Identification of in Vivo Phosphorylation Sites Required for Protein Kinase D Activation*

Protein kinase D (PKD) is activated by phosphorylation in intact cells stimulated by phorbol esters, cell permeant diacylglycerols, bryostatin, neuropeptides, and growth factors, but the critical activating residues in PKD have not been identified. Here, we show that substitution of Ser744 and Ser748 with alanine (PKD-S744A/S748A) completely blocked PKD activation induced by phorbol-12,13-dibutyrate (PDB) treatment of intact cells as assessed by autophosphorylation and exogenous syntide-2 peptide substrate phosphorylation assays. Conversely, replacement of both serine residues with glutamic acid (PKD-S744E/S748E) markedly increased basal activity (7.5-fold increase compared with wild type PKD). PKD-S744E/S748E mutant was only slightly further stimulated by PDB treatment in vivo, suggesting that phosphorylation of these two sites induces maximal PKD activation. Two-dimensional tryptic phosphopeptide analysis obtained from PKD mutants immunoprecipitated from 32P-labeled transfected COS-7 cells showed that two major spots present in the PDB-stimulated wild type PKD or the kinase-dead PKD-D733A phosphopeptide maps completely disappeared in the kinase-deficient triple mutant PKD-D733A/S744E/S748E. Our results indicate that PKD is activated by phosphorylation of residues Ser744 and Ser748 and thus provide the first example of a non-RD kinase that is up-regulated by phosphorylation of serine/threonine residues within the activation loop.

The newly identified PKD is a mouse serine/threonine protein kinase with distinct structural and enzymological properties (9). The catalytic domain of PKD is distantly related to Ca 2ϩ -regulated kinases and shows little similarity to the highly conserved regions of the kinase subdomains of the PKC family (10). Consistent with this, PKD does not phosphorylate a variety of substrates utilized by PKCs, indicating that PKD is a protein kinase with distinct substrate specificity (9,11). In contrast to all known PKCs, including mammalian, Drosophila, and yeast isoforms (12), the NH 2 -terminal region of PKD contains a pleckstrin homology domain that regulates enzyme activity (13) and lacks a sequence with homology to a typical PKC autoinhibitory pseudosubstrate motif (9). However, the amino-terminal region of PKD contains a tandem repeat of cysteine-rich, zinc finger-like motifs that binds phorbol esters with high affinity (9). Immunopurified PKD is markedly stimulated by either biologically active phorbol esters or diacylglycerol, in the presence of phosphatidylserine (11,14). The human PKC (15), with 92% homology to PKD (extending to 98% homology in the catalytic domain), is also stimulated by phorbol esters and phospholipids (16). These in vitro results indicate that PKD/PKC is a novel phorbol ester/diacylglycerolstimulated kinase.
More recently, a second mechanism of PKD activation has been identified that involves PKD phosphorylation (17). Specifically, exposure of intact cells to phorbol esters, cell-permeant diacylglycerols, bryostatin, neuropeptide agonists, and growth factors induces rapid PKD phosphorylation and activation that is maintained during cell disruption and immunoprecipitation (13,14,(17)(18)(19). PKD activity recovered from cells stimulated with these agents can be measured by kinase assays in the absence of lipid activators. Several lines of evidence including the use of PKC inhibitors and cotransfection of PKD with constitutively active mutants of PKC⑀ and PKC indicate that PKD is activated by phosphorylation in living cells through a novel PKC-dependent signal transduction pathway (14,18,19). These results revealed an unsuspected connection between PKCs and PKD and implied that PKD can function downstream of PKCs in signal transduction, but the mechanisms involved remained unclear. In particular, the critical phosphorylation sites in PKD that are responsible for enzyme activation had not been identified.
In the present study, we used a combination of mutational analysis and two-dimensional tryptic phosphopeptide mapping to identify Ser 744 and Ser 748 as activating residues within the activation loop of the PKD catalytic domain. Most protein kinases activated by phosphorylation of residues in the activation loop are RD kinases (20), i.e. kinases in which the critical aspartic residue in the "catalytic loop" (in subdomain VI B) has an adjacent arginine. Interestingly, PKD contains a cysteine residue instead of an arginine immediately adjacent to the aspartic acid in the catalytic loop (10). Thus, PKD provides the first example of a non-RD kinase that is activated by phosphorylation of serine/threonine residues within the activation loop. serum at 37°C in a humidified atmosphere containing 10% CO 2 . The cells were transfected at 40 -60% confluence in serum-free medium by using 5 g of DNA from the various plasmids and 10 l of Lipofectin reagent (Life Technologies, Inc.) per 60-mm-diameter dish, as described previously (13). The cells were used for experimental purposes 72 h later.
Site-directed Mutagenesis-The site-specific mutations within the activation loop in the catalytic domain of PKD, resulting in single amino acid substitutions, were generated by overlap PCR using pBluescript SK(ϩ)-PKD cDNA as a template. Mutants were made by using an oligonucleotide annealing from nucleotide 1694 to 1719, close to the SphI site (located at nucleotide 1754), within PKD and a sequence downstream of PKD corresponding to pBluescript SK(ϩ) near the polylinker region as external forward (5Ј-CAG TGT TCT CCC CAG TGG CAT CGG TC-3Ј) and reverse (5Ј-CTC ACT AAA GGG AAC AAA AGC TGG AGC TC-3Ј) primers, together with internal reverse and forward primers complementary to each other and containing the different residue substitutions. The sense primers containing the sequence changes encoding the desired mutation were as follows: PKD-D733A, 5Ј-G GTG AAG CTC TGT GCT TTT GGT TTT GCC CG-3Ј; PKD-S744A, 5Ј-C ATT GGA GAG AAG GCT TTT AGG AGG TCA GTG G-3Ј; PKD-S744E, 5Ј-C ATT GGA GAG AAG GAG TTT AGG AGG TCA GTG G-3Ј; PKD-S748A, 5Ј-GAG AAG TCT TTT AGG AGG GCA GTG GTG GGT AC-3Ј; and PKD-S748E, 5Ј-GAG AAG TCT TTT AGG AGG GAA GTG GTG GGT AC-3Ј. After the second PCR reaction, the amplified fragment, cut with SphI and XbaI, was used to replace the original pBluescript SK(ϩ)-PKD SphI and XbaI segment. For double or triple mutants (PKD-S744A/S748A, PKD-S744E/S748E, and PKD-D733A/ S744E/S748E), we followed the same approach, using as template pBluescript SK(ϩ)-PKD mutants that already contained a single or a double mutation, respectively. Constructs were sequenced using an Applied Biosystems automated DNA sequencer before they were subcloned into the mammalian expression vector pcDNA3 (Invitrogen) for transient expression in COS-7 cells.
Elution of PKD from Immunocomplexes-The immunocomplexes were washed once with Buffer A, twice with Buffer B (Buffer A minus Triton X-100), and twice with kinase buffer (30 mM Tris-HCl, pH 7.6, 10 mM MgCl 2 , 1 mM dithiothreitol). PKD was eluted by incubating the immunoprecipitates with 0.75 mg/ml of the immunizing peptide in kinase buffer for 30 min at 4°C.
Kinase Assay-PKD autophosphorylation was determined in an in vitro kinase assay as described previously (14) . Briefly, the immunocomplexes were washed once with Buffer A, twice with Buffer B (Buffer A minus Triton X-100), and twice with kinase buffer (30 mM Tris-HCl, pH 7.6, 10 mM MgCl 2 , 1 mM dithiothreitol), and 20 l of PKD immune complexes were mixed with 20 l of kinase buffer containing a 100 M final concentration of [␥-32 P]ATP (specific activity of 400 -600 cpm/ pmol) for 10 min at 30°C. The reaction was then stopped by adding an equal volume of 2ϫ SDS-PAGE sample buffer (1 M Tris-HCl, pH 6.8, 0.1 mM Na 3 VO 4 , 6% SDS, 2 mM EDTA, pH 8, 4% 2-mercaptoethanol, 10% glycerol) and analyzed by SDS-PAGE and autoradiography. Autoradiograms were scanned, and images were processed by Adobe PhotoShop (Mountain View, CA) Exogenous substrate phosphorylation by immunoprecipitated wild type PKD or mutants was carried out under the same conditions as the in vitro kinase assay adding a final concentration of 2.5 mg/ml syntide-2. After 10 min at 30°C, the reaction was terminated by adding 100 l of 75 mM H 3 PO 4 and spotting 75 l of the supernatant on P-81 phosphocellulose paper. Free [␥-32 P]ATP was separated from the labeled substrate by washing four times (5 min each) in 75 mM H 3 PO 4 . The P-81 papers were dried, and radioactivity incorporated into syntide-2 was determined by Cerenkov counting. Syntide-2 phosphorylation by eluted PKD was performed by mixing 10 l of the eluted PKD protein with 10 l of the phosphorylation mixture containing 2.5 mg/ml syntide-2 in the presence or absence of PDB (200 nM) and PS (100 g/ml) micelles or dextran sulfate (30 g/ml) as activators. After 10 min at 30°C, the reaction was stopped and measured by Cerenkov counting as described above.
Western Blot Analysis-For Western blot analysis of either PKD or PKD mutants expressed in transfected COS-7 cells, 50 g of total protein from cell lysates were mixed with the same volume of 4ϫ SDS-PAGE sample buffer, boiled for 10 min, and analyzed by SDS-PAGE followed by transfer to Immobilon membranes at 100 V, 0.4 A at 4°C for 4 h using a Bio-Rad transfer apparatus. The transfer buffer composition was 200 mM glycine, 25 mM Tris, 0.01% SDS, and 20% CH 3 OH. Membranes were blocked for 1 h at room temperature in 5% nonfat dried milk in PBS, pH 7.2, and incubated for 3 h with PA-1 antiserum (1:500 dilution) in PBS containing 3% nonfat dried milk. Immunoreactive bands were visualized using either 125 I-labeled protein A (0.1 Ci/ml) and autoradiography or horseradish peroxidase-conjugated anti-rabbit IgG and subsequent enhanced chemiluminescence detection.
Phosphopeptide Mapping-Two-dimensional tryptic phosphopeptide mapping on thin-layer cellulose plates was performed according to standard procedures (21). Briefly, COS-7 cells were transiently transfected with wild type PKD or the different mutants. Three days following transfection, cells were washed twice and incubated at 37°C in phosphate-free Dulbecco's modified Eagle's medium for 30 min and then metabolically labeled with this medium containing 200 Ci/ml carrier-free 32 P i for 5 h. At the end of this labeling period, cells were either left untreated or stimulated with 200 nM PDB for 10 min as indicated in the figure legends and then lysed, immunoprecipitated, washed, and analyzed by SDS-PAGE and autoradiography. The 110-kDa bands corresponding to phosphorylated PKD or mutant proteins were excised from dried gels and processed for tryptic two-dimensional phosphopeptide mapping as described previously (19). Thin-layer electrophoresis in pH 1.9 buffer (2.2% formic acid and 7.8% acetic acid in water) was performed for 30 min at a constant current of 21 mA. The solvent system for the ascending chromatography consisted of n-butanol, pyridine, acetic acid, and water at a ratio 75:50:15:60. Thin-layer plates were air-dried and subjected to autoradiography to allow visualization of radiolabeled phosphopeptides. Autoradiograms were scanned and images were processed by Adobe PhotoShop (Mountain View, CA) Materials-[␥-32 P]ATP (370 MBq/ml), 32 P i (10 mCi/ml) 125 I-labeled protein A (15 mCi/ml), and enhanced chemiluminescence reagents were from Amersham International (Buckinghamshire, United Kingdom). PDB was obtained from Sigma. Protein A-agarose was from Boehringer Mannheim. Oligonucleotide primers and synthetic peptides were synthesized at the facilities of the Imperial Cancer Research Fund. Other items were from standard suppliers or as indicated in the text.

Comparison of the Putative Activation Loop of PKD with
That of Other Protein Kinases-Recent evidence indicates that PKD functions downstream of PKCs in a novel signal transduction pathway (14,(17)(18)(19). A critical aspect in the regulation of protein kinases that function in signaling cascades is the phosphorylation of activating residues located in a region spanning the highly conserved sequences DFG (in kinase subdomain VII) and APE (in kinase subdomain VIII) of the kinase catalytic domain termed the "activation loop" or "activation segment" (20,22). As a first step to determine whether the activation loop of PKD could be involved in the phosphorylationdependent regulation of this enzyme, we compared this region with the corresponding sequence of other protein kinases (Fig.  1). Surprisingly, the activation segment of PKD exhibits homology with the highly conserved activation loop of all known members of the MEK family from mammals, amphibians, insects, and yeast ( Fig. 1) (20,22,23). In particular, the spacing of the key residues Ser 217 and Ser 221 of MEK, which are phosphorylated by the proximal kinase Raf leading to MEK activation (23), is identical to that found for Ser 744 and Ser 748 in the activation segment of PKD. In contrast, this region of PKD differs from that of other signal-transducing protein kinases, including all members of the PKC family (Fig. 1). The homology between the activation loops of PKD and members of the MEK family suggested that Ser 744 and Ser 748 could play a critical role in PKD activation. To explore this hypothesis, we mutated these residues to either alanine, to prevent phosphorylation, or glutamic acid, to mimic phosphorylation, and we also examined whether these residues are phosphorylated in vivo under conditions that promote PKD activation.
Substitution of Ser 744 and Ser 748 for Alanine Prevents PKD Activation-If Ser 744 and Ser 748 are critical target sites for activating phosphorylation events, their conversion to Ala should reduce or eliminate PDB-mediated activation of PKD. To test this possibility, we generated PKD mutants with single or double substitutions of these residues cloned in the expression vector pcDNA3 (i.e. PKD-S744A, PKD-S748A or PKD-S744A/S748A). COS-7 cells transiently transfected with pcDNA3 containing wild type PKD or the PKD mutants were treated with or without 200 nM PDB for 10 min and lysed, and the extracts were immunoprecipitated with the PA-1 antibody. The immune complexes were incubated with [␥-32 P]ATP to measure PKD kinase activity by either autophosphorylation or exogenous substrate phosphorylation of the synthetic peptide syntide-2 (14,24,25). As shown in Fig. 2B, and in agreement with previous results (13,14,18,19) , PKD isolated from unstimulated cells had low catalytic activity that was markedly activated by PDB stimulation of intact cells (ϳ10-fold increase).
In striking contrast, substitution of both Ser 744 and Ser 748 for Ala in PKD completely blocked PDB-induced kinase activation because the kinase activity detected in immunoprecipitates from cells transfected with PKD-S744A/S748A was comparable to that observed in cells transfected with the expression vector pcDNA3 alone (Fig. 2B). PDB also failed to induce kinase activity of PKD-S744A/S748A when added at various concentrations (2 nM to 2 M) or for various times (2.5-30 min; results not shown). Single substitutions of either Ser 744 or Ser 748 for Ala resulted in PKD mutants that displayed reduced activity after PDB stimulation (50% decrease in both single Ala mutants compared with PDB-stimulated PKD). In all cases, the protein expression levels of the transfected PKD mutants were comparable to that of wild type PKD, as shown by Western blot analysis ( Fig. 2A). Thus, substitution of Ser 744 and Ser 748 in the activation loop of PKD by neutral nonphosphorylatable residues prevents the activation of this enzyme by PDB in vivo.
To examine whether mutation of Ser 744 and Ser 748 causes irreversible inactivation of PKD or selectively prevents PKCdependent PKD activation in vivo, we determined the syntide-2 kinase activity of PKD, PKD-S744A, PKD-S748A, and PKD-S744A/S748A eluted from PA-1 immunoprecipitates and measured in the absence or presence of the cofactors PS and PDB (11,14) or the potent PKD activator dextran sulfate (26). As shown in Fig. 3A, the catalytic activity of either wild type PKD or the PKD mutants was stimulated by addition of either PS and PDB or dextran sulfate to the incubation mixture. Thus, substitution of Ser 744 and Ser 748 for Ala selectively prevents PKD activation in vivo via a PKC-dependent pathway but does Substitution of Ser 744 and Ser 748 for Glutamic Acid Activates PKD-Some protein kinases that are activated by phosphorylation in the activation loop including MEK can be rendered constitutively active by substitution of the phosphorylated residue(s) for glutamic acid (27). To investigate further whether Ser 744 and Ser 748 of PKD are targets for activating phosphorylations, we mutated these residues to Glu in order to partially mimic phosphorylated residues and examined whether these substitutions were sufficient to increase PKD basal activity. COS-7 cells transiently transfected with PKD and PKD-S744E/ S748E were incubated in the absence or presence of 200 nM PDB for 10 min, and kinase activity was determined after immunoprecipitation by measuring either syntide-2 phosphorylation or autophosphorylation. The expression of this mutant was verified by Western blotting (Fig. 2A). As shown in Fig. 2B, PKD-S744E/S748E exhibited a high level of basal catalytic activity (ϳ7.5-fold increase compared with unstimulated PKD), which was only slightly further enhanced by PDB stimulation of intact cells. These results support the hypothesis that Ser 744 and Ser 748 are activating residues in the catalytic domain of PKD.
Ser 744 and Ser 748 Are Phosphorylated in Intact Cells-In order to complement the mutational analysis presented in Figs. 2 and 3, we next examined PKD phosphorylation in intact cells using two-dimensional tryptic phosphopeptide mapping. The wild type PKD was immunoprecipitated from lysates of transiently transfected COS-7 cells labeled with 32 P i and incubated with or without 200 nM PDB for 10 min. The immune complexes were analyzed by SDS-PAGE and autoradiography. The bands of 110 kDa corresponding to PKD were excised from the gel and further analyzed by two-dimensional tryptic phosphopeptide mapping. Tryptic peptides derived from unstimulated wild type PKD resolved into six major phosphopeptides (Fig. 4A, WT), a, b, and c being the most abundant. PDB stimulation induced the appearance of at least three new spots (Fig. 4B, WTϩPDB, labeled as D, E, and F). Next, we analyzed the tryptic phosphopeptide maps of PKD-S744A/S748A. The map of PKD-S744A/S748A isolated from unstimulated cells contained the same spots as wild type PKD (Fig. 4C,  S744S748A).In contrast, the tryptic map of PKD-S744A/ S748A isolated from PDB-stimulated cells did not show the presence of spots D and E (Fig. 4D, S744S748AϩPDB).
In order to examine whether the new D and E phosphopeptides appearing in activated PKD maps resulted from autophosphorylation or transphosphorylation, we generated a kinase-dead PKD in which the functionally critical aspartic acid at position 733, in the DFG motif of the catalytic domain of PKD, was mutated to alanine (PKD-D733A). The PKD-D733A mutant exhibited no kinase activity either in autophosphorylation or in exogenous substrate phosphorylation assays and functioned as a dominant negative suppressing the PDB-induced activation of endogenous PKD (Fig. 3B). Two-dimensional tryptic phosphopeptide mapping of PKD-D733A isolated from unstimulated COS-7 cells showed that a major phosphopeptide (called a in wild type-PKD or PKD-S744A/S748A maps) was not present, suggesting that peptide a is generated by autophosphorylation (Fig. 5A, D733A, versus Fig. 4A). However, PDB stimulation induced the generation of the three new spots that migrated at positions equivalent to D, E, and F, indicating that these phosphopeptides arise from transphosphorylation rather than autophosphorylation (Fig. 5B,  D733AϩPDB).
To examine the contribution of Ser 744 and Ser 748 to the transphosphorylation of phosphopeptides D, E, and F induced by PDB stimulation, we generated a triple PKD mutant in which Asp 733 was substituted for Ala in PKD-S744E/S748E to produce the kinase-deficient triple mutant PKD-D733A/S744E/ S748E. The expression of this mutant in COS-7 cells was comparable to that of PKD-D733A, as shown by Western blot analysis (Fig. 3C). Mutation of Ser 744 and Ser 748 in the kinasedead mutant did not affect the pattern of the tryptic phosphopeptide map originated under unstimulated conditions (Fig. 5C, D733/S744S748E, versus Fig. 5A, D733A) but blocked the appearance of phosphopeptides D and E in response to PDB (Fig. 5D, D733/S744S748EϩPDB versus Fig. 5B,  D733AϩPDB). These results indicate that Ser 744 and Ser 748 are phosphorylated in vivo in response to PDB stimulation.
Concluding Remarks-Recently, we demonstrated that PKD is rapidly phosphorylated and activated in a variety of cell types in response to phorbol esters, cell-permeant diacylglycerols, bryostatin, neuropeptide agonists, and growth factors (13,14,(17)(18)(19). The experiments presented here were designed to identify the sites responsible for phosphorylation-dependent activation of PKD. Using a combination of mutational analysis and two-dimensional peptide mapping of in vivo labeled wild type and mutant PKDs, we identified Ser 744 and Ser 748 as activating residues in the activation loop of PKD and demonstrated that these residues are phosphorylated in intact cells in response to PDB stimulation.
Our results reveal a similarity between the mechanism of activation of PKD and that of other kinases, such as MEK (23), that are activated by phosphorylation of Ser or Thr residues located in the activation loop. However, there is also a fundamental difference. MEK and other protein kinases activated by phosphorylation of residues in the activation loop are RD kinases (20), i.e. kinases in which the critical aspartic residue in the catalytic loop, located in subdomain VI B, has an adjacent arginine. It has been hypothesized that phosphorylated residues in the activation loop form an ionic bridge with Arg of the catalytic loop, thereby stabilizing the catalytic active structure of the enzymes (20). A salient feature of the catalytic kinase domain of PKD is the presence of a cysteine residue instead of an arginine immediately adjacent to the aspartic acid in the catalytic loop (10). Therefore, our results provide the first example of a non-RD kinase that is activated by phosphorylation in the activation loop and thus could have important implications for understanding the molecular events that lead to phosphorylation-dependent activation of regulatory protein kinases.