 1 ml of ice-cold lysis buffer and were transferred to ice. PKD antibody (C-20; Santa Cruz Biotechnology) was then added, and PKD was reimmunoprecipitated for 3 h, washed as before, and then subjected to syntide-2 assays in the presence of 3 M Gö6983. Activation assays using purified PKC⑀ to transphosphorylate purified PKD (Calbiochem) were conducted similarly beginning with the addition of ATP to purified PKD in kinase buffer (20 l).
For measurements of PKC⑀ transphosphorylation/PKD autophosphorylation by 32  Autophosphorylation or syntide-2 phosphorylation reactions to assess activity of the PKD-⌬PH-S744A/S748A mutant were conducted as described previously for PKD activity measurements (28). Briefly, COS-7 cells were transfected with the different plasmids. After 72 h, cells were left untreated or treated with PDB (200 nM for 10 min) and then lysed, and PKD or PKD mutant protein was immunoprecipitated from the lysates. Immunoprecipitates were washed twice with lysis buffer and twice with kinase buffer and subjected to autophosphorylation reactions or syntide-2 phosphorylation reactions as described (28). Activity of the catalytic domain GST fusion proteins was measured by immobilization on glutathione-Sepharose beads and subsequent incubation (30°C for 10 min) with syntide-2 (2 mg/ml), together with ATP (100 M with 1 Ci/assay of [␥-32 P]ATP). Reactions were terminated by addition of H 3 PO 4 , spotted to P-81 paper, washed, and counted as above.
Western Blot Analysis-For Western blot analysis, samples or cell lysates were directly solubilized by boiling in SDS-PAGE sample buffer. After resolving by SDS-PAGE, proteins were transferred to Immobilon-P membranes (Millipore) as described previously (21). To block nonspecific protein binding, membranes were blocked by incubation with either 5% non-fat dried milk in PBS, pH 7.2 (for PKD C-20 or the antibody pS748 that recognizes PKD) (34), or 5% bovine serum albu-min/0.1% Tween 20 in PBS (for the antibody pS744 that recognizes PKD) (34) for 2-3 h at 20-25°C or overnight at 4°C. Membranes were then incubated at room temperature for 2-3 h with antisera, specifically recognizing PKD at a dilution of 1 g/ml, the phosphospecific phosphoSer 748 recognizing PKD phosphorylated at Ser 748 (1:500) in PBS containing 3% non-fat dried milk, or the phosphospecific phos-phoSer 744 raised against a peptide containing phosphoSer 744 and phos-phoSer 748 , chiefly recognizing PKD phosphorylated at Ser 744 , at 1:1000 dilution in 5% bovine serum albumin/0.1% Tween 20 in PBS. Immunoreactive bands were visualized using horseradish peroxidase-conjugated anti-rabbit IgG and enhanced chemiluminescence (ECL reagents; Amersham Biosciences) detection.
Materials-[␥-32 P]ATP (6000 Ci/mmol) was from Amersham Biosciences. Protein A-agarose and 4-(2-aminoethyl)-benzenesulfonyl fluoride hydrochloride (Pefabloc) were from Roche Molecular Biochemicals. The anti-PKD antiserum (C-20) used in Western blot analysis was from Santa Cruz Biotechnologies, Palo Alto, CA. pS744 was from Cell Signalling Technologies, Beverly, MA. Purified PKD (purity Ն 90% by SDS-PAGE), PKC (purity Ն 95%), and PKC⑀ (purity Ն 95%), produced in cells from Spodoptera frugiperda, were from Calbiochem. Opti-MEM and Lipofectin were from Invitrogen. The PKC substrate peptide, peptide ⑀, purified PDK-1, and its control substrate peptide, PDKtide, were obtained from Alexis Biochemicals. Syntide-2 and the PKD activation loop peptides were synthesized at the peptide synthesis core facility of CURE-DDRC at UCLA. The C-terminal PKD peptide, PKD CT and pS748, were synthesized by the central peptide synthesis and antibody production facility at Imperial Cancer Research Fund, London, United Kingdom. All other reagents were from standard suppliers or as described in the text and were the highest grade commercially available.

Comparison of the PKD Activation Loop Segment with the Known Substrate Recognition Sequences of PKC Isoforms-
The initial molecular cloning and expression of PKD and its human homologue PKC was followed by the molecular cloning and expression of another related protein, termed PKD2, and two other clones thus far identified only at the cDNA level (human PKC/PKD3 and a Caenorhabditis elegans clone related to PKD2). A canonical region in the catalytic portion of protein kinases bridges two characteristic amino acid motifs in subdomains VII (DFG) and VIII (APE). This region typically forms an elongated loop positioned adjacent to the active site, referred to as an activation loop/segment because of targeting of this region for regulatory control of enzyme activity by upstream protein kinases acting in a cascade fashion (54,55). Interestingly, the activation loop segment is 100% conserved among all members of the novel PKD protein kinase subfamily (Fig. 1A). We recently demonstrated that Ser 744 and Ser 748 are not phosphorylated significantly in resting cells and that phosphorylation of these residues is rapidly induced by multiple cellular stimuli in a PKC-dependent manner and is concomitant with persistent activation (34). Additional studies from this laboratory have shown that phosphorylation at these sites is also closely linked with dynamic intracellular redistributions of PKD (56 -59). Thus, PKD is regulated by PKC activity at multiple levels, and direct phosphorylation of the activation loop may represent a convergence point for these events.
The amino acid residues in the vicinity of the serine or threonine residues targeted by different protein kinases comprise characteristic substrate recognition sequences that contribute to enzyme specificity. The substrate recognition sequences of various PKC isoforms and PKD/PKC were analyzed by Cantley and co-workers (12) using directed peptide libraries. As shown in Fig. 1B, the substrate recognition sequences of PKC and PKC⑀ possess similarities to the amino acid sequence surrounding Ser 744 and Ser 748 in the PKD activation loop, suggesting that these PKCs might directly phosphorylate these sites in vivo. In particular, the sequences clustered on both sides of Ser 744 , as well as Ser 748 , in PKD correspond to favored PKC⑀ recognition sequences. 2 Similarly, the amino acids in positions adjacent to and C-terminal to Ser 744 , or those N-terminal to Ser 748 , are strikingly similar to those forming PKC recognition sequences (Fig. 1B). However, the resemblance between the PKD activation segment and the substrate recognition sequence of atypical PKC is relatively limited (Fig. 1B), consistent with our previous findings that this isoform is unable to activate PKD in cotransfection experiments (20,36,60).
Phosphorylation of PKD Activation Loop Peptides by Different PKC Isoforms-To investigate whether PKC, PKC⑀, or PKC could recognize the PKD activation segment in vitro, we synthesized a peptide, RIIGEKSFRRSVVGTPA, comprising amino acids 738 -754 of PKD (PKD738 -754) (Fig. 1C) as a substrate for phosphorylation assays. The importance of dual activation loop phosphorylation in the PKD activation mechanism has been underscored by previous studies. Specifically, we demonstrated previously that mutation of both Ser 744 and Ser 748 to glutamic acid residues to mimic dual phosphorylation in PKD produced a highly constitutively active enzyme (50). In contrast, single point mutation of Ser 744 or Ser 748 to glutamic acid residues introduce only minor alterations in PKD activity. 3 Thus, second site phosphorylation of PKD singly phosphorylated at one of the activation loop sites is necessary for full activation. Therefore, to complement assays of PKD738 -754 phosphorylation, we also synthesized peptides with individual glutamic acid changes at the positions corresponding to either Ser 744 or Ser 748 (PKD738 -754/S744E and PKD738 -754/ S748E; shown in Fig. 1C) as substrates that mimic singly phosphorylated PKD activation loop segments in phosphorylation assays.
We first examined immunoprecipitated active PKD in assays of syntide-2, ⑀-peptide, or PKD738 -754 phosphorylation, shown in Fig. 2A. Consistent with our previous studies, activated PKD phosphorylated syntide-2 strongly but ⑀-peptide to only a minimal extent (less than 5% of control), indicating that this peptide is a poor substrate for PKD ( Fig. 2A). Interestingly, PKD738 -754 was also poorly phosphorylated by PKD in these assays ( Fig. 2A), suggesting a lack of intrinsic recognition of the activation loop peptide by PKD.
Next, specific PKC immunoprecipitates (wild-type PKC⑀ or PKC from transiently transfected COS-7 cells) or PKC (a mixture of endogenous wild-type and transfected active mutant enzyme expressed in COS-7 cells) were assayed for phosphorylation of peptide ⑀, syntide-2, or PKD738 -754 (Fig. 2B). Peptide ⑀ was identified previously as an excellent model substrate for all PKCs but not PKD (36,61) and was therefore used as a normalization control to assess the relative degree of phosphorylation of PKD738 -754 by PKC⑀, PKC, or PKC. In comparison with control, and consistent with previous studies (62), syntide-2 was phosphorylated very well by the different PKC isoforms (Fig. 2B). The extent of syntide-2 phosphorylation ranged from ϳ55% (for PKC and PKC) to 75% (for PKC⑀) of peptide ⑀ phosphorylation (Fig. 2B). Consistent with intrinsic recognition of the PKD activation loop by PKC and PKC⑀, these isoforms phosphorylated PKD738 -754 to an extent between ϳ40% (PKC) and 50% (PKC⑀) of those reached in peptide ⑀ assays (Fig. 2B). In contrast, a peptide similar to PKD738 -754 but with both Ser 744 and Ser 748 replaced with glutamic acid residues was not phosphorylated significantly in parallel assays (data not shown), indicating that the single threonine residue (corresponding to PKD Thr 752 ) in these peptides was not phosphorylated by these PKCs.
As shown in Fig. 2B, PKC also phosphorylated PKD738 -754 to an extent similar to those reached by PKC and PKC⑀ (ϳ45% of the extent of peptide ⑀ phosphorylation) suggesting that at least in vitro, this isoform also intrinsically recognizes the PKD activation loop as a possible substrate. However, previous studies from this laboratory indicated that PKC neither interacts with nor activates PKD significantly (20,36,60). Taken together, these results suggest that factors other than intrinsic substrate recognition (e.g. a failure to colocalize with PKD) contribute to the inability of PKC to activate PKD in vivo. Purified PDK-1, an upstream regulator that phosphorylates the activation loop of different PKC isoforms, as well as protein kinase B, did not phosphorylate PKD738 -754 significantly (data not shown). Also shown in Fig. 2B, none of these PKCs significantly phosphorylated the C-terminal PKD peptide that contains a PKD autophosphorylation site (described under "Experimental Procedures"). These data were therefore included as a negative control for specific recognition by PKCs.
We then used the variant activation loop peptides described in Fig. 1A as substrates to examine whether PKC, PKC⑀, or PKC act as second-site PKD activation loop kinases. Because these PKCs all phosphorylated PKD738 -754 to similar extents, we used this peptide as a normalization control to facilitate comparison of relative phosphorylation levels. Shown in Fig. 2C, both PKD738 -754/S744E and PKD738 -754/S748E were well phosphorylated by PKC, indicating that this isoform can act as a second-site kinase to phosphorylate PKD equally well after an initial phosphorylation at either Ser 744 or Ser 748 . Interestingly, and in contrast with PKC, PKC⑀ phos-phorylated the two peptides differentially, demonstrating a strong preference for PKD738 -754/S748E. These data indicated that this enzyme acts preferentially as a second-site kinase to phosphorylate Ser 744 . Interestingly, PKC phosphorylated these peptides only modestly, suggesting that this isoform is relatively deficient as a second-site kinase for the PKD activation loop.
PKC⑀ Increases PKD Autophosphorylation-To further examine whether PKC-mediated PKD activation was direct, we developed in vitro activation assays using purified components. We have shown previously that certain PKCs, notably PKC, form stable molecular complexes with the PKD PH domain when cotransfected, together with PKD in COS-7 cells (36). Therefore, for these studies we selected PKC⑀, which potently activates PKD but forms complexes with PKD relatively poorly. To establish the potential of PKC⑀ to activate intact purified PKD in vitro by phosphorylation at the activation loop sites (i.e. corresponding to Ser 744 and Ser 748 in PKD), we carried out PKD autophosphorylation assays under conditions that either supported only autophosphorylation (i.e. purified PKD, together with [␥-32 P]ATP) or a combination of transphosphoryla-

FIG. 2. Peptide phosphorylation assays of immunoprecipitated PKD and PKCs.
A, PKD was immunoprecipitated from transiently transfected, PDB-stimulated COS-7 cells, and peptide phosphorylation assays were carried out as described under "Experimental Procedures" using syntide-2, peptide ⑀, and PKD738 -754 as indicated. A representative Western blot analysis to detect immunoprecipitated PKD using anti-PKD antibody is shown in the inset. B, PKC, PKC⑀, or PKC, as indicated, were immunoprecipitated from COS-7 cells transiently transfected with PKC expression constructs, and peptide phosphorylation assays were carried out as described under "Experimental Procedures" using peptide ⑀, syntide-2, PKD738 -754, or PKD CT, as indicated. Results are expressed as the percentage of the phosphorylation of peptide ⑀ and represent the mean Ϯ S.E. of at least five determinations. Representative Western blot analysis to detect immunoprecipitated PKC isoforms using specific anti-PKC⑀, PKC, or PKC antibodies are shown in the inset. C, PKC isoforms were immunoprecipitated as in B, and peptide phosphorylation assays were carried out as described under "Experimental Procedures," using PKD738 -754, PKD738 -754/S744E (744E), or PKD738 -754/S748E (748E), as indicated. Results are expressed as the percentage of the phosphorylation of PKD738 -754 and represent the mean Ϯ S.E. PKC phosphorylated 744E to a significantly greater extent than did PKC⑀, as assessed by Student's t test (p ϭ 0.04, n ϭ 9, shown with an asterisk). PKC⑀ phosphorylated S748E to a significantly greater extent than S744E (p ϭ 0.02, n Ն 9, shown with a number sign). Phosphorylation of S748E by PKC was significantly less than by PKC⑀ (p ϭ 0.04, n Ն 7, shown with a double asterisk). tion and autophosphorylation (i.e. including purified PKC⑀), measuring the total extent of PKD phosphorylation over time.
Shown in Fig. 3A, the time courses of PKD autophosphorylation with or without added PKC⑀ were dramatically different when assays were carried out in the absence of PS/PDB vesicles. This was possible, because the preparation of PKC⑀ used here possesses constitutive activity without allosteric effectors (not shown). Thus, purified PKD incubated alone autophosphorylated to only a very limited extent, reaching a maximum within the first 2.5 min of the assay and not increasing significantly thereafter. Similarly, inclusion of PKC, a negative control for PKC-mediated PKD activation, promoted PKD autophosphorylation only very weakly over a 40-min time course, despite having a potent activity demonstrated by autophosphorylation (Fig. 3A), which was not enhanced in the presence of PS/PDB (not shown). In contrast, inclusion of PKC⑀ promoted a continued strong increase in PKD phosphorylation (Fig. 3A). Collected data from multiple experiments, normalized to the initial PKD phosphorylation levels and shown in the graph in Fig. 3A, emphasized the selective enhancement of PKD autophosphorylation by PKC⑀. From these results, we conclude that catalytic activation of PKD triggered by PKC⑀ transphosphorylation permits a dramatic overall increase in PKD autophosphorylation, consistent with our proposal that PKC⑀ activates PKD by activation loop phosphorylation.
In the presence of PS/PDB vesicles that allosterically stimulate PKD enzyme activity without persistent activation, PKD autophosphorylation increased approximately linearly over a 40-min time course (Fig. 3B). Interestingly, even when PKD activity was directly stimulated, inclusion of PKC⑀ in the assay increased the initial rate of PKD phosphorylation (Fig. 3B). These results are consistent with a modest initial acceleration of PKD autophosphorylation by PKC⑀-mediated activation and then continued autophosphorylation until all the possible sites approached saturation.

PKC⑀ Rapidly and Directly Phosphorylates PKD in Both
Activation Loop Sites, Whereas PKD Autophosphorylates at Ser 748 Only-We next verified PKC⑀-mediated phosphorylation of the activation loop sites by Western analysis using the specific antibodies we characterized previously, which recognize the phosphorylated state of Ser 744 and Ser 748 in PKD (Fig. 4).
In these experiments, we incubated PKD with PS/PDB and cold ATP, either alone or together with PKC⑀, and examined activation loop phosphorylation at 30-s intervals by Western analysis using the phosphospecific antibodies, pS748 and pS744.
Experiments using pS748 to monitor Ser 748 phosphorylation over a 2.5-min time course, shown in Fig. 4, revealed the time-dependent appearance of a phosphorylated band corresponding to PKD. Thus, these data indicated that PKD could autophosphorylate at the activation loop after 1 min of incubation in the presence of activators. We also examined PKD phosphorylation in these reactions by Western analysis using the commercial antiserum that recognizes chiefly phos-phoSer 744 (and to a lesser extent phosphoSer 748 (34)). Results shown in Fig. 4 illustrate that pS744 immunoreactivity in PKD increased very little over this time course. Because this antibody binds to a limited extent to phosphorylated Ser 748 , this low degree of immunoreactivity most likely reflects little or no autophosphorylation at Ser 744 but rather corresponds to the increase in Ser 748 phosphorylation observed using pS748.
Significantly, Ser 748 autophosphorylation by isolated PKD occurred slowly, becoming detectable only after a lag of ϳ1-1.5 min of the reaction. In contrast, PKD incubated with PKC⑀ exhibited a rapid increase in pS748 immunoreactivity, detected within 30 s of the reaction (Fig. 4). Furthermore, when PKC⑀ was included in reactions with PKD, pS744 immunoreactivity increased rapidly and produced a robust signal within 1 min, indicating that transphosphorylation of PKD Ser 744 had occurred. These Western blots were also stripped and reprobed with C-20 to demonstrate that equal amounts of PKD protein were present in each sample (Fig. 4). Taken together, data in Fig. 4 showed that PKC⑀ rapidly catalyzed phosphorylation of both Ser 744 and Ser 748 in PKD, whereas autophosphorylation was slower and only occurred at Ser 748 . From these data, we conclude that transphosphorylation is the primary mechanism involved in rapid PKD activation occurring in cells in response to stimuli but that PKD autophosphorylation over a longer time could contribute to the overall increase in enzyme activity. Fig. 4 verified that PKC⑀ had phosphorylated the PKD activation loop sites in our in vitro assays using purified components. We next examined whether PKC⑀-mediated PKD activation loop phosphorylation was associated with stable increases in PKD activity by a two-stage assay. In our previous studies, we have assayed PKD in soluble form by eluting the enzyme from immunoprecipitates (50). Here, we transfected PKD or the kinase-deficient PKD mutant PKD-K618N into COS-7 cells, immunoprecipitated PKD from these cells, and then eluted the enzyme from immunoprecipitates as described under "Experimental Procedures." We then incubated the soluble, eluted PKD either alone or with purified PKC⑀ and unlabeled ATP in the first stage of the assay. After 2.5 min at 37°C, the reactions were stopped and diluted with ice-cold buffer, and PKD was reimmunoprecipitated, and its catalytic activity was measured by syntide-2 phosphorylation in the second stage of the assay (Fig. 5B).

PKC⑀-mediated Activation Loop Phosphorylation Triggers Persistent PKD Activation in Vitro-Results in
To substantiate further the role of PKC catalytic activity in the activation of PKD, we also used an inhibitor, Gö6983, of classic and novel PKCs including PKC⑀ but not PKD (63). To determine the concentrations of Gö6983 required to selectively and completely eliminate PKC⑀ activity in peptide phosphoryl- FIG. 4. PKD activation loop phosphorylation in the absence or presence of PKC⑀. Purified PKD, alone or together with purified PKC⑀ (PKD ϩ PKC⑀) were incubated with ATP in the presence of PS/PDB for the times indicated and then analyzed by SDS-PAGE and Western blot analysis using the phosphospecific antisera pS744 or pS748 as described under "Experimental Procedures." To demonstrate that equal amounts of PKD protein were present, blots were stripped and reprobed with the anti-PKD antibody C-20 recognizing total protein. Western blots (upper panels) and corresponding band intensities depicted in the graph (below), as determined by scanning and densitometry, are representative of three experiments with similar results.
ation assays under our conditions, we performed syntide-2 phosphorylation assays of PKC and PKD with increasing concentrations of the inhibitor. Results shown in Fig. 5A indicated that Gö6983 inhibited PKC⑀ in peptide phosphorylation assays in a concentration-dependent fashion, with half-maximal inhibition at ϳ300 nM and complete inhibition at concentrations above 1 M. In contrast, PKD activity was not significantly inhibited by concentrations of Gö6983 up to 5 M (Fig. 5A). Thus, in the assays of reimmunoprecipitated PKD, 3 M Gö6983 was included in the reaction mixture to eliminate any PKC⑀ activity that might have carried over.
In control experiments, we activated PKD within cells by stimulation with PDB and then immunoprecipitated, eluted, and reimmunoprecipitated the enzyme before assaying. Results (Fig. 5B) demonstrated that PKD activated within cells retained its increased activity throughout all these procedures. Eluted PKD, incubated with ATP and PS/PDB and then reimmunoprecipitated and assayed for syntide-2 activity, had a low level of activity that was very similar to the control enzyme isolated from unstimulated cells (Fig. 5B). This clearly indicates that allosteric activators do not induce a persistent increase in PKD activity and that PKD must be phosphorylated at both serines in the activation loop to stabilize the active form of the enzyme. In contrast, eluted PKD incubated with PKC⑀, in the presence of ATP and PS/PDB, then reimmunoprecipitated and assayed, had dramatically increased activity, i.e. more than 4-fold that of the control enzyme (Fig. 5B). Consistent with PKC⑀ activity in the initial incubation having been responsible for the observed increases in PKD activity, incubating PKD with PKC⑀ in the presence of 3 M Gö6983 reduced PKD activation by ϳ40% (Fig. 5B). Control Western blots, shown in Fig. 5B, demonstrated that equal amounts of PKD protein were isolated by the reimmunoprecipitation procedures.
To confirm that an increase in PKD activity, as opposed to that of a coimmunoprecipitated kinase, was responsible for the increased catalytic activity of the reimmunoprecipitated PKD, we incubated eluted kinase-deficient PKD mutant PKD-K618N protein with PKC⑀ in parallel assays and then processed the protein as before for syntide-2 assays. Shown in Fig. 5B, these immunoprecipitates contained no significant activity.
We also carried out experiments similar to those in Fig. 5B but using the insect cell-expressed and purified PKD and PKC⑀ in the initial transphosphorylations. Results, shown in Fig. 5C, demonstrate that incubation of PKD with PKC⑀ increased the activity of the subsequently immunoprecipitated PKD by more than 2.5-fold in comparison with the same amount of PKD incubated by itself and then immunoprecipitated and assayed. Again, in all these syntide-2 assays, 3 M Gö6983 was included to eliminate any possible traces of PKC⑀ activity carried over into the PKD immunoprecipitates. Similar to results in Fig. 5B, this increase could be attenuated significantly by including 3 M Gö6983 in the incubation mixture to inhibit PKC⑀ activity. Taken together, the data in Fig. 5 demonstrate that PKC⑀mediated direct phosphorylation of PKD promotes its conversion from an inactive to an active state.
Relief of Autoinhibition and Activation Loop Phosphorylation Both Contribute to PKD Activation-Autoinhibition is a central feature of the regulation of protein kinase catalytic activity (64). Previous results from this laboratory indicated that PKD mutants lacking the pleckstrin homology domain or the cysteine-rich domain were highly active in the absence of stimulation (18,19). These results suggest that in the intact kinase, the entire N-terminal regulatory domain, i.e. both the cysteinerich and PH domains, help to maintain the enzyme in an inactive, autoinhibited state, and consequently, removal of either domain facilitates activation. Consistent with this interpretation, a maltose-binding protein-PKD catalytic domain fusion protein produced in bacteria was demonstrated previously to be catalytically active (13). Mutation of Ser 744 and Ser 748 in the activation loop to glutamic acid residues introduces negative charges that mimic phosphorylation, which generates a full-length kinase that is highly active in the absence of stimulation. Reciprocally, a PKD mutant in which Ser 744 and Ser 748 in the activation loop are replaced with unphosphorylatable alanine residues (PKD-S744A/S748A) is not activated within cells by stimuli that fully activate wild-type PKD (50). Thus, we hypothesize that the mechanism of PKD activation is based on to activate the enzyme in vivo, as indicated, and then lysed, and PKD was immunoprecipitated and eluted as described under "Experimental Procedures." The eluted enzyme was then either kept on ice or incubated with ATP and PS/PDB at 37°C, either with (filled bars) or without PKC⑀ (dotted bar) and in the presence (gray bar) or absence (black bar) of 3 M Gö6983, as indicated, in PKC-PKD transphosphorylation/PKD activation reactions and then reimmunoprecipitated and assayed for syntide-2 phosphorylation activity as described under "Experimental Procedures." In the presence of Gö6983, PKD activation by incubation with PKC⑀ was significantly inhibited, as assessed by Student's t test (p ϭ 0.02, n ϭ 4, indicated by an asterisk). A control Western blot is depicted to illustrate that PKD was reimmunoprecipitated in equal amounts prior to syntide-2 assays (inset). C, purified PKD, either alone (dotted bar) or together with purified PKC⑀ (black bar) were incubated in the presence of ATP and in the presence or absence of 3 M Gö6983, as indicated, for 2.5 min at 37°C and then diluted with cold buffer, and PKD was immunoprecipitated and assayed for syntide-2 phosphorylation activity as described under "Experimental Procedures." Results presented are the mean Ϯ S.E. from at least three experiments, each performed in quadruplicate. a functional link between activation loop phosphorylation and relief of autoinhibition.
Further experiments were designed to assess the possible interdependence(s) between activation loop phosphorylation and relief of autoinhibition. We first examined the effect of activation loop phosphorylation on the catalytic activity of the isolated PKD catalytic domain using a wild-type catalytic domain fusion protein, GST-Cat, and a catalytic domain fusion protein with Ser 744 and Ser 748 mutated to unphosphorylatable alanines, PKD-GST-Cat/S744A/S748A. As shown in the inset in Fig. 6A, control GST had no catalytic activity. In contrast, and consistent with previous results using a different PKD catalytic domain fusion protein (MBP-CAT) (13), bacterially expressed GST-Cat vigorously phosphorylated syntide-2. Interestingly, an equivalent amount of GST-Cat/S744A/S748A phosphorylated syntide-2 to only a slightly lesser extent (ϳ80% of that achieved by GST-Cat). Thus, even complete prevention of phosphorylation did not drastically affect the activity of the isolated catalytic domain, suggesting that activation loop phosphorylation is not needed for PKD activation in the truncated enzyme.
To assess the effect of preventing activation loop phosphorylation under circumstances in which PH domain-mediated regulation is removed, we generated a novel mutant derived from a PH domain deletion mutant (PKD⌬PH) we described previously (18). Whereas PKD⌬PH retains a wild-type activation loop sequence, the newly generated mutant has Ser 744 and Ser 748 mutated to alanine residues (PKD⌬PH/S744A/S748A). We first compared the activity of this mutant, by both syntide-2 phosphorylation (Fig. 6A) and autophosphorylation assays (Fig. 6B), to that of wild-type PKD, PKD-S744A/S748A, and PKD-⌬PH, from stimulated or unstimulated cells. PKD from unstimulated cells had very low activity, and PDB stimulation of cells dramatically activated the enzyme by both types of assay. PKD-S744A/S748A was not active either constitutively or after cell stimulation with PDB. In contrast, PKD⌬PH from unstimulated cells was dramatically active in comparison with the wild-type PKD, and PDB stimulation of cells did not further stimulate the activity of this mutant (Fig. 6, A and B). These results with PKD, PKD-S744A/S748A, and PKD⌬PH were consistent with those we described previously (18,50).
Interestingly, like PKD⌬PH, PKD⌬PH-S744A/S748A from unstimulated cells was also very active, indicating that activation loop mutation to alanines did not abrogate the function of the enzyme. Thus, this mutant displayed constitutive syntide-2 phosphorylation activity that was ϳ5-fold higher than wildtype PKD and about 60% of PKD⌬PH (Fig. 6A). Results of analysis by autophosphorylation assays was similar (Fig. 6B). These results are consistent with the notion that removal of the PH domain releases PKD from autoinhibition. Although in these experiments, expression levels of PKD⌬PH-S744A/ S748A were, on average, slightly less than (0.75-fold) those of PKD-⌬PH (Fig. 6C), it was not clear whether this difference could fully account for the measured differences in activity between these two enzymes. Thus, it remains possible that activation loop phosphorylation could influence the overall activity in the absence of the PH domain.
We next analyzed the activation loop phosphorylation in the cell lysates from cells transfected with the enzymes assayed in Fig. 6, A and B, using the phosphospecific antibodies described above. As shown in Fig. 6C, PKD was unphosphorylated in unstimulated cells, and stimulation with PDB induced a dramatic increase in phosphorylation at Ser 748 and Ser 744 , measured using pS748 and pS744, respectively. PKD-S744A/S748A could not be detected in either unstimulated or stimulated cells, as the phosphorylation sites responsible for antibody detection are altered by mutation.
Western analysis of PKD⌬PH using pS748 and pS744 indicated that whereas the basal phosphorylation of these residues in unstimulated cells was slightly increased in comparison with wild-type enzyme, significant increases (in the case of pS748) FIG. 6. Constitutive activity and activation loop phosphorylation status of PKD PH domain deletion mutants. COS-7 cells were transiently transfected with constructs encoding PKD, PKD-S744A/ S748A, PKD⌬PH, or PKD⌬PH-S744A/S748A or with vector pcDNA3 as control. After 72 h, cells were either left unstimulated (open bars) or stimulated with PDB (filled bars) as indicated and then lysed. PKD proteins were then either immunoprecipitated from lysates and assayed for autophosphorylation or syntide-2 phosphorylation activity or analyzed directly by SDS-PAGE and Western analysis, as described under "Experimental Procedures." A, PKD activity was analyzed by syntide-2 assays as described under "Experimental Procedures." Syntide-2 phosphorylation activity of GST-PKD catalytic domain fusion proteins (WT, GST-Cat; S744A S748A, GST-Cat-S744A/S748A) is presented in the inset. Data represent the means Ϯ S.E. from three separate experiments. B, PKD proteins (S744A/S748A, PKD-S744A/S748A; ⌬PH, PKD⌬PH; ⌬PH-S744A/S748A, PKD⌬PH-S744A/S748A) were subjected to in vitro kinase (IVK) autophosphorylation assays. A representative autoradiogram (above), together with compiled data, quantitated by scanning densitometry (below), are shown. Means Ϯ S.E. from three experiments are depicted in the graph. C, cell lysates from COS-7 cells transfected with PKD proteins and treated as described in A were subjected to Western analysis using the phosphospecific antibodies pS748 or pS744 to detect activation loop phosphorylation or the PKD antibody C-20 to detect total PKD protein, respectively, as described under "Experimental Procedures." Similar results were obtained in three separate experiments. or even very dramatic increases (in the case of pS744) in immunoreactivity were produced in the protein upon PDB stimulation of cells. These results were interesting taken in the light of those in Fig. 6, A and B, as they clearly indicate that the high constitutive activity of PKD⌬PH can be dissociated from quantitative phosphorylation at the activation loop, as this event was dramatically induced during cell stimulation but was not associated with a corresponding increase in enzyme activity. Taken together, these results strongly suggest that the function of activation loop phosphorylation is to reverse an autoinhibitory effect of the PH domain.

CONCLUDING REMARKS
Previous studies from this laboratory demonstrated that pharmacological inhibition of PKC prior to cell stimulation prevents PKD activation, and cotransfection of PKC⑀ or PKC, together with wild-type PKD, activates PKD in cells via dual phosphorylation of Ser 744 /Ser 748 . As these sites also become phosphorylated in kinase-deficient forms of PKD, we have proposed that PKD activation depends on PKC-mediated transphosphorylation rather than on PKD autophosphorylation (34,50,51).
Recent additional studies elucidated dynamic movements of PKD in response to G protein-coupled receptor stimulation (56 -59) by immunocytochemistry and real-time imaging. These studies demonstrated that PKD initially translocates to the plasma membrane, where the enzyme becomes activated, and then subsequently reverse translocates away from the plasma membrane, to gain access to targets elsewhere in the cell (56 -58), including the nucleus (59). Interestingly, PKC-dependent phosphorylation of activation loop Ser 744 and Ser 748 was required for this dynamic behavior of activated PKD (58,59). Specifically, preincubation with selective PKC inhibitors prior to cell stimulation with neuropeptides including bombesin prevented reverse translocation of PKD. Thus, the actions of PKC induce multiple facets of PKD function including its persistent activation and the process by which the activated enzyme is dispatched within the cell. But as yet, information as to whether Ser 744 /Ser 748 are directly phosphorylated by PKC or by an intermediate kinase(s) has been elusive (20,36,(51)(52)(53).
Here, we investigated the involvement of direct activation loop phosphorylation in PKC-mediated PKD activation, using in vitro phosphorylation assays with immunoprecipitated or purified recombinant PKC isoforms as upstream kinases and PKD activation loop peptides and intact PKD as substrates. Novel PKC⑀ and PKC directly phosphorylated a peptide derived from the PKD activation loop in vitro to levels approaching those of model peptides including peptide ⑀ and syntide-2. To assess phosphorylation of the individual serine residues, we generated peptides in which either site was selectively replaced with a negatively charged residue and examined second-site phosphorylation of these variant peptides. Results of these assays suggest that PKC could act as a second-site kinase to phosphorylate either Ser 744 and Ser 748 whereas PKC⑀ could preferentially act as a second-site kinase for PKD Ser 744 .
Our results also indicate that both PKC⑀ and PKC transphosphorylate Ser 744 and Ser 748 in intact proteins. We therefore conclude that PKC and PKC⑀ mediate PKD activation by direct phosphorylation of Ser 744 and/or Ser 748 in pathways of importance in vivo. Although PKD did not significantly phosphorylate an activation loop peptide, Western analysis using the phosphospecific antibodies revealed that the fulllength enzyme autophosphorylated, albeit slowly, at Ser 748 . Given the known substrate specificity of PKC⑀, and our results suggesting that PKC⑀ acts as a second-site kinase for PKD Ser 744 , it appears possible that initial PKD autophosphoryla-tion at Ser 748 can, in some circumstances, positively influence or prime Ser 744 for transphosphorylation by PKC⑀.
We also examined the mechanism whereby activation loop phosphorylation triggers PKD activation. Specifically, deletion of the PH domain renders the enzyme insensitive to activation loop phosphorylation. Furthermore, mutation of Ser 744 /Ser 748 to alanines did not abrogate the catalytic activity of PKD forms lacking either the PH domain or the entire regulatory domain. Collectively, these results suggest that the major part of PKCmediated activation derives from relief of autoinhibition mediated by the regulatory domain of PKD and that activation loop phosphorylation is a mechanistic trigger for this process. Our findings complement and extend our previous studies, establishing PKC⑀ as a direct upstream kinase for PKD in a PKC-PKD signaling cascade in which PKD mediates some of the critical signaling events controlled by the PKC family.