A Phosphatidic Acid-activated Protein Kinase and Conventional Protein Kinase C Isoforms Phosphorylate p22 phox , an NADPH Oxidase Component*

Using a phosphorylation-dependent cell-free system to study NADPH oxidase activation (McPhail, L. C., Qualliotine-Mann, D., and Waite, K. A. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 7931–7935), we previously showed that p47 phox , a cytosolic NADPH oxidase component, is phosphorylated. Now, we show that p22 phox , a subunit of the NADPH oxidase component flavocytochrome b 558, also is phosphorylated. Phosphorylation is selectively activated by phosphatidic acid (PA) versus other lipids and occurs on a threonine residue in p22 phox . We identified two protein kinase families capable of phosphorylating p22 phox : 1) a potentially novel, partially purified PA-activated protein kinase(s) known to phosphorylate p47 phox and postulated to mediate the phosphorylation-dependent activation of NADPH oxidase by PA and 2) conventional, but not novel or atypical, isoforms of protein kinase C (PKC). In contrast, all classes of PKC isoforms could phosphorylate p47 phox . In a gel retardation assay both the phosphatidic acid-dependent kinase and conventional PKC isoforms phosphorylated all molecules of p22 phox . These findings suggest that phosphorylation of p22 phox by conventional PKC and/or a novel PA-activated protein kinase regulates the activation/assembly of NADPH oxidase.

Using a phosphorylation-dependent cell-free system to study NADPH oxidase activation (McPhail, L. C., Qualliotine-Mann, D., and Waite, K. A. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 7931-7935), we previously showed that p47 phox , a cytosolic NADPH oxidase component, is phosphorylated. Now, we show that p22 phox , a subunit of the NADPH oxidase component flavocytochrome b 558 , also is phosphorylated. Phosphorylation is selectively activated by phosphatidic acid (PA) versus other lipids and occurs on a threonine residue in p22 phox . We identified two protein kinase families capable of phosphorylating p22 phox : 1) a potentially novel, partially purified PA-activated protein kinase(s) known to phosphorylate p47 phox and postulated to mediate the phosphorylation-dependent activation of NADPH oxidase by PA and 2) conventional, but not novel or atypical, isoforms of protein kinase C (PKC). In contrast, all classes of PKC isoforms could phosphorylate p47 phox . In a gel retardation assay both the phosphatidic acid-dependent kinase and conventional PKC isoforms phosphorylated all molecules of p22 phox . These findings suggest that phosphorylation of p22 phox by conventional PKC and/or a novel PA-activated protein kinase regulates the activation/assembly of NADPH oxidase.
Phagocytic cells are the first line of defense against invading microorganisms (reviewed in Ref. 1). This defense is achieved, in part, by the respiratory burst, leading to the production of superoxide anion, which along with its metabolic products are toxic to invading microorganisms. The respiratory burst is mediated by the multicomponent enzyme complex, the NADPH oxidase. The two membrane-bound components (p22 phox and gp91 phox ) form the heterodimeric flavocytochrome b 558 (2). The flavocytochrome contains a putative NADPH binding site, FAD, and two hemes; thus, it possesses all of the electron machinery required to transfer two electrons from NADPH to molecular oxygen (3)(4)(5)(6)(7)(8)(9)(10). The three required cytosolic components are Rac-GTP and the phosphoproteins p47 phox and p67 phox (11) (reviewed in Ref. 12). Activation of phagocytic cells leads to the translocation of the cytosolic components to the membrane, where they interact with flavocytochrome b 558 (reviewed in Ref. 12). Assembly of the NADPH oxidase components is required for activation of electron flow, possibly by inducing a conformational change within the flavocytochrome b 558 component.
The signaling mechanisms leading to assembly and activation of NADPH oxidase are not well understood. When a ligand binds its receptor on the neutrophil membrane, a cascade of events is initiated (reviewed in Ref. 1) that includes activation of phospholipases, generation of lipid second messengers, and the activation of protein kinases. One of the phospholipases activated in response to many physiological agonists of neutrophils is phospholipase D (13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27). Phospholipase D cleaves neutrophil phospholipids to form phosphatidic acid (PA), 1 which can then be converted to diacylglycerol by PA phosphohydrolase. Numerous studies have correlated phospholipase D activation/PA production and NADPH oxidase activation (19, 28 -34). Thus, it has been hypothesized that phospholipase D and its product, PA, play important roles in the signal transduction mechanisms leading to superoxide anion production.
Recently, our laboratory developed and characterized a cellfree system for NADPH oxidase activation, which was synergistically activated by PA ϩ diacylglycerol (35,36). NADPH oxidase activation was enhanced by ATP and reduced by protein kinase inhibitors. Furthermore, PA induced the phosphorylation of several neutrophil proteins. These data suggest that phosphorylation-dependent mechanisms are involved in the activation of NADPH oxidase in this cell-free system. We have reported that p47 phox is phosphorylated in this system by a novel, cytosolic PA-activated protein kinase (36,37). Now we report that p22 phox also is phosphorylated in the system in a PA-dependent manner. Phosphorylation of p22 phox was first observed by Garcia and Segal (38) in intact cells. We have now characterized the phosphorylation of p22 phox in vitro and show that a potentially novel, PA-activated protein kinase and conventional but not other protein kinase C (PKC) isoforms are able to phosphorylate this protein.
Neutrophils from patients with chronic granulomatous disease (CGD) were prepared in a similar fashion; however, the sonication buffer contained 11% sucrose, 130 mM NaCl, 5 mM EGTA, and 1 mM phenylmethylsulfonyl fluoride. Individuals from two families (four patients total) were obtained. Two males (CBI and CBII) from one family have ϳ10% of normal cytochrome b 558 levels (46,47); whereas, one male and one female from a second family (JB and EB) have no detectable cytochrome b 558 . 2 In Vitro PA-dependent Protein Phosphorylation-Reaction mixtures (150 l) contained 50 mM Na x PO 4 , pH 7.0, 1 mM EGTA, 5 mM MgCl 2 , either neutrophil membrane fractions (3.1-6.25 g of protein) or another substrate, and a lipid activator (usually 10 M di10:0PA) (36,37). 10 M [␥-32 P]ATP (ϳ10 Ci) and neutrophil cytosol (12.5-25 g of protein) or another source of protein kinase were added, and the reaction mixture was allowed to incubate for the times indicated in figure legends. The reaction was stopped by the addition of Laemmli sample buffer (48) for analysis by SDS-PAGE, autoradiography, and densitometry. 5ϫ NaCl Solution (5 M NaCl, 25 mM EDTA, 25 mM EGTA, 5 M staurosporine, 25 mM sodium orthovanadate, 5 M microcystin, 125 mM NaF, 5 mM p-nitrophenylphosphate, 50 M benzamidine, 50 g/ml leupeptin, 50 M pepstatin, 5 g/ml aprotinin, and 5 mM phenylmethylsulfonyl fluoride) was used to quench reactions that were used for solubilization and subsequent immunoprecipitation (see below).
In Vitro Protein Kinase C-mediated Protein Phosphorylation-For phosphorylation by conventional and novel PKC isoforms, previously described conditions were used (42,44,49). For PKC experiments, 100-l reaction mixtures contained 25 mM Tris, pH 7.5, 5 mM MgCl 2 , 0.5 mM EGTA, and 1 mM dithiothreitol with or without the addition of 100 g/ml PS. For PKC reaction mixtures containing GST-p47, 25 g/ reaction whale myoglobin was added. 3 Proteins were analyzed by separation on SDS-PAGE, transferred to nitrocellulose, and analyzed by autoradiography/densitometry or PhosphorImager analysis (Molecular Dynamics, Sunnyvale, CA). Western blotting was performed to confirm p22 phox and GST-p47 protein levels (described below).
Two methods for immunoprecipitation were used. In the first method, CNBr-activated Sepharose beads were conjugated to either mAb 44.1 or the isotype control, according to manufacturer's directions (Amersham Pharmacia Biotech). Solubilized membrane was precleared with isotype control antibodies-conjugated beads and then incubated with the mAb 44.1-conjugated (ϳ150 g of solubilized membrane protein/15 l antibody-conjugated beads) at 4°C for 16 h. Alternatively, protein A-Sepharose and either the isotype control antibody or mAb 44.1 were incubated with the solubilized membrane (500 g solubilized membrane/20 g antibody) at 4°C for 16 h. The beads were washed, and the immunoprecipitated proteins were analyzed by 14 or 8 -15% SDS-PAGE.
SDS-PAGE, Autoradiography, and Densitometry-Protein samples were prepared for analysis by SDS-PAGE using Laemmli sample buffer (48) and separated on 8 -15, 7, or 14% SDS-PAGE. Gels were stained with either Coomassie Brilliant Blue R-250 or GelCode Blue Stain (Pierce), destained, dried, and analyzed by autoradiography and densitometry. In some experiments, proteins were transferred electrophoretically (51) to nitrocellulose for autoradiography and Western blot analysis. Autoradiographs were analyzed by scanning densitometry (PDI, Huntington Station, NY).
Phosphoamino Acid Analysis-Phosphorylated p22 phox was excised from polyvinylidene difluoride membranes and subjected to hydrolysis with 6 N HCl for 1.5 h at 110°C (54). Phosphoamino acids were separated by thin layer electrophoresis at 1100 V for 45 min in water:acetic acid:pyridine (189:10:1, pH 3.5) (11,54). H 3 [ 32 P]O 4 was used as a standard for the migration of inorganic phosphate. Fig. 1A (left two lanes), when neutrophil membrane and cytosolic fractions were incubated with di10:0PA, a 22-kDa protein was phosphorylated. Because this phosphoprotein was the appropriate size for the light chain of the flavocytochrome, we hypothesized it could be p22 phox . Neutrophil membrane fractions from patients with flavocytochrome b 558 -deficient CGD were substituted for normal membrane fractions in the reaction mixture. Membrane fractions from a total of four patients, two with the X-linked form of CGD and two with the autosomal recessive form, were tested. The two patients with X-linked CGD were previously characterized as having ϳ10% of the normal levels of flavocytochrome b 558 (46,47), whereas the sibling pair with the autosomal recessive form had no detectable flavocytochrome b 558 . 2 As shown in Fig. 1A (right two lanes), the substitution of neutrophil membrane fractions from a patient having ϳ10% normal levels of flavocytochrome b 558 resulted in markedly reduced levels of the phosphorylated 22-kDa protein. Similar results were obtained with membrane fractions from the other X-linked patient, and no PA-dependent phosphorylation of a 22-kDa band was observed using membrane fractions from the two patients totally deficient in flavocytochrome b 558 (data not shown).

Identification of p22 phox as a Substrate for a Neutrophil PA-activated Protein Kinase-As shown in
To further confirm that p22 phox was phosphorylated, we performed immunoprecipitation experiments using a monoclonal antibody to the protein (41). Following phosphorylation in the presence of cytosol, membrane fractions were reisolated, and solubilized proteins were subjected to immunoprecipitation with the p22 phox -specific antibody. As shown in Fig. 1B (left two lanes), a phosphorylated 22-kDa protein was immunoprecipitated from membrane fractions incubated with neutrophil cytosol in the presence of di10:0PA. Very little phosphorylation was observed in immunoprecipitates from membrane fractions incubated in the presence of neutrophil cytosol without the addition of di10:0PA. In contrast, no phosphorylated proteins were immunoprecipitated with the isotype control antibody (Fig. 1B, right two lanes).
We performed two additional experiments to verify that p22 phox is phosphorylated by a PA-dependent protein kinase present in neutrophil cytosol. First, purified, relipidated flavocytochrome b 558 was mixed with neutrophil cytosolic fractions in the absence or presence of di10:0PA. In the presence of purified flavocytochrome b 558 , di10:0PA clearly induces the phosphorylation of a protein migrating at 22 kDa (Fig. 1C, right three lanes). Because only a faint band at 22 kDa is observed in the absence of the purified flavocytochrome b 558 (Fig. 1C, left two lanes), these results indicate that the 22-kDa band is p22 phox . In the second experiment, we tested whether a recombinant form of p22 phox could undergo PA-dependent phosphorylation. We used a GST-p22 phox fusion protein containing amino acids 127-195 of p22 phox , comprising the putative cytosolic tail of the protein (55). This region of the protein also contains three protein kinase phosphorylation motifs (Wisconsin Package version 9.0, Genetics Computing Group, Madison, WI). Therefore, it might be expected that this portion of the protein would contain a PA-dependent phosphorylation site(s). As shown in Fig. 1D, in the presence of di10:0PA and neutrophil cytosolic fractions, the p22 phox peptide was phosphorylated. Thus, the cytosolic portion of p22 phox contains one or more PA-dependent phosphorylation site(s). Therefore, p22 phox can be phosphorylated by a PA-dependent protein kinase, pre-sumably present in neutrophil cytosol. In none of these experiments did we observe evidence of phosphorylation of the flavocytochrome b heavy chain, gp91 phox (data not shown).
Characterization of the PA-dependent Phosphorylation of p22 phox -To characterize the phosphorylation of p22 phox by a PA-activated protein kinase, we varied the concentration of di10:0PA from 0 to 300 M, and the time of incubation from 0 to 120 min. Optimal phosphorylation of p22 phox was obtained with incubation of neutrophil cytosol and membrane fractions in the presence of 10 M di10:0PA for 60 min (data not shown). This concentration of PA is within the range of that measured in stimulated neutrophils (56), thus indicating that phosphorylation is mediated by a physiological concentration of PA.
We next determined the lipid specificity for the induction of p22 phox phosphorylation. At 10 M, only di10:0 PA (PA) clearly induced p22 phox phosphorylation. However, the addition of 10 M PS or phosphatidylinositol (PI) caused a faint darkening over background in the 22 kDa range ( Fig. 2A, top panel). A concentration curve with PS (0 -300 M) revealed a low level of p22 phox phosphorylation with a maximum at 100 M (data not shown). We then screened various lipids at 100 M for their ability to induce p22 phox phosphorylation. At this concentration, PS, phosphatidylinositol (PI), phosphatidylglycerol (PG), and 1-oleoyl-sn-glycero-3-phosphate (LPA) induced low levels of p22 phox phosphorylation ( Fig. 2A). However, di10:0PA clearly induced the highest response.
Identification of Protein Kinases That Phosphorylate p22 phox -To identify the type of protein kinase that mediates p22 phox phosphorylation, we tested the effect of several protein kinase inhibitors. 1-(5-isoquinolinesulfonyl)piperazine, an H-7 analog, selectively blocks protein Ser/Thr kinases (42). Staurosporine inhibits both protein Ser/Thr and tyrosine kinases (57,58). GF109203X is a PKC-selective inhibitor (59). Each of these three inhibitors blocked p22 phox phosphorylation by 70 -85% (Fig. 3). In contrast, genistein, a protein tyrosine kinase inhibitor (60), had no effect on the level of p22 phox phosphorylation (Fig. 3). These data indicate that p22 phox is likely phosphorylated by a protein Ser/Thr kinase rather than a tyrosine kinase. Furthermore, the same Ser/Thr protein kinase inhibitors decrease NADPH oxidase activation by 70 -80% in the phosphorylation-dependent cell-free system (36).
A similar profile of inhibition was observed for the partially purified, cytosolic PA-activated protein kinase (37), suggesting it might be responsible for the phosphorylation of p22 phox . As shown in Fig. 4A, no phosphorylation of p22 phox was observed without the addition of neutrophil cytosol, thus indicating that the PA-activated protein kinase responsible for p22 phox phosphorylation is cytosolic. Next, we examined the ability of the partially purified PA-activated protein kinase to phosphorylate p22 phox . The protein kinase was partially purified by precipi-tation with 40% saturated ammonium sulfate followed by passage over a hydrophobic interaction column, as described previously (37). As shown in Fig. 4B (right two lanes), when the partially purified protein kinase replaced cytosol in the reaction mixture, PA-dependent phosphorylation of p22 phox was observed. Thus, at least one protein kinase that can phosphorylate p22 phox is the apparently novel PA-activated protein kinase.
We next tested the ability of PKC to phosphorylate p22 phox , based on the following rationale. Many groups, including ours, have correlated PKC activation with the activation/assembly of NADPH oxidase (Refs. 44 and 61; reviewed in Ref. 1). Sequence analysis of p22 phox revealed a PKC phosphorylation motif ((R/ K)X(S/T) or (S/T)X(R/K)) at Thr 147 (Wisconsin Package version 9.0, Genetics Computing Group). This region is in the C-terminal, putative cytosolic tail of p22 phox (55). GF109203X, a PKCselective inhibitor, diminished the phosphorylation of p22 phox by ϳ80% (Fig. 3). Garcia and Segal (38) reported in 1988 that p22 phox is phosphorylated in whole neutrophils stimulated with phorbol myristate acetate, a potent PKC activator. Therefore, we tested whether a purified preparation of PKC (rat brain) was able to induce the phosphorylation of p22 phox . According to the manufacturer (Calbiochem), the rat brain PKC preparation consisted primarily of conventional isoforms. Under conditions optimal for PKC activation, neutrophil membrane fractions were incubated in either the absence or presence of PKC and in the absence or presence of PKC activators (PS, 1,2-dioleoyl-racglycerol, and Ca 2ϩ ). As shown in Fig. 4C (top panel), the phosphorylation of p22 phox was readily apparent in the presence of PKC and its activators. A low level of p22 phox phosphorylation was observed in the absence of exogenous PKC, possibly because of slight contamination of the membrane fraction with cytosol containing endogenous PKC isoforms (44). To verify that PKC was active under these conditions, we observed increased autophosphorylation of the purified PKC (middle panel). In addition, p47 phox , a known substrate for rat brain PKC (62), underwent activator-dependent phosphorylation by PKC (bottom panel). Thus, PKC is capable of phosphorylating p22 phox .
To further characterize the phosphorylation of p22 phox by PKC, we obtained individual human, recombinant PKC isoforms known to be expressed in human neutrophils (␣, ␤I, ␤II, ␦, and ) (63-66) and tested each for the ability to phosphorylate p22 phox . Neutrophil membrane fractions were incubated with each PKC isoform under optimal assay conditions (44, 62). As shown in Fig. 5A (top panel), the conventional PKC isoforms (␣, ␤I, and ␤II), but not the novel PKC ␦, were able to phosphorylate p22 phox . In contrast, all of the PKC isoforms tested were able to phosphorylate p47 phox (Fig. 5A, bottom panel).
PKC was tested under slightly different conditions (see "Experimental Procedures"), which were recommended by the manufacturer. PKC was able to phosphorylate p47 phox (Fig.  5B) but not p22 phox (data not shown). The phosphorylation of p47 phox induced by PKC was slightly greater than that induced by PA, using cytosol as a protein kinase source. To  1 g of protein) and [␥-32 P]ATP were incubated with either neutrophil cytosol (12.5 g of protein) or the partially purified PA-activated protein kinase (0.5 g/reaction, PAPK) in the absence (Ϫ) or presence (ϩ) of 10 M di10:0PA for 60 min at 25°C. Reactions were stopped and proteins were analyzed as described in A. The scan of an autoradiograph shown is representative of four experiments using three preparations of kinase, purified as described (37). C, purified PKC from rat brain phosphorylates p22 phox . Neutrophil membrane fractions (6.25 g of protein, top panel), buffer (middle panel), or 1 g of GST-p47 (bottom panel) were incubated for 30 min at 30°C in the presence of 100 ng purified rat brain PKC (ϳ100 pmol phosphate transferred/min) and [␥-32 P]ATP, either in the presence of 10 mM EGTA (Ϫ) or in the presence of PKC activators (ϩ; 20 g/ml PS, 2 g/ml 1,2-dioleoyl-rac-glycerol, 0.6 mM CaCl 2 ). Position of each substrate (p22 phox , PKC, GST-p47) is denoted by arrows. The scans shown are representative of three experiments.
FIG. 5. Phosphorylation of p22 phox and p47 phox by various human PKC isoforms. A, phosphorylation by PKC␣, ␤I, ␤II, and ␦. Neutrophil membrane fractions (6.25 g of protein) or GST-p47 (1 g/reaction) were incubated at 30°C for 30 min in the presence of [␥-32 P]ATP and the indicated recombinant PKC isoforms (ϳ95 pmol phosphate transferred/min) either without (Ϫ, ϩ10 mM EGTA) or with (ϩ) the addition of PKC activators (20 g/ml bovine brain PS, 2 g/ml 1, 2-dioleoyl-rac-glycerol, 0.6 mM CaCl 2 for PKC␣, ␤I, and ␤II; 10 mM EGTA, 20 g/ml bovine brain PS and 2 g/ml 1, 2-dioleoyl-rac-glycerol for PKC␦). Proteins were separated on 8 -15% (p22 phox ) or 7% (GST-p47) SDS-PAGE and transferred to nitrocellulose. Blots were analyzed using the PhosphorImager followed by Western blotting for p22 phox and GST-p47. Data were normalized as the percentage of phosphorylation obtained with PKC␣, in the presence of activators, after correction for slight differences in p22 phox and GST-p47 protein levels. Data are shown as the means Ϯ S.E. (n ϭ 3-4). B, phosphorylation by PKC. 1 g of GST-p47 and [␥-32 P]ATP were mixed with neutrophil cytosol (25 g of protein), under PA-dependent phosphorylation conditions (see "Experimental Procedures"), or with PKC (95 pmol phosphate transferred/ min), under PKC phosphorylation conditions (see "Experimental Procedures"), in the presence of 10 M di10:0PA, 0.2 mg/ml PS, or no lipid, as indicated. Reaction mixtures containing cytosol were incubated at 25°C for 30 min, and mixtures containing PKC were incubated at 30°C for 30 min. Proteins were separated by 7% SDS-PAGE and transferred to nitrocellulose. Protein phosphorylation was quantitated using the PhosphorImager. All data were corrected for slight variations in GST-p47 levels, as determined by Western blot analysis. The data were normalized as the percentage of the ϩPA, ϩcytosol condition and are given as means Ϯ S.E. (n ϭ 3-5) or as the average of two experiments, as indicated. p22 phox Phosphorylation by Lipid-dependent Protein Kinases summarize, these data show differences in the ability of classes of PKC isoforms to phosphorylate p22 phox but not p47 phox .
The assays used for PKC activation differ in several respects from the assays used for phosphorylation of p22 phox by the PA-dependent protein kinase(s) in neutrophil cytosol (see "Experimental Procedures"). Therefore, we tested whether purified PKC could substitute for cytosol and mediate PA-dependent phosphorylation of p22 phox , under conditions optimal for the PA-dependent protein kinase. Little, if any, PA-dependent p22 phox phosphorylation was observed when cytosol was replaced with the purified rat brain PKC, a source of conventional PKC isoforms, or PKC (Fig. 6). However, rat brain PKC and PKC were able to mediate low levels of phosphorylation of another endogenous membrane protein, indicating that the PKC isoforms were activated (Fig. 6). Western blotting demonstrated that the phosphorylation patterns observed were not caused by variations in p22 phox protein levels (data not shown). Based on these results, it is unlikely that PKC isoforms present in neutrophil cytosol are responsible for the phosphorylation of p22 phox observed under our PA-dependent assay conditions (used in Figs. 1, 2, 3, and 4A).
We next developed an assay to determine the proportion of phosphorylated to unphosphorylated p22 phox . Using 8 -15% SDS-PAGE followed by transfer of the proteins to nitrocellulose and Western blot analysis, we observed a retardation in the migration of phosphorylated versus nonphosphorylated p22 phox . Using this assay, we have shown that all of the p22 phox becomes phosphorylated under conditions optimal for the PAdependent protein kinase (Fig. 7A) and for PKC (Fig. 7B). Furthermore, retardation of p22 phox migration was observed when p22 phox was phosphorylated with the human, recombinant, conventional PKC isoforms (data not shown). The retardation in migration was not due to lipid/p22 phox interactions because the retardation was prevented when reactions were performed in the absence of neutrophil cytosol or in the presence of protein Ser/Thr kinase inhibitors (data not shown).
Phosphoamino acid analysis also was performed on p22 phox phosphorylated by either the PA-activated protein kinase or rat brain PKC (Fig. 8). Under both conditions, only threonine residues became phosphorylated. This result along with the gel retardation data above (Fig. 7), suggest that both protein kinases may phosphorylate the same site in p22 phox .

DISCUSSION
It has been established that protein kinase-mediated reactions are critical for the activation of the neutrophil respiratory burst enzyme, NADPH oxidase, and that several enzyme components are phosphorylated during cell stimulation. Phosphorylation of p47 phox can be mediated by PKC, cAMP-dependent protein kinases, mitogen-activated protein kinases, p21-activated kinases, and a potentially novel PA-activated protein kinase (36,37,62,(67)(68)(69). Recently, it was found that p67 phox is phosphorylated by p21-activated kinase (11,70) and p40 phox is phosphorylated in stimulated HL60 cells and in vitro by casein kinase II (71). Both gp91 phox and p22 phox , components of flavocytochrome b 558 , were identified as phosphoproteins in stimulated neutrophils (38), and the phosphorylation of Rap1A is mediated by cAMP-dependent protein kinase (72). However, the mechanism by which these phosphorylation events regulate NADPH oxidase activation are not clear.
To aid in the identification of protein kinase-dependent reactions participating in the activation mechanism, we have developed a phosphorylation-dependent cell-free system, leading to the activation of NADPH oxidase (36). Previously (36,37), we identified one NADPH oxidase component, p47 phox , as a substrate in this system. Here we show that the light chain of the flavocytochrome b 558 component of NADPH oxidase, p22 phox , also is phosphorylated in this cell-free system. The phosphorylated protein migrating at 22 kDa was identified as p22 phox by several criteria (Fig. 1). Furthermore, retardation of the migration of p22 phox protein on SDS-PAGE gels was observed (Fig. 7), thus confirming that p22 phox was indeed phosphorylated.
We did not observe phosphorylation of the heavy chain of flavocytochrome b 558 in any of the reaction mixtures, although the presence of other proteins might have obscured visualization of the glycosylated protein. This differs from a report that both subunits of the flavocytochrome were phosphorylated by stimulation of intact neutrophils (38). Possibly, the protein kinase responsible for gp91 phox phosphorylation in intact cells is not activated in our PA-dependent cell-free system. Phosphorylation of p22 phox in the cell-free system was induced by PA, but not by diacylglycerol, similar to previous results examining phosphorylation of p47 phox ( Fig. 2A and Ref. 37). This suggests that the protein kinase activated by PA and responsible for phosphorylating p22 phox is the same protein kinase that phosphorylates p47 phox . Indeed, the potentially novel PA-activated protein kinase we have partially purified could substitute for cytosol as the source of the protein kinase phosphorylating p22 phox (Fig. 4B). Thus, this as yet unidentified protein kinase can utilize at least two NADPH oxidase components as substrates. Based on previous studies correlating PA levels in neutrophils with NADPH oxidase activity, (19, 28 -34), we speculate that the PA-activated protein kinase is responsible for phosphorylation of one or both of these components in intact cells. Studies to identify the PA-activated protein kinase are underway and must be completed before we can explore its function in intact neutrophils.
Interestingly, differences in lipid activator specificity for the phosphorylation of p22 phox and p47 phox were observed in vitro. Phosphorylation of p22 phox was more selective for PA, compared with other phospholipids, than was the phosphorylation of p47 phox ( Fig. 2A and Ref. 37). The basis for this difference is unclear, but it may be related to physical differences in the two substrates. Studies were performed using p22 phox as an integral membrane protein, whereas p47 phox is provided as a soluble protein. Perhaps, PA selectively induces membrane association of the protein kinase and, thus, accessibility to p22 phox . Alternatively, PA may be specifically needed to induce a required conformational change in p22 phox before phosphorylation can occur. PA is known to directly interact with flavocytochrome b 558 (73)(74)(75). Further studies will be required to understand the basis for this substrate-dependent selectivity of PA-stimulated phosphorylation.
We found that only the conventional PKC isoforms could phosphorylate p22 phox (Fig. 5A, top panel). This is in contrast to results using p47 phox as substrate, in which all classes of PKC isoforms were effective (Fig. 5A, bottom panel). It is unlikely that gross differences in substrate accessibility account for these differences, because all classes of PKC isoforms are known to become membrane-associated in the presence of their activators (reviewed in Refs. 76 -78). Indeed, we found that the autophosphorylation of each PKC isoform was stimulated by the presence of membrane fractions (source of p22 phox ) (data not shown), indicative of an interaction between the PKC and the membrane. The differences between p22 phox and p47 phox as substrates for PKC may have functional implications. Our results predict that p47 phox , but not p22 phox , could become phosphorylated by calcium-independent PKC isoforms (PKC ␦, PKC ) in the absence of an increase in intracellular Ca 2ϩ . Indeed, Dusi and co-workers (79) found that p47 phox became phosphorylated in calcium-depleted neutrophils treated with formylmethionyl-leucyl-phenylalanine and concanavalin A. Thus, these results and our data suggest that a calcium-independent mechanism for p47 phox phosphorylation and NADPH activation could involve novel or atypical PKC isoforms.
Under both PKC and PA-activated protein kinase conditions, p22 phox is phosphorylated on threonine residues and undergoes a retardation in migration by SDS-PAGE. Thus, all of the p22 phox becomes phosphorylated. According to sequence simi- FIG. 8. Phosphorylation of p22 phox occurs on a threonine residue. 25 g of neutrophil membrane was mixed with either 100 g of neutrophil cytosol (cytosol) Ϯ di10:0PA for 60 min at 25°C or 133 ng of rat brain PKC (PKC, 160 pmol of phosphate transferred/min) Ϯ 20 g/ml PS, 2 g/ml 1,2-dioleoyl-rac-glycerol, and 0.6 mM Ca 2ϩ for 30 min at 30°C. Reactions were stopped by the addition of 5ϫ NaCl buffer, and p22 phox was immunoprecipitated. The immunoprecipitates were separated on 8 -15% SDS-PAGE, transferred to polyvinylidene fluoride membrane, and subjected to autoradiography, and the band corresponding to p22 phox was excised. The p22 phox protein was hydrolyzed, and the phosphoamino acids were separated using thin layer electrophoresis, as described under "Experimental Procedures." The scan of an autoradiograph shown is representative of three experiments with rat brain PKC and four with neutrophil cytosol. The migration of phosphoserine, phosphothreonine, phosphotyrosine, and inorganic phosphate standards are shown on the right. p22 phox Phosphorylation by Lipid-dependent Protein Kinases larity alignments, two threonine residues in the p22 phox sequence are putative phosphorylation motifs (Thr 147 , PKC; Thr 132 , casein kinase II) (Wisconsin Package version 9.0, Genetics Computing Group) and are candidate phosphorylation sites.
In conclusion, p22 phox can be phosphorylated by an unknown PA-activated protein kinase as well as by conventional PKC isoforms. Furthermore, p47 phox can be phosphorylated by all of the PKC isoforms known to be expressed in human neutrophils. Thus, there is selectivity of various PKC isoforms for the phosphorylation of NADPH oxidase components. This information will aid in understanding the role of PKC isoforms in the activation of NADPH oxidase in intact neutrophils. Furthermore, we have identified a second substrate, p22 phox , for a PA-dependent protein kinase. Because multiple NADPH oxidase components are now known to be substrates for several protein kinases (Refs. 11, 36 -38, 62, and 67-72 and this paper), it is clear that the phosphorylation-dependent activation of this enzyme is complex.