Phosphatidic Acid-mediated Phosphorylation of the NADPH Oxidase Component p47-phox

Phosphatidic acid (PA), generated by phospholipase D activation, has been linked to the activation of the neutrophil respiratory burst enzyme, NADPH oxidase; however, the intracellular enzyme targets for PA remain unclear. We have recently shown (McPhail, L. C., Qualliotine-Mann, D., and Waite, K. A. (1995)Proc. Natl. Acad. Sci. U. S. A. 92, 7931–7935) that a PA-activated protein kinase is involved in the activation of NADPH oxidase in a cell-free system. This protein kinase phosphorylates numerous endogenous proteins, including p47-phox, a component of the NADPH oxidase complex. Phospholipids other than PA were less effective at inducing endogenous protein phosphorylation. Several of these endogenous substrates were also phosphorylated during stimulation of intact cells by opsonized zymosan, an agonist that induces phospholipase D activation. We sought to identify the PA-activated protein kinase that phosphorylates p47-phox. The PA-dependent protein kinase was shown to be cytosolic.cis-Unsaturated fatty acids were poor inducers of protein kinase activity, suggesting that the PA-activated protein kinase is not a fatty acid-regulated protein kinase (e.g. protein kinase N). Chromatographic techniques separated the PA-activated protein kinase from a number of other protein kinases known to be activated by PA or to phosphorylate p47-phox. These included isoforms of protein kinase C, p21 (Cdc42/Rac)-activated protein kinase, and mitogen-activated protein kinase. Gel filtration chromatography indicated that the protein kinase has an apparent molecular size of 125 kDa. Screening of cytosolic fractions from several cell types and rat brain suggested the enzyme has widespread cell and tissue distribution. The partially purified protein kinase was sensitive to the same protein kinase inhibitors that diminished NADPH oxidase activation and was independent of guanosine 5′-3-O-(thio)triphosphate and Ca2+. Phosphoamino acid analysis showed that serine and tyrosine residues were phosphorylated on p47-phox by this kinase(s). These data indicate that one or more potentially novel protein kinases are targets for PA in neutrophils and other cell types. Furthermore, a PA-activated protein kinase is likely to be an important regulator of the neutrophil respiratory burst by phosphorylation of the NADPH oxidase component p47-phox.

Phospholipase D (PLD) 1 is activated in a variety of cells by hormones and growth factors (1)(2)(3). This activation results in the generation of phosphatidic acid (PA), which can be further metabolized by PA phosphohydrolase to diacylglycerol (DG). The generation of PA by PLD in neutrophils has been linked, by us and others, to the activation of the respiratory burst enzyme, NADPH oxidase (2)(3)(4)(5)(6)(7). The enzymes involved in the activation of the oxidase, which are downstream of the generation of PA, have not been identified. We have developed a cell-free activation system for NADPH oxidase, which utilizes PA and is phosphorylation-dependent (7,8). The system requires the presence of both cellular membranes and cytosol, as well as the presence of DG. Since a variety of potential enzyme targets for PA are present, this system provides a means to identify these enzymes and their role in activation of the NADPH oxidase. We have recently shown that, during phosphorylation-dependent oxidase activation, PA, but not DG, induces phosphorylation of a wide range of proteins, in which the most prominent protein phosphorylated is the NADPH oxidase component p47-phox (phagocytic oxidase component) (8). This indicates that potential targets for PA, involved in NADPH oxidase activation, are protein kinases.
Rho-GDp dissociation inhibitor (32,33). In fact, the production of PA from PLD activation has been linked to the activation of PAK isoforms in neutrophils (34,35). Thus, while MAPK and PAK isoforms are not known to be activated directly by PA, they could be activated downstream of other protein targets of PA and ultimately be responsible for the phosphorylation of p47-phox.
We examined whether the PA-dependent protein kinase activity in human neutrophils was due to the activation of one of these known protein kinases. Our results indicate that this is not the case and that a potentially novel protein kinase, which phosphorylates p47-phox, may be the target for PA in this system. This protein kinase may therefore play a role in the activation of the NADPH oxidase.

Isolation of Blood Cells and Subcellular Fractionation
Platelets were derived from heparinized venous blood by centrifugation (300 ϫ g, 20 min, 25°C) (39). The supernatant, containing plateletrich plasma, was then centrifuged (2500 ϫ g, 15 min, 4°C) (39), and the pellet was resuspended in Buffer A (50 mM Na x PO 4 , (pH 7.0), 11% sucrose, 130 mM NaCl, 1 mM EGTA, and 0.5 mM phenylmethylsulfonyl fluoride) and processed as described below. Neutrophils, monocytes, and lymphocytes were isolated from heparinized venous blood obtained from consenting human donors by dextran sedimentation followed by Isolymph (Gallard-Schlesinger Ind., Inc., Carle Place, NY) centrifugation (40). After the Isolymph centrifugation, the interface, containing monocytes and lymphocytes, was washed with RPMI (Life Technologies) and plated (41). Once the monocytes had adhered, the nonadherent cells (lymphocytes) were collected in Buffer A at 1 ϫ 10 8 cells/ml. The monocytes were scraped off the plates and resuspended at 1 ϫ 10 8 cells/ml in Buffer A (41). Monocytes and lymphocytes were then processed as described below. The pellet after Isolymph centrifugation, containing neutrophils, was cleared of contaminating red blood cells by hypotonic lysis (42,43). Isolated neutrophils were suspended to a concentration of 1 ϫ 10 8 cells/ml in Buffer A. All cells to be analyzed were sonicated to ϳ90% breakage (visualized by microscopy) and processed as follows. To prepare cytosolic and membrane fractions, sonicates were freed of unbroken cells and nuclei by centrifugation (200 ϫ g, 10 min), loaded onto a 15-40% discontinuous sucrose gradient at a 2:1:1 (v/v/v) ratio and centrifuged at 150,000 ϫ g (SW50 rotor, 30 min) (40). Cytosolic fractions were collected from the top layer down to the 15% interface, and membrane fractions were collected from the 15-40% interface and the 40% sucrose layer. To prepare cytosolic fractions only, sonicates were centrifuged at 150,000 ϫ g (Type 50 rotor, 90 min), and the resulting supernatant was collected. Fractions were stored at Ϫ70°C. Protein was determined using the Coomassie Plus Protein protocol from Pierce, which is based on the Bradford method (44). Bovine serum albumin was used as a standard.
Cell-free Assay: Analysis by SDS-PAGE-Reaction mixtures (150 l total) contained 50 mM Na x PO 4 (pH 7.0), 1 mM EGTA, 5 mM MgCl 2 (7,8) plus one of the following: a combination of the cytosolic and membrane fractions (25:1 protein:protein ratio), the cytosolic fraction only, the membrane fraction only, partially purified enzyme, or column fractions. GST-p47-phox (1 g/reaction) or recombinant p47-phox (Rp47, 1 g/reaction) were added to the mixtures as indicated in the figure legends. Six l of [␥-32 P]ATP were added, followed by the immediate addition of activator, as indicated in the figure legends. This mixture was allowed to incubate at 25°C for 45 min. The reaction was then terminated by the addition of Laemmli sample buffer and prepared for SDS-PAGE analysis (48). Proteins were separated by 8 -15% SDS-PAGE (40), silver-stained (49), dried, and analyzed via autoradiography. Protein kinase inhibitors were added 5 min prior to the addition of [␥-32 P]ATP. For samples that were analyzed by Western blotting, reactions were performed as above with the following modifications. Cold ATP (10 M) was used in place of radiolabeled ATP. Proteins were separated by 9% SDS-PAGE and then transferred to polyvinylidene difluoride (50).
Cell-free Assay: Analysis by Phosphocellulose Binding-Reaction mixtures were the same as above, except that [␥-32 P]ATP, diluted to yield 2 ϫ 10 6 cpm/reaction, was used (final ATP concentration is indicated in figure legends), and samples were allowed to preincubate for 5 min at 25°C before the addition of sample material to be tested (51). After 30 min, 75% (112.5 l) of the reaction mixture was transferred to Whatman P-81 phosphocellulose squares; the squares were washed with 75 mM phosphoric acid and quantitated by Cerenkov counting (52,53). 1 M GTP␥S or 0.6 mM CaCl 2 were added to samples as indicated under "Results." For the mixed lipid vesicle experiments, PA, PS, PE, and PC were prepared separately by sonication (7). PE, PC, and PS were then mixed together (final assay concentration: 25, 25, and 10 M, respectively) with PA (final assay concentration: 0 -300 M). These mixtures were then used in the place of PA in the reaction mixtures. All samples were assayed in duplicate. Conditions used were linear with respect to protein concentration and time of incubation.

Saturated Ammonium Sulfate Precipitation
Cytosol (2 mg of protein) was cleared by centrifugation at 10,000 ϫ g and then adjusted with a solution of saturated ammonium sulfate at 4°C to yield 40% ammonium sulfate saturation. This mixture was kept on ice for 10 min and was centrifuged (10,000 ϫ g, 4°C, 10 min) to collect precipitated protein. The resulting supernatant was adjusted with saturated ammonium sulfate to yield 70% saturation and was kept on ice for 10 min, and the precipitate was collected by centrifugation. Both the 40 and 70% protein precipitates were resuspended in 500 l of Buffer B (Buffer A plus 2 mM dithiothreitol). These samples, plus the resulting 70% supernatant, were cleared of residual ammonium sulfate by passage over a Sephadex G-25 column (90 ϫ 15 mm) in Buffer B. The void volume fraction was collected for each sample and assayed for PA-dependent protein kinase activity using the phosphocellulose binding assay. The ammonium sulfate-free samples were also concentrated by centrifugation with Centricon-30 tubes (Amicon; Beverley, MA) and stored at Ϫ70°C.

Hydrophobic Interaction Chromatography
The 40% ammonium sulfate protein precipitate obtained from 25 mg of cytosolic protein (containing 85% of recovered activity) was resuspended on ice in Buffer B containing 20% ammonium sulfate. After 10 min, the solution was cleared by centrifugation and applied to a Ranin Hydropore column (Woburn, MA), equilibrated with Buffer B containing 10 g/ml leupeptin, 1 g/ml pepstatin A, and 20% ammonium sulfate (Buffer C). The unretained proteins were collected, and the retained proteins were then eluted with Buffer C containing no ammonium sulfate. Fractions containing the retained and unretained proteins were concentrated via centrifugation using Centricon 30 tubes and were assayed for PA-dependent protein kinase activity, using the phosphocellulose binding assay, and for the presence of various protein kinases by Western blot analysis.

Gel Filtration Chromatography
Neutrophil cytosol (ϳ120 mg of protein) was adjusted with a solution of saturated ammonium sulfate at 4°C to yield 40% ammonium sulfate. This mixture was kept on ice for 10 min and was centrifuged (10,000 ϫ g, 4°C, 10 min) to pellet precipitated protein. The precipitate was resuspended in Buffer B containing 20% ammonium sulfate. After 10 min on ice, the solution was cleared by centrifugation and applied to a Pharmacia Sephacryl S-200 column equilibrated with Buffer C containing no ammonium sulfate. Fractions (4.2 ml) were collected and assayed for PA-dependent protein kinase activity by the phosphocellulose assay. In a separate experiment, neutrophil cytosol (ϳ240 mg of protein) was subjected to 40% saturated ammonium sulfate precipitation. The resulting precipitate was resuspended in Buffer C, and the solution was cleared by centrifugation. The cleared supernatant was then subjected to hydrophobic interaction chromatography as described above. The bound proteins were eluted, concentrated by centrifugation with Centricon-10 tubes, and then applied to the Sephacryl S-200 column, as described above. Fractions were assayed for activity using the SDS-PAGE phosphorylation system. Molecular mass standards (thyroglobulin ϭ void, B-amylase ϭ 200 kDa, alcohol dehydrogenase ϭ 150 kDa, bovine serum albumin ϭ 66 kDa, carbonic anhydrase ϭ 29 kDa, and cytochrome C ϭ 12.4 kDa) were loaded and eluted separately for calibration of the column.

Phosphoamino Acid Analysis
GST-p47-phox was phosphorylated in the presence of [␥-32 P]ATP, separated by SDS-PAGE as described above, and transferred to polyvinylidene difluoride (50). The polyvinylidene difluoride membrane was then washed with H 2 O and subjected to autoradiography, and the phosphorylated GST-p47-phox band was excised from the membrane. The polyvinylidene difluoride membrane was then subjected to hydrolysis with 5.7 N HCl, at 110°C for 1 h as described previously (54). Hydrolysis was stopped by the addition of 200 l of dH 2 O, and samples were centrifuged. The supernatant was then lyophilized, suspended in phosphoamino acid thin layer chromatography analysis buffer (62.5% isobutyric acid, 1.9% n-butyl alcohol, 4.8% pyridine, 2.9% glacial acetic acid) containing 1 mg/ml phosphoamino acid standards (55), and spotted on cellulose plates. The plates were then subjected to thin layer chromatography using the phosphoamino acid buffer (55). The plates were allowed to air dry, standards were visualized with ninhydrin (55), and radiolabeled phosphoamino acids were detected by autoradiography.

Immunodepletion of Raf-1 or B-Raf from Neutrophil Cytosol
Cytosol (250 l) was diluted to 1 ml with 50 mM Tris (pH 7.0), 150 mM NaCl, 2 mM EDTA, and 1 mM EGTA, and incubated with 100 l of Protein A-Sepharose beads for 10 min, followed by removal of the beads. 10 g/ml of either anti-Raf-1 antibody, anti B-Raf antibody, or control IgG was added to 330 l of diluted, precleared cytosol, and the mixture was incubated at 4°C for 1 h. Protein A-Sepharose beads (100 l) were added again, and each mixture was incubated, with shaking, at 4°C for 1 h. The beads were pelleted by centrifugation, and the supernatants were collected and assayed for PA-dependent protein kinase activity by the phosphocellulose binding assay. Immunoprecipitates and the resulting supernatants were analyzed by SDS-PAGE, followed by Western blotting with antibodies to Raf-1 or B-Raf.

Preparation of Opsonized Zymosan
Zymosan was suspended in 154 mM NaCl, boiled for 15 min, centrifuged (10 min, 300 ϫ g, 4°C), washed once, and resuspended in phosphate-buffered saline. Two volumes of pooled human serum, pooled from four or five healthy donors, and one volume of zymosan suspension were incubated for 30 min at 37°C (56). The opsonized zymosan was then centrifuged, washed once in stimulation buffer, and resuspended in stimulation buffer for use.

RESULTS
Characterization of PA-induced Protein Phosphorylation-We have shown previously that the addition of PA to a mixture of cytosolic and membrane fractions from neutrophils results in the phosphorylation of a wide range of neutrophil proteins, including p47-phox (8). We investigated the ability of other phospholipids to induce protein phosphorylation in this system. Results are shown in Fig. 1. PC, PE, or cardiolipin (CL) did not induce protein phosphorylation above basal levels (H 2 O). Since CL was reported to activate a PA-sensitive protein kinase (19), we also examined a range of concentrations of this lipid (3 M to 1 mM). No significant phosphorylation occurred at any of the CL concentrations tested (data not shown). PS and phosphatidylinositol each induced a moderate level of protein phosphorylation. While phosphatidylglycerol induced protein phosphorylation to a higher degree than PS or phosphatidylinositol, the best activator of protein phosphorylation was PA. Indeed, the addition of PA resulted in several proteins being phosphorylated that were not phosphorylated in the presence of the other phospholipids (Fig. 1, arrows). This suggests that PA activates a protein kinase(s) not responsive to other phospholipids.
We next compared the phosphorylation pattern of cytosol stimulated with PA in vitro with cytosol isolated from opsonized zymosan-stimulated intact cells. Opsonized zymosan stimulation has been shown to induce activation of PLD, and this activation has been linked to the activation of NADPH oxidase (46,57). Increases in protein phosphorylation were harder to detect in the in vivo system due to high levels of labeled phosphate incorporated in nonstimulated cells. However, increased protein phosphorylation in stimulated samples could be observed upon longer autoradiographic exposures (Fig. 2). This indicates that a protein kinase is probably acti-vated by stimulation of neutrophils with opsonized zymosan. Several proteins appeared to be phosphorylated in both the intact cell and the in vitro system (Fig. 2, arrows). In both systems, p47-phox was phosphorylated upon stimulation. This suggests that the protein kinase activated by PA in the in vitro system is also activated in the intact cell, under conditions that result in PLD activation. Taken together, these data indicate that PA is the most effective activator of protein phosphorylation in the neutrophil cell-free system and suggest that the protein kinase activated in the cell-free system is also activated under physiological conditions. Subcellular Location of the PA-activated Protein Kinase-To determine the subcellular location of the protein kinase activated by PA, we used recombinant p47-phox (Rp47) as an exogenous substrate for the enzyme. Rp47 was added in excess to phosphorylation reaction mixtures that contained either cytosol and membranes, cytosol alone, or membranes alone. Phosphorylation of Rp47 was dependent upon the presence of PA (Fig. 3). Reaction mixtures that lacked cytosol did not result in PA-dependent phosphorylation of Rp47. In contrast, reaction mixtures that lacked the membrane fraction supported PA-dependent phosphorylation. Shown in Fig. 3 is a 5-min autoradiograph. In lanes with cytosol present, protein phosphorylation patterns similar to those seen in Fig. 1 (PA lane) were observed when longer exposures were viewed. Minimal protein phosphorylation was observed in longer exposures when reaction mixtures contained only membrane fractions (data not shown). The identical pattern was observed when the GST-p47phox fusion protein was used as the exogenous substrate (data not shown). These data indicate that the PA-activated protein kinase is cytosolic and does not require membrane components for activation.
Comparison of Protein Kinase Activation by PA and cis-Unsaturated Fatty Acids-Rat liver (19,20) and platelets (18) contain cytosolic PA-responsive protein kinases, which were identified as cardiolipin-activated and cis-unsaturated fatty acid-activated protein kinases, respectively. Recently, the hepatocyte protein kinase has been determined to be a novel protein kinase (PKN) (16), which is also the PKC-related protein kinase 1 (PRK1) (14,17,58). We tested the possibility that the protein kinase activity we observe in the neutrophil is the same or similar to these PA-responsive protein kinases. The results shown in Fig. 1 suggest the neutrophil enzyme is not a cardiolipin-activated protein kinase. We tested whether the cisunsaturated fatty acid AA could activate a protein kinase in neutrophil cytosol. Fig. 4 shows protein kinase activation as a function of PA and AA concentration, using the phosphocellulose assay. AA was a poor inducer of protein kinase activity with maximal activity of 40 Ϯ 3 pmol of PO 4 /min/mg (25 M AA, Fig. 4). In contrast, PA activated a protein kinase(s) in a dosedependent manner with a 6-fold higher maximal activity (246 Ϯ 44 pmol of PO 4 /min/mg, 30 M PA, Fig. 4). Another cis-unsaturated fatty acid, oleic acid, also elicited little protein kinase activation (35 pmol of PO 4 /min/mg, 25 M oleic acid, n ϭ 2, data not shown). Thus, the PA-responsive protein kinase found in neutrophil cytosol is poorly activated by cis-unsaturated fatty acids or cardiolipin and does not appear to be a fatty acid-activated protein kinase described by others (14, 16 -20).
Separation of the PA-activated Protein Kinase from PKC Isoforms-PA has been shown, in vitro, to activate several isoforms of PKC (9 -13), raising the possibility that the protein kinase target of PA in neutrophil cytosol is one or more of these enzymes. To test this possibility, we determined whether the PKC isoforms present in neutrophil cytosol (59 -64) co-purified with the PA-activated protein kinase. We first subjected neutrophil cytosol to sequential ammonium sulfate precipitation, as described under "Experimental Procedures." The fractions obtained were tested for the location of the PA-activated protein kinase, using the phosphocellulose protein kinase assay. The PA-activated protein kinase was recovered primarily in the 40% ammonium sulfate precipitate, which contained 85% of the total recovered PA-activated protein kinase activity (Fig. 5A). Much less activity was observed in the 70% precipitate and the resulting supernatant (13 and 2% of the total recovered activity, respectively). The same fractions were then evaluated for the presence of various PKC isoforms by Western blotting (Fig.  5B). The PKC isoforms ␣, ␤ ⌸ , ␦, and were present in unfrac- tionated cytosol that had been passed over a Sephadex G-25 column as a control for the ammonium sulfate-treated fractions (G25). Following ammonium sulfate fractionation of the cytosol and Sephadex G-25 chromatography of the fractions, the calcium-dependent PKC isoforms, ␣ and ␤ ⌸, were recovered in the 70% ammonium sulfate precipitate. Similarly, the calciumindependent isoforms, ␦ and , were recovered predominantly in the 70% ammonium sulfate precipitate, although small amounts of PKC-␦ and -were present in the 40% ammonium sulfate precipitate.
Since PKC-␦ and -remained in the 40% ammonium sulfate precipitate, we subjected the PA-activated protein kinase to an additional step of purification. The 40% ammonium sulfate pellet was resuspended in buffer containing 20% ammonium sulfate and subjected to hydrophobic interaction chromatography using a Rainin Hydropore column. The unretained proteins (Void) and the retained proteins (Bound) were collected as described under "Experimental Procedures." Activity measurements (Fig. 6A) showed that the PA-activated protein kinase was predominantly present in the retained protein fraction (Bound, 75% of the recovered PA-activated protein kinase activity). The unretained protein and the retained protein fractions were then subjected to Western blot analysis for the presence of PKC-␦ and -. Results (Fig. 6B) show that neither PKC-␦ nor -were retained on the column; these PKC isoforms were observed solely in the unretained protein fraction (Void). The retained protein fraction (Bound) was also analyzed, by Western blotting, for the presence of PKC isoforms not known to be present in the neutrophil (␥, , ⑀, ), to determine if one of these isoforms was present. No immunoreactive bands were observed (data not shown). These data indicate that the PAactivated protein kinase is not likely to be a PKC isoform.
Separation of the PA-activated Protein Kinase from PAK and MAPK Isoforms-Neutrophil cytosol contains MAPK and PAK isoforms, two classes of protein kinases recently shown to phosphorylate p47-phox, the in vitro substrate for the PA-activated protein kinase (24,25,29,31). While neither of these protein kinases are known to be directly activated by PA, it was possible that these protein kinases are targets for PA in our system or are activated downstream of the PA-activated protein kinase. Neutrophil cytosol depleted of Raf-1 or B-Raf by nondenaturing immunoprecipitation still maintained PA-dependent protein kinase activity (data not shown), suggesting that neither Raf-1 nor B-Raf is the target for PA in this system. This also suggests that MAPK isoforms are not activated via PA-mediated activation of Raf-1. However, it was possible that MAPK isoforms may be activated directly by PA. Fractions obtained by ammonium sulfate precipitation and hydrophobic interaction chromatography, during the partial purification of the PA-activated protein kinase, were analyzed for the pres-ence of MAPK or PAK by Western blotting. As presented in Figs. 5A and 6A, the PA-activated protein kinase was precipitated from a 40% saturated ammonium sulfate solution (40% cut ϭ 85% of recovered activity) and was retained by the column resin during hydrophobic interaction chromatography (Bound ϭ 75% of recovered activity). Fig. 7 shows that both PAK-2 and p42 and p44 MAPK isoforms are present in neutrophil cytosol. Upon 40% saturated ammonium sulfate precipitation, both kinase families distributed between the 40% cut and the resulting supernatant, although MAPK was predominantly in the 40% cut. In contrast to the PA-dependent protein kinase, both PAK and MAPK isoforms were found only in the unretained protein fraction eluting from the hydropore column. Similar results to those obtained with PAK-2 antibodies were obtained with antibodies recognizing PAK-1 (data not shown). Thus, the PA-activated protein kinase co-purifies with neither PAK-1 or PAK-2 nor p42 or p44 MAPK isoforms.
Molecular Weight Determination of the PA-activated Protein Kinase-The apparent molecular weight of the PA-dependent protein kinase was determined by gel filtration chromatography. Neutrophil cytosol was first subjected to ammonium sulfate precipitation and then gel filtration chromatography, as described under "Experimental Procedures." The peak of PAdependent protein kinase activity eluted at fraction 28 (Fig.  8A), which corresponds to a molecular mass of 125 kDa. We confirmed these results with more purified material obtained by subjecting neutrophil cytosol to both ammonium sulfate precipitation and hydrophobic interaction chromatography prior to gel filtration chromatography. Fractions were analyzed by the SDS-PAGE phosphorylation assay. Fig. 8B shows that, like the activity seen in Fig. 8A, the peak of GST-p47-phox phosphorylation occurred at fraction 28. These data indicate that the PA-responsive protein kinase has a maximum size of 125 kDa.
Characterization of the PA-activated Protein Kinase-Due to the low yield (4%) after gel filtration, we used the hydroporebound material to characterize the PA-activated protein kinase. We first analyzed the effect of calcium and GTP␥S on the activation of the PA-activated protein kinase. We found that calcium slightly enhanced the PA-activated protein kinase (1571 pmol of PO 4 /min/mg in the absence of calcium, 1780 pmol PO 4 /min/mg in the presence of 0.6 mM calcium, n ϭ 2). The addition of 1 M GTP␥S, sufficient to activate Rac in the NADPH oxidase system (7,8,65,66), had little effect on activation (1680 pmol of PO 4 /min/mg activity in the absence of GTP␥S, 1613 pmol PO 4 /min/mg in the presence of GTP␥S, n ϭ 2). We also verified that the partially purified PA-activated protein kinase phosphorylated recombinant p47-phox (Rp47), using SDS-PAGE and autoradiography (data not shown). The PA-activated protein kinase was also active in the presence of mixed vesicles, containing PC, PS, and PE, which mimicked the inner membrane lipid layer (67). The concentration of PA yielding maximal protein kinase activation shifted, only slightly, from 30 M PA (3830 pmol PO 4 /min/mg, n ϭ 2) in the absence of mixed vesicles to 100 M PA (3634 pmol PO 4 /min/mg, n ϭ 2) in the presence of mixed vesicles. This suggests that activation of the PA-activated protein kinase can occur in a physiologically relevant manner.
We have previously shown that NADPH oxidase activation and endogenous protein phosphorylation induced by PA plus DG is sensitive to several protein kinase inhibitors (8). Fig. 9 shows that the partially purified PA-activated protein kinase is sensitive to the same inhibitors. In the absence of inhibitors, the GST-p47-phox fusion protein was phosphorylated by the protein kinase only when PA was present. Staurosporine and C-I each completely inhibited the phosphorylation of GST-p47phox, while GF-109203X was less effective. The autoradiograph shown in Fig. 9 is a long exposure to show that little protein phosphorylation occurred in the presence of these protein kinase inhibitors. Staurosporine is known to inhibit both protein serine/threonine kinases and protein-tyrosine kinases (68). C-I, an H-7 analog, inhibits serine/threonine protein kinases but not protein-tyrosine kinases (51). This suggests that the PA-activated protein kinase is a protein serine/threonine kinase.
FIG. 5. Localization of the PA-activated protein kinase and PKC isoforms during ammonium sulfate precipitation. A, PA-dependent protein kinase activity. Neutrophil cytosol (2 mg of protein) was subjected to sequential 40 -70% sequential ammonium sulfate precipitation as described under "Experimental Procedures." PA-dependent protein kinase activity for the 40% ammonium sulfate-precipitated proteins (40%), the 70% ammonium sulfate-precipitated proteins (70%), and the resulting supernatant (Supn.), and cytosol passed over a Sephadex G-25 column (G25), which was used as a control, was determined by the phosphocellulose assay described under "Experimental Procedures." Reactions contained GST-p47-phox fusion protein (6.7 g/ml), 10 M ATP, and 100 M PA. Results are expressed as a percentage of total recovered activity and are the average of two experiments using cytosol from separate donors. B, Western blot. Fractions from the ammonium sulfate precipitation were assessed for the presence of PKC isoforms. 33% of the volume recovered from the Sephadex G-25 column for the 40% precipitate (40%), the 70% precipitate (70%), the resulting 70% supernatant (Supn.), and 1% of the unfractionated cytosol passed over a Sephadex G-25 column (G25), were analyzed by Western blotting as described under "Experimental Procedures." Std., baculovirus-expressed standards. Shown are scans of autoradiographs from one experiment, representative of two performed with cytosols from different donors.
FIG. 6. Separation of the PA-activated protein kinase from PKC-and -␦ during hydrophobic interaction chromatography. A, PA-dependent protein kinase activity. Proteins present in neutrophil cytosol (15 mg of protein) were subjected to 40% saturated ammonium sulfate precipitation; the resulting pellet was resuspended in buffered 20% saturated ammonium sulfate and subjected to hydrophobic interaction chromatography, as described under "Experimental Procedures." PA-dependent protein kinase activity was determined for the unretained protein (Void) and the retained protein (Bound) fractions by the phosphocellulose assay in the presence of GST-p47-phox fusion protein (6.7 g/ml), 200 M ATP, and 30 M PA. Results are expressed as a percentage of total recovered activity and are the average of two experiments using cytosol from separate donors. B, Western blot. Fractions from the experiment shown in A were analyzed for the presence of PKCand -␦. 134 g of the unretained protein fraction (Void), and 94 g of the retained protein fraction (Bound), representing 23 and 45%, respectively, of the total protein recovered for each fraction, were analyzed by Western blotting as described under "Experimental Procedures." Std., baculovirus-expressed standards. Shown are scans of autoradiographs from one experiment, representative of two performed with cytosols from separate donors.
To further evaluate the amino acid specificity of the protein kinase, we performed phosphoamino acid analysis on phosphorylated GST-p47-phox. Fig. 10A clearly shows that serine residues are phosphorylated and that threonine residues are not. However, this analysis did not clearly separate phosphotyrosine from inorganic phosphate. To determine if tyrosine residues were phosphorylated, we extracted each of the phosphoamino acids from the cellulose plate with methanol. This treatment did not elute inorganic phosphate (data not shown). The methanol extracts were then spotted onto cellulose plates and subjected to autoradiography. After extraction, only the phosphoserine and phosphotyrosine extracts resulted in spots, as assessed by autoradiography (data not shown). This confirmed that serine residues were phosphorylated and indicated that tyrosine residues were also phosphorylated. We confirmed tyrosine phosphorylation by Western blot analysis. The GST-p47-phox fusion protein became phosphorylated on tyrosine residues only when PA was present (Fig. 10B). It was possible that the GST portion of the fusion protein was phosphorylated by a protein kinase in the hydropore-bound material. However, we found that hydropore-bound material did not phosphorylate the cleaved GST portion of the fusion protein in the presence of PA (data not shown). In addition, recombinant p47-phox was phosphorylated on tyrosine residues, in a PA-dependent manner, using cytosol as a source for protein kinases (data not shown). Taken together, these data suggest that the hydroporebound material contains one or more protein kinases that phosphorylate serine and tyrosine residues of p47-phox in a PA-dependent manner. Since no mammalian dual specificity protein kinases that phosphorylate both serine and tyrosine residues have been identified, we speculate that the hydropore-bound material contains at least two PA-responsive protein kinases.
Cell and Tissue Distribution of PA-activated Protein Kinase-Since PA is generated in many cell types upon stimulation of PLD (2), we investigated whether cytosol from other cell types contained a protein kinase responsive to PA. Cytosols were isolated as described under "Experimental Procedures," and protein kinase activity was determined in the presence and absence of PA. Several cell types were analyzed. PA-dependent protein kinase activity (defined here as Ͼ2-fold stimulation upon PA addition) was present in all except three cell lines We further investigated the possibility that the same enzyme present in neutrophil cytosol might be present in rat brain cytosol. Rat brain cytosol was subjected to 40% saturated ammonium sulfate precipitation followed by hydrophobic interaction chromatography. The purification profile obtained was FIG. 7. Separation of the PA-activated protein kinase from PAK and MAPK during hydrophobic interaction chromatography. Proteins in neutrophil cytosol (15 mg of protein) were subjected to 40% saturated ammonium sulfate precipitation and hydrophobic interaction chromatography as described in Fig. 5 and "Experimental Procedures." Fractions were analyzed by Western blotting for the presence of PAK-2 and MAPK. The percentage of total protein and amount of protein loaded (in parentheses) of each fraction analyzed was as follows: Cyto., 0.6% (96 g); 40% Cut, 6.4% (119 g); 40% supernatant (40% Supn.), 3% (332 g); unretained protein fraction (Hydropore Void), 23% (134 g); and retained protein fraction (Hydropore Bound) 45% (94 g). Shown are scans of autoradiographs representative of two separate experiments. similar to that obtained with neutrophil cytosol (see Figs. 4 and 5). A PA-dependent protein kinase was precipitated by 40% saturated ammonium sulfate (980 pmol of PO 4 /min/mg, n ϭ 1) and was retained during hydrophobic interaction chromatography (1590 pmol of PO 4 /min/mg, n ϭ 1). This suggests that the same or a similar enzyme was present in these two sources of material. However, other PA-responsive protein kinases appear to be more prevalent in the rat brain cytosol than in neutrophil cytosol, since the supernatant from the 40% ammonium sulfate cut contained 57% of the recovered activity, compared with 15% using neutrophil cytosol. This may account for the smaller increase in specific activity observed in rat brain material (1.8-fold increase) compared with neutrophil material (5-fold increase) after elution from the Hydropore column. DISCUSSION We have recently developed a phosphorylation-dependent cell-free system for the activation of the neutrophil NADPH oxidase (7,8). In this system, PA activates one or more protein kinases that phosphorylate the NADPH oxidase component p47-phox. Here, we have eliminated known protein kinase targets for PA in this system. Our results suggest that PA activates potentially two previously unidentified cytosolic protein kinases, which phosphorylate p47-phox.
We first characterized the lipid specificity for the activation of the PA-dependent protein kinase. Previous studies indicated that DG has only minor enhancing effects on protein phosphorylation induced by PA (8), suggesting that DG-regulated PKC isoforms were unlikely to be targets for PA in this system. Fig.  1 demonstrates that PA induced a high level of protein phosphorylation and activates a protein kinase(s) that is not a target for other phospholipids, supporting the concept that PA activates a novel protein kinase. Bocckino et al. (69) first demonstrated that PA could induce protein phosphorylation in various rat tissues, and our results confirm this. Like them, we found that PC and PE induced minimal protein phosphorylation and that phosphatidylinositol and PS induce moderate levels (Fig. 1). We show here that phosphatidylglycerol can also induce protein phosphorylation (Fig. 1). In the neutrophil system, more proteins appear to be substrates for a PA-dependent protein kinase(s) than in rat tissues.
We considered the possibility that the protein kinase target for PA was one of the recently described fatty acid-activated protein kinases (18 -20), one of which has been shown to be PKN (16), known to be activated by cis-unsaturated fatty acids as well as CL (17) and identical to PRK1 (14,17). The predicted molecular size of PRK1/PKN is 120 kDa (14,15,17,58,70), similar to the apparent size of the predominant protein kinase described here (Fig. 8). PA was much more effective than cis-unsaturated fatty acids at inducing protein kinase activity ( Fig. 4 and text). In addition, CL (3 M to 1.0 mM), was unable to induce protein phosphorylation in neutrophil cytosol ( Fig. 1  and text). Furthermore, while PKN is regulated by the GTPase Rho (71,72), the PA-dependent protein kinase activity described here was independent of GTP␥S. Thus, it is unlikely that Rho could activate PKN under these conditions. These data strongly suggest that PRK1/PKN (14,15,17,58,70) and/or the fatty acid-activated protein kinases previously described (18 -20) are not responsible for the PA-dependent protein kinase activity in neutrophils. It is possible that the neutrophil PA-activated protein kinase is related to these enzymes and is an isoform poorly responsive to cis-unsaturated fatty acids and highly responsive to PA. Such a determination will require identification of the neutrophil enzyme at the molecular level.
Our second approach was to purify the PA-activated protein kinase present in neutrophil cytosol to a stage where we could distinguish the PA-activated protein kinase from other protein kinase targets. The partial purification protocol successfully separated the PA-activated protein kinase(s) from a number of other protein kinases in neutrophils known to phosphorylate p47-phox, including PKC, MAPK, and PAK isoforms (Figs. 5, 6, and 7). We did not examine fractions for the presence of PKC-, which has not been reported to be in neutrophils. Since PKChas a predicted molecular size of 115 kDa (73), it is possible that this isoform is responsible for some of the PA-dependent protein kinase activity in neutrophils. However, PKChas been reported to have a putative transmembrane domain (73) and has been isolated in the particulate material after subcellular localization (74). In contrast, the PA-activated protein kinase is cytosolic. Also, PA has been reported to have little effect on activation of PKC- (75). Taken together, these observations strongly suggest that PKCis not a PA-activated protein kinase in neutrophil cytosol.
MAPK and PAK isoforms were also candidates for the protein kinase(s) responsible for phosphorylation of p47-phox (24,25). However, neither p42 or p44 MAPK isoforms nor PAK-1 or PAK-2 isoforms co-purified with the PA-activated protein kinase ( Fig. 7 and text). In addition, depletion of cytosolic Raf-1 or B-Raf, which could lead to MAPK activation, by immunoprecipitation had little or no effect on PA-dependent protein kinase FIG. 9. Effect of protein kinase inhibitors on PA-activated protein kinase. Proteins present in neutrophil cytosol (15 mg of protein) were subjected to 40% saturated ammonium sulfate precipitation; the resulting pellet was resuspended in buffered 20% saturated ammonium sulfate and subjected to hydrophobic interaction chromatography, as described under "Experimental Procedures." Hydropore-bound material was incubated for 5 min with solvent (PA ϩ DMSO) or inhibitor (100 nM staurosporine (PA ϩ St), 300 M C-I (PA ϩ C-I), or 300 nM GF-109203X (PA ϩ GF)) and then for 30 min with H 2 O or 30 M PA in the presence of GST-p47-phox fusion protein (6.7 g/ml) and [␥-32 P]ATP (7-10 Ci). Phosphorylation was detected after SDS-PAGE by autoradiography. The arrow indicates the position of the GST-p47-phox fusion protein (GST-p47). Shown is a scan of a representative autoradiograph from two separate experiments.
FIG. 10. Determination of phosphorylated amino acids in GST-p47-phox. A, phosphoamino acid analysis. GST-p47-phox (3 g) was phosphorylated with partially purified PA-activated protein kinase and subjected to phosphoamino acid analysis via thin layer chromatography as described under "Experimental Procedures." Location of phosphoamino acid standards are indicated on the right. Shown is a scan of an autoradiograph representative of two separate experiments. B, phosphotyrosine Western blot. Partially purified PA-activated protein kinase was incubated in the presence of cold ATP (10 M), in the presence (ϩ) or absence (Ϫ) of 30 M PA and in the presence or absence of GST-p47-phox (6.7 g/ml) for 30 min as described for the cell-free SDS-PAGE phosphorylation assay. Proteins were separated by SDS-PAGE (6 g of protein/lane) and analyzed by Western blotting for phosphotyrosine as described under "Experimental Procedures." The location of GST-p47-phox is indicated by the arrow. Shown is a scan of an autoradiograph representative of two separate experiments. activity (see text). Finally, the PA-dependent protein kinase activity was GTP␥S-independent, which suggests Rac was not participating in protein kinase activation in our system. We did not examine other protein kinases in these families (24, 28 -31, 76).
These results strongly suggest that PA activates a novel protein kinase in neutrophil cytosol, which phosphorylates the NADPH oxidase component p47-phox. The PA-activated protein kinase was sensitive to the same protein kinase inhibitors that inhibited cell-free NADPH oxidase activation by PA plus DG (Fig. 9) (8). GF-109203X has been described as a selective PKC inhibitor (77), but it was partially effective against the hydropore-purified PA-activated protein kinase. This may indicate that a PA-activated protein kinase is similar to PKC; however, this remains to be determined. The PA-activated protein kinase phosphorylates p47-phox on serine residues (Fig.  10A), which has been shown to occur in vivo (25,78,79). The inhibitor results suggested strongly that the PA-activated enzyme is a protein serine/threonine kinase. However, the phosphoamino acid analysis revealed that PA also activates a protein tyrosine kinase, which phosphorylates p47-phox (Fig. 10, A  and B). Thus, the hydropore-bound material either contains at least two PA-responsive protein kinases, or one of the kinases is activated downstream of the other. It is possible that the known lability of the phosphotyrosine bond (54, 55) accounts for the apparent serine/threonine specificity observed during SDS-PAGE analysis. This also may explain the appearance of only one peak of PA-dependent protein kinase activity during gel filtration (Fig. 8). Further purification of these enzymes will be necessary to address these issues and is beyond the scope of this paper.
Tyrosine phosphorylation of p47-phox has not been reported until now. Previous studies have analyzed the phosphorylation of p47-phox after stimulation of neutrophils with phorbol myristate acetate (25,78,79). Further work needs to be done to examine if other NADPH oxidase agonists stimulate tyrosine phosphorylation of p47-phox, in vivo. In these studies, it will be important to monitor the hydrolysis of p47-phox to maximize the ability to observe potential tyrosine phosphorylation, since phosphotyrosine is more labile than phosphoserine and phosphothreonine (54,55). It is also likely that the phosphorylation of p47-phox on tyrosine residues occurs to a lesser extent than serine phosphorylation. Thus, the amount of p47-phox recovered prior to phosphoamino acid analysis and the assay conditions used are likely to be critical to observe tyrosine phosphorylation in vivo. P47-phox contains several tyrosine residues that may be a target for a protein-tyrosine kinase (80 -82). Further work is necessary to determine which tyrosine residues are phosphorylated in response to PA. Once these sites are determined, the role of such phosphorylation can be examined more closely. It remains to be determined what role, if any, tyrosine phosphorylation of p47-phox has in NADPH oxidase activation.
A primary significance of this study lies in the possibility that the PA-activated protein kinases are new and selective targets for intracellular PA generated by PLD activation. In neutrophils, the PA-activated protein kinase(s) could participate in the regulation of a variety of cell functions (1,2,83), such as activation of the NADPH oxidase. Activation of PLD and the resulting production of PA have been closely linked to the activation of the oxidase (4 -8, 84). Our previous data clearly indicate that a PA-activated protein kinase participates in the PA-and DG-dependent cell-free activation of NADPH oxidase (8). A PA-activated protein kinase retained the ability to phosphorylate p47-phox, even after three steps of purification (Fig. 8B), providing additional evidence indicating that the NADPH oxidase component p47-phox is a substrate for the enzyme. However, it remains to be determined if phosphorylation of p47-phox by a PA-activated protein kinase regulates the assembly and activation of the NADPH oxidase enzyme.
It is notable that we found PA-dependent protein kinase activity in a wide range of blood cells and hematopoietic cell lines and in rat brain cytosol, suggesting that the enzyme may be widely distributed. Activity was not apparent in the hepatoma and neuroblastoma cell lines, but the significance of these preliminary observations is not known. It is not yet clear if the same enzymes are responsible for the activity in all of these cells and in rat brain. However, like the enzyme in neutrophils, a PA-activated protein kinase in rat brain cytosol was precipitated at 40% ammonium sulfate saturation and retained on the Hydropore column. Once a PA-activated protein kinase has been purified and sequence information is available, it will be possible to use molecular and immunological approaches to definitively address the cell and tissue distribution of the enzyme.
In conclusion, we have shown that neutrophil cytosol contains potentially two novel PA-responsive protein kinases. The molecular weight of one of these protein kinases is 125 kDa. We have differentiated these enzymes from other known protein kinases that could be targeted by PA in our system and have evidence that the enzymes may be widely distributed in cells and tissues. These protein kinases phosphorylate p47-phox, a necessary component of the NADPH oxidase. Future work will focus on purifying and sequencing the PA-responsive protein kinases to identify them as either novel enzymes or known protein kinases with a new function. In addition, we plan to define the role of PA-activated protein kinases in the phosphorylation-dependent activation of the NADPH oxidase.